You are on page 1of 8

The Syntax of Temporal Relations: A

Uniform Approach to Tense and Aspect*

<

HAMIDA DEMIRDACHE & MYRIAM URIBE-ETXEBARRIA


University of British Columbia & University of California, Irvine

1. Introduction
The goal of this paper is to lay the foundations of a crosslinguistically valid
theory of the grammar of temporal relations - a theory which uniformly
derives the compositional interaction of Tense and Aspect. The theory
developed here is based on the following premise: in order to derive the
interaction between Tense and Aspect, we must first reduce Tense and
Aspect to the same set of semantic and syntactic theoretical primitives. We,
thus, establish a strict parallelism between the syntax and semantics of
Tense and Aspect. We adopt Klein's (1995) proposal that both Tense and
Aspect establish ordering relations between two times. We capture this
semantic parallelism syntactically by extending the proposals in Stowell
(1993) and Zagona (1990). In particular, we propose that both TO and
ASpo are spatiotemporal predicates (prepositional-type heads such as
*We thank Gerhard Brugg~r, Henri Davis, Rose-Marie Dechaine, Ken Hale,
Eloise Jelinek, Alazne Landa, Toshi Ogihara, Javi Ormazabal, Karen Zagona and
Martina Wiltschko for helpful comments and discussion.

-.:

THE SYNTAX OF TEMPORAL RELATIONS / 147

146/ HAMIDA DEMIRDACHE & MYRIAM URIBE-ETXEBARRIA

(with)in or after) each projecting a maximal projection in the syntax (TP and
ASP-P) and ordering two time-denoting arguments.

2. What does (Grammatical) Aspect do?


Our first step towards building a uniform theory of the grammar of Tense
and Aspect will be to determine the role of (grammatical) aspect. For Smith
(1991), Aspect conveys a viewpoint on the situation described by a sentence
- that is, a temporal perspective that focuses all or parts of the situation. In
Smith's own words,
"Aspectual viewpoints function like the lens of a camera, making objects
visible to the receiver. Situations are the objects on which viewpoint lenses
are trained. And just as the camera lens is necessary to make the object
available for a picture, so viewpoints are necessary to make visible the
Smith (1991)
situation talked about in a sentence"

Thus, in a simple past tense sentence such as Laura built a house, the event
is presented as whole, as completed - as illustrated by the temporal schema
in (1). (The slashes indicate the part of the event that is focussed by the
viewpoint; I and F indicate the Initial and Final endpoints of the event,
following Smith's notation.)

(1)

Temporal schema ofa simple past sentence


Laura built a house

LI

F]

111111111111111111

In contrast, the progressive viewpoint in a sentence such as Laura was


building a house, focuses a subpart of the event of building that includes
neither its beginning nor its culmination. The event is presented without
Initial or Final end-points, as illustrated by the temporal schema in (2).

(2)

Temporal schema ofa progressive sentence


Laura was building a house

LI

F]
1111111

Smith (1991) argues that Aspect plays a crucial role in the semantic
interpretation of a sentence,
"Continuing the analogy of a viewpoint with the lens of a camera, we shall
say that the part focused by a viewpoint is visible to semantic
interpretation. "
"What is focused has a special status, which I will call visibility. Only what
is visible is asserted..." [emphasis added]
Smith (1991)

Thus, in the progressive sentence in (2), only the time interval within the
temporal contour of the eventfocussed by Aspect is visible to the semantic
interpretation. Since tlus time does not include the endpoints of the event,
no assertion is made about whether the event of building culminated -

although the sentence is in the past. Thus, we can say Laura was building a
house but she never finished it. In contrast, in the simple past sentence fri
(1), the event is portrayed in its entirety - as having both initial and final
endpoints. Hence, Laura built a house but she never finished it is a
contradictory statement.
We conclude that the role of Aspect is to focus (pick up) an interval in
the temporal contour of the event described by an utterance. Only the
interval of the event focused by Aspect is visible to semantic interpretation.
We call the time interval in the event time of the VP that Aspect focuses,
the Assertion Time (AST-T), following Klein (1995).
"The Assertion Time is the time for which an assertion is made or to which
the assertion is confined; for which the speaker makes a statement."
Klein (1995)

Having established that the role of Aspect is to focus a time interval in the
time of the event denoted by the VP, we now turn to the role of Tense.

3. What does Tense do?


Tense has been classically analysed as establishing an ordering relation
between two times. Tense relates the time of utterance (Uf-T) and the time
at which the event (or state) denoted by the VP occurs (or holds). Thus,
PAST orders the event time before the utterance time, whereas FUTURE
orders the event time after the utterance time, as illustrated in (3).1
(3)

a.

Past:

b. Future:

EV-T < UT-T

Rosa slept

UT-T < EV-T

Rosa will sleep

3.1. The Syntax of Tense


En y (1987), Zagona (1990), Stowell (1993), Giorgi & Pianesi (1991) and
Thompson (1994) among others, propose that the syntax and semantics of
Tense can be derived from a theory of the structural representation of Tense
together with independently motivated principles of Universal Grammar (see
also Hornstein 1990 for the latter claim).
Zagona (1990) and Stowell (1993) propose to syntactically capture the
idea that tense relates two times by breaking down Tense - structurally - into
its semantic components. Zagona (1990) proposes that Tense is a head that
projects a maximal projection TP and takes two time-denoting phrases as its
arguments. The external argument of lOis a REF-T (typically UT-T in
matrix clauses) and its internal argument is the EV-T.
Stowell (1993) adopts tIus proposal and further argues that Tense is a
dyadic predicate that establishes a temporal ordering relation between its two
time denoting arguments. In particular, Stowell analyses PAST tense as a
I Note that in Reichenbach's system of Tenses - which is based on three times,
the speech time S, the event time E, and the reference time R - the relation
between Sand E is always mediated by R. Thus, Tense does not directly order E
with respect to S, rather Tense orders R with respect to S.

THE SYNTAX OF TEMPORAL RELATIONS /

148/ HAMIDA DEMIRDACHE & MYRIAM URIBE-ETXEBARRIA

temporal ordering predicate with the meaning of after (Uf-T after EV-T) and
PRESENT tense as a temporal ordering predicate with the meaning of
(with)in (Uf-T within EV-T). The phrase structure for Tense proposed by
Stowell is illustrated in (4c-d).2

(4)

Stowell's (simplified) Phrase Structure for Tense


a.
Past
b. Present

c.

EV-T
TP

~
T'

UT-T

~
TO
VP
after

~
T'

~
TO
VP

EV-T

TP

UT-T

within
VP

EV-T

(5)

Klein (995):

Aspect, like Tense, relates (orders) two times


Aspect relates the Event-time to the Assertion-time
Tense relates the Assertion-time to Speech-time

-[-I

EV-T

Aspect should also be analysed as relating two times. This is precisely the
proposal made by Klein (1995).

Note that for Klein, Tense does not directIy order the Event-time with
respect to Speech-time. In this respect, he concurs with Reichenbach (1947),
Hornstein (1990), Giorgi & Pianesi (1991) and Thompson (199'4), for
whom the syntactic priInitives of tense are identified with Reichenbach's
three time points: S (speech time), E (event time) and R (reference time).

UT-T
]1------'>

UT-T
-[-}---I->

149

VP

Stowell identifies the EV-T with Kratzer's (1991) event argument. He


departs from Kratzer, however, in analysing this event argument as a time
denoting argument (rather than an argument of spatiotemporallocation) and
in assuming that all predicates - be it stage or individual level - have a
(temporal) event argument. He follows Kratzer in analysing this temporal
argument as the true external argument of the verb. Since it is the highest
(external) argument of the verb, it is base generated in the highest subject
position of a recursive VP shell structure, as in (4).

4. Putting Tense and Aspect Together


Recall that our goal is to build a theory of the syntax and semantics of
temporal relations that uniformly derives the interaction of Tense and
Aspect. We make the following assumption: in order to derive the
compositional interaction of Tense and Aspect, we must first reduce Tense
and Aspect to the same set of semantic and syntactic theoretical priInitives.
We first establish a strict parallel between the semantics of Tense and
Aspect (section 4.1). We then establish a strict parallel between the syntax
of Tense and Aspect (section 4.2).

4.2 A Uniform Syntactic Approach to Tense and Aspect


Having reduced Tense and Aspect to tlIe same set of semantic prilnitives, we
are now in a position to reduce Tense and Aspect to the same set of
syntactic priInitives. Zagona (1990) and Stowell (1993) argue tlIat tlIe idea
tIlat Tense relates two times can be captured syntactically by breaking down
Tense - structurally - into its semantic components. We claim that the idea
that Aspect relates two times can also be captured syntactically by
decomposing Aspect - structurally - into its semantic components. In
particular, we propose that both Tense and Aspect can be broken down structurally - into tlIe same set of semantic components once we assume
that both Tense and Aspect are spatiotemporal predicates
ordering two times.
Tense has been analysed as a head that projects a maximal projection
(TP) and takes two time denoting phrases as arguments. We propose tlIat
Aspect - like Tense - is a head tIlat projects a maximal projection (ASP-P)
and takes two time denoting phrases as arguments. For Stowell, tlIe head of
TP is a spatiotemporal predicate tIlat establishes a temporal ordering relation
between its two time denoting arguments. We propose that the head of
'ASP-P is also a spatiotemporal predicate that establishes a temporal
ordering relation between its two time denoting arguments. Under this
proposal, summarized in (6) and illustrated in (7), the syntax (and semantics)
of tense lnirrors tlIe syntax (and semantics) of Aspect.

(6)

A Uniform Syntax for Tense and Aspect


a.

Both Tense and Aspect are dyadic spatiotemporal ordering


predicates taking two time-denoting phrases as arguments.

4.1. A Uniform Semantic Approach to Tense and Aspect


In order to reduce Tense and Aspect to the same set of semantic prilnitives,
we assurne that, since Tense has been analysed as relating two times,3 then

b. The external argument of Aspect (ASPO) is a reference time (the


AST-T), its internal argument is tile time of the event denoted by
tile VP (BV-T).

2 We have simplified Stowell's phrase structure for expository reasons.


3 Cf. McCawley (1973), Musan (1995), von Stechow (1992), Stowell (1993),
Zagona (1990) and references therein.

c.

The external argument of Tense (TO) is a reference time (UT-T),


its internal argument is the AST-T.

150/ HAMIDA

DEMIRDACHE & MYRIAM URlBE-ETXEBARRIA

Recall that the role of ASPect is to focus a time interval in the temporal
contour of the event described by a sentence. The time focussed by Aspect is
the Assertion time, following Klein (1995). How does Aspect pick up
(focus) an interval in the time of the event denoted by the VP? By
establishing a relation between the AST-T and the EV -T. This relation can
be an ordering relation (e.g. the AST-Tis ordered after or before the EVT) or a topological relation (e.g. the AST-T is ordered within the EV-T).
Tense then orders the time interval focussed by Aspect (that is, the AST-T)
with respect to the Utterance time.
(7)

The Phrase-Structure of Tense and Aspect


TP

~
UT-T

T'

~
TO

ASP-P

~
AST-T

ASP'

ASpo

VP

~
EV-T

VP

~
If the strict parallel that we establish between the syntax of Tense and the
syntax of Aspect is correct, then temporal and aspecnlal relations can be
reduced to a simple schemata: a prepositional-type head establishes a
spatiotemporal relation between its two arguments. We believe that the
theory proposed here opens the way to a simple account of the temporal and
aspecrual systems instantiated in natural languages.
In the next sections, we illustrate our proposal with an analysis of two
aspects: the progressive and the perfect. Tins analysis is summarized in (8).
(8)

a.

Both TO and ASPO are spatiotemporal ordering predicates.

b. The head of TP is a temporal ordering predicate with the meaning


of after for Past or (with)in for Present, as in Stowell (1993).
c.

The head of ASP is a spatiotemporal ordering predicate with_the


meaning of after for Perfect Aspect or (with)in for Progressive
Aspect.

We will argue that the proposal in (9) has two correlated conceptual
advantages over alternative analyses of the Progressive and the Perfect. First,
it derives the compositional interaction between Tense and Aspect without

THE SYNTAX OF TEMPORAL RELATIONS /

151

resorting to ad-hoc diacritics or fearures (e.g. +/- completed, +/- perfective).


In particular, within a Reichenbachian system of tenses, diacritics must be
introduced in order to distinguish between the simple past and the past
progressive (for instance), as illustrated in (9). The arrow over E in (9b)
indicates that the designated event is not punctual but spreads over time.
(9)

a.

Past:

E, R_S

Rosa slept

->

b. Past progressive:

E, R_S

Rosa was sleeping

Any model based on Reichenbach's three times can integrate the Perfect into
a system of Tense representation. The Perfect acts like 'a past tense': it
locates the event time (E) in the past with respect to the Reference time (R).
The proposal in (7c) that Past Tense and Perfect Aspect are both
spatiotemporal predicates with the meaning of after also captures this idea.
Tile tricky question is how to integrate the Progressive into a system of
Tense representation. The thesis we defend in section 5 is that the
Progressive acts like a 'present tense': both Present Tense and Progressive
Aspect are spatiotemporal predicates with the meaning of within. The theory
we propose will, thus, unifonnly derive the interpretation of the Progressive
and the Perfect from the same set of syntactic and semantic primitives:
spatiotemporal predicates ordering two times.
A further advantage of our analysis is that it derives - without additional
stipulations - the syntax and semantics of recursive aspects (e.g. the Perfect
of a Progressive), as well as constraints on recursive aspects (e.g. *the
Progressive of a Perfect) - see Demirdache & Uribe- Etxebarria (1996).
S. The Phrase Structure of the Progressive Aspect
Consider the example in (10), a past progressive sentence. (10) describes a
process: Henry was in the process of building a house. We propose that
(10) has the phrase-structure in (11).
(10)

Henry was building a house.

In (11), the EV-T is a bounded interval [tI, t2] where t2 is the time that
defines the final endpoint of the event. The progressive aspect is a
spatiotemporal predicate with the meaning of (with)in: it orders the ASTT within the EV -T. It thus picks out a time contained witlnn the interval
defined by [t!, t2]. Past tense is a spatiotemporal predicate with the meaning
of after. It orders the UT-T after the AST-T. (11) thus focuses'a subinterval
within the interval defined by the event of building. TIlis subinterval is
itself located in the past, since the UT-T is ordered (by Tense) after tIlis
subinterval (i.e. after the AST-T).

(V

152/ HAMIDA DEMIRDACHE &

(11)

THE SYNTAX OF TEMPORAL RELATIONS

MYRIAM URIBE-ETXEBARRIA

The Phrase Structure of the Past Progressive

(13)

The Present Progressive

TP

TP

~
Uf-T

T'

Uf-T

TO

after

~
AST-T

ASP'

ASpo
VP
withilJ, ~

ASpo
VP
within ~
VP

Following Smith (1991), we assume that only the time interval witllln the
time of the event denoted by tlle VP focussed by aspect is visible to the
semantic interpretation (see section 2). The time focussed by ASPect in (11)
- wlllch we have called the AST-T, following Klein (1995) - must be a
subinterval of tlle event time; that is, it must be properly' contained within
the EV-T.4 Since the AST-T does no include the endpoints of the event, no
assertion is made about whether the event of building culminated - although
the sentence is in the past. (10) CaII, thus, be followed by but he never
finished itl ... and he is still building itl ... and he finished it this summer.
The present progressive sentence in (12) has the sanle phrase structure
as the progressive sentence in (10), as illustrated in (13). The single
difference between (10) aIld (12) is tlle tense of the sentence (past vs.
present). Whereas in (11), tlle head of TP is a spatiotemporal predicate with
the meaning of after, in (13), the head of TP is a spatiotemporal predicate
witll the meaIung of within.
Henry is building a house.

The Progressive Aspect is a spatiotemporal predicate with the meaning of


(with)in: it orders tlle AST-T within tlle EV-T. It thus picks out a time
contained wit1un the time of the event denoted by the VP. Present Tense is
also a spatiotemporal predicate with the meaIllng of (with)in. It orders the
Uf-T witilln the AST-T. (13), thus, focuses a subinterval of the time that
defines the event of building. TIllS subinterval is located in the present since
it includes the Uf-T.

EV-T

This raises the question of how we define the EV -T when the event never
culminates (e.g. Max was crossing the street when he was struck dead). This issue
- and more generally, the imperfective paradox - is discussed in detail in
Demirdache & Uribe-Etxebarria (1996).

VP

5.1. The Temporal Frame Reading of the Progressive


This analysis of the Progressive as the spatiotemporal predicate (with)in .
explains why the progressive entails the existence of an event of greater
duration than (what Jesperson called) "the framed time" (see Kearns 1991).
In particular, Jesperson states that in the progressive sentence He is hunting,
"The hunting is felt to be a kind of frame around something else; it is
represented as lasting some time before a'nd possibly (or probably) also
some time after something else, which mayor may not be expressly
indicated, but which is always in the mind of the speaker... ..
Jesperson (1932)
Jesperson'sframed time is our Assertion Time. The AST-T is ordered by the
Progressive Aspect within the event time. The event time, thus, contains frames - the Assertion Time. The EV-T will be larger then tlus framed time.
5.2. Cross linguistic Evidence for our Analysis of the
Progressive
We now provide crosslinguistic support for the proposal that the
Progressive is a spatiotemporal predicate with the meaIung of (with)in.
5.2.1. French
In French, tlle Progressive is fonned by combilung all infinitive with a
complex aspectual expression contailung a locative preposition wluch is
traIlslatable as 'engaged in' aIld could be decomposed as in + along.
(14)

ASPP

within

ASP'

EV-T

T'

TO

ASPP

AST-T

(12)

/153

a.

[BE + en train de + INFINITIVE]

b. Zazie
Zazie

est
IS

en
in

train
along

ce
of

miauler
miaowing

'Zazie is engaged innuaowing' I 'Zazie is miaowing'

THE SYNTAX OF TEMPORAL RELATIONS / ISS

154/ HAMIDA DEMIRDACHE & MYRIAM URIBE-ETXEBARRIA

5.2.2. Basque

5.2.5. Basque Again

Setting aside the reduced group of verbs that fonn the progressive by
making use of a synthetic fonn, the progressive is fonned by combining
the nominalized form of the verb suffixed with the inesive (locative)
postposition -n 'in/on/at' and the verb ari 'engage'. Progressive sentences
are always intransitive.

TIus correlation between locative 'Be' and the auxiliary used in progressive
sentences also holds in Basque. In particular, progressive constructions can
be fonned by combining tile locative verb egan 'Be' with a nominalized
fonn of the verb in inesive (locative) case, as illustrated in (l8a).

(15)

a.

(18)

a.

[V + tze + LOC (IN) + ARI (engage) + AUX (BE)]

b. Amaia
leihoa
apur-tze-n
aIi
cb.
Amaia-ABs window-ABs break-NoM-LOC (in) engage Aux(is)
Lit 'Amaia is engaged in breaking the window'

'MU-en is (= estar) reading tile book'


b.

'Amaia is breaking the window'

5.2.3. St'at'imcets (Lillooet Salish)


In St'at'imcets, progressive sentences make use of the auxiliary wa7, which
van Eijk (1985) defines as in (16).
(16)

wa7: 'to be (busy with, involved in)'


''The (auxiliary) use of wa7 corresponds to the English progressive."

Davis (1996) argues that main verb wa7 only takes locative complements
and establishes the following correlation between the auxiliary use of wa7
in progressive sentences aIld the locative use of main verb wa7.
.. The primary function of the main verb wa7 is locative. Main verb wa7
unlike English be is compatible only with locative complements... Main
verb wa7 appears to mean not just be but be somewhere; in other words, just
as auxiliary wa 7 supplies temporal coordinates, main verb wa 7 supplies
spatial coordinates which may be fixed by locative deictics or prepositional
phrases .....

Our analysis of the Progressive as a spatiotemporal predicate with the


meaIung of (with)in provides a principled eXplaIlation for tIus correlation
between the locative use aIld tile progressive use of wa7 in St'<it'imcets.

5.2.4. Spanish
The SaIne correlation holds in SpaIlish between tile auxiliary verb used in
progressive constructions aIld the verb 'Be' used with locative expressions.
There are two verbs 'Be' in SpaIush: Ser and ESlar. It is tile locative 'Be'
(Eslar) that appears in progressive constructions (Eloise Jelinek, p.c.).
(17)

a.

b.

Progressive.
Jator
es ta
Jator
1s

[ ESTAR + Vgerund]
coniendo
nUUling
[ESTAR + LOC NP]

Simple Locative
Jator
esta

en

casa

Jator

at/in

home

IS

Progressive.
[V + tze + LOC (IN) EGON (ESTAR)]
Miren
liburua
irakur-tze-n
dago
Miren-ABS book-ABS read-NOM-LOC be (estar)3-SG.ABS
Simple Locative
[NP-LOC EGON (ESTAR)]
Miren
etxe-a-n
dago
Miren-ABS
house-DEf-LOC be (estar) 3-SG.ABS
'Miren is (= estar) in tile house'

In (19), we see that copular constructions with a locative NP have a


progressive interpretation. Note, also that tile wh-phrase in interrogative
progressive sentences such as (20) is a locative NP.
(19)

(20)

LaIle-a-n
dago
[NP-LOC EGON (ESTAR) ]
work-DEf-LOC be (estar) l-SG.ABS
Lit: 'S/he is inlat work' ='S/he is working'
a.

Zer-tan
what-LOC (in)

zaude?
be (estar) 2-SG.ABS

Lit: "In what are you?" =''What are you doing?"


or

b. Zer-tan
what-LOC (in)

ari
engage

zara?
be (ser) 2-SG.ABS

Lit In what are you engaged? = "What are you doing?"

5.2.6. English
'Locative inesive expressions with a progressive interpretation are also found
in English, where we Call say idiomatically I am in the middle a/washing
the dishes (as pointed out by Eloise Jelinek p.c.)
5.3. Historical Evidence for our Analysis of the Progressive
Historical evidence for our proposal that the progressive is a spatiotemporal
ordering predicate with the meaIllng of (with)in is provided by Vlach (1976)
and Bybee, Perkins & Pagliuca (1994). Vlach argues tIlat the historical
antecedent of the progressive in English is a construction in which a
preposition - eitIler on or at, later shortened to a- as in 'asleep' - combined
with a genmd(-ive) noun phrase, as illustrated in (21), quoted from Bybee,
Perkins & Pagliuca (1994).

THE SYNTAX OF TEMPORAL RELATIONS / 157

156/ HAMIDA DEMIRDACHE & MYRIAM URIBE-ETXEBARRIA

(21)

a.

He is on hunting

b.

b.

He was a-coming home

TP

According to Vlach, this locative preposition has the meaning of '(engaged)


in'; (21a), can thus be paraphrased as 'He is in the process of running.'

~
UT-T

5.4. Conclusion
Our model of temporal relations is based on three times (BV-T, AST-T, UTT). However, it cannot be reduced to the classical Reichenbachian system
(i.e. EV-T, REF-T, UT-T). Within a Reichenbachian system, a REF-T
cannot be ordered within another time - that is, a REF-T cannot refer "to a
subpart of the EV -T. Reichenbach is, thus, forced to resort to an ad-hoc
system of diacritics to derive the interpretation of the progressive - as was
illustrated in (9) above, where an arrow over E was used to indicate that the
E(vent time) is not punctual but spreads over time.
We have integrated the progressive aspect into a system of Tense
representation - and uniformly derived the compositional interaction of
Tense and Aspect - without resorting to either features such as [+/completed, +/- progressive, +/-perfective] or ad-hoc diacritics. Our analysis
rests on two assumptions: (i) ASP<> is a dyadic predicate taking two timedenoting phrases as arguments; (ii) the Progressive Aspect is a
spatiotemporal predicate with meaning of (with)in. In a progressive
sentence, the event denoted by the VP is presented without Initial or Final
end-points because ASP establishes a topological relation between the time
to which the assertion of the sentence is confined and the event time of the
VP: it orders the AST-T within the EV-T. The AST-T in (11) or (13),
thus, refers to a subpart of the EV-T which includes neither the Initial nor
Final endpoint of the event.
Under tllis analysis, the Progressive acts like 'a present tense': both
Present Tense and Progressive Aspect are analysed as spatiotemporal
predicates with the meatling of within.

6. The Present Perfeet


6.1.The Syntax of the English Pel'feet
We now integrate the English Perfect (Have Ved) into our system of Tense
representation by atlalysing the Perfect as a spatiotemporal predicate with
the meatling of after. Under this atlalysis, the Perfect is like 'a past tense':
both Past Tense and Perfect Aspect are spatiotemporal predicates with the
meatling of after. We illustrate this proposal in (22). (22b) is the phrase
structure proposed for the past progressive sentence in (22a).
(22)

a.

Laura had built a house.

The Phrase Structure of the Past Perfect

T'

TO
after

ASP.P
~

ASP'

AST-T

ASpo
VP
after
~
EV-T

VP

In (22b), the EV -T is a bounded interval [11, t2] where t2 is the final


endpoint of the event. The perfect is a spatiotemporal predicate with the
meatling of after. It orders the AST-T after the EV-T: it thus picks out a
time after the interval defined by [11, t2]. 'Laura PERF BUILD a house', thus,
describes an event of building, completed prior to some reference time (the
AST-T). Past Tense is a spatiotemporal predicate with the meaning of
after. It orders the UT-T after the AST-T. Laura had built a house, thus,
asserts that an event of building was completed prior to some past reference
time (e.g. prior to 1986).
This analysis, thus, captures the idea that the Perfect acts likes a past
tense: it locates the EV-T in the past with respect to a reference time (the
AST-T) by ordering the AST-T after the EV-T.
The Present perfect sentence in (23a) has the satne phrase structure
except that Tense is a spatiotemporal predicate with the meaning of
within. It orders the UT-T within the AST-Laura has built a house, thus,
asserts that an event of building was completed prior to a present reference
time (i.e. prior to UT-T).
For an analysis that derives the existential, resultative and continuative
readings of the English Present Perfect uniformly from the proposal that the
perfect is a spatiotemporal predicate with the meatling of after, see
Denlirdache & Uribe- Etxebarria (1996).
(23)

a.

Laura is building a house.

158/ HAMIDA

DEMIRDACHE & MYRIAM URIBE-ETXEBARRIA

b.

THE SYNTAX OF TEMPORAL RELATIONS

Selected References

The Phrase Structure of the Present Perfect


TP

~
UT-T

T'

TO
within

ASP-P
~

AST-T

ASP'

ASpo

after

VP
~

EV-T

VP

We have derived the perfective viewpoint of the perfect (i.e. the event is
presented as completed) without resorting to ad-hoc features such as, +/completed, +/- perfective, +/- terminated. The event is presented as
completed because the perfect orders a reference time (the ASS-T) after the
EV-T and, hence, after the time that defines the culmination of the event.

6.2. Cross-linguistic
Perfect

Evidence

for

our

Analysis

of

/159

the

Cross-linguistics evidence for our analysis of the perfect as a spatiotemporal


predicate with the meaning of after is provided by Bull (1960), who argues
that,
"In Hebrew, the difference between "is singing", shar, and "has sung", kvar
shar, is indicated by kvar, which also means "already". Likewise, in Hebrew
a translation of "he had already sung" is literally: "he was already after his
singing" (bar shar ~~shirato). This is paralleled in Gaelic by "he was
after his singing" (rabh se ndiaidh seinnm).
"... these languages demonstrate that both the concept of aspect and the
concept of order produce identical results. "
Bull (1960)

7. Conclusion
The uniform structural representation for Tense and Aspect proposed here is
based on a simple schemata: spatiotemporal predicates relating times.
Present is analysed as UT-T (witlz)ill AST-T, Past as UT-T after AST-T,
Progressive as AST-T (witlz)ill EV-T and Perfect as AST-T afterEV-T.
We believe that our proposal can be extended to other aspects, once we
assume that Aspects are spatiotemporal predicates and can explain why
predicates of +/- central coincidence (e.g. allative to/towards, locative
in/on/at) playa pervasive role in temporal and aspectual systems crosslinguistically (cf. Hale 1984, Fong 1996).

Bull, William. 1960. Time, Tense and the verb. University of California Press.
Brugger, Gerhard.1996. The Temporal Representation of Present Perfect Types,
ms. UCLA.
ByBee, John & Revere Perkins & William Pagliuca. 1994. The Evolution of the
Grammar. The University of Chicago Press.
Davis, Henry. 1995. Auxiliary and Main Verb Wa 7 in St'at'imcets. Paper
presented at the Fourth Annual Salish Morphosyntax Workshop, UVic.
Demirdache, Hamida & Myriam Uribe- Etxebarria. 1996. The Primitives of
Temporal Relations, ms. UBC & UCr.
En<;, MUrvet. 1987. Anchoring Conditions for Tense. Linguistic Inquiry 18.
Fong, Vivienne. 1996. A Temporal Interpretation for Locative Case.
Proceedings of WCCFL xv.
Giorgi, Alessandra & Fabio Giorgi. 1991. Toward a Syntax of Temporal
Representation, PROBUS 3-2.
Hale, Kenneth. 1985. Notes on World View and Semantic Categories: some
Warlpiri examples, in P. Muysken & H. van Riemsdijk, eds., Features and
Projections. Foris, Dordiecht.
Hornstein, Norbert. 1990. As Time goes by: Tense and Universal Grammar. MIT
Press.
Jespersen, Otto. 1932. A Modem English Grammar on Historical Principles, IV.
London: George Allen and Unwin.
Kearns, Katherine. 1991. The semantics of the English Progressive, Doctoral.
dissertation, MIT.
Klein, Wolfgang. 1995. A Time Relational Analysis of Russian Aspect.
Language, Vol. 71.4.
Kratzer, Angelika. 1995. Stage-Level and Individual-Level Predicates, in Carlson
& Pelletier, eds., The Generic Book. The University of Chicago Press.
Leder, Harry. 1991. Tense and Temporal Order, Doctoral dissertation, MIT.
van Eijk, Jan. 1983. A Lillooet-English DictiO,nary. Mount Currie, Be.
Musan Renate. 1995. On the Temporal Interpretation of Noun Phrases. Doctoral.
dissertation, MIT.
Reichenbach, Hans. 1947. Elements of Symbolic Logic. The Free Press, N.Y.
Smith, Carlotta. 1991. The Parameter of Aspect. Kluwer Academic Press.
Stowell, Tim. 1993. The syntax of Tense, ms. UCLA.
_ _ _ _ To appear. The Phrase Structure of Tense, in 1. Rooryck & L. Zaring,
eds., Phrase Structure and the Lexicon.
Thompson, Ellen. 1994. The Structure of Tense and the Syntax of Temporal
Adverbs. Proceedings of WCCFL Xlll.
Zagona, K. 1990. Times as Temporal Argument Structure. Paper presented at the
Time in Language Conference, MIT.
_ _ _ _ 1993. Perfectivity and Temporal Arguments, Proceedings of XXIII
LSRI.

You might also like