(Studies in Language Companion Series 138) Gabriele Diewald, Leena Kahlas-Tarkka, Ilse Wischer-Comparative Studies in Early Germanic Languages_ With a Focus on Verbal Categories-John Benjamins Publish

Comparative Studies in Early Germanic Languages
Studies in Language Companion Series (SLCS)

This series has been established as a companion series to the periodical
Studies in Language.
For an overview of all books published in this series, please see

http://benjamins.com/catalog/slcs
Editors
Werner Abraham Elly van Gelderen
University of Vienna / Arizona State University
University of Munich
Editorial Board
Bernard Comrie Christian Lehmann
Max Planck Institute, Leipzig University of Erfurt
and University of California, Santa Barbara
Marianne Mithun
William Croft University of California, Santa Barbara
University of New Mexico
Heiko Narrog
sten Dahl Tohuku University
University of Stockholm
Johanna L. Wood
Gerrit J. Dimmendaal University of Aarhus
University of Cologne
Debra Ziegeler
Ekkehard Knig University of Paris III
Free University of Berlin
Volume 138
Comparative Studies in Early Germanic Languages.
With a focus on verbal categories
Edited by Gabriele Diewald, Leena Kahlas-Tarkka and Ilse Wischer
Comparative Studies
in Early Germanic Languages
With a focus on verbal categories
Edited by
Gabriele Diewald
Leibniz University Hannover
Leena Kahlas-Tarkka
University of Helsinki
Ilse Wischer
University of Potsdam
John Benjamins Publishing Company

Amsterdam/Philadelphia
TM
The paper used in this publication meets the minimum requirements of
8
theAmerican National Standard for Information Sciences Permanence

of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data
International Conference on English Historical Linguistics (16th : 2010 : Pcs, Hungary)

Comparative Studies in Early Germanic Languages : With a focus on verbal categories /
Edited by Gabriele Diewald, Leena Kahlas-Tarkka and Ilse Wischer.
p. cm. (Studies in Language Companion Series, issn 0165-7763 ; v. 138)
This publication comprises the papers presented at a workshop on the Contrastive study
of the verbal categories and their grammaticalisation in Old English and Old High
German held at the 16th ICEHL in Pcs, Hungary, in August 2010.
Includes bibliographical references and index.
1. English language--Grammar, Comparative. 2. English language--Grammar,
Historical. 3. Germanic languages--Grammar, Comparative. 4. Germanic
languages--Grammar, Historical. 5. Grammar, Comparative and general--
Grammaticalization. I. Diewald, Gabriele, editor of compilation. II. Kahlas-
Tarkka, Leena, editor of compilation. III. Wischer, Ilse, 1959- editor of
compilation. IV. Title.
PE1075.I57 2013
429.56--dc23 2013020757
isbn 978 90 272 0605 3 (Hb ; alk. paper)
isbn 978 90 272 7145 7 (Eb)
2013 John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Co. P.O. Box 36224 1020 me Amsterdam The Netherlands
John Benjamins North America P.O. Box 27519 Philadelphia pa 19118-0519 usa
Table of contents
Introduction 1
Gabriele Diewald, Leena Kahlas-Tarkka & Ilse Wischer
*haitan in Gothic and Old English 17
Robert A. Cloutier
Incipient Grammaticalisation: Sources of passive constructions
in Old High German and Old English 41
Robert Mailhammer & Elena Smirnova
Passive auxiliaries in English and German: Decline versus
grammaticalisation of bounded language use 71
Peter Petr
Causative habban in Old English: Tracing the Development
of a Budding Construction 101
Matti Kilpi
Remembering (ge)munan: The rise and decline of a potential modal 127
Matthias Eitelmann
The emergence of modal meanings from haben with zu-infinitives
in Old High German 151
Anne Jger
Hearsay and lexical evidentials in Old Germanic languages, with focus
on Old English 169
Olga Timofeeva
Markers of Futurity in Old High German andOld English: A Comparative
Corpus-Based Study 195
Gabriele Diewald & Ilse Wischer
The Verb to be in the West Saxon Gospels andthe Lindisfarne Gospels 217
Christine Bolze
Aspectual properties of the verbal prefix a- in Old English
with reference to Gothic 235
Vlatko Broz
Comparative Studies in Early Germanic Languages
r ws vs. thr was: Old English and Old High German existential
constructions with adverbs of place 263
Simone E. Pfenninger
On gain and loss of verbal categories inlanguage contact:
Old English vs. Old High German 289
Theo Vennemann
Index 313
Introduction

Leibniz Universitt Hannover, Germany / University of Helsinki, Finland /
University of Potsdam, Germany
1. V
erbal categories and their diachronic development in Old English
andOld High German
This publication comprises the papers presented at a workshop on the Contras-

tive study of the verbal categories and their grammaticalisation in Old English
and Old High German held at the 16th ICEHL in Pcs, Hungary, in August 2010.
To provide a wider view on historical comparative corpus analyses in the field of
verbal categories in Early Germanic and on the interrelation between these catego-
ries in their development, the volume is expanded by some additional papers on
related topics. Although our focus is on the early stages of English and German,
other Germanic languages are also examined.
English and German are closely related. Even after their separation in the first
centuries AD they have had a long history of contact and interrelated develop-
ment. Nevertheless they have moved in different directions from a typological per-
spective. While there are certainly similarities in their increase in analytic forms,
there are even more differences in the development of the structure of grammati-
cal categories (particularly the verb).
Thus, while English developed an aspectual system, German did not. Instead
German kept and even refined a complicated system of mood marking, whereas in
English the grammatical category of mood has been lost. Even the two categories
that in both languages showed a trend in the same direction (i.e. tense and voice)
contain fundamental disparities as far as the internal structuring of the paradigms
and their values are concerned.
In order to indicate the scope of topics brought together in this volume, some
introductory notes on the verbal categories and their members in early Germanic
are useful. Early Germanic verbs were inflected for person, number, mood and
tense. The tense system was reduced to present and past compared to six or seven
tenses in Greek and Latin. The two languages under closer examination here, Old
English (OE) and Old High German (OHG), had an unmarked non-past tense
form and a past tense marked by a dental suffix or by vowel gradation respectively.
Additionally, forms of have or be could be combined with the past participle of a

verb to form perfect-like structures, although their status as periphrastic may be
rather called into question. Reference to future events was made by the indica-
tive or subjunctive in the present tense in combination with temporal adverbials
or other contextual clues. Additionally, in OE there existed particular uses of the
be/become-copula in constructions with future reference besides combinations
of (pre)modal verbs and infinitive to express futures with modal colouring. Thus,
the early Germanic languages, having been drastically reduced in their number of
tenses, seem to have used their resources in various similar but not identical ways
to create new grammatical means of expressing temporal distinctions.
A grammatical category of aspect did not exist in the early Germanic lan-
guages, although residues of aspectual markers may be attested. They become
obvious in the use of prefixed verb forms. In example (1) geworhte made is used
in a perfective sense, while the same verb without a prefix in example (2) denotes
an imperfective meaning.
(1) foram on VI dagum Crist geworhte heofonas & eoran,

because Christ made heaven and earth in VI days,
(ALFREDS INTRODUCTION TO LAWS, HC,1 p. 26f.)
& st r a hwile e mon worhte a burg t Tofeceastre mid stanwealle,

(2)
& was sitting there while they were building the fortress at Toucester with a
stone wall, (CHRONICLE MS A EARLY (O2), HC, p. (102)
A closer comparative look at the use of prefixed verbs with regard to their aspec-
tual function in OE and OHG is presented in Wischer and Habermann (2004).
Focusing particularly on ge-/gi-verbs in the OE Orosius and the OHG Tatian, the
authors come to the conclusion that the diverging path of development with
regard to aspect seems to begin already in these early stages (262). Although an
aspectual distinction can also be attested in OHG (cf. example (3)) the prefix gi- is
not used as frequently and systematically as ge- is in OE.
quando mortui audient uocem filij dei. & qui audierint uiuent. thanne thie
(3)
toton horent stemma gotes sunes inti thie sia gihorent lebent.
Then the dead (will) hear the voice of Gods son, and those who will have
heard it (will) live. (Tatian2 137, 2628)
. HC = Helsinki Corpus.
. St. Gallen Cod. 56, Masser, ed. (1994, 265389), cf. also Wischer & Habermann (2004:277).
Introduction
A similar aspectual function compared to that of ge-/gi- can be signified by the OE

verbal prefix a-, which is the topic of Broz article in this volume. Broz expands the
study to cognate forms in Gothic and even includes a comparison with Croatian
verbs, which are morphologically marked for aspect.
Besides the existence of aspectual residues in the form of verbal prefixes there
have emerged analytical constructions with have or be and past participle in OE
and OHG, which can also have a perfective sense, as in example (4).
(4) onne u as word gecweden hbbe genim one ppel

when you have spoken those words, take the apple
(QUADRUPEDIBUS, HC, p. 9)
Again, differences in frequency and use have been observed between OE and
OHG. In OE, such periphrastic constructions appear to be more frequent than in
OHG. At least in Tatian they are extremely random (cf. Wischer & Habermann
2004). In OE, habban had even developed an additional causative use, which is the
topic of Kilpis paper in this volume.
The category of mood, comprising inflectional distinctions between indica-
tive, subjunctive and imperative, interacts with notions of modality and temporal-
ity. For OHG, the functional-semantic field of modality with regard to its linguistic
expressions has recently been analyzed on the basis of Bible texts in a dissertation
by Monika Schnherr (2010). Several papers in this volume deal in depth with
specific problems relating to modality, futurity, and evidentiality in OE and/or
OHG, often from a comparative perspective.
The shift from Proto-Indo-European active structure to Germanic transitive
structure brought about the development of a new active/passive voice system.
Modern English and Modern German differ not only in the use of the auxiliary in
passive constructions, but also in their syntactic patterns. While in English even
indirect objects can become the subject of passive sentences, which is not pos-
sible in German, German, on the other hand, possesses an impersonal passive (Es
wird getanzt) and the so-called dative passive (Sie bekamen die Betriebserlaubnis
entzogen), both of which do not exist in English. The reasons for such a diverging
development may be discovered in a historical comparative analysis. This is what
Petr in the present volume attempts at with regard to the different auxiliaries used
in English and German.
These differences are all the more remarkable as the source items for each
involved construction as well as the general mechanisms of grammaticalisation
are very similar. It is assumed that these differences in the pathways and direction
of grammaticalisation can be put down to different language contacts to a substan-
tial degree, but also to differences in the original situation in the oldest attestable
periods of each language.
The papers in this volume aim to discover the early traces of the distinct devel-
opments of both languages by contrastive analyses of central verbal categories in
several early Germanic languages, with a focus on OE and OHG. Most papers are
based on empirical evidence making use of various corpus data. However, theo-
retical and philological reflections are included as well.
As the theoretical background of grammaticalisation studies on one hand,
and the empirical operational questions concerning diachronic data and their pro-
cessing on the other, are central to all papers in this volume, some introductory
remarks on both topics seem useful before dealing in more detail with the verbal
categories investigated in the contributions.
2. G
rammaticalisation, comparative diachronic linguistics
and socio-cultural/philological aspects
As all contributions of the volume are concerned with the diachronic develop-
ment, the reshaping or the rise of grammatical functions in particular linguistic
items that had no or less grammatical functions before, the red thread that unites
all investigations is the framework of grammaticalisation studies. Given that the
focus is on the evolution of verbal categories, the authors follow Brinton (1988),
Bybee et al. (1994), Krug (2000), Heine and Kuteva (2002), Aikhenvald (2004),
Diewald (2006), just to mention some of the representative works in this field. In
the centre of interest is the evolution of particular verbal categories, structures or
constructions, viewed from a comparative perspective of related Germanic lan-
guages. There are several issues that have received much attention in grammati-
calisation theory in recent years and that also constitute a major topic in several of
the papers united here. These are issues like the following ones:
contrastive investigations of different lexical sources (e.g. Petr on the ori-

gin of different auxiliary uses in English and German passive constructions;
Diewald and Wischer on source lexemes for future grams in OE and OHG)
polygrammaticalisation (e.g. Mailhammer and Smirnova arguing that the
passive is only one of several possible readings for copula constructions com-
bining be/become-verbs with past participles in Old English and Old High
German; Kilpi discussing the conditions and stages for causative and perfect
have-constructions in Old English)
constructions in grammaticalisation (e.g. Cloutier with regard to haitan-
constructions in Gothic and Old English; Kilpi on the OE causative have-
construction; Jger analysing the grammaticalisation path of the modal
haben+ zu-infinitive-construction in Old High German; Timofeeva g iving
Introduction
a survey on auditory evidential constructions in Old English and other

Old Germanic languages; Pfenninger studying OE and OHG existential
constructions)
the fuzzy edges between grammatical categories and the intercategoriality of
various linguistic forms and constructions (e.g. Bolze examining future and
habitual functions of the verb to be in two OE dialects)
interrupted processes of grammaticalisation (e.g. Eitelmann on the rise and
decline of OE (ge)munan as a potential modal; Broz presenting a study on
the aspectual function of the verbal prefix a- in OE and its cognate in Gothic)
the influence of sociolinguistic and/or regional factors or language contact
situations on particular developments (e.g. Eitelmann taking into account a
combination of language-internal and cultural change (change from an oral
to a literal society), as well as the impact of language contact (Old Norse) in
the rise and decline of gemunan as a modal verb in Old English; Vennemann
relating several of the most important structural changes and categorial dif-
ferences in the verb systems of Proto-Germanic, OE and OHG to the different
contact histories of these languages).
This means that in addition to grammaticalisation theory and its typological foun-
dations, ample use is made of the results of diachronic comparative and philologi-
cal studies and their theoretical and methodological prerequisites.
Due to these broad theoretical foundations that are shared by all authors the
individual results concerning particular categories or constructions tie in with
each other closely to give a coherent picture of the factors, stages and results of
grammaticalising language change in the area of verbal categories in English and
German.
Furthermore, the studies are mostly based on detailed corpus investigation
in order to enable the authors to identify similarities and differences in linguistic
contexts and then ultimately derive some understanding of similar or diverging
developments. The use of and access to empirical data in historical comparative
studies is, however, still a particular challenge. Thus, the next section is devoted
to an evaluation of the conditions and place of corpus studies in diachronic gram-
maticalisation research.
3. Historical comparative corpus studies
Historical study of language has become considerably more approachable during

the past few decades. Kyt (2012:1) claims that [i]t is probably not an exaggera-
tion to say that corpus linguistics is a methodology that enjoys an ever increasing
popularity world-wide today. Electronic corpora have turned out to be invaluable

tools in searching for linguistic items in large amounts of text, quickly and without
too much manual intervention. They have also provided us with new methodolo-
gies in analysing linguistic data, both qualitatively and quantitatively. Diachronic
change and synchronic variation can be approached from many angles, so that it
becomes possible to even pinpoint important moments of language change and
analyse more easily than ever before synchronic variation across social strata,
regions, text types and genres (Tyrkk et al. 2012).
Corpora are far from being online text archives today. The majority of
present-day corpora are still one-language databases, but there seems to be more
and more need for comparative corpora consisting of data from different languages
and obviously from different periods of time. With regard to the field of corpus
development, Stig Johansson pointed out on several occasions that [w]e need big-
ger corpora, better corpora, corpora with a wider range of languages, and we need
to learn to exploit the corpora in the best possible manner (quoted in Oksefell et al.
2012). Historical corpus studies still lack electronic databases of the earliest phases
of Germanic languages, but work is being done in that area too.
Interdisciplinary corpus work is nothing new today: Moreover, it is not only
linguists that find the approach increasingly attractive: corpus linguistic meth-
odology and language analyses are nowadays applied to fields beyond linguistics
proper. Professionals profiting from techniques developed in corpus linguistics
include historians, experts in law, literary critics, computer scientists and language
teachers (Kyt 2012:1).
Corpus annotation creates a particular challenge to corpus compilers at the
moment, but new methods for parsing historical texts, as well as for statistical
analysis are being developed today. Archer (2012) presents opinions between too
little and too much in terms of annotation, but comes to the conclusion that the
primary solution depends on the purpose that annotation is used for. She also calls
for a combination of manual and automatic techniques, as well as reliable retrieval
tools. Only when all these aspects are taken into account we will achieve reliable
results. The volume and diversity of digitized material available for linguists today
also creates problems, as for comparative study in particular, it would be ideal to
have data that are representative enough and comparable with each other.
Several studies in this volume make use of historical corpora. The majority of
these represent the English language and are databases of texts from different peri-
ods in the history of English, but annotated corpora have also been utilised. As the
constitution of English diachronic corpora is more advanced than that for other
Germanic languages including German the following description of available
resources is confined to the English language. For many of these corpora, basic
information can be retrieved from e.g. the Corpus Resource Database (CoRD) as
well as the ICAME Corpus Collection.
Introduction
The opus magnum within OE studies, the Dictionary of Old English Corpus
(DOEC) (Bolze, Kilpi, Timofeeva), is a complete electronic record of surviving
Old English except for some variant manuscripts of individual texts. It comprises
at least one copy of each text written in Old English, and sometimes multiple cop-
ies if of interest for dialect, date, etc. All in all there are 3,060 texts, more than 3
million running words of Old English and ca. 1 million running words of Latin.
The body of surviving OE texts encompasses a rich diversity of records, prose,
poetry, glosses to Latin texts and inscriptions. The online Dictionary of Old English
project makes use of the Corpus compiled for this purpose, and the dictionary is
currently available for letters AG.
Two major electronic sources, partly still in progress, have turned out to be
indispensable for any study on Middle English (ME), also witnessed by studies in
this volume (Eitelmann, Timofeeva). The electronic version of the Middle English
Dictionary preserves all the details of the print MED (completed in 2001), which
is based on a collection of over three million citation slips containing English
of 11001500, but goes far beyond this, by converting its contents into an enor-
mous database, searchable in ways impossible within any print dictionary. The
collection of ME texts in the Corpus of Middle English Prose and Verse, still in
progress, contains some 60 texts assembled by the Humanities Text Initiative
with the intention to develop the corpus into an even more extensive and reli-
able collection of ME electronic texts for wide use with the help of various search
mechanisms.
A well-known fact is that the (online) Oxford English Dictionary (OED) has
been compiled on historical principles and thus serves a great variety of interests
among English historical linguists. Eitelmann and Timofeeva have made extensive
use of its resources for their comparative studies in the present volume. Another
huge electronic resource is the Dictionary of the Scots Language, comprising two
old authorities within the study of Scots, namely the Dictionary of the Older Scot-
tish Tongue (Eitelmann) and the Scottish National Dictionary. Online dictionaries
have become an indispensable tool for historical linguists.
The Helsinki Corpus of English Texts (HC) is a structured multi-genre dia-
chronic corpus, which includes periodically organized text samples from Old,
Middle and Early Modern English. Each sample is preceded by a list of parame-
ter codes giving information on the text and its author. The Corpus is useful par-
ticularly in the study of the change of linguistic features in long diachrony. It can
be used as a diagnostic corpus giving general information of the occurrence of
forms, structures and lexemes in different periods of English. This information
can be supplemented by evidence yielded by more special and focused historical
corpora. (CoRD) The HC is relatively small in size, only ca. 1.5 million words,
but has indeed been supplemented with several corpora compiled by Helsinki
scholars. It has been followed by several annotated corpora, starting with the
York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE).This syntacti-

cally annotated corpus of OE prose is based on the Toronto DOEC and contains
all the major OE prose works. Each word is tagged for part of speech, and the
corpus can be searched automatically for syntactic structure, constituent order
and lexical items. Corpora of the Helsinki family have been utilised in this
volume (Broz, Cloutier, Timofeeva, Wischer and Diewald).
For a historical study of other Germanic languages there is a considerably
more limited number of corpora available, but there is a clear trend to supplement
the resources. In the present volume such online text databases as offered by the
Heimskringla Project, Project Wulfila and TITUS have been made use of (Cloutier,
Pfenninger, Timofeeva). The first one includes some 4,000 Old Norse titles (some
of them in Modern Scandinavian languages), primarily the Eddas, the Icelandic
sagas and skaldic poetry. Project Wulfila is a small digital library dedicated to the
study of the Gothic language and Old Germanic languages in general, and so far
it includes the Gothic Bible and minor fragments, but further expansion is being
worked on. TITUS=Thesaurus Indogermanischer Text- und Sprachmaterialien con-
sists of an ongoing project that contains even now a variety of early Germanic texts
and text fragments (Gothic, Nordic, Old English, Old Frisian, Old Saxon, Middle
Low German, Old Dutch, Old High German, etc.).
A synopsis of presently available corpora of diachronic stages of German,
including work in progress such as the large project DDD (Deutsch Diachron
Digital), is available online via http://texte.mediaevum.de/textkorpora.htm.
Beyond the resources referred to in that synopsis, there exists a number of small
collections for highly specific purposes which can be accessed online. An exam-
ple is the kali-korpus (cf. http://www.kali.uni-hannover.de), which has been
compiled for selective needs of research and academic teaching. It contains com-
plete glosses of selected texts and is annotated for verbal categories (for details
cf. Diewald, Lehmberg & Smirnova 2007). In the present volume, it is used as a
database in the contributions by Diewald and Wischer as well as by Mailhammer
and Smirnova.
In the past few years, several projects have been launched with the aim of
providing online databases containing material for comparative linguistic study.
Two of them shall be mentioned in this context, even though they do not form a
direct link to any of the studies of this volume: ISWOC (Information Structure and
Word Order Change in Germanic and Romance languages) aims to map develop-
ments of information structure and word order in early Germanic and Romance
languages and their modern counterparts; PROIEL (Pragmatic Resources in Old
Indo-European Languages) aims to describe and account for the so-called prag-
matic resources of these languages, and focuses e.g. on word order or the use of
participles to refer to background events.
Introduction
4. The verbal categories studied in this volume
As has become obvious from Section 1, the verbal categories underwent major
restructuring in the early Germanic languages. The studies collected in this vol-
ume try to detect these diverging developments looking at the linguistic con-
texts and sociolinguistic situations. Many of them take a comparative perspective,
often contrasting English and German with Latin, but also comparing different
Germanic languages. Moreover, the historical perspective, i.e. the comparison of
the categorial systems in different diachronic stages of a language, are addressed
in all papers, and emphasized in several of them. Table 1 shows a survey of the
verbal categories that are studied in the individual papers of this volume:
Table 1. Verbal categories and languages studied in this volume

Categories Languages Authors
Passive Voice OE/Gothic Cloutier

Passive Voice OE/OHG Mailhammer & Smirnova
Passive Voice OE ME/OHG MHG Petr
Passive Voice/Causative OE Kilpi
Modality OE ME EModE Eitelmann
Modality OHG Jger
Evidentiality OE/Gothic Timofeeva
Future Tense OE/OHG Diewald & Wischer
Future Tense OE Bolze
Aspect OE/Gothic ModE, Croatian Broz
Existential Constructions OE/OHG Pfenninger
Verbal Categories OE/OHG Vennemann
Though all verbal categories relevant in English and German are dealt with
in several papers in the volume, there are some focal points where the research
interests of the studies converge and complement each other. Thus, three papers
are concerned with passives (Cloutier, Mailhammer & Smirnova and Petr),
another one is related to passives, too, while focusing on causative constructions
with have-verbs (Kilpi), three papers are concerned with modality and evidenti-
ality (Eitelmann, Jger, and Timofeeva), and two papers investigate future mark-
ers (Bolze, Diewald & Wischer). Further topics treated are aspectual functions in
derivational verbal morphology (Broz), the question of the grammaticalisation of
constructions with existential meaning (Pfenninger), and the overall development
of the verbal grammatical systems in both German and English (Vennemann).
Typically, linguistic items (the verbal source lexemes) undergoing grammati-

calisation show affinities to several verbal categories, and their development is
characterized by phases of polysemic divergences towards distinct target cate-
gories. For example, modal verbs (more exactly their pre-modal ancestors) are
possible and often used sources for future markers as well as for epistemic and
evidential markers. Verbs meaning have may develop into causatives, perfectives
etc. Thus, beyond the categories each contribution focuses on, there are several
cross-cuttings and interrelations among the papers as to the variety of choices
individual markers and their cognates in other languages have (e.g. have-verbs
and their grammaticalisation potential treated as in Kilpi and Jger in very dif-
ferent functions).
Thus, the array of contributions presented here gives a coherent (though by
no means complete) overview of central aspects of the development of verbal
categories in the early stages of German and English, with substantial outlooks
to other languages. The commitment to a grammaticalisation approach, which is
found in every paper, furthermore ensures a comparable line of argument, uniting
empirical-descriptive thoroughness with a common theoretical framework that
is explicit enough to make findings comparable, and open enough to allow new
aspects of, e.g. historical sociolinguistics and other fields to be integrated.
The following sections present a brief summary of the content, results and
theoretical intentions of the papers united in this volume.
The first four papers in this volume address issues related to the passive voice.
Robert Cloutier in his article *haitan in Gothic and Old English, presents a
comparative study of the uses of haitan and htan in Gothic and Old English,
the only verb that retained a synthetic passive inflection in these languages. On
the basis of an analysis of data collected from the Helsinki Corpus (for Old Eng-
lish) and from Gothic Wulfila the author aims to shed light on the early functions
of haitan and htan, and from there to analyze their further development. The
Gothic data are compared to the Greek original. Cloutier presents a detailed study
of the distribution of this verb in various constructions in the two corpora, tak-
ing into account the semantics of its auxiliary-like functions, its lexical vs. copula
functions and the syntactic constructions in which it occurred.
The category of passive is also the topic in Robert Mailhammer & Elena
Smirnovas paper on Incipient Grammaticalisation: sources of passive construc-
tions in Old High German and Old English. The authors present a contrastive
study of copula constructions combining verbs denoting be and become with
past participles, which are the sources of later passives. It is shown that in the old-
est stages the passive interpretation was only one of several possible readings. All
their triggering features are listed, analysed and compared, providing the basis of
Introduction
a thorough investigation into the different courses of the development of the pas-
sives in English and German.
The contribution by Peter Petr (Passive Auxiliaries in English and G erman:
Decline versus grammaticalisation of bounded language use) relates the diverg-
ing paths in the grammaticalisation of passive auxiliaries in English and German
(be- versus become-verb) to a typological distinction at the level of macrostruc-
tural planning based on bounded (German) versus unbounded (English) sys-
tems (cf.Carroll & Lambert 2003; Rodeutscher & von Stutterheim 2005). It is
argued that English weoran disappeared in passive constructions together with
the bounded language use in English. Supported by corpus data, Petrs study
shows how different categories or structures in a language may interact and direct
language change in various ways. At the same time it exemplifies how different
subdisciplines of linguistics can benefit from each other, as e.g. psycholinguis-
tic findings (bounded versus unbounded language use) and grammaticalisation
theory.
Matti Kilpis article, Causative habban in Old English: Tracing the Devel-
opment of a Budding Construction is concerned with a construction in OE cor-
responding to the Present-Day English (PDE) type I had my shoes repaired. The
author draws his data from a complete analysis of the whole habban material of
the Dictionary of Old English Corpus (DOEC). It turns out that causative habban is
very rare in Old English and that all constructions of that kind contain the seman-
tic features of deontic or volitional modality. It is argued that this fact plays a major
role in the rise of the causative habban construction. Kilpis study does not only
present a detailed analysis of the exact contexts and factors of the rise of causative
habban using central concepts of grammaticalisation theory, it also illustrates how
different verbal categories may interact with each other, as in this case passive
voice, causativity and modality.
One of the papers addressing the topic of modality is Matthias Eitelmanns
Remembering (ge)munan The Rise and Decline of a Potential Modal. Dealing
with modal meanings and modal verbs, this paper focuses on those preterite-
presents that did not evolve further to modal functions but got lost. These verbs
in contrast to the surviving ones have not received very much attention in
diachronic linguistics as well as in grammaticalisation studies; this paper takes
care of one of them and thus closes a gap in diachronic coverage and provides an
extended perspective on processes of grammaticalisation. It takes into account a
combination of language-internal (semantic change, auxiliarisation, grammati-
calisation) and cultural change (change from an oral to a literal s ociety), as well
as the impact of language contact (acceleration of change through contact with
Old Norse).
A different aspect of modality is approached by Anne Jger in her paper

on The emergence of modal meanings from haben with zu-infinitives in Old
High German. Focusing on Notkers writings and a comparison with Latin,
she presents a classical grammaticalisation study dealing with the path from
POSSESSION to OBLIGATION in the development of haben & zu-infinitive to a
marker of modality in the OHG period. The background and previous research
referred to are naturally grammaticalisation studies, in particular Heine and
Kuteva (2002), and Haspelmath (1989) for the grammaticalisation channels of
infinitives, as well as research into Old High German and Old English. The author
presents detailed argumentation for the originality of Notkers constructions as
compared to the Latin texts on the basis of a close-up comparison of Old High
German versus Latin. Thus, Jgers study complements the discussion on the
range of grammaticalisation paths possibly entered by have-verbs as outlined by
Kilpi in this volume.
Another category closely related to and subsumable under the broad domain
of modality and evidentiality, is the topic of the contribution by Olga Timofeeva
(Hearsay and lexical evidentials in Old Germanic languages, with focus on Old
English). It concentrates on perception verbs and verbs with a say-component,
which are known to grammaticalise into morphological evidentials. Like in the
previous article by Jger, not only individual lexical items but whole constructions
are under discussion. The qualitative case study of the oldest stages up to Middle
English comes to the conclusion that OE and other Germanic languages have
very similar patterns of marking direct auditory perception and hearsay evidence,
which must have their origin in Proto-Germanic.
Modality is not only closely linked with evidentiality, but also with futurity.
The interrelation of futurity and modality is part of Gabriele Diewald & Ilse
Wischers contribution on Markers of Futurity in Old High German and Old
English: A Comparative Corpus-Based Study. It analyses those constructions in
OHG and OE that had the potential of marking future events, i.e. modal con-
structions and those with be/become-verbs. The authors rely on the framework
of grammaticalisation, uniting grammaticalisation studies on future markers
(e.g. Bybee et al. 1994) with diachronic studies on the development of futures in
German and English. The detailed analysis of syntactic and semantic as well as dis-
tributional properties of the different constructions reveals remarkable differences
with respect to the relevant items even in their earliest attested stages, the reasons
for which may be socio-linguistically conditioned. The approach allows theoreti-
cal generalizations as to the combination of language-internal factors, universal
grammaticalisation paths, and sociolinguistic conditions.
The verbal category of futurity is also the topic of Christine Bolzes paper on
The verb to be in the West Saxon Gospels and the Lindisfarne Gospels. Old E nglish
Introduction
is the only Germanic language which contains a double paradigm of the verb to be
(b-forms/s-forms) in the present tense, whereby the b-forms are used with a future
sense or to express habitual events. Bolze analyses the use of b- and s-forms in two
OE dialects, the West Saxon and the Northumbrian one, comparing them to the
Latin original, thus providing further insights into their functions in Old English
and their dialectal distribution. It closes a gap in English historical dialectology
since none of the previous studies has hitherto focused on Northumbrian dialect,
relying on such a detailed corpus-based analysis. The paper furthermore under-
lines the importance of an integrated approach to the study of verbal categories in
terms of their mutual interaction, as we are dealing here with an interrelation of
modality, futurity and habitual aspect.
While habituals can be considered to be related to imperfective aspect, there
were other linguistic means, such as verbal prefixes (ge-, for-, a-, and others), in
Old English that had a close affinity to perfective aspect. One of these prefixes
is in the focus of interest in Vlatko Brozs paper on Aspectual properties of the
verbal prefix a- in Old English with reference to Gothic. It is shown that this
item had a range of meanings in Old English, among them the expression of
perfective aspect. The paper contrasts Old English with Gothic, and in the mod-
ern stages it compares the equivalents in Modern English and Croatian. The
paper presents a first detailed study of a poorly investigated erstwhile perfective
marker; it describes the development from a grammaticalised marker to a fossil-
ised morpheme, and taking into account that English developed a different way
of expressing aspectual distinctions the renovation of the grammatical category
of aspect in English.
Simone E. Pfenningers article r ws vs. thr was: Old English and Old
High German existential constructions with adverbs of place, draws our attention
to existential constructions and their degree of grammaticalisation in Old English
and Old High German. The paper shows that notwithstanding the differences in
detail both languages in their modern stages have existential constructions with
locative adverbs that have arisen via grammaticalisation. It is assumed that both,
OE expletive r- and OHG thr-constructions, had a common origin and com-
mon general traits in their development. The study takes up a still neglected area
of diachronic research and presents a wealth of qualitative and quantitative data
on OE and OHG existential constructions. Drawing on grammaticalisation stud-
ies and constructional approaches, the paper puts forward hypotheses on why the
two languages developed differently with English acquiring one highly grammati-
calised existential construction while German did not, but developed a variety of
competing constructions for existential sentences.
The last contribution, which relates several of the most important struc-
tural changes and categorial differences in the verbal systems of Old English and
Old High German to the different contact histories of these languages, is Theo
Vennemanns survey On gain and loss of verbal categories in language contact:
Old English vs. Old High German. After giving an overview of the shared inher-
ited categories, the author discusses the shared innovated categories in English and
German, namely periphrastic perfect, future and passive voice. He argues that the
grammaticalisation of these categories is well underway in Old English and Old
High German, although the new categories had by no means been fully integrated
into the verbal systems yet. The main part of the paper, however, is devoted to the
divergent developments in English, focussing at some length on the two copulas
in Old English, which are also under closer inspection in Bolzes paper in this vol-
ume. Vennemann concludes that most of the differential innovated categories in
Old English as compared to Old High German are due to Celtic influence.
5. Summary and outlook
All papers in this volume gain their new insights from a comparative study of
language data in related languages or dialects. Generalising the results of the indi-
vidual studies concerning the verbal categories in English and German, in this con-
cluding section an attempt will be made to widen the perspective and formulate
some general principles for studying language change in a comprehensive way.
As language change is a highly complex process, it can only be tackled by
combining several perspectives concerning the data, the theoretical approaches,
and the methodologies.
1. Combining data. The comparanda, i.e. the linguistic entities compared, may
be related typologically, areally or genealogically. Typologically oriented com-
parison aims at identifying universal features, cross-linguistic tendencies,
and regularities of change. As a consequence, closely related languages in this
approach are often treated as mutually exchangeable in cross-linguistic sam-
ples so that divergencies of closely related languages tend to get obliterated.
However, the categories supplied by typological studies provide a suitable
instrument for investigating historical changes and comparing the respective
outcomes in any pair of languages, independent of genealogical or areal dis-
tance. The comparison of closely related languages, on the other hand, brings
to light the differences missed in large-scale typological comparisons, and
allows the search for language specific pathways of change.
2. Combining theoretical approaches. As language change results from multifac-
eted situations comprising linguistic and extra-linguistic factors, its investiga-
tion has to take into account theoretical models of every structural layer and
their interaction as well as of language usage and its socio-historical context.
Introduction
In this volume, this is achieved by a combination of grammaticalisation the-

ory, constructional approaches, philological as well as sociolinguistic studies.
Applying this kind of empirically and theoretically founded approach makes
it possible to account for language change in its complexities and thus pro-
vides a solid basis even for more formally oriented reflections on linguistic
structure.
3. Combining methodologies. In the last decades the rise of corpus linguistics in
diachronic investigation has proved its merits without however making close-
up qualitative and philologically informed studies dispensable. In particular,
sociolinguistic research has proved vital for the investigation of language
contact and its enormous influence on language change. Thus, methods from
sociology and natural sciences must complement the tool box for an integra-
tive investigation of language change.
In short, there is still much to be done. For one thing, the sufficiently complete
parallel and combined history of the grammatical development of the languages
dealt with here has yet to be written; for another, if such a combined linguistic
history can be supplied, it may provide valuable new insights for other disciplines
concerned with social, historical and cultural development. The present volume is
meant to be a step in this direction.
References
Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: OUP.

Archer, Dawn. 2012. Corpus annotation: A welcome addition or an interpretation too far? In
Jukka Tyrkk et al. (eds).
Brinton, Laurel J. 1988. The Development of English Aspectual Systems. Aspectualizers and post-
verbal particles [Cambridge Studies in Linguistics 49]. Cambridge: CUP.
Bybee, Joan, Perkins, Revere & Pagliuca, William. 1994. The Evolution of Grammar. Tense, Aspect
and Modality in the Languages of the World. Chicago IL: The University of Chicago Press.
Carroll, Mary & Lambert, Monique. 2003. Information Structure in narratives and the role of
grammaticised knowledge: A study of adult French and German learners of English. In
Information Structure and the Dynamics of Language Acquisition [Studies in Bilingualism
26], Christine Dimroth & Marianne Starren (eds), 267287. Amsterdam: John Benjamins.
CoRD = Corpus Resource Database. http://www.helsinki.fi/varieng/CoRD/
Diewald, Gabriele. 2006. Context types in grammaticalization as constructions. In Construc-
tions. Special Volume 1: Constructions all over case studies and theoretical implications,
Doris Schnefeld (ed.). http://www.constructions-online.de/articles/specvol1/
Diewald, Gabriele, Lehmberg, Timm & Smirnova, Elena. 2007. KALI A diachronic corpus
for the investigation of grammaticalization and semantic change. In Data Structures for
Linguistic Resources and Applications. Proceedings of the Biennial GLDV Conference 2007,
Georg Rehm, Andreas Witt & Lothar Lemnitzer (eds), 103112. Tbingen: Narr.
Haspelmath, Martin. 1989. From purposive to infinitive: A universal path of grammaticalisa-

tion. Folia Linguistica Historica 10: 287310.
Heine, Bernd & Kuteva, Tania. 2002. World Lexicon of Grammaticalisation. Cambridge: CUP.
ICAME Corpus Collection. http://nora.hd.uib.no/corpora.html
ISWOC (Information Structure and Word Order Change in Germanic and Romance languages).
http://www.hf.uio.no/ilos/english/research/projects/iswoc/index.html
Krug, Manfred. 2000. Emerging English Modals. A Corpus-Based Study of Grammaticalisation.
Berlin: Mouton de Gruyter.
Kyt, Merja (ed.). 2012. English Corpus Linguistics: Crossing Paths [Language and Computers:
Studies in Practical Linguistics 76]. Amsterdam: Rodopi.
Oksefell Ebeling, Signe, Ebeling, Jarle & Hasselgrd, Hilde (eds). 2012. Aspects of Corpus Lin-
guistics: Compilation, Annotation, Analysis [Studies in Variation, Contacts and Change in
English 12]. http://www.helsinki.fi/varieng/journal/volumes/12/introduction.html
PROIEL (Pragmatic Resources in Old Indo-European Languages). http://www.hf.uio.no/ilos/
english/research/projects/proiel/index.html
Rodeutscher, Antje & Stutterheim, Christiane von. 2005. Semantische und pragmatische
Prinzipien bei der Positionierung von dann. Linguistische Berichte 205: 2960.
Schnherr, Monika. 2010. Modalitt und Modalittsausdrcke in althochdeutschen Bibeltex-
ten. Eine korpusgesttzte Analyse [Wrzburger elektronischen sprachwissenschaftlichen
ArbeitenNr.7].http://opus.bibliothek.uni-wuerzburg.de/volltexte/2010/4690/pdf/
MonikaSchoenherrDissWespa7.pdf
Tyrkk, Jukka, Kilpi, Matti, Nevalainen, Terttu & Rissanen, Matti (eds). 2012. Outposts of His-
torical Corpus Linguistics: From the Helsinki Corpus to a Proliferation of Resources [Studies
in Variation, Contacts and Change in English 10]. http://www.helsinki.fi/varieng/journal/
volumes/10/introduction.html
Wischer, Ilse & Habermann, Mechthild. 2004. Der Gebrauch von Prfixverben zum Ausdruck
von Aspekt/Aktionsart im Altenglischen und Althochdeutschen. Zeitschrift fr Germanis-
tische Linguistik 32: 262285.
*haitan in Gothic and Old English*
Robert A. Cloutier
University of Amsterdam, Netherlands
By collecting data from various corpora, I examine and compare the use of the
Gothic and Old English reflexes of *haitan, a transitive verb that survives as a
copula-like verb in the modern Germanic languages. Between the two languages,
this verb can occur in five constructions: calling, transitive naming, infinitival
commanding, subclause commanding, and copular naming. Both Gothic and
Early Old English share the use of this verb in calling constructions whereas
the subclause commanding construction is an Old English innovation and the
copular naming construction does not appear until Late Old English. Regardless
of the language or period, however, when *haitan occurs in transitive naming
constructions, it strongly favours passive voice, which may explain its later use in
copular naming constructions. Moreover, an examination of the competitors of
Gothic haitan shows that it has strong competition from various verbs in each of its
functions, though the competition in the transitive naming construction is weakest.
1. Introduction
The descendants of the proto-Germanic verb *haitan have a rather peculiar syn-
tax, behaving in the various modern Germanic languages (with the exception of
English where it does not survive) syntactically like a copula in that they connect
a subject to a complement, as seen in example (1), and thereby letting the listener
know that the complement is an attribute of the subject. In a historical context, we
should note that the modern descendants of this verb have this property without
any overt marking of passive voice their inflections are simply that of the active
voice. This contrasts with the use of this verb in the older stages of the Germanic
languages, which will be discussed later.
(1) I am-called Robert.
a. Ik heet Robert Dutch
b. Ich heisse Robert German
c. g heitir Rbert Icelandic
* I use *haitan to represent the Proto-Germanic verb as well as the pan-Germanic lexeme.
This is to make a clear distinction between this form and the Gothic verb haitan.
Robert A. Cloutier
Because of this peculiar property, these verbs have been variously analysed in, for
instance, Modern Dutch as a copula (Haeseryn et al. 1997), an intransitive verb1
(Den Boon & Geeraerts 2008), and a raising verb2 (Matushansky 2008). These
modern forms of *haitan differ from most other copulas, however, in that they
encode something more specific: the attribution, for the most part, is not just any
attribute of the subject but specifically a name, hence the translation of this verb
into English as to be called, to be named. The ancestor of this verb in Gothic did
not have this function without overt marking of passive voice, which could have
been applied to any transitive verb, and the shift from a purely transitive verb to
the modern descendant has not received any attention in the literature.
Moreover, the etymology of this verb does not help us figure out the pos-
sible trajectory of its development because its etymology is not clearly established
(Mailhammer 2007). Traditionally, *haitan has been analysed as an ablauted
form of proto-Indo-European *keih2- meaning to move with the addition of a
dental suffix (Lotspeich 1933; Pokorny 1951; Seebold 1970; Green 1985; Kluge
2002). However, scholars have pointed out a number of problems with this etymo-
logy. There are no direct cognates of this verb in any of the other Indo-European
languages (Phillippa et al. 2009). The meanings associated with the reflexes of
this root in the other older Indo-European languages do not correspond well to
those meanings associated with *haitan (De Vries 1962). The semantic devel-
opment from the proto-Indo-European root meaning to move to *haitan to
name, to command is problematic (De Vries 1962; Boutkan & Siebinga 2005).
And the proposed dental suffix reconstructed for this etymology would be isolated
to Germanic and absent in the other branches of Indo-European (Seebold 1970;
. Den Boon and Geeraertss (2008) analysis is that the modern use of heten is as an intransi-
tive verb. Unlike the entries for zijn to be or blijken to seem, they do not claim that heten can
function as a copula.
. A raising verb has the property of allowing an argument that belongs semantically to a
subordinate clause to be realized as a constituent of a higher clause, as shown in the following
sentence:
John continues to shock Steve.
In this sentence, John is semantically the subject of to shock but is realized as the subject
of continues.
Dutch heten is analysed in Matushansky (2008) as a subject-raising verb, i.e. the semantic
subject of the subordinate small clause is syntactically realized as the subject of heten, which
is in a higher clause. This analysis, then, assumes that the underlying structure of heten is the
same as in the older stages of Germanic (where Matushansky would analyse it as an object-
raising verb), albeit with the ability to raise the subject of the subordinate small clause. A small
clause is a minimal predicate structure in which the copula is omitted.
*haitan in Gothic and Old English
Boutkan & Siebinga 2005). Addressing a number of these issues, Green (1985)
points out that many can be resolved though he does not necessarily conclude that
the traditional etymology is correct. Because of the numerous issues surrounding
the proposed Indo-European etymology, some scholars suggest that the etymol-
ogy is unclear (Philippa et al. 2009) or that the word has no Indo-European ety-
mology (Boutkan& Siebinga 2005). What is most important to take into account
is that the discussion surrounding the etymology of *haitan has never posited
medio-passive or passive voice as an inherent part of its meaning, though such
uses could be encoded through inflection or periphrastic constructions, so we
need to account for its shift from a purely transitive verb to its modern use by
investigating its use over time in the older Germanic languages.
In this study, I examine and compare the uses of Gothic haitan and Old
English htan. By examining the behaviour of these verbs in these languages, we
will have a firm grasp on the original functions of *haitan in Germanic, which
will later allow us to have a better understanding of how this verb develops over
time and then how best to analyse its current structure. The questions addressed in
this study are the following: (1) What functions do haitan and htan fill in Gothic
and Old English, i.e. how and in how many ways can it be used? (2) For each func-
tion, is there an indication that certain functions favour either active or passive
voice more strongly? An inclination toward passive voice in the naming function
might indicate a move toward the current usage. (3) How do the functions filled by
haitan and htan change over time? Competing functions might show the moti-
vation for *haitan to develop in the way it does. (4) In Gothic, what other verbs
compete with haitan in its various functions, and is there any indication that other
verbs are preferred to haitan in those functions? Strong competitors might indi-
cate pressure on *haitan to develop into its current copula-like function.
2. Methodology
2.1 Corpora and data collection

For the first part of the study in both Gothic and Old English, two electronic
corpora were used. The Gothic data were collected from the website of Project
Wulfila(2004) using the search engine available on its website. The corpus includes
the Gothic Bible (New Testament) and minor fragments, including Nehemiah,
Skeireins, Signatures, and Calendar, totalling around 67,400 tokens. All inflected
forms of haitan are included, but prefixed forms of haitan were avoided because of
resultant changes in valency and meaning. The tokens were collected by examining
the instances of haitan recorded in the concordance available on Project Wulfila
Robert A. Cloutier
and grouping them based on the construction in which they occurred, exemplified
in (2) below. Within each construction, the examples were further divided based
on the voice in which they occurred.
The Old English data were collected from texts in the Helsinki Corpus of
EnglishTexts and were divided into Early Old English (up to 950; OE1 & OE2)
and Late Old English (9501150; OE3 & OE4). Only texts that clearly fit into one
of these periods or that were ambiguous between periods within the same group-
ing were used. In total, the Early Old English corpus used for this study totalled
around 82,195 words and the Late Old English corpus around 135,770 words
(refer to the appendix for a list of the texts examined). All inflected forms of Old
English htan as well as the infinitive and past participle (with or without ge-),
whether part of a periphrastic construction or a non-finite participial construc-
tion, are included in the data. Other prefixed forms of htan are avoided, includ-
ing the lexeme gehtan to promise, in order to avoid changes in valency and
meaning and to maintain comparability with Gothic, where gahaitan is clearly a
separate lexeme from haitan. Additionally, the synthetic passive forms, htte and
htton, are included in this study and counted as passive for two reasons: to main-
tain symmetry with the Gothic data and because htte and htton also appear in
alternation with periphrastic passives illustrated below. The Old E nglish tokens
were collected using WordSmith, a concordance program, and were grouped
based on the construction in which they occurred, exemplified below. Within
each construction, the examples were further divided based on the voice in which
they occurred.
Instances of either haitan or htan were collected and grouped based on
the combination of arguments that occurred with the verb in the given context.
Among the data were five possible argument configurations as listed in (2) below,
with an example of each from either Gothic or Old English:
(2) a. the calling construction (with a direct object only)

ak an waurkjais dauht, hait unledans, gamaidans, haltans, blindans
But when you have a feast, call the poor, the maimed, the lame,
theblind (Goth, Luke 14:13)
b. the transitive naming construction (with a direct object and an
objectcomplement)
Daweid ina fraujan haiti
David calls him Lord (Goth, Luke 20:44)
c. the infinitival commanding construction (with an optional direct
object and an infinitive)
Gong hre to cirican, & hat ure seofon broor hider to me cuman
Go quickly to church and command our seven brothers to come
hereto me (OE2, bede R3.266.1)
d. the subclause commanding construction (with a subordinate clause)

& heht t he biscophade onfenge & to Breotone ferde
And commanded that he receive bishophood and go to Britain
(OE2, bede R1.254.5)
e. the copular naming construction (with a subject complement)
Rodbeard het se yldesta
The eldest was called Rodbeard (OE4, chroa2 R1086.59)
These examples, with the exception of (2e), were then divided based on whether
they were expressed in active voice or passive voice, the latter of which is explicitly
marked either synthetically through an inflectional ending or analytically through
a periphrastic construction.
Non-finite appositional participial constructions, as in (3a) below, were
counted as passive voice.
In m mynstre ws in a tid festes lifes & gemetfstes liifes
(3) a.
abbud&mssepreost elwald haten
In this monastery, there was at that time an abbot and priest of pious
and modest life called thelwald (OE2, Bede R 13.434.22)
Mid y a fter longre tiide cwom to him of Breotone fore neosunge
b.
intingan se halgesta wer & se forhfdesta, Hygebald hatte
When after a long time, a most holy and most ascetic man, called
Hygebald, came to him from Britain on a visit (OE2, Bede R 3.270.16)
This was motivated by the alternation of the past participle and the synthetic pas-
sive, as in (3b), in this construction. In both constructions, both the name and
verb form constitute a reduced relative clause that is appositional in nature and
therefore outside of the main clause, i.e. the construction can be taken out without
a major change in the overall structure or meaning of the sentence. For other non-
finite participial constructions in Gothic, the morphology on the Greek verb was
used as a guide since passive and active participles have distinct inflections.
Instances in Old English containing mon one are also included and counted
as active voice, even though functionally, this construction often serves as an
alternative to the passive voice. Counting such instances as active voice is moti-
vated by the verbal morphology of htan, which is always active in such construc-
tions, the appearance of mon in the function of subject, and the accusative case
of the direct object in instances where the accusative case is distinct from the
nominative, asin (4).
(4) fter m hiene mon het casere
after that-dat him-acc one called Caesar
after that, one called him Caesar/he was called Caesar
(OE2, orosiu R10.234.20)
Robert A. Cloutier
In (4), the direct object is the masculine singular pronoun hiene him, which can
only be the accusative case. Unfortunately, most of the direct objects and/or com-
plements are like casere Caesar in this example; its case is ambiguous between
nominative and accusative.
The second part of the study focuses on Gothic and gives a sense of the com-
petition haitan had from other verbs. To discover potential competitors for Gothic
haitan in each of its functions, I first found the Greek verbs haitan was used to
translate in each of its functions. Using Biblos.com to identify other verses in the
Bible containing these Greek verbs, I then looked back at the Gothic translation of
these Greek verses to see what other Gothic verbs were used to translate the Greek
verbs. A comparison of the frequencies of the different verbs in each function
gives an indication of the extent to which haitan is entrenched in each function as
opposed to its competitors, and an examination of the voice preferences helps us
to establish whether haitan has a stronger preference for passive voice in compari-
son to its competitors.
3. Results
3.1 Functions of Gothic haitan and Old English htan

3.1.1 Gothic
The Gothic data contain 64 instances of the verb haitan used in three construc-
tions, the distribution of which is given in Table 1 below: in calling constructions
(haitan+DO), in transitive naming constructions (haitan+DO+Comp), and in
infinitival commanding constructions (haitan+DO+V).
Table 1. Distribution of Gothic haitan

haitan + Active Passive Total
DO 12 5 17 (27%)
DO+Comp 7 33 40 (62%)
DO+V 7 0 7 (11%)
Total 26 38 64 (100%)
As seen in Table 1, haitan occurs most frequently in transitive naming construc-

tions (62%) followed by calling constructions (27%) and then infinitival com-
manding constructions (11%). This shows a preference for Gothic haitan to appear
in transitive naming constructions.
The instances of haitan occurring in calling constructions appear in both the

active and passive voices, both of which occur in (5) below.
ak an haitaizau, atgaggands anakumbei ana amma aftumistin stada,
(5)
eibie qimai saei haihait uk, qiai du us: frijond, usgagg hauhis
But when you are called, go and sit down in the lowest room so that when
he who called you comes, he may say to you, Friend, go up higher
(Goth, Luke 14:10)
The second instance of haitan in (5), haihait, is the third singular preterite indica-
tive active and is representative of the active voice examples. The first instance,
haitaizau, is the second singular present subjunctive passive; one other example
of the calling construction in the inflected passive was found. The next two
instances of passive voice in the calling construction were nominalized past par-
ticiples, asin (6).
jah insandida skalk seinana hveilai nahtamatis qian aim haitanam:
(6)
gaggi,unte ju manwu ist allata
And sent his servant at supper time to say to them who were invited: Come,
for everything is ready now (Goth, Luke 14:17)
Such examples were considered passive voice since the participle is describing a
property of a group of people, namely those who have been called/invited, and
because the Greek verb it translates is , the perfect passive participle
of to call.
This leaves one rather peculiar and unexpected instance of the passive voice:
Example (7) is an instance of a periphrastic present passive with a past participle
haitans occurring with ist, a third singular present indicative of wesan to be.
saei auk in fraujin haitans ist skalks, fralets fraujins ist; samaleiko saei freis
(7)
haitada, skalks ist Xristaus
For he that is called in the Lord, being a servant, is the Lords freeman:
likewise also he that is called, being free, is Christs servant
(Goth, Corinthians I 7:22)
When we compare these clauses to the original Greek, we see that both instances
of haitan in this sentence are translations of , an aorist passive participle
of to call that is nominative masculine singular. In the Gothic translation,
each instance is translated differently: the first with an unexpected periphrastic
construction haitans ist, the second with the inflected passive haitada. Further
evidence that the first instance is actually a periphrastic present passive is the fact
that each clause in Greek only contains one other verb, namely (he) is; the
first clause has two instances of ist. This example suggests that already in Gothic,
the synthetic passive was in the process of breaking down.
Robert A. Cloutier
All examples of the infinitival commanding construction are in the active

voice, as given in (8).
gasaihvands an Iesus managans hiuhmans bi sik, haihait galeian
(8)
siponjans hindar marein
Seeing then great multitudes about him, Jesus commanded his disciples
togo beyond the sea. (Goth, Matthew 8:18)
Of the seven instances of this construction, only two have direct objects that
represent the people being commanded, like in (8). In this example, siponjans
disciples is in the accusative plural and is the group being commanded to go
beyond the sea. In the five other instances, the person or group being commanded
is not expressed, as in (9).
(9) Iesus haihait ina tiuhan du sis
Jesus commanded him to be brought to him (Goth, Luke 18:40)
In this example, ina him is the direct object of tiuhan to guide/lead.

In the transitive naming construction, haitan occurs in more construction
types than either the calling construction or the infinitival commanding con-
struction. Example (10) shows the regularly inflected active (7 instances) and
example(11) the passive voices (22 instances).
(10) haita o ni managein meina managein meina
I will call those not my people my people (Goth, Romans 9:25)
(11) jah u, barnilo, praufetus hauhistins haitaza
And you, child, will be called the prophet of the highest (Goth, Luke 1:76)
In addition to these two, we also find five instances of periphrastic preterite pas-
sive constructions: one with the preterite of wairan to become (see 12) and four
with the preterite of wesan to be (see 13). These are expected as Gothic only has
inflectional morphology for the present passive.
(12) due haitans war akrs jains akrs blois und hina dag
For that reason, that field has been called the field of blood up to this day
(Goth, Matthew 27:8)
jah qinons ozei wesun galeikinodos ahmane ubilaize jah sauhte, jah Marja
(13)
sei haitana was Magdalene, us izaiei usiddjedun unhulons sibun
And certain women, who had been healed of evil spirits and infirmities,
Mary, who was called Magdalene, out of whom went seven devils
(Goth, Luke 8:2)
Instead of the nominalized past participle construction, we find the past participle
functioning as a reduced relative clause, as in (14); this construction occurs in six
instances.
(14) habaidedunuh an bandjan gatarhidana haitanana Barabban

And they then had a notable prisoner, called Barabbas
(Goth, Matthew 27:16)
In considering the distribution of voice in the Gothic data in Table 1 above, we

observe that each construction has a clear voice preference: both the calling and
infinitival commanding constructions prefer active voice while the transitive nam-
ing construction prefers passive voice. These observations are confirmed when
these voice preferences are compared against one another using the Fisher-Yates
test (a modification of -square adjusted for smaller amounts) as the voice distri-
butions between the transitive naming construction on the one hand and either
the calling or the infinitival commanding construction on the other are statis-
tically significant with p-values of <0.001. Comparing the values for DO versus
DO+V, we might assume that the voice distribution is not due to chance, i.e. that
it is significant; however, the Fisher Yates text shows that this is not the case, with
a p-value of only 0.146.
3.1.2 Early Old English

The data collected from the Early Old English period shown in Table 2 below show
125 instances of htan occurring in four constructions: in calling constructions
(htan+DO), in transitive naming constructions (htan+DO+Comp), in infini-
tival commanding constructions (htan+DO+V), and in subclause commanding
constructions (htan+SubC).
Table 2. Distribution of htan in Early Old English

htan + Active Passive Total
DO 2 1 3 (2%)
DO+Comp 31 45 76 (61%)
DO+V 41 0 41 (33%)
SubC 5 0 5 (4%)
Total 79 46 125
The transitive naming construction (61%) is the most frequent construction in

Early Old English followed by infinitival commanding constructions (33%), sub-
clause commanding constructions (4%), and finally, calling constructions (2%).
Even in Early Old English, htan prefers to appear in transitive naming construc-
tion by quite a margin.
The calling construction occurs very infrequently in the Early Old English
corpus. Of the two instances in the active voice, (15) is a past perfect occurring
Robert A. Cloutier
with hfde had and (16) is an infinitive occurring with sceal is obliged. The one
instance of this construction in a periphrastic past passive is given in (17).
seoan he hine to Cristes eowdome gehatenne hfde

(15)
afterwards he had called him to Christs kingdom (OE2, bede R8.124.13)
(16) & hine mon sceal swie hlude hatan grdan oe singan
and one has to call, shout, or sing it so loudly (OE2, laece R5.1.1)
(17) a ws seo fmne gehaten & fter fce Eadwine onsended

then the woman was called for and after a while Eadwine sent forth
(OE2, bede R8.120.21)
These examples are noteworthy: in this corpus, (15) is the only example of a perfect
construction with htan, (16) is one of only three instances of htan in the infini-
tive, and (17) is the only example of htan in the passive voice in a construction
other than the transitive naming construction.
All instances of the infinitival and subclause commanding constructions are
in the active voice (see 18 and 19 for the former and 20 for the latter).
(18) Gong hre to cirican, & hat ure seofon broor hider to me cuman
Go quickly to church and command our seven brothers to come here
tome (OE2, bede R 3.266.1)
(19) Seo cwen het a m cyninge t heafod of aceorfan

The queen then commanded the kings head to be cut off
(OE2, orosiu R4.76.31)
(20) se papaheht t he biscophade onfenge & to Breotone ferde
the pope commanded that he receive bishophood and go to Britain
(OE2, bede R1.254.5)
The only variation in the infinitival commanding construction is the presence or

absence of a direct object, exemplified in (18) (repetition of 2c above), which is
in the third singular present tense, and (19), which is in the third singular pret-
erite, respectively. The subclause commanding construction remains the same
in all instances; the subject of the subordinate clause is always the entity being
commanded (see 20), and the verb is in the subjunctive. Examples(19) and (20)
demonstrate the two different preterite forms of htan: hht is a vestige of the
reduplicating preterite and ht is a newer form that rises in Old English.
The transitive naming construction is not only the most frequent use of htan,
but it also has the most variation. In the active voice, htan occurs both in the
regular active voice (18 instances, see 21) and the mon construction (13 instances,
see 22).
(21) eofas we hata o VII men

we call (a group of) up to seven men thieves (OE2, lawine R13)
(22) se steorra e mon on boclden ht cometa
the star that is called (one calls) cometa in book Latin(OE2, chroa2 R892.1)
In the passive voice, we find the expected split between the periphrastic passive (30
instances, see 23) and the synthetic passive (15 instances, see 24).
(23) Se resta cyning ws Ninus haten
The first king was called Ninus (OE2, orosiu R1.60.11)
(24) Minutia hatte an wifmon
a woman who is called Minutia (OE2, orosiu R6.108.15)
Within each type of passive construction are instances of the reduced appositional
relative clauses discussed and exemplified above in (3): three of the past participles
are used in this way and four instances of htte.
Except for the calling construction, which has far too few tokens for any
accurate discussion, each of the remaining constructions with htan in Early
Old English has a clear preference for voice. The transitive naming construction
(htan+DO+Comp) prefers the passive voice whereas the infinitival and subclause
commanding constructions only occur in the active voice. Using the Fisher-Yates
test to compare the ratio of active and passive voice between each pair of construc-
tions, we find that, like Gothic, the voice ratio in the transitive naming construc-
tion in Early Old English differs significantly from the infinitival and subclause
commanding constructions, as shown by p-values of <0.001 and 0.015, respec-
tively. The ratio between the two types of commanding constructions is not statis-
tically significant, with a p-value of 0.20. These data support the hypothesis that
htan favours passive voice when it is in transitive naming constructions.
3.1.3 Late Old English

The Late Old English data contain 199 instances of htan, shown in Table 3, occur-
ring in four constructions: the transitive naming construction (htan+DO+Comp),
the infinitival commanding construction (htan+DO+V), the subclause com-
manding construction (htan+SubC), and the inherent passive naming con-
struction (htan+Comp). The Late Old English data are quite different from the
distributions we observed in both Gothic and Early Old English. While the transi-
tive naming and the infinitival commanding constructions remain the two most
frequent uses of htan and both at roughly the same frequencies as the Early Old
English period, the calling construction is completely lost and the inherent passive
naming construction makes an appearance at the same frequency as the calling
construction it seems to replace.
Robert A. Cloutier
Table 3. Distribution of htan in Late Old English

htan + Active Passive Total
Comp 7 7 (3%)
DO+Comp 29 93 122 (59%)
DO+V 76 0 76 (36%)
SubC 2 0 2 (1%)
Total 106 100 206
As in Early Old English, both infinitival (75 instances) and subclause (2 in-
stances) commanding constructions occur only in the active voice (see 25 and 26,
respectively).
(25) Datianus het hine secgan of hwilcere byrig he wre

Datianus commanded him to say from which city he was
(OE3, aelive R23)
(26) & het t hi scioldon Wynflde & Leofwine swa rihtlice geseman
swahim fre rihtlicost uhte
and commanded that they should reconcile Wynfld & Leofwine as
justly as they considered most just for ever (OE3, docu3 R11)
The infinitival commanding construction can occur with a direct object, as in(25)
or not. The two examples of the subclause commanding construction, as seen
in(26), are interesting because both occur with the modal scioldon instead of the
subjunctive as was the case in Early Old English. Two instances of the command-
ing construction were found that were ambiguous between the infinitival and
subclause constructions, an example is given in (27).
& he cw to am sacerdum: Nima is Godes scrin and ga tforan am

(27)
folce & a sacerdas dydon swa swa Iosue hi het
and he said to the priests, Take this Gods ark and go before the people,
and the priests did as Joshua commanded them (OE3, otest R3.6)
The action that Joshua is commanding the priests to carry out is expressed as a
directly quoted imperative. Later when het occurs, the commanded action is left
unexpressed, so we cannot determine whether it would have been an infinitive or
a subclause.
The transitive naming construction is the same as in Early Old English: it is
the most frequent use of htan and has the most variation. In the active voice, it
occurs both in regular constructions (25 instances, see 28) and mon constructions
(4 instances, see 29).
(28) Swaswa a haligan dydan, e we hata confessores
As the holy men did, whom we call confessors (OE3, aelet3 R139)
On am geare e man ht solarem on Lyden beo reo hund daga & fif &
(29)
syxtig daga, & syx tida
In that year that is called solarem in Latin, are three hundred sixty-five days
and six hours (OE3, byrhtf R64.8)
In the passive voice, we find both periphrastic constructions (52 instances, see 30)
and synthetic constructions (17 instances, see 31).
(30) heo ws Raab gehaten

she was called Raab (OE3, otest R2.1)
(31) Se ridda heafodwind hatte zephirus

The third main wind is called Zephirus (OE3, tempo R10.19)
Within each type of passive are instances of the reduced appositional relative
clauses discussed and exemplified in (3) above: 19 of the past participles are used
in this way and four instances of htte.
The innovation in Late Old English is the presence of copular naming con-
structions (7 instances), as demonstrated in (32).
He lfde fter him reo sunan. Rodbeard het se yldesta se oer het
(32)
Willelm Se ridda het Heanric
He left behind three sons. The eldest was named Rodbeard. the second
was named Willelm The third was named Heanric
(OE4, chroe4 R1086.59)
In (32), three men are introduced, sons of the one who has died. After each is
introduced, a small description of that son is provided. I consider these instances
of the copular naming construction. However, since the verb appears in the third
singular preterite in each instance and the subject of the initial clause is a third
singular entity (he), one could conceivably interpret these as the transitive nam-
ing construction with a subject: He left behind three sons. The eldest (he) called
Rodbeard. A clearer example of the copular construction is given in (33).
Eft cw se hlend be his haligum anddetterum, a e hata confessores

(33)
onhaligum bocum
Again the lord said to his holy confessors, those who are called confessores
in the holy book (OE3, aelet3 R26)
Robert A. Cloutier
In (33), the subject of the main clause is se hlend the lord, a third singular entity.
However, the verb hata is marked for a plural subject; the only possible subject is
his haligum anddetterum. In this example, hata serves to link or identify an Old
English term anddetterum with its Latin equivalent.
The distribution of voice preference of the transitive naming construction ver-
sus the infinitival commanding construction in Late Old English htan is statisti-
cally significant with a p-value of <0.001. This shows that the observed preference
that the transitive naming construction has for passive voice is not simply chance.
Because there is only one example of a subclause commanding construction, a
statistical test would not be a reliable indicator.
3.1.4 Comparison of Gothic and Early and Late Old English

Taken together, the data from Gothic, Early Old English, and Late Old English
show clear and interesting trends. The normalized data for the three languages are
presented side-by-side in Table 4 below.
Table 4. A comparison of Gothic and Old English, normalized to 100,000 words

Gothic Early Old English Late Old English
Active Passive Active Passive Active Passive
DO 16 7 2 1 0 0
DO+Comp 10 50 38 55 21 68
DO+V 10 0 50 0 55 0
SubC 0 0 6 0 1 0
Comp 0 0 0 0 5
Early Old English seems to be an intermediary stage between what we observe

in Gothic and in Late Old English with respect to htan: like the former, Early Old
English htan occurs in calling constructions and like the latter, it occurs in the
innovative subclause commanding construction. Besides this innovation, another
striking difference between Gothic haitan and Early Old English htan is the fre-
quency with which the verb occurs in the different constructions. This may be a
side effect of differences in the sources of data. While haitan and htan both occur
most frequently in naming constructions, the second most frequent function of
Gothic haitan is the calling construction, which is the least frequent function of
Early Old English htan. Late Old English is the period that is most different from
the other two: not only is htan not used in calling constructions, italso occurs in
inherent passive naming constructions.
A striking difference between the Gothic data on the one hand and the two
periods of Old English on the other is the subclause commanding construc-
tion, which is lacking in Gothic but already present in Early Old English. Is this
construction just a later development in the Germanic languages and therefore
not present in Gothic with any type of verb at all? Or is it present in Gothic but
incompatible, i.e. ungrammatical, with haitan? Or does the Greek original just
not provide an opportunity for such a construction to be expressed in the Gothic
translation, i.e. do all instances of commanding constructions in Greek contain
an infinitive and never a subclause? Another look through the corpus turns up
instances where subclause commanding constructions appear in both Greek and
Gothic, an e xample of which is given in (34) below.
(34)
bidjands ina ei qimi jah ganasidedi ana
asking him that would-come and would-heal the

skalk is
servant his
asking him to come and heal his servant (Luke 7:3)
As we will see in the second part of this study, bidjan is one of the verbs that com-
pete against haitan in the infinitival commanding construction. When translat-
ing the Greek verb , the subclause commanding construction is actually
much more frequent in Gothic than the infinitival commanding construction:
there are only 3 instances of the former (14%) and 18 of the latter (86%). So, while
the subclause commanding construction is grammatical in Gothic, it appears to
be incompatible with haitan, and haitans use in this construction in the other
Germanic languages must be a later innovation.
3.2 Competitors of haitan in Gothic

In this second section, I investigate verbs that compete with Gothic haitan in each
of its functions. The steps in this process include (1) looking at the Greek originals
translated by Gothic haitan in each function, (2) finding the verses in which these
Greek verbs occur, (3) looking back at the Gothic translation to see what other
Gothic verbs are used to translate the Greek verbs in each function. Again for the
competitors of haitan, I compare preferences for voice.
Table 5 shows the Greek verbs that are translated by haitan and the distribu-
tion per structure. As might be expected, Gothic haitan does not appear to be an
exact equivalent of a particular Greek verb as none of the Greek verbs appears in
all of the functions.
Robert A. Cloutier
Table 5. Greek verbs translated by Gothic haitan per construction

Gothic Greek Active Passive Total
8 5 13
haitan+DO
3 0 3
Total 11 5 16
7 23 30
0 7 7
haitan+DO+Comp
0 3 3
0 1 1
Total 7 34 41
3 0 3
haitan+DO+V 3 0 3
1 0 1
Total 7 0 7
The discussion in the rest of this section will be divided by construction type.
Table 6 shows how the Greek verbs and are translated into
Gothic when they occur in the calling construction, i.e. when haitan means to
call for, to bid, to invite, etc. and implies motion on the part of the addressee
toward the speaker. It is clear from these data that haitan is in competition with
other verbs that actually occur more frequently than haitan. In the case of ,
the Gothic verb laon occurs in 18 instances, or 47% of the total, compared to
the 13 instances of haitan, or 34%. In the case of , the verb wopjan occurs
in 9instances, or 53% of the total, compared to the 3 instances of haitan, or 18%.
Table 6. Gothic translations of the Greek calling construction

Greek Gothic Active Passive Total
laon 13 5 18 (47%)
ga-laon 1 3 4 (11%)
+DO at-laon 0 2 2 (5%)
haitan 8 5 13 (34%)
at-haitan 1 0 1 (3%)
Total 38
wopjan 8 1 9 (53%)
+DO at-wopjan 5 0 5 (29%)
haitan 3 0 3 (18%)
Total 17
These data already show strong competition for this function and suggest that
haitan may be on its way out. The contrast is even more striking if we combine
these occurrences with the prefixed forms of the verbs: a total of 24 instances of
laon (63%) versus 14 instances of haitan (37%) in the case of and a total
of 14 instances of wopjan (82%) versus 3 instances of haitan in the case of .
These data may also suggest that the lower occurrence of this DO haitan structure
in comparison to the DO Comp haitan structure shown in Table 1 above may not
be a result of an overall lower frequency of this structure in the corpus but actually
further evidence that haitan is losing this function to other verbs.
A noteworthy observation occurs in examples (35) and (36) given below. In
the examples, wopjan and at-wopjan appear as infinitives used in conjunction with
haitan. Their occurrence in these examples may be a strategy for avoiding repeti-
tion of haitan because haitan is already being used as the verb of commanding.
(35) jah haihait wopjan du sis ans skalkans

and commanded to.be.called to himself the servants
and he commanded the servants to be called to him (Luke 19:15)
(36) jah gastandands Iesus haihait atwopjan ina

and standing the Jesus commanded to.call him
and standing, Jesus commanded (them) to call him (Mark 10:49)
This avoidance in Gothic may have been reinforced by the fact that the Greek
version also has two different verbs: the verb of commanding, i.e. haitan, is ,
the third singular aorist indicative active form of , while the verb of calling,
i.e. wopjan, is a form of . These examples are also interesting because they
demonstrate a preference in Gothic for haitan, given competition between two
ofits functions, to act as a verb of commanding rather than as a verb of calling,
perhaps further evidence that the construction DO haitan is in decline.
The four Greek words shown in Table 7 below are the ones that are translated
with the construction DO Comp haitan, where the complement is most often a
name. A noteworthy observation is the fact that one of the four Greek words is the
noun meaning name. This will be relevant when examining competition
among the various Gothic forms used to translate these words. As in the case of
the calling construction, haitan has some competition with other words, but the
data show a very interesting picture. In the cases of and , haitan
is clearly preferred: when translating , haitan occurs in 30 instances, or 94%,
clearly dominating namnjan, which only occurs in 2 instances, or 6%, and there
is only one instance of the naming construction with the Greek verb ,
which was translated by haitan. These examples suggest that haitan is strongly
associated with the naming construction in Gothic. While at first glance the data
Robert A. Cloutier
from and might seem not to support this, the data warrant closer
examination, which follows.
Table 7. Gothic translations of the Greek naming construction

haitan 7 23 30 (94%)
DO Comp
namnjan 0 2 2 (6%)
Total 32
haitan 0 7 7 (47%)
DO Comp qian 5 1 6 (40%)
namnjan 0 2 2 (13%)
Total 15
namo 0 10 10 (77%)
DO Comp
haitan 0 3 3 (23%)
Total 13
DO Comp haitan 0 1 1 (100%)
When translating the verb , haitan has competition from two verbs,
namely qian and namnjan, though the competition from namnjan is negligible
with only 2 occurrences, or 13%. The competition haitan has from qian, on the
other hand, is quite considerable: the former occurs in 7 instances (47%) while
the latter occurs in 6 (40%). What is even more interesting is the distribution of
these words with respect to voice: all of the instances of haitan are in the passive
voice whereas all but one of the examples of qian are in the active voice, a statisti-
cally significant distribution with a p-value of 0.005 by the Fisher-Yates test. This
reinforces the observation from Table 1 above that haitan in transitive naming
constructions prefers passive voice even in comparison to other naming verbs, an
important characteristic that may explain its later development.
The only instance where haitan is clearly disfavoured is in translating the
Greek : it only occurs in 3, or 23%, of the examples while 10 instances, or
77%, are translated by the noun namo. However, as mentioned before,
itself is not a verb but a noun, so it is not surprising that it would be translated
with a noun in a majority of the cases. appears in two main ways in naming
constructions. Example (37) below illustrates one of these structures. The nomi-
native form of /namo appears with a genitive (or dative) pronoun (/
izos) which is then linked to a name (/Aileisabai) without a copula.
This structure appears to be grammatical, or at least permissible, in Gothic as
well as Greek.
(37)
jah qeins is us dauhtrum Aharons, jah
and wife his from the daughters Aarons and the

E
namo izos Aileisabai
name her Elizabeth
And his wife was among the daughters of Aaron, and her name
wasElizabeth (Luke 1: 5)
The second structure is shown in example (38). In this structure, a name (I/
Jaeirus) is attributed to a noun phrase ( /ains ize swna-
gogafade) by means of /namin, the dative case of /namo. In the
examples collected, it appears that this structure is appositive in nature and occurs
in both Greek and Gothic.
(38) ,
jah sai qimi ains ize swnagogafade namin Jaeirus
and behold comes one of.the synagogue.rulers by.name Jairus
And behold, one of the synagogue rulers comes, Jairus by name.
(Mark 5:22)
Gothic generally translates these structures in the same way as they appear in
Greek, but in three instances, the Gothic translation includes haitan with varying
degrees of resemblance to the original Greek structure. This is perhaps testimony
to the strong association of haitan with naming constructions as not even the verb
namnjan, which itself is derived from the noun namo and might be considered a
more appropriate choice, is used to translate when a verb appears. A closer
examination of these examples will show the different strategies employed by
Gothic for translating this structure.
In example (39) below, Gothic maintains the same structure as found in the
Greek with respect to the naming construction: both include the dative of the
noun name. The only difference is that Gothic includes haitans, the past partici-
ple of haitan, which is unexpected because example (38) above demonstrates that
this is grammatical without haitans. It is noteworthy, however, that the Gothic
translation rephrases the Greek one, a restructuring that is so dramatic that it
necessitated presenting the two versions separately, which may explain the addi-
tion of haitans.
(39) a.
beggar but certain by-name Lazarus was.placed at the

gate his wounded
And a certain beggar, Lazarus by name, was placed at his gate covered
in sores (Luke 16:20)
Robert A. Cloutier
b. i unleds sums was namin haitans Lazarus sah

but beggar certain was by-name called Lazarus he
atwaurpans was du daura is, banjo fulls
placed was at gate his wounds full
And there was a certain beggar called Lazarus by name; he was placed
at his gate full of wounds (Luke 16:20)
The original Greek is one clause whereas the Gothic translation splits this clause
into two, one specifically to introduce the beggar to the reader before continuing
with the story. The introductory nature of the clause may have contributed to, or
perhaps even necessitated, the addition of haitans.
Example (40) below is interesting because the Gothic translation restructures
the original Greek clause in a more dramatic way than the previous example, as
reflected in the presentation of both versions of the clause and in the English
translations of the clauses. Although the Greek and Gothic clauses both include a
form of the noun name, the Greek has the noun in the nominative case while the
Gothic has namin, the dative form of namo.
(40) a.
was but name of.the servants Malchus
And the name of the servant was Malchus (John 18:10)
b. sah an haitans was namin Malkus
he but called was by-name Malchus
And he was called Malchus by name (John 18:10)
The purpose of this clause is to introduce the name of one of the participants in
a narrative, similar to what we see in example (39) above. However, this clause
appears after the participant himself has already been introduced into the nar-
rative, albeit without his name; in fact, the narrative with respect to this par-
ticipant is complete at the time his name is introduced. But perhaps it is this
introductory nature of the name itself that necessitates the addition of haitans
in the Gothic.
The final example is given in (41) below. This one differs from the previous
two in that the Gothic translation employs a completely different construction.
It does not include the noun namo in any form; the original Greek construction,
a dative relative pronoun with the nominative of and without a copula, is
simply translated with haitada, the present indicative passive form of haitan.
(41)
in baurg Galeilaias sei haitada Nazarai
into city the Galilee which is-called Nazareth
into the city of Galilee, which is called Nazareth (Luke 1:26)
What is interesting is that the relative pronoun in Greek is in the dative case while
the Gothic relative pronoun is in the nominative. It is not clear in this case why the
Gothic translation uses haitan as there are other examples where a similar relative
clause is translated into Gothic with the same structure as the Greek.
The final structure in which Gothic haitan appears is the infinitival com-
manding construction. Table 8 shows the distribution of the Gothic translations of
the Greek verbs. As we can see, this construction is on the whole not very frequent.
Table 8. Gothic translations of the Greek infinitival commanding construction

haitan 3 0 3 (75%)
DO V
us-laubjan 1 0 1 (25%)
Total 4
bidjan 2 0 2 (67%)
DO V
haitan 1 0 1 (33%)
Total 3
DO V haitan 3 0 3
It appears to occur only in the active voice for all verbs, and haitan does have
competition with other verbs. Given the small numbers, however, we cannot really
estimate the extent of this competition.
4. Discussion and conclusion
The Gothic and Old English data give us a hint at the behavior of *haitan in
the earlier stages of Germanic, and the Old English data show one way in which
the later languages change its usage. When we compare the various functions of
Early Old English htan to Gothic haitan, we see that the verb in these stages of
each language behaves in the same way for the most part, the main innovation
being the use of Early Old English htan in subclause commanding constructions.
Moreover, based on the Gothic and Early Old English data, it seems that *haitan
did not originally appear in the copular naming construction in the earliest stages
of Germanic; this use is an innovation that does not appear until later in English,
namely during the Late Old English period.
These data also show that *haitan in Gothic and both stages of Old English
most often occurs in transitive naming constructions, which may explain why
a derivative of this meaning seems to survive into the other modern Germanic
Robert A. Cloutier
languages. Even more interesting is that fact that in this particular construc-
tion, *haitan strongly prefers passive voice; the distribution of voice preferences
between this and each of the other constructions is confirmed statistically and
found to be significant in each language and period. This strong and statistically
significant preference perhaps leads to *haitans later appearance in copular
naming constructions in Late Old English.
Having a closer look at the Gothic data with respect to this verb, we see that
haitan has competition with a number of other verbs. The paucity of data in infini-
tival commanding constructions does not allow any reasonable observations. In
the calling construction, however, haitan has strong competition from the verb
laon to invite, call for, which occurs around two times more frequently than
haitan. In the transitive naming construction, we see that haitan is the favoured
verb, especially in passive voice.
Future research on *haitan should focus on its use in the other older G
ermanic
languages and subsequent stages in each to get a better grasp on the original prop-
erties of this verb and the factors influencing its later development. More detailed
research on its use in Old English is also needed since htan is the only verb in Old
English that retains synthetic passive inflection; the distinction between the syn-
thetic passive use and the periphrastic passive has not yet been established.
Appendix
Early Old English Texts

codocu1: Documents (Harmer; Robertson; Birch)
conorthu: Caedmons Hymn; Bedes Death Song; The Ruthwell Cross; The Leiden Riddle
cobede: Bedes Ecclesiastical History
cobrunan: The Battle of Brunanburh
cochroa2: Chronicle Ms A Early
cocura: Alfreds Cura Pastoralis
codocu2: Documents 2 (Harmer; Robertson; Sweet-Whitelock)
colaece: Laeceboc
colaw2: Laws (Alfreds Introduction to Laws; Alfred; Ine)
coohtwu2: Ohthere and Wulfstan (Ms L)
coorosiu: Alfreds Orosius
coprefcp: Alfreds Preface to Cura Pastoralis
Late Old English Texts

coaelet3: Aelfrics First and Second Letters
coaelhom: Aelfrics Catholic Homilies (II)
coaelive: Aelfrics Lives of Saints
coaepref: Aelfrics Preface to Cath. Hom. I; II

coaepreg: Aelfrics Preface to Genesis
coapollo: Apollonius
cobenrul: The Benedictine Rule
cobyrhtf: Byrhtferths Manual
cochroa3: Chronicle Ms A Late (O3)
codocu3: Documents 3 (Harmer; Robertson; Whitelock)
codurham: The Durham Ritual
colaw3: Laws (Eleventh Century)
colindis: Lindisfarne Gospels
cootest: The Old Testament
corushw: Rushworth Gospels
cotempo: Aelfrics De Temporibus Anni
cowsgosp: West-Saxon Gospels
cowulf3: Wulfstans Homilies (O3)
cochroe4: Chronicle Ms E (O3/4)
codocu4: Documents 4 (Robertson; Robertson, Appendix)
colaw4: Laws (Late; William)
coleofri: An Old English Vision of Leofric
References
Boutkan, Dirk & Siebinga, Sjoerd Michiel. 2005. Old Frisian Etymological Dictionary. Leiden:
Brill.
den Boon, Ton & Geeraert, Dirk. 2008. Van Dale: Elektronisch Groot Woordenboek van de
Nederlandse taal, 14th edn. Utrecht: Van Dale Lexicografie.
de Vries, Jan. 1962. Altnordisches etymologisches Wrterbuch, 2nd improved edn. Leiden:
E.J.Brill.
Green, Eugene. 1985. On Early Germanic *hai-t-a. International Journal of American Linguistics
51(4): 425427.
Haeseryn, W., Romijn, K., Geerts, G., de Rooij J. & van den Toorn, M.C. 1997. Algemene
Nederlandse Spraakkunst, 2nd revised edn. 2 Vols. Groningen/Deurne: Martinus Nijhoff
uitgevers/Wolters Plantyn.
Kluge, Friedrich. 2002. Etymologisches Wrterbuch der deutschen Sprache, 24th edn. Berlin:
Walter de Gruyter.
Lotspeich, Claude M. 1933. Germanic Strong Verbs of Class VII. Journal of English and G
ermanic
Philology 32: 28192.
Mailhammer, Robert. 2007. The Germanic Strong Verbs: Foundations and Development of a
New System [Trends in Linguistics. Studies and Monographs 183]. Berlin: Mouton de
Gruyter.
Matushansky, Ora. 2008. On the linguistic complexity of proper names. Linguistics and Philoso-
phy 12: 573627.
Nielsen, Hans Frede. 2000. The Early Runic Language of Scandinavia [Studies in Germanic
Dialect Geography]. Heidelberg: C. Winter.
Robert A. Cloutier
Philippa, Marlies, Debrabandere, Frans, Quak, Arend, Schoonheim, Tanneke & van der Sijs,
Nicoline. 20032009. Etymologisch Woordenboek van het Nederlands, 4 Vols. Amsterdam:
Amsterdam University Press.
Pokorny, Julius. 1951. Indogermanisches Etymologisches Wrterbuch. Bern: A. Francke.
Project Wulfila. 2004. Antwerp: University of Antwerp. http://www.wulfila.be
Seebold, Elmar. 1970. Vergleichendes und etymologisches Wrterbuch der Germanischen starken
Verben. The Hague: Mouton.
Incipient Grammaticalisation
Sources of passive constructions
in Old High German and Old English*

University of Western Sidney, Australia / Leibniz Universitt Hannover, Germany
In this paper we deal with Old English and Old High German copula
constructions combining verbs denoting be and become with past participles,
which are traditionally analysed as periphrastic passive constructions. We
propose that these constructions should not be seen as grammaticalised passives
but rather as fully compositional structures. We investigate these constructions
from an aspectual perspective and argue that the passive is only one of several
possible readings for these constructions, though one that follows logically from
certain combinations. In particular, we show that the copula verbs act as aspect
operators that select different parts of the event structure of the past participle,
and that transitivity is the crucial factor that gives rise to passive readings. As
a conclusion, we outline a detailed corpus investigation in order to catalogue
all possible readings and then ultimately make a contribution to the different
developments of the passive in English and German.
1. Introduction
Both English and High German in their earliest stages possess constructions com-
bining copula verbs denoting be and become (OE s-copula and weoran, OHG
sn/wesan and werdan) and the past participle.1 In the traditional view r epresented
* We are grateful to the audience of the workshop Contrastive study of the verbal categories
and their grammaticalization in Old English and Old High German and to two anonymous
reviewers for their very helpful comments on earlier versions of this paper.
. We follow Vennemann (2010) in the use of the term s-copula for the Old English para-
digm of the verb be that is based on Proto-Indo-European s-roots (e.g. OE is; < PIE *h1es-)
and suppletively on wesan and b-copula for the bon paradigm. See Vennemann (2010) for
further details including the functional distribution of the two copulas in Old English as well
as a hypothesis on their ultimate origin.
by older literature and most grammars, these constructions are analysed as peri-
phrastic passive constructions, replacing the inherited Germanic synthetic passive,
of which both languages preserve only fossilised relicts (it survived only in Gothic).
It is assumed that the variant with be denotes a stative passive (GZustandspassiv),
whereas the become variant denotes a dynamic passive (GVorgangspassiv). This is
also seen as being connected to transitivity: Transitive verbs combined with either
type of auxiliary are commonly interpreted as passive constructions. Intransitive
verbs combined with be auxiliaries are usually interpreted as non-passive resulta-
tive-stative (or later as periphrastic perfect with motion verbs), whereas combina-
tions with the become auxiliary are usually not mentioned, probably because they
are seen as theoretically impossible (but see below). As passives these construc-
tions only permit the promotion of the direct object (the non-subject argument
marked by accusative case, usually the patient) of the active version to the subject
of the respective passive version but not other arguments (see e.g. Fischer & van
der Wurff 2006:153 for Old English).
However, this analysis is not uncontested, as there are problems attached to
it. In recent work, the following points have been paid particular attention to. The
main issue is whether these constructions can be seen as fully grammaticalised
passives as the traditional view maintains or whether the passive readings are
compositional and a result of the combination of copula plus past participle, a
position developed in more recent research, especially for Old High German. For
instance, the fact that both Old English and Old High German use other verbs
with similar meaning (OE becuman, weaxan; OHG beliben, ligen, stn) in place
of the above-mentioned copulas with the same or similar result, has been taken
as indication that the passive construction was not yet entrenched as a grammati-
calised passive. However, the fact that the agent is frequently expressed by a prepo-
sitional phrase, both in Old English and in Old High German, has been interpreted
as evidence for a diathetic relationship with a corresponding active sentence and
against a simple predication involving a copula and an adjective derived from a
verb (participle).
In this paper we review the discussion and investigate the compositionality of
copula constructions with the past participle in Old High German and Old E nglish
from an aspectual perspective. We argue that the passive is only one of several
possible readings for these constructions, though one that follows logically from
certain combinations of auxiliary and participle. In particular, we show that the
copula verbs act as aspect operators that select different parts of the event structure
of the past participle, and that, as has been asserted in the literature, transitivity is
key to passive readings. In addition, we outline a detailed corpus investigation in
order to catalogue all possible readings and then ultimately to make a contribution
to investigating the different developments of the passive in English and German.
This paper is structured as follows. Section 2 points out some problems and
issues with the notion of a grammaticalised passive in Old High German and Old
English. Section 3 then proposes a composite analysis of the aspectual semantics
of the copula plus past participle constructions in those languages. In Section 4
some considerations about incipient stages of grammaticalisation are presented
and discussed before Section 5 offers some conclusions and sketches pathways for
further research.
2. P
roblems with the traditional view of an Old High German
and Old English passive
This section discusses some problems attached to the traditional view that both
Old High German and Old English had a grammaticalised passive. We show that
the support for positing a periphrastic passive is slender and ambiguous, though,
as will become clear in Section 4, in both languages a grammaticalisation process
is incipient.
2.1 Old High German

In the last twenty to thirty years, several studies have proposed to treat sn/wesan/
werdan plus past participle as constructions that primarily convey meanings
related to aspect and/or aktionsart, thus they explicitly contest the traditional view
described above (see e.g. Oubouzar 1974; Eroms 2000; Kotin 2003; Schrodt 2004;
Vogel 2006; Jones 2009, inter alia).
Eroms (2000:25f), for example, treats these constructions analogously to con-
structions with past participles of intransitive verbs and integrates all of these into
a homogeneous paradigm of sn/wesan/werdan plus past participle constructions
(similar views are presented in e.g. Schrodt 2004 and Jones 2009). But even though
his aim is to highlight the close parallels between different constructions with the
copula verbs sn/wesan and werdan, he sticks to the term passive when describ-
ing structures with transitive verbs, putting them in a clear aspectual opposition
to each other which he formulates as statal (with sn/wesan) vs. mutative (with
werdan). This view seems to presuppose the existence of an active diathesis cor-
responding to both of these passive constructions. We think that such a strong
interpretation in terms of a fixed diathetic value conveyed by the Old High Ger-
man constructions with sn/wesan and werdan is rather problematic. In our view,
the passive meaning of these constructions results logically from the combination
of the transitivity, aspect and aktionsart values of their respective composite parts,
just in the same way as their different aspectual values do. In particular, we see no
evidence that the passive reading of those constructions is connected in a specific

or even paradigmatic way to a corresponding active.
The compositional character of the Old High German constructions with
wesan/sn/werdan appears to be widely recognised by now. In recent literature,
they are mostly treated as non-grammaticalised combinations of copula verbs
and past participles (Jones 2009), as Kopula-Prdikativ-Konstruktionen [copula-
predicative-constructions] (Kotin 2003), or as Prdikativphrasen [predicative
phrases] (Schrodt 2004), etc. Thus, the constructions with past participles are
assigned the status of compositional structures whose semantics are said to be
derivable and predictable from the semantics of their respective composite parts.
The view presented in this paper goes in the same direction: we argue that that the
Old High German combinations of wesan/sn/werdan with past participles cannot
be considered grammaticalised passives.
In spite of these widely shared insights, the term passive is still applied to
these constructions, and has not yet been replaced by another, more appropriate
label. Even in some recent publications which admit that the traditional classifica-
tion of these constructions as passive is too strong, they are still called passive
constructions, Passivumschreibungen, or even just Old High German passive,
and, which is even more important, they are considered to contrast paradigmati-
cally with the active forms of the respective verbs (see e.g. Schrodt 2004; Jones
2009, et al.).
In Verbindung mit den Transitiva bilden die Konstruktionen sein + Partizip
II und werden + Partizip II ein inaktivisches Pendant zu den entsprechenden
Aktivfgungen mit dem Partizip I.
[In combination with transitive verbs the constructions sein + past participle
and werden + past participle form an inactive counterpart to the corresponding
active constructions with the present participle.] (Kotin 2003:8182) [our
emphasis; RM/ES]2
The recent attempts to describe the Old High German copula constructions in
terms of their aspectual and/or aktionsart semantics show considerable differ-
ences. For example, Schrodt (2004:11ff), similarly to Kotin (2003), argues that the
semantic opposition between wesan/sn and werdan is stative vs. mutative, thus
attributing the crucial role to the aktionsart character of the copula verb. In a simi-
lar vein, Eroms (2000) calls the constructions with wesan/sn Vorgangspassiva
and those with werdan Zustandseintrittspassiva. Fritz (1994) chooses the aspec-
tual formulation Imperfektivpassiv for wesan/sn constructions as opposed to
Perfektivpassiv for werdan constructions. He also mentions that, in most cases,
. All translations are ours unless marked otherwise.

the durative semantics of sn overrides the semantics of the past participle of a

transformative or perfective verb.
Jones (2009), by contrast, proposes a more elaborate model in which each com-
ponent of the construction, i.e. the copula as well as the past participle, are assigned
a particular aktionsart value; the aktionsart of the whole periphrasis results from the
combination of these two values. In this model, werdan with a transitional as well
as with a non-transitional3 participle denotes the transition into a resultant state,
which he calls eventive. When werdan combines with a non-transitional participle
such as loved the resulting meaning is ingressive.4 A stative meaning arises when
wesan/sn is combined with a non-transitional participle like loved. The combi-
nation of wesan/sn and a transitional participle (of an active-transitional verb)
results in an eventive meaning, whereas its combination with a non-transitional
participle (of an active-transitional verb) yields a resultative meaning for the whole
construction. Thus, the combination of wesan/sn is said to be able to yield two
different meanings, depending on the aktionsart value of the past participle of the
same active verb. In this connection, Jones (2009) refers to an idea put forward by
Schrder (1955) and more recently by Abraham (1991) according to which the Old
High German past participle had both transitional and non-transitional meanings.
So, gioffanot could be interpreted either as geffnet werdend (being opened),
i.e. denoting the transition into an opened state, or as geffnet (worden) seiend
(having been opened), i.e. denoting the opened state itself. This possibility is not
allowed for in the conceptions presented by e.g. Schrodt (2004), Eroms (2000) and
Kotin (2003). However, Jones (2009) does not pay any attention to the transitiv-
ity of the verb occurring as past participle in the combination with wesan/sn and
werdan. He seems to assume that only past participles of transitive verbs combine
with the copula verbs to form a passive periphrasis. Other authors take into account
that wesan/sn and werdan often co-occur with past participles of intransitive verbs.
These constructions have recently been integrated into the Old High German ver-
bal system (see Eroms 2000; Schrodt 2004) on purely formal grounds; their exact
function is, however, still a matter of debate (see e.g. Kuroda 2010).
At this point, we can conclude that most recent work accepts the existence of
an aktionsart and/or aspectual opposition between two copula constructions that
. For Jones (2009), transitional aktionsart value denotes a verbal situation where there is a
transition, i.e. a change of state, from one situation to another, while non-transitional denotes
a situation without a transition. In this sense, this opposition is very close to our terminative
vs. non-terminative.
. Jones uses the term ingressive where we usually would expect inchoative; his definition
is transition into a non-resultant state (2009:51).
is connected to the choice of copula verb: sn/wesan [stative, imperfective, non-

transitional, non-transformative, etc.] versus werdan [change of state, perfective,
transitional, transformative, etc.]. However, this opposition is often blurred, if one
looks at the data more closely: these constructions are often used interchangeably,
without any recognisable semantic differences (cf. esp. Kotin 2003:55f; see 2.2
below for a similar observation about Old English).
Moreover, the precise semantic status of sn/wesan plus past participle in Old
High German (as well as in Modern German, see e.g. Maienborn 2007) is still
a matter of discussion. Should it be considered resultative (cf. e.g. Leiss 1992),
stative (cf. e.g. Schrodt 2004) or both, depending on the particular nature of the
verb as well as on the specific interpretation of its past participle (which presum-
ably requires that the specific context of use should be accounted for)?
Furthermore, the changes in the system of the Old High German construc-
tions with copula verbs and past participles have still not been described and
explained concisely enough. At the beginning of the Old High German period,
the construction with sn/wesan appears to have been preferred to that with
werdan. Towards the end of the Old High German period, however, the werdan-
construction becomes more frequent. In this connection, a markedness reversal
effect is sometimes mentioned: the unmarked sn/wesan-construction is said to be
replaced by the marked, i.e. mutative, werdan-construction, which subsequently
becomes the unmarked choice in the passive opposition (cf. e.g. Kotin 2003:44).
But the details and motives for this reversal process remain for the most part
obscure. Furthermore, the markedness values for these constructions remain
to be ascertained independently. Though perfective aspect cross-linguistically
more frequently forms the marked member of the opposition, ultimately this
is language-specific (see Sonnenhauser 2006 for details). Additionally, the status
and development of the Old High German impersonal passive constructions with
werdan and their role in the diachrony of the German Vorgangspassiv has not
been examined in detail and remains for the most part unclear; cf. e.g. Das his-
torische Verhltnis der unpersnlichen Passivkonstruktion zu der persnlichen
ist nicht ganz klar5 (Dal 1966:130) and The rise of the construction is obscure
(Lockwood 1968:145f).
2.2 Old English

The literature has pointed out several problems with the traditional view according
to which the construction of s-copula plus past participle denotes a stative passive
. The historical relationship of the impersonal passive construction to the personal one is
not entirely clear.
and that of weoran plus past participle denotes a dynamic passive in Old English.
The first question is whether there really are two kinds of passives in Old English;
the second question concerns the fact whether we can speak about a grammati-
calised passive at all or whether we are simply dealing with copula constructions
in which passive-like readings are purely compositional following logically from
the combination of verb plus participle. In this section we show that the notion of
a grammaticalised passive in Old English is problematic, paving the way for our
hypotheses in Section 3.
Textbooks and handbooks have noted that the functional split of stative vs.
dynamic passive exists only in theory and that it is not always reflected in the data
(see e.g. Brunner 1965:267 and Fischer & van der Wurff 2006:152). This is par-
ticularly evident from past tense examples such as (1).
(1) a. Her wear Eadwine cining ofslagan fram Cadwallan.
here got E. king killed by C.
Here King Eadwine got/was killed by Cadwalla. [c. 633, MS Parker D]
b. Her ws Osuuald ofslagen Norhymbra cining fram Pendan.
here was O. killed Northumbria king by P.
Here Oswald, king of Northumbria was killed by Penda
[c. 642, 0MS Parker D]
Different manuscript versions often have alternative versions, such as the MS

Parker A variant of (1a), which reads Her Edwine ws ofslgen (see also Visser
1973:20912092 for more extensive documentation of this phenomenon). It has
also been argued that stylistic considerations were factors in selecting one con-
struction over the other (Frary 1929). The functional overlap between the two
forms of passive constructions has been seen as instrumental in the loss of
weoran in Middle English (e.g. Klingenbiel 1937; Brunner 1962; Fischer & van
der Wurff 2006:153; but see Petr 2010 for a critical discussion).
However, all examples of functional overlap quoted in the literature are in the
past tense, and this is exactly what would be expected from the combination of
tense and aspect values of the respective combinations. Hence, this is not really a
convincing argument against two theoretically distinct passive combinations. As
the action denoted by the dynamic passive wear ofslagen literally got killed
is completed at the moment of utterance, and the significance of the progressive
action is diminished by the large temporal lens, the focus is more on the fact
that someone got killed. This is very close to a stative passive ws ofslagen was
killed, as evidenced by parallel cases in e.g. Modern German (cf. wurde erschla-
gen got killed vs. war erschlagen was killed). This is probably also the reason
why weoran plus past participle disappears about 200 years earlier in the past
tense than in the present tense (see already Visser 1973:2093 and Petr 2010:463
for quantitative data), in which there was a clearer semantic difference between
both constructions. At any rate, Petr (2010) argues that combinations of past
participle with (past tense) forms of weoran more frequently denote sudden
changes of state than combinations with be and past participle. He thinks that
this is connected to a more fundamental change in the language from bounded to
unbounded language use.
Irrespective of the difference in the past tense, in the present tense, past par-
ticiples combined with weoran are obviously distinct from past participles plus
s-copula. The former usually has future reference (similarly to constructions with
the b-copula; see also Petr & Cuyckens 2009:345), whereas the latter denotes a
state, which may be the result of a past action (see e.g. Visser 1973 for the data).
As a result of this brief discussion it certainly does not look like the two pas-
sive constructions were completely similar in function. This may be more true
for past tense contexts but not in the present tense. Thus, in principle they look
sufficiently distinct to form two passives.
The second point of contention is whether it makes sense to speak of a gram-
maticalised passive in Old English at all. Fischer and van der Wurff (2006:153)
argue that the gradual loss of inflections on participles in copula constructions is
evidence for grammaticalisation (similar suggestions have also been proposed for
Old High German). But the fact that predicatively used adjectives are not always
inflected in Old English (Brunner 1965:236) casts doubts on this hypothesis,
because participles can simply be seen as following adjectival practice. Vezzosi
(1998:56) asserts that auxiliaries experienced semantic bleaching, and although
both weoran and the s-copula are still used as non-copula-verbs at the end of the
Old English period, there is little doubt that a certain degree of desemanticisa-
tion of the finite verbs in those constructions already took place in Old English
(Petr& Cuyckens 2009:350). However, whether this actually amounts to the cre-
ation of a passive is less clear; the examples usually adduced are all in the past
tense, and there the semantic overlap is a result of the focused point of completion,
i.e. the state, due to the past tense.
Another argument for a beginning development of a passive that has been
advanced is the increasing explicit encoding of the agent of the event (Petr &
Cuyckens 2009:350). But the fact that this is possible for all copulas in Old and
Middle English (Petr & Cuyckens 2009:351) weakens this argument somewhat.
Petr and Cuyckens (2009:352) put forth a third argument, which is the
extension of possible types of aktionsart of the participle. With atelic verbs being
permissible this denotes a shift of the construction from expressing result(ative)
semantics to expressing the event itself . One has to add that they refer to transi-
tive verbs, and that they refer to a processual reading rather than to a perfect-like
reading. This becomes clear from their translation of Al es reserved uncertayne til
another worlde as Everything is (being) kept uncertain until another world. Thus,
what Petr and Cuyckens really seem to mean is that at some point the static pas-
sive reading of the s-copula and the past participle developed a dynamic passive
reading, but this seems to have been very rare in Old English (Petr & Cuyckens
2009:352 and already Visser 1973:2088). However, in our view this is not an argu-
ment in favour of an Old English passive, though it may shed some light on the
loss of weoran (as also suggested by Visser 1973:2088).
A fact that is rarely used in the discussion is that other arguments apart from
the patient (argument marked by accusative) can be used in these constructions,
such as the recipient. If that argument is marked by the dative case, it retains that
case, e.g. the type him was given a book. This suggests that there was some kind
of passive-like notion, though it would seem that the full promotion of the dative
argument to subject (marked by nominative case, i.e. ModE he was given a book)
is perhaps even more suggestive. It has been debated whether Old English already
had the latter possibility (Lieber 1979 says yes, Russom 1982 says no; cf. also
Denison 1993:110112), but the first example for this in Visser (1973:2113) dates
from 1200, and is thus very late Old English at best.
As a preliminary result of this discussion, there is some indication that
passive-like constructions began to grammaticalise in Old English, but it is not
clear at all how entrenched they were. In particular, the fact that present tense
weoran in combination with past participles usually has future-time reference
and not the meaning of a dynamic passive like the respective constructions in
Modern Dutch and German, seems to be a valid counter-argument against a fully
grammaticalised passive.
The fact that the agent of an action can be explicitly encoded with all copula
structures, points to the significance of other copula constructions with similar,
passive-like semantics. In fact, Petr (2010) argues that there was an entire net
of constructions with desemanticised copulas that were interchangeable to some
degree. Whether this was actually true and whether this played a role in the loss
of weoran constructions is a different matter, but there were a number of such
constructions in Old English (and also in Old High German). Some of these
seem relevant in this context. First, combinations of the b-copula and the past
participle are in principle equivalent to combinations of weoran and past par-
ticiple in their future use (Visser 1973:2041f and 2091). Second, becuman is to
some degree equivalent with weoran, only with a more incipient focus (Visser
1973:2030). Third, there is the quasi-passive construction ModE he had a book
given to him and variants of it, which occurs from Middle English onwards (see
Visser 1973:21552157 for discussion and further details).
To sum up, the discussion of the literature has shown that it is not clear that Old
English and Old High German really had a grammaticalised passive or whether
the passive readings arise naturally from the combination of verb plus past partici-
ple. The next section examines the predictions a compositional hypothesis would
make for the analysis of constructions with a copula-like verb and a past participle.
3. Th
eoretical considerations for the interpretation of constructions
withthe past participle in English and German
In order to assess the notion of a grammaticalised passive in Old High German

and Old English, we will analyse combinations of the main copula verbs used in
passive-like constructions, OHG sn/wesan and werdan as well as OE s-copula and
weoran, and past participles. We will argue that in Old High German and Old
English these constructions can convey passive meanings as a result of combin-
ing the aspect/aktionsart value of the copula verbs and the transitivity of the verb
supplying the past participle but not as an entrenched or fully grammaticalised
periphrastic passive. That is, we maintain that these combinations are still fully
compositional in both languages, and the passive-like readings arise from specific
combinations in a predictable way. We will investigate the parts of the construc-
tions individually, and then move on to the combinatory semantics.
3.1 The copulas: OHG sn/wesan and werdan, OE s-copula and weoran
For the purpose of calculating the semantics of copula-past participle combina-
tions, the aktionsart of the copula is pivotal. The traditional term aktionsart refers
to the inherent lexical aspect of a verb. It reflects the temporal organisation of
the event structure denoted by the verb (Bickel 1996). This time structure can
be divided into alternations of phases () and change of states or boundaries ().
Typically a change of state is preceded by a dynamic phase (dyn) and followed
by a static phase (stat) (Sonnenhauser 2006:46f; see Bickel 2000 for details on
the differentiation between dynamic and stative phases). Verbs may be catego-
rised according to their internal time structure, i.e. according to what the event
structure looks like (see Bickel 1996 and Sonnenhauser 2006 for further details).6
So, for instance, events expressing just a change of state and whose internal time
structure therefore consists of only a boundary (), such as knock, hit, etc. can be
set apart from verbs whose event structure consists only of a static phase (stat),
. It is immediately obvious that there is such an internal time structure, and therefore
having a term for this seems a good idea (see Bertinetto & Delfitto 2000 for a detailed discus-
sion). While it may be true that in some languages aktionsart does not play a role, it does not
follow from his that the concept and the term should be abandoned (pace Sasse 1991).
such as know, or verbs which comprise a dynamic phase and a change of state
(dyn, ), such as arrive, reach. There exist a variety of classification systems;
probably the most well-known is the Vendler-Dowty system (Dowty 1972 and
Vendler 1967:97121). This paper uses the time-structure approach developed in
Bickel (1996, 2000) and further elaborated on in Sonnenhauser (2006 et passim),
because of its flexibility and crosslinguistic applicability (see Mailhammer 2009
for an application to the verb system of Amurdak, a language from Northern
Australia, including references).
The internal time structure of OHG sn/wesan and the Old English s-copula
consists of a static phase (stat). Hence, their aktionsart is stative and therefore
non-terminative.7 This becomes clear from predications with adjectives or nouns,
in which OHG sn/wesan and the Old English s-copula predicates a state, quality
etc. about the subject.
By contrast, the aktionsart of OHG werdan and OE weoran is inchoative
(mutative in the terminology of Kotin 2003) and consequently terminative but
not telic. These verbs signify a change of state and consist of a dynamic phase, a
boundary and a subsequent static phase (dyn, , stat).8
(2) Inchoativity of OHG werdan/OE weoran
a. OHG
Inti sie gitruobte vvurdun.
and they afflicted got
And they got afflicted [Tatian 81, 2]
b. OE
Her Cynegils king wear gefullad fram Byrino m bisceope
here C. king got baptised by B. the bishop
on Dorkeceastre.
in Dorchester.
Here King Cynegil got baptised by Byrino, the bishop of Dorchester
[c. 635, MS Parker C]
To sum up, the two Old High German and Old English copula verbs examined
here differ fundamentally with respect to their aktionsart (internal time structure).
. Terminative aktionsarts contain a boundary in their event structure; whereas non-

terminative aktionsarts do not. A subset of terminative verbs is telic; they contain a final
boundary. See Sonnenhauser (2006) for a discussion of the terminology (with references).
. See Bickel (1996), Breu (1998) and Sonnenhauser (2006) for a detailed discussion of
how to represent and test for inchoativity. The crucial issue is that the realisation of an event
encoded by an inchoative in progress is not defeasible: e.g. it is getting dark presupposes that
it is at least a little dark (Bickel 1996:233).
While OHG sn/wesan and the Old English s-copula are stative and therefore non-
terminative verbs, consisting only of a static phase, OHG werdan and OE weoran
are inchoative, consisting of a dynamic phase, a change of state and a subsequent
phase, and therefore, terminative and atelic. As far as the aspectual values of these
copula verbs (what will become important when combinations with past parti-
ciples of other verbs are taken into account) are concerned, they serve as aspect
operators selecting particular phases and/or boundaries from the verbs internal
event structure. While OHG sn/wesan and the Old English s-copula are basi-
cally non-terminative and stative (imperfective) and focus on the static phase of
the event structure, OHG werdan and OE weoran are basically terminative and
dynamic (perfective) and select the change of state and the subsequent phase
from the verbs event structure. The next section looks at the semantics of the past
participle in OE and OHG.
3.2 Past participle

In this section we argue that the semantics of the past participle in Old High
German and Old English is primarily aspectual. The formation of the past par-
ticiple entails the selection of change of state and the following static phase; as a
result Germanic past participles are resultative-stative (perfective). Past parti-
ciples can be formed of any verb largely irrespective of aktionsart and are syntacti-
cally detransitivised and agentless.
Historically, the past participle of Proto-Germanic is a verbal noun with the
same stem-forming suffix as the infinitive, PGmc. *-na- (< PIE *-no-). The old per-
fect participle in PIE *-uos/us- (e.g. PIE *stest2- us, Att. < *hestaus, Rix
1976:235) is attested in Germanic only in relics, such as Goth. br-us-js parents,
lit. those who have carried (Rix 1976:235).
The Proto-Indo-European perfect denotes a state resulting from a pre-
ceding action expressed by the lexical component of the verb, for instance Gk
is dead/has died (e.g. Rix 1976:192; Fortson 2010:105). Its participle
is an adjective belonging to the verbal paradigm (Rix 1976:193). In Germanic
the newly created past participle is also an adjectival form associated with its
base paradigm, in this case the past tense. But its meaning is like that of the
Proto-Indo-European active perfect participle resultative-stative and indiffer-
ent to diathesis/voice (see Abraham 2000:147150 for a theoretical discussion).
As a result, the semantics of the Germanic past participle is primarily aspec-
tual (see also Abraham 2000). In aspectual terms the perfect makes an asser-
tion (called topic time, cf. Sonnenhauser 2006) about a change of state and the
following static phase. Because the preceding dynamic phase is not part of the
assertion, though it is logically presupposed, the boundedness specifications of
the aspectual lens are closed to the left and open to the right (see Sonnenhauser
2006 and Mailhammer 2009 further details, as well as Vogel 2006:126f for a
similar view for German).9 This is illustrated in Figure 1.
Aktionsart
dyn stat
Aspect
[ [
Figure 1. Resultative-stative aspectual reading of the past participle
From the aktionsart input (see 3.1 above), the past participle as aspect opera-
tor selects the culmination point and the following static phase, to make an asser-
tion about the event encoded by the verb that can be paraphrased as follows: state
resulting from a change of state.
The possibility to form a past participle is largely irrespective of aktionsart of the
verb (pace Estival 1986). The only exception seems to be the class of durative verbs,
i.e. verbs whose event structure cannot be bounded (not even coerced; see Bickel
1996:16f for the term and Sonnenhauser: 4547 for discussion), such as weigh in it
weighs a ton (weighed can only be associated with a transitive verb weigh).
(3) Modern English examples of past participles10
a. play (delimitative): played
b. knock (momentaneous): knocked
. The boundary specifications of the aspectual operator (lens) are independent of the
notion of boundary in the representation of the event structure. The former refer to bound-
edness features, i.e. the question whether the edges of the aspectual lens overlap with the
temporal boundaries of the event or not. For instance, a classical perfect, like that of Proto-
Indo-European, and a pluperfect select the same portion of the event, but the bounded-
ness features are different. The perfect includes the following stative phase and excludes the
dynamic phase preceding the change of state from the assertion, though the dynamic phase is
presupposed, i.e. it has a closed boundary to the left and an open one to the right. By contrast
the pluperfect has an open boundary to the left, because it makes an assertion to the preceding
dynamic phase, i.e. the action denoted by the verb, but it has a closed boundary to the right,
because the following static phase is not part of the assertion anymore. This can be illustrated
with Modern English, e.g. Peter has opened the door (*but its not open now). vs. Peter had
opened the door before Charles entered, but the wind had closed it again.
. The past participle in Modern English can be taken as representative for Germanic past
participles as a whole.
c. blossom (ingressive-phasal): blossomed

d. forget (telic): forgotten
e. know (stative): known
f. weigh (durative): *weighed
From a syntactic viewpoint past participles are detransitivised and agentless (see
Abraham 2000:151, 160). This is not surprising, given that participles are nominal
(adjectival) forms, expressing a state. For the discussion of the passive readings
further below, it is the transitivity of the verb that plays a crucial role in the com-
posite semantics.
To sum up, the past participle is a newly formed adjectival form in Germanic.
It is basically an aspectual operator that selects the change of state and the follow-
ing static phase from the verbs event structure, thereby asserting a resultative-
stative of the verbs lexical meaning. Consequently, the aktionsart of the verb is
effectively re-categorised as resultative-stative, but the original aktionsart is rel-
evant in the interaction with the aspectual-temporal properties of the copula
(see3.3 below).
3.3 Constructions with the past participle

This section examines constructions of the two kinds of copula verbs discussed in
Section 3.1 and the past participle in Old English and Old High German against
the backdrop of our hypothesis described in Section 3 above. We will demonstrate
that the passive-like readings in Old English and Old High German are a natural
consequence of the interaction of the transitivity and the aspectual values of the
copula and the past participle involved, and that the notion of a grammaticalised
passive is not necessary for either Old High German or for Old English. Note that
we do not, however, reject the idea that there was incipient grammaticalisation of
a passive, which continued in later periods of German and English respectively. In
combination with the past participle, which represents a resultative-stative view of
the verbs event structure, the copula verb likewise functions as aspectual modifier
with their aktionsart supplying the aspectual parameter. As will become clear, the
crucial element is the terminativity value of the copula, though particularly with
werdan/weoran tense has also to be factored in.
3.3.1 Constructions with stative copula: sn/wesan, s-copula plus past participle
As an operator the non-terminative, stative be focuses on the static phase of the
event structure, which is by definition resultative-stative, i.e. it consists of a change
of state and a following phase (see 3.1 above). Consequently the combination rep-
resents events basically as states as illustrated in Figure 2, although the completion
of the action denoted by the verb in the (recent) past is implied.
Aktionsart
dyn stat
Aspect
[ [ Assertion
Figure 2. Aspect selection of the be copula
This basic stative reading of the past participle occurs irrespective of the
aktionsart and the transitivity of the verb. However, differences in these param-
eters affect the basic meaning in different ways. As we explain further below,
transitivity is decisive for passive-like readings, this will be the focal point of the
following survey.
The combination of be with past participles of transitive terminative verbs,
such as G erschlagen slay, kill, berauben rob, results in a stative reading with a
strong implication of a recently completed action, i.e. a resultative component. The
resultative usually occurs with verbs that have no static phase in their aktionsart
input, e.g. momentaneous verbs, such as knock over: a hypothetical he is knocked
over can only mean he just got knocked over and is lying there as a result. This is
why sentences in (4) actually refer to an event time that is before the reference time
(in Reichenbachs 1947 classic terms).
(4) The passive and temporal anchoring in Modern Standard German
a. Das Fahrrad ist gestohlen (worden).
The bicycle is stolen./The bike has been stolen
b. Das Fahrrad war gestohlen (worden).
The bicycle was stolen./The bike had been stolen.
Without worden become, got (4a) and (4b) are simply predications containing
an adjective derived from a verb which could also be used attributively, e.g. ein
gestohlenes Fahrrad a stolen bicycle (see e.g. Maienborn 2007 for a convincing
analysis of the past participle as adjective). But with worden the sentences are pas-
sive referring to completed action that is responsible for a subsequent state, much
like a present perfect in Modern English.
By contrast, with non-terminative transitive verbs, there are slightly different
semantic nuances. Although the past participle gives these verbs a culmination
point (which they originally lack), and consequently coerces them into resultative-
stative, the stative aspect operator be selects again a phase rather than a boundary
and therefore emphasises a state that is the result of an ongoing event, more pre-
cisely a series of completed mini-events, giving rise to a processual or a habitual
reading. This reading (G allgemeingltige Aussagen general statements) has

been often mentioned as one of the possible readings of the Old High German
construction with sn/wesan (cf. e.g. Schrodt 2004:11; Oubouzar 1974:19). Thus, a
verb such as G plagen trouble with a delimitative aktionsart possesses a past par-
ticiple geplagt troubled, tortured, which denotes a state resulting from a previous
action. But in combination with a be copula the result is a state of being-tortured,
not a state resulting from having been tortured.
(5) Peter ist geplagt (von Alptrumen).
Peter suffers (from nightmares).
The example in (5) usually means that Peter suffers from chronic nightmares. It
could also mean that Peter suffers from several nightmares at the same time, but
normally not that Peter has suffered from a nightmare and is now awake, as this
would be for a terminative transitive verb. The way this reading can be modelled
in aspectual terms is similar to generic habituals, which depict an event repeatedly
in its totality (see Mailhammer 2009:369f and 374 for further details). These two
readings can be exemplified with examples from Old High German.
(6) Processual (a) and habitual (b) readings in Old High German
a. min tohter ubilo von themo tiuuale giuueigit ist
my daughter maliciously by this devil tortured is
my daughter is tortured maliciously by this devil [Tatian 85, 2]
b. therde mih minnot ther ist giminnot fon minemo fater
who-rel me loves who is loved by my father
this who loves me is loved by my father [Tatian 164, 6]
In (6a) the focus is on the duration of the predicated state of the subject at the time
of speaking, whereas (6b) has a generic-habitual reading. All these readings can
be accounted for by combining the aspectual operator values of the construction.
Now, the passive (or rather anticausative, see below), i.e. an agentless reading, is
a result of the deagentivisation of the participle. Basically, participles inherit the
argument structure of the verb, although in a reduced form. As the past participle
refers to a state resulting from the action denoted by the lexical component of
the verb, it is naturally predicated about the non-subject argument of a transitive
verb.11 Consequently, the subject (usually the agent) does not appear in the argu-
ment frame anymore, and the result is deagentivisation. Thus, a combination of be
and a past participle identifies the subject of the finite copula with the only argu-
ment of the participle, i.e. the object, resulting in a predication about the original
object of the verb that lacks an agent. The passive reading is just a consequence of
. The development of OE is geworden from passive to active shows that a passive reading
is not mandatory (Visser 1973:2050).
the combinatory value of a construction consisting of a finite be copula and the

past participle of a transitive verb.
By contrast, the situation is different for combinations with past participles of
intransitive verbs, irrespective of whether they are terminative (come, arrive) or
non-terminative (live, walk). These are typically more ambiguous. The past parti-
ciple is resultative-stative and agentless, and the argument of be supplies the sub-
ject. But, while the non-subject argument of a transitive verb cannot be promoted
to agent (Abraham 2000:151), the only argument of an intransitive verb can, of
course, be an agent. Hence, this constellation allows for different interpretations
depending on whether the introduced subject is seen as agent or patient (or some-
thing else). In combination with the past participle of an agentive intransitive, the
reading is not passive but active. It denotes a resultative state of the subject-agent,
e.g. OHG nu ist er queman (O II, 7, 45) lit. now he is come (common in English
until Late Modern English times) meaning now he has come.
However, in combination with the past participle of non-agentive intransitive
verbs (i.e. event verbs, e.g. happen), the reading is neither passive nor active it is
anticausative. The non-agentive subject is said to be in a particular state denoted
by the past participle, and this state is not the result of any preceding action. This
is a direct result of the verbs non-agentive nature, i.e. because the verbs sole argu-
ment is not an agent: Tnnan ist keskhen lit. then [it] is happened (see Eroms
2000; Vogel 2006, 124f).12
3.3.2 Constructions with inchoative copula: werdan/weoran plus past participle

As argued in 3.1 above, the copula in these constructions is semantically incho-
ative, i.e. its event structure consists of a dynamic phase, a change of state and
a resulting phase. In fact, it consists all of what Sonnenhauser (2006) calls the
event nucleus, i.e. the representative core of the event structure dyn, , stat.
. For example, Modern German verbs denoting motion (real or metaphorically) are
interpreted as having an agent, hence the reading of er ist gelaufen he has run is not passive
but active. But for verbs with patientive argument, such as brechen break (intr)., a combina-
tion with be results in a non-agentive state reading, cf. das Glas bricht the glass is breaking/
breaks vs. das Glas ist gebrochen the glass is broken, which can be interpreted as passive of
state or a resultative-stative, but not as active, since this would require an agentive subject. But
in Modern English this is different, as all combinations of be and past participle consistently
trigger a patientive interpretation of the subject, overriding the original constellation in the
main verb. This frequently leads to problems if the subject is not an agent, e.g. with motion
verbs, in which cases a patientive reading is coerced, e.g. the race is run (transitive run), an
adjectival predicative reading is inferred, e.g. he is gone, or the result is not attested, e.g. *he
is died meaning either he was killed or he is dead. However, historically, also the German
solution is attested with agentive subject constructions being treated as active (see Visser
1973:2042ff for details).
But as an aspect operator it is specified for adterminality (Sonnenhauser 2006); it

selects the change of state, and is therefore a perfective operator (in the sense of
a cross-linguistic meta-category, cf. Sonnenhauser 2006). In combination with a
resultative-stative participle the prediction is that the resulting reading is that of a
change into the state denoted by the participle. However, this presents a compli-
cation, because the completion of an event cannot be asserted for an event time
that is equal to the speech time (in terms of Reichenbach 1947), i.e. completion is
incompatible with present tense contexts. Consequently, an additional interaction
with tense is to be expected.
The use of werdan/weoran in the present tense combined with past parti-
ciple results in a clash that triggers an implicature, namely that the state denoted
by the past participle has not been reached yet, because its completion cannot be
asserted in the present. The logical entailment from this is that the state will be
reached in the future. This opens up the possibility for another implicature that
depends also on the original aktionsart of the verb, and this is that the change of
state is being reached gradually at the moment, i.e. inchoativity in its prototypi-
cal sense. This is usually interpreted as a process in progress. If such a processual
reading is impossible or unlikely because of the aktionsart or the semantics of the
verb (see 8b below) then the result is a future reading, i.e. a prediction about a
resultant state. Modern German shows that although both readings can be gram-
maticalised (action passive and future tense) this effect still occurs. Example (8b)
below cannot really be interpreted as passive, though it morphologically is clearly
passive and not future tense. But in general combinations of werdan/weoran in
the present tense and past participle have a processual reading, which, however,
ultimately results from the fact that the perfectivity and present tense contexts
are incompatible (see also Mailhammer 2009:371374 for further discussion and
details). Figure 3 shows the aspectual focus of OHG werdan and OE weoran.13
Aktionsart
dyn stat
Aspect
] [
Figure 3. OHG werdan/OE weoran as aspect operator
. Another frequent possibility, especially for momentaneous verbs, such as knock and hit,
is iteration, e.g. G er wird geschlagen he is being hit (see Mailhammer 2009:369 for further
details including references).
Note that the boundedness specifications of the interval that represents the
aspectual lens do not contain the boundaries of the event, because the beginning
of the change and the completion of the change need not be part of the asser-
tion made by werdan/weoran, in particular in the present tense. A sentence like
ModG es wird grn it is getting green, means that it has started to turn green and
that some of it is already green, but not necessarily that all of it is green and that
the speaker has witnessed the very beginning of the process. But some part of the
dynamic phase and the resultant state are clearly within the aspectual lens, which
is precisely why the processual and future readings can result.
By contrast, the copula in the past tense has the effect that the completion of
the event is foregrounded and that the processual interpretation comes out clearer,
only if there is a suddenly occurring event.
(7) a. Das Spiel wurde gestern gespielt.
The game was (being) played yesterday.
b. Das Spiel wurde (gerade) gespielt, als der Strom ausfiel.
The game was being played when the power went out.
As argued in 3.1 above and as is apparent from the English translation in (7a), this
is probably the reason why it appears that combinations of the past participle with
the be copula and the become copula in the past tense have identical readings, as
has been asserted in particular for Old English.
As with the be copula, the combination of werdan/weoran and a past par-
ticiple leads to different readings, which are influenced by the aktionsart and the
transitivity of the verb supplying the participle. However, while transitivity is
also crucial for the passive reading, terminativity is involved in the processual vs.
(future) completion reading.
The combination of werdan/weoran with the past participle of terminative
transitive verbs (e.g. erschlagen slay, finden find) leads to an inchoative, termina-
tive and passive interpretation: the focus is on the state resulting from the change
of state provided by become. But this is only true for the past tense; in the present
tense usually a future reading is obtained (8b), predicting the completion of the
event, thereby often effectively indicating a process (8a).14
. A reviewer remarks that cases like (8b) can also have a passive reading. We do not dispute
the availability of such a reading. In fact (8a) is such a case, as pointed out in the main text.
What we are saying, however, is that the combination of present tense werdan/weoran and
past participles of terminative verbs cannot result in a terminative reading, as opposed to
combinations of past tense werdan/weoran and past participles of terminative verbs. The
crucial point is not passive vs. non-passive but terminative vs. non-terminative, and this is
determined by the tense of werdan/weoran.
(8) a. G Der Knig wird erschlagen.

The king is being killed.
b. G Der Schlssel wird gefunden.
The key is being found, i.e. The key will be found.
a. OHG Thar wrdit fon iu fntan ein selin gibntan.
there gets by you found an ass bound
there a tied ass will be found by you [O IV, 4, 9]
b. OE a wear he gefstnod be re swiran handa to re
there got he fastened by the stronger hand to that
bre, t he hangode to eoran.
bar that he hung to earth.
then he was bound with his right hand to that bar, so that he was
hanging down to the earth
[Bickling Homilies, quoted from Petr & Cuyckens 2009:352)
The combination of werdan/weoran with a past participle of non-terminative

transitive verbs (e.g. G qulen torment, OE lufian love, OHG tuon do) results
either in a processual or an iterative-habitual reading (see Mailhammer 2009:369,
Footnote 12 for a discussion of iterativity vs. habituality in aspectual terms). This
is illustrated with two examples from Old High German.
(9) a. tho this ward sus gibrdigot, fon mo also girdinot
there that got so preached, by him so spoken
Then that was so preached, spoken by him [O II, 13, 40]
b. Wirdit thaz ouh ana wn ofto in smbazdag gidn
gets that also without doubt often in Saturdays done
This is often done without doubt on Saturdays [O III, 16, 37]
(9a) exhibits the processual reading, while (9b) shows that the combination of
werdan and gidn conveys habituality, which becomes clear from the adverb ofto
often. It is fairly typical that temporal adverbs are key to the interpretation of the
construction. This also supports the assumption that we are not dealing with a
fully grammaticalised passive in Old High German and also Old English.
As with combinations with be and the past participle, due to the deagentivisa-
tion of the main verb in the participle, transitive verbs appear as intransitive and
because the state is a result of the action denoted by the verb, the subject introduced
by werdan/weoran is seen as the non-subject argument of the originally transitive
main verb. Consequently, the combination of werdan/weoran and a past participle
of a transitive verb gives rise to a passive reading that frequently conveys a process,
hence the German term Vorgangspassiv (passive of action, i.e. a dynamic passive).
Past participles of intransitive verbs in combination with werdan/weoran are
usually interpreted as progressive, which is actually an implicature resulting from
a prediction. For example OE wear agangan literally says that the subject was in
the process of reaching a state of gone-ness. This is not only a prediction that the
subject will be gone at some point in the future, but it is a processual statement, the
subject was going at the time the assertion is made about (see Visser 1973:2041f
for Old English attestations). A similar case is OHG ward quhoman, literally
became come, meaning that someone was in the process of arriving somewhere
(see Eroms 2000; Kotin 2003 and Kuroda 2010 for the Old High German data).
Thus, in contrast to combinations with the be copula werdan/weoran and a past
participle of an intransitive verb express a process rather than a completed action.
To sum up, combinations of become copula verbs and past participles yield
different results than the be combinations investigated in 3.1 above, though the
two constructions are similar with respect to what causes the passive-like read-
ings, which is the transitivity of the verb supplying the participle. Basically, transi-
tive verbs cause passive-like readings; but the different terminativity values of the
copulas as aspectual operators impact on the readings as follows. With the basi-
cally inchoative werdan/weoran past participle constructions obtain a processual
or future interpretation, whereas be copulas focus on completed states. As hypo-
thesised, the passive readings are caused naturally by the combinations of aspec-
tual, transitivity (and partly temporal) parameters of the participating elements.
4. C
onsiderations about different incipient stages of grammaticalisation
of passive constructions in Late Old English and Old High German
In the previous section, we proposed to treat OE and OHG constructions with

copula verbs and past participles as fully compositional. They convey passive
meanings as a result of combining the aspectual values of the copula verbs with
the aspectual and transitivity values of the past participle. In this section, we
hypothesise that already at the end of the Old High German and Old English peri-
ods we can observe that German and English take different paths of the incipi-
ent grammaticalisation of the copula constructions with past participles towards
periphrastic passive constructions. We argue that the development of the German
and English passive constructions started out from structurally identical sources,
i.e. combinations of the copula verbs be and become with the past participles of
transitive verbs, as it is these combinations in which the (genuine) passive reading
was favoured.
First signs of the incipient grammaticalisation of a different auxiliary in English
and German may be detected already at this early stage in the development. While
German gradually establishes passive constructions with werden, English shows
a clear preference for be. We suggest that the divergent developments of German
and English passive constructions, especially the choice of a different auxiliary, is

mainly due to different (semantic) reinterpretations processes which were trig-
gered in specific contexts of use (possible also genre, as Petr 2010 proposes). For
German the fixation and gradual spread of the past participles of transitive non-
terminative verbs in combination with the copula werdan played a crucial role. As
shown in Section 3, it is these combinations in which the diathetic passive reading
is integrated with aspectually processual or habitual ones i.e. the interpretation
closest to the modern meaning of German Vorgangspassiv. In English the situation
is different, as weoran is only rarely attested with non-terminative transitive verbs
in Old English. Here the functional overlap between the two copula constructions
in the past tense seems to have been more important. Due to the deagentivisation
of the main verb in its past participle form, transitive verbs appear as intransitive;
and the subject of the sentence is seen as the non-agent argument of the participle
verb. Hence, we are dealing here with a genuine passive, i.e. agentless, reading of
both copula constructions.
In combinations of the copula verb (be as well as become) and a past parti-
ciple of a transitive verb the resultative meaning of the past participle is defocused
and the processual meaning of the whole construction is favoured. That is, the
aspectual readings of both variants with be or with become overlap to a
considerable extent.
(10) a. Er wird geqult/geliebt.
He is (being) tortured/loved.
b. Er ist geqult/geliebt.
He is (has been) tortured/loved
c. Er wurde geqult/geliebt.
He was (being) tortured/loved.
d. Er war geqult/geliebt.
He was (had been) tortured/loved
(10b) predicates a state about a subject that is the result of the event described
by the verb. Now, at this part the aktionsart of the verb in the participle has an
effect. For a stative verb like love the semantic overlap of the two constructions
becomes immediately obvious: (10b) is very similar to the non-bracketed version
of (10a), because a processual reading of love is less prototypical. This is presum-
ably the reason why the static passives (Zustandspassiv) in (10b) and (10d) refer
to a time before the reference time, which becomes clear from the bracketed ver-
sion, instead of denoting a state that holds true at the reference time (i.e. where
reference time and event time are simultaneous). For delimitative verbs (activity
verbs in the Vendler-Dowty system), this is slightly different, but the outcome is
similar. Again, (10b) expresses a state that is the result of a previous action denoted
by the verb. The conclusion is that someone is in a state of being tortured, which
is almost processual (reinforced by the dynamic phase in the event structure) and
which overlaps with the dynamic passive reading in (10a). In the past tense there is
a noteworthy difference, which was pointed out in 2.2 above: the past tense causes
the static reading to dominate and hence the overlap is in the static semantics.
Both (10c) and (10d) can be taken to refer to states, unless (10c) is adequately con-
textualised and even then a processual reading for stative verbs is difficult.
This means on the one hand that, in Late Old English and Late Old High
German, there exist two constructions with almost equivalent meanings. (Note
that slight interpretational and frequency differences have been attested in pres-
ent and past temporal perspectives. However further careful investigations of
diachronic data are needed to clarify this point.) The presence of the two largely
synonymous constructions creates a situation of competition: often only one of
the constructions is bound to win. On the other hand, the conditions are such
that either copula verb has its own advantages and thus the potential to win out
(in addition to other potential candidates, such as OE becuman become, see 2.2
above). The aspectual operator be selects the static phase from the event structure
and is neutral with respect to the specific interpretation of the state denoted by
the past participle: it may be interpreted as a completed as well as a constant, con-
tinuing state. The aspectual operator become focuses on the change of state and
favours a gradual attainment of the resultant state expressed by the past participle.
Both interpretations are easily compatible with a genuine passive interpretation.
In Late Old High German15 werdan in constructions with past participles of
transitive verbs becomes gradually more frequent as compared to the construc-
tions with sn/wesan.16 For the earlier periods of Old High German, the clear dom-
inance of sn/wesan-constructions has been documented (see e.g. Schrodt 2004;
Kotin 2003, and Vogel 2006). Fritz (1994), for example, gives the following figures
for Otfrids Evangelienbuch: 98 occurences of werdan-constructions versus 203 of
those with sn/wesan. In Notkers Psalter, Oubouzar (1974) detects 230 construc-
tions with sn/wesan and 226 with werdan. In the fragment of Notkers De conso-
latio philosophiae we found 58 sentences with sn/wesan plus past participle and
. For the Late OHG period, we refer here mainly to the texts by Notker von St. Gallen,
especially to his De Consolatione Philosophiae from ca. 1025, which is part of the diachronic
corpus KALI www.kali.uni-hannover.de (18 May 2011).
. A word of caution is necessary here: throughout the Old High German period, the dis-
tribution of copular construction with be and become plus past participles varies consider-
ably depending on the different temporal perspective. The figures found in the literature and
given here do not capture these differences. A closer look is still needed to clarify this point
(see below).
71 of werdan plus past participle. This suggests that already at the end of the OHG
period, the verb werdan starts to compete with or even win over sn/wesan.
For German, it should be also noted that while sn/wesan combines relatively
freely with past participles of transitive and intransitive as well as of terminative
and non-terminative verbs, werdan starts to be predominantly used with past par-
ticiples of transitive (terminative and non-terminative) verbs.
(11) werdan plus pp of a non-terminative transitive verb
Subiectiuum st tz n demo proloquio. fne demo
subjective is this at the introduction about that
eht kesprchen uurdet.
something spoken becomes
In the introduction it is the subjective that is spoken about[Notker, 5069]17
(12) werdan plus pp of a terminative transitive verb
tr s fersget hbet. tz [] beatitudo fnden neuurde.
there they denied have that [] beatitude found neg-become
They denied that beatitude is/will be found [Notker, 17035]
Moreover, the number of past participles of non-terminative transitive verbs used

with werdan (but not with sn/wesan) increases rapidly towards Late Old High
German. This not only suggests the strong tendency towards reinterpretation of
werdan as a default passive auxiliary, but also supports our hypothesis that the
contexts with werdan and past participles of non-terminative transitive verbs
served as a point of departure in the incipient phase of the grammaticalisation
process of werdan plus past participle as a preferred German passive periphrastic
construction (see also Vogel 2006:162f for a similar observation).
In Late Old English, by contrast, there seems to be a split between weoran
used in the present tense and weoran used in the past tense, of which the lat-
ter disappears 200 years earlier than the former (Petr 2010:463). In addition,
as mentioned above (see 1), there is a seeming equivalence between the two
copula (passive-like) constructions in the past tense. Moreover, weoran is very
infrequent combination with delimitative transitive verbs until the 14th century
(Petr& Cuyckens 2009:353). In addition, Old English had a whole range of dif-
ferent copula verbs, among which there was considerable semantic overlap (see
Petr & Cuyckens 2009). This suggests that the incipient grammaticalisation of a
passive construction involves a temporal split: whereas the two passive-like read-
ings have largely collapsed into one passive-construction in the past tense, the
present tense uses of copula and past participle seem to be much less entrenched,
. Here and further below, the figures specify the location in the text according to the
internal convention of the KALI-corpus. The examples can be easily tracked back.
and it is not until the replacement of weoran that we may speak of a fully gram-
maticalised passive in English (as mentioned in 2.2 above, present tense weoran
plus past participle usually has future reference).
Another important observation about the relevant contexts of change con-
cerns the explicit encoding of the agent of the event by means of a prepositional
phrase (cf. ModE by and ModG von). If the agent is explicitly mentioned, this
leads to the transitivisation of the copula construction (see also Petr & Cuyckens
2009:350): instead of encoding intransitive, resultative predication about the
subject of the sentence which is usually a non-agent they start to express a
transitive event encoding an agent and a patient. Thereby the agent is introduced
by the prepositional phrase, and the patient is expressed in the subject position.
Importantly, this (re)interpretation process takes place on the sentence level and
concerns the argument structures of the verb supplying the participle and the
respective copula. The past participle may thus be reanalysed as a main verb of
the sentence together with its argument structure, and the copula may gradually
grammaticalise into a passive auxiliary without an argument structure of its own.
This yields the relevant structural conditions in which the constructions may lose
their compositional status and grammaticalise as passive verbal periphrases.
Indeed, in Notkers text we find that werden plus past participle of transitive
non-terminative verbs (and not of sn/wesan with the same participles) is often
complemented by prepositional phrases, which can be interpreted as introducing
the agent:
(13) nde er fne lgedngen neuulle betrgen uurden
and he by deceptives neg-wants betrayed become
And he does not want to be betrayed by the deceptives [Notker, 11883]
Similarly, Old English can express the agent of the sentence with a preposi-
tional phrase, though different prepositions are used, of of , fram from and urh
through (see Petr & Cuyckens 2009:351 for examples). But one probably can-
not speak of a grammaticalised way of expressing the agent of a corresponding
active sentence (see also Petr & Cuyckens 2009:351); in particular the fact that
the agent can be expressed with all copula constructions, including those that
feature a property adjective as subject complement, supports this conclusion. (14),
taken from Petr & Cuyckens (2009:351), illustrates this.
(14) & he a lrde his apostolas, him sgde urh hwt
and he then taught his apostles, them told, through what
seo saul eadegust gewurde.
the soul happiest became.
And then he taught his apostles, he told them, how the soul would become
most blessed. [c. 971. LS 20 (AssumptMor[BlHom 13]: 159394)]
To sum up, combinations of the copula verbs be and become with past parti-
ciples of transitive verbs provide the main relevant contexts in which the passive
meaning of the whole periphrasis most probably became established and conven-
tionalised. In addition, the explicit reference to the agent of the verbal action by
means of prepositional phrases might have influenced the transitivisation of the
whole periphrasis. These relevant contexts can be traced back to the Late Old High
German and Late Old English periods, and serve as first attestations of the incipi-
ent grammaticalisation process. The only difference between English and German
is the different copula involved in the genesis of the new passive.
5. Conclusion and further research
Concluding, we would like to sketch pathways for further corpus research of

English and German passive constructions and their diachrony, building on the
theoretical considerations presented in this paper. As we have shown, and as the
data confirms, Old English as well as Old High German did not possess grammati-
calised passive periphrases. Constructions with the copula verbs be and become
plus past participles served as fully compositional structures with primary aspec-
tual meaning whose passive readings resulted from the logical combination of the
aspectual and transitivity values of their composite parts. But there is already indi-
cation of a beginning process of grammaticalisation in Late Old English and Late
Old High German.
Starting from the assumption that, approximately from the Late Old High
German and Late Old English periods onwards, specific relevant contexts of use
may be detected in which the respective copula verb (i.e. be in English, become
in German) is gradually grammaticalised as a basic passive auxiliary, a diachronic
corpus study aiming at the investigation of these constructions should concentrate
on the following points. Particularly important, in our view, is the data from the
Middle High German and Middle English periods.
First, the investigation of diachronic data of the so-called passive construc-
tions should integrate the examination of other constructions with past participles
(and other copula verbs, see 1.) as well as other constructions with the same cop-
ula verbs (e.g. with present participles, infinitives, and adjectives).
Second, the study ought to pay special attention to the aspect and aktion-
sart values of the verbs supplying the past participles. As pointed out in 4., Late
Old High German seems to favour non-terminative transitive verbs with werdan,
while Late Old English does not seem to prefer one aspectual value.
Third, temporal values of the copula verbs and frequently used temporal
expressions in the contexts of the passive constructions should be taken into
consideration. As mentioned in 4., the semantic interpretation as well as the fre-

quency distributions of constructions with be and become plus past participles
differ depending on the past and present temporal perspective, both in Old High
German and Old English. This fact might have influenced the grammaticalisation
process of these re-analysed constructions.
Furthermore, the modal semantics of the copula verbs, the transitivity values
of the main verbs, and word formation patterns of verbs supplying the past parti-
ciples would be of interest.
For the envisioned analysis of the diachronic corpus data from several periods
of German and English, not only the evidence concerning frequency and seman-
tic consistency of the investigated constructions would be collected. Additional
semantic and pragmatic features of these constructions would have to be taken
into account as well: Are there e.g. aspectual or other usages that cannot be moti-
vated in terms of passive formation? Moreover, the interaction effects with other
parameters like e.g. definiteness, information structure, genre, influence of the
Latin original (in the case of translated texts), might provide additional insights
into how the respective grammaticalisation process evolved.
Such a study would not only contribute significantly to our understanding
of the development of passive constructions in English and German but would
also yield new insights into the general areas of categorial change and grammati-
calisation, fine-detailed modelling of language change and overall theory of tense-
aspect-mood and transitivity.
References
Abraham, Werner. 1991. Aktionsartsemantik und Auxiliarisierung im Deutschen. In Neue Fra-

gen der Linguistik. Akten des 25. Linguistischen Kolloquiums, Paderborn 1990. [Bestand und
Entwicklung 1], Elisabeth Feldbusch (ed.), 125133. Tbingen: Niemeyer.
Abraham, Werner. 2000. Perfektpartizip und seine angebliche Passivbedeutung im Deutschen.
Zeitschrift fr Germanistische Linguistik 28(2): 141167.
Bickel, Balthasar. 1996. Aspect, Mood, and Time in Belhare. Ph.D. dissertation, University of
Zrich.
Bickel, Balthasar. 2000. Unlogischer Aspekt: zur Bedeutungsstruktur von Aspekt und Aktion-
sart, besonders im Belharischen. In Probleme der Interaktion von Lexik und Aspekt, Walter
Breu (ed.), 120. Tbingen: Niemeyer.
Bertinetto, Pier Marco & Delfitto, Denis. 2000. Aspect vs. Actionality: Why they should be kept
apart. In Tense and Aspect in the Languages of Europe, sten Dahl (ed.), 190225. Berlin:
Mouton de Gruyter.
Breu, Walter. 1998. Aspektuelle Verbklassen, insbesondere Inchoativa und Inzeptiva. In Slavist-
ische Linguistik 1997, T. Berger & J. Raecke (eds), 5580. Mnchen: Sagner.
Brunner, Karl. 1962. Die englische Sprache. II. Tbingen: Niemeyer.
Brunner, Karl. 1965. Altenglische Grammatik. Tbingen: Niemeyer.

Dal, Ingerid. 1966. Kurze Deutsche Syntax. Auf historischer Grundlage. Tbingen: Niemeyer.
Denison, David. 1993. English Historical Syntax: Verbal Constructions (Longman Linguistics
Library). London and New York: Longman.
Dowty, David. 1972. Studies in the Logic of Verb Aspect and Time Reference in English. Ph.D.
dissertation, University of Texas at Austin.
Eroms, Hans-Werner. 2000. Einfache und expandierte Verbformen im frhen Deutsch. In Aspe-
kte der Verbalgrammatik, Ludwig M. Eichinger & Oddleif Leirbukt (eds), 934. Hildesheim:
Olms.
Estival, Dominique. 1986. A diachronic study of the English passive. Diachronica 6: 2354.
Fischer, Olga & van der Wurff, Wim. 2006. Syntax. In A History of the English Language, Richard
M. Hogg & David Denison (eds), 109198. Cambridge: CUP.
Fortson, Benjamin W. 2010. Indo-European Language and Culture: An Introduction, 2nd ed.
Malden MA: Blackwell.
Frary, Louise G. 1929. Studies in the syntax of the OE passive with special reference to the use of
wesan and weoran (Language Disserations, no. 5). Linguistic Society of America.
Fritz, Thomas. 1994. Passivformen in Otfrids Evangelienbuch: Tempus, Aspekt, Aktionsart.
Sprachwissenschaft 19(2): 165182.
Jones, Howard. 2009. Aktionsart in the Old High German Passive with special reference to the
Tatian and Isidor translation. Hamburg: Buske.
Klingenbiel, Josef. 1937. Die Passivumschreibungen im Altenglischen. Bottrop: Postberg.
Kotin, Michail. 2003. Die werden-Perspektive und die werden-Periphrasen im Deutschen. Histo-
rische Entwicklung und Funktionen in der Gegenwartssprache. Frankfurt: Peter Lang.
Kuroda, Susumu. 2010. Das althochdeutsche Verbalsystem aus textlinguistischer Perspektive.
Paper read at the Internationalen Vereinigung fr Germanistik. Warsaw, 30 July 7 August
2010.
Leiss, Elisabeth. 1992. Die Verbalkategorien des Deutschen. Ein Beitrag zur Theorie der sprachli-
chen Kategorisierung. Berlin: Walter de Gruyter.
Lieber, Rochelle. 1979. The English passive: An argument for historical rule stability. Linguistic
Inquiry 10: 667688.
Lookwood, William B. 1968. Historical German Syntax. Oxford: Clarendon Press.
Maienborn, Claudia. 2007, Das Zustandspassiv. Grammatische Einordnung Bildungsbe-
schrnkung Interpretationsspielraum. Zeitschrift fr germanistische Linguistik 35: 83114.
Mailhammer, Robert. 2009. Towards an aspect-based analysis of the verb categories of Amur-
dak. Australian Journal of Linguistics 29(3): 349-391.
Oubouzar, Erika. 1974. ber die Ausbildung der zusammengesetzten Verbformen im deutschen
Verbalsystem. Beitrge zur Geschichte der deutschen Sprache und Literatur 95: 596.
Petr, Peter. 2010. The function of weoran and its loss in Old and Middle English. English Lan-
guage and Linguistics 14: 457484.
Petr, Peter & Cuyckens, Hubert. 2009. Constructional change in Old and Middle English copu-
lar constructions and its impact on the lexicon. Folia Linguistica Historica 30: 311366.
Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York NY: Macmillan.
Rix, Helmut. 1976. Historische Grammatik des Griechischen. Darmstadt: Wissenschaftliche
Buchgesellschaft.
Russom, Jaqueline Haring. 1982. An examination of the evidence for OE indirect passives. Lin-
guistic Inquiry 14: 677680.
Sasse, Hans-Jrgen. 1991. Aspect and Aktionsart: A reconciliation. In Perspectives on Aspect and
Aktionsart, Carl Vetters & Willy Vanderweghe (eds), 3145. Special issue of Belgian Journal
of Linguistics 6.
Schrodt, Richard. 2004. Althochdeutsche Grammatik, II, Syntax. Tbingen: Niemeyer.
Schrder, Werner. 1955. Zur Passivbildung im Althochdeutschen. Beitrge zur Geschichte der
deutschen Sprache und Literatur 77: 176.
Sonnenhauser, Barbara. 2006. Yet Theres Method in it. Semantics, Pragmatics, and the Interpreta-
tion of the Russian Imperfective Aspect. Mnchen: Sagner.
Vendler, Zeno. 1967. Linguistics and Philosophy. Ithaca NY: Cornell University Press.
Vennemann, Theo. 2010. Contact and prehistory: The Indo-European Northwest. In The Hand-
book of Language Contact, Raymond Hickey (ed.), 380405. Malden MA: Wiley-Blackwell.
Vezzosi, Letitia.1998. Passive constructions. The case of Old English. Folia Linguistica Historica
29(1): 5364.
Visser, Fredericus T. 1973. An Historical Syntax of the English Language, III.2, Leiden: Brill.
Vogel, Petra M. 2006. Das unpersnliche Passiv. Eine funktionale Untersuchung unter besonderer
Bercksichtigung des Deutschen und seiner historischen Entwicklung. Berlin: Walter de
Gruyter.
Passive auxiliaries in English and German
Decline versus grammaticalisation of bounded
language use*
Peter Petr
KU Leuven, Belgium
The passive construction constitutes a marked difference between English, which

uses the auxiliary be, and German, which uses werden become. Originally,
however, both languages used both verbs. In this paper I argue, based on evidence
from Old English, early Middle English, Tatian and Otfrid, that this situation
changed when English and German developed different systems of boundedness.
Bounded language use construes situations as completed sub-events, emphasizing
narrative progress, and making abundant use of time adverbials, which split up
an event chronologically and often take up the first position in a verb-second
system. In German when this type of bounded language use was grammaticalised,
werden grammaticalised as the only passive auxiliary, precisely because it was
already predominantly used in bounded clauses. By contrast, the bounded system
disappeared in English, as evidenced in the heavy decrease of time adverbials of
narrative progress such as a then, and the confusion of verb-second-syntax.
Weoran, the Old English cognate of werden, was highly entrenched in these
constructions, and disappeared with them. In general, my analysis shows how
the bounded-unbounded distinction makes it possible to account for a major
difference in the auxiliary system between English and German.
1. Introduction
The passive construction constitutes a marked difference between English, which

uses the auxiliary be, and German, which uses werden become. Zieglschmid
(1929, 1931), however, showed that originally both languages used both verbs.
In this paper, I argue that English lost weoran (the cognate of werden) when it
* This paper applies a hypothesis tested in a previous paper on Old English weoran vs. wesan
in general (Petr 2010b) to a contrastive analysis of these two verbs as used in the Old English
and Old High German Passive Construction. I would like to thank Ilse Wischer, Christopher
Shank and an anonymous reviewer for their helpful comments on draft versions of the paper.
Peter Petr
abandoned the bounded construal of narratives, inherited from Germanic, while

German further grammaticalised this system, with the further grammaticalisation
of werden as a consequence.
The original variation, and its loss, is illustrated in (1) and (2). In (1), Old
High German opts for werdan, while Old English has wesan. In (2) it is the other
way around, even though the context is almost identical. In a fifteenth-century
Middle High German bible printed by Mentel, both (1) and (2) now have a form of
werden, which at that time had become the only form available in German.1 While
wesan continues to exist in German, its position in the Passive Construction is
marginalised. By contrast, the Middle English Wycliffe bible uses a form of be
twice. Indeed, by the end of the fourteenth century weoran though highly fre-
quent in Old English had all but disappeared (see Petr [2010a: 89] for a detailed
frequency overview).
(1) a. Latin vulgate
Et repleti sunt timore. (Lk 5: 26)
b. Old High German
Inti vvurdun gifulte forhtu.(c830. Tatian: 54.9)
c. Middle High German
Vnd sy wurden drfullt mit vorcht.(c1466. Mentelinbibel)
d. Old English
& hig [] wron mid ege gefyllede.(c1025. Lk (WSCp))
e. Middle English
And thei weren fulfillid with greet drede.
((c1384). WBible(1) [Dc 369(2)])
And they were filled with fear.
(2) a. Latin vulgate

Ipse autem repleti sunt insipientia et conloquebantur ad invicem,
quidnam facerent Ihesu. (Lk 6: 11)
b. Old High German
Sie th gifulte uuarun nuuistuome inti sprahhun untar zuisgen, uaz
sie tatin themo heilante.(c830. Tatian: 69.7)
c. Middle High German
Wann sy wurden drfullt mit vnweysheit: vnd sy retten z einander: was
dings sy teten ihesus.(c1466. Mentelinbibel)
. Note that my analysis is restricted to the use of weoran/werdan in the Passive Construc-
tion, with some references to its use as a copula as well. German werden as the auxiliary of the
future is not discussed.
d. Old English
a wurdon hig mid unwisdome gefyllede & spcon betux him hwt
higam Hlende dydon.(c1025. Lk (WSCp))
e. Middle English
Sothli thei weren fulfilld with vnwysdom, and spaken to gidere, what
theischulden do of Jhesu.((c1384). WBible(1) [Dc 369(2)])
But they were filled with fury and discussed with one another what
they might do to Jesus.
In this paper, I argue that Germanic was a moderately bounded language. Bounded
languages construe situations as completed sub-events, emphasizing narrative
progress, and make abundant use of time adverbials (Carroll, von Stutterheim &
Nuese 2004), which split up an event chronologically and often take up the first
position in a verb-second system.
In German this bounded system became further grammaticalised through the
fixation of the verb-second system. The concomitant grammaticalisation of werden
is explained by its bounding change-of-state semantics, which denote completed
events. In English the system disappears, as shown by the heavy decrease of time
adverbials of narrative progress (a then in Old English (2d) is gone in Middle
English (2e); van Kemenade & Los 2006) and the confusion of verb-second-syntax
(Los 2009). Weoran, being highly entrenched in these constructions, therefore
disappears with them.
The concrete structure of this paper is as follows. Section 2 gives an overview
of previous studies on the development of the Passive Construction in English
and German, and argues that all of them have trouble accounting for the eventual
success of either verb. Section 3 suggests that the solution to this problem does
not lie in an account based on competition in an area of functional overlap, but,
on the contrary, it lies in closely analysing the prototypical uses of both alterna-
tives, and their different constructional environments. It will be argued that such
a close analysis may be based on the general observation that Present-Day English
and New High German differ with respect to the grammatical status of bounded-
ness; a concept in which it will be argued that weoran and werdan are closely
associated. Section 4 provides some evidence for the key role that boundedness
plays in both Old English and Old High German. Section 5 presents an analysis
of the association between boundedness and weoran/werdan. The initial situa-
tion will be analysed through a comparison of Tatian with a representative sample
of Old English texts Tatian is the only Old High German text that still shows
a clear continuation of the Germanic distribution of sein and werden as passive
auxiliaries. The different directions the two languages take at later stages is then
examined by a comparison of Otfrid, in which the earlier system has already been
Peter Petr
abandoned, with a Middle English data sample. From this analysis it will become
clear that both weoran and werdan are strongly associated with bounded lan-
guage use, which leads to the loss of the former and a strengthened position of the
latter. I will end with a brief conclusion (Section 6), in which I draw attention to
the crucial role of the constructional environment in determining the outcome of
a process of competition.
2. Previous studies
The diachrony of the Germanic Passive Construction and the distribution of its
various auxiliary verbs has already been studied extensively. Section 2.1 gives an
overview of existing work that focuses on Old and Middle English. Section 2.2
outlines some studies of Old and Middle High German.
2.1 Old English

A wealth of studies exists that are devoted to wesan and weoran in the pas-
sive construction, as well as to their differences and development (Frary 1929;
Zieglschmidt 1930; Kurtz 1931; Klingebiel 1937; Mitchell 1985; Kilpi 1989; Green
2009; Mller 2009). Early on, Frary (1929) and Kurtz (1931) argued that the two
verbs are basically in complementary distribution: wesan was used for statal pas-
sives and expressed states resulting from previous actions with present relevance
(resultative), as in (3), or with past relevance (pluperfect); weoran was used for
actional passives, focusing on the event itself, as in (4). In their view the functions
of both verbs do not overlap at all.
(3) Hi urnon on fnunge ut of issere byrig mid am

They ran in evening out of this city at time-point
e a burhgata belocene wurdon.
that the city-gates closed became
They ran out of this city in the evening, right when the city gates were

[= change of state] closed.(c1025. Josh: 2.5)
(4) Se hlend become into his apostolum. & wron

The saviour become:sbjv into his apostles & were
eahhwere a dura belocene.
yet the doors closed
The Saviour got by his apostles, and yet the doors were closed
[= no change of state]. (a1020(c995). CHom I, 16: 308.27)
Because of the lack of competition between the verbs, these studies fail to give
a language-internal account of the loss of weoran i.e. one that is not limited
to interference by contact with another language, but maps the mechanisms
of change taking place within the language itself. Instead they are forced to
appeal to external influences from either Scandinavian or Latin. Zieglschmidt
(1929:1326), however, points out that wesan and beon (perhaps not so much
is, see Petr & Cuyckens 2009) were used in actional passives in all the older
Germanic languages. For this reason, Klingebiel (1937) concludes that the uses
of wesan/beon and weoran overlap and that these verbs are involved in some
kind of language-internal competition in Old English.2 Mitchell (1985:324) goes
even further and claims that the two were in free variation, as is illustrated by(5),
where they both express an actional passive with seemingly no difference in con-
text whatsoever.
(5) (Annal 633) Her wear Eadwine cing ofslagen []

here got Edwin king slain
(Annal 642) Her was Oswald ofslagen Norhymbra cing.
here was Oswald slain Northumbrians king
In 633 king Edwin was slain. [] In 642 Oswald, king of the
Northumbrians, was slain.(c1107. ChronF: 633 & 642)
Under this assumption, a straightforward explanation for the loss of weoran is

that it was gradually ousted by the far more frequent wesan. This is characterised
by Wattie, who calls the redundant presence of weoran the only false start in
the Old English tense system (Wattie 1930:143). More recently, however, Kilpi
(1989:85) has shown that their existence, side-by-side, is not entirely redundant
by pointing out the preference of weoran for sudden changes, which are often
negatively connoted (like dying, getting slain, angry etc.), or important actions, as
pointed out by Mller (2009). Mller (2009) revived the idea of loss through com-
petition, arguing that weoran in late Old English extended to durative contexts
and in this way encroached upon the domain of wesan. Durativity is argued to be
clearly present when a durative adverb accompanies the Passive Construction. An
example with wesan is (6), one with weoran is (7) (an example that first appeared
in Denison 1993:419).
. Still, Klingebiel does not go beyond the language-external approach either the only
difference to his predecessors is that he opts for French as the major player (a view recently
revived by Green 2009).
Peter Petr
A r he lg, he [] up to heofenum locade, yder his modgeanc a

(6)
gesetedws.
Always where he lay, he [] looked up to heaven, to which place
his mind was always oriented.
(c1000(c971). LS 17.1 (MartinMor(BlHom 17)): 227.288)
(7) Hi wurdon a utan ymbsette mid romaniscum here swa
they were then outside besieged with roman army so
lange t r fela usenda mid hunger
long that there many thousand:gen.pl with hunger
wurdon acwealde
were killed
They were then from outside besieged by the Roman army so long that
many thousands were killed by hunger there.
(a1020(c995). CHom I, 28: 411.39)
According to Mller, weoran was on the verge of extending its scope to durative
contexts by late Old English. This extension brought about the bleaching of its
prototypical semantics of sudden change to such an extent that it lost its reason of
existence vis--vis the much more frequent wesan. The evidence for a claim such
as this is, however, very thin. In Petr (2010b) it is shown that durative uses with
weoran are not restricted to late Old English, but appear from the earliest texts
onwards. More importantly, the phenomenon remains exceedingly rare through-
out Old and early Middle English, with only seven clear instances out of 1334, or
0.5%, in LEON-alfa, the corpus used for this study (for details see Section 5.2). In
sum, it seems very unlikely that a peripheral use such as the durative one could
have such a great effect on weoran as a whole.
2.2 Old High German

In German there are also numerous studies that discuss the changing distribu-
tion of werdan and wesan in the Passive Construction (Lussky 1924; Zieglschmidt
1929; Fritz 1994; Kotin 2000; Jones 2009, and references therein). Generally these
studies are limited to Old High German, because after this period werdan had
become the default auxiliary in the German Passive Construction.3 Traditionally,
and similar to what has been claimed for Old English by Frary (1929), it has been
maintained that werdan was used for the actional passive, and wesan for the stative
. There is a long-standing debate on the existence of a Passive Construction with sein in

New High German, but it is generally acknowledged that this construction, if it exists at all, is
relatively marginal (see e.g. Maienborn 2007).
passive. However, for Old High German too, this claim has been abandoned since
Zieglschmidt (1929). Indeed, examples of apparently unmotivated variation in
actional passives such as the one in (8) show that such a claim is untenable.
(8) 2. Et terra mota est, et petre sciss sunt, et monumenta aperta sunt.
Inti erda giruorit uuas, inti steina gislizane uuarun, inti
and earth moved was and rocks split were and
grebir uurdun giofanotu.

tombs were/got opened

And the earth was shaken, and rocks were split, and tombs were opened.
((c830). Tatian: 5152: cf. Mt 27:51)
Only recently have extensive quantitative studies been carried out which take as
a starting point the possibility of wesan to express actional passives. Of these, the
most thorough is Jones (2009), who extensively examines the use of the two auxil-
iaries in Tatian and Isidor, and, less extensively so, in a number of other Old High
German works, including Otfrid. With respect to the past tense, Jones looks at two
developments on the way in Old High German (2009: x): (i) uuerdan auxiliarisa-
tion, i.e. the extension of werdan to stative passives (or those involving duration).
Werdan ceases to be a lexical element denoting change of state in a transparent
combination of verb and participle, and becomes part of a no longer fully compo-
sitional passive construction, in which the change-of-state semantics is lost and
only the passivizing function remains; (ii) copular alignment, i.e.the restriction
of wesan to stative passives. Originally, the combination of wesan with participle
could result either in an actional or a stative passive, depending on the nature of the
participle and the context. The absence of wesan in actional passives in some texts
suggests a process of semantic alignment of passive wesan with its stative copu-
lar use. According to Jones results, Tatian shows neither of these developments,
and wesan and werdan overlap in that they are both used in actional passives, a
situation identical to that of Old English, and one likely inherited from G ermanic
(2009:237). Isidor is argued to be ahead of its time, in that it already largely restricts
wesan to stative passives, a situation that is present even more strongly in Otfrid. At
this stage, werdan has not yet become a general passive auxiliary. The restriction of
wesan to stative passives may have had the effect that participles co-occurring with
wesan were considered adjectives unconnected to their verbal source, and this may
have triggered the subsequent extension of w erdan to stative passives.
Generally, while Jones study and those of his predecessors contain valuable
information on the distribution of wesan and werdan in Passive Constructions,
these studies remain largely descriptive, and none of them attempt a language-
internal explanation that can account for the success of werdan and the loss of
wesan in the Passive.
Peter Petr
3. A different approach: Constructional environments
Most of the studies discussed in the previous section argue that (passive) weoran
was lost due to its (existing or developing) overlap and subsequent competi-
tion with the far more frequent is/beon/wesan. However, the way in which these
studies use the notion of competition as an explanation is not very convincing.
While overlap between function words is found everywhere, the prototypical
semantics of two competing items may still remain clearly distinct (Geeraerts
2000:889), and this may help explain their existence side-by-side. This assump-
tion forms the point of departure for my own approach. The core uses of each
of the passive auxiliaries, as will be seen, each time significantly co-occur with a
specific set of clausal structures or constructions, which I refer to as their respec-
tive constructional environments. Note that these constructional environ-
ments are not restricted to the Passive Construction. In Petr (2010a, b) I have
shown that the loss of weoran occurred more or less simultaneously in all its
uses, and that all of these were strongly associated with a single constructional
environment. I will refer to this environment, which was very similar for Old
English and Old High German, as the bounded environment. Conversely, wesan
was not entrenched in this kind of environment. The correlation of the two pas-
sive auxiliaries to two different constructional environments, then, entails that
their core uses will stick to a different set of particular constructions, like cogni-
tive chunks. The combined usage of a particular auxiliary with a particular envi-
ronment is equivalent to what Langacker calls entrenched structures (1987:59)
(a simple example of this phenomenon is an idiomatic expression). When the
larger constructional environment changes, this will naturally also affect the
passive auxiliaries themselves.
The bounded environment of Old English and Old High German is mani-
fested most clearly in narratives, i.e. story-telling. The reason is that boundedness
is closely related to the (chrono)logical ordering of events, and this will be fur-
ther explained in Section 4. The narration of a series of events, now, is frequently
done in the indicative past tense. This association between past tense and narra-
tive action has been shown to be statistically significant (Biber 1991:108). Present
tense is only rarely used for this purpose, and it is used mainly in genres such as
instruction or exposition. It is only to be expected that, when the communica-
tive goals of past and present tense differ so widely, the mechanisms that have an
impact on their use will differ as well. This hypothesis is also supported by the fact
that weoran disappears in the present tense only about two centuries later than
in the past tense (Petr 2010b: 463464). For this reason, my analysis is restricted
to the indicative past tense. To make reference to the past tense clear, I refer to the
verbs at issue as follows: wear for any indicative past tense form of Old English
weoran; ws for Old English wesan; ward4 for Old High German werdan;
and was for Old High German wesan.
4. Bounded and unbounded language use
4.1 Status in Present Day English and New High German

An important distinction related to the narrative genre, which is highly relevant
for explaining the opposite developments seen in German and English, is that
found between bounded and unbounded language use. Basically, bounded lan-
guage use construes situations with the inclusion of their goal or endpoint, and
often serves to mark progress in a narrative (e.g. then he walked over to the other
side). By contrast, unbounded language use construes situations as open-ended,
often by means of progressive aspect (e.g. he was walking about) (Declerck 2007).
In recent psycholinguistic studies (Carroll & von Stutterheim 2003; Carroll &
Lambert 2003; Carroll, von Stutterheim & Nuese 2004), it is argued that both
types of use are not freely available to our minds independently of the language
we speak. Grammatical form is not viewed as a separate system which is inde-
pendent of meaning, but as one which incorporates a system of meanings which
is treated in a given language as prominent in the conceptualisation of states of
affairs (Carroll et al. 2004:185). Cross-linguistic diversity consists less in what is
possible to specify than in the relative ease with which meanings can be specified.
Depending on the availability of certain grammaticalised constructions, some lan-
guages, like German and Dutch, show a strong preference for bounded construal
of events, while other languages such as Present Day English or Arabic more easily
make use of unbounded construal.
For example, speakers of these languages, when asked to describe a narrative
sequence (when watching, for instance, a short animation film), tend to construe
this sequence very differently. For German, this means that its speakers tend to
divide the narrative action into a sequence of temporal segments, each requir-
ing an explicit temporal marker, like Auf einmal, dann in (9a) below. Temporal
anchors like these serve to set the topic time (or temporal topic). Topic time is
the time span about which an assertion is made (Klein 1994:3). By defining topic
time, these temporal adverbials co-bound (together with markers of perfectivity
and/or goals/endpoints) the event described in the clause containing them: the
. In New High German the third person past singular is wurde, but this as a later analogical
form.
Peter Petr
action is construed as reaching its endpoint or goal within the time span defined
by the topic time. These time adverbials also often provide a link to the preceding
clause, and as such typically fill the first slot in their own clause. At the same time
they also create a time after the bounded event and with this the conditions of
opening up a new interval on the time line (temporal shift). The effect of temporal
shift is that a sequence in strict terms is created, in which each situation is com-
pleted before the next one begins.
(9) (a) Shift of topic time
Auf einmal hrt der Lehmmann Wasser tropfen
(On sudden hears the clay-man water drop)
und dann grbt er nach dem Wasser
(and then digs he after the water)
bis der Sand dann unter ihm nachgibt
(until the sand then under him away-gives)
(b) Maintenance of topic time
The man is hearing the sound of dripping water
and he is digging for the water
and the sand is caving in under him
(von Stutterheim 2002:25)
In the narrative sequence given in (9a), the first event is bounded by auf einmal
suddenly, which sets a brief interruption of an unexpressed ongoing situation as
the topic time of the event hear the dripping of water. The second one is bounded
by dann then, which sets as topic time the time span starting after the hearing-
event and ending with the giving way of the sand. Following these time adverbials,
the finite verb remains in second position (so-called verb-second syntax), and the
subject usually follows this finite verb (inversion). This subject defines a second
topic, which is usually the protagonist that remains constant throughout the nar-
rative action. The perspective taken in bounded language use can be compared to
a camera looking through the eyes of the protagonist, who experiences a narrative
action as a series of bounded (complete) events.
By contrast, Present Day English makes abundant use of unbounded con-
strual in describing narrative sequences, as is illustrated in (9b) above. Besides
sequences such as those of (9b), Present Day English preserves the possibility of
construing an event as completed or bounded, depending on the viewpoint of
the speaker (see Smith 1997:9294, Carroll & von Stutterheim 2003:378). For
instance, bounded language use remains the normal way of construing first person
narratives (where by default the speaker is a participant of what happens). How-
ever, Carroll, von Stutterheim and Nuese (2004) show that the syntax and gram-
maticalised constructions of Present Day English encourage unbounded language
use. Unbounded language use typically makes use of a rigid subject-initial syntax,
and the subject is the only structural topic available. It is usually identified with the
most agentive participant (a natural topic) in the event expressed in each clause,
which is not necessarily always the protagonist. There is no structurally required
slot for defining topic time, and the events that are conceptualised are anchored
to a single point in time, which is maintained throughout the event. Each event is
described in unbounded terms by means of progressive aspect (hearing, digging,
caving in). Topic time implicitly remains the same throughout, and the time span
covered by the events is either simultaneous with topic time or is included in it.
In the example given in (9b), topic time is not conceptualised at all, but rather
it is a prolonged now. All the events described are construed as being included
in this now, and this is achieved by the use of the progressive, which denotes
events that are ongoing in such a now. The perspective taken in unbounded lan-
guage use, then, can be compared to filming from a birds eye view (Carroll, von
Stutterheim& Nuese 2004:190).
In sum, the differences in present tense descriptions are the following. Speak-
ers of Present Day English opt for a progressive form in their descriptions, linking
them to an implicit topic time. Speakers of German usually construe a narrative
action as a series of bounded, perfectively construed events, and they may alto-
gether lack an expression for the progressive. Instead, they prefer anchoring in
time (and space), which is normally realised through adverbs like dann filling the
first slot of the clause. Past tense descriptions less easily give in to unbounded
construal strategies, because bounded construal is probably more accessible as a
consequence of the completedness of the events in reality. Accordingly, G erman
need not change its strategies and behaves exactly the same in the past tense.
Present Day English, however, makes use of a hybrid system in the retelling of
an event: bounded construal is still fairly common, but unbounded strategies
regularly creep in, for instance by making use of ingressive constructions (start
Ving) or switching to unbounded progressives in the present tense (Carroll, von
Stutterheim & Nuese 2004:204211). A preference for unbounded construal in
real-time descriptions therefore also correlates to syntactic strategies in retelling
past events that are different from default bounded construal.
4.2 The bounded system of Old English and its breakdown

Old English was much like Modern German in a number of properties. First, Old
English lacked a grammaticalised progressive construction. While Old English
already possessed the be + Vende construction, the predecessor of the Present-
Day English be + Ving construction, only in exceptions did it express progressive-
ness (Killie 2008). Secondly, Old English grammar was fine-tuned for the bounded
construal of events. This is evidenced in its word order rules and its stock of time
Peter Petr
adverbials to establish topic time. The most grammaticalised of these adverbials is

a then (particular point in time), which (similarly to German dann and Dutch
toen) chops up a narrative into temporal segments, foregrounding actions within
that narrative (it has therefore been called an action marker by Enkvist 1986). As
in German, time adverbs were often put in the first position in main clauses, the
verb being in second position (verb-second syntax) and the subject inverted. In
Old English though, most time adverbials only triggered inversion if the subject
was not a pronoun. The exceptions were a and onne (which however mainly
had non-narrative functions). Whenever they occurred on their own in first posi-
tion, they consistently triggered inversion, even with pronominal subjects (Los
2009:103; Westergaard 2009:74).
The bounded system of Old English is illustrated by the biblical fragment (from
the Prodigal Son) in (10) and note that there are no less than three occurrences
of wear in this fragment, the third one occurring in a Passive Construction.
a fter feawa dagum ealle his ing gegaderude se gingra sunu, & ferde
(10)
wrclice on feorlen rice, & forspilde ar his hta lybbende on his glsan. a
he hig hfde ealle amyrrede a wear mycel hunger on am rice & he wear
wdla. [] a beohte he hine & cw, Eala, hu fela yrlinga on mines
fder huse hlaf genohne habba. [] Ic [] fare to minum fder, & ic secge
him, Eala fder, [] do me swa anne of inum yrlingum. & he aras a &
com to his fder, & a gyt a he ws feorr his fder he hyne geseah & wear
mid mildheortnesse astyrod.
Then after a few days the younger son gathered all his belongings, and
travelled abroad to a far country, and wasted there his possessions living in
his lusts. When he had wasted them all, then a great hunger came over the
country & he became a beggar. [] Then he considered to himself and said:
Why, how many servants in my fathers house have enough bread. []
I[] will go to my father, and I will tell him: hey father, [] take me as one
of your servants. And he arose then and came to his father, and when he
was still far from him, his father saw him and was stirred by mercy.
(c1025. Lk (WSCp): 1320)
In (10), a variety of time adverbials (in bold) serves to establish topic time and
divide the story into segments: a fter feawa dagum then after a few days, a he
hig hfde ealle amyrrede, a when he had wasted them all, then (causing inver-
sion as well), etc.
Present-Day English lost these time adverbials and verb-second syntax that
encouraged bounded construal. Instead it has the progressive, which encourages
unbounded construal. Hence, a transition from one system to the other must have
occurred somewhere in between. Los (2012) suggests that English developed its
preference for unbounded construal in Early Modern English. It is at this time that
SVO word order is established and the progressive function of [be Ving] explodes
in frequency at the cost of its ever more marginal stative and durative functions,
two processes which largely seem to run in parallel (Denison 1993; Killie 2008).
However, there is evidence that the rigid bounded system of Old English had
largely disappeared already by the end of the fourteenth century. First, the trans-
parency between syntax and information structure originally present in the sys-
tem of verb-second starts to break down from 1300 onwards (see van Kemenade &
Westergaard 2012). Second, there is the rapid decrease of the most typical bounding
adverb a and the obligatory inversion co-occurring with it. Already in the early
Middle English of the thirteenth century, a (realised as tho in Middle English) is
significantly less frequent than in late Old English, and in the course of the four-
teenth century its use becomes exceptional (van Kemenade & Los 2006:243244).
In Westergaard (2009:9394), for instance, it is shown how main clauses with ini-
tial a (tho)/onne decrease from about 36% of all main clauses in Old English to
15% in early Middle English and 11.3% in late Middle English.
The breakdown of the bounded system of Old English, and its immediate
impact on wear, can be illustrated by comparing the Old English Bible fragment
in (10) to its Middle English counterpart in (11), as found in the Wyclif Bible.
And not aftir many daies [] the onger sone wente forth in pilgrymage in to
(11)
a fer cuntre; and there he wastide hise goodis in lyuynge lecherously. And aftir
that he hadde endid alle thingis, a strong hungre was maad in that cuntre,
and he bigan to haue nede. [] And he turnede aen to hym silf, and seide,
Hou many hirid men in my fadir hous han plente of looues []. Y schal []
go to my fadir, and Y schal seie to hym, Fadir, [] make me as oon of thin
hirid men. And he roos vp, and cam to his fadir. And whanne he was it afer,
his fadir sai hym, and was stirrid bi mercy.
((c1384). WBible(1) [Dc 369(2)], Lk 15: 1320)
Narration in (11) still mainly proceeds by means of bounded construal, but bound-
ing adverbials signalling topic time have decreased, and an unbounded construc-
tion (he bigan to haue nede he began to have need) has crept in. Importantly,
the language of (11) illustrates that the highly grammaticalised way of constru-
ing bounded events has been lost. Specifically, a then and verb-second syntax
are entirely absent in this late Middle English version and so is wear. Middle
English does not only abandon the bounded system of Old English, it also shows
the beginning of the development of a new mixed system, with the eventual gram-
maticalisation of the progressive in Modern English as its endpoint.5
. A discussion of this new system falls outside the scope of this paper. For a brief overview,
I refer to Petr (2010a:134137).
Peter Petr
4.3 The bounded system of German and its grammaticalisation

While English gave up its bounded system, German grammaticalised it further.
This is evident in the current characteristics of German: obligatory verb-second,
the finite verb being placed after any one but no more than one constituent
in first position; moreover, this constituent is often a temporal or argumentative
adverb linking the preceding sentence to the current one, as is illustrated in (9b).
According to Fleischer, Hinterhlzl & Solf (2008) and Axel (2009), however, the
verb-second system was not yet fully established in Old High German. Based on
a substantial body of word order patterns deviating from the Latin source text
in an extensive number of Old High German texts, both studies argue that in
less-established regions verb-second was still in competition with another, older
system, where the verb was generally in first position, and sentence type was indi-
cated by a number of particles.
First, the system of verb-second was neither as strict nor as extensive as it is
in New High German. Similar to Old English, instances may be found where pro-
nominal forms appear in front of the verb when there is already an adverb such as
a in the prefield as well (2b) is an example in case. In addition, elements trig-
gering inversion at this stage were largely limited to topics or focused constituents.
The extension of verb-second to any kind of constituent is still rare, though there
are occasional instances of elements triggering inversion and which are neither a
topic or in focus, such as the sentence adverb certainly in (12).
(12) endi chiuuisso ist christus in dheru selbun salbidhu chimeinit
and certainly is Christ in this same salve meant
and certainly Christ is meant in this same salve
et utique christus ipsa unctione monstratur(a800. Isidor: 144)
Therefore, already at this stage, German has grammaticalised the verb-second sys-
tem further than English.
Yet this process of grammaticalisation only reached its conclusion in Middle
High German. In Old High German, there are still a number of occurrences of V1,
which come in two types. First, V1 appeared when existential or presentational con-
structions were involved, including idiomatic happen-constructions such as (13).
(13) Uuard th thaz
(it) happened then that
In New High German, most clauses of this type have added es in first position.
This expletive element, which correlates with the further obligatorification of
verb-second in all sentence types (except questions), developed only in Middle
High German. Interestingly, Tatian sometimes uses an early alternative for exple-
tive es, i.e. the temporal adverb th, which is the same word as the th found after
the finite verb in (13). The significance of this becomes clear when we first look
at the second use of V1 in Old High German, in declarative main clauses with a
topical subject such as (14).
(14) & sanatus est puer in illa hora;
uuard tho giheilit ther kneht in thero ziti.
became then healed the servant in that time
His servant was then healed in that hour. (c830. Tatian: 183,7)
In New High German, this type of sentence would be ungrammatical. Already in

Old High German it was less common than the verb-second structure. Signifi-
cantly, the majority of these V1-clauses contains th in post-finite position. Axel
(2009:3536) suggests that th may have belonged to a system of particles indicat-
ing sentence type and served as a kind of narrative-emphatic marker (ahypothesis
very similar to what van Kemenade & Los 2006 argue for Old English). Even more
specifically, Fleischer, Hinterhlzl and Solf (2008) show that in Tatian verb-first is
mostly used at an episode turn, introducing new material in the story. Axel points
out that sentence particles encoding sentence-type, mood or aspects of informa-
tion structure are not really a typical feature of verb-second languages. If th was
a narrative-declarative marker, it was not necessary to have an element in first
position to mark the non-interrogative nature of the clause. The loss of sentence
particles may thus have contributed to the generalisation of verb-second word
order already going on during Old High German. The Old High German verb-first
declaratives with th, then, are probably vestiges of this older system.
The presence of th in V1-clauses and the subsequent loss of this sentence-
type may at first seem at odds with the hypothesis that it is an important bounding
adverbial in bounded construal. However, my own data show that th was also
common in pre-finite position, as in (15).
(15) Et repleti sunt omnes in sinagoga ira hc audientes.
Th vvurdun sie gifulte alle in theru samanungu
then were they filled all in the synagogue
gibuluhti thisu horente.
anger:dat this hearing
Then they were all filled with anger, in the synagogue, when they
heardthis.(863x871. Tatian: 78.115.9 (cf. Lk 4: 28))
Similar to Old English, th may have combined its sentence-type defining function
with the bounding function of encoding topic-time. With the further grammati-
calisation of the verb-second system, in which word order defined sentence-type,
the primary function of th became that of indicating topic-time. This in turn may
have enabled a further homogenisation of the verb-second system, involving a
fixed sentence structure in narratives of a temporal or argumentative topic in first
position, the verb in second position, and the subject-topic then following.
Peter Petr
5. C
onvergence and divergence in the development of the English
andGerman passive
5.1 Introduction
In this section I will analyse the available data on the various passive auxiliaries
in Old English and Old High German from the perspective of a possible loss
of the bounded system in Old English versus its further grammaticalisation in
Old High German. To concretely test the association of wear/ward in the
Passive Construction with boundedness, I will chart a number of formal and
semantic variables. The formal variables involve clause type and word order. It
is predicted that wear/ward are used more often in main clauses, where the
advancement of the plot line mainly takes place, and that they are more strongly
associated with XVS-word order, where X is a temporal or argumentative adverb
linking the preceding sentence (chrono)logically to the current one. This second
variable is directly related to the semantic variable, which predicts a high asso-
ciation between wear/ward and certain bounding time adverbs, absent from
ws/was. The analysis will show that originally both wear and ward indeed
showed a preference for declarative main clauses, and strongly correlated with
inverted word order and bounding time adverbs such as Old English a and its
Old High German cognate th. Significantly, the analysis will also show that the
later divergence of English and German is not so much the consequence of a
divergence in the Passive Construction itself indeed, the two auxiliaries con-
tinue to behave in a very similar way in both English and German , but the loss
of wear in English and the further grammaticalisation of ward in G erman
may be explained as the consequence of the loss and further strengthening of
their respective constructional environments, when the bounded system was
lost in English and was grammaticalised further in German.
5.2 Corpora and data

The English data is drawn from the prose material from LEON-alfa (described in
Petr 2010a). Following the makers of the prose corpus of PPCME, I also included
the LEON-alfa sample of the Ormulum, which, even though metrical in form,
is generally much more like prose than like poetry. From these data, an exhaus-
tive sample of wear-passives has been extracted, and a 10% sample of all ws-
passives, randomly selected.
For German I have extracted all passives from two Old High German gospel
harmonies. The first is the translation of Tatians gospel harmony, written c830
and preserved in a contemporary manuscript (see Krotz 2007). The second is the
metrical poem by Otfrid, written between 863 and 871, and preserved in a number
of contemporary manuscripts (see Krotz 2011). The electronic version of the texts
has been taken from the Titus database (Gippert, Javier & Korn 2011). The choice
of Tatian is motivated by the fact that this is the only Old High German text in
which the was-passive is still a common alternative to the ward-passive. While
the translation generally sticks closely to its Latin original, it has been shown in
a number of studies that it is definitely not a gloss (e.g. Fleischer, Hinterhlzl &
Solf 2008; Petrova & Solf 2009:123126). Specifically, it occasionally deviates from
Latin word order or has words that have no equivalent in the original, such as,
significantly, th then. Moreover, if it can be shown that there is a significant dif-
ference in word order preferences between ward and was-passives within the text
itself, the effect of the overall influence from Latin word order is no longer of great
significance. As a comparison, Otfrid has been selected. Being written only about
half a century later, it is not temporally that far removed from Tatian. Yet its idiom
is generally already noticeably more modern than that of Tatian, as discussed in
Jones (2009), whose arguments have been presented in Section 2.2. The choice of
Otfrid is also somewhat problematic because it is a metrical poem but again, it
has been used in syntactic research with interesting results, for example in Wicka
(2009) whose work on cliticisation phenomena argued that in general the struc-
ture of medieval poetry is much closer to spoken language patterns than is usually
the case with contemporary poetry (Wicka 2009:45).
Tables 1 and 2 below give an overview of the frequency of the various types of
passives in the English and German data.
Table 1. Old English data, frequencies per million words

9511050 10511150 11511250
Wear 365 (11%) 791 (14%) 64 (2%)

Ws 2818 (89%) 5066 (86%) 2533 (98%)
[Raw number of tokens] [202] [187] [119]
Table 2. Old High German data, frequencies per million words

Tatian Otfrid
Ward 1579 (38%) 903 (67%)

Was 2625 (62%) 436 (33%)
[Raw number of tokens] [197] [86]
The tables give normalised frequencies per million words, as well as the rela-
tive share of each verb in the totality of passive constructions. While this invites
Peter Petr
direct comparison between periods and languages, caution is encouraged. The fre-
quency peak shown by the English period 10511150 is likely due to differences
in style and genre with the surrounding periods, although what exactly causes the
difference remains unclear. The markedly lower number of passives in Otfrid as
compared to Tatian can be similarly explained. Otfrid is in verse and deviates more
from the Latin original, and this may account for its low number of passives. The
fact that passive ws consistently outnumbers wear in Old English still suggests
that the subsequent development is primarily a matter of frequency. However, the
relative share of wear increased in late Old English before dropping drastically
in Middle English, and this increase is not random, as will be seen below. More-
over, in the Old High German Tatian was is also almost twice as frequent as is
ward, and yet the verb dropped to half the frequency of ward in Otfrid. Generally,
the data seems to suggest that in Germanic *was was the default verb in passive
constructions (or the precursors thereof, assuming that at that stage participles
were still fully adjectival), but that the etymon *war was gaining ground as an
alternative. This process had advanced further in German than in English, where
it presumably came to a halt during the eleventh and twelfth centuries due to some
factor other than frequency.
5.3 Clause type

The first parameter I will look at is that of clause type. Tables 3 and 4 show that
both English wear and German ward originally show a far higher preference
for main clauses than do ws and was. However, their subsequent developments
reveal opposite patterns in both languages. While consistently outnumbered
by ws, wear increased its overall share in main clauses in Old English from
less than a sixth (15%) to a quarter (24%). In Middle English, wear drastically
dropped in frequency, but remained somewhat productive precisely in main
clauses. By contrast, in the Old High German Otfrid ward is already on its way
to being extended to the same contexts as was, and is no longer restricted to any
particular clause type.
Why, then, was this alternative passive auxiliary so successful precisely in main
clauses? The etymology of Germanic *wer is suggestive here (see e.g. Picket et al.
2006:2054). The verb is related to Latin vertere turn, a sudden change of direc-
tion. Hence, its semantics was very suitable to express progress in the plotline that
involved unexpected changes. Relative clauses are usually descriptive in nature,
adding background information to a topical NP, as in They chose his brother Har-
thacnut, who was a Danish citizen. As such, they typically do not denote changes
of state. By contrast, (past tense) main clauses often express events that mark prog-
ress in a narrative (they provide foregrounded information), and therefore will
Table 3. Clause types co-occurring with Old English wear and ws

9511050 10511150 11511250
100%
90%
80%
70%
60%
Relative clause
50% Subordinate clause
Main clause
40%
30%
20%
10%
0%
RD
RD
RD
S
S
EA
EA
EA
W
W
W
W
Table 4. Clause types co-occurring with Old High German ward and was
Tatian Otfrid
100%
90%
80%
70%
60%
Relative clause
50% Subordinate clause
Main clause
40%
30%
20%
10%
0%
RD
RD
S
S
EA
EA
W
W
W
W
often be about changes of state, precisely the kind of semantics shown by *war.
Finally, subordinate clauses are usually temporally anterior to the main clause
(Then Herod found out that he had been deceived). In such cases, their focus is
often on the resulting state of the action rather than the action itself, which may
explain why stative ws/was originally was more frequent in this clause type.
Peter Petr
Around the tenth century, it would seem that Old High German was more
advanced in this development, ward being already more or less as frequent as
was, and increasing rapidly. All the same, Old English seemed to be heading into
a similar direction in the eleventh century, but then the development was checked
when the constructional environment of the passive construction began to change.
As a consequence, wear became more and more restricted to its prototypical
use, which then remained productive the longest. This accounts for the increas-
ing share of main clauses in its distribution, as it is here that its change-of-state
semantics presumably worked best. By contrast, Old High German Otfrid already
shows a more even distribution over clause type between ward and was. This in
turn suggests that in Old High German ward is grammaticalising even further,
with semantic bleaching of its change-of-state semantics as a consequence, as well
as the extension to contexts that used to be the domain of was. I will now turn in
more detail to the various constructional environments found in these declarative
main clauses, where the change presumably started.
5.4 Time adverbs

The preference of Old English wear and Old High German ward for main
clauses points to their foregrounding function, being mainly used in narrative
action. In this section I will further examine main clauses containing wear/
ward vis--vis ws/was in order to test to what extent it is typical for wear/
ward to occur with bounding time adverbials in this clause type, as for instance
in (16)(19). This is done by contrasting the behaviour of wear/ward to that
of ws/was in main clauses. Typical instances of ws/was are (20)(21), which
provide stative background information to the narrative.
Old English
(16) Heo hine freclice bat. a wear heo sona fram
She him heavily beat. Then got she suddenly from
deofle gegripen.
devil seized.
She beat him heavily. Then was/got she suddenly seized by the devil.
(c1025. GD 1 [C]: 4.31.1)
(17) Wear a se munuc micclum afyrht.
got then the monk much frightened
Then the monk got very frightened.(a1020(c995). CHom II, 10: 83.87)
Old High German
(18) Inti sliumo uuard tho giheilit sin rf.
and immediately got then cured his leprosy
And immediately then his leprosy got cured.(c830. Tatian: Mt 8 3)
(19) Tho wurtun sie gidruabte \ zwivalemo muate

then got they startled doubting:dat mind:dat
Then they got troubled with a doubting mind. (863x871. Otfrid: 11.19)
Old English
(20) He ws mid wildeora fellum gegerwed.
he was with wild.animals:gen skins:dat adorned
He was adorned with the skins of wild animals. (c1000. Alex: 35.2)
Old High German
(21) Senu tho uuib thiu habeta geist unmahti hahtuzehen
look then woman who had spirit unstrong eighteen
ir, inti uuas nidargineigit.
years and was down.bent
And look, this woman who had a powerless spirit for eighteen years, and
was bent over.(c830. Tatian: Lk 13 10)
(16) and (19) have clause-initial time adverbials and a subject immediately
following the finite verb, conforming to the boundedness template of Modern
German. In (17) both the time adverb and the subject appear after the finite verb,
which is in first position. (18) has an initial time adverb, but combines this with
a rather uncommon final position of the subject. This variation shows that there
is no (or not yet a) one-to-one-relationship between bounding time adverbs and
word order (initial position + subject following the finite verb). Word order at
this stage of both Old Germanic languages is arguably in a stage of transition,
in which the slot preceding the finite verb is gaining in weight. Indeed, the data
shows that this slot is not yet obligatory, as there are still quite a few instances of
verb-first order in both the Old English period 9511050, as illustrated by (17)
(5.3% of the 130 English passives sampled for the period 9511050 still have this
order) and Tatian (13 out of 115 instances, or 11.3%). Arguably, time adverbials
that occur innon-clause-initial positions also generally have a bounding func-
tion, as in (17)(18), even if it is unclear to what extent they topicalise the time
span for which the assertion holds.
The method I used for comparing differences in association strengths to time
adverbs between wear and ws was that of distinctive collexeme analysis, the
analysis of alternating pairs of constructions and their relative preferences for
words that can (or should be able to) occur in both of them (Gries & S tefanowitsch
2004:101). In this case study, the pair of constructions referred to consists of any
main clause in which wear is the finite verb in the past tense and any main clause
containing ws. Given the limited amount of data, the various time adverbs have
been grouped together into larger semantic categories, functioning as higher-
level collexemes. Using such collexeme categories increases the average frequency
Peter Petr
of each collostruct (i.e. each construction-specific collocational pattern), which

makes significant semantic distinctions easier to spot. A list of categories occurring
in my data, with some representative members, is given in Table 5 (spelling has
been regularised).6 The grammaticalised items a/th then (single occurrence)
and onne/thanne then (iterative/generic, each time when X, then), including
clausal adverbials headed by these items, have been assigned separate categories
(tho and then respectively).
Table 5. List of categories of time adverbials

Old English Old High German
after_x fter (X), sian after X V-ed

again eft
all_of_a_sudden soon (so X V-ed), forriht sum sliumo
e X V-ed, fringa, frlice, rae
always a
ere r, beforan, iu r X V-ed
finally t nextan risten
from_time of re tide (e X V-ed) fon theru ziti
never nfre fon werelti ni
noadverbial [No time adverbial is present]
now nu nu
often oft
on_time on X, at X, in X, ymbe midne mit thiu X V-ed when X V-ed,
dg around midday, the Xth in themo sehsten manude in the
eare the Xth year, her here sixth month, in mitteru naht in
(in chronicle entries) the middle of the night,
other silice temporarily
repeatedly swa oft swa X V-ed
still et yet noh still
then onne thanne
tho a (X V-ed) th (X V-ed)
throughout_period eond a seofon niht throughout
those seven nights
within_x binnan X in fierzug inti in sehs iaron in
46years
. Note that this list is shorter than the one in Petr 2010a, b, because the current data on the
Passive Construction are only a subset of the data used there. Also, the Old High German data
did not yield any new categories.
Tables 69 summarise the results of the distinctive collexeme analysis. Tables6

and 7 give the results of the two first Old English periods. The third period
(11511250) is discussed in summary form below. Tables 8 and 9 give the results
for Tatian and Otfrid respectively.
For Old English, all categories have been included that have a significance of
P < 0.05. Unfortunately, it is not possible with this method to plot the develop-
ment of the constructional strength of a particular time adverb over time, because
results fluctuate greatly with changes in the raw frequencies of both collostructs
for each period examined. The tables can still be compared with regard to their
relative rankings of time adverbials, and this turns out to be quite revealing.
Table 6. Old English, 9511050

Wear P-value Ws P-value
on_time 0.003 NoAdverbial <0.001

tho 0.003
again 0.004
never 0.027
now 0.027
all_of_a_sudden 0.033
(N of Time Adverbials co-occurring with wear = 98; ws = 500)
Table 7. Old English, 10511150

Wear P-value Ws P-value
all_of_a_sudden <0.001 NoAdverbial <0.001

tho <0.001 then 0.045
after_x 0.004 ere 0.047
again 0.004
(N for wear = 116; ws = 330)
The two first Old English periods represented in Tables 6 and 7 are very
similar. In both periods, time adverbials typically used for single foregrounded
events, including the grammatical action marker (Enkvist 1986) a then, are very
strongly associated with wear, while ws typically has no accompanying time
adverbial at all or combines with then, which in Old English refers to habits and is
best translated by each time, or ere, which evokes the sense of a pluperfect, some-
thing that had happened before the main action expressed, and therefore does not
belong to the foreground. The situation is less clear in the last period 11511250,
Peter Petr
where only all_of_a_sudden appears in the list of significant collexemes of

wear (P = 0.025) Yet the overall situation is still quite similar. Instances of tho
are still skewed towards co-occurrence with wear, with a P-value of 0.12, but not
reaching the stricter level of significance of 0.05, yet not too far removed from it
either. The behaviour of ws is harder to fathom. Lack of an adverbial occurs still
more often with ws, but this preference is not significant at all (P = 0.42). These
results can be accounted for, first by the lack of data for wear, which at this stage
had already become fairly uncommon, and secondly, because ws was most likely
already extending its use to the former domain of wear and on its way to becom-
ing the general auxiliary of the passive.
The results for Old High German generally mirror those of Old English.
Unfortunately, due to the limited amount of data for Old High German, the strict
significance level of 0.05 is rarely reached. Therefore I give the results of all catego-
ries that had at least three instances, and are skewed to either verb.
Table 8. Old High German, Tatian

Ward P-value was P-value
on_time 0.110 NoAdverbial <0.001

at_once 0.155
tho 0.185
(N ward = 65; was = 55)
Table 9. Old High German, Otfrid

Ward P-value was P-value
tho 0.035 ere_x 0.159

NoAdverbial 0.324
(N ward = 39; was = 19)
5.5 Word order

Besides their association with certain types of time adverbials, wear and ward
also seem to be associated with inverted word order, as in (16) and (19), here
repeated as (22) and (23).
(22) Heo hine freclice bat. a wear heo sona fram
She him heavily beat. Then got she suddenly from
deofle gegripen.
devil seized.
She beat him heavily. Then was/got she suddenly seized by the devil.
(c1025. GD 1 [C]: 4.31.1)
(23) Tho wurtun sie gidruabte \ zwivalemo muate

then got they startled doubting:dat mind:dat
Then they got troubled with a doubting mind. (863x871. Otfrid: 11.19)
The variable of word order is relevant because verb-second syntax and bound-
edness seem to go hand in hand (as implied in Los 2009:104106, and made
explicit in Los 2012), as an unbounded system is typically subject-initial. While
there are many instances of a/th or other bounding time adverbials that do
not cause inversion, the lexeme causing inversion is very often one of them.
Where the collostructional analysis with various time adverbials pins down the
most conspicuous semantic conspiracy between wear/ward and bounded-
ness, inversion constitutes an interesting additional testing variable. Specifi-
cally, the association between wear/ward and inversion gives insight in the
global effect of the formal side of the bounded system on the different develop-
ments of wear and ward. Inversion does not only occur with time adverbials
in first position, but also with other items such as spatial (frequently r there)
or argumentative (urh t/X through that/X) adverbs or adverbial prepo-
sitional phrases. The similarity in bounding function of temporal and spatial
adverbials is established and further elaborated for German in Carroll and
Lambert (2003:169170). For Old English, their function as discourse-anchors
has been established by Los (2009:104). To summarise briefly, spatial adverbi-
als perform a function similar to those of time adverbials, which is also par-
tially bounding, in that they define the topic space, i.e. the space within whose
boundaries a certain assertion holds. Argumentative prepositional phrases are
less transparently bounding structures, but they arguably limit a clause to a
certain region in argumentative space. Put differently, all the possible fillers of
the first position-slot share a function of structuring information, in anchoring
the clause to the preceding one and, in that way, locating it in time, space or
argumentative space.
Given the relevance of verb-second and concomitant inversion for a bounded
system, a strong association of wear/ward with verb-second would once again
confirm that their opposite developments are the consequence of the opposite
developments of that system in English and German. Table 10 gives the significance
of the different frequencies with which wear and ws co-occur with inverted
subjects in main clauses in Old and early Middle English. Table 11 does the same
for ward and was in Tatian and Otfrid. The test of significance used is the Fisher
Exact Test, with the left-tailed P-values as indicators of significance, which pre-
dict the chance probability that wear/ward would occur even more in inverted
contexts, or ws/was even less in non-inverted ones (Langsrud 2004).
Both English and German show very similar behaviour, except for the over-
all frequency of the two auxiliaries. In terms of word order preference, neither
Peter Petr
Table 10. Word order differences between wear and ws-Passives in main clauses
9511050 10511150 11511250
Wear Ws Wear Ws Wear Ws
No inversion 47 (61%) 230 (60%) 54 (61%) 230 (79%) 6 (40%) 280 (62%)
Inversion 30 (39%) 150 (40%) 34 (39%) 60 (21%) 9 (60%) 170 (38%)
[Excluded] 7 80 5 10 10 0
P-value 0.58 <0.001 0.07
Table 11. Word order differences between ward and was-Passives in main clauses
Tatian Otfrid
Ward Was Ward Was
No inversion 17 (41%) 17 (49%) 16 (43%) 10 (63%)

Inversion 24 (59%) 18 (51%) 21 (57%) 6 (38%)
[Excluded] 21 18 3 4
P-value 0.35 0.16
nglish nor German show significant correlations with any one passive auxiliary
E
in the earliest data. Still, already at this stage inversion is more common with
ward in Tatian. In addition, in English the lack of a significant association with
inversion seems to be restricted to the Passive Construction. Overall, there is
already a strong association between inversion and wear at this stage (Petr
2010b). Whether or not passive wear was influenced at a later stage by copular
and other uses of wear, the verb shows a highly significant association with
inversion in the next period (10511150). This association remains strong up to
the point where the verb wear disappeared along with inversion. For German,
Otfrid suggests that the association between ward and inversion grew stronger
as well, although it does not quite reach the level of significance. In later German
ward remains as the only passive auxiliary. At this stage, the grammatical sys-
tem of German allows the first position to be filled by any one constituent, with
frequent inversion of the subject as a consequence. The general picture is there-
fore similar to that obtained from the distinctive collexeme analysis with time
adverbs discussed in the preceding section. Both wear and ward seem to be
going through the same kind of development, only in English this results in the
loss of wear as a consequence of its constructional environment being lost,
whereas in German the opposite holds.
6. Conclusion
In this article I have shown that there are significant correlations between wear
and ward and bounded language use. Bounded language use here means the set
of constructions cross-linguistically most commonly used to construe narratives
in chronological fashion, with each event marked off from the preceding and fol-
lowing one by means of a topical time adverb and a perfective verb phrase. I have
analysed three types of evidence for these correlations. First, both wear and ward
are preferred in main clauses, though this preference disappears in German when
ward is generalised as the only passive auxiliary. In addition, when appearing in
main clauses, both verbs are strongly associated with bounding time adverbs as well
as with inverted word order, again two hallmarks of bounded construal. Overall,
then, the constructional environments with which both verbs are associated are very
similar in both languages. At the same time, these constructional environments go
through very different developments in the two languages. The Old English system
of boundedness is lost in Middle English, whereas that of Old High German is fur-
ther grammaticalised in Middle High German. The result of these opposite develop-
ments is that wear altogether disappears in English (also in other uses than the
passive), and that ward generalises as the default passive auxiliary.
From a theoretical point of view, the results of this contrastive analysis add to
our understanding of the role of a words surroundings in its history. Specifically,
rather than trying to explain the different developments as a result of different out-
comes of competition in overlapping functions between wear/ward vs. ws/
was, overlap in use has been argued to be of only secondary importance. Instead,
it is precisely the prototypical specificity of both verbs that causes their develop-
ment to be different. In their prototypical uses, the two auxiliaries are tied to two
different environments, like cognitive chunks, and which consist of distinct sets of
constructions and lexemes. Given this association, it is only natural to assume that
the development of these environments is of primary importance for the history of
the two auxiliaries themselves. In other words, the history of function words such
as the passive auxiliaries at issue here apparently depends on the broader gram-
matical system of the language, and cannot be explained in isolation on the basis
of their shared use in Passive Constructions alone.
References
Axel, Katrin. 2009. The verb-second property in Old High German: Different ways of filling the
prefield. In Information Structure and Language Change: New Approaches to Word Order
Variation in Germanic, Roland Hinterhlzl & Svetlana Petrova (eds), 1744. Berlin: Walter
de Gruyter.
Peter Petr
Biber, Douglas. 1991. Variation across Speech and Writing. Cambridge: CUP.
Carroll, Mary & Lambert, Monique. 2003. Information Structure in narratives and the role of
grammaticised knowledge: A study of adult French and German learners of English. In
Information Structure and the Dynamics of Language Acquisition [Studies in Bilingualism
26], Christine Dimroth & Marianne Starren (eds), 267287. Amsterdam: John Benjamins.
Carroll, Mary & von Stutterheim, Christiane. 2003. Typology and information organisation:
Perspective taking and language-specific effects in the construal of events. In Typology and
Second Language Acquisition, Anna Ramat (ed.), 365402. Berlin: Mouton de Gruyter.
Carroll, Mary, von Stutterheim, Christiane & Nuese, Ralf. 2004. The language and thought
debate: A psycholinguistic approach. In Multidisciplinary approaches to language pro-
duction [Trends in Linguistics: Studies and Monographs 157], Thomas Pechmann &
Christopher Habel (eds), 183218. Berlin: Mouton de Gruyter.
Declerck, Renaat. 2007. Distinguishing between the aspectual categories (a)telic, (im)perfec-
tive and (non)bounded. Kansas Working Papers in Linguistics 29: 4864.
Denison, David. 1993. English Historical Syntax: Verbal Constructions. London: Longman.
Enkvist, Nils E. 1986. More about the textual functions of the Old English adverbial a. In Lin-
guistics across Historical and Geographical Boundaries: In Honour of Jacek Fisiak on the
Occasion of his Fifiteth Birthday, Dieter Kastovsky & Aleksander Szwedek (eds), 301309.
Fleischer, Jrg, Hinterhlzl, Roland & Solf, Michael. 2008. Zum Quellenwert des althoch-
deutschen Tatian fr die Syntaxforschung. Zeitschrift fr germanistische Linguistik 36:
210239.
Frary, Louise G. 1929. Studies in the Syntax of the OE Passive, with Special Reference to the Use of
Wesan and Weoran [Language Dissertation No. 5]. Washington DC: Linguistic Society
of America.
Fritz, Thomas. 1994. Passivformen in Otfrids Evangelienbuch: Tempus, Aspekt, Aktionsart.
Sprachwissenschaft 19(2): 165182.
Geeraerts, Dirk. 2000. Salience phenomena in the lexicon: A typology. In Meaning and Cogni-
tion. A Multidisciplinary Approach [Converging Evidence in Language and Communica-
tion Research 2], Liliana Albertazzi (ed.), 79101. Amsterdam: John Benjamins.
Gippert, Jost, Javier, Martnez & Korn, Agnes. 2011 [1995]. Titus: Thesaurus Indogermanischer
Text- und Sprachmaterialien. University of Frankfurt. http://titus.uni-frankfurt.de/indexe.
htm (7 November 2011).
Green, Eugene. 2009. Synonymy and obsolence: The example of Old English weoran. Presented
at Studies in the History of the English Language 6 (SHEL 6), Banff, AB.
Gries, Stefan T. & Stefanowitsch, Anatol. 2004. Extending collostructional analysis: A corpus-
based perspective on alternations. International Journal of Corpus Linguistics 9(1): 97129.
Jones, Howard. 2009. Aktionsart in the Old High German Passive: With Special Reference to the
Tatian and Isidor Translations. Hamburg: Buske.
van Kemenade & Los, Bettelou. 2006. Discourse adverbs and clausal syntax in Old and Middle
English. In The Handbook of the History of English, Ans van Kemenade & Bettelou Los
(eds), 22448. Oxford: Blackwell.
van Kemenade & Westergaard, Marit. 2012. Syntax and information structure: Verb-second
variation in Middle English. In Meurman-Solin, Anneli,Jose Lopez-Couso, Maria Jose &
Bettelou, Los (eds), 87120.
Killie, Kristin. 2008. From locative to durative to focalized? The English progressive and PROG
imperfective drift. In English Historical Linguistics 2006, Vol. I: Historical Syntax and Mor-
phology. Selected Papers from the Fourteenth International Conference on English Historical
Linguistics (ICEHL 14, Bergamo, 2125 August 2006) [Current Issues in Linguistic Theory
295], Maurizio Gotti, Marina Dossena & Richard Dury (eds), 6988. Amsterdam: John
Benjamins.
Kilpi, Matti. 1989. Passive Constructions in Old English Translations from Latin: With Special
Reference to the OE Bede and the Pastoral Care [Mmoires de la Socit nophilologique de
Helsinki 49]. Helsinki: Socit nophilologique.
Klein, Wolfgang. 1994. Time in Language. London: Routledge.
Klingebiel, Josef. 1937. Die Passivumschreibungen im Altenglischen. Bottrop: Postberg.
Kotin, Michail L. 2000. Das Partizip II in hochdeutschen periphrastischen Verbalfgungen
im 9.15. Jh. Zur Ausbildung des analytischen Sprachbaus. Zeitschrift fr Germanistische
Linguistik 28(3): 319345.
Krotz, Elke. 2007. St. Gallen, Stiftsbibl., Cod. 56 http://www.handschriftencensus.de/16961
(23December 2011).
Krotz, Elke. 2011. Otfrid von Weienburg: Evangelienbuch. http://www.handschriftencensus.
de/werke/1285 (23 December 2011).
Kurtz, Georg. 1931. Die Passivumschreibungen im Englischen. Ohlau: Dr. Hermann Eschenhagen.
Langacker, Ronald. 1987. Foundations of Cognitive Grammar, Vol. 1: Theoretical prerequisites.
Stanford CA: Stanford University Press.
Langsrud,yvind.2004.FishersExactTesthttp://www.langsrud.com/stat/fisher.htm
(21December 2011).
Los, Bettelou. 2009. The consequences of the loss of verb-second in English: Information struc-
ture and syntax in interaction. English Language and Linguistics 13(1): 97125.
Los, Bettelou. 2012. The loss of verb-second and the switch from bounded to unbounded
systems. In Meurman-Solin, Jose Lopez-Couso & Los (eds), 2146.
Lussky, Georg F. 1924. Uuerdan und Uuesan mit dem Partizip Passiv in der Althochdeutschen
Tatianbersetzung. The Journal of English and Germanic Philology 23(3): 342369.
Maienborn, Claudia. 2007. Das Zustandspassiv: Grammatische Einordnung Bildungsbe-
schrnkung Interpretationsspielraum. Zeitschrift fr Germanistische Linguistik 35(12):
83114.
Meurman-Solin, Anneli, Lopez-Couso, Maria Jose & Los, Bettelou (ed.). 2012. Information
structure and syntactic change in the history of English [Oxford Studies in the History of
English 2]. Oxford: OUP.
Mitchell, Bruce. 1985. Old English Syntax, Vol. I: Concord, the Parts of Speech and the Sentence.
Oxford: Clarendon Press.
Mller, Torsten. 2009. Construction grammar, grammaticalisation and the Old English wurthe-
passive. Presented at Middle and Modern English Corpus Linguistics (MMECL), Inssbruck.
Petr, Peter & Cuyckens, Hubert. 2009. Constructional change in Old and Middle English
Copular Constructions and its impact on the lexicon. Folia Linguistica Historica 30,
311365.
Petr, Peter. 2010a. On the interaction between constructional & lexical change: Copular, Pas-
sive and related Constructions in Old and Middle English. Ph.D. dissertation, University
of Leuven.
Petr, Peter. 2010b. The functions of weoran and its loss in the past tense in Old and Middle
English. English Language and Linguistics 14(3): 457484.
Petrova, Svetlana & Solf, Michael. 2009. On the methods of the information-structural analysis
of historical texts. A case study on Old High German. In. Information Structure and Lan-
guage Change: New Approaches to Word Order Variation in Germanic, Roland Hinterhlzl&
Svetlana Petrova (eds), 121160. Berlin: Walter de Gruyter.
Peter Petr
Picket, Joseph P. et al. 2006. The American Heritage Dictionary of the English Language, 4th edn.
Boston MA: Houghton Mifflin.
Smith, Carlota. 1997 [1991]. The Parameter of Aspect, 2nd edn. [Studies in Linguistics and
Philosophy 43]. Dordrecht: Kluwer.
von Stutterheim, Christiane. 2002. Konzeptualisierung und Versprachlichung von Ereignisse-
quenzen.http://www.idf.uni-heidelberg.de/fileadmin/user_download/antrag_KVE.pdf
(20 November 2009).
Wattie, James MacPherson. 1930. Tense. English Studies 16: 12143.
Westergaard, Marit. 2009. Word order in Old and Middle English: The role of information struc-
ture and first language acquisition. Diachronica 26(1): 65102.
Wicka, Katerina S. 2009. From Phonology to Syntax: Pronominal Cliticization in Otfrids Evange-
lienbuch [Linguistische Arbeiten 530]. Tbingen: Max Niemeyer.
Zieglschmid, A.J. Friedrich. 1929. Is the use of wesan in the periphrastic Actional Passive in the
Germanic languages due to Latin influence? Journal of English and Germanic Philology 28:
360365.
Zieglschmid, A.J. Friedrich. 1930. The disappearance of Werdan in English. Philological Quar-
terly 9: 11115.
Zieglschmid, A.J. Friedrich. 1931. Werdan and wesan with the passive in various Germanic
languages. Germanic Review 6(4): 389396.
Causative habban in Old English
Tracing the Development of a Budding Construction*
Matti Kilpi
University of Helsinki, Finland
The budding causative use of Old English habban have has so far received
little attention in the literature compared to other Old English periphrastic
causatives. The construction with habban represents indirect passive causation
and corresponds to the Present-Day English construction of the type I had
my shoes repaired. The study is based on the entire habban material, c. 12,600
instances, from the Dictionary of Old English Corpus. The material yields 19 (20)
causative habban instances. After a brief look at the diachronic and dialectal
breakdown of the data, the discussion focuses on syntactic features, such as
word-order and the inflection/non-inflection of the past participle, and semantic
features, e.g. the roles of the causer, patient and causee, the presence of volitional
or deontic modality in all the instances, and the telicity/atelicity of the action
expressed by the verb phrase. The article concludes with a discussion of the
origin of the construction. A new hypothesis concerning the triggering of the
grammaticalisation process of causative habban is presented and viewed in
thelight of Diewalds context-sensitive grammaticalisation scenario.
1. Introduction
1.1 The structure of the paper

This article is organised as follows:
Section 1 (Introduction) describes the causative passive construction formed with
habban and the corpus studied. It further lists all the causative habban instances
* Sections 12.4 and 3 represent a reworking of Kilpi 2010, Sections1, 2.12.3, 2.5 and 3.
My thanks are due to Peter Lang Verlag for permission to use these sections in this article.
Sections 2.5 and 2.6 containing my hypothesis on the rise of causative habban are new in all
essentials. The research reported here was supported in part by the Academy of Finland
Centre of Excellence funding for the Research Unit for Variation, Contacts and Change in
English at the Department of English, University of Helsinki. I am grateful to Leena Kahlas-
Tarkka, Hugh Magennis, Olga Timofeeva and Turo Vartiainen, as well as the two anonymous
referees, for helpful comments on various draft versions of this study.
Matti Kilpi
found in the corpus. The Introduction ends with an overview of previous work
done on causative habban.
Section 2 (Discussion) first briefly addresses the diachronic and dialectal
breakdown of the instances but subsequently focuses on the syntax and semantics
of the causative construction. The syntactic features analysed include word-order
and the inflection/non-inflection of the past participle. In the domain of seman-
tics, attention is paid to the presence of volitional or deontic modality in all the
instances, the telicity/atelicity of the action expressed by the causative construc-
tion and further to the ease or difficulty of retrieving the causee and the animacy/
inanimacy of the patient. Possible support from sources outside the Old English
(OE) construction for the causative reading in the case of a number of instances,
provided e.g. by Latin originals, is also taken into account. The section is capped
up by a new hypothesis concerning the rise of causative habban. This hypothesis
is viewed in the light of Diewalds context-sensitive grammaticalisation scenario.
Section 3 (Final remarks) makes a very brief excursion into post-OE develop-
ments, pointing out that the real proliferation of the causative haven construction
in Middle English only becomes possible when the ambiguity between haven per-
fects and causative haven constructions is resolved as a result of the word-order
change from SOV to SVO in haven perfects.
1.2 The construction

The only type of OE causative construction with habban I have found, and the one
discussed in this study, is the causative passive construction corresponding to the
Present-Day English (PDE) type I had my shoes repaired. Typically, this construc-
tion follows the pattern
NP1 + HAVE + NP2 + past participle (PTCP)
The complex predication seen here can profitably be analysed by having recourse
to the concept of small clause (SC).1 Following this approach, my PDE example
will be analysed as
(1) I had [my shoes repaired]
with the SC in square brackets. The form of HAVE in the matrix clause is always
active, both in OE and PDE (Palmer 1988:195; Huddleston & Pullum 2002:1235),
but in the SC the voice is passive. The corresponding active construction with
HAVE (e.g. the type I had Tom repair my shoes) only arises in late Middle English
(see Visser 1973: 2073).
. See e.g. Miller (2002) who operates with this term.

The OE causative passive construction with habban (and the corresponding PDE
have construction) can have three participants: the causer, the patient and the
causee. Example (2) from PDE contains all three:
BNC CKD 311 Some of the ladies [CAUSER] from the finest houses have
(2)
their fancy shoes [PATIENT] made by Hari Morgan [CAUSEE].
At least in OE, the presence of the causee in the construction is rare: there is only
one example of this (example (23)) in my causative habban material.
Before starting the detailed discussion, let us have a look at one OE example of
causative habban. It is an abbreviated version of example (22):
(3) lfld gswytela on is gewrite hu ho [CAUSER] wile habban gefadad
hir hta [PATIENT] . makes it clear in this document how she wants
to have her possessions disposed of Chapter 1486 (Whitelock 15)
This is a very straightforward and unproblematic example of causative habban.

The action of disposing of the property can only take place after lflds death;
thus she can only give instructions to the executors of the will as to how this must
be done.
1.3 The corpus studied

The corpus I have studied is the whole habban material of the Dictionary of Old
English Corpus (DOEC), comprising about 12,600 instances. The total word count
of the DOEC is about 3.5 million running words. I have analysed the habban
instances by using the Microsoft Excel program and a Microsoft Access database
with 24 parameters, one of which is causative habban.
1.4 The causative habban instances

In the whole habban material there are only 19 or 20 instances which I have inter-
preted as representing causative habban. The different totals, 19 or 20, are due to
the fact that examples (16) and (17) represent one and the same instance from
Alfreds Pastoral Care as recorded in two different manuscripts.
(4) Or 6 B9.2.7 (31.150.22)] fter m he gegaderade fierd, & wolde faran
on Perse & bebead, onne he eft wre eastane hamweard, t mon hfde
anfiteatrum geworht t Hierusalem, t he mehte Godes eowas on don,
t hie dior rinne abite. and gave a command, on his way home from
the east, that they should have an amphitheatre built in Jerusalem, so that
he could put Gods servants there to be devoured by beasts.
(5) LS 28 (Neot) B3.3.28 1. (7) Swa eac ne mihte Sanctus Neotus behydd beon
ne bedigelod, a a God hine geupped habben wolde. So likewise St Neot
could not be hidden or kept concealed, when God wanted to have him
manifested.
Matti Kilpi
(6) MCharm 1 A43.1 1. (17) And hbbe him gworht of cwicbeame feower
Cristes mlo and awrite on lcon ende: Matheus and Marcus, Lucas and
Iohannes.
And let him have four Christs crosses of rowan made for him, and let him
write at each end: Matthew and Mark, Luke and John.
(7) HomM 4 (Ass 5) B1.5.4 1. (60) Hyt ws a gehende heora Eastertide,
andhi woldon habban one halgan easterdg geblodegodne welhreowlice
mid s hlendes blode. Their Easter time was then at hand, and they
wanted to have the holy Easter day made bloody with the Saviours blood in
a cruel way.
(8) WHom 4 B2.1.4 (12) Ac God gescyrt his dagas for ra ingan e him
gecorene synd & he gehealden habban wile. But God will shorten his
daysfor the sake of those who are his chosen ones and whom he wants to
have/keep preserved.
(9) WHom 5 B2.1.5 (109) Ac for ra gebeorge e him syn gecorene & e
hehabban wyle gehealden & geholpen, he forde ne eodfeond.
Butfor the protection of those who are his chosen people and whom
hewants to have preserved and helped, he will destroy the public enemy
(10) WHom 6 B2.2.1. (143) Full mycel wundor hit ws t t mden gebr
cild e nfre nahte urh hmeding weres gemanan; ac se e hf rihtne
geleafan & understent Godes mihta, he mg ful georne witan t hit ws
Gode ydde a he hit swa gedon habban wolde. It was a very great
miracle that the virgin, who had never had carnal knowledge of man, gave
birth to a child; but he who has true faith and understands Gods powers
may know full earnestly that it was easy for God when he wanted to have it
done that way.
(11) HomS 34 (Scragg Verc 19) B3.2.34 (99) y we beoda Godes bebode &
eallra his haligra, t nan ara cristenra manna e is gehyre him beforan
rdan oe elles hwara hit him gereccan, ne geristlce he is fsten to
abrecenne, be am e he wille him for Gode geborgen habban. Therefore
we proclaim with the decree of God and all his saints that no Christian who
hears this read in his presence or elsewhere communicated to him should
presume to break this fast, according as he wishes to have/keep himself
protected before God.
(12) LS 9 (Giles) B3.3.9 (377) Se Godes man heom s blielice getiode foron
e drihten hit wolde habban swa idon. The Gods man graciously granted
that to them, as the Lord wanted to have it done in that way.
(13) LS 34 (SevenSleepers) B3.3.34 (230) Ac God lmihtig Scyppend e is
ealra gesceafta Wealdend, e his gecorenan bi milde swa modor bi hire
agenum cilde, he sylf as seofon halgan bebead t hi swa slepon for his
micclan wundrum e eft he gedon habban wolde.
But God the Almighty Creator, who is the ruler of all created beings, he
who is kind to his chosen ones like a mother is to her own child, he himself
commanded these seven holy men to sleep in that way because of the great
miracles which he wanted to have performed later
(14) LS 34 (SevenSleepers) B3.3.34 (364) Ac lmihtig God se milda e lcne
mann mid fulre mildheortnysse underfeh e hine mid fulre eadmodnysse
geseh, a he s caseres mycclan hreowsunga geseah, him eac sona t
hreow and his t sarlice anginn, and hine a na lengc ahwnedne habban
nolde, ne he eac a na lengc geafian ne mihte t his halige folc lge on
gedwylde,
But the mild Almighty God, who receives with perfect loving-kindness
everyone who seeks him with full humility, when he saw the emperors deep
repentance, at once that, and also Theodosiuss sad behaviour, aroused his
compassion and he did not want to have/keep him afflicted any longer nor
was he any longer able to allow that his holy people would remain in error.
(15) GenPref B8.1.7.1 (105) God gesceop us twa eagan & twa earan, twa
nosyrlu, twegen weleras, twa handa & twegen fet, & he wolde eac habban
twa gecynyssa on isre worulde gesett, a ealdan & a niwan; for an e he
de swa swa hine sylfne gewyr, & he nnne rdboran nf, ne nan man
ne earf him cwean to, Hwi dest u swa? God created for us two eyes and
two ears, two nostrils, two lips, two hands and two feet, and he also wanted
to have two testaments established in this world, the old and the new;
(16) CP B9.1.3. (33.217.18) Form ws suie ryhtlice beboden Ezechiele m
witgan t he scolde one Godes alter habban uppan aholodne t he
meahte on healdan a offrunga & a lac e mon brohte to m weobude;
Therefore the prophet Ezechiel was very rightly commanded to have
Gods altar hollowed out up above that it might hold the offerings and gifts
brought to it. [Latin: ad Ezechielem in altari fieri fossa praecipitur,
cited from BTS s.v. -holian].
(17) CP (Cotton) B9.1.3.1 (33.216.18) Form ws swie ryhtlice beboden
Ezechiele m witgan t he sceolde one Godes alter habban uppan
aholodne t he meahte on healdan a offrunga & a lac e mon brohte to
m weobude. [For the translation and the Latin original, see (16) above]
(18) GD 1 (H) B9.5.8.2 (9.60.13) Hwer se ankenneda Godes sunu, ,
onysum inge wolde gefyllan, t he gefyllan ne mihte, a a he
wolde habban forsuwod t, t na f orholen beon ne mihte? Did the
only-begotten Son of God,, want to fulfil in this thing what he could not
fulfil, when he wanted to have passed over in silence what could not be
concealed?
(19) WPol 3 (Jost) B13.3 (20) t eallswa Crist a earce beclysde and rinne
geheold, t t he gehealden habban wolde, swa scylon a cyricweardas
rihtlice beclysan and gehealden ealle a ing, e to cyrican gebyria. That
just as Christ closed the ark and kept there what he wanted to have/keep
preserved, in the same way the church-keepers must rightly shut up and
keep all the things that belong to the church.
Matti Kilpi
(20) WPol 3 (Jost) B13.3 (21) And ealswa Crist into re earce gelaode, a
ehe rinne gehealdenne habban wolde, swa man sceall laian Godes
folc mid bellhrincge into Godes huse , And just as Christ invited into
thearkthose whom he wanted to have/keep preserved there, in the same
way Gods people must be invited with the ringing of bells into Gods
house,
(21) LawIICn B14.30.2 (80.1) & forga lc man minne huntno, locehwr
ic hit gefriod wylle habban, be fullan wite. And let every man let my
hunting alone where I wish to have/keep it preserved, on pain of the
full penalty. [Latin translation of the OE in Quadripartitus: Et abstineat
omnis homo a uenariis meis, ubicumque pacem eis haberi uolo, super
plenam witam. [Jurasinski 2004:135] and in Consiliatio Cnuti: Et dimittat
meam uenacionem, ubicumque illam pacificatam esse uolo, super plenam
forisfactum. [Jurasinski 2004:135].
(22) Chapter 1486 (Whitelock 15) B15.6.5 (1) lfld gswytela on is gewrite
hu ho wile habban gefadad hir hta for Gode & for world. lfld
makes it clear in this document how she wants to have her possessions
disposed of before God and the world.
(23) ChronD (Cubbin) B17.8 (1067.28) Se forewitola Scyppend wiste on r
hwt he of hyre gedon habban wolde, foran e heo sceolde on an
lande Godes lof ycean & one kyng gerihtan of am dweliandan pe &
gebegean hine to beteran wege & his leode samod, & alegcean a uneawas
e seo eod r beeode, eallswa heo syan dyde. The foreknowing Creator
knew beforehand what he wanted to have done by her, because she was
to increase the praise of God in the country and to guide the king from
the path of error and to turn him to a better way, as well as his people,
and to set aside the sins that the nation formerly committed, which she
consequently did.
1.5 Previous studies

Old English causative habban has been a relatively neglected area of OE syntax.
Visser (1973: 2118) has a long list of examples, which, however, begins with Law-
mans Brut. Thus no OE examples are given. Mitchell discusses the possibility of
OE habban being used in a causative function in a number of sections and dis-
cusses a couple of instances without unequivocally recognising them as represent-
ing causative habban (see Mitchell 1987: 726 and 731 in particular). In light of
this, it is understandable that he does not discuss causative habban in a section of
its own. Two relatively recent standard works on the history of English, CHEL 1
(1992) and Denison 1993, have nothing on OE causative habban.
By way of contrast, Andrzej ckis 2008 doctoral dissertation Grammati-
calisation paths of HAVE in English and its revised version (cki 2010) contain
iscussions of causative HAVE. ckis ideas about the rise of the causative use
d
have provided an important stimulus for me when I have been developing my own
hypothesis concerning causative habban.
There is very little dictionary evidence on the OE construction studied here.
The BTS recognises it (s.v. habban XVII (2)), but the only example given of this
use is actually non-causative. OED3 gives examples of causative have only from
Middle English (ME) onwards.
2. Discussion
2.1 Diachronic and dialectal breakdown

Let us first consider the diachronic and dialectal breakdown of the instances. Only
three of them, examples (4), (16) and (17), come from earlier, that is pre-950 OE.
We could equally well say that pre-950 OE is represented by two instances only
as examples (16) and (17) contain the same Pastoral Care instance from two dif-
ferent manuscripts. All the rest represent later Old English, including some very
late examples, such as example (23). The raw figures themselves tell us nothing
about the relative frequency of causative habban in earlier as opposed to later
Old English. The DOEC is not a structured corpus, and thus it would be difficult
to produce statistics of the type we can easily get from corpora like the Helsinki
Corpus. For our purposes the most important result of a diachronic survey is that
causative habban occurs in both early and late Old English.
A look at the dialectal distribution of the texts in which the instances of caus-
ative habban occur shows that West Saxon predominates here: possibly all the
texts containing causative habban could be broadly labelled as West Saxon, early
or late. This does not exclude the possibility of some admixture of other dialects,
as for instance in example (18), which comes from the anonymous revision of
Wrferths translation of Gregorys Dialogues and which may still contain some
Mercian features of the original translation. On the basis of the material at hand, it
will not be possible to make any hypotheses about the possible dialectal distribu-
tion of causative habban.
2.2 Syntactic properties of causative habban constructions

Let us first consider the word-order found in causative habban constructions.
Instead of trying to present a detailed analysis of the word-order patterns in these
constructions, I will here concentrate on that segment of the word-order which
Ifound to be of decisive importance when I studied the inflection of the past par-
ticiple in my article dealing with the OE perfect (Kilpi 2007). I refer to the relative
order of the object and the past participle.
Matti Kilpi
Before having a look at the causative constructions, let us consider an example

of the OE possessive perfect:
(24) ChronA (Bately) B17.1 (893.27) Ac hie hfdon a heora stemn gesetenne
But they had completed their service
In this construction, the object heora stemn comes before the past participle
gesetenne and this participle, which is syntactically an object complement, is
inflected. In my 2007 article on perfects I point out that in the period 850950,
with the word-order object participle, inflection is still found in 31% of all those
participles that could potentially have inflection, while for the period 9501050 the
corresponding percentage of inflection is 22%. With the word-order participle
object, the percentage of inflection is 17% for the earlier period and 0% for the
later period (Kilpi 2007:329). It is thus quite clear that the word-order object
participle is one that is most likely to trigger inflection on the participle.
Table 1. Relative order of object and past participle in relation to the inflection
of the participle in causative habban constructions
Relative order of object and past participle
a. O () PTCP: 17 instances
b. PTCP () O: 3 instances
Inflection of the PTCP in groups (a) and (b)
a. Inflection possible and attested: 5 instances
Inflection possible, not attested: 4 instances
Inflection not possible: 8 instances
b. Inflection possible, not attested: 1 instance
Inflection not possible: 2 instances
A look at the section Relative order of object and past participle in Table 1
shows that the word-order O () PTCP is found in all but three instances ((6),
(18) and (22)). This order is the same as in the possessive type of perfect seen
in example (24). With the order O () PTCP, the inflection of the past parti-
ciple is easily triggered, whether we have a perfect construction or a causative
construction. The section Inflection of the Past Participle in groups (a) and (b)
highlights the fact that in group (a), out of the nine instances in which inflection
of the past participle is possible more than half, i.e. five, have inflection. In group
(b) there are no examples with inflection. There is thus a clear analogy between the
word-order and inflectional patterns of possessive perfects and causative habban
constructions.
Before turning to a discussion of the semantic properties of causative habban

constructions, I must briefly mention two more points of syntactic interest.
The verb in the past participle is usually one that takes an accusative object
but it can also be one that is construed with a non-accusative object, as helpan in
example (9) and beorgan in example (11). In (9) there are in fact two co-ordinated
past participles, gehealden and geholpen, the former of which takes an accusative
object. This juxtaposition of two verbs which are differently construed is made
easy by the neutrality with regard to case of the patient e.
The causee is typically left unexpressed: there is only one instance, example(23),
where it is present in the form of the agent of hyre by her. An agent phrase is also
possible in Present-Day English causative have constructions, too, cf. example (2)
above.
2.3 Semantic properties of causative habban constructions

All the instances that I have interpreted as representing causative habban belong to
the category intended (controlled) causation as opposed to unintended (uncon-
trolled) causation (for further details about these terms, see Givn 1980:335).
Unlike the stative habban of perfect constructions, habban occurring in causative
constructions is always dynamic. In view of the causation being intended it is
natural that the causer is animate and personal in all the instances. The patient is
always present in the constructions and can be either animate or inanimate.
All the causative habban constructions represent mediated or indirect causa-
tion as opposed to direct causation, to use Givns terminology (1980:335). As
already pointed out in Section 2.2, the causee is only once expressed in my material
(see example (20); for the terms causer, causee and patient, see Polinsky 1995).
It was only after I had completed my analysis of the entire habban data that I
became aware of a feature that is shared by all the instances of causative habban.
That is the presence of volitional or deontic modality.
Volitional modality is more common, occurring as it does in 16 of the
instances (examples (5), (7), (8), (9), (10), (11), (12), (13), (14), (15), (18), (19),
(20), (21), (22), and (23)). It is expressed by a finite form of pre-modal willan (or
the contracted negative nyllan) followed by the infinitive of habban.
Deontic modality is found in four instances (examples (4), (6), (16), and (17)).
It takes three different realisations: the pre-modal *sculan followed by the infinitive
habban is found in examples (16) and (17): t he scolde one Godes alter hab-
ban uppan aholodne that he should have Gods altar hollowed out up above The
second type of realisation involves a form of bebeodan command in the matrix
clause, see example (4) .and bebead,, t mon hfde anfiteatrum geworht t
Hierusalem, and gave a command,, that they should have an amphitheatre
Matti Kilpi
built in Jerusalem. Finally, there is one example, example (6), where the deontic
modality is realised by the jussive present subjunctive (for the term jussive, see
Mitchell 1987: 883): And hbbe him gworht of cwicbeame feower Cristes mlo
And let him have four Crists crosses of rowan made for him.
The presence of volitional and deontic modality in the constructions is
striking. What is its contribution to the causative use of habban?
I will argue in Section 2.5. that volitional and deontic modalities co-occurring
with dynamic habban have a pivotal role in the rise of the causative habban
construction.
2.4 An analysis of the OE instances listed as causative in Section 1.3

In this section I will analyse all 20 instances listed in Section 1.3, discussing factors
that led me to the causative reading and assessing the likelihood of it being correct.
In the analysis, I will pay attention to a number of factors, of which at least (a),
(b) and (d) seem to have an important role in the analysis:
a. The telicity vs. atelicity of the action expressed by the verb occurring in the
past participle;
b. The ease or difficulty of retrieving the causee (which, with one exception, is
not expressed in my material);
c. The animacy vs. inanimacy of the patient;
d. Possible support for the causative interpretation supplied by Latin originals,
or existing Latin or Present-Day English translations, of the OE constructions.
Support from other sources.
Point (a) requires a few words of explanation. The classification of situations as

presented by Huddleston and Pullum (2002:118f.) gave me reason to pay atten-
tion to the telicity vs. atelicity of the action expressed by the verb occurring in the
past participle. A majority of my causative instances, 13 in all, can be classified
as having telic action expressed by the verb in the past participle, i.e. the action
represents a process which has an inherent terminal point (Huddleston & Pullum
2002:120). In the remaining seven instances I have interpreted the verbs in the
past participle as expressing atelic action.
As will be pointed out in Section 2.5., the presence of deontic or volitional
modality co-occurring with dynamic habban in the causative constructions stud-
ied effectively excludes a perfect or perfect infinitive reading. After examples (4)
and (5) below, I am not going to mention this property of the modal expressions in
connection with the remaining examples, although it applies equally to all of them.
I will now discuss all the instances, starting from the ones that show telicity in
the past participle.
Example (4):
Action expressed by the V in the PTCP: Telic
Causee: There is a tripartite causal chain here (for the term and a syntactically
slightly different example, see Hollmann (2005:208)): Emperor Julian gives a com-
mand (he bebeadt), the causee being the indefinite mon. This mon has the
double role of causee and causer: it is the causer in the causative habban construc-
tion mon hfde anfiteatrum geworht. The causee of this construction has, typi-
cally, been left unexpressed but is easy to retrieve: if mon refers to the authorities
in Jerusalem, the unexpressed causee can be understood to refer to the people who
were responsible for the actual construction work.
Patient: Inanimate
Other comments: The presence of deontic modality (bebeadt + pres.subj.
hfde) and dynamic habban effectively rules out the possibility of interpreting
hfdegeworht as a past perfect (see the discussion in Section 2.5.). The caus-
ative construction in the Latin original is simpler than the OE one, having as
it does only one causer and an ACI construction with the passive infinitive of
exstruere: Nam et amphitheatrum Hierosolymis exstrui jussit For he (i.e. Julian)
ordered an amphitheatre to be built in Jerusalem. [Latin cited from Sweet 1959
[1883]: 287].
Example (5):
Causee: God is here the causer; the unexpressed causee, or causees, could be seen
as Gods people on earth carrying out his will. Ultimately, however, the causee
remains elusive: this seems to be typical of a number of other instances where the
causer is a person of the Godhead.
Patient: Animate, personal
Other comments: A perfect infinitive interpretation of geupped habben is excluded
due to the presence of volitional modality and dynamic habban (see the discussion
in Section 2.5.).
Example (6):
Causee: A carpenter is the obvious suggestion for the unexpressed causee.
Patient: Inanimate
Other comments: An alternative translation by Gavin Chappell (n.d.): And let
the man have four crosses of quickbeam made for him, and write upon each end:
Matthew and Mark, Luke and John.
Matti Kilpi
Example (7):
Causee: There are two candidates here: the noun phrase in the prepositional phrase
mid s hlendes blode with the Saviours blood or alternatively an unexpressed
causee referring to the men who crucified Jesus. As the mid phrase is a typical
instrumental phrase used in passive constructions (see Kilpi 1989:1478), I prefer
to regard it as an intermediary (for this term used in connection with causative con-
structions, see Denison 1993:280), and choose the latter option, the unexpressed
animate and human causee.
Patient: Inanimate
Other comments: -
Examples (10), (12) and (13):

Action expressed by the Vs in the PTCPs: Telic
Causees: All three instances have God as the causer. In each case it is hard to say
anything specific about the causee (cf. my comment on example (2)).
Patients: Inanimate in all three instances
Other comments: -
Example (15):
Causee: The men who composed2 the Old and the New Testament
Patient: Inanimate
Other comments: -
Examples (16) and (17):

Action expressed by the V in PTCP: Telic
Causee: The worker(s) who made the cavity on top of the altar
Patient: Inanimate
Other comments: Examples (16) and (17) represent one and the same instance
transmitted in two different MSS. The presence of the double passive (fieri
praecipitur is orderedto be made) in the Latin source text strongly supports the
causative reading of the habban construction. The causer he of the causative con-
struction he sc(e)olde one Godes alter habban uppan aholodne refers anaphorically
. The verb (ge)settan is highly polysemous; I have here adopted the sense to compose
offered by Marsden (2004:128) for the occurrence of this verb in this particular sentence.
to Ezechiele, who receives Gods command (ws beboden Ezechiele); thus there is a
tripartite causal chain as in example (4).
Example (18):
Causee: The two men cured from blindness
Patient: Inanimate
Other comments: -
Example (22):
Causee: the executor(s) of the will of lffld
Patient: Inanimate
Other comments: This instance is a very straightforward and unproblematic
example of causative habban. The action of disposing of the property can only take
place after lfflds death; thus all she can do is to give instructions as to how this
is to be done.
Example (23):
Causee: The agentive phrase of hyre by her [i.e. Margaret]. This is the only instance
where the causee is present in the construction.
Patient: Inanimate
Other comments: The rest of the sentence after the causative construction lists all
the tasks God had reserved for Margaret and which she duly carried out.
The thirteen examples so far discussed have been ones containing a verb in
the past participle which presents the action as telic, i.e. having a punctual ter-
minal phase. In mediated causative constructions the idea of viewing the action
as telic seems to be natural. It is, however, obvious that a mediated causative con-
struction is possible even when the action is atelic, as in the following Present-Day
English examples:
BNC CSI 127 Campbell suggests that this is due to their own recognition, as
(25)
vulnerable women, of the problem of male violence and thus wish to have it
controlled.
(26)
Wheeler 2007 A group that stopped a proposed development of the Civil
War site seeks to have it preserved as a park
In the remaining seven instances I give an atelic reading to the verb in the past
participle.
Matti Kilpi
Example (8):
Action expressed by the V in the PTCP: Atelic
Causee: See the comment on example (5)
Other comments: Although I regard the translation of the habban construction
by PresE have as fully possible, one should also consider the possibility that hab-
ban has the dynamic meaning keep here, in which case the phrase should be
translated and whom he wants to keep preserved. Visser (1973: 2120) seems to
have adopted the latter line on interpretation, although he has both the verb have
and the verb keep in his translation: he would no longer have (keep) him afflicted.
Whichever translation, have or keep is adopted, the causative reading of habban
holds (see Section c. on p. 116 below for the causative use of PresE keep).
Example (9):
Action expressed by the Vs in the PTCPs: Atelic
Other comments: What is syntactically noteworthy about this instance is that there
are two, coordinated, past participles with a different verbal rection, healdan taking
an accusative object and helpan a dative object. Their juxtaposition here is facili-
tated by the fact that the relative particle e marking the patient is not case sensitive.
Example (11):
Causee: Mitchells (1987:922) interpretation of be am e as an agentive phrase
(by which) would lead us to regard this phrase as the causee. I have taken a differ-
ent line: following DOE s.v. be, big prep., conj. and adv. II.D, I interpret be am e as
the subordinator meaning according as (see Kilpi & Timofeeva 2011). This means
that the causee in my interpretation remains as vaguely defined as in example (5).
Other comments: Again, as in example (8), habban can be translated by keep; cf.
Vissers (1973: 2120) translation he would no longer have (keep) him afflicted.
Example (19):
Causee: The causer here is Christ in the function of the hostiarius doorkeeper of
Noahs ark. As Wulfstan is here presenting a typological interpretation of the ark it
remains unclear whether Noah would qualify as the causee within this framework.
Patient: Inanimate
Other comments: -
Example (20):
Patient: Animate, personal and non-personal
Other comments: -
Example (21):
Causee: The subjects of the ruler
Patient: Inanimate
Other comments: BT s.v. huntna gives a causative translation for this construc-
tion: And let every man leave my hunting alone where I wish to have it preserved.
There are two Latin translations of this construction available: one in Quadripar-
titus and another in Consiliatio Cnuti. The former in particular renders the caus-
ative idea inherent in the OE: Et abstineat omnis homo a uenariis meis, ubicumque
pacem eis haberi uolo; here the phrase pacem eis haberi uolo I wish peace to be
observed concerning them contains the idea of mediated causation (Jurasinski
2004:135).
The above analysis gives rise to a few observations:
a. The action expressed by the verb in the past participle is telic in a majority of
the instances, 13 in all. As already pointed out, this is natural and expected,
as in mediated causation the causer could be thought to start from the default
assumption that what is commissioned will also be fulfilled. The 7 instances
in which the action is atelic show that indirect causation with habban can
also involve processes with duration but with no inherent terminal phase. Of
these two types, I am inclined to see OE causative habban constructions with
telic action expressed by the past participle as representing more prototypical
indirect causation than those where the action is atelic.
b. The 12 instances in which the causer is one of the persons of the Godhead dif-
fer from those with an earthly human causer in that in the case of the former
it is often difficult to pinpoint the causee in a precise way (cf. my comments
on examples (5), (8), (9), (10), (12), (13), (19) and (20)). Exceptions to this
difficulty are examples (14), (15), (18) and of course example (23), the only
instance where the causee is expressly mentioned. It seems to me that, even
Matti Kilpi
in those cases in which the causee is elusive, we have instances of mediated

causation. At the risk of entering the realm of metaphysics, it could be thought
that divine intervention in earthly events takes more typically hidden forms
when compared to indirect causation between human beings.
c. In all but one of the 7 instances in which the process expressed by the verb
in the past participle is durative and atelic, it would be possible to translate
the form of OE habban by keep. When discussing one of these instances,
example (14), I already pointed to Vissers translation of the OE. The remain-
ing instances in which keep could translate the form of habban are examples
(8), (11), (19), (20), and (21). In example (9), it is only the presence of the past
participle geholpen co-ordinated with gehealden that rules out the use of keep
as a translation equivalent.
The possibility of choosing keep as a translation of habban in examples (8),

(11), (14), (19), (20) and (21) then arises from the durative and atelic process
expressed by the verb in the past participle (ahwnan, (ge)beorgan, (ge)friian
and (ge)healdan [3x]). However, this does not mean that these instances can-
not represent c ausative habban. The verb keep itself can be causative, cf. OED3
s.v. keep v.: 24. With complement: To preserve, maintain, retain, or cause to
continue, in some specified condition, state, place, position, action, or course
[emphasis added]. When used in this sense, keep can have a past participle as its
complement.
Causative have constructions can occur in Present-Day English when the
action is atelic (see examples (25) and (26)). In my interpretation they are also
possible in OE and accordingly I have included the OE atelic instances discussed
here in my selection of causative habban constructions. It is likely, though, that
they represent less prototypical indirect causation than the 13 instances where the
action expressed by the past participle is telic.
2.5 A hypothesis concerning the rise of causative habban

In Kilpi 2010 I do not yet present any hypothesis of my own concerning the rise of
causative habban. In that article I point out, however, that as dynamic and stative
uses of habban coexist in a variety of constructions, I can see no reason why the
language could not have developed parallel constructions, one with stative habban
and the other with dynamic habban, making use of an identical, or near-identical,
frame (2010:62).
Before presenting my own theory concerning the rise of causative h abban
I must briefly address the possible role of contact influence in this process.
anguage contact indeed often plays a role in the rise of syntactic structures. In
L
the case of causative habban, however, this does not seem very likely. Celtic influ-
ence is apparently ruled out: there do not seem to be any obvious Celtic parallels
for causative habban (Anders Ahlqvist, p.c.). As to possible Latin influence, the
OLD and the DMLBS do not recognize a causative use for habere with a passive
SC. Neither is habere found in the list of Latin causative verbs entering a variety
of syntactic constructions in Lehmann (forthcoming), a list which contains the
following verbs: efficere, curare, compellere, impellere, inducere, suadere, iubere,
persuadere, sinere, and facere. On the other hand, it is worth noting that Pinkster
in his 1987 article on the evolution of the Latin future and perfect auxiliaries gives
a causative interpretation to a verb phrase containing habere in his example(17)
from Ciceros De re publica. He offers the following two translations, he had
them divided or he caused them to be divided, for habuitdiscriptum in the
predication Romulushabuit plebem in clientelas principum discriptum (Pinkster
1987:220). At least the latter of the two translations is clearly causative. It may be
best, however, to regard the causative reading suggested by Pinkster as a context-
based interpretation which does not in itself allow us to regard habere as a gram-
maticalised causative verb.
Rather than seeking the impulse for the rise of causative habban in language
contacts, I am willing to see this development as a language-internal process. The
rest of this section will be dedicated to my own hypothesis concerning the devel-
opment of causative habban in OE.
The syntactic frame from which the causative use of habban could be seen to
rise is one which has the following elements:
Subject (human; semantic role agentive)

lexical HABBAN, the lexical meaning of which is roughly to possess
a direct object noun phrase (accusative or non-accusative Od, cf. Mitchell
1987: 1565)
a passive past participle of a transitive verb functioning as an object
complement; the implied subject of the participle is ambiguous in that it
allows two possible readings:
a. it is identical with the subject of HABBAN
b. it is not identical with the subject of HABBAN
This represents the construction traditionally regarded as the starting-point for,

or the early version of, the Old English perfect. But note that the above frame
assumes the ambiguity concerning the identity or non-identity of the subject of
HABBAN and the implied subject of the past participle.
Matti Kilpi
This ambiguity is seen in examples (27) and (28):
(27) CHom II, 26 (213.5) And sende his eowan to am gelaodum het
secgant hi comon. for an e he hfde a ealle his ing gegearcode
Andhe [therich man] sent his servant to those invited and ordered him
tosay that they should come because he had then prepared all his things
[ because he had then all his things in the state of having been
prepared]
(28) LawIIAtr (3.2) Gyf he his scip upp getogen hbbe oon hulc geworhtne
oon geteld geslagen: t he r fri hbbe & ealle his hta. If he [the
owner of a trading vessel] has hauled his ship up onto dry land, or built a
hut or pitched a tent, he and all his possessions shall have peace there [If
he has his ship in the state of having been hauled up onto dry land, or a
hut in the state of having been built, or a tent in the state of having been
pitched]
Theoretically we can hypothesize that in example (27) the person referred to by

the pronoun he has personally carried out the activities of preparing a feast. It can,
however, be argued with good reason that this person, the rich man in the gospel
who invited people to his feast, did not do the preparations personally; they were
carried out by his servants. Similarly in example (28) the hauling of the ship, the
building of a hut, or the pitching of a tent were obviously carried out by the ships
crew, not the owner, but at least on a theoretical level the ambiguity concerning the
identity of the subject of HABBAN and the implied subject of the past participle
remains.
As can be seen, this is the construction which is traditionally regarded in the
literature as the starting-point for, or the early version of, the OE perfect, the hall-
marks of which are present in all the italicized constructions in examples (27)
and (28): the word-order Od-PTCP and the inflection of the past participle where
morphologically possible. But note that the frame suggested here is explicit about
the ambiguity concerning the identity or non-identity of the subject of HABBAN
and the implied subject of the past participle.
In his discussion of causative HAVE, cki (2010) comes to conclusions which
show that his starting-point for the rise of causative HAVE is basically identical
with mine. Analysing the causative structure NP1 + HAVE + NP2 + past participle
he says It is hypothesised that the causative structure with a past participle is an
offshoot of the stative HAVE construction developed through evoking contextual
implicatures in which the referent of NP1 is in a superior position enabling the
referent to cause an action to be performed by giving instructions and inviting
the addressee to infer the causative meaning of the proposition. (2010:205). The
main difference between his hypothesis and mine is that he does not discuss the
co-occurrence of volitional or deontic modality with the verb-phrase containing
HAVE, while in my theory concerning the rise of causative HABBAN these modal-
ities have a pivotal role.
We already saw that all the OE causative habban instances attested in the
DOEC have either volitional or deontic modality co-occurring with them. I argue
that the causative use of habban arose from constructions of the type seen in
examples (27) and (28) by the addition of volitional or deontic modality.
Before the addition of modality, the HABBAN construction seen in examples
(27) and (28) can be described as follows:
S + HABBAN (Act, Stat) + Od + PTCP (Pass, Trans)

S = human
S = agentive
S =/ implied S of PTCP
Time reference: anterior to the time of orientation
Again I can refer you to example (27):
CHom II, 26 (213.5) And sende his eowan to am gelaodum het secgan t
hi comon. for an e he hfde a ealle his ing gegearcode And he [the rich man]
sent his servant to those invited and ordered him to say that they should come
because he had then prepared all his things [ because he had then all his things
in the state of having been prepared]
By way of contrast, the causative HABBAN construction is represented by the

following:
S + MOD (volitional/deontic) + HABBAN (Act, Dyn) + Od + PTCP (Pass, Trans)

S = human
S = agentive
S implied S of PTCP
Time reference: posterior to the time of orientation
Example (22) provides a good illustration of the causative construction:
Chapter 1486 (Whitelock 15) B15.6.5 (1) lfld gswytela on is gewrite hu

ho wile habban gefadad hir hta for Gode & for world. lfld makes it clear
in this document how she wants to have her possessions disposed of before God
and the world.
Matti Kilpi
In addition to the presence of volitional or deontic modality, the prerequisites of

the causative function of HABBAN include the following changes:
stative HABBAN dynamic HABBAN

time reference anterior to the time of orientation time reference posterior
to the time of orientation
in the structure serving as the starting-point, both the identity and the
non-identity of S and the implied S of PTCP are possible. Only non-identity is
possible in causative constructions
As will be shown later, the above formalized description given for the causative
construction will have to be modified in order to arrive at the ambiguous structure
serving as the critical context for the rise of causative HABBAN.
I will now briefly discuss these three transitions under A, B and C, partly by
extrapolating from what we know about Present-Day English.
A. With added modality, stative habban dynamic habban

In Present-Day English, modal verbs with intrinsic meanings tend to co-occur
with dynamic rather than stative verbs. I cite Biber et al. 1999: Intrinsic modality
refers to actions and events that humans (or other agents) directly control: mean-
ings related to permission, obligation, or volition (intention). There are two typ-
ical structural correlates of modal verbs with intrinsic meanings: (a) the subject of
the verb phrase usually refers to a human being (as agent of the main verb), and
(b) the main verb is usually a dynamic verb, describing an activity or event that can
be controlled. (Biber et al. 1999:485)
I decided to test, with a limited OE material, of Biber et al.s statement that
modals with intrinsic meanings tend to co-occur with dynamic verbs. I studied all
the combinations in the DOEC of either willan or *sculan co-occurring with hab-
ban. Outside the causative instances, there are
379 instances of willan + habban. Out of them, 354, or 93.4 per cent have dynamic
habban.
Similarly, excluding the causative instances, there are

298 instances of *sculan + habban. Out of them 214, or 71.8 per cent, have
dynamic habban.
The process of stative, possessive habban being replaced in causative construc-

tions by dynamic habban, when volitional or deontic modality is added, is seman-
tically smooth in light of the preponderantly dynamic use of habban when it
co-occurs with willan or *sculan outside the causative construction.
B. With added modality, time reference anterior to the time of orientation time
reference posterior to the time of orientation
It is quite obvious that the presence of volitional or deontic modality in OE caus-
ative habban constructions is of pivotal importance in changing the time reference
from the past into the future.
What Huddleston and Pullum (2002) say about volitional and deontic modal-
ity with reference to Present-Day English is in my opinion also true about the situ-
ation in Old English. The following two citations both refer to the applicability of
both types of modality to future situations.
Volitional modality:
Volition is better regarded as an implicature overlaid upon futurity an

implicature deriving from the assumption that the subject-referent is in control
(Huddleston & Pullum 2002:193)
Deontic modality:
Deontic modality generally applies to future situations: I can oblige or permit

you to do something in the future, but I cant oblige or permit you to have done
something in the past. (Huddleston & Pullum 2002:184)
The presence of volitional or deontic modality in the Old English causative habban
constructions rules out the possibility of the action being placed to a time anterior
to the time of orientation: it can only be posterior.
C. In the structure serving as the starting-point, both the identity and the non-
identity of S and the implied S of PTCP are possible: with added modality, only
the non-identity is possible in causative constructions
This is an obvious development: as causative habban represents mediated or indi-
rect causation, a causee different from the causer is a default assumption, although
this causee is seldom explicitly present in the construction. In my data there is only
one instance of the causee being expressed by an agentive phrase.
Consider example (29), in which the subject of habban is identical with the
implied subject of the past participle:
(29) ThCap 1 (Sauer) 44,399.13 & him huru nan mon onfon ne sceal butan his
scriftes leafe m he sceal habban r geandet eal t he wi Godes willan
geworhte s e he geencan mge & fter his dome betan. And no one
must receive it (i.e. the Eucharist) except by the permission of his confessor,
to whom he must have previously confessed all that he did against Gods
will, all that he can think of, and he must atone for it according his (i.e. the
confessors) judgment.
Matti Kilpi
Here the construction is definitely non-causative. The following features attested

here are all ones that differ from those expected in causative habban constructions:
habban is stative;
the time reference is anterior to the time of orientation;
the subject of sceal habban is identical with the implied subject of geandet.
The construction habbangeandet is an early instance of the perfect infinitive.

This example is a good illustration of the fact that the presence of deontic does not
automatically trigger a causative interpretation: it is also required that habban is
dynamic and that the subject is non-identical with the implied subject of the past
participle.
2.6 Th
e triggering of the grammaticalisation process: the rise of causative
habban in the light of Diewalds context-sensitive grammaticalisation
scenario
I will wind up the discussion by making an attempt to apply Gabriele Diewalds
context-sensitive grammaticalisation scenario to my hypothesis concerning the
rise of causative habban. Diewald distinguishes three chronologically ordered
stages in the rise of grammatical functions: the preconditions of grammaticalisa-
tion are placed in untypical contexts, while the triggering of grammaticalisation
takes place in what Diewald calls the critical context. The final consolidation of
the grammaticalisation process takes place in an isolating context.
Given the very infrequent occurrence of causative habban in the whole of the
DOEC, I would not venture to divide the attested instances between the first stage,
the untypical contexts and the second stage, the critical context. My very tentative
assumption here is that they all could be said to belong to the second stage. On the
other hand, they have not yet reached the final consolidation of the grammaticali-
sation process. This in my opinion only takes place in the ME period, when the
presence of volitional or deontic modality is no longer obligatory but optional with
causative haven, and after the gradual word-order change from SOV into SVO has
started to disambiguate between perfects and causative haven constructions.
I then assume that the twenty instances of OE causative habban are representa-
tives of the critical context, where the triggering of grammaticalisation takes place:
Diewald (2006:20) defines the critical context as follows: [T]he critical context
is a highly ambiguous structure, which through morpho-syntactic opacity allows
several options for its interpretation, among them the new grammaticalized
meaning.
Here is my proposal for the critical context out of which causative habban arises:
S + MOD (volitional/deontic) + HABBAN (Act, Stat/Dyn) + Od + PTCP

(Pass, Trans)
S = human
S = agentive [semantic role]
S /= implied S of PTCP
Time reference: anterior or posterior to the time of orientation
In addition to causative constructions, this frame also allows for the existence of
non-causative constructions of the type seen in example (29).
On the basis of the preceding discussion, we can now suggest a way of resolv-
ing the opacity of the critical context to allow for the rise of causative habban:
S implied S of PTCP
HABBAN Active, Dynamic
Time reference: posterior to the time of orientation
An excellent example of a causative habban construction which fulfils all these

criteria is example (22):
Chapter 1486 (Whitelock 15) B15.6.5 (1) lfld gswytela on is gewrite hu

ho wile habban gefadad hir hta for Gode & for world. lfld makes it clear
in this document how she wants to have her possessions disposed of before God
and the world.
Here the subject, lfld, is different from the implied subject of gefadad, which
must refer to the executor(s) of the will;
HABBAN is dynamic;
The time reference is posterior to the time of orientation, which is the time when
the will was made.
3. Final remarks
A possible reason for the slow emergence of causative habban constructions in

Old English is their far-reaching similarity, at the level of surface structure, with
habban perfects, particularly those with the SOV word order. With the word-order
change SOV SVO in what Visser calls true perfects the ambiguity begins to fall
away (Visser 1973: 2001; Denison 1993:348).
Matti Kilpi
This also facilitates the rise of a causative haven construction without the pres-
ence of volitional or deontic modality. The earliest MED example of this new type
is from The History of the Holy Rood-tree, dated about 1175:
(30) C1175 (?OE) HRood (Bod 343) 4/24 He [i.e. Moses] hfde an fet to
amanum iwroht He had a vessel made for that purpose alone
[Napiers translation]
But it is worth noting that in the MED, almost half of the causative haven con-
structions of the type discussed here are still ones in which either willen or shulen
co-occurs with haven. They thus testify to the continuation of the OE pattern side
by side with the emerging construction without the co-occurrence of volitional or
deontic modality.
References
Biber, Douglas, Johansson, Stig, Leech, Geoffrey, Conrad, Susan & Finegan, Edward. 1999.
Longman Grammar of Spoken and Written English. Harlow: Longman.
BNC = The British National Corpus, version 3 (BNC XML Edition). 2007. Distributed by Oxford
University Computing Services on behalf of the BNC Consortium. http://www.natcorp.
ox.ac.uk/
BT = Bosworth, Joseph T. Northcote Toller. 1964 [1898]. An Anglo-Saxon Dictionary. Oxford:
OUP.
BTS = Toller, T. Northcote. 1921. Supplement to Bosworth-Tollers Anglo-Saxon Dictionary.
Chappell, Gavin. n.d. Translation of Metrical Charm 1: For Unfruitful Land. http://www.north-
vegr.org/misc%20primary%20sources/anglo-saxon%20metrical%20charms/001.html
CHEL 1 = Hogg, Richard M. (ed.). 1992. The Cambridge History of the English Language, Vol. 1:
The Beginnings to 1066. Cambridge: CUP.
Denison, David. 1993. English Historical Syntax: Verbal Constructions. London: Longman.
tions. Special Volume 1: Constructions all Over Case Studies and Theoretical Implications,
Doris Schnefeld (ed.). http://elanguage.net/journals/constructions/article/view/24/29
DMLBS = Latham, R.E. & D.R. Howlett (eds). 1975-. Dictionary of Medieval Latin from British
Sources. Oxford: OUP.
DOEC = Dictionary of Old English Corpus. Toronto: University of Toronto. http:://quod.lib.
umich.edu/o/oec/
Givn, Talmy. 1980. The binding hierarchy and the typology of complements. Studies in Lan-
guage 4: 333377.
Hollmann, Willem B. 2005. Passivisability of English periphrastic causatives. In Corpora in Cog-
nitive Linguistics: Corpus-based Approaches to Syntax and Lexis, Stefan T. Gries & Anatol
Stefanowitch (eds), 193223. Berlin: Mouton de Gruyter. http://www.lancaster.ac.uk/staff/
hollmann/WBH-passivisability.pdf
Huddleston, Rodney & Pullum, Geoffrey K. 2002. The Cambridge Grammar of the English Lan-
guage. Cambridge: CUP.
Jurasinski, Stefan. 2004. The Rime of King William and its analogues. Neophilologus 88: 131144.
http://link.springer.com/article/10.1023%2FB%3ANEOP.0000003818.88235.27/
Kilpi, Matti. 1989. Passive Constructions in Old English Translations from Latin With Special
Reference to the OE Bede and the Pastoral Care. {Mmoires de la Socit Nophilologique
de Helsinki 49]. Helsinki: Socit Nophilologique.
Kilpi, Matti. 2007. Auxiliation in progress: Diachronic grammaticalisation changes in Old
English and Early Middle English HAVE perfects. In Change in Meaning and the Mean-
ing of Change [Mmoires de la Socit Nophilologique de Helsinki 72], Matti Rissanen,
Marianna Hintikka, Leena Kahlas-Tarkka & Rod McConchie (eds), 323343. Helsinki:
Socit Nophilologique.
Kilpi, Matti. 2010. Causative habban in Old English. In e comoun peplis language [Medieval
English Mirror 6], Marcin Krygier & Liliana Sikorska (eds), 4365. Frankfurt: Peter Lang.
Kilpi, Matti & Timofeeva, Olga. 2011. Semantic polyfunctionality and grammaticalization of
the Old English subordinator be m e: A corpus-based study. In Connectives in Syn-
chrony and Diachrony in European Languages [Studies in Variation, Contacts and Change
in English 8], Anneli Meurman-Solin & Ursula Lenker (eds), Helsinki: VARIENG. http://
www.helsinki.fi/varieng/journal/volumes/08/kilpio_timofeeva/
cki, Andrzej M. 2008. Grammaticalisation paths of HAVE in English. Ph.D. dissertation,
University of Silesia, Katowice.
cki, Andrzej M. 2010. Grammaticalisation Paths of Have in English. [Studies in English Medi-
eval Language and Literature 24]. Frankfurt: Peter Lang.
Lehmann, Christian. Forthcoming. Latin causativization in typological perspective. In Actes
du 13me Colloque International de Linguistique Latine [Collection dtudes Classiques],
Muriel Lenoble & Dominique Longre (eds). Louvain: Peeters. http://www.christianlehm-
ann.eu/pub/latin_causativization.pdf
Marsden, Richard. 2004. The Cambridge Old English Reader. Cambridge: CUP.
Miller, D. Gary. 2002. Nonfinite Structures in Theory and Change. Oxford: OUP.
Mitchell, Bruce. 1985 [1987]. Old English Syntax, III. Oxford: Clarendon Press.
OED = Oxford English Dictionary. Online version. http://www.oed.com
OLD = P.G.W. Glare (ed.). 1982 [1992]. Oxford Latin Dictionary. Oxford: Clarendon Press.
Palmer, F.R. 1988. The English Verb, 2nd edn. London: Longman.
Polinsky, Maria. 1995. Double objects in causatives. Studies in Language 19(1): 129221.
Sweet, Henry (ed.). 1883[1959]. King Alfreds Orosius [EETS O.S. 79]. London: OUP.
Visser, F.T. 196373. An Historical Syntax of the English Language, IIII. Leiden: Brill.
Wheeler, Linda. 2007. The second Battle of Shepherdstown A group that stopped a proposed
development of the Civil War site seeks to have it preserved as a park. Los Angeles Times
November 18, 2007. http://articles.latimes.com/writers/linda-wheeler
Remembering (ge)munan
The rise and decline of a potential modal
Matthias Eitelmann
Johannes Gutenberg-University Mainz, Germany
Old English (ge)munan is one of the preterite-present verbs that became obsolete
in the (strongly debated) more or less radical change into modal auxiliaries.
Contrasted with those verbs that replaced the preterite-present in its lexical sense
from Middle English onwards, an etymological analysis reveals gemunan to
indicate an act of memory the function of which is not so much to (individually)
reminisce about the past, but rather to (collectively) assess the present against
the backdrop of the past. The preterite-present experienced a renaissance due
to interlingual influence from Old Norse as mun was reintroduced and used
as a modal especially in the northern dialects of English (cf. most prominently
Sc. maun), moving steadily along the grammaticalisation cline just like the
other fully-fledged modal auxiliaries. The eventual decline of mun not only in
Standard English but also in most regional varieties can be explained from a
functional perspective which also bears implications for current changes affecting
the Present-day English paradigm of modality expressions.
1. Introduction
Constituting the primary source for the modal auxiliaries of Present-day English,
the Old English preterite-present verbs of Proto-Germanic origin have received
a great amount of scholarly attention. Yet, this is only true for those preterite-
presents that actually evolved into Standard English modals while those that
did not have rather been neglected. The losers of the evolutionary process, so to
speak, are six verbs that eventually became obsolete: witan (to know), dugan
(to be of value), unnan (to grant), urfan (to need), munan (to think) and
bengan (to suffice), with bengan and unnan dropping out of the paradigm
already in early Middle English and witan only disappearing completely at the
end of the 19th century.
One approach that does take the lost preterite-present verbs into consideration
is given with Lightfoot (1974). In his seminal paper on the diachronic analysis of
the English modals, he points out the loss of the aforementioned preterite-present
Matthias Eitelmann
verbs, or rather the non-pre-modals (Lightfoot 1974:236) as he calls them, in

connection with a series of unrelated changes that also involve
the loss of direct object constructions with preterite-present verbs,

the loss of past tense functions for past tense forms and
the resistance against adopting the to-infinitive which otherwise extended its
territory at a rapid pace.
Together with several simultaneous changes, these developments supposedly

amounted to a catastrophic change at the beginning of the 16th century, namely
the radical restructuring of the verbal paradigm resulting in the formation of a
new verbal category, MODAL.
However, the loss of the non-pre-modals leading to the isolation of the pre-
modals was not as straightforward as Lightfoots notion of a cataclysmal change
implies. This is, somewhat contradictorily, realised by Lightfoot himself when
he admits that dugan as one of the lost non-pre-modals survived well into the
19thcentury, behaving schizophrenically (Lightfoot 1974:237) both as an inde-
pendent verb and as a modal similarly to dare (another verb from the preterite-
present group). Also, he acknowledges that the now uncommon mun (from OE
munan) survives into recent times, yet only as a modal (ibid.). Thus, in order
to defend his assumption of a radical change turning the whole system of verbs
upside down, he conveniently counts the latter verb to the group of pre-modals
and at the same time shuns the first without any further justification. As for the
exact reasons why only some of the preterite-present verbs formed a distinct verbal
category while others were lost, Lightfoot (1991:147) suggests that semantic fac-
tors must have played a major role. Yet, he fails to notice that the entire preterite-
present paradigm displays a high degree of semantic coherency, with all meanings
revolving around the cognitive concepts of volition, intentionality and necessity
(cf. Fischer 2007:203, Footnote 2).
Frequently, albeit briefly, the issue of the lost preterite-presents has been
addressed in critical responses to Lightfoots approach which centre around
the correction of mainly two aspects (for concise summaries see McMahon
1994:116129 or Fischer 2007:161177). On the one hand, it has been criticised
that Lightfoot exaggerated the homogeneity of the verbal group in Old English by
putting pre-modals and full lexical verbs on a par. As Warner (1993) has shown,
however, the Old English preterite-presents constitute a verbal subcategory with
distinctive features that anticipate the peculiar nature of the Present-day English
modals (for a similar view of the Old English preterite-present verbs as a sub-
class see Colman 1992). These properties, e.g. their occurrence in ellipses or in
impersonal constructions, distinguish the preterite-presents from full lexical

verbs already in Old English and increasingly so in Middle English and not
only the later modal auxiliaries but also the lost preterite-presents. The second
point of criticism concerns the radical nature of the change which should rather
be conceived of as a gradual development with some pre-modals taking the early
lead in the change to modal auxiliaries (such as mtan, sculan), others lagging
behind (such as cunnan) and yet some others developing more verb-like char-
acteristics (such as the so-called quasi-modals durran and agan). In this context,
the lost preterite-presents serve to illustrate that they, too, developed modal char-
acteristics as is evidenced in regional dialects, an observation that undermines
the straightforwardness of the auxiliarisation process (cf. Plank 1984:311ff. and
Nagle & Sanders 1996:256). Still, what is lacking is a detailed account of the lost
preterite-presents that discusses their undeniable potential as modal auxiliaries
in relation to the other members of the very same verbal subclass that underwent
the full process of auxiliarisation.
The present paper seeks to contribute to the full story of the modal auxiliaries
by concentrating on one of the lost preterite-presents, namely munan. Section 2
investigates the semantic implications of Old English (ge)munan by contrasting
it with Present-day English verbs that are generally offered as renditions for (ge)
munan in Anglo-Saxon dictionaries. Section 3 turns to the post-Old English era
and explores occurrences of man/mun/maun in Middle English and Early Modern
English as well as in regional dialects of the 18th and 19th centuries. In Section4,
the observed development of modal senses is then put into the context of the
grammaticalisation framework in order to assess the degree to which mun evolved
modal characteristics like the genuine modal auxiliaries. Section 5 discusses rea-
sons why mun eventually declined and became obsolete even though it had moved
along the grammaticalisation cline to a considerable extent. The conclusion in
Section 6 provides a summary of the findings and highlights their implications for
the grammaticalisation of the modal auxiliaries and current developments within
the modal field.
2. Re(-)membering ( ge)munan: A semantic analysis of OE (ge)munan
Old English munan and its derivative gemunan go back to the Indo-European
etymon *men-, the core meaning of which is reconstructed as to think, to be men-
tally active. Cognates of this root can be found all over the Indo-European linguis-
tic area, ranging from Old Indian (mnyat think) over Old Persian (mainyhay
think) to Greek ( remember) and Latin (meminisse remember), to name
Matthias Eitelmann
just a few (see Pokorny 1959, s.v. 3. men-). In the Germanic languages, the ety-
mon is manifested in the group of preterite-present verbs. Apart from Old English,
attested cognates include (cf. Birkmann 1987):
Gothic *munan think and gamunan remember,
Old Saxon *munan think and *farmunan dispise,
Old High German firmonen dispise,
Old Icelandic muna remember and munu will, shall,
Old Faroese munna probably be,
Old Swedish mona remember and munu think, shall, seem,
Old Norwegian *munu will, shall and
Old Danish muna/mona remember and munu/monu will, shall.
Basically, the Indo-European etymon *men- took two paths of development in
the Germanic languages: It either preserved its lexical semantics, or it developed
modal characteristics early on, with both forms partially co-existing in the North
Germanic sub-branch.
With regard to Old English, the meaning of munan is, as (1) and (2) illus-
trate, purely lexical and closely connected to the reconstructed semantic core of
the Indo-European etymon think, be mentally active:
(1) Til mon tiles ond tomes meares.
A good (one) thinks of his good and tame horse.
(Maxims I, 141 [OED, s.v. mone])1
(2) t hine God s cynedomes weorne munde
That God thought him to be worthy of the kingdom
(Psalter C. 50, 150 [Bosworth & Toller 1898, s.v. munan])
However, occurrences of the simplex munan in Old English prose and poetry
are extremely rare in contrast to the more complex derivative gemunan. As can
be deduced from the examples provided in Bosworth and Toller 1898 and 1921,
munan is clearly outnumbered by gemunan, with merely seven attestations for
munan as opposed to over sixty for the prefixed form. As evidence from glosses
and Latin source texts shows, gemunan is the Old English rendition of Latin recor-
dari, memorari, meminisse and meditari. It seems safe to say, therefore, that gemu-
nan, other than the simplex form, denotes the act of remembering in Anglo-Saxon
society:
. Dictionary sources for quotations are provided in brackets. The abbreviations are as
follows: OED (Oxford English DictionaryOnline), DOST (A Dictionary of the Older Scottish
Tongue ed. Adam Aitken), EDD (English Dialect Dictionary ed. Joseph Wright), MED (Middle
English Dictionary ed. Shermann Kuhn).
(3) Ic gemune e recordor tui

(lfrics Grammar 41 [Bosworth/Toller 1898, s.v. ge-munan])
(4) Gemunan his halegan cynesse memorari testamenti sui sancti
(Gospel of St. Luke 1, 72 [Bosworth & Toller 1898, s.v. ge-munan])
Accordingly, dictionary entries (Bosworth & Toller 1898/1921; Clark Hall 1960)
agree on remember as an appropriate rendition, but also offer (somewhat weaker)
bear in mind, be mindful or think, consider. Translations thus usually revolve
around two closely connected cognitive activities of the human mind, namely
thinking and remembering. Yet, it would be oversimplifying to treat gemunan
as an exact equivalent of Present-day English verbs denoting the act of memory,
especially when taking their respective etymologies into account. A brief con-
trastive look at those verbs that replaced gemunan in Middle and Early Modern
English the most common ones being remember (first attestation according to
the OED ca. 1300), recollect (1559), recall (1582) and remind (1645) may help us
to assess the semantics of gemunan more adequately.
The earliest lexical competitor of gemunan, i.e. remember, enters the early
Middle English lexicon as a borrowing from Old French remembrer, which
itself goes back to Latin (re)memorari or rather to the Vulgar Latin form (sibi)
(re)memorare. Like Old English gemunan, Latin (re)memorare originates from
the Indo-European etymon *men-, to think, to be mentally active, and results
from a threefold derivational process, as illustrated in Figure 1:
(re-) mem- -or -are

men-
Figure 1. Derivation of Latin (re)memorare (own diagram)
While the optional attachment of the prefix re- implies in Latin that some-
thing is not only memorised but memorised again in other words, brought back
into ones memory , in Old French (and subsequently in English) this special case
of memorisation turns into the common case, implying that every act of memory
is a reintegration into mens, i.e. the mind.
Other Early Modern English additions to the semantic field conceptualise the
act of memory in a similar manner, all of them emphasising the act of memory
as a kind of reintegration into the mind, which is, after all, the literal meaning of
Matthias Eitelmann
remind. Recollect and recall start out as collocations in which the literal senses
of the verbs re-collect (pick up again) and re-call (call back) are still prevalent,
namely recall to ones mind and recollect to memory. It takes considerable time until
both of them are used autonomously in the sense of remember without the local
context of the to-complement:
(5) The seller Who recollecting how commodious The bargaine was for
me Did kindly offer me the peny-worth.
(1615, J. Stephens Satirical Ess. ii 13 [OED, s.v. recollect, v.2])
(6) Here the latter clause is so inferred, that it recalls the former.
(1634, J. Barton, Art Rhetorick vi. 26 [OED, s.v. recall])
What is immediately striking about all these verbs is that they all have the iterative
prefix re- in common, a property that strongly characterises the act of memory
associated with them. Past events obviously have to be called back to mind in a
more or less conscious effort that varies in degrees depending on the respective
verb. What this implies is a clear boundary between now and then, which makes
it necessary for an otherwise externalised and distanced past to be pieced together
in the process of remembering. It is important to note in this context that the
cognitive concept that underlies the modern verbs of remembering befits liter-
ate societies that are indeed capable of externalising the knowledge about their
past in written form, i.e. in manifestations of cultural memory (sensu Assmann
1992). The reiteration of past events stored in the cultural memory is then usu-
ally practiced in the act of reading, which is rather performed individually (and
privately) as opposed to the collective (and public) act of memory characteristic
of oral societies.
The concept of memory denoted by Old English gemunan differs crucially in
this respect, an observation that is closely connected to its inherent characteristics
as a preterite-present verb. As typical of preterite-present verbs, munan implies
a shift of focus from a past action to its present effect, with the psychological
emphasis on the state attained (Prokosch 1939:188). This logical transfer is most
prominently represented in witan, the perfect forms of which infiltrated the pres-
ent tense paradigm accompanied by the semantic-pragmatic shift
(7) ic wt I have seen ergo I know
In the case of munan this means that a preceding process of thinking, i.e. the
action of having been mentally active, results in the present state of keeping
something in ones thoughts. The derivation gemunan emphasises this logical link
even more strongly since the prefix ge- generally fulfils a resultative or perfective
function when attached to verbs (cf. gernan gain by running or geascian learn
by asking, Kastovsky 1992:380). In other words, similarly to the p rototypical
preterite-present verb witan, the semantic-pragmatic implications of munan can

analogously be described as
(8) ic man I have thought ergo I keep in my thoughts/keep in mind
In this respect, gemunan conceptualises the act of memory in a way typical of oral
societies in which memory is evoked collectively through repetitive retellings in
order to keep the oral societys past alive and guarantee its continuity and distinc-
tive identity. What is characteristic of oral societies is that only those events form
an integral part of the collective memory which have an imminent significance for
the societys present and its self-awareness, whereas anything deemed irrelevant is
excluded. In other words, for a pre-literate society which has no access to written
records and thus no means to externalise its collective memory in the sense of a
cultural memory, the only way to preserve knowledge is provided by constant
evocations of the past. This practice is illustrated in the following passage from the
heroic poem Beowulf which, after all, depicts such an oral society depending on
oral memorisation in the absence of written records. Here, the scop, i.e. the Anglo-
Saxon poet who recites traditional songs for his audience, is described as
(9) guma gilphlden, gidda gemyndig

se e ealfela ealdgesegena
worn gemunde
a man filled with high words, mindful of [traditional] tales, who bore in
mind a multitude of all kinds of old traditions (Beowulf, ll. 868870a)
In the cited passage, the poet improvises a lay on Beowulf s heroic feat by drawing
on traditional material that has been handed on orally as the term eald-gesegena
(old-sayings) implies. His continuation of the old traditions is described twice
explicitly as an act of memory, first by gemunde and second by the related adjective
gemyndig (mindful). As we know from studies on oral poetics, the poet does not
remember the old stories individually as an expert; on the contrary, he can be sure
that his audience has a shared knowledge of the traditional material so that they
can indulge even more in his creative variations and innovations. In this regard,
gemunan denotes the act of memory as an evocation of the past in the light of
which the present, namely Beowulf s success, is assessed.
With its preterite-present morphology and the derivation via the resultative
prefix ge-, gemunan can thus be taken to be the form-symbolic expression of the act
of memory as conceptualised in oral societies. The identity of form and function
is, however, successively distorted as gemunan also comes to serve as the denota-
tion of the act of memory in the literate society for which a different concept of
memory applies. It might therefore be no coincidence that the preterite-present
verb would ultimately be replaced by other verbs denoting the act of memory.
Matthias Eitelmann
3. The rising modality of munen in the post-OE periods
Munen continues to be used in the lexical sense of bear in mind/remember in

Middle English (cf. (10)) but becomes obsolete by the middle of the 14th century.
The last occurrences of munen, used in the same syntactic contexts as the Old
English equivalent, date from around 1350 (cf. (11)). It is only in the Present-day
English idiomatic expression to keep something in mind that at least the nominal
derivation gemynd ([collective] memory) has been retained.
(10) Ne munnde nohht tatt illke follc/att crist comm till onn eore.
That same folk did not bear in mind that Christ had come to earth
(?c1200, Ormulum, l. 9357 [OED, s.v. mone])
(11) [Androcheus] preied hym at he wolde mone at he was his broer sone.
[Androcheus] beseeched him that he would bear in mind that he was his
brothers son (a1338, R. Mannyng, Chron., l. 4811 [OED, s.v. mone])
While Middle English munen ceased to be used in its full lexical sense, the Indo-
European etymon experiences a kind of renaissance as it resurfaces, especially in
northern dialects, in a new function and in new syntactic environments. Com-
plemented by an infinitive, Middle English mun/man displays, similarly to other
preterite-present verbs, a finer shade of meaning than the simplex verb can
express (Mitchell 1985:421); in other words, it has adopted a modal function and
starts down the road to auxiliaryhood (Bolinger 1980:297).
The re-emergence of mun as a modal auxiliary is generally taken to be a bor-
rowing from Old Norse munu (cf. Visser 1969:1441; Fai 1989:295f.) which had
developed future modal semantics comparable to shall/will early on. This function
is, after all, attested in Old Swedish, Old Danish and Old Icelandic (cf. Barnes
2008:153). However, we should not overlook the fact that both the Old Norse
modal auxiliary and the Old English preterite-present originate from the same
Indo-European etymon. It does not seem implausible that the Anglo-Saxon
word already in usage merged with the Old Norse cognate with which it shared
homophonous and near-homophonous forms. Thus, it would take too narrow a
view to assume that an already existing word was replaced by a cognate and that
any further developments are to be strictly kept apart from the Old English coun-
terpart. Rather, we should expect strong interlingual influence to be at work in a
scenario in which two sister languages come into close contact with each other.
The introduction of mun as a function word as well as the loss of munen in its old
lexical sense can therefore be assumed to have been facilitated after a brief period
of co-existence (see below). What is more, we can expect an accelerated develop-
ment of modal characteristics that exceed the Old Norse usage of mun as a future
auxiliary.
3.1 mun/mon in Middle English

A first clue to the modal functions of Middle English mun is provided by manu-
scripts which exist in different dialectal versions. In these, we can often observe
that mun varies with other auxiliaries established at that time such as will or shall.
For example, the southern version of the Cursor Mundi quoted in (12) substitutes
mon for wolt, while mun varies with both sall and wille in southern versions of the
Passion cited in (13).
(12) And at ic hope ou mon it by
And that I hope you will be it
(a1325, Cursor Mundi 2036 [MED, s.v. mnen v.(2)])
(13) If ou haue done till him vnright, On i self mun fall e plight
If you have wronged him, the plight will befall yourself
(a1425, The northern Passion 85/842 [MED, s.v. mnen v.(2)])
The exact connotations of mun as a future auxiliary may vary from context to
context, especially if it co-occurs with another future auxiliary. In (14), e.g. mon
contrasts with schall in that the first displays a strong sense of volitionality absent
in the latter:
(14) To nyt we mon geyt a chyld at schall owre londus weld.
Tonight we will get a child that shall rule our lands.
(c1450, Sir Gowther [Visser 1969:1443])
Mun indicates the speakers intention to get a child in the near future of tonight
while the use of schall in the relative clause refers to the consequences of this action
in the remote future. It is in volitional contexts such as these that the modal func-
tion of mun appears to be most closely connected to the Indo-European etymon
men-, i.e. its basic meaning to think was metaphorically extended to the meaning
to intend and came to function as a future marker.2
In (15), on the other hand, where mon co-occurs with wil, it is wil that has voli-
tional sense (he who is willing to orient) whereas mon displays non-volitionality:
(15) He at his wille til God wil sette, Grete mede arfor mon he gette.
He who will orient his will towards God will receive great reward for this.
(a1400, The pricke of conscience 96 [MED, s.v. mnen v.(2)])
. It is interesting to note in this context that we can observe an incipient development of a

modern parallel in German, i.e. gedenken. A derivative of denken think, it originally meant
commemorate and recently adopted modal senses when complemented by a zu-infinitive
(cf.Was gedenkst du zu tun? What do you intend to do? > What will/are you going to do?).
In this modal sense, gedenken would be a perfect translation of mon in, for example, (14).
Matthias Eitelmann
Here, mon specifies the succeeding verb not as an action to be controlled by the
subject ( he is willing to receive) but as the expected outcome of the circum-
stances mentioned before (= he will undoubtedly receive). In this respect, then,
mun expresses certain futurity, i.e. the future course of events can be inferred from
the linguistic or extra-linguistic context. This function is also prevalent in the next
two examples where a volitional reading can be ruled out:
(16) If we waste it ful grevosly mon we be demed and punyst.

If we waste it , full grievously will we be doomed and punished.
(c1349, Rich. Rolle, Form of Living [Visser 1969:1442])
(17) Help we drowne! Now mon we dy for all oure dede.

Help we are drowning! Now we will die for all our deeds.
(a1460, Towneley Plays 77 [MED, s.v. mnen v.(2)])
In (16), the prospect of doom and punishment is dependent on the condition

stated in the preceding if-clause but is surely not an outcome intentionally planned
by the subject; in (17), death is the inevitable fate of the drowning sinners. What
these usages of mun thus entail is an inferred sense of obligation and necessity
with respect to the future, which endows them with a latent deontic modality.
Deontic modality is even stronger in contexts of second person pronouns:
(18) ou mun pay my rawunsun

you must pay my ransom
(?c1425, Avowynge of King Arthur 367 [OED, s.v. mun])
(19) When I ame dede, ou mon regne after me.

When I am dead, you must rule after me.
(c1440, The prose life of Alexander 9/14 [MED, s.v. mnen v.(2)])
In (18), the speaker imposes an obligation on the addressee, whereas in (19), the
necessity of reigning is linked to a specific condition mentioned in the when-
clause. This strong deontic use is also prevalent in (20) in which the verb of
motion is omitted, thus forming an exception to the otherwise compulsory verbal
complementation.
(20) Thow mon to Paris to the King.

Thou must [go] to Paris to the King.
(c1475, Rauf Coilear 425 [Visser 1969:1443])
A further usage of mon can be identified in contexts in which it adopts the function
previously fulfilled by the subjunctive, an observation that ties in with the general
rise of periphrastic modal constructions as a compensatory strategy for the loss
of subjunctive inflections (see Fischer & van der Wurff 2006:144). In (21), mon
is used to compensate for the subjunctive in order to clearly mark the c ontents of
the at-clause as something only stated in the books, which is not necessarily a
factual truth:
(21) The bokes says at he mon Venge his fader bane.
The books say that he will take revenge on his fathers murderer.
(c1440, Perceval 567 [MED, s.v. mnen v.(2)])
The past tense monde is also frequently used in constructions which formerly
made use of subjunctives such as when expressing the future in the past (cf. (22)).
In this respect, monde behaves similarly to the past tense of willan, which had
started to be implemented in such contexts already in Old English (cf. Hopper &
Traugott 2003:48).
(22) Ne munnde nohht tatt illke follc habbenn mihht To follhenn all
hisslare.
That same people would not have the power to follow all his teachings.
(?c1200, Ormulum 9357 [MED, s.v. mnen v.(2)])
To sum up, Middle English mun primarily serves as a marker of futurity for both
volitional and non-volitional actions. In those contexts in which volitionality is
ruled out as the action described cannot be controlled by the subject, mun approx-
imates deontic modality. What seems to be completely absent as yet, though, are
usages of mun that display epistemic modality in a narrow sense.
3.2 mun/man in Early Modern English

In the period between 1500 and 1700, we can observe various changes in the use
of mun as a modal auxiliary, both in terms of frequency and functions. On the
one hand, mun continues to be used in its function as a future marker indicating
volitional actions as in (23) and (24), but to a far lesser extent.
(23) I mun be maried a Sunday.
I will be married on a Sunday.
(c1553, Udal, Roister Doister III, iv, 151 [Visser 1969:1443])
(24) I ame and man contineu best freind to you baith.
I am and will continue [to be] a best friend to you both.
(c1580, James VI, Facs. Nat. MSS. III lxxiv. [DOST, s.v. man, maun])
On the other hand, mon is now increasingly restricted to the narrower sense of
indicating certain futurity in non-volitional contexts. (25) is particularly interest-
ing in that the quotation again contrasts the usage of the two future auxiliaries will
and mun, with the first being used to speculate about a reaction in the future and
the second going beyond speculation by referring instead to a certain outcome.
Matthias Eitelmann
(25) What will awd Maaster say to this? I mun ner see the Face of him I wot.
What will old Master say to this? I shall never see the face of him [again]
Iknow. (1688, T. Shadwell, Squire of Alsatia I.i.7 [OED, s.v. mun])
(26) For I man pay thame I am adebtet to

For I will [have to] pay them [back] to whom I am indebted
(1596, Aberdeen Council Letters I. 65 [DOST, s.v. man, maun])
More frequently than in Middle English, mun marks pure deontic modality,
expressing necessity and obligation both on the part of the speaker as in (27) and
on the part of the addressee as in (28).
(27) be is phrais ve man onderstand the equalite of e Sone with e Father.
By this phrase we must understand the equality of the Son with the
Father.(c1560, Litil Tracteit 124, 8 [Visser 1969:1443])
(28) ow man remowe all sik fleschlye Iugements.
You must remove all sick fleshly judgments. (c1560, Kennedy,)
Furthermore, starting in Early Modern English, mun/man also comes to display

epistemic modality, expressing the inferred or presumed certainty of a fact just
like must which we would have to substitute man for in each of the following sen-
tences in order to provide a Standard English rendition. This is not to say of course
that subjective senses were completely absent before 1500; after all, any statement
made about the future course of events involves subjectivity on the part of the
speaker. What is new about the epistemic uses of mon evolving in Early Modern
English is that they concern a subjective assessment of a present situation and
the speakers confidence [] in the truth of the proposition expressed (Coates
1983:18).
(29) Veralie He man be God.
Truly, he must be God. (1533, J. Gau, The Richt Vay to the Kingdom of
Heuine 54/6 [DOST, s.v. man, maun])
(30) He man be vaileand, For he is cummit of nobill parentell.
He must be valiant for he descends from noble parents.
(c1540, Clariodus: a Metrical Romance III 1414 [DOST, s.v. man, maun])
What is more, an epistemic reading of mun ist most evident in juxtaposition with
a perfect infinitive as in (31) where there is no doubt about its interpretation as
an evidential of belief or induction (for a similar example concerning the perfect
infinitive in juxtaposition to must see Chafe 1986:266).
(31) Quhilk visiounis man haif procedit of your dewilisch airt of wichcraft
Such visions must have proceeded from your devilish art of witchcraft.
(1591, Criminal Trials [DOST, s.v. man, maun])
As we can see, mun prevails in the functions already attested for Middle English,
but at the same time develops new modal characteristics as it comes to be used
as a marker of epistemic modality. Furthermore, we can observe a change in the
regional distribution of mun-instances as mun is almost completely ousted by the
core modals in the rising standard of Southeastern English while it is retained in
northern varieties as witnessed in the 18th and 19th centuries.
3.3 mun/man/maun in northern dialects

As Upton, Parry and Widdowson (1994) demonstrate in their survey of English
dialects, mun/man prevails as a modal auxiliary especially in Northern England
(widespread dialectal usage is recorded for Lancashire, Cheshire, Northumberland,
Cumberland, Durham, Westmoreland and Yorkshire), but also to a lesser extent in
the dialects of the East Midlands (Derbyshire, Lincolnshire) and the West Midlands
(Shropshire, Staffordshire). All in all, mun/man covers a dialectal area from the
northern regions of England to regions as far south as Oxfordshire and Berkshire
(cf. EDD, s.v. mun).
For the present purpose, it will suffice to explore Scottish English in order
to assess paradigmatically to what extent modal tendencies of the Middle and
Early Modern English periods continued if not intensified in the regional variet-
ies. In this northern variety of English, maun has been added to the grammatical
inventory of modal auxiliaries which is markedly different from that of Standard
English anyway (cf. Miller 2008:303307).
The most prevalent use of maun is its function as a marker of deontic modality.
In 1st person contexts, maun frequently expresses a participant-external necessity
independent from the will of the speaker. In those contexts in which maun refers
to the future, it again expresses a fixed futurity or an inevitable outcome retaining
the overall dominant sense of necessity and obligation. Like Middle English (20),
the verb of motion may be omitted as in (34).
(32) But fare ye weel, Edie, I maun be back to the evening service.
(1816, Scott, Antiquary XXVII, 289 [Visser 1969:1444])
(33) Father said I maun try and get a place.

(1848, E.C. Gaskell, Mary Barton I.iv.45 [OED, s.v. maun])
(34) Weve lots to dae, baith you and I, And I maun to the sea.
We have lots to do, both you and I, and I must [depart] to the sea.
(1933, Border Mag. (Dec.) 178 [The Scottish National Dictionary, s.v. maun])
In 2nd person contexts, the speaker expresses a high degree of insistence towards
the addressee(s).
Matthias Eitelmann
(35) Desultory reading is the bane o lads. Ye maun begin with self-restraint and
method, my man. (1850, Ch. Kingsley, Alton Locke [Visser 1969:1444])
(36) If we wid interest dem, Ye mann spaek in English too.

If you want to interest them, you must speak in English, too.
(1913, J.J.H. Burgess, Rasmies Bddie 11 [OED, s.v. maun])
A similar sense of obligation and necessity is also dominant in 3rd person contexts
such as these:
(37) And farbye they maun labour until sic time as the hairst be gatherit in.
And far away they must work until the moment the harvest is
gatheredin. (1969, G.M. Brown, Orkney Tapestry 41 [OED, s.v. maun])
(38) Whats dune in the body maun be answered in the spirit.

Whats done in the body must be answered in the spirit.
(1816, Scott, Antiquary XXVII [Visser 1969:1443])
Finally, epistemic uses of maun followed by the bare infinitive as in (39) or (40)
or by the perfect infinitive as in (41) can also be found in Scottish English, again
expressing the inferred or presumed certainty of a fact:
(39) Ye maun be in an unco [= uncouth] hurry, said a little man on
horseback, riding up close to him.
(1848, J. Boyce, Shandy MGuire i.5 [OED, s.v. maun])
(40) It mun be drouthy work preachin. (1925, C.P. Slater, Marget Pow 166 [The
Scottish National Dictionary, s.v. maun])
(41) Ye maun hae been terrible bonny in thae days!
You must have been terribly bonny in these days!
(1894, S.R. Crockett, Lilac Sunbonnet 34 [OED, s.v. maun])
This brief survey of Scottish English usages of maun demonstrates a continuation

of the modal developments already discernible in the transition from Middle to
Early Modern English, with a steady increase of mun/man/maun as a marker of
deontic modality at the expense of its function as a marker of volitional futurity.
However, it should be noted that there are also dialects which differ in this respect,
namely in that they use mun/man almost entirely with a future force (EDD, s.v.
mun). Still, as its function is primarily to indicate certain or fixed futurity, the
aspect of necessity is clearly prevalent.
3.4 Summary: The auxiliaryhood of mun/man/maun

It is striking that mun/man/maun displays exactly the same properties as
thosepreterite-present verbs that underwent the full process of auxiliarisation.
This implies that third person singular does not take the regular inflectional
ending -s (*he muns), non-finite forms do not exist (*muning), and mun does
not co-occur with other modal auxiliaries (*shall mun).3 Also, mun can no lon-
ger take direct objects, its past tense forms no longer function as past tenses,
and it takes a bare infinitive without to as a verbal complement.
Furthermore, mun shares some major formal characteristics with the other
modal auxiliaries which have become known as the so-called NICE-properties
(cf. Huddleston 1976:333): (1) the negation particle not is placed immediately
after the auxiliary and may be cliticised to nt (Negation), (2) they can switch their
position with the subject without triggering DO-support (Inversion), (3) they can
recur without a full verb (Code), and (4) they can be used in emphatic affirmation
(Emphasis). The first three of these properties also apply to mun/man/maun. The
negative particle not is placed after mun/man/maun as in (42), eventually leading
to contracted forms of the negative operator such as munnet (43) or maunt (44) in
northern English dialects.
(42) where I am e moun not come.

Where I am you must not come.
(c1380, Wyclif, John 7,36 [Visser 1969:1443])
(43) Thou munnet expect to see me.

You mustnt expect to see me (1867, Lonsdale Mag. 312 [EDD, s.v. mun])
(44) Yo maunt forget ta call

You mustnt forget to call. (1882, Senior, Smithy Rhymes [EDD, s.v. mun])
In interrogative clauses, mon is inverted with the subject; no do-operator is needed:
(45) Mon nocht everie man beir his awin burdyng?

Must not every man bear his own burden?
(1558, Q. Kennedy, Tractive 131 [DOST, s.v. mon])
Mon also occurs in elliptical structures in which the infinitival complement is

absent but can be retrieved from the linguistic context:
(46) Prey we al to that blyssyd Son That he vs help wan we not mon.
Lets pray all to the blessed son that he help us when we will/can not
[help ourselves] (a1500, His body 118 [MED, s.v. mnen v.(2)])
. Admittedly, one double modal usage is recorded for c1436: he shal mone lightlych lese
his lond. (MED, s.v. mnen v.(2)); no double modals involving mun/man are attested in Early
Modern English.
Matthias Eitelmann
Against the backdrop of these striking parallels between mun/man/maun and the
other modal auxiliaries, the question is now how the observed tendencies fit into
the larger picture of the preterite-present verbs losing their lexical semantics and
obtaining modal characteristics in the process of grammaticalisation.
4. The grammaticalisation of mun/man
The rising modality of mun/man from the early Middle English period onwards
is no coincidence but part of a universal series of changes that involved all the
members of the preterite-present group (with one exception, namely witan). As
Warner (1993:183) speculates, the semantic coherency of the [Old English]
modal group [] was itself a factor which led speakers and learners to prefer
developments of this group to other forms of expression. Due to their striking
semantic-pragmatic potentiality the preterite-presents are perfectly qualified to
serve as modality markers. The beginning of the 13th century is pivotal not only
because most preterite-presents had obtained modal characteristics by then but
also because only then did strongly subjective epistemic readings develop for will
and must (cf.Traugott 1989:42 and Warner 1993:162). With frequency being one
of the main triggers for language change (cf. Krug 2003), it is not surprising that
the increasing use of the preterite-presents as modality markers leads to drastic
changes concerning both their lexical and grammatical status.
The adoption of modal characteristics at the expense of lexical semantics is
accompanied by the interaction between syntactic and semantic-pragmatic fac-
tors usually encountered in grammaticalisation. As concerns the changes Middle
English mun underwent, it has to be admitted that the development differs from the
grammaticalisation of the core modals in one crucial respect, namely in that Old
English munan seems to have become obsolescent in early Middle English while
mun resurfaces as a Scandinavian borrowing in northern dialects. In other words,
it is primarily due to strong interlingual influence from Old Norse that the (re-)
emergence of mun as a future auxiliary is promoted. Thus, the development into
a modal auxiliary rather ties in with an auxiliarisation process that had already
affected the Old Norse cognate munu. Still, it is essential to note that lexical munen
obviously coexisted with mun functioning as a grammatical marker for some time,
in a scenario typically known as divergence (Hopper 1991: 24). Such a co-existence
is nicely illustrated by (10) and (22) (repeated in (47) and (48) for convenience):
(47) Ne munnde nohht tatt illke follc att crist comm till onn eore.
(?c1200, Ormulum, l. 9357 [OED, s.v. mone])
(48) Ne munnde nohht tatt illke follc habbenn mihht To follhenn all hiss
lare.(?c1200, Ormulum 9357 [MED])
If we accept the occurrences from the Ormulum as evidence for a brief period of
co-existence between lexical munen and modal mun, we can sketch the grammati-
calisation process as follows with the caveat that some of the incipient changes
are already anticipated in Old Norse, thus considerably facilitating the introduction
of mun as a future auxiliary and accelerating the further adoption of modal char-
acteristics. While Old English (ge)munan exclusively licensed nominal or clausal
complements, Middle English mun is predominantly used with infinitival comple-
ments, which is the syntactic prerequisite for its functioning as a modal auxiliary.
Due to the increasing frequency of such syntactic contexts, mun and the succeeding
infinitive are perceived less and less as a combination of two autonomous verbs but
rather as a construction-type (cf. Fischer 2007:193). Also, mun eventually loses
its property to license direct objects which might correlate with its replacement
by other verbs of Romance origin denoting the act of memory. This steady loss of
autonomy characteristic of grammaticalisation goes along with the decategorialisa-
tion of lexical mun as visualised in the formal cline of modality (see Figure 2).
Lexical > Vector > Auxiliary > Clitic

verb verb
Figure 2. Formal cline of modality (adapted from Fischer 2007:182)
The crucial point in the development of munen to a grammatical marker of

modality is constituted by the second stage where mun functions as a vector verb.
The earliest instances of infinitival constructions can be hypothesised to have been
ambiguous in that mun could be interpreted as either still possessing full lexical
semantics or as already forming a constituent part in a complex predication. As
for the lexical, i.e. pre-grammaticalised sense of mun in such constructions, we can
agree with Visser (1969:1441), who suggests a prehistoric meaning to intend,
which is clearly connected to the reconstructed meaning of the Indo-European
etymon *men- to think, to be mentally active. The semantic-pragmatic shift of
mun on its way to auxiliaryhood would then imply:
(49) I mun go
I think/consider + go > I intend to go > I will go > I must go
As soon as the infinitival complementation becomes compulsory, the auxiliary

stage is achieved. Consequently, mun is progressively affected by semantic bleach-
ing as it loses its syntactic independence to the infinitive. In those contexts in
which the former lexical sense of mun no longer provides a clue to the inter-
pretation of its function, the stage of ambiguity is left behind; instead, due to its
Matthias Eitelmann
receding metaphorical extension, the reanalysis of mun as a modal auxiliary

p
becomes overt. The following two sentences in which mone functions as a marker
of futurity are both examples of such contexts in which mone is without doubt
semantically bleached.
(50) I wene that we deye mone For hunger, is dere is so strong.
I fear that we will [have to] die of hunger, this drought is so strong.
(c1300, The Lay of Havelok the Dane 840 [MED, s.v. mnen v.(2)])
(51) Thine eris mone irk To here the syne that I cane vyrk
Your ears will irk [i.e. feel vexed] to hear the sin that I can work.
(a1400, Legends of the Saints xviii, 428 [DOST, s.v. mon])
In (50), the subject cannot possibly have the intention to die, and in (51), any
cognitive sense of mone is ruled out because of the inanimate subject mine eris. It
is particularly in 3rd person contexts that functional changes are triggered. Again,
mun shares a parallel development with other modal auxiliaries in this respect,
for they, too, started to develop more grammatical senses in exactly these con-
texts (see, e.g. Aijmer 1985:13f. for the development of will). The extension from
human to non-human subjects causes the change into a pure, i.e. non-volitional
future, with the grammaticalised meaning thus being no longer a merely prag-
matically induced implicature, but a semiotic unit independent of the older, more
lexical meaning (Diewald 2002:104).
As for a more precise sketch of the order in which the modal functions of mun
emerged, we have already observed a clear trend towards deontic modality, with
epistemic modality developing relatively late in the Early Modern English period.
An analysis of the Middle English and Early Modern English data retrieved from
the OED and Visser (1969: 1369) as well as the Middle English Dictionary and the
Dictionary of the Older Scots Tongue confirms this intuitional observation made in
the previous chapter (see Figure 3).
For the analysis of modal functions, 114 Middle English instances of mun/
man in the present tense were collected and contrasted with a total of 91 Early
Modern English instances of mun/man. In Middle English, mun/man serves as a
marker of futurity in 83% of all cases, which can be differentiated more accurately
into 25% indicating volitional future and 58% occurring in non-volitional con-
texts with a strong deontic sense. Pure deontic modality, i.e. without future refer-
ence, is given in 17%. This scenario changes drastically in Early Modern English
with expressions of pure deontic modality increasing to 54%. Future markers, on
the other hand, decline to 32%, with volitional contexts dropping to merely 8%.
Expressions of epistemic modality, which only evolved in Early Modern English,
already constitute 14%, thus outscoring mun/man in its function as a marker of
volitional futurity. However, we can safely assume that epistemic mun/man was
100%
17% 14%
90%
80%
70%
60% 54% Epistemic modality
58% Deontic
50% Future deontic
40% Future volitional
30%
20% 24%
25%
10%
8%
0%
Middle English Early Modern English
(n = 114) (n = 91)
Figure 3. Modal functions of mun in Middle English and Early Modern English
even more widespread than the numbers suggest as epistemic usages must have
been considerably more frequent in spoken speech than in the surviving written
records (cf. Warner 1993:174).
In this respect, then, mun/man also moved along the semantic cline of gram-
maticalisation (see Figure 4) to a respectable degree, following the general trend
observed by Langacker (1985) that meanings become more and more subjective
in the course of time. The development of epistemic modality from root modal-
ity takes its course as inferences, at first probably of a rather associative nature,
frequently re-occur so that conversational implicatures are routinised and finally
conventionalised (cf. Traugott 1989:50).
Fully lexical > Root modality: > Epistemic

dynamic/deontic
Figure 4. Semantic cline of modality (adapted from Fischer 2007:182)
To sum up, mun/man/maun has been grammaticalised in a process that is no

different from the grammaticalisation of the proper modal auxiliaries. Therefore,
a question that inevitably poses itself concerns the question which factors con-
tributed to the eventual loss of mun as a modal auxiliary even though it had the
best potential to go all the way down auxiliaryhood , or put differently, why mun
did not develop into a fully-fledged modal but remained in its embryonic status.
Matthias Eitelmann
5. The decline of modal mun/man
According to Lightfoot (1974:237), it was a mere coincidence that only the pre-
modals survived while all the other members of the preterite-present group
became obsolete. This accidental view, however, is not confirmed if we take the
grammaticalisation perspective into account.
First of all, the grammaticalisation perspective has shown that there are,
strictly speaking, no non-pre-modals. As the paradigmatic case study of mun
demonstrated, the former preterite-present verb considerably moved along the
various grammaticalisation clines and developed modal characteristics that
indeed match those of must, shall or may. Interestingly, this is also true for the
other lost preterite-present verbs as long as they fulfilled the syntactic prerequisites
for adopting modal functions. Witan which never took bare infinitival comple-
ments accordingly never developed modal characteristics and instead turned into
a full verb which was finally ousted by know at the beginning of the 20thcentury
(cf.OED, s.v. wit, v.1). In contrast, dugan and urfan both started their way down
the grammaticalisation clines as they ceased to take direct object complements
and occurred more and more frequently with bare infinitives. As befits their orig-
inal lexical semantics, dugen/dow be of value comes to mark dynamic modal-
ity expressing an even stronger ability than can as late as Early Modern English
(cf.52) while urfen/arf need is used as a marker of deontic modality already in
Old English (cf. 53).
(52) Ye may not, ye cannot, ye dow not want Christ.
(1637, Rutherford Letters I. 203 [OED])
(53) t u him on-drdan ne earft.
so that you need not fear him. (Beowulf l. 1674 [OED])
The grammaticalisation of practically the whole preterite-present group to modal

expressions of necessity and possibility now leads to a scenario of competition
and rivalry which is not untypical of grammaticalisation. Thus, in the domain of
deontic modality, mun has early rivals in shall and arf, and the scenario becomes
even more complicated with the development of mot/must into a marker of neces-
sity, which also takes its course in Middle English (cf. Solo 1977). Yet, while the
initial stages of grammaticalisation may allow diversity and variability, the num-
ber of alternants is eventually reduced since grammatical forms whose referential
meaning has bleached have no need for synonyms (Fischer 2007:186). Similarly
to dugen/dow or urfen/arf which were ousted by their competitors can/may and
must/need, mun was easily dispensable due to the much more frequent supra-
regional shall (cf. Plank 1984:311f.) Such a process of downsizing the range of
synonomy is exactly what Hopper & Traugotts notion of specialisation implies,
i.e. the process of reducing the variety of formal choices available (2003:116).
It is therefore no surprise that the formal choices within the domain of deontic
modality would be reduced, resulting in the loss of mun/man in the rising stan-
dard of southeastern English as early as 1500.
Contrary to Standard English, mun/man/maun gained a relatively strong
foothold in the northern dialects of English and was even further grammaticalised
as it developed contracted negatives in -nt. Yet, after its hey-day in the 18th and
19th centuries, mun/man/maun rapidly declined in the regional varieties as well;
even in Scottish English maun is now on its last legs soon to be replaced by huv
tae and the like (Britain 2007:118f.). On that score, maun meets the same destiny
as other modals once frequent in Older Scots such as dow and sall. What is more,
it follows a similar trend currently affecting its standard English equivalent must,
which is also increasingly replaced by periphrastic expressions such as have to or
need to (cf. Leech 2003).
One reason for the ousting of mun/man/maun in the regional varieties is surely
prescriptive pressure which makes language users favour the supraregional modal
auxiliaries, a choice which ultimately leads to the decreasing frequency of dialectal
variants. From a functional point of view, we must also concede that the opposi-
tions within the field of deontic modality are not clear-cut enough. Even though
the English Dialect Dictionary (s.v. mun) claims that there is a well-defined dif-
ference of meaning between [mun and must], with mun implying physical, must
moral necessity, such a distinction is most often not discernible in the sample
sentences provided. If, however, there is no well-established division of labour
between grammatical variants, and if the principles of economy, clarity and com-
municative efficiency are potentially violated, it is quite natural that in the end one
variant will lose out to another that is more frequent and more readily accepted.
What is important to emphasise, though, is that mun/man/maun was only
lost in a gradual process that was far from uniform across all English dialects.
Actually, there is no intrinsic reason why mun/man/maun was ousted by the
other modal auxiliaries on the contrary, it is not only as qualified as the other
members of the preterite-present subgroup but also displays a modality almost
as well-developed.
6. Conclusion
The aim of the present paper was to remember (ge)munan with respect to both its
lexical semantics in Old English and its modal functions in the post-Old English
periods. The study of this lost preterite-present verb has proven worthwhile for
four main reasons.
Matthias Eitelmann
First, Old English munan and its more frequent derivative gemunan are par-
ticularly interesting in that they denote the cognitive activities of thinking and
remembering in a different way from the Present-day English equivalents remem-
ber, recollect or recall. A contrastive analysis of the semantic content of (ge)munan
thus suggests different conceptualisations of memory in oral and literate memorial
cultures.
Second, the development of munan after the Old English period has illustrated
paradigmatically that can, shall, may and must were not the only preterite-presents
to emerge as modals. Obviously, the intrinsic semantic-pragmatic property char-
acteristic of the preterite-presents, i.e. the shift from a past action to its present
effect, qualifies both the preserved preterite-presents and the lost ones to develop
into modal auxiliaries, which all of them did to a certain extent.
Third, the investigation of munan as an evolving modal auxiliary showed it
to follow the expected paths of grammaticalisation. As the formal cline of gram-
maticalisation predicts, the originally lexical verb adopted modal characteristics
via a transitional stage in which the infinitive constructions can be ambiguously
interpreted, either as consisting of two independent verbs or as the first modifying
the second. In contexts in which the first interpretation is ruled out, it can safely
be assumed that metaphorical extension has set in resulting in the paradigmati-
cisation of the construction. The decategorialisation of mun to an auxiliary goes
along with syntactic restrictions so that infinitival complements become com-
pulsory. Moreover, as is expected from the semantic cline of grammaticalisation,
deontic modality precedes epistemic modality. In these respects, then, the case
study of mun strongly supports the tantalizing idea of unidirectionality (Tabor&
Traugott 1998:229).
Fourth, the present paper provided a more cautious approach to the loss of
mun not in the sense of Lightfoots cataclysmal change that left the core modals
isolated in a rather abrupt manner but in the sense of a gradual change. This
implies that mun (as well as the other lost preterite-presents urfan and dugan)
steadily adopted more and more modal characteristics, thus starting to compete
with other expressions of deontic modality.
Such a view that reveals the field of modality to have always been characterised
by variability and rivalry bears implications for current changes affecting English
modals. In particular, American English experiences the impending loss of shall
and may as well as a rapid decline of must in its functions as an expression of both
deontic and epistemic modality. However, this is not at all surprising with new
modal expressions constantly emerging (such as wanna or gotta, cf. Krug 2000) or
well-established periphrastic constructions extending their functional range (such
as have to which has recently come to be used as an expression of epistemic modal-
ity). In other words, we can observe a functional overlap similar to the situation
resulting from the early grammaticalisation of the preterite-present subgroup, a
scenario which will predictably lead to the ousting of some functionally equivalent
variants for reasons of economy and efficiency.
A full comprehension of the grammaticalisation of the pre-modals can only
be achieved if we take all verbs into account that had the potential to undergo
the auxiliarisation process. Instead of treating the now obsolete preterite-presents
marginally, their inclusion provides us with a more complete picture of this com-
plex change.
References
Aijmer, Karin. 1985. The semantic development of will. In Historical Semantics, Historical Word-
formation, Jacek Fisiak (ed.), 1121. Berlin: Mouton de Gruyter.
Aitken, Adam J. 1973. A Dictionary of the Older Scottish Tongue, Vol. IV: MN. Aberdeen:
Aberdeen University Press. [= DOST].
Assmann, Jan. 1992. Das kulturelle Gedchtnis. Schrift, Erinnerung und Identitt in frhen Hoch-
kulturen. Mnchen: Beck.
Barnes, Michael. 2008. A New Introduction to Old Norse, Part I: Grammar, 3rd edn. Exeter: Short
Run Press.
Birkmann, Thomas. 1987. Prteritoprsentia. Morphologische Entwicklungen einer Sonderklasse
in den altgermanischen Sprachen. Tbingen: Max Niemeyer.
Bolinger, Dwight L. 1980. Wanna and the gradience of auxiliaries. In Wege zur Universalienfor-
schung: Sprachwissenschaftliche Beitrge zum 60. Geburtstag von Hansjakob Seiler, Gunter
Brettschneider & Christian Lehmann (eds), 292299. Tbingen: Gunter Narr.
Bosworth, Joseph & Toller, Thomas Northcote. 1898. An Anglo-Saxon Dictionary. Oxford: OUP.
Bosworth, Joseph & Toller, Thomas Northcote. 1921. An Anglo-Saxon Dictionary. Supplement.
Oxford: OUP.
Britain, David. 2007. Grammatical variation in England. In Language in the British Isles, David
Britain (ed.), 75104. Cambridge: CUP.
Chafe, Wallace. 1986. Evidentiality in English Conversation and Academic Writing. In Evi-
dentiality: The Linguistic Coding of Epistemology, Wallace Chafe & Johanna Nichols (eds),
261272. Norwood: Ablex Publishing Corporation.
Clark Hall, J.R. 1960. A Concise Anglo-Saxon Dictionary, 4th edn. With a Supplement by Herbert
D. Merritt. Toronto: Toronto University Press.
Coates, Jennifer. 1983. The Semantics of the Modal Auxiliaries. London: Croom Helm.
Colman, Fran. 1992. A touch of (sub-)class? Old English Preterite-Present verbs. In History
of Englishes: New Methods and Interpretations in Historical Linguistics, Matti Rissanen,
Ossi Ihalainen, Terttu Nevalainen & Irma Taavitsainen (eds), 241261. Berlin: Mouton de
Gruyter.
Diewald, Gabriele. 2002. A model for relevant types of contexts in grammaticalisation. In
New Reflections on Grammaticalisation [Typological Studies in Language 49], Gabriele
Diewald& Ilse Wischer (eds), 103120. Amsterdam: John Benjamins.
Fai, Klaus. 1989. Englische Sprachgeschichte. Tbingen: Francke.
Fischer, Olga & van der Wurff, Wim. 2006. Syntax. In The History of the English Language,
Richard Hogg & David Denison (eds), 109198. Cambridge: CUP.
Fischer, Olga. 2007. Morphosyntactic Change. Functional and Formal Perspectives. Oxford: OUP.
Matthias Eitelmann
Hopper, Paul J. & Traugott, Elizabeth C. 22003. Grammaticalization. Cambridge: CUP.

Hopper, Paul. 1991. On some principles of grammaticization. In Approaches to Grammati-
calization, Vol. I: Focus on Theoretical and Methodological Issues [Typological Studies in
Language 19(1)], Elizabeth Closs Traugott & Bernd Heine (eds), 1735. Amsterdam: John
Benjamins.
Huddleston, Rodney. 1976. Some theoretical issues in the description of the English verb.
Lingua 40: 331383.
Kastovsky, Dieter. 1992. Semantics and vocabulary. In The Cambridge History of the English Lan-
guage, Vol. 1: The Beginnings to 1066, Richard M. Hogg (ed), 290408. Cambridge: CUP.
Krug, Manfred. 2000. Emerging English Modals. A Corpus-Based Study of Grammaticalization.
Krug, Manfred. 2003. Frequency as a determinant in grammatical variation and change. In
Determinants of Grammatical Variation in English, Gunter Rohdenburg & Britta Mondorf
(eds), 767. Berlin: Mouton de Gruyter.
Kuhn, Sherman M. 1975. Middle English Dictionary, Vol. 9: Metalli-Nywrbe. Ann Arbor MI:
University of Michigan Press. [=MED].
Langacker, Ronald W. 1985. Observations and speculations on subjectivity. In Iconicity in Syn-
tax [Typological Studies in Language 6], John Haiman (ed.), 109150. Amsterdam: John
Benjamins.
Leech, Geoffrey. 2003. Modality on the move: The English modal auxiliaries 19611992. In
Modality in Contemporary English, Roberta Facchinetti, Manfred Krug & Frank Palmer
(eds), 223240. Berlin: Mouton de Gruyter.
Lightfoot, David. 1974. The diachronic analysis of English modals. In Historical Linguistics, John
M. Anderson & Charles Jones (eds), 219249. Amsterdam: North Holland.
Lightfoot, David. 1991. How to Set Parameters: Arguments from Language Change. Cambridge
MA: The MIT Press.
McMahon, April. 1994. Understanding Language Change. Cambridge: CUP.
Miller, Jim. 2008. Scottish English: Morphology and syntax. In Varieties of English, 1: The British
Isles, Bernd Kortmann & Clive Upton (eds), 299327. Berlin: Mouton de Gruyter.
Mitchell, Bruce. 1985. Old English Syntax, Vol. I. Oxford: Clarendon Press.
Nagle, Stephen J. & Sanders, Sara L. 1996. Downsizing the preterite-presents in Middle English.
In Advances in English Historical Linguistics, Jacek Fisiak & Marcin Krygier (eds), 253261.
Oxford English DictionaryOnline, The (2011). Oxford: OUP. [= OED].
Plank, Frans. 1984. The modals story retold. Studies in Language 8: 305364.
Pokorny, Julius. 1959/1969. Indogermanisches etymologisches Wrterbuch, 2 Vols. Bern: Francke.
Prokosch, Eduard. 2009 [1939]. A Comparative Germanic Grammar. Surrey: Tiger Xenophon.
Solo, Harry Jay. 1977. The Meaning of *motan. A Secondary Denotation of Necessity in Old
English? Neuphilologische Mitteilungen 78: 215232.
Tabor, Whitney & Traugott, Elizabeth C. 1998. Structural scope expansion and grammatical-
ization. In The Limits of Grammaticalization [Typological Studies in Language 37], Anna
Giacalone Ramat & Paul J. Hopper (eds), 229272. Amsterdam: John Benjamins.
Traugott, Elizabeth C. 1989. On the rise of epistemic meanings in English: An example of
subjectification in semantic change. Language 65: 3155.
Upton, Clive, David Parry & Widdowson, J.D.A. 1994. Survey of English Dialects: The Dictionary
and Grammar. London: Routledge.
Visser, Frederikus T. 1969. An History of the English Language, Part III, First Half: Syntactical
Units with Two Verbs. Leiden: Brill.
Warner, Anthony R. 1993. English Auxiliaries. Structure and History. Cambridge: CUP.
Wright, Joseph. 1923. The English Dialect Dictionary, Vol. IV: MQ. London: OUP. [= EDD].
The emergence of modal meanings from haben
with zu-infinitives in Old High German*
Anne Jger
Leibniz Universitt Hannover, Germany
The grammaticalisation path from possession to obligation which describes

the development of haben to a marker of modality is well-established (cf. Heine&
Kuteva 2002), but opinions differ on when exactly these modal readings came
about. Haspelmath (1989) argues for a dating no earlier than Middle High
German, but a careful study of Notkers writings reveals evidence of modal
haben+ zu-infinitive in Old High German. Following Ebert (1976), this study
identifies four stages of grammaticalisation of haben with zu-infinitive. The pivotal
point is reached as soon as haben + zu-infinitive is combined with nominal
complements in the genitive or dative case which clearly evince modal meanings.
Notkers Old High German texts contain several instances of this use. Finally, the
comparison with the original texts demonstrates that haben + zu-infinitive does
not derive from Latin, but rather is used independently, even in Old High German.
1. Introduction
New verbal categories and paradigms dont just crop up from nowhere, but emerge
from very specific sources. In the case of verbal markers of modality, the gram-
maticalisation paths that lead to the emergence of modal meanings are similar
to each other in a number of languages. One such grammaticalisation path leads
from verbs of possession to markers of obligation and has been identified not
only in Indo-European languages such as German, English, Latin and Spanish, but
also in Nyabo, Koyo, Kagbo and Yoruba (Heine & Kuteva 2002:243244).
In Present-Day English, have is used in combination with to and an infinitive
in order to express meanings of obligation or necessity:
(1) You have to leave now.
* This study was made possible through the Interuniversity Attraction Poles Program
(Project GRAMIS P6/44) of the Belgian Science Policy. It is based on a paper presented at
the International Conference on Grammaticalization and (Inter)Subjectification in Brussels
in November 2010.
Anne Jger
In Present-Day German, haben + zu-infinitive can be used to encode meanings

of necessity or obligation, as well as meanings of possibility. Sentences (2) and (3)
are examples of these uses (cf. Helbig & Buscha 2005:112):
(2) Ich hab-e mit dir zu red-en.
I have-prs.1sg with you to talk-inf
I have to talk to you.
(3) Was hast du zu bericht-en?
What have.prs.2sg you to report-inf
What can you tell me?
While there is little doubt that have and haben encode modal meanings in Present-
Day English and Present-Day German respectively, there has been some discus-
sion about the circumstances surrounding the emergence of these modal uses.
Fischer (1997, 1994) claims that the change to modal meaning probably occurred
sometime between late Middle English and early Modern English, whereas Harris
and Campbell (1995) put this development sometime after the reanalysis of
English modal verbs to auxiliaries (which they date back to early Middle English at
the latest). Furthermore, Brinton (1991) and ecki (2010) maintain that have with
to-infinitival already codes modal meanings in late Old English (cf. Krug 2000:74;
OED 2011; Visser 2002:1477, 1481; Bosworth & Toller 1980:492). Clearly, there is
no consensus as to when the change took place, but the data seem to suggest that
the conditions necessary for its onset were reached either in late Old English or in
early Middle English.
In German, a group of modal verbs emerges at roughly the same time
namely during the transition from Old High German to Middle High German
(cf.Birkmann 1987). These modal verbs are mostly made up of former preterite-
presents that started expressing modal meanings. Out of the Old High German
preterite-presents wizzan, heigan, unnan, kunnan, thurfan, giturran, sculan,mugan,
muozzan and ganah (cf. Birkmann 1987:129) and the modal verb wellen, only
mgen, soln, mezen, kunnen, durfen and wllen (cf. Paul 2007:296) are used as
modal verbs in Middle High German.1 All six verbs share the same characteristics:
They express modal meanings, exhibit a special inflection that distinguishes them
from other verbs2 and combine with infinitives instead of zu-infinitives.
. Old High German lemmata are spelled conforming to Schtzeichel (1995). In cases where
Schtzeichel lists more than one variant of the lemma, the elaborate one is given (e.g.: wizzan
instead of wizan). Middle High German lemmata are spelled conforming to Lexer (1992). If
Lexer lists more than one variant of a lemma, only the first one is provided.
. The distinctive features in the inflection of the modal verbs go back to their status as
former preterite-presents. Even wllen, an old optative form, adopted a special inflection with
its integration into the paradigm (cf. Paul 2007:272274).
The emergence of modal meanings from haben with zu-infinitives in Old High German
The expression of modal meanings notwithstanding, these characteristics

are not shared by haben: It does not inflect like the modal verbs and it is only
combined with zu-infinitives not bare infinitives. In fact, the involvement of
the zu-infinitive is crucial to the grammaticalisation of haben + zu-infinitive to
a marker of modality since it adds another semantic layer to the meaning of the
phrase. The result is the emergence of modal readings and both zu and the infini-
tive contribute to this.
Abraham (2004:125) states that zu, purely on the basis of its lexical direc-
tional meaning, inevitably elicits a modal or future reading. The grammaticalisa-
tion of the infinitive also includes modal readings. Haspelmath (1989) traces the
semantic changes of the infinitive and postulates the grammaticalisation channel
in Figure 1:3
Benefactive
Irrealis- Irrealis- Realis- (Realis-
Allative Purposive
directive potential non-factive factive)
Causal
Figure 1. Semantic changes involved in the grammaticalisation of the infinitive

(Haspelmath 1989:298)
In the grammaticalisation of modal meanings, the transition from irrealis-

directive to irrealis-potential meaning is the pivotal step. If Haspelmath (1989:299)
is to be believed, this step is taken sometime in Middle High German because it is
only then, according to him, that the infinitive is used in complement clauses with
irrealis-potential modality. Following this, haben should not appear in conjunc-
tion with the zu-infinitive in Old High German texts, and if it did, there should not
be any modal meaning involved.
Demske (2001) and Kotin (1999) argue against Haspelmaths position and cite
passages from Tatian as modal that include haben + zu-infinitive:
(4) simon, ih hab-en thir sih uuaz ciqued-an-ne
Simon I have-prs.1sg you something to.tell-inf-dat (T 238.16)
Simon, I have to tell you something
Simon, ich habe dir etwas zu sagen (Demske 2001:70)
. Haspelmath (1989:298299) defines complements of manipulative and desiderative verbs

as irrealis-directive, complements of evaluative predicates and modal predicates as irrealis-
potential, complements of verbs of thinking or of utterance as realis-non-factive, and comple-
ments of verbs of cognition or of evaluative predicates as realis-factive. Note that complements
of evaluative predicates may be either irrealis-potential or realis-factive.
Anne Jger
In Present-Day German, this example of haben with zu-infinitive would indeed

express obligation or necessity. Nevertheless, there are two arguments as to
why one might not accept a quote like this from Tatian as a true case of modal
haben + zu-infinitive. First, one might argue that the actual reading of (4) may
very well be:
(5) Simon, I have something to tell you.4
In this case, modal meanings would not necessarily be involved. Rather, a mean-
ing of possession as it is caused by haben as a full verb would be likely. Any modal
meaning could very well only be implied or even just be inferred in retrospect
(cf.Krug 2000:80 and Mitchell 1985:402 for a comparable supposition in regard
to modal have to in Old English). If this should be true, the Old High German
phrase would be ambiguous at best (between todays modal and the original pos-
sessive reading).
Second, one might assume that haben + zu-infinitive is no more than a trans-
lation from Latin and that it does not exist as an Old High German expression in
its own right. In fact, the Latin version of example (4) reads:
(6) Simon habeo tibi aliquid dicere. (T 238.16)
The construction habeo + infinitive5 allows for a modal reading and Coleman
(1971:216217) and Adams (1991) describe the development of the meaning of
habeo + infinitive as illustrated in Figure 2:
Purpose/destination > Possibility > Obligation/necessity > Futurity > Futurity-in-the-past >
Conditioned unreality
Figure 2. The semantic change of habeo + infinitive
According to Adams (1991) and Coleman (1971), the transition from pur-
posive to potential readings of possibility and of obligation or necessity that is
under discussion for Old High German haben + zu-infinitive is already attested
in habeo + infinitive from classical Latin.6 It follows that all competent Old High
German authors who work regularly with classical Latin texts should be aware
. Indeed, Bosworth and Toller (1980:492) allows for both readings of the Old English
translation of example (4), which is quoted under the heading to have as a duty or thing to
be done.
. In accordance with most dictionaries, Latin verbs are cited in the inflected form of 1st
person singular present indicative active.
. Coleman (1971) also touches on the fact that there is an even older construction +
infinitive from Ancient Greek that may have influenced the rise of habeo + infinitive. While
this is certainly interesting, it will play no further part in this study.
of this. Semantic interference must thus be seen as a very real possibility. ecki
(2010:143) also comments that Latin may have had some impact on the rise
of obligative meanings in English have with to-infinitive. In light of these facts,
the status of Old High German haben + zu-infinitive is controversial. While the
Tatian texts show a higher degree of autonomy from Latin than glosses like the
Benediktinerregel, it is difficult to judge the true meaning of a phrase that is as
ambiguous as haben + zu-infinitive when it appears as a translation of habeo +
infinitive. In order to prove that there are indeed some first instances of haben+
zu-infinitive with modal meaning in Old High German, three separate points
have to be made. First, we have to be certain that haben is not used as a full verb
indicating possession and that the zu-infinitive is not just an attribute. Second, it
has to be shown that many of the instances where haben + zu-infinitive is used are
indeed modal in meaning, and, third, it must be verified that these cases are not
simply verbatim translations from Latin. Hence an in-depth study of the situation
in the crucial language period is needed.
The choices of material are limited since most writers had gone back to w riting
in Latin by the end of the Old High German period and there were only a few who
still used the vernacular at the beginning of the eleventh century (cf. Young &
Gloning 2004:26). One of these authors was Notker III. of St Gallen, a teacher
at the school of his monastery. He translated classical Latin texts into Old High
German and wrote commentaries on them in order to give his students the chance
to learn about the complex topics of the artes liberales in their mother tongue
(cf. Braungart 1987; Sonderegger 1987; Hellgardt 1979). Although Notker kept
his translations close to their Latin originals, he made sure that they were phrased
in appropriate Old High German. The glossed passages therefore allow for well-
founded speculation about specific meanings of individual lemmas, while the
commentaries that deviate from the Latin text provide an impression of the state-
of-affairs of Notkers original Old High German.
Five of Notkers longer Old High German texts remain intact: De Nuptiis
Philologiae et Mercurii, De Interpretatione, Categorieae, De Consolatione Philoso-
phiae and the Psalter.7 But since several of the manuscripts of Notkers Psalter were
written by scribes in the 12th century and show linguistic features (e.g. regular use
of Perfect) that are characteristic of that time (cf. Tax 1979: XVIXIX, XXXI), the
Psalter is not included in this study. All in all, there are 901 instances of haben in
the other four texts (cf. Sehrt & Legner 1955), but only 15 of these combine haben
and zu-infinitive. That adds up to 1.66% of all instances of haben. Following this,
. Sonderegger (2003:139) estimates the number of Old High German lemmas that can be
found in these texts to be about 8000.
Anne Jger
it is immediately apparent that haben + zu-infinitive was not yet all that frequent
in use during Notkers times as was to be expected. As a matter of fact, haben is
mostly used with nominal complements, but the instances where it is associated
with zu-infinitives are instructive for the way how this came about.
From his estimation of the situation in Early New High German, Ebert (1976)
draws several conclusions about the historical development of modal character-
istics noticeable with haben. He hypothesises that haben must have appeared in
the meaning possess, have in a shared-object construction at first, must then
have appeared in construction with verbs that take accusative object[s], with a
blending of shared-object and modal characteristics only to be used as a modal
predicate with the infinitive of intransitive verbs or verbs with dative or genitive
objects after that (Ebert 1976:113). Although Notkers texts alone do not allow
for conclusions about the diachronic order in which these stages were reached, it
is possible to identify all three of them among the examined material. This puts
Haspelmaths timeline into question and provides a basis for the re-evaluation of
the status of haben + zu-infinitive in Old High German.
2. Different uses of haben + zu-infinitive in late Old High German
Type 1: [haben + [nominal complement (acc.) + [zu-infinitive]]]

Originally, haben is used as a full verb indicating possession. The item in the
possession of the subject appears in the accusative case and sometimes there is
a zu-infinitive involved as well. Example (7) from De Consolatione Philosophiae
illustrates this use:
(7) be d sia uula bechnnest s hb-o h
If you them well recognize.prs.2sg so have-prs. 1sg I
zt tr dia uurn ze_og-en-ne
time.acc.sg you art.acc.sg truth.acc.sg to_show-inf-dat
(Tx 142, 89; cf. P 168,22)
If you recognize them well, (then) I have time to show you the truth.
Obviously, cases like this cannot be regarded as examples of haben + zu-infinitive

because the zu-infinitive is not governed by haben but by the nominal complement
(cf. Ebert 1976:113 stage 1). Type (1) rather comprises haben + [nominal comple-
ment + [zu-infinitive]]. A good portion of the instances of haben in Notkers writ-
ings conform to this type and there is much to be learned from including these
cases in the survey, even though they do not qualify as haben + zu-infinitive.
The zu-infinitives in clauses that are governed by nominal complements point
to conditions that have not been realized yet and are irrealis-prospective. This
irrealis-prospective or irrealis-directive quality (cf. Haspelmath 1989:298) is most

noticeable in sentences in which the zu-infinitive depends on nouns such as uullo
(will), which express a volition of future realisation. Other nouns do not evoke the
same prospective quality. Some of them induce irrealis-potential readings instead.
This is important for the reading of the zu-infinitives, as will be demonstrated
below.
Among the nouns found in complement position in Notkers writings are:
uullo (will/desire), geuult (power), zt (time), ugen (eyes) and geuunehite
(habit).8 While uullo is mostly irrealis-prospective and geuunehite habitual, zt,
ugen and geuult allow readings of potentiality. Such a modal notion is especially
prevalent in geuult. Consider:
(8) Ter mn ne=hb-et nehin-en geuult
art man.nom.sg.m neg=have-prs.3sg neg.one-acc.sg power.acc.sg
rche ze_sn-ne
rich to_be.inf-dat (Tx 76, 2627; cf. P 88,19)
Man does not have the power to be rich.
The power to do something might still be irrealis-prospective, but not the power to
be something. In combination with infinitives of stative verbs like sein, the mean-
ing is no longer prospective or final it is irrealis-potential. So, while sentences
conforming to patterns of type (1) are no examples of haben + zu-infinitive they
prove that it is already possible to use zu-infinitives in contexts of potentiality in
Old High German.9
Type 2: [haben + [nominal complement (acc.) + zu-infinitive]]
Type (2) involves haben in combination with a nominal complement in the
accusative case and the zu-infinitive. Compare:
(9) Uuz hb-o h n fne den
What.acc have-prs.1sg I now of art.dat.pl
lge-breu-en ze_sg-en-ne ?
lie-letters-dat.pl to_say-inf-dat (Tx 25, 1012; cf. P 31,4)
What do I have to say now about the deceitful letters ?
. According to Brinton (1991:15), nominal objects such as power, will and time are also
common in Old English have + object + to V constructions.
. This use of haben with zt or geuult and zu-infinitive is not exclusive to Notker. There
are also several instances in Tatians Evangelienharmonie from the 9th century: e.g. Lk 1,57;
Lk 5,24; Joh 1,12 (cf. Sievers 1966:18, 33, 76). For further evidence see Behaghel (1924:334).
Anne Jger
At first glance, there is no big difference between example (9) and examples (7)
and (8), since all of them have haben in combination with an accusative object and
a zu-infinitive. But whereas the zu-infinitives in (7) and (8) are clearly governed
by the nouns in complement position, the structure in (9) is ambiguous to that
effect that the accusative may be governed either by the infinitive or by haben. At
the same time, there is ambiguity between the irrealis-prospective and the irrealis-
potential meaning. It is only the context that makes it clear that a non-modal read-
ing is to be preferred over a modal one. This is not always the case. Haben may
also appear in combination with the zu-infinitive and a complement clause in the
function of an accusative object. Compare:
(10) s hb-est tu n ze_bechnn-en-ne. uur d
so have-prs.2sg you now to_recognize-inf-dat where you
sa hl-n sl-st
them get-inf shall.sbjv-prs.2sg (Tx 148, 89; cf. P 175,30)
so you have to realize now, where you ought to get them.
In a sentence such as this, which lacks a concrete object that can be in possession
of the subject, there is no doubt about the irrealis-potential meaning: Haben + zu-
infinitive clearly expresses obligation or necessity and is used in the same fashion
as sculan or mugan (must) with infinitive.
Type 3: [haben + [[nominal complement (genitive/dative)] + zu-infinitive]]
Type (3) has haben in combination with a complement in the genitive or
dative case and a zu-infinitive. For example:
(11) Tr st fne du slg. uunda er
pron.m.nom.sg be.prs.3sg therefore blessed because he
dz frder10 neht-es ne=hb-et ze_gr-on-ne
further nothing-gen.sg neg=have-prs.3sg to_crave-inf-dat
(Tx 191, 1415; cf. P 239,19)
He is blessed therefore, because he has to crave nothing else.
Here, a different set of circumstances is in place and the syntactic structure is no

longer ambiguous: If the nominal complement were still dependent on haben,
it would have to appear in the accusative, but this is not the case: nehtes is a
genitive and does not agree with the case requirements of haben. Instead, there
is agreement with the requirements of the zu-infinitive ze_gronne, which may
. Sehrt & Legner (1955:58) rate dz frder as an adverb instead of classifying dz as an

accusative (with nehtes in the position of genitive attribute). The syntactic structure of the
sentence supports this interpretation.
govern either accusatives or genitives. Thus it is obvious that the zu-infinitive

does not depend on the nominal complement any longer. The opposite is true:
the nominal complement is now governed by the zu-infinitive. This has repercus-
sions on the semantic level, as the irrealis-potential reading takes hold over the
irrealis-prospective one.
Sentences (12) and (13) have a nominal complement in the dative case:
(12) N hb-est tu mr ze_dnch-n-ne. lso dr

Now have-prs.2sg you me to_thank-inf-dat like pron.nom.sg.m
frmide got nuz-et.
foreign good.acc.sg use-prs.3sg (Tx 49, 89; cf. P 59,11)
Now you have to thank me, like someone who uses goods that belong to

someone else.
(13) Nh ten fter nh chm-ent-n. ne=hb-et

Nor art.dat.pl after after come-ptcp-dat.pl neg=have-prs.3sg
s ze_dnch-on-ne
she to_thank-inf-dat (Tx 271, 23; cf. P 362,5)
Nor must she thank those who follow her
Both mr in (12) and ten after nh chmentn in (13) are datives functioning as
an indirect object. In cases like this, where there is only an indirect object, the
zu-infinitive takes the complement position, so that haben + zu-infinitive signi-
fies the obligation to act to the benefit or detriment of the indirect object. This
evokes a reading of necessity. The zu-infinitive is no longer irrealis-prospective,
but clearly irrealis-potential. There are only these two instances in Notkers writ-
ings in which haben has a zu-infinitive that does not govern accusative objects,
and both instances include t(h)ankn. But the mere fact that haben + zu-infinitive
is used with verbs like t(h)ankn shows that a new stage of grammaticalisation
has been reached. At the same time, it is clear that even in Old High German the
zu-infinitive was not confined to irrealis-prospective meanings and that haben +
zu-infinitive had already acquired a modal reading.
Type (4): [haben + zu-infinitive]
In the wake of t(h)ankn, another type of haben + zu-infinitive is introduced
that can be characterized as type (4). It includes zu-infinitives of intransitive verbs.
Consider the following example from the Alexanderlied:
(14) Hren, wir ne hab-en niuht ze bt-en-e,

Gentleman.nom.pl we neg have-prs.1pl not to wait-inf-dat
(Kinzel 1884: 76)
Gentlemen, we dont have to wait
Anne Jger
The first instances of type (4) of haben + zu-infinitive appear in Middle High
German (cf. Behaghel 1924:333 for further examples). The three types of haben
+ zu-infinitive that are attested in Notkers writings thus conform closely to the
three stages of diachronic development defined by Ebert (1976:113). The fact that
haben was already used with objects (in genitive or dative case) that are governed
by the zu-infinitive clearly demonstrates that haben + zu-infinitive was already
well on its way to grammaticalisation in late Old High German.
These findings are in agreement with the timelines of Brinton (1991) and
ecki (2010): Like have with to-infinitival starts coding modal meanings in Old
English, the first instances of haben + zu-infinitive with modal meaning appear in
late Old High German. Note that this is roughly at the same time that haben + past
participle is first used for conveying anteriority or perfect. And while this change,
like any other, progressed over long periods of time, it is Notker once again who is
credited with the first unambiguous examples of haben + past participle (cf. hl
2009:275).
But it still has to be taken into account that Notkers Old High German writ-
ings are translations and commentaries of texts that were originally written in
post-classical Latin. So, in order to judge the true degree of grammaticalisation
of haben + zu-infinitive in late Old High German, we need to make certain that
this apparent innovation is more than a mere literal translation of a Latin phrase
or construction. There are two ways to demonstrate this: First, by showing that the
corresponding Latin phrase(s) may also be translated by other means, and second,
by verifying that Old High German haben + zu-infinitive is also used without
prompting from the original Latin text.
3. Modes of expression in Latin
There are two phrases in Latin that correspond to German haben + zu-infinitive.
One is habeo + infinitive (cf. Coleman 1971) as seen in the previous example from
Tatian, and the other is habeo + gerundive (cf. Eilers 1992:142; Pertsch 1994:279).
If German haben + zu-infinitive were merely a copy of a Latin phrase or construc-
tion, one of these two would be the source. But a closer look at the corpus data
reveals that there is no such correspondence in Notkers translations. Although the
Latin original includes several instances of habeo + infinitive/gerundive, none of
them is translated with haben + zu-infinitive. Notker uses other means of expres-
sion, e.g. modal verb + infinitive. Consider (15):
(15) S i animalia inquam considerem. qu habent aliquam uolendi nolendique
naturam
be h tu lb-end-en chus-o chd
If I art.acc.pl.n live-ptcp-acc.pl.n choose-prs.1sg say.prt.1sg
h. tu natrlicho mg-en uull-en.

I art.nom.pl.n naturally can-prs.3pl want-inf
nde ne=uull-en
and neg=want-inf (Tx 166, 811; cf. P 200,26)
If I consider the animals, I said, that may want or not-want by nature
This offers two important insights: On the one hand, the fact that Notker uses
modal verbs in order to convey the meaning of habeo + infinitive/gerundive
demonstrates that he regards the Latin phrase as modal. This shows that habeo +
infinitive/gerundive does indeed fulfil similar functions as haben + zu-infinitive
in its modal use. On the other hand, it must be noted that the emergence of
haben + zu-infinitive in its new modal reading cannot be caused by a lack of
expressions appropriate for the translation of habeo + infinitive/gerundive if
modal verb + infinitive is available.
Since none of the instances of haben + zu-infinitive as used by Notker is
based on either habeo + gerundive or habeo + infinitive, the Old High German
phrasing cannot be caused by trying to recreate the Latin counterpart. As a matter
of fact, there is not just one Latin phrase or construction that Notker translates
by using haben + zu-infinitive, but a variety of them. One that actually employs
habeo is habeo + object (accusative):
(16) Habes gratiam. uelut usus alienis.

N hb-est tu mr ze_dnch-n-ne.
Now have-prs.2sg you me to_thank-inf-dat
lso dr frmide got nuz-et.
like pron.nom.sg.m foreign good.acc.sg use-prs.3sg
(Tx 49, 89; cf. P 59,11)
Now you have to thank me, like someone who uses goods that belong to
someone else.
Pertsch (1994:278) observes that Latin habeo + object (accusative) is used as an

emphatic paraphrase of a verbal notion. Habes gratiam can thus be understood as
a reinforced expression of to thank. Gegenschatz & Gigon (1969:4749) translate
habes gratiam with Present-Day German Du schuldest Dank (you owe thanks/
gratitude). And of course there is a close semantic correspondence between owe
and must which is also reflected in the fact that Old High German sculan is rou-
tinely used to express both meanings.
Example (16) also shows that Notker is in the business of adequately express-
ing the meaning of the Latin original rather than merely glossing its syntax.
Otherwise the translation would involve dank habn instead of ze_thankn
habn and read be rewarded for instead of have to thank (cf. Schtzeichel
1995; Kbler 1993).
Anne Jger
There are other examples where Notker uses haben + zu-infinitive. (17) also
plays on the semantic adjacency between owe and must:
(17) Nihil uero debet posterioribus.
Nh ten fter nh chm-ent-n.
Nor art.dat.pl after after come-ptcp-dat.pl
ne=hb-et s ze_dnch-on-ne
neg=have-prs.3sg she to_thank-inf-dat (Tx 271, 23; cf. P 362,5)
Nor must she thank those who follow her
Like sculan, Latin debeo can be used both as a full verb and as a modal auxiliary. In
its full verb use, debeo means to owe. As an auxiliary, it means must and is usually
translated with sculan + infinitive, as can be seen in example (18):
Sed cum in minimis quoque rebus inquit. uti in timeo placet nostro platoni.
(18)
diuinum debeat implorari presidium. quid nunc faciendum censes. ut
mereamur reperire sedem illius summi boni?
Sd ber platoni dnchet in_snemo boche timeo. dz man ih in lzzeln
dngen sle gtes hlfo flhn. uuz st ns ze_tonne dnne. uunest tu. dz
uur irfren mozn da huestt. tes frdersten gotes ?
(Tx 148, 1119; cf. P 176,6)
dz man ih in lzzeln dingen
that one even in little.dat.pl thing.dat.pl
sl-e gtes hlfo flh-n
shall.sbjv-prs.3sg god.gen.sg help.gen.sg desire-inf
that one should desire gods help even in little things
By using ze_thankon haben for debeo in (17), Notker makes the transition from
a reading of debt with inferences of obligation to a meaning of necessity. As
demonstrated above, the syntactic structure leaves no doubt about this interpreta-
tion (cf. (12)).
Another Latin source for haben + zu-infinitive includes attinet + infinitive:
(19) Nam quod attinet de compositis falso literis dicere
Uuz hb-o h n fne
What.acc have-prs.1sg I now of
den lge-breu-en ze_sg-en-ne ?
art.dat.pl lie-letters-dat.pl to_say-inf-dat (Tx 25, 912; cf. P 31,4)
What do I have to say now about the deceitful letters ?
Literally, attinet de dicere would have to be translated with it is of impor-

tance to say about. In this case, Notkers hbo ze_sgenne is less modal and
more possessive in meaning (cf. (9)). The same cannot be said about habest
ze_bechnnenne in (20):
(20) nunc superest ut agnoscas.

s hb-est tu n ze_bechnn-en-ne. uur d
so have-prs.2sg you now to_recognize-inf-dat where you
sa hl-n sl-st
them get-inf shall.sbjv-prs.2sg (Tx 148, 69; cf. P 175,30)
so you have to realize now, where you ought to get them
Latin superest ut means it is left that and implies some kind of necessity. Notkers
use of hbest ze_bechnnenne without a concrete object reflects this. Conse-
quently, as stated above, there is no doubt about the irrealis-potential reading of
this example.
Examples (16), (17), (19) and (20) prove that there are several different sources
for Notkers haben + zu-infinitive. And, contrary to popular belief, none of the
sources is Latin habeo + infinitive/gerundive. Instead, the Latin phrases which
Notker translates with haben + zu-infinitive all describe some kind of predisposi-
tion to an action that can be interpreted as an obligation or a necessity. This, rather
than a certain syntactic structure, is the common denominator between the differ-
ent Latin expressions.
In addition to these examples, there is even one instance of haben + zu-
infinitive that has no apparent source in Latin since it appears in one of Notkers
commentaries on the original text. This example is ambiguous between the irrealis-
prospective and the irrealis-potential reading:
(21) ne strt ne=hb-et si neht ze_ton-ne

without conflict neg=have-prs.3sg she nothing to_do.inf-dat
(Tx 55, 2324; cf. P 66,3)
without conflict, she does not have anything to do
without conflict, she does not have to do anything
The ambiguity stems from the indistinct status of neht, which may be either an
indefinite pronoun in the accusative case or an adverb. If neht were an indefinite
pronoun, this would be an example of haben + complement (acc.) + zu-infinitive
which in turn could express an irrealis-prospective or an irrealis-potential mean-
ing. But if neht were an adverb, this would be another instance of haben with a
zu-infinitive in complement position. In this case, the reading would have to be
irrealis-potential.11 There are good reasons for both interpretations and the con-
text allows for both of them as well.
. Sehrt & Legner (1955:391) classify neht as an indefinite pronoun or a pronominal adverb
without case distinction.
Anne Jger
In evaluating the importance of (21), one important fact has to be taken into
account: While there is ample evidence of haben + zu-infinitive in Notkers Old
High German without explicit syntactic counterparts in Latin, the lack of an iden-
tifiable Latin source is no ultimate proof for freedom from Latin influence. The
library of the monastery of St Gallen offered Notker access to Latin commentar-
ies on all kinds of classical texts, and it stands to reason that he consulted them
when he translated and interpreted the important works for his students. Furrer
(1971:18) referring to both content and syntax maintains that Notkers com-
mentaries in De consolatione philosophiae are based at least in part on the Latin
writings of Remigius and other scholars. Therefore, a certain degree of interfer-
ence from Latin is possible in all parts of Notkers writings although the exact
extent cannot be determined. Nevertheless, the fact that Notker uses haben + zu-
infinitive in passages in which he is free from the need to translate word for word,
plus the fact that even those passages which paraphrase Latin have quite different
Latin sources, all point to a high degree of independence from Latin. This is proof
that even in Old High German, haben + zu-infinitive is a phrase in its own right
and is more than a verbatim translation of Latin.
4. Old High German sein + zu-infinitive
Haben is not the only Old High German verb that develops modal meanings when
it is combined with zu-infinitives. The same happens with sein (to be),12 but at a
much faster pace so that Notkers writings already offer quite a collection of exam-
ples for sein + zu-infinitive. In De consolatione philosophiae for example, there are
as many as 48 instances of sein + zu-infinitive just in 3rd person singular present
indicative (cf. Sehrt & Legner 1955; Tax 198690). Many of these appear as trans-
lations of Latin sum + gerundive, but they are also quite often used without Latin
prompting. Compare:
Latin sum + gerundive and Old High German sein + zu-infinitive:

Quorum quidem exercitus tametsi numerosus est. tamen spernendus est.13
(22)
Tro hre no s mchel ne=st.
Their army.nom.sg never so mighty neg=be.prs.3sg
. Likewise, Old English wesan/beon, the counterpart of sein, is used for expressing neces-
sity and obligation (cf. OED 2011; Visser 2002:14451446).
. Example (22) is also cited in Eilers (1992:142).
z ne=s ze_uerches-en-ne.
it neg=be.prs.sbjv.3sg to_reject-inf-dat (Tx 17, 46; cf. P 20,29)
Their army is never so mighty, that it could not be rejected
Sed quoniam descendunt in te iam fomenta mearum rationum! puto vten-
(23)
dum esse paulo ualidioribus. s. remediis.
Uunda dh th u ingn-t mn-e rd-.
Since you then already enter-prs.3pl my-nom.pl speech-nom.pl
mt ten h th nz n fsc-ta. s st
with art.dat.pl I you until now sooth-pst.1sg so be.prs.3sg
n ze_uhen-ne uun-o h
now to_turn.to.inf-dat think-prs.1sg I
ze_strch-er-en rd-n.
to_strong-cpv-dat.pl speech-dat.pl (Tx 76, 711; cf. P 87,29)
Since my speeches, with which I have soothed you until now, are now sink-
ing in, we should now turn to stronger arguments, I think.
Original Old High German sein + zu-infinitive:

(24) Tra nh st tz ze_uuz-en-ne. tz status
Furthermore be.prs.3sg that to_know-inf-dat that status
nde constitutio. l in st
and constitutio all one be.prs.3sg (Tx 60, 67; cf. P 70,26)
Furthermore, one must know that status and constitutio are one and
thesame
It is important to note that the zu-infinitive in these cases is no longer confined

to irrealis-prospective meanings, but takes on irrealis-potential meanings as well.
In fact, sein + zu-infinitive is already established as a means for expressing neces-
sity in Notkers writings (cf. Nf 1979:180181). This is further proof that the Old
High German infinitive is not restricted to irrealis-prospective meanings and it
indicates where haben + zu-infinitive is headed.
5. Summary
Although the grammaticalisation of haben + zu-infinitive is still in its early stages

in (late) Old High German, and even though haben + zu-infinitive does not
appear very often in this early period of development, there are some instances of
it in Notkers writings that are clearly modal. This means that Haspelmaths time-
line is not entirely correct.
Anne Jger
With sein the use of the zu-infinitive with irrealis-potential reading is already
established in Notkers Old High German, and even with haben, there are clear
signs that the zu-infinitive is no longer simply irrealis-prospective in meaning. In
most cases, there is at least some degree of ambiguity and sometimes even actual
modality involved.
The emergence of the modal meanings can be traced by looking once more at
the different types of appearances for haben + zu-infinitive (see Figure 3).
Type (1): [haben + [nominal complement (acc.) + [zu-infinitive]]]

Type (2): [haben + [nominal complement (acc.) + zu-infinitive]]
Type (3): [haben + [[nominal complement (gen./dat.)] + zu-infinitive]]
Type (4): [haben + zu-infinitive]
Figure 3. The four stages of grammaticalisation of haben with zu-infinitive
Type (1) is no example for haben + zu-infinitive. But, depending on the noun
in the position of the object, sentences of this type may encode weak modal mean-
ings. Type (2) is structurally and semantically ambiguous. Since the object may
depend on haben or on the zu-infinitive, the irrealis-prospective and the irrea-
lis-potential readings both offer equally sound interpretations. Type (3) involves
haben + zu-infinitive in combination with a complement in the genitive or dative
case. Due to the case requirements of haben, the nominal complement must
depend from the zu-infinitive. This evokes a reading of necessity or obligation.
Type (4) is not yet attested in Notkers Old High German texts. It involves zu-
infinitives of intransitive verbs that are used without any nominal complements.
Examples of this type possess a strong reading of necessity or obligation and are
unambiguously modal.
The pivotal point in the process of grammaticalisation is reached as soon as
haben + zu-infinitive appears with nominal complements in the dative or genitive
case. This is where ambiguity ends, where haben is well on its way to an auxiliary
and where haben + zu-infinitive turns into a marker of modality.
Sources
Gegenschatz, Ernst & Gigon, Olof (eds). 1969. Boethius: Trost der Philosophie: Lateinisch und
Deutsch, 2nd edn. Zrich: Artemis.
Kinzel, Karl (ed.). 1884. Lamprechts Alexander nach den drei Texten mit dem Fragment des
Alberic von Besanon und den lateinischen Quellen [Germanistische Handbibliothek 6].
Halle: Waisenhaus.
[P] = Piper, Paul (ed.). 1882. Die Schriften Notkers und seiner Schule, Vol. 1: Schriften philoso-
phischen Inhalts [Germanischer Bcherschatz 8]. Freiburg: Mohr.
Pertsch, Erich (ed.). 1994. Langenscheidts Handwrterbuch Lateinisch Deutsch. Berlin:

Langenscheidt.
Sievers, Eduard (ed.). 1966. Tatian. Lateinisch und altdeutsch mit ausfhrlichem Glossar [Bib-
liothek der ltesten Deutschen Literatur-Denkmler 5], 2nd edn. Paderborn: Schningh.
[T] = Masser, Achim (ed.). 1994. Die lateinisch-althochdeutsche Tatianbilingue Stiftsbibliothek St.
Gallen Cod. 56 [Studien zum Althochdeutschen 25]. Gttingen: Vandenhoeck & Ruprecht.
Tax, Petrus W. (ed.). 1979. Notker der Deutsche. Der Psalter. Psalm 150 [Althochdeutsche Text-
bibliothek 84. Die Werke Notkers des Deutschen 8]. Tbingen: Niemeyer.
[Tx] = Tax, Petrus W. (ed.). 198690. Notker der Deutsche. Boethius, De consolatione Philosophiae
[Altdeutsche Textbibliothek. Die Werke Notkers des Deutschen 13]. Tbingen: Niemeyer.
References
Abraham, Werner. 2004. The grammaticalization of the infinitival preposition: Toward a theory
of Grammaticalizing Reanalysis. Journal of Comparative Germanic Linguistics 7: 111170.
Adams, John N. 1991. Some neglected evidence for Latin habeo with infinitive: The order of the
constituents. Transactions of the Philological Society 89: 131196.
Behaghel, Otto. 1924. Deutsche Syntax: Eine geschichtliche Darstellung, Vol. 2: Die Wortkassen
und Wortformen: B: Adverbium, C. Verbum [Germanische Bibliothek 1: Sammlung ger-
manischer Elementar- und Handbcher 10(2)]. Heidelberg: Winter.
Birkmann, Thomas. 1987. Prteritoprsentia: Morphologische Entwicklungen einer Sonderklasse
in den altgermanischen Sprachen [Linguistische Arbeiten 188]. Tbingen: Niemeyer.
[Bosworth/Toller] = Toller, Thomas Northcote (ed.). 1980. An Anglo-Saxon Dictionary. Based
on the Manuscript Collections of the late Joseph Bosworth. Supplement. Oxford.: OUP (Repr.
1st ed. 1921).
Braungart, Georg. 1987. Notker der Deutsche als Bearbeiter eines lateinischen Schultextes:
Boethius De Consolatione Philosophiae. Zeitschrift fr deutsche Philologie 106: 215.
Brinton, Laurel J. 1991. The origin and development of quasimodal have to in English, paper
presented at the workshop on The Origin and Development of Verbal Periphrases, 10th
International Conference on Historical Linguistics, Amsterdam, August 16, 1991. http://
faculty.arts.ubc.ca/lbrinton/haveto.pdf (November 2011).
Coleman, Robert. 1971. The origin and development of Latin habeo + infinitive. The Classical
Quarterly 21: 215232.
Demske, Ulrike. 2001. Zur Distribution von Infinitivkomplementen im Althochdeutschen.
In Modalitt und Modalverben im Deutschen, Reiner Mller & Marga Reis (eds), 6186.
Hamburg: Buske.
Ebert, Robert P. 1976. Infinitival Complement Constructions in Early New High German [Lin-
guistische Arbeiten 30]. Tbingen: Niemeyer.
Eilers, Helge. 1992. Notkers Wiedergabe und Gebrauch lateinischer Konstruktionen in seiner
bersetzung der Consolatio (Prosa im I. Buch) des Boethius. In Althochdeutsch, Syntax
und Semantik: Akten des Lyonner Kolloquiums zur Syntax und Semantik des Althoch-
deutschen, Yves Desportes (ed.), 115151. Lyon: Universit Lyon.
Fischer, Olga. 1994. The development of quasi-auxiliaries in English and changes in word order.
Neophilologus 78: 137164.
Fischer, Olga. 1997. On the status of grammaticalisation and the diachronic dimension in expla-
nation. Transactions of the Philological Society 95: 149187.
Anne Jger
Furrer, Dieter. 1971. Modusprobleme bei Notker. Die modalen Werte in den Nebenstzen der
Consolatio-bersetzung [Das Althochdeutsche von St. Gallen 2]. Berlin: Walter de Gruyter.
Harris, Alice C. & Campbell, Lyle. 1995. Historical Syntax in Cross-linguistic Perspective
[Cambridge Studies in Linguistics 74]. Cambridge: CUP.
Haspelmath, Martin. 1989. From purposive to infinitive: A universal path of grammaticaliza-
tion. Folia Linguistica Historica 10: 287310.
Heine, Bernd & Kuteva, Tania. 2002. World Lexicon of Grammaticalization. Cambridge: CUP.
Helbig, Gerhard & Buscha, Joachim. 2005. Deutsche Grammatik: Ein Handbuch fr den Ausln-
derunterricht. Berlin: Langenscheidt.
Hellgardt, Ernst. 1979. Notkers des Deutschen Brief an Bischof Hugo von Sitten. In Befund
und Deutung: Zum Verhltnis von Empirie und Interpretation in Sprach- und Literatur-
wissenschaft: Hans Fromm zum 26.5.1979, Klaus Grubmller, Ernst Hellgardt, Heinrich
Jelissen & Marga Reis (eds), 169192. Tbingen: Niemeyer.
Kbler, Gerhard. 1993. Wrterbuch des althochdeutschen Sprachschatzes. Paderborn: Schningh.
Kotin, Michail L. 1999. Die Basisrelationen des Deutschen und die Auxiliarisierung von Haben,
Sein und Werden. Zeitschrift fr Deutsche Philologie 118(3): 391419.
Krug, Manfred G. 2000. Emerging English Modals: A Corpus-based Study of Grammaticalization
[Topics in English Literature 32]. Berlin: Mouton de Gruyter.
ecki, Andrzej M. 2010. Grammaticalisation Paths of have in English [Studies in Medieval Lan-
guage and Literature 24]. Frankfurt: Lang.
Lexer, Matthias. 1992. Mittelhochdeutsches Taschenwrterbuch. Stuttgart: Hirzel (Repr. 3rd print.
Leipzig: Hirzel. 1885).
Mitchell, Bruce. 1985. Old English Syntax, Vol. 1: Concord, the Parts of Speech, and the Sentence.
Nf, Anton. 1979. Die Wortstellung in Notkers Consolatio. Untersuchungen zur Syntax und ber-
setzungstechnik [Das Althochdeutsche von St. Gallen 5]. Berlin: Walter de Gruyter.
[OED] = Oxford English Dictionary. 2011. Oxford: OUP. http://www.oed.com/ (December
2011).
hl, Peter. 2009. Die Entstehung des periphrastischen Perfekts mit haben und sein im
Deutschen eine lngst beantwortete Frage?. Zeitschrift fr Sprachwissenschaft 28: 265306.
Paul, Hermann. 2007. Mittelhochdeutsche Grammatik, 25th edn. Tbingen: Niemeyer.
Schtzeichel, Rudolf. 1995. Althochdeutsches Wrterbuch, 5th revised and supplemented edn.
Tbingen: Niemeyer.
Sehrt, Edward H. & Legner, Wolfram K. (eds). 1955. Notker-Wortschatz: Das gesamte Material
zusammengetragen von Edward H. Sehrt und Taylor Starck. Halle: Niemeyer.
Sonderegger, Stefan. 1987. Notker der Deutsche als Meister einer volkssprachlichen Stilistik. In
Althochdeutsch, Vol. 1: Grammatik. Glossen und Texte, Rolf Bergmann, Heinrich Tiefen-
bach & Lothar Voetz (eds), 839871. Heidelberg: Winter.
Sonderegger, Stefan. 2003. Althochdeutsche Sprache und Literatur: Eine Einfhrung in das lteste
Deutsch: Darstellung und Grammatik, 3rd revised and supplemented edn. Berlin: Walter
de Gruyter.
Visser, Frederikus Theodorus. 2002. An Historical Syntax of the English Language, 4th edn.
Leiden: Brill.
Young, Christopher & Gloning, Thomas. 2004. A History of the German Language through Texts.
London: Routledge.
Hearsay and lexical evidentials
in Old Germanic languages, with focus
on Old English
Olga Timofeeva
University of Zurich, Switzerland
This article offers a paradigmatic survey of auditory evidential constructions

in Old English: direct-perception constructions accusativus cum infinitivo
(ACI) introduced by the auditory (ge)hieran to hear ((ge)hieran+ACI) and
hearsay-evidence constructions, consisting of the verb (ge)hieran with the
infinitive of a verb of utterance ((ge)hieran+Inf), followed by a compliment
clause, a prepositional clause, or a parenthetical. Comparative data from other
Old Germanic languages suggests a common origin of both constructions. It is
further hypothesised that these two do not go back to the same Proto-Germanic
construction: (ge)hieran+ACI is more likely to have arisen from the reanalysis
ofthe verbal noun in I heard his speech into an ACI with a verb of speaking
Iheard him speak, while (ge)hieran+Inf could have developed from I heard the
story into I heard (the) say with the verb of saying.
1. Introduction
It is a well-known fact that Germanic languages, and Western European languages

more generally, do not encode evidentiality in the verbal morphology (Chafe
1986; Aikhenvald 2004:355; Whitt 2010:13; Wiemer 2010:6267). What these
languages do instead is signify information source by various lexical means: per-
ception verbs (e.g. see, hear), modal verbs (e.g. must), verbs of utterance (e.g.
speak, say), sentential adverbs (e.g. apparently, supposedly), etc. (Diewald &
Smirnova 2010b:16).1 Among these means, perception verbs and verbs with a
say-component are ones that occur most frequently cross-linguistically and often
. For a survey of these lexical means in English and German, see Whitt (2010:1439). A
classification of hearsay lexemes in other European languages (defined geographically and
including Turkish, Armenian, Georgian, and Estonian) is given in Wiemer (2010:69113); see
also other contributions to Diewald and Smirnovas 2010a volume.
Olga Timofeeva
grammaticalise into morphological evidentials (Aikhenvald 2004:271274). It is

the relation between these two groups (perception and utterance) that I investigate
here and illustrate with data from the Old Germanic languages, which seldom
come into typological discussions of evidentiality in literature. More specifically I
address Old English auditory perception verb (ge)hieran2 hear and its infinitival
complements and compare these findings to cognates of (ge)hieran in other Old
Germanic languages and their complementation patterns. I, thus, share a broader
understanding of evidentiality and evidential markers, one that includes not only
grammatical mark-up but also lexical means (see Diewald & Smirnova (eds.)
2010a in general and Wiemer (2010:60) in particular), such as verbs of perception
and verbs or verbal phrases denoting hearsay evidence. Following existing clas-
sifications, I distinguish between two basic groups of evidentials, with two sub-
groups within the second group:
1. Direct/first-hand perception
2. Indirect perception
a. Indirect/second-hand inference
b. Indirect/second-hand hearsay evidence (Chafe 1986:261264; Willett
1988:57; Aikhenvald 2004:25, 6366; Plungian 2010:3538).
I address these groups in more detail in the relevant sections of this article, illus-
trate them with Old English (OE) examples and supply comparative material from
other Old Germanic languages if it is available. The aim of this comparison (pre-
sented in Sections 3.1 and 3.2) is to establish whether auditory evidential con-
structions in OE and other Old Germanic languages developed from a common
stock, to analyse the specific construction types in which (ge)hieran and its cog-
nates are used, and to find out what specific evidential meanings are bound to
these constructions. These findings are summarised in Section 3.3. Further in 3.4,
I attempt a more detailed analysis of the verbs of utterance that are used in eviden-
tial constructions in combination with (ge)hieran in Old English and address the
co-lexicalisation of hear-say in English and German, which is to lay the ground
for my discussion of the origin of the direct auditory perception and hearsay evi-
dence constructions in Proto-Germanic in Section 3.5. The bulk of my data pre-
sented in Section 3 deals with hear+Inf constructions which would in present-day
English correspond to something like Ive heard say that he is a nice chap. These
are typically treated in literature as accusativus cum infinitivo (ACI) construc-
tions with ellipsis of the accusative (see Note13 below). Thus, a full construction
would be Ive heard people say that he is a nice chap. In 3.5 I suggest that hear+Inf
. I use the spelling (ge)hieran to refer collectively to both the prefixed verb gehieran and
the simplex hieran.
Hearsay and lexical evidentials in Old Germanic languages, with focus on Old English
and hear+ACI could have developed independently of each other at a stage when
Proto-Germanic infinitives were still construed as verbal nouns. Corpus data for
the study is introduced in Section 2.
2. The data
OE data for this study come from the Dictionary of Old English Web Corpus
(DOEC), an online database consisting of at least one copy of every surviv-
ing Old English text (c. 7001150) and containing about 3,033,000 words, and
from the syntactically annotated York-Toronto-Helsinki Parsed Corpus of Old
English Prose (YCOE). For comparative Old Germanic material, I consulted
Heimskringla Project, an online database of some 2,000 Old Norse texts (primar-
ily the Eddas, the Icelandic sagas and skaldic poetry); Project Wulfila, an online
database of surviving Gothic texts (Gothic Bible and minor fragments), with
syntactic annotation; and TITUS online database (Thesaurus Indogermanischer
Text- und Sprachmaterialien) for Old High German and Old Saxon. The chief
criterion for choosing these resources on the Old Germanic languages was their
availability. In selecting Old Germanic examples for this study, I limit my upper
time frame to the end of the OE period and only include those whose composi-
tion can be dated to before 1150 to be able to compare roughly contemporary
developments. Later material is quoted on a few occasions when this criterion
cannot be met. It is not my intention here to attempt a statistical evaluation of
auditory and hearsay evidentials in Gothic, Old High German, Old Saxon, and
Old Norse, but rather to suggest a typology of cognate constructions in these
languages and to check whether they could have developed from a common
stock. It is to be hoped that with the emergence of new electronic resources for
these languages, the results of this study could be revisited and corroborated
with a more extensive analysis of data going beyond Old English. To trace the
long-term diachronic development of English auditory and hearsay evidentials,
I also examined the Corpus of Middle English Prose and Verse (CME) and the
Oxford English Dictionary (OED).
3. OE (ge)hieran: Its complements and cognates
To obtain OE data from the YCOE, I ran CorpusSearch Programme with several
input commands, such as
node: IP*
query: ((IP-MAT dominates IP-INF*)
AND (geher*|gehyr*|gehir*|gehier* precedes IP-INF*))
Olga Timofeeva
I also searched for simplex forms of hieran (with spelling variants) preceding the
infinitival clause (IP-INF) and for the reversed order of constituents infinitival
clauses followed by (ge)hieran. I then supplemented these data with proximity
searches in the DOEC for collocations of (ge)hieran with the infinitives that had
been yielded by the YCOE, so as to obtain the data from verse texts and to make
sure that I get the constructions which might have been missed by the parser.3 This
procedure yielded a total of 166 hits. The complete list of verbs used as infinitival
complements of (ge)hieran is as follows:
secgan to say (106 tokens)

rdan to read, explain (23)
sprecan to speak (15)
nemnan to name, call (7)
reccan to tell, explain (7)
cwean to say (2)
cyan to tell (2)
maelian to speak (2)
hrutan to snore (1)
huntan to hunt (1)4
Among these the source of information (the direct object of the perception verb)
is implicit (henceforth (ge)hieran+Inf construction) in 144 tokens (or 87 per cent
of the data), as below:
(1) Hwt, we nu gehyrdon secgan hwylc hit is on helle
what we now hear-pret-pl say-inf which it is in hell
to bionne (HomS 4 222)
to be
Lo, we have now heard say what it is like to be in hell
In the remaining 22 tokens (13 per cent), we have a full accusativus cum infinitivo
(ACI) construction (henceforth (ge)hieran+ACI construction):
Of re tide, e ic rest mid e on isum westene eardode,
(2)
of that time that I first with thee in this wasteland lived
ic e gehyrde sprecan on fenne and on
I thou-acc hear-pret-1sg speak-inf on evening and on
. This double-checking proved to be useful in my previous studies: Timofeeva (2010) and

Kilpi & Timofeeva (2011).
. I examine the example with huntan in my dissertation (Timofeeva 2010) on p. 135. For
the reasons explained there, I consider it very marginal for the present analysis.
renmergen, ic nat mid hwne (LS 10.1 (Guth) 20.73)

early-morning I not-know with whom
At the time when I first lived with you in the wilderness, I heard you speak
in the evening and early morning I did not know with whom
In what follows I survey these complementation patterns in more detail. Compara-

tive material from other Germanic languages: Gothic, Old High German (OHG),
Old Saxon (OS), and Old Norse (ON) is provided (if available) for each OE pattern.
3.1 Direct perception

Auditory evidentials are part of a bigger subsystem of sensory or non-visual per-
ception. They introduce information acquired through hearing and signal direct
access to this information (Chafe 1986:267; Willett 1988:57; Plungian 2010:37).
Thus hearing and the event that is heard are construed as happening simultane-
ously. Compare examples from present-day English and German: I hear Karen
singing; Ich hre Karen singen (Whitt 2010:9).
3.1.1 (ge)hieran+ACI construction

Syntactically this evidential value can be coded by full ACI constructions in OE
(similarly to modern and present-day English (Whitt 2010:140145)). Infini-
tival complements in these constructions are typically verbs of sound emission
(Timofeeva 2010:135), as below:
a he
(3) ne cyrcward gehyrde ofer eall
when he that-acc churchwarden-acc hear-pret-3sg over all
hrutan, a ne wnde he him nanes incymes (Leof 31.34)
snore-inf then not thought he him no entrance
when he heard the churchwarden snore very loudly, he thought that it
would not be possible for him to enter [the church]
More specifically and overwhelmingly so 20 out of 22 occurrences the comple-

ments are verbs of utterance (see Table 1):
Hwt a cempan a hine gelhton. for an e
(4)
what the soldiers then him seized for that that
hi gehyrdon hine be am cyrclicum
they hear-pret-pl he-acc of the churchly
mamum sprecan
(CHom I,29 420.70)

treasures speak-inf
Lo, then the soldiers seized him, for they heard him speak of the treasures
of the church
Olga Timofeeva
Direct perception of St Lawrences words in this homily is taken as evidence

against him. He is seized by the soldiers of the prefect of Rome precisely because
they have heard him mention the treasures of the church in his prayer (CHom
I,29 420.68).5 Both here and in (3), the focus is on hearing itself rather than on
reported information (cf. Whitt 2010:140).
Table 1. Direct auditory perception: (ge)hieran+ACI constructions

secgan 10
sprecan 7
maelian 2
reccan 1
hrutan 1
huntan 1
Total 22
Previous research has shown that ACIs with verbs of direct perception are
a common feature in Old Germanic languages (Callaway 1913:185186; Scheler
1961:92; Mitchell 1985: 37413743). Accordingly, we find similar uses of hrian
and heyra with an ACI complement in the ninth-century OS (Heliand) and ON
(the Eddic Old Lay of Guthrun, whose composition is dated to before 1000, but the
surviving manuscript to 13th century):
(5) quun that sie ina selon seggian gihrdin,

[they] said that they he-acc self-acc say-inf hear-pret-3pl
that he [mahti] teuuerpen thena uuh godes
that he [could] knock-down that shrine of-god
(Heliand lx.50735074)
they said that they had heard himself say that he could knock down the
shrine of god
(6) heyrir
then hear-pres-2sg thou
hrafna gjalla,
ravens-acc cry-inf
rnu
gjalla

eagles-acc cry-inf
. To prevent the church property being seized by the prefect, St Lawrence distributed it
among the poor and suffering of Rome, for which he was imprisoned and martyred.
zli fegna,

carrion happy
varga jta
wolves-acc howl-inf
of veri num (Gurnarkvia in forna 8.38)
of husband thine
then you (will) hear / the ravens cry, / the eagles cry / happy with carrion, /
the wolves howl / about your husband
Here again direct auditory perception helps to interpret facts and to verify
rumours: the guilt of Christ is established by the law-interpreters who have heard
him promise to destroy the old temple (5), while the death of Sigurd is revealed to
Gudrun by the cries of wolves and birds of carrion at his corpse (6).
3.1.2 (ge)hieran+Inf
First-hand auditory perception can also be coded by an (ge)hieran+Inf construc-
tion. In these cases, context typically suggests habitual events or immediate past
reference, as below:
On am halgan godspelle e ge gehyrdon nu
(7) rdan us
on that holy gospel that ye hear-pret-pl now read-inf us
seg be Lazare (Hom 6 1)
says of Lazar
in the holy gospel that you have heard [me] read just now we are told
aboutLazar
Example (7) reflects a common medieval situation when people would listen to
books being read to them but could not read themselves. Thus, here we can envis-
age a preacher who has just finished reading the Gospel in Latin (which is sug-
gested by the adverb nu) and is now going to explain and comment on its contents
in English. A direct reference to the preacher is, however, omitted.
A similar use of hrian taking an infinitive without the accusative NP is found
in the OS Heliand (ninth century):
(8) Manag fagonoda
many rejoiced
uuerod aftar [them] uuha: gihrdun uuilspel mikil
people at [the] shrine hear-pret-3pl good-news great
fon gode seggean (Heliand vi.526528)
about god say-inf
many rejoiced among the people at the shrine/[as] they heard say great
andwelcome news about God
Olga Timofeeva
This example refers to prophetess Anna whose words about the birth of the Sav-
iour have been quoted in the preceding verses (cf. Luke 2.3638). Thus, the source
of information is omitted but can be retrieved from previous context.6
3.1.3 (ge)hieran+Inf+PP
This structure can be seen as an extension of (ge)hieran+Inf, in which the source
of information is indicated overtly by a prepositional phrase with fram from, of
from, or urh through followed by an animate human noun:
Ond monige men a e as ing gehyrdon secgan

(9)
and many men when they these things hear-pret-pl say-inf
from m forsprecenan were, wron brnde
from that abovementioned man were kindled
in geleafan (Bede 4 23.330.16)
in faith
And in many people who had heard this man speak about these things faith
was kindled
Similarly in the early-thirteenth-century Yngvars Saga:
Glmr hafi numit at fur snum, en rir hafi numit af

(10)
Glum had taken from father his, and Thorir had taken from
Klkku Smssyni, en Klakka hafi heyrt segja ina
Klakka Samsson, and Klakka had hear-PART2 say-inf among
fyrri frndr sna (Yngvars saga vfrla xiv)
older of-family his
Glum had got it from his father, and Thorir had got it from Klakka
Samsson, and Klakka had heard it told by the elders in his family
Because my data on this type is limited to 3 occurrences of (ge)hieran+secgan+PP

in OE, it is rather difficult to conclude whether it is equivalent to the ACI
construction: does I heard say from him imply I heard him say, and thus is it a clear
case of direct evidentiality? As in 3.1.2. the interpretation depends on the focus.
If it is placed on the information and its source, Examples (9) and (10) qualify as
hearsay evidence; if, on the other hand, the focus is on hearing, they are direct
auditory evidence (mark the importance of multiple witnesses in (9) and possible
contrast between hafi numit and hafi heyrt segja in (10)).7
. An alternative interpretation is proposed in 3.5.

. Whitt is inclined towards the former interpretation (2010:142, 156158).
3.1.4 Gothic data

My searches in the Project Wulfila database revealed that Gothic cognate of (ge)
hieran (ga)hausjan can take accusativus cum participio (ACP) as complement.
There are five constructions of this type with participle 1 (John 7.32; Luke 18.36;
Mark 12.28, 14.58; and Thessalonians II 3.11) and two constructions with parti-
ciple 2 (Luke 4.23, 9.7) in the Gothic New Testament.
(11) hausidedun an Fareisaieis o managein
hear-pret-3pl then Pharisees this-acc multitude-acc
birodjandein bi ina ata (John 7.32)
murmur-part1-acc about him this
then the Pharisees heard people murmuring such things about him
Closer examination showed, however, that all seven occurrences of the ACP trans-
lated similar participial constructions of the Greek original. There are no attesta-
tions of (ga)hausjan+ACP in the original Gothic Skeireins, a commentary on the
Gospel of St John, to ascertain whether this construction could be used in Gothic
independently of a Greek source text. Further my analysis of ACPs as comple-
ments of perception verbs in OE suggests that these constructions are also mostly
found in translations from Latin (Timofeeva 2010:139141). It seems therefore
that the Gothic data on ACPs should best be treated as ambiguous and not be
included in the typology of auditory evidential constructions in the Old Germanic
languages.
3.2. Indirect perception

Indirect evidence requires either inference, reasoning, interpretation of the per-
ceived information or verbal report, hearsay, folklore (Chafe 1986:268; Willett
1988:5758). Both types of indirect evidence first have to be acquired through
some form of perception (typically, visual or auditory) in order to be interpreted
or passed on as hearsay (cf. Whitt 2010:1011).
Inferential evidence in modern and present-day English is typically coded by
see with a finite clause as complement (Boulonnais 2010:1822; Whitt 2010:40,
5761):
Leaphorn could see Dashee was hating this. Hed pushed his uniform
(12)
cap back on his head. His face was flushed. His forehead was beaded with
perspiration. (T. Hillerman, 1998, quoted in Boulonnais 2010:19)
This use of see involves a semantic shift see realise/understand and includes
seeing some signs or symptoms and interpreting their meaning. In (12) Leaphorn
observes the cap, flushed face and perspiration of Dashee and concludes that the
Olga Timofeeva
latter is uncomfortable and hates the situation he is in. Similarly in OE, inferential
evidence is coded by visual (ge)seon see (see Timofeeva 2010:138141). Because
inference indicated by auditory (ge)hieran is not attested in my data, I do not
discuss this evidential meaning any further, but concentrate instead on hearsay.
According to Whitt (2010:136), hear+finite complement clause with or without
that-complementiser is the default encoding of hearsay evidence:
I heare she is worthe a thousande pounde and more. (Helsinki Corpus:

(13)
ceplay1a, Nicholas Udall, Roister Doister (15001570), p. L. 131, quoted
inWhitt 2010:137)
The event has not been heard directly nor can being worth a thousand pound and
more be virtually heard but acquired from a second-hand unspecified source.
Importantly for the purposes of the present study, Whitt remarks that hearsay
sometimes is explicitly marked as such, i.e. the speech-act verb say occurs right
alongside the perception verb hear (2010:138):
I heard say that your husband would now put you in your hood, and silke
(14)
gowne, I pray you is it true? (Helsinki Corpus: cefict2B, Thomas Deloney,
Jack of Newbury (1619), p. 70, quoted in Whitt)
The content of the proposition in the complement clause was reported by a non-
specified individual(s) and perceived through hearing. We are now going to see
what precursors this encoding of hearsay evidentiality had in OE and its cognate
languages.
3.2.1 (ge)hieran+Inf+t- / wh-clause

There are three ways to indicate hearsay evidence in OE, in which (ge)hieran co-
occurs with a verb of utterance. The first one is to use a finite form of (ge)hieran
with an infinitive (typically of the verb secgan, see Table 2) and a finite comple-
ment clause introduced by t8 (24 occurrences):
(15) ic gehyrde secgan t u wre gleaw ron (Gen 41.15)
I hear-pret-1sg say-inf that thou were skilful thereon
I have heard say that you are skilful in that [in interpreting dreams]
Foram we gehyra
(16) reden on am godspelle t Crist
for-that we hear-pres-pl read-inf on that gospel that Christ
. Cross-linguistically this is often the underlying etymological structure of morphological

hearsay evidentials (Anderson 1986:285).
cwde t he were weig and sofasnes and lyf (Solil 1 51.11)

said-subj that he were way and truth and life
For we hear it read in the Gospel that Christ said that he was the way
andtruth and life
The speaker (pharaoh) in (15) indicates that his information derives from an
unspecified source. Although it has been perceived through hearing, auditory
perception here is the means by which this information has been acquired rather
than direct evidence for it. Similarly in (16), the focus is on the report of what
Christ said, rather than on the fact that the Gospel was read and listened to. The
proposition in the complement clauses may or may not be true (mark also the
subjunctive of u wre and Crist cwde).
Further, there are 9 occurrences of finite wh-complement clauses after (ge)
hieran+secgan, see example (1) which I reproduce here for convenience:
(1) Hwt, we nu gehyrdon secgan hwylc hit is on helle to bionne (HomS 4 222)
Table 2. Hearsay evidence: (ge)hieran+Inf+t- / wh-clause

secgan 30
cyan 1
rdan 1
reccan 1
Total 33
These encodings of hearsay evidence are attested in other Old Germanic lan-
guages, compare examples with OHG hrran and ON heyra below:9
(17) Ik gihorta at seggen, (Hildebrandslied 13)

I hear-pret-1sg that say-inf
at sih urhettun non muotin:
that each-other challengers alone met
Hiltibrant enti Haubrant untar heriun tuem
Hildebrand and Hadubrand between armies two
I have heard it say / that warriors met each other alone / Hildebrand and
Hadubrand, between two armies
. See also example (317) in Whitt (2010:156).

Olga Timofeeva
In ON (again I quote one of the Eddic lays recorded in the 13th century) the com-
plement clause can be introduced by hv:
(18) Heyra ek segja
hear-pret-1sg I say-inf
sgum fornum,
in sagas old
hv mr of kom
how maiden came
til Mornalands (Oddrnarkvia 1.14)
to Morningland
I have heard say / in old sagas / how a maiden came / to Morningland
Because both tokens occur in the opening lines of the poems (cf. also examples19,
24, 26, 29, and the conclusion of Yngvars Saga in 10), this suggests a common
Germanic strategy to mark hearsay when a poet was about to begin his narrative
and wanted to evoke the authority of the oral tradition (cf. a similar observation
on Germanic and non-Indo-European heroic poetry in Bowra 1952:40).
3.2.2 (ge)hieran+Inf+PP
Another way to introduce indirect evidence is to use a combination of hear and
say together with about prepositional phrase (be, fram, of, ymbe). PPs in this con-
struction do not introduce the source of information (as in 3.1.3), but rather the
focus of the report transmitted via hearsay. Only secgan and sprecan are attested in
these constructions (see Table 3).
(19) Hyrde ic secgan gen bi sumum fugle
hear-pret-1sg I say-inf also about some bird
wundorlicne [***] (Part 12)
wonderful
I have also heard it say about a certain wonderful bird
In re stowe ws gewuna, t man hwilum ymb
(20)
in that place was custom that one sometimes about
fisc gehyrde sprecan & r ns nfre
fish hear-pret-3sg speak-inf and there not-was never
nan gesewen (GD 1 (C) 1.11.16)
none seen
In that place, people were accustomed to hearing reports about fish but
none of them had ever seen a fish
Exotic animals partridge (19) and fish (20) have never been observed in the
communities implied in (19) and (20), but people know about them from folklore
and other second-hand accounts.
Table 3. Hearsay evidence: (ge)hieran+Inf+about-PP

secgan 24
sprecan 6
Total 30
Hearsay markers with prepositional phrases are often found either within a
relative clause as in (21) or in a superordinate clause that introduces a relative
one as in (22), where hearsay evidence is made even more tentative by the use of
the pre-modal verb magan.
a ohte ic t t wre seo helle e ic oft on life
(21)
then thought I that that were the hell that I often in life
embe secgan gehyrde(CHom II, 23 200.31)
about say-inf hear-pret-1sg
Then I thought that this was Hell about which I have often heard say
Magon
(22) we nu gehyran secgan be suman halgan men
may-pres-pl we now hear-inf say-inf about some holy man
se ws on gastlice gesyhe gelded (HomU 55 (Thorpe) 69)
that was in spiritual vision led
Now we can hear say a story about a holy man who was instructed in a
spiritual vision
Again the emphasis is on the objects/persons Hell and a holy man that are
introduced or described by the PPs, while the knowledge about them is clearly
second-hand.
3.2.3 (ge)hieran+secgan in swa-parentheticals

Yet another way to encode hearsay is a parenthetical clause connected by means
of swa so, as:
To issum dagum urh haligne gast is fsten &
(23)
to these days through holy ghost this fast and
as gangdagas geset weron, swa we oft on bocum
the gang-days set were as we often in books
gehyrdon secgan (HomS 33 (Frst) 26)
hear-pret-pl say-inf
this fasting and the procession days have been observed until these days
through the holy spirit, as we have often heard say in books
We can only hypothesise a generic reference to some wise men of the books
whose authority is evoked here to emphasise the importance of fasting and reli-
gious processions. Intensifying adverbs such as oft often, gelome frequently,
Olga Timofeeva
and solice indeed, truly can be used to emphasise the reliability of rumours
or tradition.
(24) swa we solice secgan hyrdon (Beo 273)

as we truly say-inf hear-pret-pl
as we have truly heard say
swa we oft & gelome secgan gehyrdon(HomS 24.1 (Scragg) 33)

(25)
as we often and frequently say-inf hear-pret-pl
as we have often heard say
A total of 12 swa-parentheticals (only with secgan) are attested in the DOEC.

There are several attestations of so-parentheticals in the early-thirteenth-
century manuscripts of the Song of the Nibelungs. The examples from the oldest
manuscript B are lines 901, 3801, 7151, 7611, and 10561, all of them containing
the following formula:
(26) so wir horen sagen

as we hear-pres-pl say-inf
as we hear say
As we see, first person prevails in these constructions. Whitt observes that the use
of hearsay markers with as-parentheticals in English (and with wie-parentheticals
in German) signals a heightened degree of intersubjectivity among parenthetical
constructions There is a greater awareness of a larger speech community, and
therefore an implicit acceptance of alternate possibilities of perception, evidence,
and interpretations thereof (2010:150151, at 151, cf. 164165).
3.2.4 Gothic data

There is only one example of hausjan+ACI encoding hearsay evidence in the
surviving Gothic texts.
(27) unte gairnjands was allaize izwara jah unwunands,

for longing he-was (after) of-all you and distressed
in izei hausidedu ina siukan (Philippians 2.26)
in that hear-pret-2pl he-acc sick-inf/-adj-acc
for he longed after all of you and was distressed, because you had heard
thathe had been sick
My gloss in (27) and the analysis of the compilers of the Wulfila Project indicate
that the status of siukan is ambiguous: it can be interpreted both as an infinitive
and as a weak masculine adjective in the accusative plural. Interestingly, hausid-

edu ina siukan translates a that-finite complement clause of the Greek source.
Nevertheless, even if this example represents original Gothic usage, it is not pos-
sible to decide how it fits into the Old Germanic system of auditory evidentials,
let alone to reconstruct whether anything like hausjan+*sagjan or hausjan+qian
had a place in it.
3.3 Some conclusions

This survey has shown that in terms of lexical means OE and other Old Germanic
languages (with the exception of Gothic, for which we do not have enough data)
seem to have similar ways of marking direct auditory perception and hearsay evi-
dence. In fact the degree of overlap is high enough to suggest common origin
of both hear+ACI and hear+Inf constructions in Proto-Germanic (or at least in
West and North Germanic). More corpus research on individual languages and
more comparative work on their later stages, therefore, would be very welcome as
both will help to reveal diverging tendencies and to explain the differences of the
present-day situation (see, e.g. Whitt 2010).
3.4 Verbs of utterance and their syntactic context

It may have become clear already that different verbs of utterance favour different
types of syntactic context. In this section, I provide a brief survey of the structures
in which verbs of utterance occur or which they take as complements. I begin with
the less frequent and proceed to secgan.
3.4.1 cwean
Both examples of cwean have a to-PP complement occurring in a (ge)hieran+Inf
construction. One of them is followed by direct speech:
Me wran mine tearas for hlafas, ger ge on dg ge on
(28)
to-me were my tears for bread, both on day and on
niht; onne ic gehyrde to me cwean, Hwr is in
night; then I hear-pret-1sg to me say-inf, where is thy
God, e u to hopast? (PPs (prose) 41.3)
God, that thou hope-to
I had my tears for bread night and day; and then I heard it spoken to me,
Where is your God in whom you put hope
No other verb in my data takes a to-PP. Both occurrences of (ge)hieran+cwean

imply direct auditory perception.
Olga Timofeeva
3.4.2 cyan
On one occasion cyan takes t-clause as complement:
(29) Hwt, we t hyrdon urh halige bec
what, we that hear-pret-pl through holy books
tacnum cyan, t twegen mid him
tokens tell-inf, that two with him
gerowedon, ond he ws ridda sylf
suffered, and he was third self
on rode treo (El 852)
on rood tree
Lo, we heard it told through symbols in the holy Gospels that two men
suffered with him, and he himself was the third on the rood
The source of information is introduced by urh-PP. Mark also that here (ge)hieran-
cyan occurs again in the opening lines of the narrative, when St Helena is going
to give a brief account of the crucifixion story to wise Jewish men in Jerusalem.
Thesecond instance of cyan is found in an extended phrase where (ge)hieran takes
four infinitives as complements, while each infinitive takes an NP as direct comple-
ment in turn.
onne we gehyron
(30) Godes bec us beforan reccean
when we hear-pres-subj-1pl Gods books us before explain-inf
& rdan, & godspell secggean, & his wuldorrymmas
and read-inf, and gospel say-inf, and his heavenly-glories
mannum cyan (HomU 20 (BlHom 10) 73)
to-men tell-inf
when we would hear Gods books explained and read before us, and gospel
said, and his heavenly glories made known to people
3.4.3 maelian
The two instances of maelian occur within ACI constructions (see Table 1), indi-
cating direct auditory perception:
(31) onne ic sigedrihten,
then I victory-lord
mihtigne god, mlan gehyrde
mighty god speak-inf hear-pret-1sg
strangre stemne, and me her stondan het (GenA,B 523)
strange voice-acc and me here stand ordered
then, o Lord of victory, mighty God, I heard a strange voice speak, which
ordered me to stand here
3.4.4 nemnan
Among the seven instances of (ge)hieran-nemnan, the following pattern emerges:
nemnan occurs in a relative clause (4 instances) or a temporal clause (3 instances)
and takes a direct object:
(32) Swelce one mran morgensteorran,
just-as that greater morning-star
e we ore naman fensteorra
that we other name evening-star
nemnan hera, u genedest one
name-inf hear-pres-1pl thou force that-one
t he re sunnan si bewitige (Met 4.13)
that he that suns journey observe
Just as You force this greater morning-star [moon], which we also hear
named by another name evening-star, to observe the journey of the sun
onne u gehyrst nemnan one fder. onne
(33)
when thou hear-pres-2sg name-inf that father, then
understenst u t he hf sunu (CHom I, 20 339.128)
understandest thou that he has son
when you hear Father mentioned, then you understand that he has a Son
In 5 instances, direct objects are proper names as in (32). Only one example with
nemnan in my data is an ACI construction (Beo 2023).
3.4.5 reccan
Reccan is typically used in combination with another infinitive: secgan (3instances)
and rdan (2 instances, see example (30)). It thus produces a somewhat tautological
effect, which is, however, characteristic of many OE texts (see Koskenniemi 1968).
Reccan is used once within an ACI construction (LS 13 (Machutus) 17r.17) and
once to introduce a t-clause (Bo 35.98.25).
3.4.6 sprecan
Like maelian, sprecan has a tendency to be used in ACI constructions (see
Table1), as in example (2), which I reproduce here for convenience:
Of re tide, e ic rest mid e on isum westene eardode, ic e gehyrde sprecan
(2)
on fenne and on renmergen, ic nat mid hwne (LS 10.1 (Guth) 20.73)
This probably has to do with the fact that both verbs are semantically verbs of
speaking rather than saying/quoting and are thus better suited to participate in
direct auditory perception rather than hearsay (I return to this distinction shortly).
Cf. also example (4).
Olga Timofeeva
3.4.7 rdan
As has been observed in Section 3.1.2 (example (7)), (ge)hieran-rdan reflects a
situation specific to predominantly illiterate communities.10 The recurrence of this
combination in the sources (particularly in homilies) points to the importance of
reading aloud as part of religious life and instruction.
On am halgan godspelle e ge gehyrdon nu rdan us seg be Lazare
(7)
(Hom 6 1)
The agent who does the actual reading can generally be reconstructed as priest or
preacher. The distinction between those people in the audience who could read
themselves and those who could not is made explicit in 6 instances out of 23:
onne we bec rda
(34) oe rdan gehyra,
when we books read-pres-1pl or read-inf hear-pres-1pl,
onne sprec God to us urh a
then speaks God to us through those
gastlican rdincge (LS (Pr Moses) 60)
spiritual readings
when we are reading the [holy] books or hear them read for us, God speaks
to us through those spiritual readings
Rdan typically (20 occurrences out of 23) takes an accusative NP as comple-

ment, which means either scripture, gospel or book (normally synonymous to
the former two) or then story, account (referring to the Bible or lives of saints).
Alternatively (3 occurrences) information source can be marked by an of- or
on-PP (as in (7)).
3.4.8 secgan
Examples with secgan amount to some 64 per cent of my data and naturally pres-
ent a greater variety of patterns, most of which, however, have featured in the
above sections: ACIs in 3.1.1, t- and wh-clauses in 3.2.1, PPs in 3.2.2, and
swa-parentheticals in 3.2.3. Many other examples are combinations of the above
patterns. I would like, however, to concentrate here on patterns that differ from
those already discussed.
On four occasions (ge)hieran-secgan is used in an abbreviated negative clause
without any complement or PP:
. On the notions of mediated literacy and vocality (Vokalitt) which describe such
communities and the role of educated elites in them, see Schaefer (1992). I am grateful to
Dr.Annina Seiler Rbekeil for this reference.
Sian Romane t gesawan t him mon swelcne wrenc to

(35)
then Romans that saw that them one such trick to
dyde swelcne hie r ne gesawon ne secgan
did such they before not saw nor say-inf
ne hirdon (Or 4 1.84.15)
not hear-pret-pl
Then the Romans realized that one has played such a trick at them, as they
had never seen or heard told of before
Negation eliminates all possible evidence here: such a thing was never witnessed
visually and never heard of in a story, thus there is neither direct perception nor
hearsay evidence for it.
Yet another pattern emerges in late OE:
(36) a e cyng Willelm geherde et secgen, a wear
when the king William hear-pret-3sg that say-inf then became
he swie wra (ChronE 1066.35)
he very angry
when king William heard/learnt this, he got very angry
a herde
(37) gelric biscop et gesecgon. a
when hear-pret-3sg gelric bishop that say-inf then
amansumede he ealle a men a t yfel dde
excommunicated he all those men who that evil deed
hfden don (ChronE 1070)
had done
when bishop gelric heard/learnt this, he excommunicated all the men
who had done this wicked deed
In both instances, et refers anaphorically to some piece of news reported via hear-
say in the previous context. This use in my data is mostly limited to the late annals
of the Anglo-Saxon Chronicle (6 occurrences altogether). According to my very
preliminary research into ME infinitival phrases with similar structure, it gradu-
ally becomes more frequent; however, more corpus work and statistical checking
would have to be done on this point. If my observations are correct, they will also
have some bearing for the following hypothesis.
The fact that anaphoric patterns become slightly more frequent towards and
during ME probably signals that semantic change is underway. What these new
patterns do is state that someone got informed about a certain fact. Although the
implication still is that the information was received through an auditory medium,
hear-say no longer introduces a report but rather follows it. This change may
point to a semantic development that involves a shift from hear-say to learn/get
Olga Timofeeva
informed. This shift is also observable in more frequent (ge)hieran+Inf patterns.

The following late OE example of a synonymous pair heard-say learnt may
point to the same conclusion about the semantic change:
Mid y he gehyrde
(38) secgan and he leornode be
when he hear-pret-1sg say-inf and he learnt about
am ancerum (LS 10.1 (Guth) 2.106)
that hermit
When he found out and learnt about the hermit
Further the word order of hear and say constituents becomes increasingly more
bound, which also suggests that the two verbs may have been construed as one
semantic unit, as below:
We iherden scgen urh sumne wisne mon t

(39)
we hear-pret-pl say-inf through some wise men that
moyses becom he to are stowe e inemnd is
Moses came he to that place which named is
quinquaginta finicas (LS 5 (InventCrossNap) 3)
fifty palm-trees
We have learnt from wise men that Moses came to a place called fifty
palm-trees11 [Helim]
And eallswa se cynincg Carolus of Francene rice

(40)
and also the king Charles of Franks kingdom
geherde secgan of his micclan godnesse (LS 9 (Giles) 400)
hear-pret-3sg say-inf of his great virtue
And also king Charles of France heard it told about his great virtue
It has been observed cross-linguistically that the complement verb can be raised
into the main clause and the two predications merged together to share one set
of grammatical relations and one phonological outline. Thus the two verbs co-
lexicalise (Noonan 1985:7376; Givn 2001: ii, 5963; Song 2001:278). I sug-
gest that this process begins in OE and extends into the ME period. Univerbated
heardsay spellings appear from ME2 onwards, so that conceptual and syntactic
proximity, already observable in OE, results in lexical and semantic unity and the
emergence of a new lexical item. Because the OED dates the first attestation of
the noun hear-saying to 1340, I assume that the ultimate co-lexicalisation of the
. Some mistake: Example 15.27 and Num 33.9 talk about seventy palm-trees.
verb hearsay should have taken place between 1250 and 1340 (but again more
work is needed to verify this). Interestingly enough there is comparative evidence
to support this scenario: in Middle High German the noun hrsagen is attested
from 1362 onwards (Mittelhochdeutsches Handwrterbuch, s.v. hrsagen).12 Thus,
the development of German Hren-sagen might also have started with two verbs,
which co-lexicalised into one and were later on converted into a noun.
To sum up, this survey has revealed that direct auditory perception in OE can
be coded most unambiguously by the (ge)hieran+ACI construction. Statistically
the verbs of speaking sprecan and maelian are used more frequently in ACIs
than other verbs. The evidential meaning of the (ge)hieran+Inf construction (and
its extensions) is predominantly hearsay evidence, although in 39 per cent of the
occurrences of (ge)hieran+Inf, the agent of the infinitive can either be retrieved
from previous context or reconstructed from our background knowledge of medi-
eval routine practices (such as the reading practices discussed in 3.4.7) or overtly
introduced by a PP (3.1.3). In these cases it is possible to interpret (ge)hieran+Inf
as constructions that encode direct perception (but see Section 3.5 below). In
terms of text types, two observations can be made: hearsay evidential construc-
tions are prominent in poetry where they are used to refer to the oral tradition
(which is in line with Aikhenvalds observations (2004:310315)), and in sermons
and homilies where they are used to evoke the authority of the Bible, the Church
Fathers, etc. Secgan is the most frequent verb, it is used in all types of construc-
tions discussed above, indicating both direct perception and hearsay evidence.
Its frequency seems to have contributed to the semantic shift hear-say get
informed, learn and the ultimate univerbation of hearsay in the Middle English
period. Comparative data from other Old Germanic languages suggests that these
evidential constructions and some of their diachronic developments might not be
unique to Old English.
3.5 Origin
It has been observed that the division of infinitival complements into ACIs and
infinitives without the accusative subjects seems to coincide with the semantic
division of verbs into speaking and saying/quoting respectively. I would like to
argue that these divisions correspond to deeper structural differences than may at
first sight seem. The emergence of such structures as OE He heard say is normally
. Cf. also DWB Online, s.vv. Hrensagen, Hrsage, and Hrsagen. Swedish also has hrsgen
hearsay, rumours, which is a clear case of loan translation from German, dated to as late as
1865 (Ordbok ver svenska sprket, s.v. hra).
Olga Timofeeva
seen as an ellipsis of He heard people say.13 However, I do not see the existence He
heard people say as a necessary precursor of He heard say.
It is well known that Indo-European infinitives go back to verbal nouns, and
Germanic infinitives (those that give rise to the so-called uninflected form) go
back to verbal nouns in the accusative -onom (e.g. Beekes 1995:251). Reconstruc-
tions of the origin and evolution of ACI constructions in Indo-European also sug-
gest that infinitives were initially more nouny and only later developed tense, voice
and mood characteristics (Coleman 1985:308311, 327, 332). Moreover, the accu-
sative in ACIs is the result of reanalysis of some other case genitive, accusative,
dative, or even locative (see Woodcock 1959:17; Coleman 1985:310311, and a
summary in Timofeeva 2010:7778).
Further, Ross (1973/2004) advocates the existence of a linear squish between
verbs and nouns that allows for a quasi-continuous hierarchy between the two
categories, whose boundaries become increasingly blurred towards the centre of
the squish. Ross sees the nouniness squish as.
that > for to > Q > Acc Ing > Poss Ing > Action nominal > Derived nominal
(40)
> Noun
These types of complements can be exemplified as follows:

(41) a. that = that-clauses (that Max gave the letters to Frieda)
b. for to = for NP to V X (for Max to have given the letters to Frieda)
c. Q = embedded questions (how willingly Max gave the letters to Frieda)
d. Acc Ing = [NP + Acc] V + ing X (Max giving the letters to Frieda)
e. Poss Ing = NPs + ing X (Maxs giving letters to Frieda)
f. Action Nominal ({Maxs/the} giving of the letters to Frieda)
g. Derived Nominal ({Maxs/the} gift of the letters to Frieda)
h. Noun (spatula)(Ross 1973/2004:1)
Rosss quasi-hierarchy suggests that some squish of nouniness, some squishoid

system, rather than the clear-cut categories of verb and noun, is part of every
speakers language competence (1973/2004:389391). This explains the diver-
sity of nouny and verby forms synchronically. I believe, however, that nouniness
squish can also explain why Indo-European verbal nouns could be reanalysed as
infinitives in the first place, and, thus, account for the diachronic developments.
If there is indeed a nounverb continuum, we might expect the fuzzy verbals and
nominals in its middle portion to evolve towards more verby or more nouny cat-
egories. Suppose we see Rosss (g) through (d) chronologically:
. Cf. OED, s.v. hear 3b: Hence, by ellipsis of such objects as people, persons, some one,
before the infinitives say, speak, talk, tell, the phrases to hear say, hear tell, etc. [emerge].
(42) Derived Nominal > Action Nominal > Poss Ing > Acc Ing
If we apply this scenario to the category of speaking nounverb continuum, we

could reconstruct the development of hear+ACI as:
(43) I heard his speech > I heard his speaking > I heard his speak > I heard him
speak
in which nouns gradually acquire verby features and develop into infinitives,
while the possessive is reanalysed as accusative. This construction from the outset
focuses on sounds (including sounds of speech) and source/agent of the sounds,
hence direct perception evidentials are more likely to develop from this pattern.
The stages in the development of constructions with verbs of saying/quoting
can be reconstructed as follows:
(44) I heard {his/the} story14 > I heard {his/the} saying >
a. > I heard his say > I heard him say
b. > I heard (the) say
For this type, sounds (media) are much less relevant than stories (results), hence
hear+Inf constructions coding hearsay evidence are probably more likely to
develop here (44b). However, this does not exclude the possibility of a parallel
development of ACIs with say (44a) by analogy with ACIs with speak or inde-
pendently of them.
One further piece of evidence to support the development I heard the story > I
heard say is that (ge)hieran+Inf but not (ge)hieran+ACI allows for a PP extension.
The emergence of constructions with PPs indicating the source of information can
be seen as:
(45) I heard this story from my dad > > I heard say from my dad
Thus the answer to my own query in 3.1.3 is that I heard say from him and I heard
him say are not equivalent, neither structurally nor semantically. The following
reconstruction can be proposed for (ge)hieran+Inf+PP indicating the focus of the
report:
(46) I heard a story about a bird > > I heard say about a bird
All in all, I suggest that the ellipsis of the accusative in the ACI construction can-
not account for the emergence of (ge)hieran+Inf, which is different from the ACI
both in structure and in evidential semantics. Moreover, chronologically, and my
corpus evidence with the preponderance of (ge)hieran+Inf over (ge)hieran+ACI
occurrences seems to support this, (ge)hieran+Inf was probably the first to emerge,
. Etymologically saga would probably be more appropriate here.

Olga Timofeeva
as cross-linguistically hearsay evidence is marked more often than direct auditory

perception (Cinque 1999:85; Aikhenvald 2004:23, 3134, 7578). As the infini-
tives became more verby, however, hear+Inf is likely to have been reanalysed as
ACI with ellipsis, or even as something less grammatical than the ACI, and eventu-
ally became obsolescent when new lexical means, e.g. hearsay adverbs reportedly,
allegedly, etc. came into being. I believe that in the case of hear+say construction,
univerbation of hearsay in ME2 and the emergence of the noun hearsay in early
ME3 can be taken as the terminus post quem this structure became less analys-
able. These stages are, however, open to debate and further research.
Acknowledgements
The early stages of this study were supported by the Academy of Finland Centre of
Excellence funding for the Research Unit for Variation, Contacts and Change in
English at the Department of Modern Languages, University of Helsinki, to which
I am most grateful. My special thanks are due to Richard J. Whitt, who sent me
his excellent book, Evidentiality and Perception Verbs in English and German, and
to Dominique Boulonnais, who read and commented on the draft of the article
and kindly helped me with some references. I would also like to thank the audi-
ence, participants, and organisers of the Verbal Categories workshop at the 16th
International Conference of English Historical Linguistics in Pcs in 2010, and the
anonymous reviewers of this volume for their helpful feedback.
References
Electronic sources
CME = The Corpus of Middle English Prose and Verse. 2006. University of Michigan. http://
quod.lib.umich.edu/c/cme/
DOEC = Dictionary of Old English Web Corpus. 2009. Ed. by Antonette diPaolo Healey, with
John Price Wilkin & Xin Xiang. University of Toronto. http://tapor.library.utoronto.ca/
doecorpus/
DWB = Das Deutsche Wrterbuch von Jacob und Wilhelm Grimm auf CD-ROM und im Internet.
19982011. University of Trier. http://woerterbuchnetz.de/DWB/
Heimskringla Project. Ongoing. Ed. by Jon Julius Sandal & Carsten Lyngdrup Madsen. www.
heimskringla.no
OED = The Oxford English Dictionary Online. 2010. OUP. http://www.oed.com
Mittelhochdeutsches Handwrterbuch. 19982011. Ed. by Matthias Lexer. University of Trier.
http://woerterbuchnetz.de/Lexer/
Project Wulfila. 2004. University of Antwerp. http://www.wulfila.be/
TITUS = Thesaurus Indogermanischer Text- und Sprachmaterialien. 2008. Ed. by Jost Gippert,
Javier Martnez & Agnes Korn. http://titus.fkidg1.uni-frankfurt.de
YCOE = York-Toronto-Helsinki Parsed Corpus of Old English Prose. 2003. Ed. by Ann Taylor,
Anthony Warner, Susan Pintzuk & Frank Beths. University of York. http://www-users.
york.ac.uk/~lang22/YcoeHome1.htm
Secondary sources
Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: OUP.
Anderson, Lloyd B. 1986. Evidentials, paths of change, and mental maps: Typologically regular
asymmetries. In Chafe & Nichols (eds), 273312.
Beekes, Robert S.P. 1995. Comparative Indo-European Linguistics: An Introduction. Amsterdam:
John Benjamins.
Boulonnais, Dominique. 2010. Verbes de perception et modes de complmentation verbale. In
Perception et structures linguistiques: Huit tudes sur langlais [Rivages linguistiques], Jean-
Charles Khalifa & Philip Miller (eds), 1589. Rennes: Presses Universitaires de Rennes.
Bowra, Cecil Maurice. 1952. Heroic Poetry. London: Macmillan.
Callaway, Morgan Jr. 1913. The Infinitive in Anglo-Saxon. Washington DC: Carnegie Institution.
Chafe, Wallace. 1986. Evidentiality in English conversation and academic writing. In Chafe &
Nichols (eds), 261312.
Chafe, Wallace & Nichols, Johanna (eds). 1986. Evidentiality: The Linguistic Coding of Epistemol-
ogy [Advances in Discourse Processes 20]. Norwood NJ: Ablex.
Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Cross-linguistic Perspective [Oxford
Studies in Comparative Syntax]. Oxford: OUP.
Coleman, Robert. 1985. The Indo-European origins and Latin development of the accusative
with infinitive construction. In Syntaxe et Latin, Actes du IIme Congrs International de
Linguistique Latine, Aix-en-Provence, 2831 mars 1983, Christian Touratier (ed.), 307342.
Aix-en-Provence: Universit de Provence.
Diewald, Gabriele & Smirnova, Elena (eds). 2010a. Linguistic Realization of Evidentiality in
European Languages [Empirical Approaches to Language Typology 49]. Berlin: De Gruyter
Mouton.
Diewald, Gabriele & Smirnova, Elena. 2010b. Introduction. Evidentiality in European lan-
guages: The lexical-grammatical distinction. In Diewald & Smirnova (eds), 114.
Givn, Talmy. 2001. Syntax: An Introduction, 2 Vols. Amsterdam: John Benjamins.
Kilpi, Matti & Timofeeva, Olga. 2011. Semantic polyfunctionality and grammaticalization of
the OE subordinator be m e: A corpus-based study. In Typology of Adverbial Connec-
tives [Studies in Variation, Contacts and Change in English 8], Anneli Meurman-Solin
& Ursula Lenker (eds). VARIENG ePublications. http://www.helsinki.fi/varieng/journal/
volumes/08/kilpio_timofeeva/
Koskenniemi, Inna. 1968. Repetitive Words Pairs in Old and Early Middle English Prose. Turku:
Turun yliopiston julkaisuja, B, 107.
Mitchell, Bruce. 1985. Old English Syntax, 2 vols. Oxford: Clarendon Press.
Mittelhochdeutsches Handwrterbuch. 1872. 1 Band. Ed. by Matthias Lexer. Leipzig: Verlag von
S. Hirzel.
Noonan, Michael. 1985. Complementation. In Language Typology and Syntactic Description,
Vol. 2: Complex Constructions, Timothy Shopen (ed.), 42140. Cambridge: CUP.
Olga Timofeeva
Ordbok ver svenska sprket utgiven av Svenska Akademien. 1933. 12 Bandet. Lund: A.-B. Ph.
Lindstedts Univ. Bokhandel.
Plungian, Vladimir. 2010. Types of verbal evidentiality marking: An overview. In Diewald &
Smirnova (eds), 1558.
Ross, John Robert. 1973/2004. Nouniness. In Three Dimensions of Linguistic Research, Osamu
Fujimura (ed.), 137257. Tokyo: TEC. Reprinted in Fuzzy Grammar: A Reader, Bas Aarts,
David Denison, Evelien Keizer & Gergana Popova (eds) 351422. Oxford: OUP.
Schaefer, Ursula. 1992. Vokalitt: Altenglische Dichtung zwischen Mndlichkeit und Schriftlich-
keit [ScriptOralia 39]. Tbingen: Gunter Narr.
Scheler, Manfred. 1961. Altenglische Lehnsyntax: Die syntaktischen Latinismen im Alteng-
lischen. Ph.D. dissertation. Berlin: Free University.
Song, Jae Jung. 2001. Linguistic Typology: Morphology and Syntax. Harlow: Longman.
Timofeeva, Olga. 2010. Non-finite Constructions in Old English, with Special Reference to Syn-
tactic Borrowing from Latin. Ph.D. dissertation, Mmoires de la Socit Nophilologique
de Helsinki LXXX, Helsinki: Socit Nophilologique.
Whitt, Richard J. 2010. Evidentiality and Perception Verbs in English and German [German
Linguistic and Cultural Studies 26]. Oxford: Peter Lang.
Wiemer, Bjrn. 2010. Hearsay in European languages: Toward an integrative account of gram-
matical and lexical marking. In Diewald & Smirnova (eds), 59129.
Willett, Thomas. 1988. A cross-linguistic survey of the grammaticalization of evidentiality.
Studies in Language 12(1): 5197.
Woodcock, Eric Charles. 1959. A New Latin Syntax. London: Methuen & Co.
Markers of Futurity in Old High German
andOld English
A Comparative Corpus-Based Study

Leibniz Universitt Hannover, Germany / Universitt Potsdam, Germany
This paper is a comparative corpus-based study of constructions that had the

potential of marking future events in Old High German (OHG) and Old English
(OE), i.e. modal constructions and those with be/become-verbs. Given the fact
that both languages stem from a common source and probably had similar source
lexemes for future grams, they nevertheless took diverging paths to develop
a future tense, with werden in German and will/shall in English. The paper
aims at comparing the earliest attestable stages of the two languages, i.e. Old
High German and Old English, to find out whether there are language internal
differences with regard to the patterns of use of the possible source items. The
database for our studies consists of OHG and OE text material dating from
790 to1155, which we consider to be maximally comparable with respect to
chronology, text type and content.
1. Introduction
Today, German and English, two West-Germanic languages, make use of different
linguistic devices to refer to future events. In Present Day German (PDG), besides
several less grammaticalised means, the werden & infinitive construction has been
grammaticalised as a future marker,1 as in example (1):
Der Bund wird im laufenden Jahr rund 80 Milliarden Euro neue Schulden
(1)
machen so viel wie nie zuvor. (18.05.2010 Sddeutsche.de)
The federal state will take on new debts of about 80 billion in the running
year so many as never before.
. The PDG werden & infinitive construction, beside its function as a future marker, has
other uses in the domain of modality and evidentiality, which, however, will not be inves-
tigated here. Disregarding the question whether PDG should be attributed a full-fledged
grammatical future category at all, it is taken for granted here that the werden & infinitive
construction does behave like a future marker in some of its PDG uses, and, moreover, that it
is the only serious candidate for this function in PDG.
In Present Day English (PDE), on the other hand, this function is performed
among other linguistic devices, mainly by the modals shall and will, as in (2)
and(3):
However, we have grave concerns that this will have serious consequences
(2)
forthe securityfinancial assets. (27 May 2010 The Times)
There we shall see one another as we really are, when all imperfection has
(3)
been wiped away. (Roderick Strange, 13 November 2009 Times Online)
Another major difference between PDG and PDE is that while in German future
marking is optional, in English it is to a large extent obligatory.
The question that derives from this observation is: Why did two closely
related languages, which shared the same source items in Proto-West Germanic,
come to choose different source items for the grammaticalisation of future mark-
ers and why did they follow different grammaticalisation paths? As far as the
diachronic lexical starting points are concerned, both languages at least at first
sight seem to have had enough in common to develop similar future markers
from cognate lexical sources. As is well known, in the earliest attested stages of
the history of German and English both types of source lexemes, i.e. werdan
on the one hand, and sculan/wellan2 on the other, were available as potential
sources for future grams. The historical development of these forms in both lan-
guages, however, was divergent. Assuming that beyond reasons of language
contact, which are not the focus of this contribution there are language internal
reasons for this divergent development, in particular different patterns of usage
and different frequencies of the source items (cf. Bybee 2010), we conducted a
corpus-based study comparing the earliest attestable stages of the two languages,
i.e. OldHigh German and Old English.3
The database we created for this study consists of OHG and OE text mate-
rial dating from 790 to 1155, which is intended to be maximally comparable
with respect to chronology, text type, content etc. The size of the corpus is about
80000 words for both German and English. These texts were analysed according
to the source lexemes that were available for future grams: wellan, sculan, werdan,
and OE beon. This paper presents the results of this study and pinpoints several
. For reasons of simplicity we refer to the various formal occurrences of these source
lexemes in OHG, MHG, OE and ME by citing the common etyma werdan, wellan and sculan.
. Preliminary considerations on this endeavour were discussed in Diewald, Habermann,
Lutz & Wischer 2002.
Markers of Futurity in Old High German andOld English
language internal factors for the diverging development of future markers in the
history of German and English.
The paper is structured along the following lines: The next section describes
our theoretical framework and gives a survey of earlier studies on the topic.
Section 3 briefly comments on the empirical procedure. In Section 4 our data will
be presented. Section 5 discusses our findings, and finally, in Section 6 we will
summarise the results and draw some conclusions with respect to the question
formulated in the beginning.
2. Theoretical background
The rise of the German werden-construction and its use as a future marker
has been the topic of a large number of studies since the nineteenth century.
On the basis of an empirical analysis, Westvik (2000:246ff.) suggests that its
use as a future marker emerged in the first decades of the 14th century in the
East Middle G erman and Upper High German dialect areas. In the course of
the 14th and 15th centuries the construction expanded across the Upper High
German and East Middle German dialects (cf. also Schmid 2000). This process
obviously interrupted the grammaticalisation of the modals wollen and sollen,
which, at that time were well on their way to becoming future auxiliaries. In the
17thcentury, finally, the replacement of wollen and sollen by werden was com-
pleted (Bogner 1989:82).4
It is only very recently that some studies on the subject have been published
which indicate a remarkable shift of interest insofar as they do no longer try to
treat the rise of a grammatical marker in isolation, but to consider its language
internal interaction with other grammatical markers on the one hand and com-
parative aspects with related languages on the other. Among them is the study
by Kotin (2003), who takes up the question of the rise of werden in connec-
tion with the development of the other German auxiliaries, and a very inspiring
paper by Harm (2001), who as far as we know for the first time raises the
issue of a comparative perspective with the aim of shedding light on the rise
. There are several highly divergent suggestions as to the exact date and origin of the
werden & infinitive-construction; see Kleiner (1925), Saltveit (1962), Schieb (1981), Walther
(1980), Leiss (1985), Schmid (2000) to quote only some of them. Some disagreement seems to
be due to heavy differences in the quantity and quality of the diachronic language data used to
propose a hypothesis. As Westvik (2000) provides an excellent survey on this, it is not neces-
sary to report the research history anew. In addition to Westviks (2000) study, a further survey
on that topic can be found in Harm (2001:290ff.).
of German werden in opposition to the modals in other Germanic languages.

Harm (2001:294) also suggests taking into account the possibility of polygenesis
of the werden-construction in several places in the linguistic area of German,
which would render the notorious where-question redundant (further recent
publications in the environment of our topic are Krmer 2005; Smirnova 2006;
Hacke 2009; Rogler 2010).
In English, on the other hand, the cognate of German werden Old and
MiddleEnglish weoran was given up in the Late Middle English period. Instead,
the will/shall & infinitive construction was grammaticalised as a future marker.
Thus, from a seemingly comparable historical starting point, the languages made
opposing choices. The history of will and shall as prime exponents of futurity in
English is discussed in more or less detail in most handbooks on English his-
torical syntax and in many treatments on the history of English modals (cf. e.g.
Jespersen 1909; Mustanoja 1960; Kisbye 1971; Berndt 1982; Arnovick 1990; van
Kemenade 1993). OE weoran is mainly dealt with in connection with passive
markers, however, it is often claimed that passive constructions with weoran in
the present tense have a future connotation (Visser 196373: 1918; Mitchell 1985:
755; K ilpi 1989:61f.). Although the use of OE beon for future states of being or
statements of eternal truth has attracted increased scholarly attention in recent
years (cf. Kilpi 1992, 1993, 1997; Lutz 2009; Wischer 2010), the development of
will and shall into future grams in English is more or less taken for granted without
considering other potential alternative sources in OE.
In our definition of a grammaticalised future marker we follow Bybee, Perkins
and Pagliuca (1994:244), who characterise a simple future as: [a] prediction on
the part of the speaker that the situation in the proposition, which refers to an
event taking place after the moment of speech, will hold. This is to say that a
simple future is a grammatical marker by which the speaker asserts that the event
expressed in the proposition will occur at a time yet to come. It indicates temporal
distance from the moment of speech, whereby the direction on the time line is
opposed to that of past tense markers. Or, to put it briefly, a simple future encodes
a prediction.
Concerning the regularity of grammaticalisation processes, the following
facts have to be taken into account:
First, as we know from studies on grammaticalisation paths (especially Bybee,
Perkins & Pagliuca 1994:253), the most frequent sources of future grams are
movement verb constructions, followed by constructions with be/become-verbs,
which correspond to the German werden-future and the OE use of beon. By con-
trast, the development of future auxiliaries from modals (modal futures), as in
the case of English will and shall, is much less widely attested crosslinguistically.
Nevertheless, English has followed the modal path.
Second, for the development of modal futures, the successive steps of semantic
change leading from lexical sources with modal meaning to the notion of simple
future are summarised as in Figure 1 (cf. Bybee, Pagliuca & Perkins 1991:29 and
Bybee, Perkins & Pagliuca 1994:254266):
Obligation (deontic modality)

Desire (volitional modality) intention future
Root possibility (dispositional modality)
Figure 1. Development of modal futures
Third, the second grammaticalisation path for future grams that needs to be
taken into account here is the development from constructions with be, become,
mentioned above. Bybee, Perkins and Pagliuca (1994:263) treat them together
with have/possession constructions as a subtype of obligation futures, although
they suggest that there exist two alternative pathways for them, one path through
obligation and a direct one with no intermediate obligation state, cf. Figure 2:
be, become, have/possession obligation
intention future
be, become, have/possession predestination
Figure 2. Development of be/become-futures
3. Empirical analysis (methodology)
The principles concerning the design of the language corpora used and the basic
considerations that have led to the building of our data-base are the comparability
and the quantity of texts. Wherever possible we use maximally comparable texts in
German and in English with respect to chronology, text type, content etc. The size
of the English as well as the German corpus is about 80 000 words each. The length
of the individual text sections varies according to availability.
The German corpus comprises the following text material dating from 790 to
1155 (see reference section for exact bibliographic data):
IS = Isidor (ca. 790): about 5 100 word forms altogether, free translation from
Latin;
TA = Tatian (ca. 830): about 13 000 word forms, starting from the beginning
of the text, gloss from Latin;
OT = Otfrid von Weienburg: Das Evangelienbuch (863/71): about 13 200

word forms, starting from the beginning of the text;
NO = Notker von St. Gallen, translation of Boethius: De consolatione phi-
losophiae (1025): about 15 000 OHG word forms from books 3 to 4 to match
the section in the Helsinki corpus; the segment chosen comprises ca. 36 000
word forms altogether, roughly 21 000 of them Latin;
KA= Kaiserchronik (1135/55): ca. 9 000 word forms, starting from the begin-
ning of the text (pages 79108 of the edition by Schrder);
SP = Speculum ecclesiae deutsch (12th century): about 13 500 word forms
starting from the beginning of the text, translation (free compilation) from
Latin;
AL = Alexanderlied (Vorauer Alexander) (ca. 1140/50): about 9 000 word
forms, starting from the beginning of the Vorauer manuscript, translation
from French (Roman d`Alexandre).
The German diachronic data can be accessed via www.kali.uni-hannover.de.

The English corpus contains those sections from the following texts that are
included in the Diachronic Part of the Helsinki Corpus of English Texts (for exact
bibliographic data see reference section). It comprises text segments dating from
8801120:
AB = Alfreds Boethius (ca. 880): about 11 000 word forms, West Saxon dia-
lect, translation from Latin;
AC = Alfreds Cura Pastoralis (ca. 885): about 18 000 word forms, West Saxon
dialect, translation from Latin;
AO = Alfreds Orosius (ca. 885): about 9 000 word forms, West Saxon dialect,
free translation from Latin;
WG = West Saxon Gospels (ca. 990): about 10 000 word forms, West Saxon
dialect, translation from Latin;
LG = Lindisfarne Gospels (ca. 960): about 9 000 word forms, Northumbrian
dialect, gloss from Latin;
C1 = Chronicle MS E (ca. 9701050): about 9 000 word forms, West Saxon
dialect;
C2 = Chronicle MS E (ca. 10701120): about 9 000 word forms, West Saxon
dialect;
GG = Gregory the Great (manuscript dating from ca. 1 100; original from ca.
885): about 5 000 word forms, West Saxon dialect, translation from Latin.
These texts were analyzed completely. Additional language data as well as exam-
ples taken from earlier studies were used when it seemed illuminating for our
argumentation (e.g. data from the 16th century).
4. Th
e data: Source lexemes in OHG and OE:
Distribution andfrequency
The source lexemes playing a role in the development of future marking in both
languages are wellan, sculan, werdan and OE beon. These lexemes show remarkable
differences with respect to frequency from the earliest stages. Tables 1a and 1b
compare werdan in OHG and OE:
Table 1a. Frequency of werdan in the German corpus

IS TA OT NO KA SP AL Total
werdan S 98 105 67 150 60 132 70 682

frequency/1,000 words 19,2 8,1 5,1 10,0 6,7 9,8 7,8 8,8
Table 1b. Frequency of weoran in the English corpus

AB AC AO WG LG C1 C2 GG Total
weoran S 33 54 44 15 50 20 49 9 274
frequency/1,000 words 3,0 3,0 4,9 1,5 5,6 2,2 5,4 1,8 3,4
The absolute numbers as well as the frequencies per 1,000 words show that
German werdan is much more frequent than its English cognate.5 However, in
contrast to OHG, weoran has a rival form in OE, which partly overlaps with
it functionally, namely beon. As is known, in OE there are two competing verbs
with the meaning to be, to exist: wesan and beon, which finally led to a suppletive
paradigm for the verb to be in Modern English. In Old English, wesan usually
refers to the real present situation, while beon is used to denote general truths or
future events.
Thus, while in OHG there is an opposition between the two lexemes werdan
and wesan, in OE there are three lexemes, weoran, wesan and beon, to share
the same functional space. In Table 2, a comparison between the OHG Tatian
. The following detail from one of the OHG texts may support this observation: the OHG
Tatian, which altogether contains 765 verb types with 11 082 tokens, the verb werdan (and
furwerdan) appears 362 times, and thus makes up 3.3% of the total token frequency of verb
forms (Sommer 1994:45, 84).
and the OE Lindisfarne Gospels rendering the same Latin text shows that where
OHG uses werdan or a form of wesan/sin (to be) to translate Latin futures, OE
prefers beon:
Table 2. The rendering of Latin futures in the OHG Tatian

and the OE Lindisfarne Gospels6
Latin Tatian Lindisfarne Gospels
II.14 Et erit tibi gaudium Inti her ist thir gifehu & bi gefea e
et multi in gaudebunt. inti manage in mendent. & monigo in bion glde.
II.15 Erit enim magnus Her ist uurlihho mihhil Bi foron micel befora
coram fora
et spiritu sancto inti heilages geistes uuirdit & gaaste halge gefylled bi
replebitur gifullit
II.20 Et ecce eris tacens Inti nu uuirdist suigenti & heono u bist suigend
quo haec fiant, in themo thisu uuerdent, of m as gewores
quae implebuntur in thiu thar gifultu a e gefylled bion
tempore uuerdent in ziti. ontid
III.35 et quod nascetur thaz thar giboran & tte acenned bi halig
sanctum uuirdit heilag,
vocabitur filius dei. thaz uuirdit ginemnit gotes bi geceid sunu godes.
barn.
III.45 quoniam uuanta thiu uuerdent foron erh-geendad bion
perficientur ea quae gifremitu thiu thar a ae
Table 3a shows that considering the mere frequency of occurrences in OE

beon is much more frequent than weoran, and this despite the fact that in Table
1b all occurrences of weoran (including the past forms) were counted, while beon
(in Table 3a) occurs only in the present tense and in the infinitive. If the past forms
of weoran are excluded from the frequency analysis as in Table 3b, an even higher
discrepancy can be observed.
However, it must be noted that in both cases, for beon and weoran, the
dominant use is not that of marking futurity. Of the 191 beon-occurrences in
Alfreds Boethius, only 6 are used with a clear future reference,7 cf. example (4),
. Luke I, 14ff. (Lindisfarne Gospels I: 14, 15, 20, 35, 45 [Skeat (ed.) 1874, pp. 1723]; Tatian
2,6; 2,9; 3,7; 4,4 [Sievers (ed.) 1892/1966:1417].
. Kilpi (1989) also notices that b-auxiliaries in OE passives are rarely used for the future
in the Cura Pastoralis and Bede. For the Gospel of St Matthew in the Corpus Manuscript and
Table 3a. Frequency of beon in the English corpus

beon S 191 248 9 35 57 12 3 31 586

frequency/1,000 words 17,4 13,8 1,0 3,5 6,3 1,3 0,3 6,2 7,3
Table 3b. Frequency of beon and weoran excluding the past forms in the English corpus
beon S 191 248 9 35 57 12 3 31 586

weoran S 26 41 1 1 2 0 2 1 74
the others rather refer to statements of general truth, as in example (5). Simi-
larly, of the 26 weoran-occurrences in the same text, only 1 can be consid-
ered to have a clear future meaning, example (6), the others rather express a
current or g eneral change of state, as in example (7). It is interesting to note
that although the original meaning of beon was very similar to that of weoran,
namely become, in Old English it had already almost completely lost its inher-
ent dynamic sense.
(4) & for m ege hi beo simle swie earme. (AB, 117.28)
and because of that fear they will always be very miserable.
Fory mon cwi be sumum goode t hit ne sie ful good, form him bi
(5)
hwshwugu wana; (AB, 34.82.28)
Concerning any form of good we say that it is not perfect good, inasmuch
as it lacks something;
Gif u onne nne stan toclifst, ne wyr he nfre gegadrod swa he r ws;
(6)
(AB, 34.92.28)
If therefore thou cleavest a stone it will never be united as it before was,
swa swa of re s cym t wter innon a eoran, & r [{afersca{];
(7)
cym onne up t m welme, wyr onne to broce, onne to ea, onne
andlang ea, o hit wyr eft to s. (AB, 34.86.22)
Even so from the sea the water makes its way into the earth, and there
grows fresh; then it comes up at the spring, becomes a brook, then a river,
then follows the course of the river until it comes again to the sea.
Rushworth 1, however, he notes that most of the instances of beon in passive constructions
refer to the future. He concludes that the text type must have influenced the use of beon/wesan.
For a similar conclusion cf. Bolze in this volume.
Tables 4 and 5 are concerned with the modal source lexemes in OHG and OE.
Tables 4a and 4b give the frequency counts for wellan, and Tables 5a and 5b those
for sculan respectively.
Table 4a. Frequency of wellan in the German corpus

wellan S 2 15 30 62 36 30 54 229
frequency/1,000 words 0,4 1,2 2,3 4,1 4,0 2,2 6,0 2,9
Table 4b. Frequency of willan in the English corpus

AB CP AO WG LG C1 C2 GG Total
willan S 84 120 41 37 15 42 34 16 389

frequency/1,000 words 7,6 6,7 4,6 3,7 1,7 4,7 3,8 3,2 4,9
Table 5a. Frequency of sculan in the German corpus

sculan S 18 3 52 46 36 122 42 318

frequency/1,000 words 3,5 0,2 3,9 3,1 4,0 9,0 4,7 4,1
Table 5b. Frequency of sculan in the English corpus

AB CP AO WG LG C1 C2 GG Total
sculan S 23 83 19 5 1 43 12 19 205
frequency/1,000 words 2,1 4,6 2,1 0,5 0,1 4,8 1,3 3,8 2,6
These figures show that both modals were present with relative frequency in
the corpora. While wellan is slightly more frequent in the English corpus com-
pared to the German one, sculan predominates slightly in German, compared to
the English corpus. This, however, may be due to the particularities of one text in
OHG, the Speculum Ecclesiae (SP), with 122 instances of sculan. This text is a col-
lection of sermons, which is highly instructive and therefore contains extremely
many deontic uses of sculan.
According to their frequency of occurrence the potential source lexemes for
future grams in both languages show the following distribution, cf. Table 6 and
Figure 3:
Table 6. Frequency of occurrence of potential source lexemes in OHG and OE

Werdan Wellan Sculan Beon Total
OHG 682 229 318 --- 1231

OE 274 389 205 586 1454
1600
1400
1200
1000
OHG
800 OE
600
400
200
0
Werdan Wellan Sculan Beon Total
Figure 3. Total numbers of each potential source lexeme in OHG and OE
Already a look at the mere numerical relations points at different conditions

for the grammaticalisation of each individual source lexeme in the two languages.
In the following section we will discuss our findings, including semantic and con-
structional factors for the development of future grams in English and German.
5. Discussion
Beyond frequency, there are further factors which are relevant for the divergent
development of future markers in German and English, one is the degree of auxi
liarisation, others are the inherent semantics and the constructional distribution
of each item in contrast to the respective competitors. Since we cannot discuss all
this at full length, we will concentrate on some of the most salient features.
5.1 The high degree of auxiliarisation of the OE modals

Earlier studies8 have proven that the German and English modals differ con
siderably as to the degree and the time of their auxiliarisation. While the English
modals had already been highly auxiliarised in OE (cf. also Wischer 2006), the
German modals never reached this degree (for details see Diewald 1999).
. For English cf. Wischer (2006); for German cf. Diewald (1999).
In our OE corpus, 84% of all willan tokens and 96% of sculan function unam-
biguously as auxiliary cf. examples (8) and (9).
Hu ne meaht u gesion t lc wyrt & lc wudu wile weaxan on m lande
(8)
selest e him betst gerist (AB 91.13)
Canst thou not see that each plant and each tree will grow best in land that
suits it best
(9) Ac t is swie dyslic & swie micel syn t mon s wenan scyle be Gode,
oe eft wenan t nig ing r him wre oe betre onne he oe him
gelic. (AB 84.18)
But it is very great folly and sin to think thus of God, or to believe that
anything existed before Him, or was better than or like unto Him.
All of these examples are complemented by an infinitive, and in all of the cases it
is the bare infinitive. Although the syntax of Old English does not allow a definite
categorisation of willan and sculan as auxiliaries since the word order is still rather
flexible and the NICE properties9 are not applicable yet, the semantics, however, is
often a clear indicator of their auxiliary status. As in examples (8) and (9) above, an
interpretation in terms of their original lexical verb meaning does not make sense.
Another criterion that sets willan and sculan apart from other lexical verbs is
the absence of non-finite forms. There is no single occurrence of a non-finite form
of these two verbs in our text corpus, neither in the infinitive nor in the present or
past participle. Furthermore, the negation of willan differs from that of most lexi-
cal verbs. The proclitic negative particle ne often merges with the verb stem giving
forms like nylle or nolde. This is not possible with sculan because of its initial [].
On the other hand, the initial [w] cannot be the only reason for the fusion of ne,
since it never occurs with weoran or wilnian.
Thus, it should have become obvious that in their syntactic behaviour the
majority of OE willan and sculan have diverged from the small residue of lexical
willan and sculan and have adopted auxiliary status.
5.2 Semantic and constructional aspects of the source items

5.2.1 Modal futures
As mentioned before, futures arising from modal sources have to pass a stage with
intentional semantics, thus presupposing a conscious, intentional entity that exerts
some influence on the event described by the proposition. In modals like will this
. Cf. Huddleston (1976:333): N: they can be Negated by a following not/nt; I: in Interroga-

tive clauses they undergo subject-verb inversion; C: they occur in post-verbal ellipses (Code)
instead of do; E: and they can carry Emphatic stress instead of using do; i.e. they function as
operators.
modal source is internal, which means it is co-referential with the subject of the
sentence: the subject of the sentences is the source of the volition and at the same
time the source of the intended action. In modals like shall, on the other hand,
the modal source is external, i.e. different from the subject of the sentence (see
Diewald 1999:93111 for an extensive treatment).
Now, it follows from the definition of a simple future given in the beginning
that to encode a purely temporal prediction, it must be devoid of any intentional
meaning. For the development of future markers from modals this requires finally
the abstraction of the semantic feature [+intentional].
The German modals wollen and sollen never reached the stage of encoding
purely temporal notions. They never completely lost their intentional component,
although they have reached a very high frequency since the OHG period (for
details see Diewald 1999:321334).
It is interesting to note that as early as in the 16th century Veit Dietrich, who
in an edition of his sermons and educational texts makes ample use of modal
wollen- and sollen-constructions, always uses werden & infinitive for predictions,
prophecies and so on, i.e. in cases which would naturally afford a simple future
without modal connotations, as in example (10):
wer an mich glaubt / d wirt leben / ob er gleich stuerb / Vnd wer da lebt /vnd
(10)
glaubet an mich / der wirt nimmermehr sterben. (VD 7921ff.)
He that believes in me, yet shall he live, though he would die. And whoever
lives and believes in me shall never die.
Research on contemporary 16th century texts (especially on Luther) by Diewald

and Habermann (2005) strongly supports the observation that in this period,
rhetorically trained authors made a clear distinction between pure future (always
encoded by werdan) and modal future uses.
Summarizing this brief excursion into the 16th century, we may say that the
fact that the German modals semantically presuppose a modal source, i.e. an
instigator of the state of obligation, volition etc. impeded their interpretation as a
simple future tense marker. This means that, although these verbs, in the centuries
we are talking about, could be used to express future time reference, they were
not optimally suited to this function, and the interpretation of a modal as primar-
ily referring to future time always remained a conversational implicature in the
German modals.10
. As Harm (2001:297, 299) rightly points out, this, of course, does not imply that modals
in general are bad candidates for future grams (in the light of the findings of many gram-
maticalization studies and the development of English this obviously cannot be true); it only
shows that in the case of German they were not good enough, as there was a better candidate
for this function.
Quite different from the situation in OHG, as early as in OE, willan and scu-
lan in some uses had become quite close to future markers, expressing a mere
prediction,11 cf. examples (11) and (12):
Nu u miht ongitan hu hefig & hu earfoe is is eall to gerecanne; ac ic sceal
(11)
eah hwthwugu his onginnan e to tcanne, form ic hbbe ongiten t
hit is swie micel lcedom inre sorge, gif u ises auht ongitst, eah hit me
lange to lranne sie. (AB 127.21)
Now, thou canst perceive how heavy and how difficult it is to explain all
this; but nevertheless I will set to work to teach thee somewhat, for I have
noted that it is a powerful remedy for thy sorrow if thou understandest,
aught of this, though it be a long task for me to teach thee.
he cu t s Halgan Gstes lar wille fleon leasunga.
(12) (AC 243.14)
he said that the doctrine of the Holy Spirit will flee falsehood.
Although, even up to today, will and shall have not completely lost their modal
colouring, will/shall + infinitive constructions are the closest approximations to a
colourless, neutral future (Greenbaum & Quirk 1990:57), and they could be used
in such functions even in the earliest attested texts.
A check on the expressions of what comes closest to a mere future in the
section of Alfreds Boethius revealed the following frequency:
willan (11)
present indicative (5)
beon (4)
sculan (3)
present subjunctive (1)
weoran (1)
Here willan clearly dominates with 44% of all future expressions.

Although in Middle English, futurity is still much more frequently referred to
by the simple present than in Modern English, will and shall are gaining more and
more ground as future markers. Now shall becomes the usual means of indicating
futurity, while will continues to carry a strong volitional meaning until the end of
the Middle English period. In Late Modern English, however, will had become the
most frequently-used future marker, particularly in less formal registers.
5.2.2 Be/become-futures
German werdan, unfortunately, is not included in the language sample used by
Bybee, Perkins and Pagliuca (1994). There is, however, one passage (262ff.), where
. Kisbye (1971:111), Berndt (1982:148149), Jespersen (1909:275276), and Mustanoja

(1960:489), Denison (1993:304).
German werden is mentioned as possibly belonging into a subclass of source lex-

emes, of be and have constructions which follow a direct path [towards future]
with no intermediate obligation stage (263), i.e. werden is classified together with
the Latin -b- future, which derives from an Indo-European be-verb. For such
constructions the authors posit the following path (ibid.):
be, become, have/possession > predestination > intention > future
Apart from the fact that we do not believe that an intention-stage is necessary on
the path to futures (cf. also Ziegeler 2006; Hilpert 2007:38; Wischer 2008), or that
a sense of predestination must precede the future meaning, we would argue that
become-sources should be treated separately from be- and have/possession-
sources because of their inherent aspectual ingressive sense, which is lacking in
the other two.
In his very inspiring investigation, Fritz (2000:43) describes the semantic
structure of werdan as encoding the contrast between an original state and a final
state whereby the focus is on the transformative moment, i.e. on the feature of a
change of state; he also points to the fact that werdan, as opposed to the modals,
does not imply an instigator (or source) of the change ofstate.
Consequently, werdan never had an intentional meaning that would have to
be bleached out on its way to becoming a future. Thus, for become-sources
we would suggest the following semantic path (cf. also Diewald & Habermann
2005:237f.):
become/ingressive > future
Due to its lack of semantic restrictions, werdan has always been compatible with
subjects and predicative elements of any kind, which make it a good candidate for
auxiliarisation processes.
Furthermore, it can be observed that throughout the history of German, wer-
dan has displayed a high constructional variability. It has always been used simul-
taneously in a range of syntactic functions spanning from full verb via copula to
auxiliary. The construction types in which werdan occurred in our OHG corpus
are summarised in Table 7a:12
. Harm (2001:298) gives a fine-grained account of the various uses of werdan, which,
among other things, includes werdan in so-called Funktionsverbgefgen (phrasal verbs),
that is, in combination with prepositional phrases (e.g. MHG. ze leide werden to turn into
grief (for somebody), see Harm 2001:298). Furthermore, OHG werdan is also used in posses-
sive constructions, as in Tatian (2, 2): Inti ni uuard in sun lat.: et non erat illis filius and they
did not have a son. For our purpose, however, it is sufficient to distinguish the four classes
inTable 7a.
Table 7a. Construction types of OHG werdan

+ Past P 86 73 27 100 23 91 41 441

Copula 6 9 19 14 23 30 18 119
Intrans. 6 9 20 33 14 6 11 111
+ Pres P 0 2 1 1 0 0 0 9
+ Inf 0 0 0 2 0 0 0 2
S 98 105 67 150 60 132 70 682
The construction types for OE weoran and beon are listed in Tables 7b and c:
Table 7b. Construction types of OE weoran

AB CP AO WG LG C1 C2 GG S
+ PastP 11 32 27 1 1 10 31 5 118
Copula 18 17 6 4 2 6 7 3 63
Intrans. 4 3 11 10 47 4 11 1 91
+ PresP 2 2
S 33 54 44 15 50 20 49 9 274
Table 7c. Construction types of OE beon

AB CP AO WG LG C1 C2 GG S
+ PastP 38 114 2 13 34 2 19 222

Copula 126 115 5 22 13 10 3 10 304
Intrans. 24 15 6 1 46
+ PresP 1 1 2 4 1 9
+ to-inf 2 3 5
S 191 248 9 35 57 12 3 31 586
For a comparison of the lexical and periphrastic construction types of

werdan/beon in OHG and OE see Table 8:
Table 8. Comparison of the lexical and periphrastic construction types in OHG and OE
Construction Type OHG werdan OE weoran OE beon
Intransitive 16% 34% 7%

Copula 17% 22% 52%
S Lexical Verb 33% 56% 59%
(Continued)
Table 8. (Continued)
Construction Type OHG werdan OE weoran OE beon
+ PastP 65% 43% 38%
+ PresP 1% 1% 2%
+ to-Inf 1%
+ bare Inf 0%
S Potential Aux 66% 44% 41%
As can be seen, the potential auxiliary uses of werdan were much higher in
OHG than in OE. It even occurred though quite rarely with a bare infinitive in
OHG, cf. example (13).
Snter thaz giscrib min wirdit bzira sin, bazent sno gati thio mino
(13)
mssodati. (Otfrid V 25,45)
Sondern das Geschreib mein wird besser sein, ben seine (des Lesers)
Qualitten meine Versumnisse.
But my writing should be better, his (the readers) qualities atoning for my
failures.
6. Summary
Our comparative analysis of potential source lexemes for future grams in an OHG
and an OE text corpus provides evidence that the two languages display remark-
able differences with respect to the relevant items even in their earliest attested
stages.
These differences concern the frequency of usage, the value (in the sense of de
Saussure) of each item, i.e. the relations to its closest competitors/neighbours in the
word field, the semantic features, and the structural and constructional factors.
While from the perspective of cross-linguistic relevance (and frequency) of
grammaticalisation paths for futures, the be/become-futures surpass the modal
futures and thus should win out in a situation where both sources are present in
a language as werden did in German , Modern English opted for the modal
future due to disadvantages of OE werdan in terms of frequency and competition
by beon.
This is a clear indication that general grammaticalisation paths, invaluable
as they are for an overall evaluation of probabilities, have to be checked carefully
against the specific internal linguistic situation in a given synchronic stage of any
language under investigation.
We have shown that these internal factors provide a powerful motive for the
development against the more frequent universal pathways, and may promote a
divergent development of closely related languages with almost identical starting

conditions.
In this context, however, it is necessary to look at sociolinguistic factors as
well, in particular at language contact, which we did not do in this paper. So, this
remains a task for future work.
As to possible grammaticalisation paths for futures, we suggest, on the basis of
our findings, to refine previous models and to distinguish an additional grammati-
calisation path for become-sources of futures as set apart from be/have-sources.
Sources
OE
AB = Alfreds Boethius in: Sedgefield, Walter John (ed.). 1968. King Alfreds Old English Ver-
sion of Boethius De consolatione philosophiae. Darmstadt: Wissenschaftliche Buchgesell-
schaft, pp. 7479, 8289, 8994, 117124, 126135.
AC = Alfreds Cura Pastoralis in: Sweet, Henry (ed.). 1988. King Alfreds West Saxon Ver-
sion of Gregorys Pastoral Care. Part I. Early English Text Society. London: OUP, Kraus
Reprint, Millwood, N.Y. pp. 39, 2353, 237271, 387427.
AO = Alfreds Orosius in: Sweet, Henry (ed.). 1883. King Alfreds Orosius. Part I. Early Eng-
lish Text Society. London, pp. 5878, 102120, 228238.
WG = West Saxon Gospels in: Skeat, Rev. Walter W. (ed.). 1878. The Gospel According to Saint
John. In Anglo-Saxon and Northumbrian Versions Synoptically Arranged, with Collations
Exhibiting all the Readings of all the mss. Cambridge: CUP, pp. 12112.
LG = Lindisfarne Gospels in: Skeat, Rev. Walter W. (ed.). 1874. The Gospel According to Saint
Luke. In Anglo-Saxon and Northumbrian Versions Synoptically Arranged, with Collations
Exhibiting all the Readings of all the mss. Cambridge: CUP, pp. 15107.
C1 = Chronicle MS E (ca. 9701050) in: Plummer, Charles (ed.). 1952. Two of the Saxon
Chronicles Parallel. Vol. I, Ms. E, Oxford: Clarendon, pp. 119177.
C2 = Chronicle MS E (ca. 10701120): in: Plummer, Charles (ed.). 1952. Two of the Saxon
Chronicles Parallel. Vol. I, Ms. E, Oxford: Clarendon, pp. 205225, 239250.
GG = Gregory the Great in: Hecht, Hans (ed.). 1900. Bischofs Wrferth von Worcester
bersetzung der Dialoge Gregors des Groen ber das Leben und die Wunderthaten ital-
ienischer Vter und ber die Unsterblichkeit der Seelen. Bibliothek der angelschsischen
Prosa V, Leipzig: Georg H. Wigands Verlag, pp. 3443, 7079, 123133.
OHG
IS = Isidor, Pariser Codex (um 790): Hench, George Allison (Hg.). 1893. Der ahd. Isidor:
Facsimile-Ausgabe des Pariser Codex nebst critischem Texte der Pariser und Monseer
Bruchstcke. Mit Einleitung, grammatischer Darstellung und einem ausfhrlichen
Glossar. Mit 22 Tafeln. Straburg: Trbner [Quellen und Forschungen zur Sprach- und
Culturgeschichte der germanischen Vlker 72]. P. 244.
TA = Tatian, Evangelienharmonie (ca. 830): Sievers, Eduard (Hg.). 1966. Tatian. Lateinisch
und altdeutsch mit ausfhrlichem Glossar. 2. neubearbeitete Ausgabe. Nachdruck der
Ausgabe 1892. Paderborn: Schningh. [Bibliothek der ltesten deutschen Literaturden-
kmler 5]. P. 1385.
OT = Otfrid von Weienburg, Das Evangelienbuch (863/71): Erdmann, Oskar (Hg.). 1973.
Otfrids Evangelienbuch. 6. Aufl. besorgt von Ludwig Wolff. Tbingen: Niemeyer.
[Altdeutsche Textbibliothek 49]. P. 148.
NO = Notker von St. Gallen, De consolatione philosophiae (um 1025): Tax, Petrus W. (Hg.).
1988. Die Werke Notkers des Deutschen. Neue Ausgabe. Begonnen von Edward H.
Sehrt und Taylor Starck. Fortgesetzt von James C. King u. a. Bd. 2. Boethius: De Con-
solatione Philosophiae. Buch III. Tbingen: Niemeyer. P. 142222.
KA = Kaiserchronik (1135/55): Schrder, Edward (Hg.). 1984. Die Kaiserchronik eines
Regensburger Geistlichen. Unvernd. Nachdr. der Ausg. Hannover 1892. Mnchen:
Monumenta Germaniae Historica. (Monumenta Germaniae Historica. Deutsche
Chroniken. Scriptores qui vernacula lingua usi sunt, 1,1). P. 79124.
SP = Speculum ecclesiae deutsch (12. Jh.): Melbourn, Gert (Hg.). 1944. Speculum ecclesiae.
Eine frhmittelhochdeutsche Predigtsammlung (Cgm 39). Mit sprachlicher Einleitung
neu herausgegeben. Lund: Gleerup/ Kopenhagen: Munksgaard. [Lunder germanist-
ische Forschungen 12]. P. 137.
AL = Alexanderlied (Vorauer) (um 1140/50): Kinzel, Karl (Hg.). 1884. Lamprechts Alex-
ander. Nach dem Fragment des Alberic von Besancon und den lateinischen Quellen.
Halle: Waisenhaus. [Germanische Handbibliothek 6]. S.26172.
ENHG
VD = Veit Dietrich. Etliche Schrifften fr den gemeinen man/von vunterricht Christlicher
lehr vnd leben/vnnd zum trost der engstigen gewissen. Nrnberg 1548. Herausgege-
ben und mit einer Einleitung versehen von Oskar Reichmann. Assen: Gorcum, 1972
[Quellen & Forschungen zur Erbauungsliteratur des spten Mittelalters u. der frhen
Neuzeit 5].
References
Arnovick, Leslie Katherine. 1990. The Development of Future Constructions in English. The Prag-
matics of Modal and Temporal Will and Shall in Middle English [Berkeley Insights in Lin-
guistics and Semiotics 2]. Frankfurt: Peter Lang.
Berndt, Rolf. 1982. A History of the English Language. Leipzig: Enzyklopdie.
Bogner, Istvan. 1989. Zur Entwicklung der periphrastischen Futurformen im Frhneuhoch-
deutschen. Zeitschrift fr Deutsche Philologie 108: 5686.
Bybee, Joan, Pagliuca, William & Perkins, Revere D. 1991. Back to the future, In Approaches to
Grammaticalization, Vol. II: Focus on Types of Grammatical Markers [Typological Studies
in Language 19(2)], Elizabeth Closs Traugott & Bernd Heine (eds), 1758. Amsterdam:
John Benjamins.
Bybee, Joan, Perkins, Revere & Pagliuca, William. 1994. The Evolution of Grammar. Tense, Aspect
and Modality in the Languags of the World. Chicago IL: The University of Chicago Press.
Bybee, Joan. 2010. Language, Usage and Cognition. Cambridge: CUP.

Denison, David. 1993. English Historical Syntax. London: Longman.
Diewald, Gabriele. 1999. Die Modalverben im Deutschen. Grammatikalisierung und Polyfunk-
tionalitt. Tbingen: Niemeyer.
Diewald, Gabriele, Habermann, Mechthild, Lutz, Angelika & Wischer, Ilse. 2002. The gram-
maticalization of future markers in German and English: A comparative study. Paper given
at the New Reflections on Grammaticalization 2 Conference in Amsterdam, April 2002
[manuscript].
Diewald, Gabriele & Habermann, Mechthild. 2005. Die Entwicklung von werden & Infinitiv
als Futurgrammem: Ein Beispiel fr das Zusammenwirken von Grammatikalisierung,
Sprachkontakt und soziokulturellen Faktoren. In, Grammatikalisierung im Deutschen,
Torsten Leuschner, Tanja Mortelmans & Sarah De Groodt (eds), 229250. Berlin: Walter
de Gruyter.
Fritz, Thomas. 2000. Wahr-Sagen: Futur, Modalitt und Sprecherbezug im Deutschen. Hamburg:
Buske.
Greenbaum, Sidney & Quirk, Randolph. 1990. A Students Grammar of the English Language.
Hong Kong: Longman.
Hacke, Marion. 2009. Funktion und Bedeutung von werden + Infinitiv im Vergleich zum
futurischen Prsens. Heidelberg: Germanistische Bibliothek; 34 (Teilw. zugl. Bamberg,
Otto-Friedrich-Univ., Diss., 2008).
Harm, Volker. 2001. Zur Herausbildung der deutschen Futurumschreibung mit werden +
Infinitiv. Zeitschrift fr Dialektologie und Linguistik 68, 288307.
Hilpert, Martin. 2007. Germanic Future Constructions: A Usage-Based Approach to Gram-
maticalization. Ph.D. dissertation, Rice University. http://scholarship.rice.edu/bitstream/
handle/1911/20683/3257333.PDF?sequence=1
Huddleston, Rodney. 1976. Some theoretical issues in the description of the English verb. Lin-
gua 40: 33183.
Jespersen, Otto. 1909. A Modern English Grammar on Historical Principles. Heidelberg: Winter.
Kilpi, Matti. 1989. Passive Constructions in Old English Translations from Latin [Mmoires de la
Societe Neo-Philologique a Helsingfors 49]. Helsinki: Univerisity of Helsinki.
Kilpi, Matti. 1992. Dictionary of Old English: BEON. With attested Spellings by Robert Millar
Using Materials Assembled by Haruko Momma. Toronto: Pontifical Institute of Mediaeval
Studies.
Kisbye, Torben. 1971. An Historical Outline of English Syntax, Vol 1. Aarhus: Akademisk
Boghandel.
Kleiner, Mathilde. 1925. Zur Entwicklung der Futur-Umschreibung werden mit dem Infinitiv.
University of California Publications in Modern Philology 12: 1101.
Kotin, Michail. 2003. Die werden-Perspektive und die werden-Periphrasen im Deutschen.
Frankfurt: Peter Lang.
Krmer, Sabine. 2005. Synchrone Analyse als Fenster zur Diachronie. Die Grammatikalisierung
von werden + Infinitiv [LINCOM Studies in Germanic Linguistics 23]. Mnchen: Lincom.
Leiss, Elisabeth. 1985. Zur Entstehung des neuhochdeutschen analytischen Futurs. Sprachwis-
senschaft 10: 250273.
Lutz, Angelika. 2009. Celtic influence on Old English and West Germanic. English Language and
Linguistics 13(2): 227249.
Mitchell, Bruce. 1985. Old English Syntax, Vol. I. Oxford: OUP.
Mustanoja, Tauno F. 1960. A Middle English Syntax, Part 1. Helsinki: Socit Neophilologique.
Rogler, Gudrun. 2010. Die Versprachlichung des Konzepts werden: mutative Prdikate und ihre
Funktion im Sprachsystem: eine Untersuchung zum Spanischen und Deutschen unter Ein-
beziehung eines sprachtypologischen Hintergrunds. Aachen: Shaker Verlag. (Zugl.: Bielefeld,
Univ., Diss., 2009).
Saltveit, Laurits. 1962. Studien zum deutschen Futur. Die Fgungen werden mit Partizip des Pr-
sens und werden mit dem Infinitiv in ihren heutigen Funktionen und in ihrer geschichtlichen
Entwicklung. Oslo: Norwegian Universities Press.
Schieb, Gabriele (ed.). 1981. Beitrge zur Erforschung der deutschen Sprache, 1. Bd. Leipzig:
Bibliographisches Institut.
Schmid, Hans Ulrich. 2000. Die Ausbildung des werden-Futurs. berlegungen auf der Grund-
lage mittelalterlicher Endzeitprophezeihungen. Zeitschrift fr Dialektologie und Linguistik
67: 627.
Sommer, Thomas. 1994. Flexionsmorphologie des Verbs im althochdeutschen Tatian. Mnchen:
tuduv.
Smirnova, Elena. 2006. Die Entwicklung der Konstruktion wrde + Infinitiv im Deutschen:
eine funktional-semantische Analyse unter besonderer Bercksichtigung sprachhistorischer
Aspekte. Berlin: Walter de Gruyter.
van Kemenade, Ans. 1993. The history of English modals: A reanalysis. Folia Lingustica
Historica. 13(12): 143166.
Visser, Fredericus T. 19631973. An Historical Syntax of the English Language. Brill: Leiden.
Walther, Claus. 1980. Untersuchungen zu Hufigkeit und Funktionen des deutschen Futurs
(werden + Inf.) in hochdeutschen Texten zwischen 1450 und 1750 (mit einem Ausblick ins
Niederdeutsche). Ph.D. dissertation, Humboldt-Universitt zu Berlin.
Westvik, Olaf Jansen. 2000. ber Herkunft und Geschichte des werden-Futurs. Eine
Auseinandersetzung mit neuerer und neuester Forschung. In Raum, Zeit, Medium
Sprache und ihre Determinanten. Festschrift fr Hans Ramge zum 60. Geburtstag, Gerd
Richter, Jrg Riecke & Britt-Marie Schuster (eds), 235261. Darmstadt: Hessische

Historische Kommission.
Wischer, Ilse. 2006. Markers of Futurity in Old English and the Grammaticalization of shall and
will. Studia Anglica Posnaniensia 42: 165178.
Wischer, Ilse. 2008. Will and Shall as markers of modality and/or futurity in Middle English.
Folia Linguistica Historica 29(1): 125143.
Wischer, Ilse. 2010. On the use of beon and wesan in Old English. In English Historical Lin-
guistics 2008, Vol. I: The History of English Verbal and Nominal Constructions [Current
Issues in Linguistic Theory 314], Ursula Lenker, Judith Huber & Robert Mailhammer (eds),
217235. Amsterdam: John Benjamins.
Ziegeler, Debra. 2006. Omnitemporal will. Language Sciences 28(1): 76119.
The Verb to be in the West Saxon Gospels
andthe Lindisfarne Gospels
Christine Bolze
University of Cambridge, UK
This paper examines the distribution and use of the twofold present
tenseparadigms of the Old English verb bon to be in the late West Saxon
Gospels (WSG) and the Northumbrian gloss to the Lindisfarne Gospels
(LiGl). Theanalysis confirms the frequently claimed semantic distinction of
the paradigms. Itfurthermore shows that the choice of a form of OE bon
mostlydepended on the tense and mood of its Latin equivalent, but it also
illustrates that the translators of the two Gospel versions took the context into
account. Quantitative differences of the forms in the two manuscripts are due
to multiple glosses in Lindisfarne, the use of alternative forms to OE bon in
the WestSaxon Gospels andthe partly different morphology of the verb in
Northumbrian.
1. Introduction
The Old English verb bon to be shows a partially twofold morphology: it con-
sists of a dual paradigm in the present indicative and subjunctive, as well as in
the imperative, infinitive and present participle. Table 1 illustrates its West Saxon
structure (cf. Brunner 1942:371374).
The forms presumably developed out of four Indo-European roots: the *bheu-
root, the *es-root, the *er/*or-root and the *wes-root (cf. Brunner 1942:371;
Hogg& Fulk 2011:309). The *bheu-root has the meaning of to become and in Old
English exhibits a separate paradigm. This phenomenon is unparalleled in West
Germanic and has been ascribed to Celtic-Germanic language contact (cf.Keller
1925; Schumacher 2007; Lutz 2009 and Wischer 2010).
Handbooks state a semantic distinction between the two present indicative
paradigms: b-forms are said to imply futurity, whereas non-b-forms refer to the
present moment (cf. Campbell 1959:350351; Brunner 1962:276; Hogg & Fulk
2011:309310). Occasionally the limits of this distinction are acknowledged and
further semantic functions of the b-paradigm are proposed. Brunner (1962:276)
Christine Bolze
Table 1. The twofold paradigm of OE bon in West Saxon

Present indicative Present subjunctive Imperative
s/r-root b-root s-root b-root w-root b-root
Sg 1 eom beo
2 eart bist, byst sie/si/sy beo wes beo
3 is bi, by
Pl sint beo sien/sin/syn beon wesa beo
sindon/-un
Infinitive: wesan, bon
Present Participle: wesende, beonde
mentions that the b-forms haben oft, aber nicht immer, futurischen Sinn,1
Campbell (1959:350) argues that they express an invariable fact and an iterative
extension into the future, Hogg and Fulk (2011:310) refer to their use in habitual
conditions, and Kilpi (1992: E.13) argues that they also occur in actional pas-
sive constructions and with durative and generic references.
Regarding the twofold paradigm in the subjunctive, the information in
grammars and handbooks is sparse. It has been argued that the b-paradigm also
indicates futurity in the subjunctive (Brunner 1942:371). Furthermore, Brunner
(1962:277) and Campbell (1959:351) claim a dialectal distinction of the b- and
s-subjunctives. They declare that there are only b-subjunctives in East Mercian,
i.e. in the gloss to the Gospel of St Matthew in Rushworth 1 (Bodleian, Auct. D. ii).
Mitchell (1985 I: 263, fn. 147) restates the assumed dialectal difference between
the subjunctive paradigms. By contrast, Kilpi (1989:73) and Wischer (2010:230)
refute it by showing that s-subjunctives are more frequent than b-subjunctives in
Rushworth 1.
In-depth studies on the use and function of the twofold paradigm confirm
that indicative b-forms are preferred to imply futurity (Jost 1909; Exter 1911;
Kilpi 1989, 1993; Wischer 2010). Kilpi (1989:69) and Wischer (2010:230)
also demonstrate the future reference of subjunctive b-forms. Nevertheless, no
previous study examined the forms in northern Old English (Northumbrian)
in comparison to the southern dialect; i.e. West Saxon. Table 2 shows that the
structure of the verb in Northumbrian varies from West Saxon (cf. Brunner
1942:371374):
. [the b-forms] frequently but not always have a future implication.

The Verb to be in the West Saxon Gospels and the Lindisfarne Gospels
Table 2. The twofold paradigm of OE bon in Northumbrian

Present indicative Present subjunctive Imperative
s/r-root b-root s-root b-root w-root b-root

Sg 1 am biom
2 ar bist sie/se (bia, bie) wes
3 is bi
Pl sint bion/-un
sindun/-n bioon/-un sie/se wosa
aron/-un bia
Infinitive: wosa, (bian)
Table 2 indicates that in contrast to the morphology of the verb in West

Saxon (cf. Table 1), a full separate b-paradigm is only attested in the present
indicative. There is no imperative b-form, and the infinitive bian and the two
forms bia and bie, which are assumed to be present subjunctives, are evinced
only once and have therefore been parenthesized. Furthermore, in the Nor-
thumbrian paradigm there are the r-roots aron/arun in the plural, which do not
occur at all in West Saxon.
In the following, I examine the distribution and use of the present indicative
and subjunctive paradigms in the Lindisfarne Gospels and earliest preserved ver-
sion of the West Saxon Gospels. I scrutinize in what semantic environments the
forms of OE bon occur and discuss to what extent the Latin Gospels, the transla-
tion type or dialectal differences in the structure of OE bon might have triggered
them. The West Saxon Gospels are dated to the early 11th century and preserved
in the Corpus Manuscript (CCCC 140). The Old English Lindisfarne Gospels are
preserved in Aldreds 10th century Northumbrian gloss to the Latin Gospels in the
Lindisfarne Manuscript (BL, Cotton Nero D iv.). I collected the present tense forms
of OE bon from the Dictionary of Old English Corpus (DOEC).2 As regards the
Lindisfarne gloss, the Latin text is included in DOEC. Concerning the West Saxon
Gospels, I used the Vulgate edition by Weber and Gryson (2007) to compare the
Old English forms to their Latin counterparts.
. The study is based on the forms appearing in Lk (WSCp), LkGl (Li), Jn (WSCp), JnGl (Li),
Mk (WSCp), MkGl (Li), Mt (WSCp) and MtGl (Li).
Christine Bolze
2. The present indicative
2.1 Distribution
There are 3497 present indicative b- and non-b-forms in both manuscripts, 1647
in the West Saxon Gospels and 1850 in the Lindisfarne Gospels. Table 3 shows their
distribution:3
Table 3. Frequency of present indicative b- and non-b-forms in the West Saxon Gospels
and the Lindisfarne Gospels (starred figures are significant)
WSG % LiGl % Sum 2 score p-value
non-b-form 1251 75.96 1281 69.24 2532 5.42 0.02

b-form 396 24.04 569 30.76 965 14.23 0.0002*
1647 100 1850 100 3497 19.65 <.0001*
The figures illustrate that there are more non-b- than b-forms in both man-
uscripts. This result is in line with the findings of previous research (cf. Kilpi
1993:98 and 1997:90; Wischer 2010:222). Furthermore, the quantitative differ-
ence of present indicative forms in the two Gospel versions is statistically signifi-
cant at p <.0001. Of particular interest is the higher frequency of b-forms in the
Lindisfarne Gospels (30.76%) compared to the West Saxon Gospels (24.04%), which
is of statistical significance at p = 0.0002. The following analyses will elucidate
these differences.
2.2 Time reference

A present indicative form of OE bon can be identified as a future marker when
it refers to a Latin future tense form (cf. Kilpi 1989, 1993). In the West Saxon
Gospels, 271 (68.43%) of the 396 b-forms and 7 (0.55%) of the 1251 non-b-forms
refer to a Latin future tense form. This is similar in the Lindisfarne Gospels, in
which 327 (57.46%) of the 569 b-forms and only 1 (0.07%) of the 1281 non-
b-forms render a Latin verb inflected for the future tense. Thus, more than half of
the b-forms in each manuscript imply futurity based on their translation of Latin
future tense forms. Example (1) illustrates this:
(1) Mt 22.28
L in resurrectione (ergo) cuius erit [3sg fut.ind.act.] de septem uxor
WSG hwylces ra sufona by t wif on am riste
. Instances in which his represents is are included in the figures. Instances in which is is a
scribal mistake for his, ic and in are excluded from the figures.
LiGl in erist (foron) hus bi of m seofonum t wif

(Therefore), in the resurrection, whose wife will she be of the
seven?
This confirms the assumed future implication of the indicative b-paradigm.

Non-b-forms are favoured in sentences without a future reference. They are pre-
dominantly used to translate Latin present tense forms: 921 (73.62%) of the 1251
non-b-forms in the West Saxon Gospels and 987 (77.04%) of the 1281 non-b-forms
in the Lindisfarne Gospels render a Latin verb in the present tense; cf. (2):
(2) Jn 4.23
L sed venit hora et nunc est [3sg pres.ind.act.]
WSG ac seo tid cym & nu is
LiGl ah cuom sio tid & nu is
But the hour/the time has come, and now it is.
Overall, the distribution of the forms in the two Gospel versions verifies the
claimed preference for b-forms in future references and the preference for non-b-
forms in statements with a reference to the present time.
2.3 Differences between the Gospel versions

The significantly increased ratio of b-forms in Lindisfarne has two main causes:
firstly, the use of alternative forms to express futurity in the West Saxon Gospels,
and secondly, the occurrence of multiple glosses involving two or more b-forms in
the Lindisfarne Gospels. The latter phenomenon is exemplified by (3):
(3) Mk 16.16
L qui vero non crediderit condemnabitur [3sg fut.ind.pass.]
WSG solice se e ne gelyf, se bi genyerod
LiGl see uutedlice ne gelefe gehened bi genirad bi
Truly, the one who does not believe will be condemned.
Multiple glosses of this sort are relatively frequent in the Lindisfarne Gospels and
thus increase the total frequency of b-forms compared to the West Saxon Gospels.
The use of alternative forms to indicate futurity in the West Saxon Gospels is, how-
ever, a more significant cause of the increased share of b-forms in Lindisfarne. The
West Saxon Gospels frequently show full verbs inflected for the present indicative
when Aldred employs b-forms; cf. (4):
(4) Jn 8.24
L dixi (ergo) uobis quia moriemini [2pl fut.ind.dep.] in peccatis uestris
WSG ic eow sde t ge swelta on eowrum synnum
LiGl ic cue iuh tte gie bion deada in iuero synno
I (thus) told you because/that you will die in your sins.
Christine Bolze
Brunner (1962:288) points out that the present indicative is a common way to
indicate a future reference in Old English due to the absence of separate inflections
for the future tense.
Moreover, the West Saxon Gospels frequently use geweoran (to happen,
to become) to indicate futurity, in particular to render future and present tense
forms of L fieri (to become). The equivalent verses in Lindisfarne usually show
b-forms as in (5) and (6):
(5) Mk 13.19
L erunt enim dies illi tribulationes tales quales non fuerunt ab
initiocreaturae quam condidit Deus usque nunc neque fient
[3pl fut.ind.]
WSG solice on am dagum beo swylce gedrefednessa swylce ne
gewurdon of frymme re gesceafte e god gesceop o nu ne na
negewura
LiGl bion foron dagas a costungo uslico sulce suelco ne
woeronfrom fruma s sceftes one gesceop god wi nu c
nebion
For in those days there will be such tribulation as has not been from
the beginning of the creation that God created until now, and never
will be.
(6) Mk 4.11
L qui foris sunt in parabolis omnia fiunt [3pl pres.ind.]
WSG am e ute synt ealle ing on bigspellum gewura
LiGl ae uta sint in bispellum alle bion
For those who are outside everything will be in parables.
The preference for OE geweoran in the West Saxon Gospels and for b-forms in the
Lindisfarne Gospels to convey the implied future reference of L fieri might be a dia-
lectal difference, or an individual translating preference. In any case, it contributes
to the increased ratio of indicative b-forms in the Lindisfarne gloss.
Furthermore, there are occasionally b-subjunctives in the West Saxon Gospels
when the corresponding verses in Lindisfarne show indicative b-forms. This is pre-
sumably due to the absence of a Northumbrian b-paradigm in the subjunctive
(cf.Table2). An example is in (7):
(7) Lk 21.6
L haec quae videtis venient dies in quibus non relinquetur lapis super
lapidem qui non destruatur [3sg pres.subj.pass.]
WSG as ing e ge geseo a dagas cuma on am ne bi stan lfed ofer
stan, e ne beo toworpen
LiGl as ae gie gesegon gesea cyma dagas on m ne bi forleten

stan ofer stane see ne bi tostrogden
These things that you see, the days will come in which a stone will
not be left on a stone that will not be destroyed.
Hence, the significantly increased number of indicative b-forms in Aldreds gloss

to the Lindisfarne Gospels is to some extent due to the northern dialect in which it
is preserved: the non-availability of subjunctive b-forms in Northumbrian contrib-
utes to the higher share of indicative b-forms in the Lindisfarne Gospels. Overall,
we noticed three alternative means to express futurity in the West Saxon Gospels:
full verbs inflected for the present indicative, OE geweoran, and b-subjunctives
of OE bon.
2.4 Aspectual references

2.4.1 Durative and iterative references
As indicated above, it is claimed that b-forms are preferred in certain aspectual
references, such as in durative, iterative, habitual and generic references. Kilpi
(1993:106107) notices b-forms in conjunction with linguistic markers imply-
ing a durative or an iterative sense. For instance, adverbs such as OE oft often
and onne whenever, every time imply an iterative sense. A durative implication
can be inferred when the verb occurs with adverbs and adverbials such as OE
simle always, ece ever, eternally, heononfor from now on and a hwile e while,
aslong as. Table 4 shows how frequently such markers occur in conjunction with
b- and non-b-forms in the two Gospel versions.
Table 4. Frequency of b- and non-b-forms with iterative and durative markers

in the West Saxon Gospels and the Lindisfarne Gospels
WSG % LiGl %
iterative
non-b-form 0/1251 0.00 0/1281 0.00
b-form 11/396 2.77 11/569 1.93
durative
non-b-form 10/1251 0.79 14/1281 1.09
b-form 14/396 3.53 17/569 2.98
The figures imply that there are not many iterative or durative markers in
combination with present tense forms of OE bon in the Old English Gospels.
They also indicate that iterative references exclusively occur with b-forms. By con-
trast, durative markers can be detected in conjunction with b- and non-b-forms.
Christine Bolze
This verifies Kilpis observations in the Helsinki Corpus (Kilpi 1993:106107).

Nevertheless, the comparison with Latin implies that the choice of the para-
digm mostly depends on the tense and mood of the equivalent Latin form: Latin
future tense forms are rendered by b-forms, and Latin present tense forms are
rendered by non-b-forms. B-forms in sentences that can be described as itera-
tive predominantly refer to the formally ambiguous L fuerit (s/he will/may have
been), which may denote a future perfect indicative active or a perfect subjunc-
tive active; cf. (8):
(8) Jn 2.10
L omnis homo primum bonum vinum ponit et cum inebriati
[pastpart.] fuerint [3pl futperf.ind.act/perf.subj.act.] (tunc) id
quoddeterius est
WSG lc man syl rest god win & onne hig druncene beo t e
wyrse by
LiGl aelc mon one forma rist t god uin sette & miy indrungno
bion (onne) t tte wurresta bi
Every man serves the good wine first, and when(ever)/each time they
are drunk, (then) that which is worse.
Thus, despite the iterative quality of the verse, it is likely that L fuerint and its pos-
sible future reference prompted the b-constructions in the Gospels. There is in fact
only one iterative b-form in which the direct Latin equivalent is unlikely to have
prompted it:
(9) Lk 11.21
L cum fortis armatus custodit atrium suum in pace sunt [3pl pres.ind.
act.] ea quae possidet
WSG onne se stranga gewpnud his cafertun gehealt, onne beo on
sibbe a ing e he ah
LiGl miy se stronga woepenberend gehalda ceafertun his in sibb bion
a ae agnage
When(ever) a strong, armed man guards his palace, the things that
he possesses are in peace.
In (9), the iterative sense is implied by correlated onneonne in the West

Saxon Gospels and by miy in the Lindisfarne Gospels. There is a clear iterative
quality in this verse, and it is possible that it prompted the b-forms in the Old
English Gospels although the corresponding Latin form is in the present tense.
Nonetheless, the tradition of the older Latin Bible indicates that this verse is also
attested with the 3pl fut.ind.act. erunt (cf. Jlicher et al. 1970); thus, a potential
future implication of the b-forms in (9) can yet be hypothesized. Regarding the
West Saxon Gospels, this variation in the Latin Gospels is significant since it is
unknown and impossible to determine what Latin Gospel text(s) the Anglo-
Saxon translator(s) used (cf. Liuzza 2000:49). It is also relevant for explaining
the b-form in the Lindisfarne gloss. Ross (1932, 1981) provides evidence that
Aldred seems to have occasionally followed another Latin manuscript while
glossing Lindisfarne. It is hence possible that the Anglo-Saxon translators had
access to manuscripts in which the Latin present tense form in (9) occurred in
the future tense and that the b-forms are hence meant to indicate futurity.
The translators dependence on Latin is also visible in sentences with durative
markers. The verses in (10) and (11) show this:
(10) Mk 9.19
L generatio incredula quamdiu apud vos ero [1sg fut.ind.act.]
WSG ungeleaffulle cneorys swa lange swa ic mid eow beo
LiGl cnewreso ungeleaffull a huile mi iuh ic beom
Faithless generation, how long shall I be with you?
(11) Jn 9.4
L me oportet operari opera eius qui misit me donec dies est
[3sg pres.ind.act.]
WSG me gebyra to wyrceanne s weorc e me sende a hwile e
hitdg is
LiGl me gedfna t ic geuyrco uoerca his see sende mec a huile
isdge
We must work the works of him who sent me while/as long as it
isday.
The idea of durativity is not quite the same in (10) and (11). However, they contain
markers which according to Kilpi (1993:106) imply a durative quality. The sen-
tences in (10) and (11) illustrate that such durative markers can occur with b- and
non-b-forms, depending on the tense of their Latin counterparts. This suggests
that it is unlikely that a durative marker alone influenced the choice of the para-
digm of OE bon. The following section strengthens the necessity to consider the
Latin when examining the semantics of the Old English forms.
2.4.2 Habitual and generic references

Kilpi (1989:97 and 1993:107110) notices a preferred use of b-forms in generic
sentences; i.e. with references to a kind or class, and in conjunction with generic
pronouns of the type see (the one who). The occurrence of b-forms in habitual
statements is recently shown in Schumacher (2007), Lutz (2009) and Wischer
(2010). Table 5 illustrates the occurrence of b- and non-b-forms in sentences with
a generic and habitual quality in the Old English Gospels. As in Kilpi (1989,
1993), the analysis covers b- and non-b-forms in the 3sg and 3pl.
Christine Bolze
Table 5. Frequency of b- and non-b-forms in generic and habitual references

in the West Saxon Gospels and the Lindisfarne Gospels (3sg and 3pl)
WSG % LiGl %
generic
non-b-form 109/1251 8.71 93/1281 7.25
b-form 71/396 17.92 106/569 18.62
habitual
non-b-form 37/1251 2.95 31/1281 2.41
b-form 43/396 10.85 42/569 7.38
The figures reveal that c. 18% of the total share of b-forms in the two Gos-
pel versions can be described as generic. The frequencies of habitual b-forms are
lower: 10.85% of the total share of b-forms in the West Saxon Gospels and 7.38% of
that in the Lindisfarne Gospels have a habitual implication. Furthermore, there are
more b- than non-b-forms in generic and habitual references. Their choice is once
again mostly dependent on Latin: b-forms tend to occur when there is a future
tense form in the corresponding Latin verse; cf. the verses with a generic quality
in (12) and (13):
(12) Mt 8.12
L filii autem regni eicientur [3pl fut.ind.pass.] in tenebras exteriores
WSG witodlice ises rices bearn beo aworpene on a ytemestan ystro
LiGl suna uutedlice rices bion gedrifen in yostrum wytmesto
Then the sons of the kingdom will be thrown into the outer d
arkness.
(13) Mk 13.16
L et qui in agro erit [3sg fut.ind.act.] non revertatur retro tollere
vestimentum suum
WSG & se e bi on cere ne cyrre he ongean t he his reaf nime
LiGl & see on lond bi ne eft gecerres on bcg to niommanne woede his
And the one who will be in the field, turn not again behind to take
his cloth.
The assumption that the Latin rather than the generic referent triggered these
b-forms is strengthened by (14) and (15):
(14) Mt 10.24
L non est [3sg pres.ind.act.] discipulus super magistrum nec servus
super dominum suum
WSG nys se leorningcniht ofer hys lareow, ne eow ofer hys hlaford
LiGl ne is egn ofer one laruu nec ea esne ofer hlaferde his
The disciple is not above his teacher, nor the servant above his master.
(15) Mk 9.40
L qui (enim) non est [3sg pres.ind.act.] aduersum uos pro uobis est
[3sg pres.ind.act.]
WSG se e nis agen eow se is for eow
LiGl see (foron) ne is wi iuih fore iuih is
(For) the one who is not against you is for you.
In (14), the generic reference is to se leorningcniht the disciple in the West Saxon
Gospels and to egn in the Lindisfarne Gospels. It is generic since it does not refer
to one certain disciple but to disciples in general. In (15) there is the same generic
pronoun as in (13). Nevertheless, unlike in (12) and (13) there are non-b-forms
in (14) and (15); they reflect the corresponding Latin present tense forms. The
generic quality of the sentence did not seem to be significant for the choice of the
paradigm of OE bon. My findings therefore do not corroborate Kilpi (1993:110),
who pointed out that future and generic references of the b-forms in the Helsinki
Corpus are almost always mutually exclusive. The differences between Kilpis
and my results are likely to be due to text type and the fact that Kilpi does not
compare all the b- and non-b-forms in the Helsinki Corpus to their Latin corre-
spondents; he only considers the Latin included in the Corpus. Since my analysis
is based on a comparison of each Old English form to its Latin equivalent, it is
sensible to assume that Kilpi might have come to a different conclusion had he
fully examined the tense of the corresponding Latin forms.
Concerning the distribution of the forms in habitual references in the two Gos-
pel versions, one cannot ignore the influence of Latin either. The habitual b-forms
render Latin future tense forms and/or can also be interpreted as future references,
whereas the non-b-forms refer to Latin present tense forms. The verses in (14)
and (15), for instance, have a habitual quality and show non-b-forms, presumably
due to the Latin present tense verbs. By and large, my data does not allow one
to conclude that the indicative b-forms are marked for habituality as those in the
Celtic languages. In Old Irish and Middle Welsh it is the main function of the sepa-
rate present indicative b-paradigm of to be to indicate habituality (for examples
cf.Schumacher 2007:187188). The b-forms in the Old English Gospels, however,
are predominantly used to imply futurity. The presumed habitual quality of the Old
English b-forms thus needs to be further examined.
Overall, the use of the two present indicative paradigms of OE bon in the
West Saxon Gospels and the Lindisfarne Gospels verifies the preference for b-forms
to refer to the future. Their claimed use in certain aspectual references could also
be observed; however, they usually translate Latin future tense forms nevertheless.
It is therefore often uncertain to what extent an implied durative, iterative, generic
or habitual quality of a sentence influenced the translators choice of the paradigm
of to be in the Old English Gospels.
Christine Bolze
3. The present subjunctive
3.1 Distribution
Table 6 illustrates the frequency of the present subjunctive forms of OE bon in the
two Gospel versions.
Table 6. Frequency of present subjunctive b- and s-forms in the West Saxon Gospels
andthe Lindisfarne Gospels
WSG % LiGl %
s-form 172 82.29 228 99.13

b-form 37 17.71 2 0.87%
210 100 230 100
The figures reveal that s-forms occur far more frequently than b-forms. This
finding supports Kilpi (1997:90) and Wischer (2010:229). Of particular interest
is the increased share of s-forms in Northumbrian: there are 228 s-subjunctives in
the Lindisfarne Gospels as opposed to 172 in the West Saxon Gospels. As indicated
in Table2 above, there is no separate subjunctive b-paradigm in Northumbrian.
However, the two b-forms bie and bia in the Lindisfarne Gospels are generally con-
sidered to be b-subjunctives (cf. Brunner 1942:373; Campbell 1959:350, Hogg &
Fulk 2011:312).
3.2 Time reference

As indicated above, Brunner (1942:371) implies that not only indicative but also
subjunctive b-forms have a future reference. Subjunctive b- and s-forms occasion-
ally correspond to Latin future tense forms: 4 (10.81%) of the 37 b-subjunctives and
2 (1.16%) of the 172 s-subjunctives in the West Saxon Gospels translate a Latin verb
inflected for the future tense. This confirms that b- rather than s-subjunctives occur
in future references. In the Lindisfarne Gospels, 17 (7.45%) of the 228 s-forms gloss
a Latin future tense form. Interestingly, they frequently occur in multiple glosses
with a present indicative b-form. Example (16) illustrates this:
(16) Mt 23.11
L qui maior est uestrum erit [3sg fut.ind.act.] minister uester
WSG see eower yltst sy beo se eower en
LiGl see heist maas is iuer bi sie embihtmonn iuer
He who is greatest of you will be your servant.
The verse in (16) has a future implication that is indicated by the Latin 3sg fut.
ind.act. erit (s/he will be). It also expresses the speakers opinion: it can be read
as a wish or desire. This frequently triggers the subjunctive mood in Old E nglish
(Mitchell 1967:146). The combination of a future implication and a personal
opinion might have prompted the b-subjunctive in the West Saxon Gospels and
the double gloss of a b-indicative and an s-subjunctive in the Lindisfarne Gospels.
Since there was no separate subjunctive b-paradigm in Northumbrian, Aldred
might have used the b-indicative to render the future tense of the Latin verb and
the s-subjunctive to convey the implied wish or desire.
Furthermore, the two b-forms bia and bie in the Lindisfarne Gospels, which
are considered to be subjunctives, appear in doublets with s-subjunctives as (17)
and (18) demonstrate:
(17) Mt 20.27
L et qui uoluerit inter uos primus esse erit [3sg fut.ind.act.] uester seruus
LiGl & see wlle betuih iuh formest foruuost wossa sie bia iuer ea
egn
And the one who will be first among you will be your servant.
(18) Mk 10.44
L et quicumque uoluerit in uobis primus esse erit [3sg fut.ind.act.]
omnium seruus
LiGl & sua hu see wlle in iuh formest wosa bie se allra rl esne
And whoever will be first among you will be servant of all.
The verses are semantically similar, and as in (16) the future implication of L erit
and the expression of a wish are likely to have prompted Aldred to employ a b- and
an s-form. However, in (16) we observed a double gloss of an indicative b-form
and a subjunctive s-form. This is also the case in (19) and (20):
(19) Mk 9.35
L si quis uult primus esse erit [3sg fut.ind.act.] omnium nouissimus
etomnium minister
LiGl gif hua wlle formest wosa bi sie allra hltmst & allra
embehtmonn
If anyone wants to be first, he will be last of all and servant of all.
(20) Mk 10.43
L quicumque uoluerit fieri maior erit [3sg fut.ind.act.] uester minister
LiGl sua hua see wlle wosa maara hera bie sie iwer hera
embehtmonn
Whoever wants to become/to be greater will be your servant.
The verse in (20) represents Mk 10.43 and shows the only instance of the disyl-
labic form bie in the Lindisfarne Gospels. Brunner (1942:373) and Hogg and Fulk
(2011:312) interpret it as a 3sg pres.ind.act. The form bie in (18) appears in the fol-
lowing verse, i.e. Mk 10.44. Therefore, it is debatable whether bie in fact represents
a subjunctive. It is possible that Aldred forgot to add the final - in Mk 10.44, and
Christine Bolze
thus bie in (18) might be a glossing mistake and denote a present indicative b-form
as bie in (20). By and large, the fact that the five verses in (16) to (20) are semanti-
cally similar and for the most part show a double gloss involving an s-subjunctive
and a b-indicative in reference to the Latin 3sg fut.ind.act. erit in the Lindisfarne
Gospels, it may be that bie and bia rather denote indicative than subjunctive forms.
Moreover, doublets of indicative b- and subjunctive s-forms are a major cause of
the increased frequency of s-subjunctives in the Lindisfarne Gospels which Table6
implied.
The subjunctive b-forms in the West Saxon Gospels which do not translate
Latin future tense forms nevertheless frequently have a future implication. They
render, for instance, Latin subjunctives as we noted in (7). They also refer to the
formally ambiguous L fuerit (s/he will/may have been). The verse in (21) provides
an example:
(21) Mt 18.3
L amen dico vobis nisi conversi [past part.] fueritis [2pl futperf.
ind.act./perf.subj.act.] et efficiamini sicut parvuli non intrabitis in
regnum caelorum
WSG solice ic secge eow buton ge beon gecyrrede & gewordene swa swa
lytlingas ne ga ge on heofena rice
Truly, I say to you, unless you turn and become like children, you
will not enter the kingdom of heaven.
The subjunctive b-construction is likely to have been chosen to convey the future
tense and the possible subjunctive mood of the Latin form. Interestingly, the
West Saxon Gospels show indicative and subjunctive b-forms translating L fuerit
in different types of verses: the sentence in (8) had a future-iterative quality and
showed a construction with a b-indicative. In sentences such as in (21), which
do not have a future quality but can rather be described as open conditions due
to OE buton unless (cf. Mitchell 1985 II: 826), there are usually subjunctive
b-constructions. This indicates that although the Latin form was important for
the translators choice of forms of OE bon, the context was significant, too.
There are also b-subjunctives translating Latin passive infinitives in the West
Saxon Gospels that may be interpreted as future implications; cf. (22):
(22) Jn 3.14
L et sicut Moses exaltavit serpentem in deserto ita exaltari [pres.pass.
inf.] oportet Filium hominis
WSG & swa swa Moyses a nddran up ahof on am westene swa gebyra
t mannes sunu beo up ahafen
And as Moses lifted up the serpent in the desert, so must/will the
Son of Man be lifted up.
Overall, 28 (75.67%) of the total share of b-subjunctives in the West Saxon Gos-
pels can be given a future reading. This confirms a preference for subjunctive
b-forms to indicate futurity and to express an action that is not bound to the pres-
ent moment. The examples demonstrated that b-subjunctives are predominantly
used in propositions that imply futurity and require the subjunctive mood of the
verb. By contrast, s-forms are usually employed when a future interpretation of the
sentence is impossible. The verse in (23) demonstrates this:
(23) Mt 18.9
L bonum tibi est un oculum in vitam intrare quam duos oculos
habentem mitti [pres.pass.inf.] in gehennam ignis
WSG betere e ys mid anum eage on life to ganne onne u si mid twam
asend on hellefyr
It is better for you to enter life with one eye than having two
eyes/with two (and) be thrown into the fire of hell.
3.3 Differences between the Gospel versions

In addition to the absence of a subjunctive b-paradigm in Northumbrian and
Aldreds employment of indicative b-forms and doublets of indicative b- and
subjunctive s-forms to substitute for it in the Lindisfarne Gospels, there are more
differences in the use of the subjunctive forms between the two Gospel versions.
For instance, the West Saxon Gospels prefer s-forms in verses dealing with the
ancestry of Jesus as in (24):
(24) Mt 27.40
L si Filius Dei es [2sg pres.ind.act.] descende de cruce
WSG gyf u sy Godes sunu ga nyer of re rode
LiGl gif sunu godes ar ofstig astig of rode
If you are the Son of God, come down from the cross.
The West Saxon Gospels reveal the subjunctive s-form sy in reference to the Latin
2sg pres.ind.act. es (you are), whilst there is the indicative non-b-form ar in the
Lindisfarne Gospels. Verses of the type as in (24) imply doubts concerning Jesus
descent. Old English uses subjunctives in conditional clauses expressing doubts,
uncertainties and conditions contrary to fact (cf. Mitchell 1967:146). The West
Saxon Gospels express this doubt by using an s-subjunctive despite the corre-
sponding Latin present indicative form. By contrast, the Lindisfarne Gospels con-
vey the indicative mood of the Latin verb by rendering it with a non-b-indicative;
the possible interpretation of the verse as a doubt or uncertainty did not prompt a
subjunctive here. This shows a closer translation of the Latin in Lindisfarne com-
pared to the West Saxon Gospels.
Christine Bolze
The West Saxon Gospels furthermore insert s-subjunctives to translate the

interjection osanna (hosanna) and the phrase pax vobis (peace to you); cf. (25)
and (26):
(25) Mt 21.9
L osanna Filio David
WSG hal sy u Dauides sunu
LiGl la hl usic sunu dauies
Hosanna to the Son of David/You be safe, Son of David.
(26) Jn 20.26
L venit Iesus ianuis clausis et stetit in medio et dixit pax vobis
WSG se Hlend com belocenum duron & stod tomiddes him & cw,
syeow sib
LiGl cuom se hlend bityndum durum & stod to middes & cu
sibbiuh
Jesus came, the doors being shut, and stood among them and said:
Peace to you/Peace be with you.
Expressions such as hal sy (you be safe) in (25) and sy eow sib (peace be with
you) in (26) appear like fixed formula in the West Saxon Gospels; i.e. they generally
occur with s-subjunctives. By contrast, Aldred does not insert a form of OE bon
at all, but the gloss in the Lindisfarne Gospels reflects the Latin syntax. This again
demonstrates that the gloss in Lindisfarne is more faithful to the Latin syntax than
the West Saxon prose translation of the Gospels, and it shows that the translation
type to some extent influences the distribution of present tense forms of OE bon.
4. Conclusion
Overall, the data allows us to confirm the claimed future indication of the
b-paradigm of OE bon in both Gospel versions. It became obvious that the
choice of the forms mainly depended on Latin; it is therefore essential to consider
the Latin when discussing the semantics of the forms of the twofold present tense
paradigm.
In the analysis of the present indicative, we identified a predominant use of
b-forms implying futurity, whereas non-b-forms were preferred for references
without a future implication. The significantly higher share of indicative b-forms
in the Lindisfarne Gospels had various reasons. It was due to their occurrence in
multiple glosses in the Lindisfarne Gospels and the use of alternative forms to
express futurity in the West Saxon Gospels, such as full verbs inflected for the pres-
ent tense, OE geweoran (to happen) and subjunctive b-forms. We also observed
b-forms in certain aspectual references; nevertheless, their Latin counterparts fre-

quently were in the future tense and most likely triggered them.
In the analysis of the present subjunctive, we detected that subjunctive b-forms
in the West Saxon Gospels mostly occurred in future references. The two b-forms
bia and bie in the Lindisfarne Gospels also implied futurity. We suggested that they
might not denote subjunctive but indicative b-forms. We furthermore noted that
the higher share of s-forms in the Lindisfarne Gospels was primarily due to the
absence of a subjunctive b-paradigm in Northumbrian. There were frequently
double glosses of subjunctive s-forms and indicative b-forms to substitute for it.
Hence, the translation type and the deviating structure of the verb in Northum-
brian compared to West Saxon contributed to the different distribution of the
forms in the two Gospel versions. Further research, in particular on Aldreds use
of OE bon in multiple glosses, is of interest to gain a more profound understand-
ing of the semantic distinction of the twofold present tense paradigm of to be.
References
Brunner, Karl. 1942. Altenglische Grammatik. Halle: Niemeyer.

Brunner, Karl. 1962. Die englische Sprache: Ihre geschichtliche Entwicklung, Vol 2, 2nd edn. Halle:
Niemeyer.
Campbell, Alistair. 1959. Old English Grammar. Oxford: Clarendon.
Dictionary of Old English, Antonette di Paolo Healey (ed.). http://www.doe.utoronto.ca/
Exter, Otto. 1911. Beon und wesan in Alfreds bersetzung des Boethius, der Metra und der Solilo-
quien. Kiel: Fienke.
Hogg, Richard M. & Fulk, Robert D. 2011. A Grammar of Old English, Vol. II: Morphology.
Oxford: Wiley.
Jost, Karl. 1909. Beon und wesan, eine syntaktische Untersuchung [Anglistische Forschungen
26]. Heidelberg: Winter.
Jlicher, Adolf, Matzkow, Walter & Aland, Kurt. 1970. Itala: das Neue Testament in altlateinscher
berlieferung, 4 vols, 2nd edn. Berlin: Walter de Gruyter.
Keller, Wolfgang. 1925. Keltisches im englischen Verbum. Anglica: Untersuchungen zur eng-
lischen Philologie, I: Sprache und Kulturgeschichte: 5566.
Kilpi, Matti. 1997. On the forms and functions of the verb be from Old to Modern English. In
English in Transition. CorpusBased Studies in Linguistic Variation and Genre Styles [Topics
in English Linguistics 23], Matto Rissanen, Merja Kyt & Kirsi Heikkonen (eds), 87120.
Kilpi, Matti. 1993. Syntactic and semantic properties of the present indicative forms of the verb
to be in Old English. In Early English in the Computer Age. Explorations through the Helsinki
Corpus [Topics in English Linguistics 11], Matti Rissanen, Merja Kyt & Minna Palander
Collin (eds), 97116. Berlin: Mouton de Gruyter.
Kilpi, Matti. 1992. BEON. With attested spellings by Robert Millar, using materials assembled
by Haruko Momma. Dictionary of Old English. Toronto: Pontifical Institute of Mediaeval
Studies.
Christine Bolze
Kilpi, Matti. 1989. Passive Constructions in Old English Translations from Latin: with special
reference to the OE Bede and the Pastoral Care [Mmoires de la Societe Neo-Philologique a
Helsingfors 49]. Helsinki: Socit Nophilologique.
Liuzza, Roy. M. 2000. The Old English Version of the Gospels, Vol 2 [Early English Text Society
314]. Oxford: OUP.
Lutz, Angelika. 2009. Celtic influence on Old English and West Germanic. English Language and
Linguistics 13(2): 227249.
Mitchell, Bruce. 1967. An Old English syntactical reverie. Neuphilologische Mitteilungen 68:
139149.
Mitchell, Bruce. 1985. Old English Syntax, 2 Vols. Oxford: Clarendon.
Ross, Alan S.C. 1932. The errors in the Old English gloss to the Lindisfarne Gospels. Review of
English Studies 8: 385394.
Ross, Alan S.C. 1981. The use of other Latin manuscripts by the glossators of the Lindisfarne and
Rushworth Gospels. Notes and Queries 226: 611.
Schumacher, Stefan. 2007. Die Deutschen und die Nachbarstmme: Lexikalische und struk-
turelle Sprachkontaktphnomene entlang der keltisch-germanischen bergangszone. In
Johann Kaspar Zeu im kultur- und sprachwissenschaftlichen Kontext (19. bis 21. Jahrhun-
dert), Kronach 21.7.-23.7.2006 [Keltische Forschungen 2], Hans Hablitzel & David Stifter
unter redaktioneller Mitarbeit von Hannes Tauber (eds), 167207. Wien: Praesens.
Weber, Robert & Gryson, Roger (eds). 2007. Biblia Sacra Vulgata, 5th edn. Stuttgart: Deutsche
Bibelgesellschaft.
Wischer, Ilse. 2010. On the use of beon and wesan in Old English. In English Historical Linguis-
tics 2008. Selected papers from the fifteenth International Conference on English Historical
Linguistics, (ICEHL 15), Munich, 2430 August 2008, Vol. I: The History of English Ver-
bal and Nominal Constructions [Current Issues in Linguistic Theory 314], Ursula Lenker,
Judith Huber & Robert Mailhammer (eds), 217235. Amsterdam: John Benjamins.
Aspectual properties of the verbal prefix a- in
Old English with reference to Gothic
Vlatko Broz
University of Zagreb, Croatia
This paper is a corpus-based study of the Old English verbal prefix a- which is
no longer productive in English today, but survives in a few lexical relics such
as arise, awake or ashamed. After a brief discussion of previous research and the
etymology of this prefix, the paper investigates a range of meanings and functions
that the verbal prefix a- had in early English, showing that it was in an advanced
stage of grammaticalisation and that its primary function was to express
perfective aspect. The prefix is contrasted with its cognate in Gothic, as well as
itsequivalents in Modern English and Croatian, a Slavic language that marks
aspect morphologically.
1. Introduction
In the literature the meaning of the Old English prefix a- has always been con-
sidered as vague and elusive. Therefore it comes as a surprise that it has been so
under-researched, compared to the prefix ge-, which has generated significant
interest amongst scholars and students.
The verbal prefix a- is no longer productive in Modern English but survives
in a number of lexicalised forms such as the verbs arise, awake or abide. For
present-day speakers of English the meaning of this fossilised morpheme is not
transparent.
Quirk and Wrenn (1957:109) say that a- is
used to modify verbs; in many cases it changes the aspect from durative to
perfective, in many it is a mere intensifier and in many others it appears to have
no semantic function.
Hogg also finds the prefix a- problematic:

It is not always possible to give a clear indication of meaning to some of the
prefixes. Thus a-, a verbal prefix found in verbs such as acalan become frozen, is
clearly an intensifier of calan become cold, but afysan and fysan can both mean
drive away. (Hogg 2002:106)
Vlatko Broz
The view that in many cases it does not seem to add to or alter the meaning of the
verb is also present in the works of Hiltunen (1983), Brinton (1988) and E lenbaas
(2007), as well as Kastovsky who also notes that it is difficult to give precise
semantic patterns to this prefix, arguing that it is questionable whether it was
still productive in Old English in view of its many shades of meaning reflect-
ing its different origins, namely as a reduced form of of-, on- or un- (Kastovsky
1992:378).
Mitchell admits that prefixes such as a-, be-, for-, ge-, of- and to- do have
aspectual properties, but he insists that this is not the sole function of any of these
prefixes (1985:367). Such a claim is of little help, as marking aspect is never the
only function of any prefix in any Indo-European language.
2. Hypothesis and theoretical framework
The paper starts from the hypothesis that English has several well-developed
systems of aspect, one of which is expressed by verbal prefixes and post-verbal
particles (Brinton 1988). The aspect system of verbal prefixes was abandoned
during the Middle English period, parallel to which a new aspect system was
emerging that of post-verbal particles (Hiltunen 1983). It is also hypothesised
in this paper that by and large Old English verbal prefixes expressed aspect in a
similar way to that in Slavic languages (Broz 2011) even though many scholars
tried to prove otherwise (Limar 1963; Mitchell 1985:364; Szemernyi 1987).
The approach taken in this paper is essentially eclectic. In order to account for
the phenomenon of aspect in the diachrony of English, a wide range of theories
have been combined. There is no single theory, school or movement in the field of
aspect. Studies on aspect have their traditions, such as Slavic or Anglo-American,
but even these traditions feature a number of widely differing angles, interpreta-
tions or methods. All these theories could be called theories of aspect, but it is
important to stress that they are not characterised by adherence to any central
positions or principles. I have combined them with some more recent theories
such as grammaticalisation theory (Hopper & Traugott 1993) and lexicalisation
theory (Brinton & Traugott 2005). These two could be subsumed under a more
general theoretical framework, which is that of Cognitive Linguistics. However,
Cognitive Linguistics is not a theory but an approach to language that places
meaning in focus, blurs the boundaries between grammar and lexicon, as well as
the boundaries between pragmatics and semantics. It studies cognitive mecha-
nisms and principles of human categorisation that in turn account for a wide range
of linguistic phenomena.
Aspectual properties of the verbal prefix a- in Old English with reference to Gothic
3. Data
For the purpose of this research the data are drawn from The York-Toronto-Helsinki
Parsed Corpus of Old English Prose (YCOE) part of LEON-alfa (Petr 2010:6878).
The concordancing programme used to search the corpora was Abundantia Verbo-
rum (AV). This corpus is tagged for parts of speech, which enables a more efficient
query. However, this still does not imply that the prefixed verbs in this research
were easy to obtain. Sifting or manually discarding all the undesired combina-
tions is a painstaking and laborious process, particularly in the case of the prefix
a-, as this single initial letter can also be an element of other non-prefixed verbs
beginning with a- (e.g. acsian to ask) or other prefixes such as and- or an- (e.g.
andswarian to answer).
Furthermore, there is frequent homonymy, so verbs like adrygan to dry and
adreogan to do, perform share a number of identical forms, as well as verbs like
agytan to understand and ageotan to pour, which count 6 and 19 tokens respec-
tively. Such situations are numerous and present serious problems in counting
tokens and types.
However, this corpus was not sufficient in order to perform a comprehensive
investigation of the phenomenon of aspectuality as expressed by verbal prefixes.
All the instances of prefixed verbs had to be checked against their simplexes, for
which all other available sources were used. They include the Toronto Dictionary
of Old English (DOE), which was unfortunately still at the time of writing this
paper only available up to the letter G, and the online edition of Bosworth-Toller
Anglo-Saxon Dictionary which was digitised at the Charles University in Prague.
4. Previous research
Two avenues of research have converged in this paper, one is that on aspect, and
the other is on verbal prefixes and particles. The only previous work that unites
these two areas in diachrony is Brintons The Development of English Aspectual
Systems. Aspectualizers and Post-verbal Particles, offering an original discussion of
the systems that mark aspect in English, both from a synchronic and a diachronic
point of view, but unfortunately the study is not corpus-based. Brinton sees verbal
prefixes and particles as belonging to the lexical or secondary system of aspect
marking in English, rather than to the grammatical or primary system. This point
of view has been challenged in Broz (2011), as both verbal prefixes and particles
can also show grammaticalised properties, i.e. express grammatical aspect rather
than lexical aktionsart (cf. Cappelle 2005:423).
Vlatko Broz
Wischer and Habermann (2004) explore the use of prefixes to express aspect
and aktionsart, but only in one Old English text (Orosius). They focus on the prefix
ge- as the main perfectiviser, giving only one example of another prefix that can
also act as a perfectiviser the prefix for-.
Aspect is one of the most challenging areas in linguistics. Aspectual systems
differ dramatically across languages and consequently this field of research is the
most prolific and diverse area in linguistics. One can often come across the state-
ment that English is a tense rather than an aspect language. English does not
mark aspect morphologically like the Slavic languages do but by means of syntax.
Scholars agree about only one aspectual opposition in English today progressive
vs. non-progressive in the tense system. All the other instances of aspectual mark-
ing in English, both synchronic and diachronic, have been dogged by controversy.
Aspect is a semantic category which Comrie (1976:3) defines as different ways of
viewing the internal temporal constituency of a situation. This definition is widely
accepted and is the one referred to in this paper.
Aspect studies originated in grammatical analyses of the Slavic languages,
where all verbs are said to have two complete sets of forms, perfective and imper-
fective. The distinction between these two is an obligatorily marked category of a
verb. However, the semantics and morphology of aspect in Slavic languages are far
from regular. Studying these irregularities can help to account for aspect in Old
English.
Many grammars try to explain the morphology of aspect in Slavic by stating
the general rule according to which the presence of a prefix automatically makes a
verb perfective (e.g. Sili 1978:49). This is in keeping with another rule according
to which all simplex or prefixless verbs are imperfective by default. Previous stud-
ies of aspect in early Germanic languages applied the basic rule of perfectivisation
imported from Slavic, realizing that it works in only one third of the cases, accord-
ing to one count.1 The earliest proposal of the existence of the category of aspect in
Germanic languages was Streitbergs (1891) theory that has remained a controver-
sial subject until this day, generating two kinds of work on aspect in Gothic: those
who refuted it (e.g. Beer 1915; Mirowicz 1935; Goedsche 1940; Wolfgang Krause
. The Czech scholar Beer (1915) disputed Streitbergs analysis of the Gothic Bible by
counting all the instances where the prefix ga- does not perfectivize. Beer claimed that 66%
are exceptions and that Wulfila only added the prefix ga- to imitate Greek compound verbs.
The Serbian linguist Pudi (1956:391) rejects Beers claim that Wulfila used the prefix ga- at
whim in his lengthy analysis, showing that the prefix ga- did act as a perfectivizer.
1953; Pilch 1953; Scherer 1954; Lindemann 1970; Szemernyi 1987)2 and those
who accepted it (e.g. Rice 1932; Hirt 1934; Pudi 1956; Lloyd 1979; West 1982;
Maxi Krause 1987; Coleman 1996; Leiss 2000).
Many linguists were led to believe that aspect was not a category of the verb in
Germanic languages. A closer look at aspect in Slavic, however, reveals that there
are many verbs that do not enter into aspectual pairs, that many verbs particularly
loanwords are biaspectual, that many prefixless verbs are in fact perfective and that
many prefixed verbs can also be imperfective. Furthermore, imperfective forms
can express perfective meaning, as the following two Croatian3 sentences show:
(1) Jesi itao Zloin i kaznu?
be.aux read.past.ptcp.impf.msg crime.acc and punishment.acc
Have you read Crime and Punishment?
(2) Juer sam gledao Avatara.
yesterday be.aux watch.past.ptcp.impf.msg Avatar.acc
Yesterday I saw Avatar.
Previous studies of Old English verbal prefixes concentrated on the prefix ge- as
the only perfectiviser, whereas this research has shown that other prefixes also
have a perfectivising function, just like in Slavic, as these Old English and Croatian
aspectual pairs illustrate:
Old English Croatian

(3) etan geetan jesti pojesti
eat-impf eat-perf eat-impf eat-perf
. Some of the opponents of Streitbergs theory expressed their views vehemently and
engaged in fierce debates with the proponents in the form of articles such as Szemernyi 1987
with a reply in Lloyd 1990. Szemernyi dismisses Streitbergs view as erroneous and unten-
able (1987:4), concluding his very brief discussion with the following words:
Aspect was certainly not a morphological category in Gothic and it is beyond all
doubt that Streitbergs thesis is simply not tenable. This negative verdict applies to
the other Germanic dialects also.
Szemernyi (1987:4fn) also refers to Mirowicz (1935:48), who quoted Trnkas statement
(1929:48) that the theory of aspect in Germanic is die grte wissenschaftliche Fiktion (the
greatest science fiction).
. Croatian is taken as an example of a Slavic language not only because it happens to be the
authors mother tongue, which provides direct native speaker insights necessary for a contras-
tive analysis with Old English, but also because it is transitional with regard to aspect, sharing
features of both eastern and western Slavic aspectual zones (Dickey 2000:1).
Vlatko Broz
(4) writan awritan pisati napisati

write-impf write-perf write-impf write-perf
(5) swelgan forswelgan gutati progutati

swallow-impf swallow-perf swallow-impf swallow-perf
For every verb there is only one prefix that changes only its aspectual meaning
without affecting the lexical meaning of the verb. There is no rule or predictable
pattern for the combination of prefixes with verbs in order to only change the
aspectual meaning. One verb can come in combination with several prefixes, but
only one prefix changes its aspect exclusively. Other prefixes change the lexical
meaning of the verb in addition to assigning perfective aspect.
One of the most frequent perfectivizing prefixes in Slavic, po-, has a number
of different meanings as the following Croatian examples show:
(6) piti popiti grammaticalised (changing aspect)

drink-impf drink-perf
(7) skoiti poskoiti jump lexicalized (changing aktionsart)

jump-perf jump at once-perf
(8) rei say porei deny lexicalized (changing the lexical

say-perf deny-perf meaning)
In popiti, the prefix is grammaticalised, i.e. its function is to perform the grammat-
ical operation known as perfectivisation. In other words, the imperfective aspect
is changed into perfective. In poskoiti, the prefix is lexicalised as the simplex verb
skoiti already is perfective (its imperfective pair is skakati), so it is changing the
aktionsart. In porei, the prefix radically changes the lexical meaning of the verb.
Old English prefixes show similar characteristics, where their meanings and func-
tions range from fully grammaticalised to fully lexicalised, as will be shown for the
prefix a- later in this paper.
However, while Croatian operates with 18 such prefixes, Old English had only
seven prefixes that could perform an aspectual function among other things, as
first identified by De la Cruz (1975). These prefixes were a-, be-, ge-, for-, of-, on-
and to-. The prefixes be- and to- have recently been studied by Petr (2005), who
identified their four different functions, only one of which is aspectual. Therefore,
Broz (2011) looked up the remaining five different prefixes in the two corpora and
analyzed in detail the three most frequent ones: The verbal prefix ge- which is four
times more frequent than a-, which is in turn twice as frequent as the prefix for-.
Table 1 shows the number of observations and the number of tokens after the files
were cleaned of all spurious hits:
Table 1. Corpus results for early English prefixes:

initial hits vs. Tokens
Prefix Period Initial hits Tokens
1 ge- OE 19,652 16,878

ME 5,434 4,332
2 a- OE 5,175 4,091
ME 4,198 1,201
3 for- OE 2,812 2,071
ME 1,712 1,575
4 on- OE 2,560 2,212
ME 225 94
5 of- OE 823 407
ME 341 114
5. Etymology
In order to examine the semantic categories expressed by the Old English prefix a-,
we should try to trace back its etymology to its original spatial senses. All verbal
prefixes are assumed to have developed from prepositions or adverbials with a
spatial (locational/directional) and/or temporal meaning.
In the first systematic analysis of the functions of verbal prefixes, De la Cruz
(1975:47) labels the prefixes a-, ge-, on- and to- as pure prefixes without an etymo-
logical prepositional counterpart. The Old English prefix a- lost its prepositional
counterpart before the written documents of Old English, but it can be traced back
in Gothic. The Gothic cognate of a- is the preverb us, which can be attested by
comparing the translations of Gospels in the examples such as
(9) Gothic us-wairpan OE aweorpan throw away
(10) Gothic us-dreiban OE adrifan drive out.
It was phonologically unstable even in Gothic, as can be seen in its allomorph ur

in the examples such as
(11) Gothic ur-rinnan OE arinnan, ayrnan go out and

(12) Gothic ur-reisan OE arisan go up, arise.
In Gothic, the allomorph ur was used if the verbal stem began with an r. This pho-
nological change is known as rhotacism. Once the s changes into r, it is a small step
to lose the r in a subsequent stage of Germanic. Phonological reduction occurs
commonly in grammaticalisation processes.
Vlatko Broz
The Gothic prepositional counterpart of this prefix is us, which meant out
of or away from. Lehmann (1986:380) says its Proto-Germanic reconstruction
is *uz, but he goes on to list several possible PIE origins, such as Uhlenbeck (1906
TNTL 25:302) who notes that the etymology is uncertain, possible from PIE
*aw- *w- away from, down (Skt avs, va downward, Lat au away in au-fer,
au-fugi), but Brugmann (1904:463), 468, Schmidt (1889: 219) and Jacobsohn
(1920 ZVS 49:195) consider that the possible origin is PIE *ud-s.
Blaek (2001:2425) elaborates this possible origin, offering attestations from
a handful of languages, but also brings it into connection with Gothic ut out and
uta outside and OE, OFris, OSax t (cf. Lehmann 1986:384). He notes that the
reflex of PIE *uds is Germanic *uz, which he thinks are both the origins of Gothic
us and uz-, ON r as well as OE or- as a privative prefix that denotes origin or
antiquity as in oreald (Modern German uralt). In the Slavic branch he identifies it
as OCS vz/vs for; up, whose Modern Russian reflex is vy- out, a frequent per-
fectivizing prefix, or Croatian uz- as in uzeti to take, rather than OCS iz from, out
of reconstructed as ProtoSlavic *jz, going back to PIE *eghs, the reflexes of which
are Latin ex and Greek x (Blaek 2001:18), but with no reflexes in Germanic.
It is really interesting that Germanic should not have a cognate with Latin ex
or Croatian iz and that Gothic us, ut, ur and uz- should all be reflexes of a single
Proto-Germanic or PIE preposition. Investigating more on this issue would take
us well beyond the scope of this paper, but it is worthwhile to point out that the
proposed etymologies are somewhat dubious and should be further investigated.
The Modern German cognate is the inseparable prefix er-. However, an impor-
tant difference between the German er- and the Old English a- is that the German
prefix eris still productive today and has preserved some of the idea of forth,
from within, or up as in the German verbs ersteigen climb, rise up (OE astigan)
or erhitzen to heat (OE ahatian).
To summarise this short discussion, the transition from *uz to a- shows a
drastic reduction in the phonological structure, but its original sense was estab-
lished as a source prefix (=out, out of ) whose inherently source oriented meaning
had already been bleached by the time of the first Old English written data.
6. Aspect in early Germanic
Let us consider a few examples from Gothic and compare them with the situation
in Old English.4
. These examples have been selected from Bosworths edition of Gospels in Gothic and Old
English (1865).
(13) Gothic
jah anaks insaiwhandans
and suddenly look round.pres.ptcp.nom.pl
ni anaseis ainohun
no any more one.indf.pron
gasewhun, alya Iesu ainana mi sis.
see.3.p.pl.pret except Jesus.acc alone.acc with refl.pron.dat
Old English
And sona a hi besawon, hi nanne hi mid
and soon when they around-looked they none one with
him ne gesawon, buton one Hlend sylfne mid him.
them not ge-saw.3.pl but the saviour self with them
And suddenly, when they had looked round, they saw no man any more
save Jesus only with themselves. St. Mark 9.8
In the above two examples we have the Gothic verb saihwan and the Old English
verb seon prefixed with ga- and ge- respectively, both used in a situation that was
completed in its entirety in the past. There is no reason to doubt that ga- and ge- in
this case signal perfectivity.
The next example features several prefixed verbs which could all function as
markers of aspectuality.
(14) Gothic
ushofun an ana stain arei was.
lift.3.p.pl.pret then that.acc stone.acc where be.3.p.sg.pret
I Iesus uzuhhof augona iup jah qa:
and Jesus lift.3.p.sg.pret eyes.acc up and say.3.p.sg.pret
atta, awiliudo us, unte andhausides mis
father, thank.1.p.sg.pres you.dat that.conj hear.2.p.sg.pret I.dat
Old English
a dydon hig aweg one stan. Se Hlend ahof his
then did they away the stone the saviour a-lifted his
Eagan up, and cw, Fder, ic do ancas e,
eyes up and said father I do thanks you
foram u gehyrdest me.
because you ge-heard me
Then they took away the stone from the place (where the dead was laid).
And Jesus lifted up his eyes, and said, Father, I thank you that you have
heard me. John 11.41
Vlatko Broz
We see that Wulfila used the same verb for taking the stone away (lifting it) and
for lifting up the eyes, whereas in Old English a phrasal verb was used in the first
case (don aweg, literally do away) and the cognate verb ahebban in the second
case. In the second case both Gothic and Old English verb are re-inforced by
the particle up which indicates direction, whereas uzuh- in Gothic and a- in
Old English could both be markers of perfectivity. The verb andhausjan is a pre-
fixed verb just like the Old English verb gehiran, both meaning to hear. Their
simplex versions hausjan and hiran mean to listen. If we attach a prefix to this
simplex verb, we could argue that we are changing its aspect, more precisely its
resultativity.
The following example also shows a verb which is perfectivised with the prefix,
cognates both in the prefix and the stem.
(15) Gothic
Bieh an uswoh fotuns ize, jah
so then us-wash.3.p.sg.pret feet.acc their.gen and
nam wastjos seinos; anakumbjands
take3.p.sg.pret garments.acc their set down.pres.ptcp.n.sg
aftra, qa du im
again say.3.p.sg.pret to they.dat
Old English
Syan he hfde hyra fet awogene, he nam his reaf; and
after he had their feet a-washed he took his robe and
a he st, he cw eft to him
then he sat he said again to them
So after he had washed their feet and had taken his garments and had
seated himself again he said to them. John 13.12
It is very likely that the Gothic verb us-wahan wash was the perfective pair of
wahan and its Old English corresponding verb awean (in the past participle
form) is its cognate both in the prefix and the stem. The fact that the prefix a- in
the verb awean to wash was understood as an inflectional (grammatical) rather
than a derivational (lexical) prefix can be confirmed by an extract from Aelfrics
Grammar where Latin verb paradigms are presented along with a translation into
Old English:
(16) lauo ic wea, laui ic woh, lautum awogen, sume cwea
lavo I wash lavi I washed lavatum a-washed some say
lotum oe lauatum
lotum or lavatum Gram 139.3
In other words, the Latin verb in the present tense lavo I wash corresponds to Old
English wea, the Latin perfect tense lavi I washed corresponds to the Old E nglish
preterite woh and the supine forms lautum, lotum or lavatum (from which the
past participle is formed) correspond to awogen, in which case the prefix a- was
added, rather than the prefix ge-.
It may at first seem absurd to claim that the prefix a- or us- was experienced
as an inflectional prefix rather than a derivational one, but other linguists have also
noted inflectional properties of other verbal prefixes. Specifically, Hogg (2002:105)
says that the prefix ge- can come close to being an inflectional marker rather than
a prefix, which certainly does make sense if we remind ourselves that grammar
and lexicon form a continuum of linguistic knowledge (Langacker 1987:3). Cases
like verbal prefixes a- or ge- show that we cannot speak of their strictly grammati-
cal or strictly lexical function. This is in line with one of the hypotheses proposed
in this paper that prefixes were grammaticalised and that even though they
occurred before the stem or root, they exhibited inflectional properties rather than
derivational. We have a similar situation in Modern German or Dutch where the
past participle marker ge- is fully grammaticalised, and therefore is classified as
an inflectional affix rather than a derivational one, even though it appears at the
beginning of the word.
The following example contains verbs that are interesting from the perspec-
tive of aspect.
(17) Gothic
Yah anaaiauk sandyan ridyan i eis yah ana
and continue.3.sg.pret send.inf third and they also he.acc
gawondondans, uswaurpun.
wound.ptcp.pl. us-cast.3.pl.pret
Old English
a sende he riddan, a wurpon hig ut one gewundodne.
then sent he third then cast him out the wounded
And again he sent a third: and they wounded him also, and cast him out.
Luke 20.12
The translations into Gothic and Old English are somewhat different the Gothic
text contains the verb anaaukan continue, whereas in Old English the construc-
tion a a (when then) is used. The verb uswairpan is a cognate of Old
English aweorpan.
The Old English verb sendan is used without the prefix a-, which would make
it imperfective. However, it is possible that the context gives it a perfective reading,
Vlatko Broz
as will be discussed in the next section. As aspect is also a matter of syntax, the
structure a a sets a frame whose verbs can only have a perfective meaning.
Another interesting feature of this example is that weorpan is also not used with
the prefix a- (aweorpan normally means to cast out) but is used with the particle
ut out instead. The prefixes ga- and geare both added in the past participle form,
re-inforcing its resultative meaning.
A few lines later in the text, we have the verb uswairpan cast out again, but
this time the Gothic verb is in the past participle form, whereas the Old English
one is in the preterit, and vice versa for usquiman and ofslean to kill:
(18) Gothic
Jah uswairpandans ina ut us amma weinagarda
and cast out.pres.ptcp.pl. he.acc out from the.dat vineyard
usqemun.
kill.3.p.pl.pret
Old English
And hig hine of am win-gearde awurpon, ofslegene.
and they him of that vine-yeard a-cast kill.ptcp
So they cast him out of the vineyard, and killed him. Luke 20.15
The context demands a perfective reading for both verbs and therefore prefixes are
attached in the above example.
The five examples [(13)(15) and (17)(18)] of contrastive analysis between
Gothic and Old English point to many similarities in the way prefixed verbs were
used and what function these prefixes performed. In this regard, Old English has
more in common with its early relative Gothic than with its descendant Middle
English, not to mention the English we speak today. The Gothic us-prefixed verb
typically has an a-prefixed counterpart in English, both occurring in perfective
contexts and exhibiting many features typical of grammaticalisation (e.g. phono-
logical reduction us- a-, lexical items assuming a grammatical function, bleach-
ing, frequency, changes occurring in small structural steps) if we imagine Gothic
as an earlier stage of Old English.
7. Semantics of the prefix a-
The original spatial sense of the prefix a- should best be seen in cases where it is
attached to verbs of motion. The following three examples feature such verbs.
(19) a halwendan men cwdon, & a geleafsuman, a e to

the holy men said and the believing that which to
urum Drihtne coman a he to heofonum astigan wolde
our Lord came that he to heaven ascend wanted
The holy and believing who came to our Lord as he was about to ascend
toheaven c971.coblick, HomS_46_[BlHom_11]:117.14.1481
(20) a asprang hire hlisa and wisdom and gelrednys, geond
then sprang her fame and wisdom and learning through
ealle a ceastre, foram heo ws on eawum gefrtwod
all the town because she was in virtues adorned
Then sprang her fame and wisdom and learning throughout all the town,
because she was adorned with virtues c1010.coeuphr, LS_7_[Euphr]:29.29
(21) a aras he from m slpe
then arose he from the sleep
Then he arose from sleep c897.cobede, Bede_4:25.344.15.3456
If we take a closer look at the contexts in which the verbs astigan ascend, asprin-
gan jump and arisan arise are used in the above three examples, we can see that
they are used metaphorically. However, they must have had concrete meanings as
well, just as the Modern English verb spring still has today. Due to the metapho-
ricity of these verbs, we cannot claim that we have clear cases of literal spatial-
directional meanings of the prefix a- in the examples (19)(21). As a matter of fact,
the literal spatial-directional sense of the prefix can rarely be encountered in the
corpus. The most frequent function of this prefix found in the corpus is actually
that of an aspect marker, as the following pair of examples shows, featuring the
most frequent a-prefixed verb awritan.
(22) Esdras se writere awrat ane boc, hu t folc com ongean
Ezra the scribe a-wrote a book, how that people came back
fram Chaldea lande to Iudea lande
from Chaldea land.dat to Judea land.dat
Ezra the scribe wrote a book, how that people returned from Chaldea
to Judea Let 4: 726.292
(23) ara abbuda str & spel isses
the.gen.pl abbot.gen.pl histories and stories this.gen.sg
mynstres [] on twam bocum ic awrat.
monastery.gen.sg in two.dat.pl books.dat.pl I a-wrote.
The history and account of the abbots of this monastery [], I wrote down
in two books. c897. cobede, Bede_5:22.484.15.4856
Vlatko Broz
For a number of examples I have contrasted the prefixed verb with its simplex
counterpart in context to see whether the prefixless variant of the verb would
express imperfective aspect.
(24) Se Hlend abeah nyer & wrat mid his finger on
the healer bent down and wrote with his finger.dat on
re eoran. a hig urh-wunedon hine axsiende
the earth.dat when they continued him questioning
a aras he upp & cw to him;
then arose he up and said to them
Loca hwylc eower si synleas wurpe rest stan on hi.
look if you be sinless cast first stone at her
& he abeah eft & wrat on re eoran;
and he bent again and wrote on the earth.dat
a hig is gehyrdon a eodon hig ut an fter anum.
when they this heard then went they out one after one.dat.pl
& he gebad ar sylf & at wif stod r
and he remained there self and the woman stood there
on middan;
in middle
The healer stooped down and wrote with his finger on the earth. When
they kept on questioning him, he straightened up and said to them, if any
one of you is without sin, let him be the firs to throw the stone at her. Again
he stooped down and wrote on the ground. When they heard this, they
began to go away one after another, until only he was left and the woman
was still standing there. Jn. (WSCp) 8.69
The verb writan is an atelic verb. However, its inherent semantic aspect is affected
if it is used in a sentence with a direct object. In example (24) the verb writan
occurs twice in its prefixless variant. There is no direct object that is normally
required in the predication of this verb acting as an endpoint. The absence of the
prefix a- could signify that the action is not finished or complete, and can have a
progressive reading. The Modern English translation does not use the continuous
tense here, as the Past Simple tense can also give a progressive reading, but it is
clear from the context that the action of writing was not complete.
Another instance where the simplex verb is used giving an imperfective read-
ing on the sentence level is evident in example number (25):
(25) Se halga godspellere swa be him wrat & cw
the holy evangelist thus about them wrote and spoke
The holy evangelist thus wrote and spoke about them
c971. coblick, LS_12_[NatJnBapt[B1Hom_14]]: 161.26.2055
Even though the action of writing was most likely in reality completed in the past,
this is no sufficient reason to use the prefix. The context enables either form, the
perfective or the imperfective, but the latter was chosen most likely for stylistic
reasons. This sentence was basically a summary of what the holy evangelist did,
so more context in case of this example would not help us to reduce the aspectual
ambiguity of the verb in question.
The following example also shows the imperfective form of the preterite in a
context where the perfective meaning does not fit.
(26) Witodlice gif ge gelyfdon on moyse. ge gelyfdon eac me;
indeed if you believed in Moses you believe also me
Solice he wrat be me. gif ge his stafum ne
truly he wrote of me if you his letters.dat not
gelyfa. hu gelyfe ge minum wordum;
believe how believe you my.dat.pl words.dat
Indeed if you believed Moses, you would believe me. Truly, he wrote of me.
But if you do not believe his writings, how will you believe my words?
c990 John V.47
If we contrast the examples (22) to (26) with their Croatian translations, we will
see that writan consistently corresponds to pisati and awritan to napisati. Even the
problematic example number (25) is likely to be translated with pisao rather than
napisao.
A further set of examples feature the verb fyllan to fill, make full, supply in its
prefixed and prefixless form:
(27) Eac is land ws swie afylled mid munecan
also this land was much a-filled with monks
Also this land was abundantly supplied with monks A.S. 1086
(28) a ya weollon & ymbsweopon & ghwonan t
the waves seethed and round-swept and on all sides that

scip, fyldon t heo him nnigra gesynta wendon

ship filled that they them none safety.gen.pl turned
The waves seethed and swept round them, and filled the ship on every side,
so that they utterly despaired of safety.c897. Bede 3.13.200.14
It is also possible to translate fyldon with the progressive were filling the ship, but
it is not necessary, as even the simple form in Modern English in the right context
can convey the imperfective meaning.
As late as the Peterborough Chronicle (mid-12th century), prefixed verbs
were still used to express perfective meaning and prefixless increasingly for
both perfective and imperfective, as the prefixes were dying out. The following
Vlatko Broz
example contains the verb fyllan, whose meaning can be both perfective and
imperfective:
(29) hi suencten suye e uurecce men of e land mid
they oppressed much the wretched men of the land with
castel weorces; a e castles uuaren maked, a
castle works when the castles were made then
fylden hi mid deoules & yuele men.

filled they with devils and evil men

They greatly oppressed the poor men of the land with castle building work;
when the castles were made, then they filled the land with devils and
evil men. ChronE 1137.14
The verb fedan to feed behaves as an imperfective verb when without a prefix, as
we see in example (30):
(30) Se goda mete ger de, ge one lichoman fede ge

the good food either does both the body feeds and
t mod glada to lcere hlo,
tha mindt gladdens for each health
Good food does two things: it feeds the body and it gladdens the mind, for
the health of each. HomU 11 (Verc 7) 95
If a prefix is attached to this verb, it is perfectivised, as example (31) demonstrates.

In fact, we have a situation with two different prefixes attached to the same verb
stem, both acting as perfectivisers.
(31) r [on am toweardan life] we beo gefedde. and we r
there in the future.dat life.dat we are ge-fed and we there
nnne ne afeda
no one not a-feed
(BG fed).
There [in the future life] we shall be fed, and we shall feed no one.
c995. CHom II, 34 258.88
Very often in the literature, Old English prefixes such as a-, for- or be- are said to
add telicity to verbs (e.g. Brinton 1988:202204; Deh 2002:6; van Kemenade &
Los 2003:79; Petr & Cuyckens 2008:144), rather than perfectivity. If they add
telicity, it means we regard aspectual meaning arising from the unit of the verb
and the prefix as lexical, rather than grammatical. However, if we regard pre-
fixes as constructions which exhibit a predictable meaning, then we could argue
that at the sentence level they express perfective aspect, rather than telic aktion-
sart. Furthermore, if we claim that prefixes have grammaticalised, then we speak
of grammatical aspect rather than lexical aspect, so the term perfective is more
appropriate in most of these cases.
Telicity, as originally coined by Garey (1957) in a description of aspect in
French, refers to the inherent meaning of the verb, i.e. telic verbs have an endpoint.
For example, the verb arrive is telic, whereas the verb play is atelic. The inherent
meaning of the verb can be modified by grammatical means, so if we use a telic
verb such as arrive in a continuous tense, we can change the inherent temporal
property of the verb on the sentence level:
(32) Peter was arriving.
As we can see, the situation no longer has either culmination or an endpoint. The
verbs inherently telic value has been dissociated by means of syntax. It is argued
here that verbal prefixes are much like function words (grammatical words), a
closed set of words with little lexical meaning which serve to express grammatical
relationship with other words in a sentence. However, verbal prefixes such as a-,
ge- and for- are not words but morphemes, so they affect the meaning of the verb
on the morphological rather than on the syntactic level. In this regard they exhibit
more inflectional rather than derivational properties, as they express grammatical
and not lexical meaning.
In order to account for the cases where the simplex verb expresses perfective
meaning, I have applied Jakobsons theory of semantic markedness.5 According
to this theory the perfective aspect is semantically marked, while the imperfec-
tive is semantically unmarked.6 This explains why Slavic imperfective forms can
also express perfective actions, as in the previously mentioned example (1) Jesi
itao Zloin i kaznu? Similar situations exist in Spanish and Italian for progressive
tenses, as well as in Early Modern English before the progressive aspect became
fully grammaticalised and no longer just optional. A well-known example from
Shakespeares Hamlet demonstrates that in Early Modern English the present
. Markedness theory refers to a relation of asymmetry between elements of linguistic or

conceptual structure. Originally, the structuralist Trubetzkoy applied it to phonology, but
soon it was extended to morphological, syntactic, semantic and lexical oppositions. An un-
marked form is the default form. In lexical opposites, for example, the unmarked tiger can
refer to a male or female, while tigress is marked since it can refer only to females.
. Jakobson (1957:[1971:137]) describes the perfective aspect as semantically marked
(concerned with the absolute completion), while the imperfective aspect is semantically
unmarked (noncommittal with respect to completion or noncompletion). In other words,
imperfective verbs, being semantically unmarked, can be employed in certain situations to
express perfective actions.
Vlatko Broz
s imple form is semantically unmarked and can take a progressive reading (just as
in Modern Spanish or Italian):
Polonius: What do you read, my lord?
Hamlet: Words, words, words. Hamlet, II.2.191192
In Present-Day English, Poloniuss sentence would not be considered grammati-

cally correct in terms of usage, and instead the progressive would be used, as in
What are you reading (cf. Rissanen 1999:216).
Aspectual studies are also famous for a proliferation of categories. Accord-
ing to the classification adopted in this paper, which most closely follows that
of Comrie (1976), the main aspectual opposition is perfective vs. imperfective.
This opposition is further subclassified into different types, so the categories
such as resultative, completive, instantaneous are regarded as types of perfec-
tive aspect. Whether a prefixed verb expresses resultative or completive action
depends on the inherent semantics of the verb and not on the prefix. So for
example, the verb awritan could be classified as resultative, whereas asingan as
completive:
(33) sian he on sie ws, he asong lce dge tuwa his
when he on journey was he a-sang each day twice his
saltere ond his mssan
psalms and his mass
When he was on a trip, every day he sang twice his psalms and his mass.
c1000. Mart 5 Se 25, A.8
The essential property of the completive aspectual category is the emphasis on the
totality of the situation, rather than on its final stage which is the distinguishing
property of the resultative category. But again, the prefix functions as an aspectual
marker, that of a completeness that the psalms were sung in their entirety, from the
beginning to the end. Examples (34) and (35) also show that.
(34) Leo ws asungen, gleomannes gyd.
song was a-sung minstrel.gen lay
The song was sung, the minstrels lay.c1015. Beo 1159
(35) a se Wisdom a is leo swie lustbrlice &

when the wisdom then this song very pleasantly and
gesceadwislice asungen hfde, a hfde ic a giet
wisely a-sung had then had I then yet
hwylchwugu gemynd on minum mode re unrotnesse
some memory on my mood the sadness
e ic r hfde
that I before had
When wisdom had sung the song so pleasantly and wisely, I still
remembered some of the sadness that I had felt previously.
c950. Bo 36.103.23
The prefix a- marks the completeness of the action: the song was sung through to
its end. The next two examples show the verb singan without the prefix, i.e. in the
imperfective meaning.
(36) a mid am e he his gebedu sang. a tr t hors
then with that he his prayers sang then tore the horse
t c of re cytan hrofe. and r feoll adune
the thatch from the cottage roof and there fell down
swilce of am hrofe wearm hlaf mid his syflinge.
such from the roof warm loaf with his food
Then while he was singing his prayers, the horse tore the thatch from the
roof of the cottage, and there fell down, as from the roof, a warm loaf with
some food. c995. CHom II, 10 82.54
Example (36) features the verb singan in the preterite tense without the prefix because
it signifies an action that is ongoing or in progress. The act of singing prayers was
not completed, but was interrupted by another action, that of the horse tearing the
thatch from the roof of the cottage. The verb tr tore is prefixless but its meaning is
perfective. Its perfectivity is signalled by the adverbial conjunction a then.
Example (37) shows that the simplex verb singan in the present tense cannot have
the prefix a- because it would interfere with the progressive aspect of the situation:
(37) Hwt is is folc e us hlude singe?
what is this people that thus loudly sing.3.sg.pres
Who are these people who are singing so loudly?
c971. Blickl. Homl. 149, 30
In a similar vein, example (38) features the verb singan to sing in the past parti-
ciple form without any prefix, as part of a passive construction. The meaning of the
verb could be interpreted as imperfective:
(38) On hwylcum tidum alleluia sceole beon sungen
in which times hallelujah shall be sung
In which hours will Hallelujah be sung? c1025. Cobenrul, BenR:16.6.21.85
The reason why an imperfective is used here is because the information whether
the Hallelujah will be sung in its entirety is not in focus or is irrelevant. Moreover,
Vlatko Broz
there is more stress on the adverbial on hwylcum tidum in which hours, which
suggests that the action will take place on several occasions or could function as a
time adverbial that marks perfectivity. Given that it is a monastic rule, it is crucial
that the Hallelujah is sung in its entirety on these occasions.
A further aspectual meaning encountered in the prefix a- is variously called
instantaneous, momentaneous or punctual, and is commonly regarded as a sub-
type of perfective aspect. For example, when prefixed with a-, the verb feallan
denotes an instantaneous action, a short-lasting punctual single event.
(39) a his here geseah t he mid y horse afeoll

when his army saw that he with the horse a-fell
when his army saw that he fell down together with his horse
c925. Or 3 7.64.30:
When the subject is in plural, such as the noun tearas in number (40) the action
is no longer instantaneous, as the tears did not fall simultaneously at once but one
after another and can be viewed as a series of events of falling. For this reason, the
Modern English translation of the simplex verb could also have the verb in the
progressive aspect:
(40) & he
[Joseph] wear swa swie astyred, t him feollon
and he [Jospeh] was so very a-stirred that him fell.3.pl
tearas of his eagan for broor ingon
tears of his own for brother things
And he w+as so strongly stirred that his tears fell for his brothers sake
c1000. cogenesiC, Gen_[Ker]:43.30.333
Alternatively, the simplex verb, being semantically unmarked, can also have a
perfective reading. In Croatian, both perfective and imperfective are possible (suze
su mu pale [perfective] and suze su mu padale [imperfective] his tears fell).
Number (41) is another example of the prefix a- signifying an instantaneous
action:
(41) ahleopon a ealle & hiene mid heora metseacsum ofsticedon

a-jumped then all and him with their daggers stabbed
inne on heora gemotrne.
inside in their meeting-place
then [the consuls and the senate] all jumped up, and stabbed him [Julius
Csar] with their daggers in their senate-house c925. Or 5 12.129.2
Even though the subject is in plural form, the prefix a- modified the meaning of
the simplex verb adding the notion of an instantaneous action all the consuls
jumped up at the same time and this happened very quickly. In a translation of this
sentence into Croatian, the perfective verb skoiti jump can be further modified
with the prefix po-, in which case its meaning denotes a very short action. In this
case the prefix po- is not a perfectiviser since the verb is already perfective, but as a
derivational prefix modifying what is traditionally called the aktionsart. A similar
function of a- can be observed in the following example:
(42) a ahleop se lichoma sona up of am wtere, ond t

then a-leapt the body soon up from that water and the
heafod on ore stowe.
head on other place
Then the body immediately leapt up from that water and the head on that
place. c1000. Mart_5_[Kotzor]:Oc31, A.13.208
Example (43) features the same verb without the prefix:
(43) a sealde se cyning him sweord, t he hine mid gyrde; &

then gave the king him sword that he him with gird and
nom his spere on hond & hleop on s cyninges
took his spear in hand and jumped on the kings
stedan & to m deofulgeldum ferde.
stallion and to the devilish idols.dat fared
Then the king gave him a sword to gird on and took his spear in his hand
and jumped on the kings stallion and set out to the devilish idols.
c897. cobede, Bede_2:10.138.4.132730
The adverbial conjunction a sets the frame of the narrative in example (43),
making it optional for verbs to take prefixes. The verbs in this sentence signify a
sequence of actions in perfective reading.
The simplex verb hleapan apart from jumping also means dancing, which
shows that the prefixless version of the verb has a durative quality, as dancing can
be perceived as a series of events of jumping:
(44) onne beo geopenode blindra manna eagan, and deaffra

then are opened blind mens eyes and deaf
manna earan
mens ears
gehyra; onne hleap se healta swa swa heort,
hear.3.pl.pres then leap.3pl.pres the halt as hart
Then shall be opened the eyes of blind men, and the ears of deaf men shall
hear; then shall the lame leap as a hart c995. Homl. Th. ii. 16, 18.
Vlatko Broz
(45) ltan scralletan sceacol se e hleape ngl neome

let shrill schackle which leaps nail harmonioussounds
cende bi him neod micel
generate be him eagerness great
let the shackle which leaps/dances be shrill, the nail produce harmonious
sounds, great is his eagerness c970. coexeter R 8385
Finally, the last proposed aspectual property of the prefix a- is ingressive or incho-
ative aspect, which focuses on the beginning of a situation. There is a group of
Old English a- prefixed verbs which are derived from adjectives and can only be
translated analytically into Modern English with the verbs such as become or grow.
These verbs express ingressive aspect as their meaning refers to the moment of
entering into a state of whichever adjective from they were derived.
acealdian to grow cold
(46) ahatian to become hot
adeorcian to become dark
Some of them can also be rendered into Modern English by adding the verbalizing
suffix -en, which is considered to be an ingressive marker in Present-Day English:
(47) ahyrdan harden afyrhtan frighten
Number (48) is a contextualised example of one such verb:

(48) seo sunne asweartade, ond se dg ws on eostre niht
the sun blackened and the day was in dark night
gecierred fram midne dg o non
turned from middle day until noon
the sun turned black and the day turned into the dark night from midday
to the ninth hourc1000. Mart 5 Ma 25, A.13
The simplex verbs cealdian, hatian, deorcian and fyrhtan exist with the same mean-
ing as well (hyrdan is attested only with a causative meaning according to BT), but
they seem to be more frequent with the prefix.
Although ingressive verbs refer to the moment of the beginning of the action,
they are considered to be a subcategory of perfective, just like resultative. In order
to determine the aspectual meaning of this group of verbs, I tried translating
them into Croatian, as this language exhibits typological similarity to Old English.
Aspectual nuances are very difficult to explain in ones mother tongue, let alone in
a foreign or even a dead language. When I tried to pinpoint what kind of aspectual-
ity was sensed in the Croatian translation equivalents to the examples in (46)(48),
Irealised that they could at the same time refer both to the beginning of the action,
as in entering a state of being whatever the adjective means, and to the result of the
action, in which case they would belong to the category of resultativity.
8. Co-existence of prefixes and particles
One of the phenomena observed in the corpus is the co-existence of prefixes and
particles, as in example (49):
(49) Gif mon folcleasunge gewyrce, & hio on hine geresp weore,
if ma public slander makes and it on him proved becomes
mid nanum leohtran inge gebete onne him mon aceorfe a
with no lighter things amends than him man a-cuts the
tungon of
tongue off
If anyone utters a public slander, and it is proved against him, he shall make
amends on no easier terms than that his tongue be cut off
LawAf_1:32.103
In 17% of the cases, the prefix a- co-occurs with the particle exhibiting a similar
meaning. This percentage is too small to make any conclusive claims that there
was a change underway, but it can serve as a piece of evidence regarding the ongo-
ing change. In combination with other evidence, it does contribute to the overall
picture of prefixes losing their semantic and phonological structure and therefore
had to be reinforced by an adverbial particle, which in the end prevailed, building
a system of phrasal verbs which can be seen in example (50):
(50) He cearf of heora handa and heora nosa.

he cut off their hands and their noses
He cut off their hands and noses O.E. Chron. an. 1014
One and the same sentence can feature both the prefixed and the prefixless version
of the verb, as we see in example (51):
(51) Gif u scyle aceorfan oe asnian unhal lim of halum

if you should a-cut or a-sever unhealthy limb from healthy
lice onne ceorf u t on am gemre s halan lices
body then cut you that on the side the.gen healthy body.
If you should cut off or remove an unhealthy limb from a healthy body, cut
it on the side of the healthy body. c1025. Lch_II_[1]:35.4.1.1060
In the above set of examples we see the verb ceorfan7 cut in all four possible com-
binations: prefixed in the first part of example (51), prefixless in the second part
. The verb ceorfan is the ancestor of Modern English carve.

Vlatko Broz
of example (51), with a particle only (50) and prefixed with particle (49). Inciden-
tally, in all the three examples the objects are body parts, but this only makes it
easier to compare them, make generalisations and conclusions.
In all three examples the meaning of the verb is perfective (a body part is
successfully amputated in all cases), except for the second instance of (51) where
ceorf is very likely imperfective, as the focus is on the action and not on its entirety.
In the first of these three examples (49), the prefix or the particle one of these
two seems to be redundant. Or perhaps the action is only intensified by the par-
ticle, as the prefix on its own was too weak. This intensification can also be inter-
preted as a higher degree of resultativity.
9. Contrastive analysis
In order to gain a better understanding of the function of the Old English prefix
a-, one of the methods was contrastive analysis. I translated a randomised sam-
ple of 200 tokens into Modern English and Croatian. Both comparisons yielded
interesting results.
As the aspect of Croatian verbs is morphologically marked, translations into
Croatian can immediately confirm if the aspectual meaning in Old English is per-
fective or not. A sample of 14 verbs has been extracted from the token translations,
listed in Table 2 in the infinitive form:
Table 2. Old English Croatian verb correspondences

acennan zaeti conceive agyfan odustati give up
aclnsian oistiti clean up asecgan izgovoriti say, speak out
acwican oivjeti revive asendan poslati send (out)
adelfan iskopati dig (up) asettan postaviti set, put
afedan nahraniti feed asiftan prosijati sift
aflowan istei flow away asmorian uguiti suffocate
afindan saznati find out aswellan natei swell up
As many as 98% of Old English tokens are translated with a perfective verb in
Croatian. Moreover, all of them are prefixed, as can be seen in the table.
Translating the tokens into Modern English has showed that in as many as
74% of cases, the meaning of the prefix a- can be conveyed with a post-verbal
particle.
10. Conclusion
The aim of this paper has been to examine the meanings of the Old English prefix
a-, the main function of which is claimed to be expressing perfective aspect. It
also gives an account how aspect is expressed in early English by means of verbal
prefixes, and thus contributes to an understanding of aspect in diachrony. Verbal
prefixes and particles are a prominent example that blurs the boundaries between
grammar and lexicon. As a result they also blur the boundaries between gram-
maticalisation and lexicalisation, as well as the boundaries between inflection and
derivation. They show that there is a gradual rather than a discrete distinction
between them.
The verbal prefix a-, as any other verbal prefix, is traditionally thought to be
a derivational prefix. However, ample evidence in this paper shows many of its
inflectional properties. Its decline and loss coincides with the decline and loss of
all other inflectional suffixes that were used to mark case, gender, verb person
and other grammatical functions, as English was transforming from a more syn-
thetic type of language to a more analytic type. The verbal prefix a- is one of the
seven prefixes that played an important role in the synthetic system of aspectual
marking, which was gradually replaced by two analytical systems, one of post-
verbal particles and the other of progressive and perfect tenses.
The prefix a- shows some typical features of grammaticalisation. The inves-
tigation of its etymology showed that once it was a content item which changed
into a grammatical word and then reduced to an inflectional affix. It follows the
grammaticalisation path (cf. Booij & van Kemenade (2003:4):
independent preverb > left member of verbal compound > prefix > (zero)
It starts as an independent preverb, as attested in Sanskrit (Broz 2011:6) and

then we have it attested in Gothic as a left member of verbal compounds. In Old
English, we find it in the prefix stage, whereas by the end of the Middle English
period it reaches the final stage of grammaticalisation where all semantic and pho-
nological content is lost and the morpheme is reduced to zero or nothing. The fact
that in many cases its meaning is not readily identifiable points to the fact that its
semantic properties were undergoing the process of bleaching.
The prefix a- had a stronger capacity to express aspectuality than the prefix
ge-. However, frequency charts may be distorting the overall picture. The prefix
ge- is four times more frequent than the prefix a-, but in more than half the cases
its meaning is not aspectual.
Languages change because their speakers re-interpret data in inter-generational
transmission of language. The reason why aspectual prefixes died out could be the
Vlatko Broz
fact speakers interpreted them as redundant since simplex versions of prefixed verbs
had the capacity to convey both imperfective and perfective meanings.
References
Beer, Antonn. 1915. Ti studie o videch slovesnho dje v gottin. ast prvni: djiny otzky. Prag.
Blaek, Vclav. 2001. Indo-European prepositions and related words. SPFFBU A 49: 1532.
Booij, Geert E. & van Kemenade, Ans 2003. Preverbs: An introduction. In Yearbook of Morphol-
ogy, Geert E. Booij & Jaap van Marle (eds), 3360. Dordrecht: Kluwer.
Bosworth, Joseph. 1865. The Gothic and Anglo-Saxon Gospels in Parallel Columns with the Ver-
sions of Wycliffe and Tyndale. London: John Russell Smith.
Brinton, Laurel J. 1988. The Development of English Aspectual Systems. Aspectualizers and Post-
verbal Particles [Cambridge Studies in Linguistics, 49). Cambridge: CUP.
Brinton, Laurel J. & Traugott, Elizabeth Closs. 2005. Lexicalization and Language Change.
Cambridge: CUP.
Broz, Vlatko. 2011. Diachronic Analysis of Aspectual Preverbs and Post-verbal Particles in
English. Ph.D. dissertation, Leuven & Zagreb.
Brugmann, Karl. 1904. Kurze vergleichende Grammatik der indogermanischen Sprachen.
Strassburg: Trbner.
BT: Joseph Bosworth & T. Northcote Toller. 1898 (main volume) & 1921 (supplement). An
Anglo-Saxon dictionary, Based on the Manuscript Collections of the late Joseph Bosworth.
Oxford: OUP. Also in the digital edition: http://bosworth.ff.cuni.cz/
Cappelle, Bert. 2005. Particle Patterns in English. A Comprehensive Coverage. Ph.D. disserta-
tion, KU Leuven.
Coleman, Robert. 1996. Exponents of futurity in Gothic. Transactions of the Philological Society
94: 130.
Comrie, Bernard. 1976. Aspect. Cambridge: CUP.
Deh, Nicole. 2002. Particle Verbs in English. Syntax, Informational Structure and Intonation
[Linguistik Aktuell/Linguistics Today 59]. Amsterdam: John Benjamins.
de la Cruz, Juan. 1975. Old English pure prefixes: Structure and function. Linguistics 145: 4781.
Dickey, Stephen. 2000. Parameters of Slavic Aspect: A Cognitive Approach. Stanford CA: CSLI.
DOE: Dictionary of Old English: A to G on CD-ROM. 2008. Angus Cameron, Ashley Crandell
Amos, Antonette di Paolo Healey et al. (eds). Toronto: Pontifical Institute of Mediaeval
Studies.
Elenbaas, Marion. 2007. The Synchronic and Diachronic Syntax of the English Verb-Particle
Combination [LOT Dissertations 149]. Ph.D. dissertation, Radboud University Nijmegen.
Garey, Howard B. 1957. Verbal Aspect in French. Language 33(2): 91110.
Goedsche, C. Rudolf. 1940. Aspect versus Aktionsart. The Journal of English and Germanic
Philology 39(2): 189196.
Hiltunen, Risto. 1983. The Decline of the Prefixes and the Beginning of the English Phrasal Verb
[Annales Universitas Turkuensis, Series B., Vol.160]. Turku: University of Turku.
Hirt, Hermann. 1934. Handbuch des Urgermanischen, Teil III: Abriss der Syntax. Heidelberg:
Carl Winters Universittsbuchhandlung.
Hogg, Richard. 2002. An Introduction to Old English. Edinburgh: EUP.
Hopper, Paul J. & Traugott, Elizabeth Closs. 2003 [1993]. Grammaticalization. Cambridge: CUP.
Jacobsohn, Hermann. 1920. Zwei Probleme der gotischen Lautgeschichte, II: Zum gotischen
Satzsandhi. KZ 49: 129-218.
Jakobson, Roman. 1957 [1971]. Shifters, verbal categories, and the Russian verb. Selected Writ-
ings, Vol. 2, 130147.The Hague: Mouton.
Kastovsky, Dieter. 1992. Semantics and vocabulary. Cambridge History of the English Language,
Vol. 1. Cambridge: CUP.
van Kemenade, Ans & Los, Bettelou. 2003. Particles and prefixes in Dutch and English. In
Yearbook of Morphology 2003, Geert E. Booij & Jaap van Marle (eds), 79117. Dordrecht:
Kluwer.
Krause, Wolfgang. 1953. Handbuch des Gotischen. Mnchen: Beck.
Krause, Maxi. 1987. Smantique et syntaxe des prverbes en Gotique, 4 Vols. Thse de doctorat
nouveau rgime, Universit Paris Sorbonne.
Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1: Theoretical Prerequisites.
Stanford CA: Stanford University Press.
Lehmann, Winfred P. 1986. A Gothic Etymological Dictionary. Leiden: E.J. Brill.
Leiss, Elisabeth. 2000. Artikel und Aspekt. Die grammatischen Muster von Definitheit. Berlin:
Walter de Gruyter.
Limar, L.S. 1963. K voprosu o roli glagolnix pristavok v svjazi s vidovym znaeniem glagolov (na
materiale drevneanglijskogo). Uenye Zapiski, xxviii, . 2-ja, Moscow.
Lindemann, John William Richard. 1970. Old English Preverbal ge-: Its Meaning. Charlottesville
VA: The University Press of Virginia.
Lloyd, Albert L. 1979. Anatomy of the Verb: The Gothic Verb as a Model for a Unified Theory
of Aspect, Actional Types, and Verbal Velocity [Studies in Language Companion Series 4].
Amsterdam: John Benjamins.
Lloyd, Albert L. 1990. A reply to Oswald Szemernyi: The origin of aspect in the Indo-European
languages. Glotta 68: 12931.
Mirowicz, A. 1935. Die Aspektfrage im Gotischen. Wilno: Nakladem Towarzystrwa Prsyjaci-L
Nauk.
Mitchell, Bruce. 1985. Old English Syntax, 2 Vols. Oxford: Clarendon Press.
Petr, Peter. 2005. On the Variables Determining the Life Span of English Prefix Constructions:
A Case Study of the Two Prefixes be- and to- (NHG zer-). MA thesis, University of Leuven.
Petr, Peter & Cuyckens, Hubert. 2008. Bedusted, yet not beheaded: The role of be-s construc-
tional properties in its conservation. In Constructions and Language Change, Alexander
Bergs & Gabriele Diewald (eds), 133170. Berlin: Mouton de Gruyter.
Petr, Peter. 2010. On the interaction between constructional & lexical change: Copular Passive
and related Constructions in Old and Middle English. Ph.D. dissertation, University of
Leuven.
Pilch, Herbert. 1953. Das AE. Prverb ge-. Anglia LXXI: 129139.
Pudi, Ivan. 1956. Prefiks ga- u gotskom jeziku. Prilog uenju o glagolskom vidu. Sarajevo: Nauno
delo.
Quirk, Randolph & Wrenn, Charles Leslie. 1957. An Old English Grammar. 2nd edn. London:
Methuen.
Rice, Allan Lake. 1932. Gothic Prepositional Compounds in their Relation to their Greek Origi-
nals. Ph.D. dissertation, University of Amsterdam.
Rissanen, Matti. 1999. Syntax. In The Cambridge History of the English Language, Vol. III,
Richard Hogg (ed.) Cambridge: CUP.
Vlatko Broz
Scherer, Philip. 1954. Aspect in Gothic. Language 30(2): 211223.

Schmidt, Johannes. 1889. Die Pluralbindungen der indogermanischen Neutra. Weimar: Hermann
Bhlau.
Sili, Josip. 1978. An Approach to the Study of Aspectuality in the Croatian Literary Language.
In Kontrastivna analiza engleskog i hrvatskog ili srpskog jezika [Contrastive Analysis of
English and Serbo-Croatian]. Volume 2. Rudolf Filipovi (ed.), 4270. Zagreb: Institute of
Linguistics, Faculty of Philosophy, University of Zagreb.
Streitberg, Wilhelm. 1891. Perfective und imperfective actionsart im Germanischen. Beitrge
zur Geschichte der deutschen Sprache und Literatur 15: 70177.
Szemernyi, Oswald. 1987. The origin of aspect in the Indo-European languages. Glotta 65:
118.
Trnka, Bohumil. 1929. Some remarks on the perfective and imperfective aspects in Gothic.
Donum Natalicium Schrijnen. Verzameling van Opstellen door Oud-leerlingen en Bevriende
Vakgenoten Opgedragen aan Mgr. Prof. Jos. Schrijnen bij Gelegenheid van zijn Sestigsten
verjaardag, 3 mei 1929. Nijmegen-Utrecht: Dekker en Van De Vegt.
Uhlenbeck, Christianus C. 1906. Aanteekeningen bij Gotischen etymologien. TNTL 25:
245306.
West, Jonathan, 1982. The semantics of preverbs in Gothic. Indogermanische Forschungen 87:
138165.
Wischer, Ilse & Habermann, Mechthild. 2004. Der Gebrauch von Prfixverben zum Ausdruck
von Aspekt/Aktionsart im Altenglischen und Althochdeutschen. Zeitschrift fr Germanis-
tische Linguistik 32: 262285.
r ws vs. thr was
Old English and Old High German existential
constructions with adverbs of place
University of Zurich, Switzerland
Even though we can observe striking differences in the isolating contexts of

Modern English and Modern High German existential constructions, both
languages feature existential constructions with locative adverbs that are the
result of long processes of grammaticalisation. In Old English (OE), expletive r
diverged from the locative r as a result of semantic and syntactic reanalysis,
which led to the development of the English existential there-construction.
Ina similar way, Old High German (OHG) thr, through grammaticalisation,
diverged from its locative origin and gave rise to the existential da-construction.
It is suggested in this paper that there was a common origin: both r and
thr evolved as a compromise in the conflict between pragmatic and syntactic
structure in OE and OHG, respectively. The aim is to contribute to the still
small number of qualitative and quantitative studies of OE and OHG existential
constructions.
1. Introduction
(1) There is silence.

Es tritt Stille ein. (MUR 59)
Es entsteht Stille. (MUR 61)
Es folgt Stille. (WHI 53)
Es herrscht Schweigen. (WHI 43)
It has been well-documented in the literature that English contents itself with a sin-
gle existential construction that is able to express various propositions and that has
a wide range of potential applications. The English existential there-construction
(hereafter ETC) was first defined by Jespersen (1924:155) as the construction in
which there appears as an unstressed, non-deictic and non-referring element func-
tioning as the syntactic subject that introduces a postponed, foregrounded NP into
the discourse.1 The locative there, on the other hand, is a deictic component, with
reference to the speech situation (Johansson 1997:304). By contrast, from Modern
High German (ModHG) we are familiar with a situation in which the spatial and
temporal existence of an entity is specified, i.e. the entitys exact position and/or
physical state is concretely determined and described. ModHG features a bewilder-
ing variety of existential constructions that include a large number of verbs that
may be used to express existence besides their original lexical meaning meanings
ranging from very specific and concrete (as exemplified in (1)) to rather imprecise
and undetermined.
The main question that will be addressed in this paper is whether High
German, being closely related to English, has ever shown a tendency towards
developing an existential construction that is similar in form and function to
the ETC. The development of the impersonal expletive there in the English ETC
from the more contentful and less functional locative adverb there has frequently
been described as a case in point of grammaticalisation, as it underwent all the
steps of change that are characteristic of grammaticalisation, from extension of
meaning to phonetic reduction. In search for a common root of English and
High German existential constructions with erstwhile locative adverbs, I will
thus use the emergence and subsequent beginning grammaticalisation of the
Old English (OE) ETC as a base from which I will proceed to examine Old High
German (OHG) existential constructions with a locative adverb. Particular
emphasis will be put on the semantic changes of r and thr, for the follow-
ing reasons: firstly, it will be shown in this paper that desemanticisation is the
dominant process at stake in OE and OHG existential constructions with a loca-
tive adverb. Furthermore, a discussion of the semantic changes is important for
our understanding of the grammaticalisation of existential constructions with
locative adverbs to try to assess the extent to which the literal meanings of the
adverbs (and the whole constructions) have weakened. Finally, the High German
existential da-construction did not develop beyond this stage of the grammati-
calisation process, while the E nglish existential there-construction was further
grammaticalised. This is not to say that desemanticisation represents the sole
or main indicator of grammaticalisation (cf.Section2), but simply that it is the
dominant process at stake in OE and OHG existential constructions. Thefocus
. Note that the scope of this study embraces only ETCs containing the verb be; it is com-
monly accepted (Givn 1993; Huddleston & Pullum 2002; Long 1961; Quirk et al. 1985; inter
alia) that most clauses with there as their subject have be as their verb, which is why the
presentational there-constructions with verbs other than be will be ignored in this paper. The
same accounts for the distinction between High German existential constructions.
r ws vs. thr was
of this paper thus lies on grammaticalisation from a synchronic point of view,

but a few points will be made from a diachronic perspective, i.e. from what
we know of the further development of English and High German existential
constructions. Due to the relatively few attestations of OE and OHG existential
constructions with adverbs, my study will be largely qualitative in nature. Note,
however, that although the grammaticalisation path of English and G erman
existential constructions have been tackled before, sometimes even from a con-
trastive perspective (see Pfenninger 2009), the present study claims novelty due
to its quantitative approach: the linguistic features involved in the earliest stages
of the grammaticalisation process are not just identified and classified, as has
been done before, but they are also quantified. This can shed light on the origins
and roots of the beginning divergence of the English and German existential
constructions and also adds value to the general discussion of grammaticalisa-
tion processes.
At this point it is important to mention that OHG does not provide the
researcher with an ideal selection of texts from which to choose, as it has not pre-
served such a wide and rich variety of literature as has OE (see Fleischer 2006);
this makes it difficult to construct a corpus that is large and homogeneous enough
to yield data that are actually representative and suitable to describe and analyse
syntactic OHG structure. Consequently, in order to provide a suitable body of
material for my study of the frequency of OHG existential thr-constructions,
in addition to more neutral prose texts, I included translations as well as poetic
texts, knowing that poetic devices such as rhyme, alliteration and meter may all
alter the normal syntactic structures of a language and that OHG syntax can be
heavily affected by Latin in translated texts (cf. Bernhardt & Davis 1997; Masser
1997; Meineke 2001; Sonderegger 2003; inter alia). With this approach Ifollow
Fleischer (2006), who stated that [j]e mehr aus mglichst verschiedenen Denk-
mlern stammenden Belege fr eine bestimmte Struktur vorhanden sind, desto
sicherer wird, dass die entsprechenden Strukturen fr das Althochdeutsche
tatschlich angenommen werden knnen und desto sicherere Aussagen knnen
zur Distribution gemacht werden (44). Despite these methodological caveats,
the only significant (and negative) aspect of translated and poetic texts for this
study is that the subject pronoun is sometimes omitted (e.g. in order to meet the
requirements of meter in poetic texts), which is not the case in native texts in
which the subject pronoun is found, even in the earliest texts. The positive side of
this is that certain translations are so free that there are numerous cases where we
encounter OHG constructions that do not occur in the Latin original. In other
words, when we find an existential construction in the OHG translation but not
in the original text, it must be because of German grammar, which of course
is the ultimate proof that these constructions are not a borrowed phenomenon
(i.e. their existence was not encouraged by a Latin model) but that they actually
occurred in natural speech in OHG.2
Breivik (1990:13), who examines the use and non-use of expletive there both
in contemporary English and earlier English, argues that the amount of poetry
he included in the material would not make a significant difference in the over-
all results; he chose the following OE texts for his study: lfrics Lives of Saints,
The Old English Version of Bedes Ecclesiastical History of the English people, The
Blickling Homilies, Two of the Saxon Chronicles Parallel, The Exeter Book, and
King Alfreds Orosius. In a similar manner, I focused on a selection of monu-
mental OHG data (both native and (preferably freely) translated prose texts as
well as poetic texts) and included the following texts in my analysis: Isidor, Die
lateinische-althochdeutsche Tatianbilingue, Heliand, Otfrids Evangelienbuch, Die
Schriften Notkers und seiner Schule, as well as a collection of minor OHG monu-
ments that is available from the online TITUS corpus.3
The paper is organised as follows. In Section 2, I briefly describe the theo-
retical background on which this analysis is based. Section 3 outlines the isolat-
ing contexts of ModE and ModHG existential constructions. Section 4 provides
an outline of the initial stages of the grammaticalisation path of the OE existen-
tial r-construction, while in Section 5 I report the results of my analysis of
the extent of grammaticalisation of OHG existential thr-constructions. Finally,
Section 6 presents a summary of the main findings.
2. Theoretical framework: Constructions and grammaticalisation
There is a wide range of processes that have been proposed to describe the gram-
maticalisation of forms and constructions (cf. Diewald 2006; Fischer, Norde &
Perridon 2004; Heine 2002; Heine & Kuteva 2005, 2006; Hopper & Traugott 2003;
among many others). The following four mechanisms described by Heine and
Kuteva (2005) have proven to be particularly useful for the present study:
a. Extension, i.e. the rise of new meanings when linguistic expressions are
extended to new contexts
b. Desemanticisation (or semantic bleaching), i.e. loss in meaning content
. Cf. Bernhardt and Davis (1997:19 ff.) study on OHG clauses that differ in word order
from the Latin original, which revealed that, in over half of the instances investigated, the
OHG word order patterns differed from those of the Latin.
. Cf. Section6.1 for more information about the editions and publication dates.
r ws vs. thr was
c. Decategorialisation, i.e. loss in morphosyntactic properties characteristic of

lexical or other less grammaticalised forms
d. Erosion (or phonetic reduction), i.e. loss in phonetic substance.
Ideally, grammaticalisation involves all of the four parameters (ad). As some

examples in this study will show, however, this is not always so; there are cases
in which grammaticalisation is restricted to only one or two of these parameters,
often desemanticisation (or bleaching) as a natural consequence of extension
(Heine & Kuteva 2005:89).
In the early stages of grammaticalisation, we can commonly observe a process
called generalisation, which is characterised by an increase in the polysemy of
a form due to context-induced reinterpretation (Heine, Claudi & Hnnemeyer
1991:65 ff.). Context-induced reinterpretation refers to the process where the old
contexts in which a form or a construction can occur have been generalised to
contexts that were unavailable before, which still offers the potential for ambi-
guity that allows for the structure to continue to be analysed as before, and at
the same time for a new analysis to evolve, and then to coexist with the earlier
analysis (cf.Hopper & Traugott 2003:52). The lexical split that separates the evolv-
ing grammatical item from its lexical mother is commonly referred to as diver-
gence in the literature (see Fischer, Nolde & Perridon 2004:10; Hopper & Traugott
2003:3, 11). The persistence of older forms and meanings alongside newer forms
and meanings, whether derived by divergence from the same source or by renewal
from different sources; this is sometimes called layering (which is often used as a
synonym of polysemy) (see Hopper 1991).
Extension is a prerequisite for desemanticisation to take place, since the exist-
ing structure is more frequently used in the new contexts, which leads to a loss of
lexical meaning: the older, more concrete meaning bleaches to yield the newer,
more abstract meaning (Heine & Kuteva 2006:58 ff.). Note, however, that, strictly
speaking, the notion of loss is misleading here because we are not talking about
a decay of meaning but rather about a semantic shift in meaning (from specific,
lexical to broad, grammatical).
In most cases, the grammaticalising form will belong to another word class
than the lexical item it derives from (Fischer, Norde & Perridon 2004:10). The
concept of decategorialisation refers to the loss of grammatical properties associ-
ated with the source category, i.e. the shift from one category status to another,
correlated with a shift from prototypical membership of a category to less proto-
typical membership of a new category (cf. Brinton & Traugott 2005:25). Again,
we are not dealing with the deterioration of a form but its functional shift from
one kind of role to another in the organisation of discourse (Hopper & Traugott
2003:108).
Finally, phonetic reduction (or erosion) is usually the last process to occur
when a grammaticalisation process takes place; it can involve segmental elements
(e.g. the loss of phonetic segments) or suprasegmental elements (e.g. loss of stress)
(Heine & Kuteva 2006:62).
The role of constructions within grammaticalisation has been widely attested
in the literature (e.g. Diewald 2002, 2006; Bergs & Diewald 2008; Bybee& Torres
Cacoullos 2009). The general tenor is that constructions provide the framework
within which lexical items can be reanalysed in a way which may lead to their
grammaticalisation. When analysing the overall process of this, Diewald (2002)
identifies three stages and their corresponding contexts of change, which precede
grammaticalisation:
i. Untypical contexts
ii. Critical contexts
iii. Isolating contexts
Diewalds stage I corresponds to what is sometimes referred to as bridging con-

text in the literature (cf., for example, Heine 2002). Untypical contexts represent
the preconditions of the grammaticalisation process, as they give rise to pragmatic
inferences in favor of potential new meanings and functions (conversational
implicature). In the so-called critical contexts (sometimes referred to as switch
contexts), we can observe additional structural ambiguity when the use of the
construction can still be understood in the literal fashion but structural differences
provide the potential for new interpretations, thus allowing for grammatical inter-
pretations (cf. also Eckardt 2006:52; Hopper & Traugott 2003:52). At this stage we
can often observe multiple structural and semantic opacity (Diewald 2006), since
several senses may be implied in a given context. This is why semantic change is
known to be difficult to analyse quantitatively, as it is often not possible to stay
objective and cite linguistic examples that are clearly unambiguous, particularly if
you aim at classifying occurrences into a dichotomous set, as will be illustrated in
this paper. Finally, in what Diewald calls isolating contexts, the new item can be
called conventionalised once we witness its unambiguous use with new meaning
and function, being isolated from the older lexical meaning and use. This is when
the process of grammaticalisation can be said to be completed.
3. ModE and ModHG existential constructions in isolating contexts
In the literature on the ModE ETC (e.g. Huddleston & Pullum 2002:1393), there
is commonly a distinction made between extended ETCs, which contain an exten-
sion in the form of a locative prepositional phrase, and non-extended ETCs, which
r ws vs. thr was
only contains there, the verb be, and the subject NP. Based on the assumption that
non-extended ETCs exemplify best that even without complementation, ETCs
can be described as having various implicit meanings, I distinguished between 3
major semantic subgroups of non-extended ETCs in an earlier study (Pfenninger
2009:240 ff.):4
A. Bare ETC
(2) There was indeed a Miss Counihan [somewhere in the world].
Es gab in der Tat eine Miss Counihan. (MUR 3)
Indeed, a Miss Counihan existed.
B. Locative ETC
(3) There was a certain pathos [in his voice].
Ein gewisses Pathos schwang in seiner Stimme. (LAD 46)
A certain pathos was in his voice.
C. Enumerative ETC
(4) Theres Wilder, of course. Theres Denise. Theres Eugene, whos living with his
daddy this year in Western Australia.
Da ist einmal natrlich Wilder. Dann Denise. Dann ist da Eugene, der dieses
Jahr bei seinem Papa in Westaustralien ist. (WHI 39/40/41)
The bare ETC, as in (2), asserts nothing but general, permanent existence; be can
be paraphrased with exist and we could add a locative extension with a very gen-
eral meaning, for instance in the world or in the universe. By contrast, the locative
ETC is implicitly locative because it points to a setting in a narrative sense, i.e. to
something that takes place in our minds and not literally on stage (cf. Bolingers
(1977) textual deixis vs. deixis ad oculos). Accordingly, be can be paraphrased
with occur, appear, take place. This type of ETC exemplifies well the principle of
persistence (Hopper 1991) outlined above; as will be shown in this paper, the
property of persistence of meaning derives in part from the fact that locative there
for a long time coexisted and was polysemous with the newer use as an expletive.
When the expletive finally diverged from its lexical counterpart it did not result in
an across-the-board re-semanticization (Hopper & Traugott 2003:97) of there;
instead, a new meaning was added and thus new distributional possibilities were
opened up for the form. The slight locative flavor can be understood as a continu-
ation of its original lexical meaning.
Finally, the enumerative ETC asserts neither location nor existence, but lists
some entities (and thus is sometimes referred to as non-existential list-construction,
. The examples come from a sample of 1,000 ETCs and their 1,000 ModHG translations,
selected from twentieth-century literature (cf. Pfenninger 2009).
cf. van Gelderen 1991:315). Most scholars (e.g. Hannay 1985; Lumsden 1988) accept
Milsarks (1974:124) claim that, in such sentences, the argument is the whole list, so
that the quantified status of the NP is not relevant to the acceptability of the ETC.
We are not interested in the existence or identity of one particular item of the list,
but rather in the identity of any person who may have satisfied the description [of
a given setting] (Hannay 1985:117). This is also true even if the list consists of
only one item. It is often suggested (cf. Breivik 1989, 1990; Hannay 1985; Lumsden
1988; among many others) that, in enumerative ETCs, the expression there is, or
more commonly theres, functions as a presentative formula or signal that is more
or less synonymous with Lets not forget, I could mention, etc. The reduction of
there is to theres as a result of the routinisation (idiomatisation) of the construction
(cf. Hopper & Traugott 2003:72) represents the last stage of the grammaticalisation
of the ETC a process which is also reflected in the frequent absence of number
agreement between the NP (the new information) and the verb. At this final stage of
grammaticalisation, expletive there and locative there are found in isolating contexts;
that is, they have become context dependent, insofar as there are contexts that favor
one use to the exclusion of the other (see Diewald 2006). We can observe the loss
of distinctive stress: in contrast to its fully stressed counterpart, the expletive there
is unaccented due to the speakers focus on the postponed subject NP, i.e. the new
information. Furthermore, there is is reduced to the routinised formula theres in
informal speech.
As the translations in (24) show (cf. also (1) above), the linguistic situation
is much more complex in ModHG. While the ETC is able to express numerous
propositions, the same semantic categories are encountered in ModHG, yet with
separate (personal or impersonal) constructions for each of them that specify the
various assertions of the different types of ETCs. The construction that is often
believed to be the usual idiom employed to express general existence in ModHG
is the es gibt-construction, whose grammaticalisation path from the personal
geben-construction with give meaning to impersonal construction with exist
sense in early ModHG has been well-documented in the literature (cf. Gaeta 2005;
Newman 1998; Pfenninger 2009). However, verbs with a lexically full meaning
are usually preferred to relatively neutral and general verbs such as geben in es
gibt. As is commonly accepted (cf. Brinkmann 1962; Newman 2002; Serra Borneto
1996; Wandruszka 1969), verbs of spatial existence as well as motion verbs are
used to an almost excessive degree in ModHG existential constructions, compared
to other Germanic languages or the Romance languages. The latter do not require
that degree of specificity when referring to the existence of an object. The lexical
verbs that are chosen exhibit lexical polysemy, that is, they are extended to abstract
senses to help conceptualise the existence of some entity. They are considered
polysemous only if they can be used in existential sentences without at the same
r ws vs. thr was
time specifying the posture/motion/state of the subject referent; in (1a) above, for
example, eintreten enter does not signify the activity of entering but the emerging
existence of the NP Stille silence. Thus, eintreten is, to some extent, bleached of its
original dynamic meaning, and conveys existential meaning. It has been reported
in the literature that in this posture/motionexistential polysemy, the posture/
motion meanings are historically primary, the locative/existential meanings being
later developments.5
The existential da-construction, as in (3) above, is the usual idiom to express
spatial existence in ModHG. In this construction, the adverbial da is not consid-
ered in isolation, but is rather seen as an integral part of the sein-construction, and
the locative function is assigned to the whole construction. Despite its similarity
to the ETC, however, the two constructions have little in common. Even though
da, like the expletive there, has to be considered non-deictic6 and semi-referential
and the whole construction does not refer to a specific location, da implicitly puts
boundaries to the world in which something exists. Hammer (1971:219222, cited
in Newman 1996:162), summarises the following semantic distinctions:
a. The es gibt-construction is used

i. to denote existence as such, without reference to a particular place;
ii. to denote existence in a large area.
b. The existential da-construction is used

i. to denote permanent or temporary presence in a definite and limited place;
ii. to denote temporary presence in a large area.
The es gibt-construction has to do with the existence of an entity, while the exis-
tential da-construction has to do with the presence of some entity (Newman
1996:162163). Pfenninger (2009) thus proposes to consider the existential
da-construction as an equivalent of the locative ETC, in which the locative asser-
tion is incorporated. Furthermore, in contrast to the widely spread ETC, the
da-construction is generally described as being rarely used nowadays (Clark
1978:117). Furthermore, the syntactic function of da is different from there.
Lenerz (1992) describes the syntactic function of the ModHG preposed adverb da
. For a detailed description of the grammatical path of posture verbs, see Lichtenberk
(2002) and Newman (2002). For instance, Lichtenberk (2002:310) suggests the following
grammatical path of posture verbs: (1) posture > (2) locative/existential > (3) aspectual.
. I point here, again, to Redders (1990) term Deixis in Leere and her explanation that
der Verweisraum von da ist im Leeren fixiert (138). She emphasizes that da can function as
deictic or anaphoric adverb (138).
as follows: Obwohl also da und Topik-es in satzeinleitender Position vorkom-

men knnen, ist ihre Analyse verschieden: da verhlt sich entsprechend wie ein
Subjekt-es (Lenerz 1992:127). However, as Mohr (2004:152) states correctly,
the true nature of da in general is hard to determine. Da can only be considered
expletive-like, since it does not display the typical properties of expletives: it
is clearly not restricted to the sentence-initial position of main clauses (neither
in presentational nor in impersonal sentences) and thus differs from the exple-
tive topic es. Since it is very difficult to distinguish between adverbial and exple-
tive da when investigating real data (particularly from a semantic point of view),
Iconsider it safest to refer to the kind of da relevant to my discussion as a quasi-
expletive, which has a similar but not exactly the same distribution as an expletive
and which has a locative reading and may occur together with a(nother) locative
element. What is important to bear in mind, however, is the fact that da does
not have the same grammatical status as the expletive subject there and that the
existential da-construction has by no means grammaticalised to the same extent
as the English ETC.
In the following, an attempt will be made to uncover how OE and OHG exis-
tential constructions have fared with respect to grammaticalisation in its earliest
stages. I will particularly focus on spotting onset or so-called bridging contexts
of grammaticalisation, i.e. untypical and critical contexts where the transmission
and use of new constructions such as the OE existential r-construction and the
OHG existential thr-construction can be investigated.
4. Development in Old English
4.1 Untypical contexts in OE

The development of the ETC, as it is analysed in the well-known traditional descrip-
tions, perfectly exemplifies Diewalds three stages of development outlined above. It
is often pointed out that the untypical context of r developed due to pragmatic
aspects rather than purely syntactic principles. Two main aspects of the linguistic
situation in OE seem to be of relevance to the onset of grammaticalisation of r:
(a) OE word order has mostly been characterised as consisting of numerous co-
existing patterns which are characterised by a degree of word order flexibility that
is exploited for discourse strategies (cf., e.g. Fischer 1992; van Kemenade, Milicev&
Baayen 2008; Stockwell 1984; Traugott 1992). (b) It is frequently discussed in the
literature on (West)-Germanic syntax that English is a language that has always
favored light elements clause-initially and heavy elements in clause-final posi-
tion, at least in clauses with indefinite subjects. For instance, Koopman (1997:313)
r ws vs. thr was
roposes that it is only after a few adverbs those which van Kemenade, Milicev
p
and Baayen (2008) refer to as discourse partitioners that subjects of whatever
type invert. This points to a clear syntactic influence of the preposed adverb. Apart
from then, which is probably the most common preposed adverb in OE that
triggers subjectverb inversion, inversion also often occurs after preposed adverbs
of place such as locative r. As illustrated in Pfenninger (2009), r frequently
functions as a transitional adverb that refers to a locative element or word group in
the preceding s sentence. The desire for a transition that is pragmatically motivated
would therefore be a logical reason for the frequent occurrence of r at the begin-
ning of the sentence. Indeed, many scholars (cf. van Kemenade, Milicev & Baayen
2008; Stockwell 1984; T raugott 1992; inter alia) point to the pragmatic use of the
V-2 order that signals that a new segment of information is beginning, as is appar-
ent in the f ollowing example (taken from Pfenninger 2009):
se ellengst earfolce rge geolode, s e in strum bd, t h
(5)
dgoragehwm dram gehyrde hldne in healle; r ws hearpan swg,
swutol sang scopes. (Beowulf ed. Jack, 8690)
Then the fierce creature with difficulty suffered the time, he who dwelt in
darkness, that he each of the days heard loud joy in the hall; there was the
sound of a harp, the clear song of a poet.
In (5), the preposed locative r refers back to a given topic, that is, the adverbial
phrase of location in healle in the hall; the verb follows the locative adverb, as
is typical; then, a new character (the sound of a harp) is introduced. Sentences
such as (5) testify to the fact that the continued function of [locative adverbVS],
i.e. [rVS], is to introduce a new or surprising subject (see also Traugott
1992:278 ff.). We can observe some kind of extension of context, which, as out-
lined above, characterises grammaticalisation processes in their early stages. Even
though r is not used in new contexts (i.e. new constructions or in combination
with new elements) it keeps its initial position followed by the verb and the
NP the pragmatic function gains weight and there is a development of poly-
semy, i.e. a development from referential meaning to non-referential meaning, or
from an orientation on the content level to an orientation on the discourse level
(cf. Sweetser 1990). This can result in semantic ambiguity, since either of the two
senses may be implied in the given context. This will be discussed in more detail
in the next section.
4.2 Critical contexts in OE

A lot has been written on how the expletive r semantically split off from
the locative r (cf. Bolinger 1977; Breivik 1989, 1997, 2005; Johansson 1997;
inter alia), which is why I do not go into great detail here. In a nutshell, we can
observe the following steps in the desemanticisation process (examples taken

from P
fenninger 2009):
hldne in healle; r ws / hearpan swg, swutol sang scopes
(6)
(Beowulf ed. Jack 33, 8690)
there was the sound of harps, the clear song of a singer

Nealles mid gewe[a]ldum wyrmhord [b]r[c], / sylfes willum, s e
(7)
him sregesced, / ac for randlan [ow] nthwylces / hlea bearna
heteswengeas flah, / rn[e]s earfa ond r inne feal[h], / secg synbysig.
Sna [onfun]de / t [] am gyste [gryr]ebrga std; / hwre [earm]
sceapen [.. /.. .] sceapen / [..] hyne se fr begeat. /
Sincft [] r ws swylcra fela / in m eor[h]se rgestrona, / sw h
on gerdagum gumena nthwylc, / eormenlfe elan cynnes, / anchycgende
r gehdde, / dore mmas. (Beowulf ed. Jack 157f., 22212236)
He did not deliberately break into the serpents hoard, with his own desire,
he who grievously harmed him, but for dire distress, the slave of someone
of the mens sons fled from hostile blows, in need of a dwelling, and made
his way there in, the guilty man. Soon he discovered that [] the intruder
was facing horror; whether wretched [. .] created [..],
then befell him the sudden attack. A precious vessel [] there were
many ofsuch ancient treasures in the earth-house, just as in former times
someone as an immense legacy of a noble kind deep in thought had hidden
there, theprecious treasures.

Ic s bames mg ae for eorlum elu secgan; r ws hlin ond c ond
(8)
se hearda w ond se fealwa holen.(Riddles ed. Williamson 301, 78)
I can easily tell easily before men recount the lineage of the tree; there was
maple and oak and the bitter yew and the dark holly.
In (6) (earlier quoted as (5)), r is clearly locative, referring back to in healle;

due to the sentence-initial, preverbal position of r, this is where the precondi-
tions of the grammaticalisation process develop, as described above. In (7) r
refers back to r inne, i.e. to the cave where the events take place, while in am
eorsele is added in order to make the referent of r explicit. According to Breivik
(1989:63), the addition of such a second locative leads to a semantic reanalysis of
the locative r: r becomes redundant as an indication of place [and] loses its
stress as well as its locative meaning. r can be interpreted with its new (exis-
tential) as well as its old (locative) meaning, which is typical of the critical context
functions found during Diewalds (2006) stage II. Finally, in (8), there is no loca-
tive phrase in the preceding sentence to which r could refer; the construction
r ws has acquired existential rather than locative meaning, and r is now
r ws vs. thr was
prospective rather than retrospective. Divergence is shown here to be a natural

outcome of the process of grammaticalisation (cf. Hopper & Traugott 2003:119): it
begins as a fixing of a lexical form in a specific context (in our case in clause-initial
position), where the form takes on a new meaning. The grammaticalising item, in
this case the adverb r, will then belong to another word class (here an expletive
topic) than the lexical item it derives from (Fischer, Norde & Perridon 2004:10
ff.). Due to the frequent use with its new meaning and function, r is bleached,
which is reflected in Breiviks (1990:100) observation that at this stage, introduc-
tory r [] can be left out whether or not the story enters on its new phase. Note
that the traditional meaning and function of r persist: with the development
of new function and meaning of r we still find r with its traditional loca-
tive meaning and adverbial function, which leads to the above-mentioned effect
of layering. Furthermore, since the new grammatical meaning is not completely
separate from the older, more lexical, meaning (see Diewald 2006:5), the gram-
maticalisation process cannot be considered consolidated yet.
As indicated above, the grammaticalisation of r occurred within the
framework of the whole construction; thus, it was accompanied by the gram-
maticalisation of OE wesan be. Figure 1 shows the so-called verb-to-affix cline
(Hopper & Traugott 1993), which has a lexical verb (or verbum substantivum
(Lehmann 1982:34)) as its starting point which develops into an auxiliary and
eventually an affix.
lexical be>copula be>auxiliary be>clitic be
Figure 1. Grammaticalisation cline of be
Stage 1 refers to wesan as it appeared in the most common OE existential con-

struction, as in sume gedwolmen wron there were some heretics (lfrics Preface
to Genesis). Wesan functioned as a full verb with a locational-existential exist, be
at, be there sense. As such it was an ideal source for the development of copula and
auxiliary constructions. I have outlined above that in OE, wesan was frequently
accompanied by adverbials such as locative r. In late OE, when the ETC was
slowly emerging, this locative assistance could also be performed by the expletive
r. I suggest it was in these contexts, i.e. where the locative-existential meaning
of be was specified by an additional constituent in its VP, that be lost its semantic
complexity. Once this additional constituent represented a copular complement
(a subject complement or an adverbial), be became reanalysed as a copular verb,
which by definition links the subject of a sentence with a predicate. The copular
complements can change the meaning of the verb (Quirk et al. 1985:1171), which
means that be had, at this stage, fully lost its verbal properties as an existential
verb. In those cases where the predicate complement is a nominalised verb form,
be functions as an auxiliary marking aspect, as in I am swimming. After stage 4,
i.e. once the reanalysis of be as an auxiliary has occurred, be can undergo changes
typical of auxiliaries, such as phonetic reduction. Interestingly, Breivik (1990) sup-
ports the hypothesis that be as a fully lexical existential verb survived in the ETC;
he suggests that the be which occurs in ECs [= existential constructions] is a full
verb, on a par with appear, exist, occur, etc. (52), based on the fact that expletive
there cannot combine with copular be; for example, we cannot say *There is the
weather beautiful.
4.3 Frequency of OE existential r-constructions

Even though it is commonly stated that in OE, ETCs did appear sometimes even
in positions where the V-2 rule did not require them, they were very rare until
late OE. Furthermore, their interpretation is very difficult due to their semantic
ambiguity, which is a common problem with critical contexts (Diewald 2006).
In his analysis of the use and non-use of expletive there both in contemporary
English and earlier English, Breivik (1990) tries to identify the factors governing
the presence and absence of there in existential constructions. Unfortunately, his
results are presented in absolute frequencies only, i.e. no relative or normalised
frequencies are given. Furthermore, he does not analyse the semantics of OE r
quantitatively, e.g. by comparing the frequencies of locative r vs. existential
r; instead, he focuses on a comparison between different OE existential con-
structions (in main as well as subordinate clauses). Table 1 presents a summary
of the distribution of OE existential constructions between r-constructions7
and wesan-constructions. For the sake of completeness, I also provide the relative
frequencies and the overall number of existential r-constructions used per ten
thousand words.
Table 1 reveals that even though the existential wesan-construction repre-
sented the dominant pattern of OE existential constructions, we can find some
unambiguous instances of r as a dummy subject (in main clauses as well as
subordinate clauses). Furthermore, a scrutiny of Breiviks data shows that by no
means all instances of r-insertion can be explained in terms of a purely struc-
tural condition like the V-2-constraint; that is, dummy subjects tended to slowly
appear optionally in positions where the V-2-constraint does not require them
(cf. Section 4.1 above). It has to be mentioned, however, that Breiviks count of
wesan-constructions includes structures with preverbal locative r, as in r
wr East Engla folces seo yld ofslagen (Chronicles, quoted in Breivik 1990:194).
. Note that instances where existential r co-occurs with a passive VP are excluded here.
r ws vs. thr was
Table 1. Distribution of OE existential constructions between r-constructions

andwesan-constructions (partly adopted from Breivik 1990:198)8
Texts8 No. words Exist. wesan- Exist. r- Total Frequency per
constructions constructions 10,000 words
lfric 100,193 150 39 189 3.87

(79.37%) (20.63%) (100%)
Bede 40,383 111 1 112 0.25
(99.11%) (0.89%) (100%)
Blickling 42,506 47 13 60 3.06
(78.33%) (21.67%) (100%)
Chronicles 40,641 37 5 42 1.23
(88.1%) (11.9%) (100%)
Exeter Book 46,543 37 4 41 0.86
(90.24%) (9.76%) (100%)
Orosius 51,020 86 21 107 4.12
(80.37%) (19.63%) (100%)
5. Development in OHG
5.1 Untypical contexts in OHG

Like OE, OHG was a language with an underlying SOV order, with the V-2 rule at
work (cf. Robinson 1997). Even though V-1 word order is not considered a com-
mon pattern in OHG, it nevertheless was common in existential constructions, as
demonstrated in the following example in which OHG prefers the V-1 order over
the SOV order in the Latin original (cf. also Fleischers (2006:43 ff.) examples):
(9) Vidua autem quaedem erat
In ciuitate illa
uuas thr ouh sum uuitua
In thero burgi (Tatian ed. Masser 201, 2)
There was also a widow in this castle
The fact that V-1 order often occurs to introduce new paragraphs or sections
wenn ein neuer Gedanke eingefhrt wird oder ein neuer Erzhlabschnitt
beginnt (Schrodt 2004:199) resembles the situation in OE described above,
where the placement of a preposed adverb at the beginning of a new paragraph
. I used the online versions of these texts from the Labyrinth Library (Georgetown
University).
was a discourse-pragmatic choice and cannot be solely attributed to stylistic

means of transition. Like OE and r, OHG th and thr represented the most
common preposed adverbs that triggered subjectverb inversion (cf. Dittmer &
Dittmer 1998). Thr (with its alternative forms dhar, dr, thr) was frequently used
to introduce new information, i.e. to indicate a change of context, as it refers back
to the time or action given in the previous discourse, thereby playing an important
linking or orientation role in the discourse (cf. Burridge 1993):
Tho sprh er fora theru mnigi slumo thara inggini, lugnit es lles, qaud
(10)
niwsti with thes mnnes. Thar was fur thuruh thz wnta iz filu klt was.
(Otfrid ed. Erdmann 68, 911)
He [Petrus] immediately contradicted in front of all the people; he d enied
everything and said that he didnt know anything about this person.
Therewas a fire because it was very cold
In (10), thr also introduces a new topic (the fire) that exists (or rather is burn-
ing) close to the people. It clearly differs in meaning from the locative instances
of thr which either refer back to a locative element in the preceding context, as in
(11), or function as translations of Latin ibi there, as in (12):
Than uuas im Iohannes fon i siuguhdi auuahsan an nero uustunni;
(11)
tharni uuas uuerodes than mr(Heliand ed. Behaghel, 37, 859862)
From youth on, Johannes grew up in a desert; there, nobody else lived
(12) et die tertio nuptiae facte sunt
In canan galileae.
& erat mater ihesu ibi,
uocatus est autem ibi & ihesus.
In thritten tage btloufti gitano
uuarun in thero steti thiu hiz canan galilee.
thar uuas thes heilantes muoter. (Tatian ed. Masser 177, 1316)
On the third day, the bride and groom were led to a hut called Galilaeus.
There, there was also the mother of Jesus.
Whereas the Latin ibi stands at the end of the clause, its translation with the loca-
tive thr is preponed to make a connection with the preceding context; this clearly
indicates the retrospective character of the latter in the OHG translation (cf.
Masser 1997).
Besides the semantic and pragmatic similarity between r and thr, there are
also syntactic parallels: as a result of the V-2 order, which was well on the way to
becoming generalised in OHG (Haiman 1974), it is observed that thr was used
in initial position in those constructions that tended to constitute violations of the
V-2 order, namely existential and presentational constructions (Lenerz 1992:106).
Like r, thr appears sentence-initially and thus inverts the usual declarative
r ws vs. thr was
order subjectverb but still satisfies the V-2 constraint and the thematic struc-
ture, i.e. it overcomes a potential conflict between discourse strategies and the V-2
order in the language, which is particularly important in existential constructions
that entail the movement rightwards of logical (typically new) subjects and senten-
tial subjects, respectively (Burridge 1993).
5.2 Critical contexts in OHG

In the next step of the grammaticalisation process, we can observe the semantic
bleaching of thr. As Breivik (1990) already observed in his analysis of expletive
r vs. locative r, it is not a straightforward task to identify those critical con-
texts in which the locative meaning of OHG thr is weaker than in others. The
consideration of the wider discourse context is thus essential in any study of exis-
tential constructions with expletives, as the context strongly determines the mean-
ing and function of these preposed adverbs and thus helps to distinguish between
locative and existential r/thr-constructions. In Pfenninger (2009) Iapplied the
following selection criteria, which are also applied to the quantitative analysis in
this paper:
a. Thr is considered a locative adverb in those cases where it refers back to a

locative element in the preceding context and/or can be replaced by a locative
element or phrase, as an helliu in hell or thar innan in there.
b. Thr is considered a non-referring expletive when the conditions in (a) are
not fulfilled and thr is introducing new information, usually in the shape
of an indefinite subject; furthermore, the presence of a second locative also
indicates the expletive function of thr.
Since we have to be careful with OHG sentences that feature initial thr + wesan,
in order not to overlook a possible retrospective function of thr, the ultimate
proof that we are dealing with an existential thr-construction is given in situa-
tions when a new chapter opens or a new paragraph9 is introduced with thr. This
is the case, for instance, in the famous fairy-tale openings:
(13) XV. DE OBVIATIONE ET BENEDICTIONE SYMEONIS.

Thar was ein mn alter, zi slidon gizlter; er was thononti thr gte filu
mnag jar.(Otfrid ed. Erdmann 34, 12)
Once upon a time there was an old man
. Note that since we are dealing with editions and not manuscripts here, there is the
possibility that the paragraph structure might have been introduced by the editor.
(14) XII. ERAT HOMO EX PHARISAEIS HICHODEMUS NOMINE.

Thar was ein mn fruater joh dilthegan gater(Otfrid ed. Erdmann 78, 14)
Once upon a time there was a wise man
Both (13) and (14) illustrate vividly that thr must have prospective instead of the
usual retrospective function, since there is no locative element to which it could
refer, as it constitutes the first element of a new chapter. Consequently, we can
state that the insertion of thr appears to have a discourse motivation because thr
represents the topic that introduces some piece of information that is partially or
completely new to the reader.10 We could even go one step further and claim that
since the use of thr-constructions in text-initial position seems to allow only one
reading, that is, the new existential reading, this might be considered evidence for
an isolation context.
The fairy-tale openings in Otfrid are also revealing if analysed from a differ-
ent perspective: I mentioned above that, in OHG, V-1 order was frequently used
at the beginning of new paragraphs and for reasons of transition (i.e. s emantic
connection with the preceding sentence). In Breiviks (1990) data, existential wesan-
constructions with V-1 order constitute the most dominant pattern of OE existen-
tial constructions; (15) below exemplifies such a case in my OHG data. If we now
compare (13) and (14) above with (15) below (all taken from the same source), it
becomes clear that the initial thr in (13) and (14) is redundant, i.e. its meaning is
not strong enough for it to change or influence the meaning of the whole sentence:
(15) XVII. DE STELLA ET ADVENTU MAGORUM.

Nist mn nihein in wrolti thaz sman al irsgeti, wio manag wntar wurti
zi theru drhtimes gibrti.(Otfrid ed. Erdmann 36, 2930)
There is no man in this world who
(13), (14) and (15) have the same denotations, with or without the expletive thr;
they all represent fairy-tale openings (i.e. open a new chapter) and thus contain new
information. The construction [thr + wesan] appears to have acquired existential
rather than locative meaning, and thr is prospective rather than retrospective.
The only (syntactic) difference is that, in (15), the verb stands in initial position,11
. Note that while High German existential and presentational constructions can be equiv-
alent with respect to existentiality (i.e. when all information provided is new) and thus cannot
be treated separately, only constructions with wesan be are considered here.
. Many scholars (e.g. Bernhardt & Davis 1997; Robinson 1997) suggest that the negative
nature of ni in preverbal position (as in nist) is not that strong and thus consider the word
order of sentences starting with nist as V-1 and not as V-2 order.
r ws vs. thr was
whereas (13) and (14) meet the V-2 requirements, which, at the time of Otfrid, had
started to slowly but surely dominate the OHG word order (cf. R obinson 1997).
This is why the sentence-initial position had to be filled by another element, pref-
erably a lighter one than a focused NP. It is thus safe to assume that OHG had been
in need of an expletive element long before the expletive es appeared in MHG texts
for the first time and that this expletive element was the (almost) non-referential,
prospective thr. Also, it has to be mentioned that no chapter openings or fairy-
tale beginnings could be found in the texts analysed in which th or another pre-
posed adverb introduces the first sentence.
Finally, further evidence for the existence of an expletive thr is found in sen-
tences with a second locative that indicates the bleached locative meaning of thr,
as in (16):
(16) Draco iste quem formasti ad illedendum ei.

[Diz mre ist michel] [] Dr ist nne dser zligo draco. serpens antiquus
den du ze huhe habest kemachot.(Psalter ed. Tax 386, 24)
There is this dangerous dragon in there
Sentences such as these (cf. Pfenninger 2009 for more examples) usually start
with the existential thr-construction (highlighted in bold) followed by the sec-
ond locative (underlined); it is obvious that the latter is clearly the semantic
equivalent of the former, which means that thr has no or at least insufficient use
as a locative.
5.3 Frequency of OHG existential thr-constructions

Following the criteria just outlined, we find a number of incipient existential thr-
constructions in OHG material;12 Table 2 presents a summary of the frequency
of OE existential constructions, differentiating between thr-constructions and
wesan-constructions in six major OHG texts, while Table 3 shows the frequency
of the adverb thr in combination with wesan, both in locative and existential
thr-constructions.
. Note that only constructions with the sequence [thr + wesan + indefinite subject-NP]
were analysed according to their locative or existential meaning. Existential sentences with
thr-insertion and a passive VP, in which the whole of the VP occurs before the subject-NP,
were not considered.
Table 2. Frequency of OHG existential constructions (thr-constructions vs.

wesan-constructions)
Texts No. words Exist. wesan- Exist. thr- Total Frequency per
constructions constructions 10,000 words
Isidor 5,029 6 1 7 1.99

(~ 790 AD) (85.71%) (14.29%) (100%)
Tatian 48,406 22 3 25 0.62
(~ 830 AD) (88%) (22%) (100%)
Heliand 46,493 26 13 39 2.80
(~ 830 AD) (66.67%) (33.33%) (100%)
Otfrid 72,012 52 12 64 1.67
(~ 865 AD) (81.25%) (18.75%) (100%)
Notker 188,281 94 11 105 0.58
(~ 9501022 AD) (89.52%) (10.48%) (100%)
Miscellaneous* 28,607 12 2 14 1.05
(~ 750900 AD) (85.71%) (14.29%) (100%)
*This refers to the text collection of minor OHG monuments that is available from the TITUS
(Thesaurus Indogermanischer Text- und Sprachmaterialien) corpus, which is a joint project of the
Institute of Comparative Linguistics of the Johann Wolfgang Goethe-Universitt, Frankfurt am Main,
and the electronic database called Bibliotheca Augustana.
Table 3. The semantics of OHG thr in [thr + wesan + subject NP]*

Texts No. words Locative Existential Total
Isidor 5,029 2 (66.67%) 1 (33.33%) 3 (100%)

Tatian 48,406 9 (75%) 3 (25%) 12 (100%)
Heliand 46,493 14 (51.85%) 13 (48.15%) 27 (100%)
Otfrid 72,012 21 (63.64%) 12 (36.36%) 33 (100%)
Notker 188,281 28 (71.79%) 11 (28.21%) 39 (100%)
Miscellaneous 28,607 16 (88.89%) 2 (11.11%) 18 (100%)
*The texts under investigation were accessed on the electronic database TITUS.
The figures confirm my initial impression that the frequency of existential thr
in my OHG data is low. Table 2 shows that, like in OE, existential constructions
without an additional dummy subject were clearly the dominant pattern in OHG,
but that OHG existential thr occurred with lower frequencies than OE existential
r (cf. Table 1 above); comparing the overall frequencies of existential r and
existential thr (in relation to the size of the text samples) with a chi-square contin-
gency test, it is revealed that this difference in occurrence is significant (X2=19.0,
df = 1, p < 0.001). The OE construction seems to be more advanced, which sup-
ports the hypothesis that it began to grammaticalise earlier than its OHG equiva-
lent, i.e.that it was already further grammaticalised at that time.
r ws vs. thr was
Table 3 shows the number of constructions where, on the one hand, thr
appears with its literal meaning and, on the other one, where the original meaning
appears, in my judgment, to be completely absent. Such an analysis of constructions
with wesan in combination with locative vs. existential thr provides a reasonably
systematic basis to compare the different forms. For instance, it becomes obvious
that the sequence [thr + wesan + subject NP] rarely occurs in the data, which is
not surprising if we consider that V-1 order was a popular option in OHG. Also,
it reveals that existential thr seems to occur in approximately one third of this
rare construction type. Thus, the implication is that whenever thr was inserted
in existential wesan-constructions (presumably to meet the V-2 constraint), it was
quite often used as a quasi-expletive with a very weak locative meaning.
Of course, many of these assumptions remain hypothetical, since the kind
and amount of OE and OHG data under investigation are so different. However,
we have in fact seen enough evidence to speculate that, even though the majority
of the OHG thr-constructions had locative connotations, there is no doubt that
there is evidence for an earlier situation in which thr conveyed existence and not
pure location. Thus, the growing tendency in OHG to use thr as a quasi-expletive
parallels the development of the grammaticalisation process of the expletive r
in OE.
6. Conclusion
It has been shown in this study that the English and High German existential
constructions with locative adverbs displayed similar features in the first stages of
their long processes of grammaticalisation. Since High German is closely related
to English, it is not surprising that it has an existential element, quasi-expletive da
in the existential da-construction, which seems to have derived in a very similar
way to the ModE expletive there in the ETC, namely through grammaticalisation
of a fully stressed locative adverb. Both existential r and thr semantically split
off from their locative equivalents, as their locative meaning became bleached.
The constructions r ws and thr was acquired existential rather than a locative
meaning, and r and thr thus began to be used with prospective rather than ret-
rospective function. Furthermore, not only did the single elements in those con-
structions undergo grammaticalisation, but the whole constructions themselves,
i.e. the sequences [rVS] and [thrVS]. The quantitative analyses provided
in this study indicate the extent to which grammaticalisation was occurring in the
development of r and thr as expletives.
It is well-known that the further development of the High German existential
da-constructions did not fully parallel that of the ETC, as the sequence [thr+
wesan + subject NP] did not become stored and processed as a prefabricated
phrase (cf. Pfenninger 2009). It has been demonstrated in this paper that already
in OHG, this construction was relatively rare compared to its OE counterpart. As
far as its further development is concerned, it is suggested in Pfenninger (2009)
that the demise of the High German existential thr-construction and the preva-
lence of the English r-construction are related to major changes that occurred
in the history of these two languages. Whereas High German grammaticalised
V-2, SVO order was the direction that ME took approximately from the twelfth
century on, which gives rise to the first major syntactic difference between there
and da: there is reanalysed as the syntactic subject, whilst da keeps its status as a
semi-expletive or quasi-expletive that displays expletive-like qualities but does not
always appear in topic position. In the case of High German existentials, gram-
maticalisation is thus restricted to one or two parameters, i.e. desemanticisation as
a consequence of extension. This set the ground for the subsequent development
of a variety of existential constructions in the history of High German (such as the
EGC, es + lexical verb, etc.) that specify the various assertions of the different types
of English ETCs. By contrast, due to a much higher degree of grammaticalisation
as far as the ETC is concerned, English lost the more lexical constructions and
uses the fully grammaticalised ETC that is able to express numerous propositions.
Sources
Beowulf: Jack, George (ed.). 1994. Beowulf. A Student Edition. Oxford: Clarendon Press.
Bibliotheca Augustana. htp://www.hs-augsburg.de/~harsch/augustana.html (31 October 2010).
Heliand: Heliand und Genesis, 10th edn., Otto Behaghel (ed.). Tbingen: Max Niemeyer, 1969.
Isidor: Der althochdeutsche Isidor. Nach der Pariser Handschrift und den Monseer Fragmenten. In
Altdeutsche Textbibliothek 63, Hans Egger (ed.). Tbingen: Max Niemeyer, 1964.
Miscellaneous:
Die kleineren althochdeutschen Denkmler. Edited by Elias Steinmeyer. 2nd edn. Berlin
andZrich,1963.http://titus.unifrankfurt.de/texte/etcs/germ/ahd/klahddkm/klahd.
htm?klahd002.htm (31 October 2010).
lteste Deutsche Dichtung und Prosa. Ausgewhlte Texte althochdeutsch neuhochdeutsch.
Edited by Heinz Mettke. Frankfurt am Main: Rderberg, 1976. http://titus.unifrankfurt.
de/texte/etcs/germ/ahd/klahddkm/klahd.htm?klahd002.htm (31 October 2010).
Althochdeutsches Lesebuch. Zusammengestellt und mit Wrterbuch versehen von Wilhelm Braune.
Edited by Karl Helm. Halle: Niemeyer, 1949.
LAD:
Lawrence, David Herbert. 1993 [1928]. Lady Chatterleys Lover. London: Penguin Books.
Lawrence, David Herbert. 2003 [1960]. Lady Chatterleys Lover. Trans. Maria Carlsson. 29th ed.
Reinbek bei Hamburg: Rowohlt.
Labyrinth Library (Georgetown University). http://www8.georgetown.edu/ departments/
medieval/labyrinth/labyrinth-home.html (31 October 2010).
MUR:
r ws vs. thr was
Beckett, Samuel. 1957. Murphy. New York: Groove Press.

Beckett, Samuel. 2003 [1959]. Murphy. Trans. Elmar Tophoven. 19th ed. Hamburg: Rowohlt.
Notker: Die Schriften Notkers und seiner Schule, 2nd edn. Edited by Paul Piper. Freiburg and
Tbingen: Akademische Verlagsbuchhandlung, 1883. http://www.archive.org/stream/
dieschriftennotk02notkuoft/dieschriften notk02notkuoft_djvu.txt (31 October 2010).
Otfrid: Otfrids Evangelienbuch. In Oskar Erdmann (ed.). 3rd ed. Altdeutsche Textbibliothek 49.
Tbingen: Max Niemeyer, 1957.
Psalter: Notker der Deutsche. Der Psalter. In Petrus W. Tax (ed.). Altdeutsche Textbibliothek 84.
Tbingen: Max Niemeyer, 1979.
Riddles: The Old English Riddles of the Exeter Book, Craig Williamson (ed.). Chapel Hill NC:
University of Carolina Press, 1977.
Tatian: Die lateinische-althochdeutsche Tatianbilingue. Stiftsbibliothek St. Gallen Cod. 56 [Studien
zum Althochdeutschen, Band 25], Achim Masser (ed.). Gttingen: Vandenhoeck and
Ruprecht, 1994.
TITUS Thesaurus Indogermanischer Text- und Sprachmaterialien. http://titus.uni-frankfurt.
de/indexe.htm?/texte/texte.htm (31 October 2010).
WHI:
DeLillo, Don. 1999 [1984]. White Noise. London: Picador.
DeLillo, Don. 2001 [1997]. White Noise. Trans. Helga Pfetsch. 2nd ed. Reinbek bei Hamburg:
Rowohlt.
References
Bergs, Alexander T. & Diewald, Gabriele (eds). 2008. Contexts and Constructions [Construc-
tional Approaches to Language 9]. Amsterdam: John Benjamins.
Bernhardt, Karl A. & Davis, Graeme. 1997. The Word Order of Old High German. Lewiston: The
Edwin Mellen Press.
Bolinger, Dwight. 1977. Meaning and Form. London: Longman.
Breivik, Leiv E. 1989. On the causes of syntactic change in English. In Language Change. Contri-
butions to the Study of its Causes, Leiv Egil Breivik & Ernst Hkon Jahr (eds), 2968. Berlin:
Mouton de Gruyter.
Breivik, Leiv Egil. 1990. Existential There. A Synchronic and Diachronic Study, 2nd edn. Oslo:
Novus.
Breivik, Leiv Egil. 1993. On the interrelation of syntax, semantics, and pragmatics: A study of
locative/temporal expressions in existential sentences in the LOB corpus. In From lfric
to the New York Times. Studies in English Corpus Linguistics, Udo Fries, Viviane Mller &
Peter Schneider (eds), 110. Amsterdam: Rodopi.
Breivik, Leiv Egil. 2005. On the rhetorical function of existential there. http://www.hum.uit.no/
nordlit/6/breivik.html (4 June 2005).
Brinkmann, Hennig. 1962. Die deutsche Sprache. Gestalt und Leistung. Dsseldorf: Pdago-
gischer Verlag Schwann.
Brinton, Laurel J. & Traugott, Elizabeth C. 2005. Lexicalization and Language Change.
Cambridge: CUP.
Burridge, Kate. 1993. Syntactic Change in Germanic. Aspects of Language Change in Germanic.
With Particular Reference to Middle Dutch [Current Issues in Linguistic Theory 89].
Amsterdam: John Benjamins.
Bybee, Joan & Torres Cacoullos, Rena. 2009. The role of prefabs in grammaticization. How the
particular and the general interact in language change. In Formulaic Language, Vol. 1: Dis-
tribution and Historical Change [Typological Studies in Language 82], Roberta Corrigan,
Edith A. Moravcsik, Hamid Ouali & Kathleen M. Wheatley (eds), 187217. Amsterdam:
John Benjamins.
Clark, Eve V. 1978. Locationals: existential, locative, and possessive constructions. In Univer-
sals of Human Language, Vol. 4: Syntax, Joseph J. Greenberg (ed.), 85126. Stanford CA:
Stanford University Press.
Diewald, Gabriele. 2002. A model for relevant types of contexts in grammaticalization. In New
Reflections on Grammaticalization [Typological Studies in Language 49], Ilse Wischer &
Gabriele Diewald (eds), 103120. Amsterdam: John Benjamins.
tions, Special Vol. 1: Constructions all over Case Studies and Theoretical Implications,
Doris Schnefeld (ed.). http://elanguage.net/journals/constructions/article/view/24/29
Dittmer, Arne & Dittmer, Ernst. 1998. Studien zur Wortstellung Satzgliedstellung in der altho-
chdeutschen Tatianbersetzung. Gttingen: Vandenhoeck and Ruprecht.
Eckardt, Regine. 2006. Meaning Change in Grammaticalization: An Enquiry into Semantic Anal-
ysis. Oxford: OUP.
Fischer, Olga. 1992. Syntax. In The Cambridge History of the English Language, Vol. II: 10661476,
Norman Blake (ed.), 207298. Cambridge: CUP.
Fischer, Olga, Norde, Muriel & Perridon, Harry (eds). 2004. Up and Down the Cline: The
Nature of Grammaticalization [Typological Studies in Language 59]. Amsterdam: John
Benjamins.
Fleischer, Jrg. 2006. Zur Methodologie althochdeutscher Syntaxforschung. Aus Anlass des
Erscheinens von: Richard Schrodt, Althochdeutsche Grammatik, II: Syntax. Beitrge zur
Geschichte der deutschen Sprache und Literatur 128: 2569.
Gaeta, Livio. 2005. Hilfsverben und Grammatikalisierung: Die fatale Attraktion von geben. In
Grammatikalisierung im Deutschen, Torsten Leuschner, Tanja Mortelmans & Sarah De
Groodt (eds), 193209. Berlin: Walter de Gruyter.
van Gelderen, Elly. 1991. To be and indices. In Issues in Germanic Syntax, Werner Abraham,
Wim Kosmeijer & Eric Reuland (eds), 307325. Berlin: Mouton de Gruyter.
Givn, Talmi. 1993. English Grammar. A Function-based Introduction, Vol. 2. Amsterdam: John
Benjamins.
Haiman, John. 1974. Targets and Syntactic Change. The Hague: Mouton.
Hannay, Michael. 1985. English Existentials in Functional Grammar. Dordrecht: Foris.
Heine, Bernd. 2002. On the role of context in grammaticalization. In New Reflexions on Gram-
maticalization [Typological Studies in Language 49], Ilse Wischer & Gabriele Diewald
(eds), 83102. Amsterdam: John Benjamins.
Heine, Bernd & Kuteva, Tania. 2005. Language Contact and Grammatical Change. Cambridge:
CUP.
Heine, Bernd & Kuteva, Tania. 2006. The Changing Languages of Europe. Oxford: OUP.
Heine, Bernd, Claudi, Ulrike & Hnnemeyer, Friederike. 1991. Grammaticalization. A Concep-
tual Framework. Chicago IL: University of of Chicago Press.
Hopper, Paul J. 1991. On some principles of grammaticalization. In Approaches to Grammaticali-
zation, Vol. 1 [Typological Studies in Language 19(1)], Elizabeth Closs Traugott & Bernd
Heine (eds), 1735. Amsterdam: John Benjamins.
Hopper, Paul J. & Traugott, Elizabeth Closs. 1993. Grammaticalization. Cambridge: CUP.
r ws vs. thr was
Hopper, Paul J. & Traugott, Elizabeth C. 2003. Grammaticalization, 2nd edn. Cambridge: CUP.
Huddleston, Rodney & Pullum, Geoffrey K. 2002. The Cambridge Grammar of the English Lan-
guage. Cambridge: CUP.
Jespersen, Otto. 1924. The Philosophy of Language. London: George Allen and Unwin.
Johansson, Stig. 1997. A corpus study of existential clauses: Register variation and discourse-
function. In To Explain the Present. Studies in the Changing English Language in Honour
of Matti Rissanen, Terttu Nevalainen & Leena Kahlas-Tarkka (eds), 303318. Helsinki:
Socit Nophilologique.
van Kemenade, Ans, Milicev, Tanja & Baayen, R. Harald. 2008. The balance between syntax and
discourse in Old English. In English Historical Linguistics 2006 [Current Issues in Linguistic
Theory 295], Maurizio Gotti, Marina Dossena & Richard Dury (eds), 321. Amsterdam:
John Benjamins.
Koopman, Willem F. 1997. Topicalization in Old English and its effects. Some remarks. In
Language History and Linguistic Modelling. A Festschrift for Jacek Fisiak on his 60th Birth-
day, Vol I: Language History, Raymond Hickey & Stanislaw Puppel (eds), 101111. Berlin:
Mouton de Gruyter.
Lehmann, Christian. 1982. Thoughts on Grammaticalization. A Programmatic Sketch. Volume1
[Arbeiten des Klner Universalien-Projekts 48]. Kln: Universitt zu Kln, Institut fr
Sprachwissenschaft.
Lenerz, Jrgen. 1992. Zur Theorie syntaktischen Wandels: Das expletive es in der Geschichte des
Deutschen. In Erklrende Syntax des Deutschen, 2nd edn, Werner Abraham (ed.), 99136.
Tbingen: Gunter Narr.
Lichtenberk, Frantisek. 2002. Posture Verbs in Oceanic. In The Linguistics of Giving, J. N
ewman
(ed.), 269314. Amsterdam: John Benjamins.
Long, Ralph B. 1961. The Sentence and its Parts. A Grammar of Contemporary English. Chicago
IL: University of Chicago Press.
Lumsden, Michael. 1988. Existential Sentences. Their Structure and Meaning. London: Croom
Helm.
Masser, Achim. 1997. Syntaxprobleme im althochdeutschen Tatian. In Semantik der syntak-
tischen Beziehungen. Akten des Pariser Kolloquiums zur Erforschung des Althochdeutschen
1994, Yvon Desportes (ed.), 123140. Heidelberg: Carl Winter.
Meineke, Eckhard 2001. Einfhrung in das Althochdeutsche. Paderborn: Ferdinand Schningh.
Milsark, Gary L. 1974. Existential Sentences in English. New York NY: Garland.
Mohr, Sabine. 2004. Clausal Architecture and Subject Positions: Impersonal Constructions in
the Germanic Languages. Ph.D. dissertation, University of Stuttgart.
Newman, John. 1996. Give: A Cognitive Linguistic Study. [Cognitive Linguistics Research 7].
Berlin and New York: Mouton de Gruyter.
Newman, John. 1997. The origin of the German es gibt construction. In The Linguistics of Giving
[Typological Studies in Language 36], John Newman (ed.), 307325. Amsterdam: John
Benjamins.
Newman, John (ed.). 2002. The Linguistics of Sitting, Standing, and Lying [Typological Studies in
Language 51]. Amsterdam: John Benjamins.
Pfenninger, Simone E. 2009. Grammaticalization Paths of English and High German Existential
Constructions. Bern: Peter Lang.
Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey & Svartvik, Jan (eds). 1985. A Compre-
hensive Grammar of the English Language. Harlow: Longman.
Redder, Angelika. 1990. Grammatiktheorie und Sprachliches Handeln: denn und da. Tbingen:
Max Niemeyer.
Robinson, Orrin W. 1997. Clause Subordination and Verb Placement in the Old High German
Isidor Translation. Heidelberg: Carl Winter.
Schrodt, Richard. 2004. Althochdeutsche Grammatik, II: Syntax. Tbingen: Max Niemeyer.
Serra Borneto, Carlo. 1996. Liegen and stehen in German: a study in horizontality and ver-
ticality. In Cognitive Linguistics in the Redwoods: The Expansion of a New Paradigm in
Linguistics, E. H. Casad (ed.), 459505. Berlin: Mouton de Gruyter.
Sonderegger, Stefan. 2003. Althochdeutsche Sprache und Literatur. Eine Einfhrung in das lteste
Deutsch. Darstellung und Grammatik, 3rd edn. Berlin: Walter de Gruyter.
Stockwell, Robert P. 1984. On the history of the verb-second rule in English. In Historical Syntax,
Jacek Fisiak (ed.), 575590. Berlin: Mouton de Gruyter.
Sweetser, Eve. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Seman-
tic Structure [Cambridge Studies in Linguistics 54]. Cambridge: CUP.
Traugott, Elizabeth C. 1992. Syntax. In The Cambridge History of the English Language, Vol. I:
The Beginnings to 1066, Richard M. Hogg (ed.), 168286. Cambridge: CUP.
Wandruszka, Mario. 1969. Sprachen- Vergleichbar und unvergleichbar. Mnchen: R. Piper & Co.
On gain and loss of verbal categories
inlanguage contact
Old English vs. Old High German*
Theo Vennemann
University of Munich, Germany
The theory of language change has in recent years increased its explanatory
repertoire by pointing out the role of language contact in determining which
paths of development are entered and followed under specified conditions. In
particular, language shifting as unmonitored second language learning is
recognized as a powerful mechanism for introducing new verbal categories into
language systems as well as leading to the loss of verbal categories from language
systems. In this paper I will relate several of the most important structural
changes and categorial differences in the verb systems of Proto-Germanic,
Old English and Old High German to the different contact histories of these
languages, among them: (1) the reduction of the Proto-Indo-European TAM
system (TAM for tense, aspect, mood) to half its size in Proto-Germanic, (2)the
existence of a double copular paradigm in Old English (and again in Irish
English) but not in German; (3) a number of properties of English but not of
German attributed to Celtic influence by Filppula, Klemola, and Paulasto (2008),
such as the loss of the affected possessor construction and the rise of the verbal
noun in -ung/-ing and the progressive based on it.
1. Introduction
In Old English and Old High German we find both identical and different basic
verbal categories. Identical verbal categories come in two varieties: They may be
inherited, and they may be owed to shared innovation. Different verbal categories
may be owed to differential loss, i.e. loss in one of the two languages but not in the
other. Thus, Old English had a 1st and 2nd person dual of the personal pronoun;
*I would like to thank Stephen Laker (Kyushu University, Japan), Robert Mailhammer
(Arizona State University, Tempe), Iva Welscher (University of Munich) and an anonymous
referee for reading manuscript versions of this paper and for making valuable suggestions.
Theo Vennemann
Old High German did not: it had lost this category. (English, too, lost it early in
the Middle English period.) On the other hand, different verbal categories may
also be owed to differential innovation. The question arises in each case whether
the changes involved reflect ordinary internal language change or the effects of
language contact. In the present paper I will deal especially with such changes
for which I have reasons to assume an effect of language contact. I will not only
discuss TAM categories (TAM for tense, aspect, and mood) but understand the
term verbal category loosely so as to include e.g. voice categories and categories
of verbs.
2. Shared inherited categories
The following were the TAM categories of Old English and Old High German
before they began developments of their own, i.e. the major verbal categories of
Proto-Germanic:
T (tense): present, preterite
A (aspect): none
M (mood): indicative, optative, imperative
The so-called present tense, which carries no inherent temporal meaning but by
default indicates the time of the speech-act, is the continuation of the Proto-Indo-
European simple imperfective aspect, likewise traditionally called the present
tense.
The preterite of the so-called strong verbs continues the Proto-Indo-European
perfect, which had the same imperfective aspectual force as the present but in
addition characterized the focused state as the result of a past event.1 The latter
property led to its re-interpretation as a past tense, following a universal unidirec-
tional path of semantic development.2 In the course of losing this double-layered
. Cf. Meier-Brgger 2010: F 202, 2; S 307: Der Perfektstamm bezeichnet eine Art von resul-
tativem Aspekt. Wie der Indikativ Prsens nimmt das Perfekt Bezug auf die Zeitstufe der
Gegenwart; dabei bezeichnet das Perfekt den nach einem vorausgegangenen Verbalgeschehen
erreichten Zustand am Subjekt (die Ziege hat gefressen = die Ziege ist satt). [The perfect stem
indicates a kind of resultative aspect. Just like the indicative present the perfect refers to the
present time; in so doing the perfect indicates the state of the subject reached after the event
expressed by the verb (the goat has eaten = the goat is full up).]
. Speaking about categories (as opposed to individual forms and to extended usage), a
perfect may become a preterite, but I know of no case of a preterite becoming a perfect.
On gain and loss of verbal categories in language contact
semantics present imperfective on the surface, but with an embedded reference

to a past event the perfects of the individual verbs for the most part preserved the
embedded semantics and became preterites. But a minority preserved the upper
layer present imperfective semantics; they became the so-called perfecto-present
or preterite-present verbs. Formally the Germanic preterites and preterite-presents
reflect the Proto-Indo-European perfect both in its ablaut behavior (with different
grades in the singular and plural indicative) and in its residual reduplication (in
class VII).
As to the moods, indicative and imperative continue the same Indo-European
categories. The same holds for the optative, except that in Germanic, especially
inthe individual Germanic languages, the category has acquired additional func-
tions and is therefore usually named differently (e.g. subjunctive, Konjunktiv,
Mglichkeitsform).
Concerning the TAM categories of Proto-Indo-European, the various intro-
ductions to the field show slight differences. The ten categories identified in
Mailhammer 2006 are probably close to the mark. In addition to the five categories
already mentioned (present, perfect, indicative, optative, imperative), there was a
perfective category, the aorist, and the injunctive, a form without any temporal ref-
erence which is nonetheless placed among the temporal categories by some linguists
(e.g. Mailhammer 2006) but among the moods by others (e.g. Beekes 1995:245).
The preterite, formed with a prefix +h1e- (called the augment), occurs only in
the following Indo-European branches: Greek, Phrygian, Armenian, Indo-Iranian
(cf. Meier-Brgger 2000: F 213). Whether Proto-Indo-European possessed a plu-
perfect is uncertain (cf. Beekes 1995:239).
Table 1 summarizes the development of the Germanic TAM categories from
the Proto-Indo-European TAM system. The left column shows the Proto-Indo-
European categories as considered in Mailhammer 2006, the second and third col-
umns show the Germanic TAM categories. Those Indo-EuropeanTAM categories
which survived into Proto-Germanic are printed recte, those lost in the process,
in italics.
The most remarkable changes occurred in the aspectual group. The aorist, the
only perfective category, was lost without a trace. The perfect took over some of
its load and formed a new opposition with the inherited present imperfective, an
aspectual opposition of present imperfective vs. past perfective which, still in pre-
historic times, developed into a temporal opposition of present vs. preterite. The
latter change, a natural kind of development, was probably invited or facilitated by
the loss of all old temporal categories, to the extent that they existed in the source
language. In the modal group the subjunctive was lost, its functions taken over by
the optative and by the indicative.
Theo Vennemann
Table 1. Major verbal categories (TAM categories) in Proto-Indo-European,

Proto-Germanic, and Phoenician3
Indo-European Germanic Phoenician
Aspectual Aspectual Temporal Aspectual
Imperfective:
Present Imperfective > Present Imperfective
Perfect Perfective > Preterite Perfective
Perfective:
Aorist
temporal
Injunctive
Imperfect
Pluperfect
modal modal modal

Indicative Indicative Indicative
Optative Optative Jussive3
Subjunctive
Imperative Imperative Imperative
This remarkable reduction of the TAM system in Germanic has often been com-
mented on. Except for general remarks bringing the possibility of language contact
(substrate influence) into play, no specific explanation can be found in the intro-
ductions to Germanic. The only explanation in existence presupposes the theory
that Germania in her prehistory was for several centuries under intensive Semitic,
namely Phoenician (Carthaginian), influence (for which cf. Vennemann 2000, 2004a
and earlier articles in Vennemann 2003). The application to the TAM system itself
. The functions of the optative and the jussive moods are rather similar. Their mean-
ings overlap in the expression of wishing. This is especially clear in the comparison on the
Internet page Grammatical mood, http://en.wikipedia.org/wiki/Grammatical_mood
(22 September 2010): The optative mood expresses hopes, wishes or commands and has
other uses that may overlap with the subjunctive mood. The jussive mood expresses
pleading, insistence, imploring, self-encouragement, wishing, desiring, intention, com-
manding, purpose or consequence. It is thus natural for a speaker of a language with the
jussive mood to equate it with the optative of a language to be learned.
was proposed by Mailhammer (2006). He compared the Proto-Indo-European and

Germanic TAM categories to those of Carthaginian Phoenician (Punic) and found
that the latter are essentially the same as those Proto-Indo-European categories that
survived into Germanic. The assumption is that Punic second-language learners
of pre-Germanic learned the pre-Germanic (Proto-Indo-European) TAM system
through the filter of their own native language, which meant that they understood
and learned only those categories that they knew from their previous, native experi-
ence, ignoring all others. The subsequent events then were the same as in the gener-
alization and functionalisation of ablaut which I sketched in Vennemann 1998 and
which Mailhammer (2006, 2007) worked out in great detail:
In this way they [the superstratal speakers of Atlantic Semitic, later identified
as Carthaginian Phoenician (Punic)] created dialects of Germanic that were
likewise characterized by a powerful verbal and deverbal ablaut system. These
new dialects were spoken in part by members of those social groups that included
the former superstratum speakers, which suggests that they became prestige
dialects. As prestige dialects they superseded in time all the other pre-Germanic
dialects and formed the basis for the emerging Proto-Germanic language.
(Vennemann 1998:42)
3. Shared innovated categories
The two TAM categories that both English and German have innovated are the
periphrastic perfect and the periphrastic future, together with their derived subcat-
egories such as the past perfect (pluperfect) and the future perfect (second future).
Also a new voice category, the passive, begins developing in both Old English and
Old High German. Both languages show periphrastic expressions leading to the
rise of these new categories (cf. e.g. Blake 1996:99f; Besch & Wolf 2009:150). The
fact that the infinite forms involved in these constructions are increasingly used
without their agreement inflection shows that the grammaticalisation of these
TAM categories is well underway. Besch and Wolf specifically say about the Old
High German passive and perfect:
Wichtig ist, da es bei all diesen Fgungen zunchst nicht um eine Unterscheidung
von Diathesen oder Tempora ging. Der Vorgang, der zur Verfassung des
Subjekts fhrt, die eben durch das Partizip ausgedrckt wird, ist nur dadurch
in der Aussage enthalten, dass den Verben werdan und wesan kein Adjektiv oder
Substantiv, sondern ein Verbaladjektiv folgt. (Rupp 1956 282) Je nach Valenz
des mutativen Verbs entsteht eine passivische oder perfektivische Fgung.
(Besch & Wolf 2009:150)
Theo Vennemann
[It is important to realize that the purpose of all these constructions is not a
differentiation of voices or tenses. The process leading to the specific state
of the subject, which is expressed precisely by the participle, is contained in
the predication only by virtue of the fact that the verbs werdan and wesan are
followed by a verbal adjective rather than an adjective or noun. (Rupp 1956:282)
Depending on the valency of the mutative verb the result is a passive or a
perfect construction.]
The full integration of the new categories into the verbal systems postdates Old
English and Old High German. This assessment is in harmony with Mailhammer
and Smirnovas conclusion (this volume):
As we have shown and as the data confirms, Old English as well as Old High
German did not possess grammaticalised passive periphrases. Constructions
with the copula verbs be and become plus past participles served as fully
compositional structures with primary aspectual meaning whose passive
readings resulted from the logical combination of the aspectual and transitivity
values of their composite parts. But there is indication of a beginning process of
grammaticalisation in Late Old English and Late Old High German.
(Mailhammer & Smirnova, this volume, p. 66)
4. Differential innovated categories
4.1 Two copulas

I said in the Introduction that I will understand sets of verbal categories loosely
both in the sense of grammatical categories expressed on verbs and of lexical
categories of verbs. The first example I will discuss in this section belongs to the
second group: the category of copular verbs.4
Proto-Germanic, as indeed Proto-Indo-European, had only a single copula.5
This may not strike anyone as peculiar, because the same is true for Contemporary
. This exposition follows Vennemann 2010: Section 2.

. Like several other Indo-Europeanlanguages (Italic, Celtic, Lithuanian, Slavic, Greek, Old
Indic), Proto-Germanic may have had forms derived from Proto-Indo-European +bheu-/+bh-
to become. There is no evidence, however, that these forms developed within Proto-Germanic
into a copular paradigm meaning to be the way this happened in Old English (and possibly in
all of West Germanic, see below). Seebold (1970: s.v. *beww-), who does make this assumption,
faces the problem that no such paradigm and, indeed, no b-forms meaning to be at all, exist
in Gothic and in Old Norse; he writes, without further discussion, beseitigt im Gotischen und
Nordischen [eliminated in Gothic and Old Norse]. In the light of all the problems Seebold
English as well as the English of Shakespeare and of Chaucer. It is also true see
Table 2 for historical North and East Germanic, as evident in the single Old
Norse and Gothic present indicative paradigms translating the forms of English to
be. But it is not true for Old English.
Table 2. A single copula in Proto-Indo-European

and in Proto-Germanic (the s-paradigm)6
Old Icelandic Gothic
em im (I) am
es(t) is (thou) art
es ist (he/she/it) is
erom sijum (we) are

ero siju (you) are
ero sind (they) are
4.1.1 Two copulas in Old English

All Old English dialects had, from the time of their earliest attestation, two cop-
ulas, the s-copula7 and the b-copula,8 each with a complete present indicative,
optative, and imperative paradigm of which I show the West Saxon indicative sub-
paradigms in Table 3.
Table 3. Two copulas in Old English (West Saxon)
(1970: s.v. *beww-) has accounting for +bheu-/+bh- to become in West Germanic, it seems
equally possible, and to me more likely, that Gothic and Old Norse reflect the Proto-Germanic
situation and that the West Germanic b-forms are entirely owed to Celtic influence. Krahe
(1967: 98) too traces the West Germanic copular b- directly to Proto-Indo-European. E.g.
he derives Old English bo via West Germanic 1st sing. pres. *biu from Proto-Indo-European
*bheu. This is not cogent; cf. Old Irish buu/.bu below, Note10.
. I call this paradigm the s-paradigm because it is, with certain irregularities, the etymo-
logical continuation of the present of the Indo-European copula, based on the root +h1es-: sing.
+h s-mi, +h s-si > +h si, +h s-ti, plur. +h s-ms, +h s-t, +h s-nti (cf. Sihler 1995: 492).
1 1 1 1 1 1 1
. See the preceding note.

. I call this copular paradigm the b-paradigm because it is based on the Indo-European
root +bheu- and Proto-Indo-European +bh became +b in Germanic (and Celtic, see below).
Whereas +h1es- inflected as a present/imperfect, +bheu- was aoristic, probably because its
original meaning was to become rather than to be (Sihler 1995: 491). A very detailed
description of the West Germanic, especially Old English copula paradigms is Flasdieck 1937.
Theo Vennemann
s-paradigm b-paradigm
eom bo (I) am
eart bist (thou) art
is bi (he/she/it) is
sind(on) bo (we/you/they) are
Of these, the s- or eom-paradigm is recognizably a formal continuation of the

Germanic paradigm also reflected in Old Norse and Gothic, while the b- or bo-
paradigm is an innovation. As to the meaning of these two copulas in Old English,
Campbell writes:
bo expresses what is (a) an invariable fact, e.g. ne bi swylc cwenlic eaw [Beowulf
1940] such is not a queenly custom, or (b) the future, e.g. ne bi e wilna gad
[Beowulf 660] you will have no lack of pleasures, or (c) iterative extension into
the future, e.g. bi storma gehwylc aswefed [Phoenix 1856] every storm is always
allayed (i.e. on all occasions of the flight of the Phoenix, past and to come); eom
expresses a present state provided its continuance is not especially regarded, e.g.
wlitig is se wong [Phoenix 7] the plain is beautiful. (Campbell 1959:350)
The DOE (s.v. bon) too lists three different types of usage distinctions for the
s- and the b-paradigm, (a) present vs. future, (b) statal vs. actional, and (c) non-
durative vs. durative.9 The distinction did not last in the history of English.
Rather, it is abandoned in early Middle English, earlier in Northern than in
Southern and Southwestern texts (see MED s.v. bn, OED s.v. be, Jost 1909:139f;
Brunner 1962:2779). (Lutz 2009:233, Note 19).
4.1.2 Two copulas in Celtic

This un-Germanic twofold paradigm for the copula was explained as a contact
phenomenon as early as 1925, when Keller pointed to the formal similarity of
the bo-paradigm with the b-paradigm of the Celtic languages, for which see the
Middle Welsh paradigm in Table 4.
Whereas the Celtic s-paradigm rarely shows its s anymore, owing to intense
phonological change, the b-paradigm is clearly recognizable as such in all Insular
. Cf. also Bolze (this volume).

Table 4. Two copulas in Middle Welsh

s-paradigm b-paradigm
wyf bydaf (I) am

wyt bydy (thou) art
yw byd (he/she/it) is
ym bydwn (we) are

ywch bydwch (you) are
ynt bydant (they) are
Celtic languages. As to the origin and meaning of the Celtic paradigms, Lewis
and Pedersen state:
The paradigm of the verb to be consists in Italo-Celtic of forms of the roots *es-
and *bheu-. In Celtic a pres[ent] stem *bhw-, *bhwije-, derived from *bheu-, also
appears. This latter present denotes either a praesens consuetudinale or a future, a
natural development from an orig[inal] meaning to become (Lat. f). The same
root is also used in the subjunctive. (Lewis & Pedersen 1989: 476.1)
The paradigms added to the inherited s-paradigm in Old English and in Middle
Welsh are remarkably similar both as to form and to meaning, cf. Table 5.
Table 5. Comparison of the Middle Welsh and the Old English b-paradigms
Middle Welsh Old English
b-paradigm b-paradigm
bydaf bo (I) am
bydy bist (thou) art
byd [bi] bi he/she/it) is
bydwn bo (we) are

bydwch (you) are
bydant (they) are
All forms in the paradigm of both languages begin with a b- followed

by a front vowel; note especially the phonetic identity of the most frequently
used forms, the 3rd person singular. And the meanings formulated by the
specialists (a) an invariable fact or (b) the future or (c) iterative exten-
sion into the future in Old English and a praesens consuetudinale or a future
Theo Vennemann
for Celtic are close enough to invite the idea that the innovations did not arise
independently.10
Kellers 1925 proposal was published in German, as was Preusler 1956 where
numerous Celtic features of English, including the b-paradigm, are discussed.
Flasdieck (1937), likewise publishing in German, thought that the preservation
of the b-forms in England may have been furthered by the contact with the Celtic
population.11 But he rejects Kellers idea that the Old English forms were taken
over from Brittonic, viz. that they originated in the manner of thinking and speak-
ing of Anglicized Britons.12 The reasons for his assumption of support but not
carry-over from Celtic are, however, not very convincing.13
. Ahlqvist (2010), in his section To be in Celtic and English (pp. 5058), lists all the rel-
evant paradigms of the Insular Celtic languages (pp. 5355), which show that certain forms of
the Irish b-paradigm are even more similar to the Old English one. This becomes interesting
especially in the context of Schrijvers view that the Celtic contact language of Anglo-Saxon
was an old form of Irish rather than of Welsh (Schrijver 2009:208): The original language
of the shifting population [shifting from Celtic to Anglo-Saxon] can now be identified as
a variety of Celtic which was ancestral to Old Irish. E.g. the Old Irish 1st person singular
consuetudinal present indicative is buu in the absolute andbu in the conjunct paradigm
(Ahlqvist 2010:55). Trudgill (2010:2527), in his socio-linguistic perspective, approvingly
cites Ahlqvist 2010, as well as other authors relating the Old English and Celtic b-paradigms
as a contact phenomenon.
. In der neuen Heimat mag die Erhaltung berdies gefrdert worden sein durch das
Zusammentreffen mit der kelt[ischen] Bevlkerung. (Flasdieck 1937:332f.) [In the new
homeland, the preservation (of the b-forms) may furthermore have been supported by the
encounter with the Celtic population.].
. Die altenglischen Formen und Funktionen der Wurzel *bheu, die den anderen ger-
manischen Dialekten fremd sind, entstanden im Munde und im Denken von englisch sprech-
enden Briten. (Keller 1925:60) [The Old English forms and functions of the root *bheu, which
do not occur in the other Germanic dialects, originated in the mouths and the thinking of
English-speaking Britons.] Using Kellers own words, Flasdieck replied: Indes wird man nicht
die a[lt]e[nglischen] Formen mit Keller als Entlehnung aus dem Brit[ischen], entstanden im
Denken und Sprechen von englisch redenden Briten, ansehen drfen. (Flasdieck 1937:333)
[However, one will not be allowed to consider the Old English forms with Keller as borrow-
ings from British, arising from the thinking and parlance of English-speaking Britons.] By
contrast, Preusler (1956:324) cites Kellers statement approvingly (m.e. mit recht [in my
opinion with justification]).
. Auch das A[lt]s[chsische] kennt die lngere Erhaltung von *bheu; der Abbau von *bheu
erfolgt gerade im Norden Englands; berdies fehlt dem A[lt]e[nglischen] ein Prt[eritum] zu
*bheu, wie es das Kelt[ische] kennt. (Flasdieck 1937:333) [In Old Saxon too, *bheu was pre-
served longer; it was in the north of England where the decline of *bheu happened. Besides,
Old English lacks a preterite of *bheu, which Celtic has.].
In any event, Keller (1925), Flasdieck (1937), and Preusler (1956) were three
authors reckoning with Celtic influence in the development of the Anglo-Saxon
b-paradigm, all three writing in German. The earliest proposal in English I have
found is Tolkiens (1963:3032).14 Tolkien refers to none of his German predeces-
sors. As it happened, he himself did not fare much better: [Tolkiens account]
seems not to have been, either noticed very much, or, perhaps, rather, not taken
very seriously by English language scholars. (Ahlqvist 2010:52).
As is well known, the continental West Germanic languages too have b-forms
in their copular paradigm; see Table 6. But what they do not have is two distinct
copular paradigms, such as the Old English ones; nor do we possess any evidence
that there ever were two distinct paradigms.
Table 6. A single copula in Continental West Germanic, combining s- and b-forms

Frisian Old Saxon Old High German
bim bium bim

bist bist
is is ist
sind sind(um) birum

birut
sint
There are two interpretations of this situation. Schumacher (2007) sees the
b-forms in Frisian and German as evidence for a separate, contact-induced prehis-
toric pan-West Germanic b-paradigm of Celtic origin which was conflated with
the inherited s-paradigm before the earliest Frisian and German attestation but
preserved in Anglo-Saxon owing to the continued contact with Celtic. This view is
considered unlikely by Lutz (2009). She assumes borrowing of individual b-forms
into the pre-West Germanic s-paradigms, arguing as follows: C ontact-induced
grammatical categories do not result from borrowing but through language-
shifting; therefore Schumachers assumption that a second copular paradigm
was transferred from Celtic into pre-West Germanic presupposes that the West
Germanic peoples are for the most part Celts who learned Germanic. But there is
. I had planned to quote at some length from Tolkiens excellent article. This was rendered
superfluous by Ahlqvist 2010 where Tolkiens entire account of the b-paradigm, including
footnotes, is reproduced verbatim on pp. 50f.
Theo Vennemann
no independent evidence for this to be true except of course in the case of the
English. On the contrary: the evidence we have for prehistoric Celtic-Germanic
contacts lexical evidence has been interpreted as pointing toward a dominant
position of the Celts in relation to the early Germans. Thus Schuhmachers view
cannot be maintained, unless one overturns the traditional view of the relation-
ship between the Celtic and Germanic populations and assumes language shift-
ing instead of, or besides, borrowing.
I need not take sides in this controversy here. But I would like to contribute
one argument possibly useful in resolving this issue that neither Schuhmacher
nor Lutz took into consideration: English and Celtic share other syntactic prop-
erties which have been interpreted as transfers from the Celtic substrate to the
Anglo-Saxon superstrate in language shifting. But all of these only unfold in the
history of the language, reaching some sort of stable grammatical status only in
Middle or Early Modern English.15 Not so the double copular paradigm: It is
fully established in Old English from the very beginning and in all dialects; and
it is the only Anglo-Saxon morpho-syntactic feature traced to Celtic that has this
property.
One way to interpret this singular feature is to assume that it belonged to
Anglo-Saxon from the beginning of its existence, i.e. that the Anglo-Saxons
brought it to the Isles from the Continent. The simplest way to embed this
hypothesis is to assume that Continental West Germanic possessed the double
copular paradigm at the time of the invasion of Britain by several of those West
Germanic tribes in the middle of the 5th century. Those West Germanic tribes
staying on the Continent then conflated the two copular paradigms into a single
one when the contact with Celtic faded and eventually stopped, so that when
their languages were first committed to writing they only each possessed the one
mixed paradigm each shown in Table 6. By contrast, the West Germanic tribes
leaving for Britain renewed contact with Celtic in their new environments, thus
stabilizing and preserving their dual copular paradigm for several centuries,
until their contact with Celtic too faded and they too conflated the two copular
paradigms into a single one. E.g. the paradigm resulting in the standard lan-
guages consists mostly of the Germanic s-paradigm forms but conspicuously
contains three b-forms, the infinitive and imperative be, the present participle
and gerund being, and the past participle been.16 This interpretation combines
elements of both Schuhmachers and Lutzs lines of argumentation: It assumes
. Cf. Section 3.2 below.

. Stephen Laker informs me that some dialects settled only on the b-forms, especially in
the southwest.
the two copular paradigms for all prehistoric West Germanic languages and
accepts the consequence that the West Germanic population consists to a con-
siderable extent of Celts who shifted from Celtic to Germanic. West Germanic is
Germanic in the mouths of Celts.17
The Frisian and German b-forms are important for yet a very different
reason: They show that not only Insular Celtic but also Continental Celtic had
b-forms, and by implication that the Celtic separate b-paradigm already existed,
and hence originated, on the Continent.18 It did not originate in the Isles but
was taken there by those Celts who left the Continent for the Isles. This has
no further importance for the question we are discussing here, which is why
Old English and Old High German differ with regard to the copular category:
That question is answered both by Schuhmachers and by Lutzs interpretation,
namely by reference to the specific substratal Celtic influence which English was
exposed to in the Isles and which all West Germanic languages remaining on the
Continent, including Old High G erman, were exempt from. However, the con-
clusion that Celtic already possessed the double copular paradigm on the Con-
tinent is important for the question which arises next, namely how the Celtic
languages themselves acquired the double paradigm. The answer published in
my 2010 handbook article is: Celtic acquired it in the same way that Anglo-Saxon
acquired it, namely on its pre-Celtic, pre-Indo-European substrate, as suggested
by the well-known fact that not only Celtic but also all western Romance lan-
guages had once, or still have, two copular paradigms (best known, of course,
from Spanish where the correct use of ser and estar is one of the great hurdles for
second language learners). See Table 7.
. The Celts referred to here were, of course, speakers of Continental Celtic, which was
an Indo-European language developed on a Vasconic substratum (cf. Vennemann 2010 and
references given there). The Celtic languages of the Isles, Insular Celtic, were very dif-
ferent, having further developed on the pre-Celtic Hamito-Semitic substratum of the Isles,
according to the theory of Morris Jones 1900; Pokorny 19271930; Gensler 1993 and others,
cf. several of the chapters of Vennemann 2003. Thus the Celtic implied in my mnemonic
sentences West Germanic is Germanic in the mouths of Celts and English is Coastal West
Germanic in the mouths of Celts (Vennemann 2004b) must be understood accordingly for
these formulas to make sense, with Continental Celts in the first sentence and Insular Celts
in the second.
. The actual attestation is meager, but forms with the basic meaning of be do seem to have
occurred in Gaulish inscriptions, cf. Lambert 1997: buetid in deuuorbuetid, compos du verbe
etre (p. 146), bue = irl. bes, qui serait, subj. prs. 3sg. *buet + s(e), particule relative (p. 67);
also p. 63, in particular 3rd sing. future bissiet il sera.
Theo Vennemann
Table 7. Romance languages with two copulas

Portuguese
Galician
Castilian
Catalan
Italian
Sicilian
Old French
not Romanian
western Romance languages
Thus, the two-copula syntax is a feature of those western Indo-European

languages see Table 8 for which a Vasconic substrate is likely for independent
reasons.
Table 8. Languages with two copulas in Europe

West Germanic
(Western) Celtic
Western Romance
Languages with Vasconic substrate
Basque has two copulas, izan and egon, which are used in similar ways as
Spanish ser and estar. E.g. Etxepare (2003) begins the section Copular construc-
tions with the following description:
Basque makes a distinction between stage-level predications (those which
attribute some transitory property to the subject of predication) and individual-
level predications (those which attribute some standing property to the subject
of predication) in the auxiliary selected to express them. Transient properties
are assigned by the verb egon be in a location, whereas standing properties are
assigned through the verb izan be. The distinction is reminiscent of the one
found in Spanish between ser and estar Izan is also used in equative sentences.
(Etxepare 2003:365)
de Azkue (1984: s.vv.) provides, besides French tre for both, the following Spanish
equivalents, each as the first (10) of several translations, where (c) indicates that
the item is comn toda la lengua:
IZAN 10 (c) ser
EGON 10 (c) estar
Thus it is quite conceivable that this non-Indo-European typological trait was

imported into West Indo-European by the Old European language shifters shift-
ing from Vasconic to Indo-European.
In concluding this section, I would like to point to the fact that there exists
very fine indirect confirmation for the substrate account: Whereas Standard
English gave up the Old English double copula syntax by conflating the two copu-
lar paradigms into one, Irish English has in recent centuries developed new dou-
ble copula paradigms on its Celtic substratum in Ireland, see Hickey 2007:141,
173, 213237.
4.2 Other categorial differences

I have dealt with the double copula paradigm at some length because it is peculiar
among the categorial features differentiating the Old English verbal system from
the Old High German one.
One property which sets this feature aside is the fact that it has not survived
into Modern English, for which I have no explanation, especially since it has
remained fully alive in Celtic, as well as in the western Romance languages except
French. Another property that sets the double copula paradigm feature aside
among the features attributed to Celtic influence is the fact that it is the only one
that is fully fledged even in the earliest Old English. This has been remarked on, at
least in comparison with one other such feature, by Filppula, Klemola, and Paulasto
(2008:60): There is a clear chronological difference between the emergence of the
OE is/bi distinction [i.e. the emergence of the double copula syntax, T.V.], which
is already found in OE literary sources, and the rise of the P[rogressive] F[orm] in
M[iddle] E[nglish].
But what really makes the double copula paradigm feature special is the fact
that it is the only feature attributed to Celtic influence that can be traced to the
Continent and there, at least in my analysis, to pre-Indo-European Old European,
with the same categorial differentiation found in Basque. All other categorial
Old English innovations attributed to Celtic influence are Insular Celtic features,
i.e. such features that have parallels not in Continental Old European but in the
Semitic languages and must therefore be assumed to have originated in the Celtic
Isles. The following grammatical features of English are those considered likely
to be of Celtic origin by Filppula, Klemola, and Paulasto (2008):
1. The internal vs. external possessor construction (pp. 3040)

2. The Old English distinction between *es- and *bheu-forms of the verb be
(pp.4042)
3. The Northern Subject Rule (pp. 4249)
Theo Vennemann
4. Periphrastic do (pp. 4959)

5. The progressive (or -ing) form (5972)
6. The cleft construction (pp. 7284)
7. Relative clause structure (pp. 8494)
8. Other grammatical features with possible Celtic origin (pp. 95118)
8.1. The development of self-forms as intensifiers and reflexives (pp. 9597)
8.2. Comparative nor (pp. 98102)
8.3. The Cumbric score (pp. 102105)
8.4. Pronoun exchange and other related phenomena (pp. 105118)
I will briefly look at them in this order, even though not all of them fall clearly
under the heading of verbal categories.
Ad 1. The internal vs. external possessor construction

English has lost nearly all dative constructions except those governed by three-
place verbs such as to give, She gave him some money. In particular, it lost the sym-
pathetic dative of the external possessor construction, i.e. the dative in locutions
such as (1) and (2), where German still uses the possessor-phrase external dative
but English has to use the possessor-phrase internal genitive.
(1) Dann stach ihm jemand die Augen aus.
then gouged him someone the eyes out
DATIVE
Then someone gouged out his eyes.
GENITIVE
(2) Dann wurde Balders Pferd das/sein Bein verrenkt.
Then was to Balders horse the/its leg wrenched
DATIVE
Then the leg of Balders horse was wrenched.19
GENITIVE
Old English, by contrast, did have the external possessor construction with the
sympathetic dative, as in (3), and so did, of course, Old High German, cf. (4),
together with all Continental European languages.
(3) a sticode him mon a eagan ut20
then gouged him someone the eyes out
DATIVE
Then someone gauged out his eyes.
. I.e., Then Balders horse sprained its leg.

. Orosius, late 9th century.
(4) du uuart demo balderes uolon sin uuoz birenkit21

then was to Balders horse its leg wrenched
DATIVE
Then Balders horse sprained its leg.
English lost the construction in the Middle English period: This construc-
tion, common in OE , is comparatively infrequent in ME and loses ground
steadily. (Mustanoja 1960:98) Following the lead of Pokorny (192730: 16.253)
in Vennemann 2002, I explained the loss of the external possessor construction in
English by the language shift of the Insular Celtic substrate speakers, who did not
have the external possessor construction, and the loss of the external possessor
construction in Insular Celtic by the language shift of the pre-Celtic Semitic sub-
strate speakers of the Isles, who did not have the external possessor construction
either; no ancient Semitic language had it. A related topic is the excessive use
English makes of possession marking, as in (5).
(5) Peter put his hands in his pockets.
Not: Peter put the hands in the pockets.
This is completely un-Germanic, cf. the German equivalent of (5) in (6).

Peter steckte die Hnde in die Tasche. ??Peter steckte seine Hnde in seine
(6)
Tasche.
But it is both Insular Celtic and Semitic. Returning to the list of grammatical
features of English considered likely by Filppula, Klemola, and Paulasto (2008)
to be of Celtic origin, we can skip point 2 because we have discussed it at length.
Ad 3. The Northern Subject Rule

This special subject-verb agreement rule of the Northern and North-Midland dia-
lects, fully established when attested in Middle English, was undoubtedly a dialect
feature of oral Old English. It too has been attributed to Insular Celtic influence,
cf. Filppula, Klemola, and Paulasto 2008:4649, and further to Semitic influence,
cf. my 2001 paper.
Ad 4. Periphrastic do
Periphrastic do is first attested in Middle English and only develops into a gram-
matical system in Modern English. Even though the evidence for a Celtic origin
of the construction is strong (cf. Filppula, Klemola & Paulasto 2008:59) and its
. Second Merseburg charm, 9th or 10th century. The possessive genitive sin its is here
used redundantly, as is also possible in Modern German; der the would be equally gram-
matical (and, indeed, better style).
Theo Vennemann
occurrence at least in oral Old English is predicted by the theory, I will not treat it
here where the focus is on Old English.
Ad 5. The progressive (or -ing) form

The progressive based on the verbal noun (terminating in -unge or -inge), unique
in the Germanic world, is not a feature of Old English but only of Middle and
Modern English. What is found in Old English is a construction with the present
participle terminating in -ande or -ende. However, the latter construction is not
specifically English:
The origin of the E[xpanded] F[orm] in Old English can be explained in several
ways. Of special interest are constructions with a predicative adjective and
with an appositive participle, , and the influence of periphrastic locutions
resembling the E[xpanded] F[orm]. These factors, however, were also present in
other Germanic languages in which the E[xpanded] F[orm] was not developed
and thus do not explain the unique position of Old English in this respect.
(Nickel 1966:392)
More specifically, Aronstein said, nearly a hundred years ago, with reference to the
combination of wesan/bon and the present participle in -ende:
Die verbindung der verba des seins mit dem part[izip] praes[ens] findet sich in
allen indoeuropischen sprachen. [] Im Mittelhochdeutschen ist sie nicht selten
bis in das 15. jahrh[undert] und verschwindet dann mehr und mehr. [] Im
Neuhochdeutschen ist der gebrauch ausgestorben. (Aronstein 1918:5f.)
[The combination of the verbs of being with the present participle is found in all
Indo-European languages. [] It is not infrequent in Middle High German until
the 15th century but then disappears more and more. [] Its use dies out in New
High German.]
Aronstein explains the remarkable fact of its survival in English by reference to a

development specific to this language:
Im Englischen dagegen ist sie nicht blo aus der lteren sprache in die neuere
bergegangen, sondern hat eine groe erweiterung erfahren, sowohl was die
hufigkeit als den umfang des gebrauchs angeht. Der uere anla hierzu liegt
offenbar darin, da im Englischen wohl durch den einflu des Franzsischen, wo
part. pres. und gerundium zusammenfielen, gegen ende des 12. jahrh. das a[lt]
e[nglische] part. prs. auf -ende, m[ittel]e[nglisch] -inde oder -ende zuerst in den
sdlichen mundarten und allmhlich auch im norden durch die form auf -ing
(ae. -ung) verdrngt worden ist, die zugleich die endung des verbalsubstantivs
war und dem deutschen -ung in teilung, wirkung etc. entspricht. So wren also
verschiedene funktionen in einer form vereinigt worden, die auf diese weise eine
besondere bedeutung erlangt htte. (Aronstein 1918:5f.)
[In English, by contrast, it did not merely pass on from the older into the more
recent language but was greatly expanded, both in its frequency and the scope of
its use. The external cause for this seems to have been the influence of French,
where the present participle and the gerund were the same: Toward the end of the
12th century, beginning in the southern dialects but gradually spreading to the
north, the Old English present participle in -ende, Middle English -inde or -ende,
was ousted by the form in -ing (Old English -ung), which was at the same time
the termination of the verbal noun and which corresponds to the German -ung in
teilung, wirkung etc. In this way different functions were united in a single form,
which thus attained special significance.]
This account is correct in its descriptive part. But the ascription of the loss of the
present participle form in -inde/-ende to French influence, namely to the fact that
in French the present participle in -ant (e.g. achetant buying) was also used as a
grondif (e.g. en achetant in/by/while buying), is less convincing. Together with
Preusler (1956:327331, 331334), Filppula, Klemola, and Paulasto (2008:5972)
and others, I consider Celtic influence more likely:
In my view the essential English innovation consists in the victory of the Celtic-
motivated verbal noun construction (suffix -ung/-ing) over the Anglo-Saxon
present participle construction (suffix -inde/-ande), where even the frequent use
of the latter may have been provoked by attempts to integrate the Celtic aspect
into English. (Vennemann 2001:355)
Even though the frequency of use of the Expanded Form with the present parti-
ciple in Old English may have been a consequence of the Celtic language shifters
looking for an equivalent of their own progressive forms, the origin of the form
itself should not be sought in Celtic, for the reason given further above and for yet
another reason: Insular Celtic does not possess a present participle.
Though more frequent in Old English than in the other Germanic languages,
even the Expanded Form with the present participle, as a forerunner of the Middle
English Expanded Form with the verbal noun, does not appear to have been gram-
maticalised, as Nickel writes in his book on this topic: Within the Old English
verbal system the E[xpanded] F[orm] is a locution still in process of development.
(Nickel 1966:391).
Whereas I consider a Celtic origin of the English Expanded Form with the
verbal noun ascertained,22 the source of the Celtic construction itself is not so
. The origin of the English progressive in the contact with Celtic is one of the earliest and
certainly the most frequently cited example of this line of research. Cf. most recently Ahlqvist
2010:5863; Trudgill 2010:28f.
Theo Vennemann
clear. It may have been a substratal feature because at least Old Egyptian has simi-
lar constructions (cf. Vennemann 2001:355).
Ad 6. The cleft construction

Clefting barely begins in Old English, only gaining momentum in Middle English
and further down to the present language. Filppula, Klemola, and Paulasto
(2008:83) attribute this development to Celtic influence. Clefting is also very
common in Egyptian (Coptic, see Reintges 2003) and common enough in Semitic
languages for Huehnergard and Pat-El (2007:338) to tentatively attribute it to
Proto-Semitic.
Ad 7. Relative clause structure

Contact clauses, i.e. relative clauses without a relative pronoun, and related phe-
nomena such as preposition stranding, though occasionally occurring in Old Eng-
lish, become a regular feature of the language much later. Even if this development
was subject to Celtic influence (which is debated, cf. Filppula, Klemola & Paulasto
2008:9094), it therefore remains outside the scope of this paper. Certain similari-
ties between Semitic and Celtic relative clause constructions need not therefore be
discussed either.
Ad 8. Other grammatical features with possible Celtic origin

The phenomena discussed in this section do not fall under the heading of verbal
categories.
5. Conclusion
In this paper I looked at a number of verbal categories of Old English and Old
High German with an eye to the role of language contact in their development.
In the first section, the remarkable fact that the Proto-Indo-European system
of TAM categories was reduced to half its size in Germanic was discussed within
the theory of prehistoric language contacts between pre-Germanic and Phoeni-
cian (Punic, Carthaginian). Tabulating the categories of Proto-Indo-European,
Proto-Germanic, and Phoenician according to Mailhammers explanatory pro-
posal of 2006, shows the Phoenician system acting like a filter in the transmission
of the Proto-Indo-European categories to Proto-Germanic: Only those categories
that were apprehended by the Phoenician language shifters on the basis of their
own native language were learned by them; since this language variety was spo-
ken by the former Phoenician superstrate speakers and thus possessed prestige, it
became the basis of the developing Proto-Germanic language.
In the second section, I briefly mentioned those categories that were inno-
vated in both languages by using nominal/adjectival parts of verbal paradigms as
predicates with copulas or quasi-copular verbs, fully grammaticalising into the
future and the perfect as well as the passive voice only in the later periods.
In the third section, I looked at a selection of verbal category features that
were innovated in Old English, thus differentiating this language from Old High
German, namely only those verbal category features for which a language-contact
explanation has been offered. Of those only a single one had been grammaticalised
before the beginning of the Old English period, the double copular paradigm, the
only such feature for which an origin on the Continent and in Vasconic has been
proposed. All other categorial features, all of them with an origin in the Isles and
traceable to a Semitic source, only fully grammaticalised in Middle or Modern
English, although they probably were involved in processes of grammaticalisation
already in Old English.
References
Ahlqvist, Anders. 2010. Early English and Celtic. The Australian Celtic Journal 9: 4373.
Aronstein, Philip. 1918. Die periphrastische Form im Englischen. Anglia 42(NF 30): 184.
de Azkue, Resurreccin Mara. 1984. Diccionario vasco espaol francs. 1st edn. 1905, 2nd
edn. (reprint) 1969. Bilbao: Euskaltzaindia. (Reprint).
Beekes, Robert S. P. 1995. Comparative Indo-European linguistics: An introduction. Amsterdam:
John Benjamins.
Besch, Werner & Wolf, Norbert Richard. 2009. Geschichte der deutschen Sprache: Lngsschnitte
Zeitstufen Linguistische Studien [Grundlagen der Germanistik 47]. Berlin: Erich Schmidt.
Blake, Norman Francis. 1996. A history of the English language. Houndmills: Macmillan Press.
Brunner, Karl. 1962. Die englische Sprache: Ihre geschichtliche Entwicklung, Vol. 2: Die Flexions-
formen und ihre Verwendung, 2nd edn. Tbingen: Max Niemeyer.
Campbell, Alistair. 1959. Old English grammar. Oxford: Oxford University Press.
[DOE] Cameron, Angus, Amos, Ashley C. & di Paolo Healey, Antonette (eds). 1986-. Dictionary
of Old English. Toronto: University of Toronto Press.
Etxepare, Ricardo. 2003. Valency and argument structure in the Basque verb. In A Grammar of
Basque [Mouton Grammar Library 26], Jos Ignacio Hualde & Jon Ortiz de Urbina (eds),
363426. Berlin: Mouton de Gruyter.
Filppula, Markku, Klemola, Juhani & Paulasto, Heli. 2008. English and Celtic in Contact [Rout-
ledge Studies in Germanic Linguistics]. New York NY: Routledge.
Flasdieck, Hermann M. 1937. Das altgermanische Verbum Substantivum unter besonderer
Bercksichtigung des Altenglischen. Englische Studien 71: 321349. (With a Nachschrift,
Englische Studien 72(1938): 158160).
Gensler, Orin David. 1993. A Typological Evaluation of Celtic/Hamito-Semitic Syntactic Par-
allels. Ph.D. dissertation, University of California, Berkeley. (Available from University
Microfilms International, Ann Arbor, Michigan, no. 9407967).
Theo Vennemann
Hickey, Raymond. 2007. Irish English: History and Present-day Forms [Studies in English Lan-
guage]. Cambridge: CUP.
Huehnergard, John & Pat-El, Naama. 2007. Some aspects of the cleft in Semitic languages. In
Studies in Semitic and General Linguistics in Honor of Gideon Goldenberg [Alter Orient und
Altes Testament 334], Tali Bar & Eran Cohen (eds). Mnster: Ugarit-Verlag.
Jost, Karl. 1909. BEON und WESAN. Eine syntaktische Untersuchung [Anglistische Forschungen
26]. Heidelberg: Carl Winter.
Keller, Wolfgang. 1925. Keltisches im englischen Verbum. In Anglica: Untersuchungen zur
englischen Philologie [Festschrift fr Aloys Brandl], Vol. 1, 5566. Leipzig: Mayer &
Mller.
Krahe, Hans. 1967. Germanische Sprachwissenschaft, Vol. II: Formenlehre [Sammlung Gschen
780]. Berlin: Walter de Gruyter.
Lambert, Pierre-Yves. 1997. La langue gauloise: Description linguistique, commentaire
dinscriptions choisies [Collection des Hesprides]. Paris: ditions Errance.
Lewis, Henry & Pedersen, Holger. 1989. A Concise Comparative Celtic Grammar, 3rd edn, second
impression with the supplement of 1961 by Henry Lewis. [First edition 1937.] Gttingen:
Vandenhoeck and Ruprecht.
Lutz, Angelika. 2009. Celtic influence on Old English and West Germanic. In Special issue
on Re-evaluating the Celtic Hypothesis, Markku Filppula & Juhani Klemola (eds). English
Language and Linguistics 13(2): 227249.
Mailhammer, Robert. 2006. On the origin of the Germanic strong verb system. Sprachwissen-
schaft 31: 152.
Mailhammer, Robert. 2007. The Germanic Strong Verbs: Foundations and Development of a
New System [Trends in Linguistics: Studies and Monographs 183]. Berlin: Mouton de
Gruyter.
Mailhammer, Robert & Smirnova, Elena. this volume. Incipient grammaticalization: Sources of
passive constructions in Old High German and Old English, p. 4169.
[MED] Kurath, H. & Kuhn, Sherman M. 19522001. Middle English dictionary. Ann Arbor MI:
University of Michigan Press.
Meier-Brgger, Michael. 2010. Indogermanische Sprachwissenschaft. 9th ed. [de Gruyter Studi-
enbuch]. Berlin: Walter de Gruyter.
Morris Jones, John. 1900. Pre-Aryan syntax in Insular Celtic. In The Welsh people, John Rhys &
David Brynmor-Jones (eds), Appendix B, 617641. London: T. Fisher Unwin.
Mustanoja, Tauno F. 1960. A Middle English Syntax, Part 1: Parts of Speech [Mmoires de la
Socit Nophilologique de Helsinki 23], Helsinki: Socit Nophilologique.
Nickel, Gerhard. 1966. Die Expanded Form im Altenglischen: Vorkommen, Funktion, und
Herkunft der Umschreibung beon/wesan + Partizip Prsens [Kieler Beitrge zur Anglistik
und Amerikanistik 3]. Neumnster: Karl Wachholtz.
[OED] Simpson, John A. & Weiner, Edmund S. C. (eds). 1989. The Oxford English dictionary,
20 Vols., 2nd edn, reset with corrections, revisions, and additional vocabulary. Oxford:
Clarendon Press.
Pokorny, Julius. 19271930. Das nicht-indogermanische Substrat im Irischen. Zeitschrift fr
Celtische Philologie 16: 95144, 231266, 363394; 17: 373388; 18: 233248.
Preusler, Walther. 1956. Keltischer Einflu im Englischen. Revue des Langues Vivantes 22:
322350.
Reintges, Chris. 2003. The syntax and semantics of the Coptic cleft constructions. University
of Sussex Working Papers in Linguistics and English Language, LxWP4/03. https://www.
sussex.ac.uk/webteam/gateway/file.php?name=reintges-cleft.pdf&site=1 (6 August 2010).
Rupp, Heinz. 1956. Zum Passiv im Althochdeutschen. Beitrge zur Geschichte der deutschen
Sprache und Literatur 78: 265286.
Schrijver, Peter. 2009. Celtic influence on Old English: Phonological and phonetic evidence.
English Language and Linguistics 13: 193211.
Schumacher, Stefan. 2007. Die Deutschen und die Nachbarstmme: Lexikalische und struk-
turelle Sprachkontaktphnomene entlang der keltisch-germanischen bergangszone. In
Johann Kaspar Zeu im kultur- und sprachwissenschaftlichen Kontext (19. bis 21. Jahrhun-
dert), Kronach 21.7.23. 7. 2006, Hans Hablitzel & David Stifter (eds), 167207. Wien:
Praesens Verlag. [English version: Lexical and structural language-contact phenomena
along the Germano-Celtic transition zone. In Kelten am Rhein: Akten des dreizehnten
Internationalen Keltologiekongresses/Proceedings of the Thirteenth International Congress of
Celtic Studies, 23. bis 27. Juli 2007. in Bonn, Part 2: Philologie: Sprachen und Literaturen,
Stefan Zimmer (ed.) [Beihefte der Bonner Jahrbcher 58.2], 247266. Mainz am Rhein:
Philipp von Zabern, 2009.].
Seebold, Elmar. 1970. Vergleichendes und etymologisches Wrterbuch der germanischen starken
Verben [Janua Linguarum, Series Practica 85]. Den Haag: Mouton.
Sihler, Andrew L. 1995. New Comparative Grammar of Greek and Latin. Oxford: OUP.
Tolkien, John Ronald Reuel. 1963. English and Welsh. In Angles and Britons [ODonnell
Lectures], 141. Cardiff: University of Wales Press.
Trudgill, Peter. 2010. Investigations in Sociohistorical Linguistics: Stories of Colonisation and
Contact. Cambridge: CUP.
Vennemann, Theo. 1998. Andromeda and the apples of the Hesperides. In Proceedings of the
Ninth Annual UCLA Indo-European Conference, Los Angeles, May 23, 24, 1997. [Journal
of Indo-European Studies Monograph Series 28], Karlene Jones-Bley, Angela Della Volpe,
Miriam Robbins Dexter & Martin E. Huld (eds), 168. Washington, DC: Institute for the
Study of Man. (Reprinted as Chapter18 in Vennemann 2003).
Vennemann, Theo. 2000. Zur Entstehung des Germanischen. Sprachwissenschaft 25: 233269.
(Reprinted as Chapter1 in Vennemann 2012).
Vennemann, Theo. 2001. Atlantis Semitica: Structural contact features in Celtic and English.
In Historical Linguistics 1999. Selected Papers from the 14th International Conference on
Historical Linguistics, Vancouver, 913 August 1999. [Current Issues in Linguistic Theory
215], Laurel Brinton (ed.), 351369. Amsterdam: John Benjamins. (Reprinted as Chapter5
in Vennemann 2012).
Vennemann, Theo. 2002. On the rise of Celtic syntax in Middle English. In Middle English from
Tongue to Text: Selected Papers from the Third International Conference on Middle English:
Language and Text, held at Dublin, Ireland, 14 July 1999. [Studies in English Medieval
Language and Literature 4], Peter J. Lucas & Angela M. Lucas (eds), 203234. Bern: Peter
Lang. (Reprinted as Chapter10 in Vennemann 2012).
Vennemann, Theo. 2003. Europa Vasconica Europa Semitica [Trends in Linguistics: Studies
and Monographs 138], Patrizia Noel Aziz Hanna (ed.). Berlin: Mouton de Gruyter.
Vennemann, Theo. 2004a. Phol, Balder, and the birth of Germanic. In Etymologie, Entlehnungen
und Entwicklungen: Festschrift fr Jorma Koivulehto zum 70. Geburtstag, Irma Hyvrinen,
Theo Vennemann
Petri Kallio & Jarmo Korhonen (eds), 439457. Helsinki: Socit Nophilologique.
(Reprinted as Chapter19 in Vennemann 2012).
Vennemann, Theo. 2004b. Sprachgeburt durch Sprachkontakt: Die Entstehung des Englischen.
In Sprachtod und Sprachgeburt [Mnchner Forschungen zur historischen Sprachwis-
senschaft 2], Peter Schrijver & Peter-Arnold Mumm (eds), 2156. Bremen: Hempen.
(Reprinted as Chapter 18 in Vennemann 2012)
Vennemann, Theo. 2010. Contact and prehistory: The Indo-European Northwest. In The Hand-
book of Language Contact [Blackwell Handbooks in Linguistics], Raymond Hickey (ed.),
380405. Malden MA: Wiley-Blackwell.
Vennemann, Theo. 2012. Germania Semitica [Trends in Linguistics: Studies and Monographs
259], Patrizia Noel Aziz Hanna (ed.). Berlin: De Gruyter Mouton.
Index
Terms for Diewald, Kahlas-Tarkka, Wischer: Comparative Studies in Early Germanic Languages
A mutative 4344, 46, 51, auxiliarisation 11, 77, 129, 140,

a- (OE) 235260 293294 142, 149, 205, 209
ablaut 18, 291, 293 telic/atelic 48, 52, 110,
generalisation and 113116, 248, 250251 B
functionalisation of 293 terminative/ backgrounding 8, 88, 90
ACI construction 111, 170, non-terminative 51 Basque 302303
171176, 184185, 189191 transformative 4546, 209 b-copula 41, 4849, 295
accusative 2122, 24, 42, 49, Aldred 219, 221, 223, 225, 229, see also be; beon (OE);
109, 114, 117, 156159, 161, 231233 beo-paradigm
163, 170, 175, 183, 186, American English 148 be 25, 1013, 18n., 2324,
189191 analogy 191 4142, 48, 5457, 5963,
accusative/non-accusative Anglo-Saxon 129130, 133134, 6667, 7172, 81, 83,
object 109, 114, 156, 187, 298301, 307 164, 198199, 201202,
158159, 161 aorist 23, 33, 291292, 295n. 208212, 217233, 264n.,
see also object Armenian 169, 291 269, 275276, 280n.,
active 3, 17, 19, 2128, 30, aspect 23, 9, 13, 4143, 4647, 294n., 295297, 298n.,
3234, 37, 4245, 52, 56n., 50, 5253, 55, 58, 6667, 300, 301n., 302303
57, 65, 102, 123, 224 140, 223227, 235260, benefactive 153
adjective 42, 48, 5152, 55, 276, 289290, 294, 307 beon (OE) 196, 198, 201203,
6566, 77, 133, 183, 256, grammatical 237, 251 205, 208, 210211
294, 306 inchoative/ingressive see also b-copula
adverb 60, 81, 8384, 158n., 209, 256 bo-paradigm 296
163, 175, 192, 223, 264265, instantaneous 252, 254 see also b-copula
271, 273, 275, 277279, 281 perfective/imperfective 13, borrowing 131, 134, 142,
see also sentence adverb 46, 235, 240, 245246, 298300
durative 75 248254, 258259, 290 boundedness 5253, 59, 7197
intensifying 181 progressive 47, 60, 79, Brittonic 298
locative 13, 95, 263284 8183, 238, 248249,
temporal 60, 82, 84, 86, 251254, 259, 289, 304, C
9094, 9697 306307 Carthaginian 292293, 308
adverbials 8283, 85, 9295, auditory evidential Castilian 302
223, 241, 253255, 257, construction 5, Catalan 302
271272, 275 169170, 177 causation 101, 109, 115116, 121
see also time adverbials see also evidentiality causative 101124
spatial 95 augment 291 causee 101103, 109116, 121
agreement 158, 270, 293, 305 auxiliaries 34, 1011, 42, causer 101, 103, 109, 111112,
aktionsart 4345, 48, 5056, 48,6162, 6466, 7197, 114115, 121
5859, 62, 66, 237238, 117, 127, 129, 134135, Celtic 14, 117, 217, 227, 289,
240, 255, 294 137,139145, 147148, 294305, 307308
durative/non-durative 45, 152,162,166, 197198, Continental 301
5354, 7576, 83, 116, 218, 202n., 206, 209, 211, Insular 298, 301, 303,
223225, 227, 235, 255, 296 275276, 302 305,307
Index
Celtic influence 14, 117, 289, D existential construction 5, 9,

295, 299, 301, 303, 305, da 207, 269, 271272, 284 13, 84, 263284
307308 see also quasi-expletive da expletive 264, 266, 269273,
cleft construction 304, 308 da-construction, existential 275276, 279281, 283284
collocation 92, 132, 172 264, 271272, 283 extension 48, 144, 148, 176, 189,
collostructional analysis 95 dative 3, 3437, 49, 114, 156, 191, 267, 284, 296297
complement/complementation 158160, 166, 190, of meaning 264, 266
171192, 269 304305 of scope 76
infinitival 143 sympathetic 304 to new contexts 77, 84, 90,
nominal 151, 156159, 166 dative construction 304 266, 273
verbal 136 dative object 114, 156, 160
see also object complement; see also object F
subject complement decategorialisation 143, foregrounding 59, 82, 88, 90,
complement pattern 170, 173 148,267 93, 263
construction 25, 913, 15, deixis/deictic 263264, fossilised morpheme 13, 235
1938, 5061, 73, 76n., 269,271 French 75, 131, 251, 302303, 307
7981, 8384, 9193, 97, desemanticisation 48, 264, Frisian 8, 299, 301
101124, 128129, 136137, 266267, 274, 284 future/futurity 3, 9, 1213, 58,
143, 148, 154, 156, 157n., double copular paradigm 289, 121, 136137, 139140, 144,
160161, 169177, 180, 300301, 309 154, 217218, 220233
182185, 189192, 195, dual 289 second 293
197199, 205211, 224, dugan (OE) 127128, 146, 148 volitional/
230, 245, 250, 266268, dynamic 42, 47, 4953, 57, non-volitional 144145
293294, 302308 5960, 63, 109111, 114, see also sculan (OE);
see also cleft construction; 116, 120, 122123, 146, sculan (OHG); shall;
existential construction; 203, 271 wellan (OHG);
object construction; will; willan (OE);
passive construction; E werden-future
possessor construction, Early Modern English 7, future grams 4, 196, 198199,
internal vs. external 57, 82, 129, 131, 137141, 204205, 207, 211
constructional environment 144146, 152, 251, 295, 300 future marker 10, 135, 137, 144,
74, 7879, 86, 90, Egyptian 308 195212, 220
9697 English 1, 37, 914, 1718, future perfect 224, 293
contact clause 308 20, 3637, 4142, 5062,
context-sensitive 6567, 7197, 101, 106, G
grammaticalisation 127, 131, 139, 141, 147148, Galician 302
scenario 101102, 122 151152, 155, 164, 169171, Gaulish 301
Continental European 175, 182, 192, 195201, generic 56, 92, 181, 218, 223,
languages 304 203205, 207n., 235238, 225227
conventionalisation 66, 241, 246,259, 263266, genitive 34, 156, 158160, 166,
145,268 276, 283284, 290, 293, 190, 304, 305n.
conversational 295296, 298301, possessive 305
implicature 207, 268 303307 genitive object 156, 160
Coptic 308 see also American English; see also object
copula (verb) 2, 4, 10, 14, 16, Early Modern English; genre(s) 67, 62, 67, 78, 88
1819, 21, 29, 34, 3638, Irish English; Late Modern see also narrative;
4152, 5467, 72n., 77, English; Middle English; sermons
96, 209210, 275276, Old English; Scottish German 1, 36, 814, 17,
294303, 309 English 4142, 46, 4962, 64,
critical context 120, eom-paradigm 218, 296 6667, 7197, 130, 135n.,
122123, 273276, evidential/evidentiality 3, 9, 151152, 160, 169, 173, 182,
279281 12, 169192, 195 189, 192, 195201, 204205,
cwean (OE) 183 see also auditory 207209, 211, 242,
cyan (OE) 184 evidential construction 264265, 280, 283284,
Index
293294, 298299, 301, Hamito-Semitic substratum of 160164, 175, 177, 202,

303309 the Isles 301 209, 219221, 224233,
see also Middle High haitan/gahaitan (Goth) 4, 10, 242, 244245, 265266,
German; Modern 17n., 1920, 2238 277278
German; Old High htan (OE) 1931, 3738 lexicalisation 170, 188, 236, 259
German have 24, 910, 12, 102103, Lithuanian 294
Germanic 16, 89, 1213, 106107, 109, 114, 116, locative/locative meaning 13,
1719, 37, 42, 5254, 118119, 151152, 154n., 155, 190, 264, 268276, 278283
7274, 77, 88, 127, 130, 169, 157n., 160, 199, 209, 212
173, 180, 183, 190, 239n., have to 147148, 154 M
241246, 272, 291293, hearsay evidence 12, 169192 markedness theory 251
295296, 298300, (ge)heran (OE) 171192 maelian (OE) 184
305308 metaphorical extension
see also Proto-Germanic; I 144, 148
West-Germanic Icelandic 8, 17, 171 Mercian 107, 218
Germanic languages 12, 46, idiomatisation 270 Middle English 7, 12, 4749,
89, 12, 17, 19, 31, 3738, imperative 3, 28, 218219, 66, 7274, 76, 83, 88, 95,
75, 91, 130, 169192, 198, 290292, 295, 300 97, 102, 107, 127, 129131,
238239, 270, 291, 306307 Indo-Iranian 291 134139, 142146, 152, 171,
gerund 300, 307 inference(s) 145, 162, 170, 189, 198, 208, 236, 246,
Gospels 12, 184, 241, 242n. 177178, 268 259, 290, 295296, 300,
Latin 219, 224 infinitive 2, 4, 12, 20, 26, 28, 31, 305, 307309
Lindisfarne 217233 33, 52, 66, 109111, 122, 128, Middle High German 66,
West Saxon 217233 134, 135n., 138, 140141, 72,84, 97, 151153, 160,
Gothic 35, 810, 13, 1738, 42, 143, 146, 148, 151166, 189, 306
130, 171, 173, 177, 182183, 171172, 175176, 178182, Middle Welsh 227, 296297
235260, 294296 184185, 189192, 195, modal/modality 3, 9, 1113,
grammaticalisation 1, 35, 197n., 198, 202, 206208, 101102, 109111, 119122,
915, 4167, 7197, 211, 217219, 230, 258, 300 124, 127149, 151166,
101102, 106, 122123, 127, see also to-infinitive; 195, 199
129, 142146, 148149, zu-infinitive see also pre-modal;
151, 153, 159160, 165166, injunctive 291292 root modality
196198, 205, 236, 241, intensifier 235, 304 deontic 101102, 109111,
246, 259, 264268, 270, intentional meaning 206207, 119124, 136140, 144,
272275, 279, 282284, 209 146148
293, 309 Irish 298, 303, 310 epistemic 137139,
grammaticalisation path 12, Irish English 289, 303 144145, 148
106, 151, 196, 198199, isolating context 122, 268272 volitional 11, 109111,
211212, 265266, 270 Italian 251252, 302 119124, 199
grammaticalised 13, 4144, Italic 294 modal futures 198199,
4750, 54, 58, 60, 6566, Italo-Celtic 297 206208, 211
117, 143145, 147, 195, 198, iterative 60, 92, 132, 218, modal source 204, 206207
237, 240, 245, 250251, 223225, 227, 230, 296297 Modern/Present-Day
264, 267, 272, 282, 284, English 3, 7, 11, 13, 53,
294, 307, 309 J 55, 57, 73, 7981, 83,
Greek 1, 10, 2123, 3137, 129, jussive 110, 292 102, 109110, 113, 116,
154, 177, 183, 238n., 242, 120121, 127129, 131, 134,
291, 294 L 137141, 144148, 151152,
language shifting 289, 300 170, 173, 177, 196, 201,
H Late Modern English 208 208, 211, 235, 247249,
habban (OE) 3, 11, 101124 Latin 1, 7, 9, 1213, 27, 30, 67, 251252, 254, 256, 257n.,
haben 4, 12, 151166 72, 75, 84, 8788, 102, 258, 263, 266,268272,
habitual 5, 13, 5556, 60, 62, 105106, 110112, 115, 276, 294295, 300, 303,
157, 175, 218, 223, 225227 117, 129131, 151, 154155, 305306, 309
Index
Modern/New (High) object construction 156 passive construction 34,

German 3, 4647, 55, see also construction 1011, 24, 27, 30, 4167,
5758, 73, 76n., 7981, object-raising 18n. 7177, 82, 8688, 90, 92n.,
8485, 91, 152, 154, 156, obligation 12, 120, 136, 9697, 101103, 112, 198,
161, 195196, 242, 245, 264, 138140, 151152, 154, 203n., 218, 253
268272, 305306 158159, 162164, 166, 199, see also construction
mun/man (EModE) 137139, 207, 209 past perfect 25, 111, 293
142145 Old Egyptian 308 see also perfect; pluperfect
mun/man/maun 139142 Old English 1, 45, 78, 1014, patient 42, 49, 57, 65, 101103,
(ge)munan (OE) 127149 1738, 4167, 7179, 8195, 109115
munen/mun/mon (ME) 97, 101125, 127134, 137, perception 12, 169170,
134142 142143, 146148, 152, 154, 172185, 187, 189, 191192
must 138, 142, 146148, 158, 157, 160, 164, 169192, direct 169, 173177, 187,
161162, 169 195212, 217219, 222225, 189, 191
227, 229, 231, 235260, indirect 170, 177183
N 263284, 289309 see also verbs of perception
narrative 7273, 7880, 82, 85, Older Scots 147 perfect 2, 4, 14, 23, 26, 42, 48,
88, 90, 97 Old European 303 5253, 55, 102, 107111,
narrative action 7881, 90 Old French 131, 302 117118, 122123, 132, 138,
necessity 128, 136, 138140, Old High German 1, 4, 8, 10, 140, 155, 160, 224, 245, 259,
146147, 151152, 154, 1214, 4167, 7173, 7679, 290294, 309
158159, 162166 8492, 94, 97, 130, 151166, see also past perfect;
negation/negative 141, 147, 171, 173, 195212, 263284, pluperfect
206, 280n. 289309 perfectiviser 238239, 250, 255
nemnan (OE) 185 Old Icelandic 130, 134, 295 periphrastic do 304305
NICE properties 206 Old Indic 294 periphrastic expressions
nominative 2123, 34, Old Irish 227, 295, 298 147, 293
3637, 49 Old Norse 5, 8, 11, 127, 134, Phoenician 292293, 308
Northern Subject Rule 303, 142143, 171, 173, 294296 Phrygian 291
305 see also Scandinavian pluperfect 53, 93, 291293
Northumbrian 13, 218219, optative 152, 290292, 295 see also past perfect;
222223, 228229, Ormulum 86, 134, 137, 142143 perfect
231, 233 Orosius 2, 238, 266, 277, 304 Portuguese 302
Notker 12, 6365, 151, 155157, Otfrid 63, 73, 77, 8688, 9091, possession 12, 151, 154156,
159166, 282, 285 9396, 278282 158,199, 209, 305
nouniness squish 190 see also possessive
P possessive 108, 120, 154,
O participle 8, 21, 23, 4150, 162,191, 209n.
object 56, 107108, 156, 157n., 5265, 108, 117, 121122, see also possession;
166, 190n., 258,270 160, 177, 294, 300, 306307 possessive genitive;
concrete 158, 163 present participle 44, 218, possessor construction,
direct 2022, 24, 26, 28, 42, 300, 306307 internal vs. external
117, 128, 141142, 146, 172, past participle 24, 10, 21, possessor construction,
185, 248 2324, 27, 29, 35, 4165, internal vs. external
indirect 3, 159 101102, 107110, 113, 303305
nominal 157n. 115118, 121122, 160, see also construction
see also accusative/ 244246, 253, 294, 300 possibility 146, 152, 154155,
non-accusative object; particle 8485, 114, 141, 206, 191, 199
dative object; 236237, 244, 246, post-verbal particle 258
genitive object 257259 potential/potentiality 142,
object complement 20, passive (voice) 1730, 32, 34, 153154, 157159, 163,
108, 117 3638, 4350, 7197 165166
see also complement/ inherent 19, 27, 30 praesens consuetudinale 297
complementation stative 7677 prefix, verbal 235260
Index
pre-modal 10, 109, 128129, reccan (OE) 185 strong verbs 290
146, 149, 181 reduplication 291 subject 3, 1718, 21, 26, 2930,
see also modal/modality referential/non-referential 146, 42, 49, 51, 5657, 6062,
prepositional phrase 176, 271, 273, 281 65, 8082, 85, 91, 9596,
180181 reflexive 304 117118, 120123, 136137,
preposition stranding 308 relative clause 21, 24, 27, 37, 141, 144, 156, 158, 189,
present indicative 23, 36, 135, 181, 185, 304, 308 206n., 207, 209, 254, 263,
154, 164, 208, 217228, resultative 42, 4546, 5255, 264n., 265, 269273, 275,
230232, 295, 298 5758, 62, 65, 74, 132133, 278279, 281n., 282284,
see also present tense 246, 252, 256, 290 290n., 294, 302, 305
consuetudinal 298 rise of the verbal noun in dummy 276, 282
present subjunctive 23, 110, -ung/-ing 289 subject complement 21, 65, 275
208,217219, 222224, Romance languages 8, 270 see also complementation/
228233 Western 301303 complement
see also present tense Romanian 302 subjunctive 23, 26, 28,
present tense 2, 13, 26, 4749, root modality 145, 199 136137, 179, 291292, 297
5859, 6465, 78, 81, 132, see also modal/modality see also present
144, 198, 219, 221225, 227, subjunctive
232233, 245, 253, 290 S substrate/substratal 300303,
see also present indicative; Scandinavian 8, 75, 142 305, 308
present subjunctive see also Old Norse substrate influence 292
preterite tense 11, 2324, 26, s-copula 41, 46, 4852, 54, 295 superstrate 300, 308
29, 127130, 132134, 140, Scottish English 139140, 147 suppletive paradigm 201
142, 146149, 152, 245, 249, sculan (OE) 109, 120, 129, sw-parentheticals 181182
253, 290292, 298 204206, 208
preterite-present (verb) 11, sculan (OHG) 152, 158, T
127130, 132134, 140, 142, 161162, 196, 201, 204205 TAM 289291
146149, 152, 290 secgan (OE) 181182, 186189 TAM categories 290293, 308
preverb 241, 259 Second Merseburg charm 305 TAM system 291293
pronoun 22, 34, 82, 118, 136, sein 44, 73, 76, 157, 164166, Tatian 23, 51, 56, 7273, 77,
163, 225, 227, 265, 304 211, 298 8491, 9396, 153155, 157,
personal 289 sein-construction 271 160, 199, 201202, 209,
relative 3637, 308 self-forms 304 277278, 282
prospective 156159, 163, semantic bleaching 48, 76, telicity/atelicity 50, 101102,
165166, 275, 280281, 283 90,143, 246, 259, 110, 250251, 294
Proto-Germanic 5, 12, 17, 52, 266267, 279 see also aktionsart
127, 169171, 183, 242, semantic role 117, 123 thar was (OHG) 263284
289295, 308 Semitic 292293, 301, 303, 305, there 29, 95, 291
see also Germanic 308309 time adverbial 73, 7982,
Proto-Indo-European 3, 18, 41, sentence adverb 84, 169 86,9097, 254
5253, 289295, 308 see also adverb see also adverbial
Punic 293, 308 sermons 189, 204, 207 to-infinitive 128, 155
purpose/purposive 153154, see also genre(s) see also infinitive
292n. shall 134, 196, 198, 207208 Tolkien 299
Sicilian 302 topicalisation 91
Q sn (OHG) 41, 4346, 5052, transitivity 4143, 45,
quasi-expletive da 283 5457, 6365, 202, 294 50,5455, 59, 61,
see also da Slavic 236, 238240, 242, 6667, 294
251,294 r ws (OE) 263284
R Spanish 151, 251252, 301302 t-clause 178180
rdan (OE) 186 sprecan (OE) 185 urfan (OE) 127, 146, 148
reanalysis 144, 152, 169, 190, stative 42, 4447, 5055, 5758,
276 6263, 83, 8990, 109, 116, U
semantic 274 118, 120, 122, 157 untypical context 272279
Index
V werdan (OHG) 41, 4346, will 134135, 137,

Vasconic substrate 302 5052, 54, 5764, 66, 142, 144, 196, 198,
vector verb 143 7274, 7677, 79, 196, 206, 208
verbal noun 52, 169, 289, 201202, 205, 207211, willan (OE) 109, 120121,
306307 293294 137, 204, 206, 208
verbs of perception 170 werden-future 198 witan (OE) 127, 132133,
see also perception see also future/futurity 142, 146
verbs of utterance 169170, wesan (OE) 71n., 72, word order 8, 81, 8384,
173, 183189 7479, 164n., 201203, 8687, 91, 9497,
volition 120121, 128, 135, 137, 218, 275277, 101102, 107108, 118,
157, 207 294, 306 122123, 188, 266, 272,
wesan (OHG) 41, 4346, 277, 280281
W 5052, 5457, 6365, SOV/SVO 102,
wellan (OHG) 196, 201, 72, 279283, 122123, 277
204205 293294 verb-second 73, 80, 8285,
Welsh 296298 West Germanic 183, 195196, 95, 97
weoran/geweoran (OE) 11, 217, 294295, 299302 see also OV; SVO
41, 4752, 54, 5762, see also Germanic
6465, 7176, 7879, 198, West Saxon 1213, 107, Z
201204, 208, 210211, 217233, 295296 zu-infinitive 12, 151166
222223, 232, 294 wh-clause 178180 see also infinitive

(Studies in Language Companion Series 138) Gabriele Diewald, Leena Kahlas-Tarkka, Ilse Wischer-Comparative Studies in Early Germanic Languages_ With a Focus on Verbal Categories-John Benjamins Publish

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Studies in Language Companion Series 138) Gabriele Diewald, Leena Kahlas-Tarkka, Ilse Wischer-Comparative Studies in Early Germanic Languages_ With a Focus on Verbal Categories-John Benjamins Publish

Uploaded by

Copyright:

Available Formats

Comparative Studies in Early Germanic Languages

Studies in Language Companion Series (SLCS)

For an overview of all books published in this series, please see

John Benjamins Publishing Company

theAmerican National Standard for Information Sciences Permanence

Library of Congress Cataloging-in-Publication Data

International Conference on English Historical Linguistics (16th : 2010 : Pcs, Hungary)

2013 John Benjamins B.V.

Gabriele Diewald, Leena Kahlas-Tarkka & Ilse Wischer

This publication comprises the papers presented at a workshop on the Contras-

Additionally, forms of have or be could be combined with the past participle of a

(1) foram on VI dagum Crist geworhte heofonas & eoran,

& st r a hwile e mon worhte a burg t Tofeceastre mid stanwealle,

A similar aspectual function compared to that of ge-/gi- can be signified by the OE

(4) onne u as word gecweden hbbe genim one ppel

contrastive investigations of different lexical sources (e.g. Petr on the ori-

a survey on auditory evidential constructions in Old English and other

3. Historical comparative corpus studies

Historical study of language has become considerably more approachable during

popularity world-wide today. Electronic corpora have turned out to be invaluable

York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE).This syntacti-

4. The verbal categories studied in this volume

Table 1. Verbal categories and languages studied in this volume

Passive Voice OE/Gothic Cloutier

Typically, linguistic items (the verbal source lexemes) undergoing grammati-

A different aspect of modality is approached by Anne Jger in her paper

5. Summary and outlook

In this volume, this is achieved by a combination of grammaticalisation the-

Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: OUP.

Haspelmath, Martin. 1989. From purposive to infinitive: A universal path of grammaticalisa-

John continues to shock Steve.

2.1 Corpora and data collection

(2) a. the calling construction (with a direct object only)

d. the subclause commanding construction (with a subordinate clause)

3.1 Functions of Gothic haitan and Old English htan

Table 1. Distribution of Gothic haitan

As seen in Table 1, haitan occurs most frequently in transitive naming construc-

The instances of haitan occurring in calling constructions appear in both the

All examples of the infinitival commanding construction are in the active

In this example, ina him is the direct object of tiuhan to guide/lead.

(14) habaidedunuh an bandjan gatarhidana haitanana Barabban

In considering the distribution of voice in the Gothic data in Table 1 above, we

3.1.2 Early Old English

Table 2. Distribution of htan in Early Old English

The transitive naming construction (61%) is the most frequent construction in

seoan he hine to Cristes eowdome gehatenne hfde

(17) a ws seo fmne gehaten & fter fce Eadwine onsended

(19) Seo cwen het a m cyninge t heafod of aceorfan

The only variation in the infinitival commanding construction is the presence or

(21) eofas we hata o VII men

3.1.3 Late Old English

Table 3. Distribution of htan in Late Old English

(25) Datianus het hine secgan of hwilcere byrig he wre

& he cw to am sacerdum: Nima is Godes scrin and ga tforan am

(30) heo ws Raab gehaten

(31) Se ridda heafodwind hatte zephirus

Eft cw se hlend be his haligum anddetterum, a e hata confessores

3.1.4 Comparison of Gothic and Early and Late Old English

Table 4. A comparison of Gothic and Old English, normalized to 100,000 words

Active Passive Active Passive Active Passive

Early Old English seems to be an intermediary stage between what we observe

3.2 Competitors of haitan in Gothic

Table 5. Greek verbs translated by Gothic haitan per construction

Table 6. Gothic translations of the Greek calling construction

3. Historical comparative corpus studies

4. The verbal categories studied in this volume

5. Summary and outlook

2.1 Corpora and data collection

d. the subclause commanding construction (with a subordinate clause)

3.1 Functions of Gothic haitan and Old English htan

3.1.2 Early Old English

3.1.3 Late Old English

3.1.4 Comparison of Gothic and Early and Late Old English

3.2 Competitors of haitan in Gothic

4. Discussion and conclusion

2.1 Old High German

2.2 Old English

3.2 Past participle

3.3 Constructions with the past participle

3.3.2 Constructions with inchoative copula: werdan/weoran plus past participle

5. Conclusion and further research

2.1 Old English

2.2 Old High German

3. A different approach: Constructional environments

4. Bounded and unbounded language use

4.1 Status in Present Day English and New High German

4.2 The bounded system of Old English and its breakdown

4.3 The bounded system of German and its grammaticalisation

5.2 Corpora and data

5.3 Clause type

5.4 Time adverbs