Professional Documents
Culture Documents
1. Introduction
This article explores some differences in the interpretation of singular and
plural noun phrases in Persian and English under the assumption that
noun phrases can consist of at least three distinct layers: DP, the locus of
definiteness, NumP, the locus of number marking, and NP, the lexical or
substantive layer.1 One of the virtues of this articulated structure is that it
parallels that of clauses, which also contain at least three layers: CP, TP,
and VP. Based on consideration of the differences that exist between the
two languages, I will make several claims. First, with respect to the
internal syntax of noun phrases, I will argue that Persian lacks a NumP
projection and that number marking is instead connected to the DP
layer.2 Second, with respect to the external syntax of noun phrases, I will
argue that Persian allows bare NPs in argument positions while English
arguments must minimally be NumPs. Third, I will show that the Persian
indefinite enclitic ±i is associated with higher functional structure than its
English counterpart a.
* I have discussed the data and the ideas in this article with so many people and, in some
cases, so long ago that I risk leaving someone out if I attempt to list each person by name. I
am sincerely grateful to all. I am particularly grateful to the audiences at Allameh Tabataba'i
University in Tehran and Islamic Azad University at Khorasgan where I presented earlier
versions of this work in 2001. I would like to acknowledge my research assistants Nima
Sadat Tehrani and Saeed Ghaniabadi who have painstakingly gathered data and provided
judgements, along with my long-standing Persian consultant, Azar Ghomeshi. I would also
like to thank the two anonymous reviewers and my colleague H. C. Wolfart for their helpful
comments. All errors are my own.
1
The most frequently cited references for DP and NumP are Abney (1987) and Ritter
(1991), respectively. There are many other relevant references, however. For a brief overview
of the literature on functional categories within the noun phrase see Lyons (1999, section 8.2)
or Vangsnes (2001:253±54).
2
I will show later that this `layer' includes quantifiers as well.
Studia Linguistica 57(2) 2003, pp. 47±74. # The Editorial Board of Studia Linguistica 2003.
Published by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford OX4 2DQ, UK, and
350 Main Street, Malden, MA 02148, USA
48 Jila Ghomeshi
2. Background Assumptions
2.1. Genericity
For the purposes of this article I want to distinguish generic noun phrases
from non-referential nouns. Generic noun phrases refer to kinds rather
than to objects. As the following examples show, in English definite
singular count nouns, bare plural count nouns and bare mass nouns can
be generic, or kind-referring:
(2) a. The lion is a predatory cat.
b. Lions are predatory cats.
c. Gold is a precious metal. [Krifka et al. 1995:5.5]
Krifka et al. (1995:10) state that some predicates can only take kind-
referring arguments. For instance the subject argument of be extinct
and the object argument of invent are always kinds. These predicates
can be used to show that indefinite noun phrases do not normally refer
to kinds:
# The Editorial Board of Studia Linguistica 2003.
50 Jila Ghomeshi
6
Dabir-Moghaddam (1992) takes these examples from Phillott (1919:455, 459) and
points out that they reflect a different pronunciation from today's standard Persian. I
have, therefore, left the transcription as given by Dabir-Moghaddam but have changed the
glosses to conform to the system I am using here. Note also that the Persian sentence in (6b)
is not passive and so may perhaps be more accurately translated as: `Do you know how they
kill sheep?'
(17) a. *salts [unless we are talking about kinds of salt or packets of salt]
b. *nñmñk-ha [ " " ]
salt-pl
Third, mass nouns in English and Persian cannot appear with the
indefinite article (excluding the taxonomic or understood-quantity
reading):
(18) a. *a salt
b. *nñmñk-i
salt-ind
These facts, taken together, suggest that both English and Persian have
a lexical distinction between count and mass nouns. Coercion ± the
interpretation of count nouns as mass (the Universal Grinder) or mass
nouns as count (the Universal Sorter) ± can be understood as an
overriding of this lexical specification.8 That is, in English, it is not the
case that mass nouns can't appear with plural marking or the indefinite
8
What I am calling a `lexical' specification of count vs. mass may actually be a conceptual
one, as one reviewer points out. Coercion effects would then be deviations from the normal
syntactic use for a given noun (cf. Vangsnes 2001). I intend what I say in this paper about the
syntax of count vs. mass to be compatible with both views.
article. It's just that the resulting interpretation will have to be taxonomic
or of the understood-quantity type. Similarly, it is not the case that count
nouns can't appear as bare singulars, but when they do they will be
interpreted as mass nouns. These facts can be captured if, in addition to
the lexical distinction of count vs. mass, there is also a grammatical
distinction between count (singular or plural) and mass. As the following
diagrams show, I assume the relevant information is encoded within the
Number Phrase:9
(19)
of measure, which they call massifiers, and those that simply name the
unit in which the entity denoted by the noun naturally occurs, which they
call count classifiers:
Mandarin (Cheng & Sybesma 1999:514.10, 11)
(21) a. san ping jiu b. san ba mi
three bottle liquor three handful rice
`three bottles of liquor' `three handfuls of rice'
c. san wan tang
three bowl soup
`three bowls of soup'
(22) a. san ge ren b. san zhi bi
three cl people three cl pen
`three persons' `three pens'
c. san ben shu
three cl book
`three books'
Evidence that there is a semantic (or lexical) distinction between count
and mass nouns comes from the modification marker de. This marker can
intervene between mass classifiers and their nominal complements but
does not appear following count classifiers:
Mandarin (Cheng & Sybesma 1999:515.12, 516.13)
(23) a. san bang (de) rou b. liang xiang (de) shu
three cl-pound de meat three cl-box de book
`three pounds of meat' `three boxes of books'
(24) a. ba tou (*de) niu b. jiu gen (*de) weiba
eight cl-head de cow nine cl de tail
`eight cows' `nine tails'
c. shi zhang (*de) zhuozi
ten cl de table
`ten tables'
Like Chinese, Persian employs classifiers in order to make nouns
countable. That is, numerals must be accompanied by a classifier even
if the nominal complement is a count noun:
(25) a. se kilo gusÏt b. se livan ab
three kilo meat three glass water
`three kilos of meat' `three glasses of water'
c. se-ta nñmñk
three-cl salt
`three salts'
(26) a. se nñfñr kargñr b. se ÅÏjeld ketab
three person worker three volume book
`three workers' `three books'
# The Editorial Board of Studia Linguistica 2003.
56 Jila Ghomeshi
One of the main points of this paper is that plural marking in Persian,
unlike in English, is not located within a syntactic projection such as
Number Phrase. Generally, the kinds of affixes that correspond to
syntactic nodes such as tense, Case, and number, are inflectional. For
this reason it is worth noting that plural marking in Persian meets some of
the criteria for being a derivational rather than inflectional affix.
In Persian most affixes affect stress placement in the words to which
they attach, while a handful of affixes do not. The latter category includes
the case marker ±ra, the indefinite marker ±i, the pronominal enclitics and
the agreement affixes. By virtue of their position (±ra, ±i, and the
pronominal enclitics are all phrasal affixes) or function (agreement) this
collection of affixes can be classified as inflectional. This makes stress one
determinant for the split between inflectional and derivational affixes.
The plural marker ±ha, as a stress-attracting affix, thus falls on the
derivational side of the split (see also Ghomeshi 1996, Kahnemuyipour
2000).
While ±ha is the most commonly used plural maker it is not the only
one. For example, Lazard (1992 [1957] ) states that nouns denoting
animate entities take their plural with ±an (e.g. mñrd `man', mñrd-an
`men'; ñsb `horse', ñsb-an `horses'. There are also a handful of inanimate
nouns that fall into this category, such as derñxt `tree', derñxt-an `trees').
Further, Lazard notes that in formal language, words of Arabic origin
will take their plural form according to Arabic rules, yielding a variety
of ways to mark plurality (e.g. mosafer `traveller', mosafer-in `travellers';
# The Editorial Board of Studia Linguistica 2003.
Plural marking, indefiniteness and the NP 57
10
In the colloquial spoken language it is far more usual for a definite noun phrase to
appear with a demonstrative or the colloquial definiteness marker. This marker is discussed
in section 4.
that the contribution of the plural marker is unclear. Given that the
configuration in (33) is not possible, plural marking forces a referential
reading for predicate nominals.12
The definiteness effect of plural marking is also found in noun phrases
involving numerals. Generally plural marking and overt numerals cannot
cooccur, as shown in (35b). As we see in (35c), however, if a noun phrase
involving numerals is definite, plural marking can (and for some, must)
occur:
(35) a. se-ta ketab xñrid-ñm.
three-cl book buy.past-1sg
`I bought three books.'
b. *se-ta ketab-ha xñrid-ñm.
three-cl book-pl buy.past-1sg
c. se-ta ketab(-ha)-ro xñrid-ñm.
three-cl book(-pl)-om by.past-1sg
`I bought the three books.'
In subject position the same fact holds. For a noun phrase containing a
numeral to be construed as definite, plural marking is obligatory:13
(36) a. se-ta ketab ru-ye miz bud.
three-cl book on-ez table be.past.3sg
`Three books were on the table.'
b. se-ta ketab-ha ru-ye miz bud-ñnd.
three-cl book-pl on-ez table be.past-3pl
`The three books were on the table.'
In this section I have shown the link between plural marking and
definiteness. On the basis of the data shown so far, it may seem that plural
marking is definite. I will show in the next section that this hypothesis is
too strong.
12
Hopper & Thompson (1985:162) show that classifiers have the same effect in Man-
darin, cuing a referential reading for the predicate nominal.
13
The fact that plural marking can be used to signal definiteness is the relevant point here
but it is actually rarely used in this way. The most common and natural way to indicate
definiteness in translating an English sentence like (35b) is with the demonstratives in `this/
these' and an `that/those'.
4. Indefiniteness in Persian
The Persian indefinite enclitic ±i can occur on noun phrases in subject
(40a) and direct object (40b) positions, as well as on noun phrases that
occur as objects of prepositions (40c):
(40) a. mñrd-i amñd. b. mñrd-i did-ñm.
man-ind come.past.3sg man-ind see.past-1sg
`A man came.' `I saw a man.'
c. ketab-o be pesñr-i dad-ñm.
book-om to boy-ind give.past-1sg
`I gave the book to a boy.'
Thackston (1983) claims that ±i means `a certain, a particular' thing, or
# The Editorial Board of Studia Linguistica 2003.
Plural marking, indefiniteness and the NP 61
have in turn been said to have a `topic' reading (Cresti 1995). In a similar
vein GivoÂn (1981) characterizes referentiality as a signal that the referent
of an argument noun phrase is going to play some role in subsequent
discourse. While it may be difficult to identify which of these three
properties, `referentiality', `topicality', or `specificity', an ±i-marked
noun phrase in Persian takes on (and it may be all three, depending on
the context) what is certain is that a bare noun has none of these
properties.
So far we have seen that the indefinite enclitic cannot appear on
unmodified predicate nominals, that it occurs in contexts in which
English any/some/no occur, and that its presence can result in a specific
reading for a noun. For these reasons, I suggest that ±i is a quantitative
indefinite determiner which heads a QP. Further evidence for this
proposal comes from the fact that this enclitic cooccurs with a number
of quantifiers and that it obligatorily occurs on kñs `person' and cÏiz
`thing' to form `someone' and `something':
(49) a. hñr ketab-i b. hicÏ ketab-i
each book-ind no book-ind
`each/every book' `no book'
c. kñs(-an)-i d. cÏiz(-a)-i
person(-pl)-ind thing(-pl)-ind
`someone (pl)' `something(s)'
The focus of this section has been the indefinite enclitic ±i. Persian has
another indefinite marker, ye(k)16 `one', however, as well as another use
for ±i. Before concluding this section, I will briefly discuss these two
phenomena though in each case further research is required to determine
how they can be accommodated within the analysis presented in this paper.
The numeral yek `one' is used as a marker of indefiniteness which may
be unsurprising given that indefinite articles often evolve from one (Lyons
1999:95). Yek can cooccur with ±i or replace it entirely. This means that
there are three ways in which indefiniteness can be expressed:
(50) a. ketab-i b. ye ketab c. ye ketab-i
book-ind a book a book-ind
`a book' `a book' `a book'
There are two facts that suggest that yek should be considered a pre-
nominal counterpart to ±i rather than a cardinal numeral. First, in its
indefinite use it cannot be followed by a classifier. This makes it unlike all
other numerals which do have to appear with a classifier:17
16
In conversation the last consonant is usually dropped and yek is pronounced ye. From
here on, I will use the citation form in the text, however.
17
Yek can be used as a cardinal numeral (`one') in which case a classifier is required. The
choice of classifier depends on the following noun.
modified by a restrictive relative clause must appear with ±ra, the marker
that prototypically appears on definite direct objects:
(55) mñn ketab-i-*(ro) ke ñli pisÏnñhad kñrd xñrid-am.18
I book-iÇ-*(om) that Ali suggestion do.past.3sg buy.past-1sg
`I bought the book that Ali suggested.'
While uniting indefinite and relative uses of ±i is highly desirable in the
interests of elegance and simplicity, this seems to lead to the view that the
same morpheme can function both as an indefinite marker under Q0 and
as a definite determiner under D0 (cf. Karimi 2001). It remains to be seen
whether such a seemingly contradictory view can be maintained.
can appear neither on the head noun nor at the end of the noun phrase
that contains modifiers:
(62) a. ketab-e bozorg-o xñrid-ñm.
book-ez big-om buy.past-1sg
`I bought the big book.'
b. *ketab-ñ-ye bozorg-o xñrid-ñm.
book-def-ez big-om buy.past-1sg
c. *ketab-e bozorg-ñ-ro xñrid-ñm.
book-ez big-def-om buy.past-1sg
One exception to the generalization that only unmodified nouns partici-
pate in this construction is if the noun and following adjective form a
compound. That is, (62c) above is possible if no Ezafe vowel intervenes
and the main stress on the [N Adj] sequence falls on the suffix, as shown
in (63a).21 (Stress is indicated in capitals) (62a) is repeated as (63b) to
show that in the `normal' case, stress falls on the last syllable of the
adjective:
21
Samiian (1983) proposes a restructuring rule that converts an [N AP] string into a
complex noun that appears under an N0-node. I am adopting her insight that such strings
are compounds and are generated under N0.
7. Conclusion
In this article I have made several claims. First I have argued that Persian
differs from English in that it lacks a NumP projection:
(71)
Persian English
(73) a. [qp [np sñg-a]-i] did-ñm. b. *I saw [NumP a dogs].
dog-pl-ind see.past-1sg
`I saw some/certain dogs.'
Finally, I have argued that plural marking in Persian is licensed only on
noun phrases contained within D/QPs. This proposal accounts for the
fact that plural nouns are construed as definite unless an overt marker of
indefiniteness appears. What unites these claims is the view that the
interpretation of noun phrases is determined by the amount of syntactic
structure that is present and that comparable morphemes across
languages may correspond to this syntactic structure in different ways.
I would like to end this article with a quote showing that the general-
ization about plural marking I have put forth is not my own and has been
around for at least two centuries:
An interesting and insightful discussion of what nowadays is called
genericity was made by the Persian Ibrahim in his manual of 1841, written
for the East India Company's students in Haileybury, England. [ . . . ] The
bulk of his grammar is contained in eight dialogues between a Persian and
an open-minded Persian-speaking Englishman in Iran. In one of the
dialogues the Englishman uses the plural in the phrase: /vaxti ke asb-ha
amade and, soma ma-ra xabar konid/ ``When the horses are ready give me
notice.'' Politely, the Persian explains at length that a /lafz-e mofrad/ a single
word [i.e. the singular ± JG] does not always denote a unit. ``But don't we
need two horses?'' ``Yes,'' replied the Persian, now returning to English for
a lengthy discussion: ``But we didn't talk about specific horses, nor had we
talked earlier about certain horses to which that expression could have
referred; we just wanted to ride out on a horse''. [Windfuhr 1994:11]
References
Abney, S. P. 1987. The English noun phrase in its sentential aspect. PhD
dissertation, MIT.
Browne, W. 1970. More on definiteness markers: Interrogatives in Persian.
Lingustic Inquiry 1, 359±363.
Bunt, C. H. 1985. Mass terms and model-theoretical semantics. Cambridge Studies
in Linguistics 42. Cambridge: Cambridge University Press.
Campbell, R. 1996. One(s): The lonely number. Proceedings of NELS 26, ed.
K. Kusumoto, 43±55. University of Massachusetts: GLSA.
Campbell, J. 1998. The power of myth. Translated into Persian by Abbas
Mokhber. Tehran: NñsÏre mñrkñz.
Cheng, L. L. & Sybesma, R. 1999. Bare and not-so-bare nouns and the structure
of NP. Linguistic Inquiry 30, 509±542.
Chierchia, G. 1998. Reference to kinds across languages. Natural Language
Semantics 6, 339±405.
Cresti, D. 1995. Indefinite topics. PhD dissertation, MIT.
Dabir-Moghaddam, M. 1992. On the (in)dependence of syntax and pragmatics:
Evidence from the post-position ±raÅ in Persian. Cooperating with written texts,
ed. D. Stein, 549±573. Berlin: Mouton de Gruyter.
# The Editorial Board of Studia Linguistica 2003.
Plural marking, indefiniteness and the NP 73