1. Classifiers and NPE !
!
It has been argued that Cantonese classifiers can license NPE (see (5) and (6))
(e.g., Cheng & Sybesma 2009). However, we observe that the presence of
classifiers alone does not guarantee the legitimacy of NPE (see (7) and (8)). (SFP
= Sentence Final Particle; Exp = Experiential marker)!
!
(5) Indefinite [Cl+N] à NPE!
Ngo !maai-zo !zi !bat, !keoi !dou !maai-zo !zi [e ].!
I ! !buy-Asp !CL !pen !he !also !buy- Asp !CL!
‘I bought a pen, he also bought a pen.’!
!
(6) Indefinite [Num+Cl+N] à NPE!
Ngo maai-zo saam-bun syu, keoi !dou !maai-zo !saam-bun [e ].!
I ! buy-Asp three-CL book he !also !buy-Asp !three-CL!
‘I bought three books, he also bought three books.’!
!
(7) Definite [Cl+N] à *NPE!
Mingzai !jiging !taijyun! !pin !man ! !laa.!
Ming !already !read.finish !CL !article !SFP!
Daanhai !ngo !zung !mei ! !taijyun! !pin !*(man) !aa.!
but !I !still !not.yet !read.finish !CL !article !SFP!
‘Ming has already finished reading the article, but I haven’t finished reading
the article.’!
!
(8) Definite [RC+Cl+N] à *NPE!
Nei !gindou !bin-zek !gau !aa?!
you !see ! !which-CL !dog !SFP!
Ngo !gindou !ngaau-gwo !Mingzai !zek !*(gau).!
I ! !see ! !bite-Exp !Ming !CL !dog!
‘Which dog did you see? I saw the dog that bit Ming.’!
!
The data above show that when Cl is present, NPE is licensed in certain cases,
but not in others. This confirms the proposal that head-government is just the
necessary condition but not the sufficient condition for licensing ellipsis. !
!
!In the [Cl + N] sequence in (5), the numeral appears to be absent; we
consider that there is a covert numeral ‘one’. This numeral can be realized as jat
‘a’ (see (9)). !
!
(9) Ngo maai-zo !jat-zi ! bat, !keoi !dou !maai-zo !jat-zi [e].!
I ! buy-Asp !one-CL pen !he !also !buy-Asp !one-CL!
‘I bought a pen, he also bought a pen.’!
!
If there is a covert numeral in the indefinite noun phrase in (5), the four cases in
(5) to (8) can be divided into two groups: in indefinite noun phrases, NPE can be
licensed; in definite noun phrases, NPE is not possible. In indefinite noun phrases
in Cantonese, there is a numeral (either overt of covert); in definite Cantonese
noun phrases there is no numeral. The structures for the two types of noun
phrases are illustrated below.!
!
(10) a. b.!
!
!
!
!
!
!
!
!
!
Summary!
!
In sum, when the numeral (either overt or covert) is present, the specifier position
of the classifier is filled, and NPE is possible. We conclude that there are two
types of Cl projections in Cantonese, as shown in (10) and (11). Our observation
also indirectly supports Saito, Lin, & Murasugi’s (2008) proposal that NPE in
Mandarin Chinese is licensed by a Cl head with a Spec filled by NumP. !
!
In the last section, we observed that Cantonese numeral jat ‘one’ can be covert.
This is not always possible. When there is a modifier in the indefinite noun phrase,
jat ‘one’ cannot be covert (see (11a, b)). In this section, we explore why this is so.
(GE = Cantonese modification marker ge)!
!
(11) a. *[ngaau-gwo !Mingzai !ge] ! !zek !gau !!
bite- Exp !Ming ! !GE ! !CL !dog!
b. [ngaau-gwo !Mingzai !ge] !jat !zek !gau !!
bite- Exp !Ming ! !GE !one !CL !dog!
‘a dog that bit Ming’!
!
!
Specificity and Numerals!
!
Cheng and Sybesma (1999) propose that the specific reading for a [Num + Cl + N]
sequence requires the phrase to undergo Quantifier Raising (QR) (see Diesing
1992). An empty numeral cannot QR, and thus the specific reading is unavailable
when the numeral is covert. Only with an overt numeral can the noun phrase
undergo QR and receive a specific interpretation. !
!
For instance, a bounded predicate requires that its object be specific. In such
cases, the numeral ‘one’ must be overt. Example (12) is taken from Cheng and
Sybesma (1999), with slight modifications (see also Sybesma 1992). The bounded
predicate chi-wan ‘finish eating’ indicates that the noun phrase in the object
position refers to a specific cookie. The numeral cannot be covert.!
!
(12) Wo !chi-wan-le ! !*(yi)-kuai ! !binggan. (Mandarin)!
I ! !eat-finish-Asp! !one-CL ! !cookie!
‘I finished eating a cookie.’!
!
!The proposal that a non-overt numeral does not QR predicts that such a
phrase does not have scope ambiguities. This is confirmed by the contrasting
sentences in (13) and (14). In (13), the numeral ‘one’ is overt. The sentence can
be interpreted either as ‘there is a specific dog such that we all want it’ or as ‘the
number of the dogs that each of us wants is one’. In (14), the numeral is covert;
the sentence only has the latter interpretation. !
!
(13) a. Women !dou !xiang !yao ! !yi-zhi !gou. !(Mandarin)!
we ! !all !want !have ! !one-CL !dog!
b. Ngodei !dou !soeng!jiu ! !jat-zek!gau. !(Cantonese)!
we ! !all !want !have ! !one-CL !dog!
Both: ‘We all want a dog.’ (any dog / a specific dog)!
!
(14) a. Women !dou !xiang !yao ! !zhi !gou. !(Mandarin)!
we ! !all !want !have !CL !dog!
b. Ngodei !dou !soeng!jiu ! !zek !gau. !(Cantonese)!
we ! !all !want !have !CL !dog!
Both: ‘We all want a dog.’ (any dog)!
(Note: (14b) can also mean ‘we all want the dog’. Only the indefinite reading
! is relevant here.)!
!
!It is possible that, the numeral must QR when there is a modifier in the
noun phrase; and hence, the numeral must be overt (see (11a, b)). Yet why a
[Num + Cl + N] sequence obligatorily undergoes QR when it is modified? It has
been observed that when there is a modifier at the left periphery of [Num + Cl +
N], this noun phrase is always specific (Zhang 2006). Since QR is required to
derive the specific reading for the noun phrase, the numeral of an indefinite noun
phrase must be overt when it is modified. !
!
!!
!
Summary!
!
In this section, we discussed the overtness of numeral jat ‘one’ in Cantonese. We
considered the proposals of relevant studies and suggested that the overtness of
jat ‘one’ in Cantonese is related to specificity. !
!
Based on our novel observation that classifiers in Cantonese do not license NP-
ellipsis when they are definite, we propose that there are two types of Cl
projections in Cantonese, one with a specifier filled by the numeral and one with
no specifier. Theoretically, the discussion on NP-ellipsis in Cantonese supports the
proposal that ellipsis can be licensed when the specifier position of the head in
question is filled. We also examined the numeral jat ‘one’ and suggested that it
may be covert only when the noun phrase is non-specific. !
!
We have not discussed other elements of Cantonese noun phrases, such as
demonstratives and the most puzzling modification marker ge. My current
research is targeting a comparison of noun phrases and NPE in Mandarin,
Cantonese, Japanese, and Turkish. Hopefully, a comparative study will lead to a
better understanding of noun phrase structures.!
!
!
!
!
!
!
Acknowledgements!
!
Part of this paper has been presented at The Fifth International Conference on
Formal Linguistics, Guangdong University of Foreign Studies, China, with Hoi-ki
Law as the second author. Most of the Cantonese data are provided by Hoi-ki Law
and are verified by Cat Fung and Boji Lam. I would like to thank the following
people for their valuable comments: Candice Cheung, Richard Larson, Hoi-ki Law,
Paul Law, Haoze Li, Jiahui Yang, and the anonymous reviewers. Any mistake is
my responsibility.!
!
!
!
!
!
!
References!
!
Alexiadou, Artemis, and Gengel, Kirsten. 2008. NP ellipsis without focus
movement/projections: The role of classifiers. Paper presented at Workshop on
Interface-Based Approaches to Information Structure, University College London.!
!
Cheng, L.-S. Lisa and Sybesma, Rint. 1999. Bare and not-so-bare nouns and the
structure of NP. Linguistic Inquiry (4): 509-542.!
!
Cheng, L.-S. Lisa and Sybesma, Rint. 2009. De as an underspecified classifier:
First explorations. In Essays in Linguistics, ed. Lu Jianming et. al, 39:123-156. !
!
Diesing, Molly. 1992. Indefinites. Cambridge, Mass.: MIT Press. !
!
Lobeck, Anne. 1990. Functional heads as Proper Governors. Proceedings of the
North East Linguistic Society 20: 348-362.!
!
Lobeck, Anne. 1995. Ellipsis: Functional heads, licensing, and identification. New
York: Oxford University Press. !
!
Saito, Mamoru and Keiko Murasugi. 1990. N’-deletion in Japanese: A preliminary
study. Japanese/Korean Linguistics (1): 258-301. !
!
Saito, Mamoru, T.-H. Jonah Lin, and Keiko Murasugi. 2008. N’-Ellipsis and the
Structure of Noun Phrases in Chinese and Japanese. Journal of East Asian
Linguistics (3): 247-271.!
!
Sybesma, Rint. 1992. Causatives and accomplishments: The case of Chinese ba.
Doctoral dissertation, HIL/Leiden University.!
!
Zhang, Niina. 2006. Representing specificity by the internal order of indefinites.
Linguistics (1): 1-21.!
!
!
Abstract
NPE in Cantonese
Overtness of the Numeral
Conclusion
Classifiers and Numerals in Cantonese
Yang, Jing
Dept. of Linguistics, University of Connecticut
Copyright Information Here!
NP-Ellipsis (NPE)
Based on the studies of Saito & Murasugi (1990) and Lobeck (1990, 1995), Saito,
Lin, & Murasugi (2008) propose that ellipses are licensed only by heads with filled
specifiers. In the tree diagram below, the specifier position of XP is filled by ZP;
the ellipsis of YP is licensed. !
!
(1)!
!
!
!
!
!
!
!
For instance, when the specifier of DP is occupied by a genitive, the NP book can
be elided (see (2a)). In (2b), a is the determiner and Spec, DP is empty; the NP
cannot be elided.!
!
(2) a. !I saw Mary’s book, but I didn’t see [DP John’s [NP book]].!
b. *!I translated a book, but I didn’t edit [DP a [NP book]].!
!
!Following this proposal on ellipsis, Saito, Lin, & Murasugi (2008) argue that
Mandarin Numeral Phrases (NumPs) should occupy Spec, ClP, since the NPs
after classifiers can be elided in Mandarin (see (3)) (see also Alexiadou & Gengel
2008 on other languages). (Asp = Aspect; CL = Classifier)!
!
(3) Ta !mai-le liang-zhi !bi, !wo !ye !mai-le !liang-zhi [e].!
he !buy-Asp two-CL !pen !I !too !buy-Asp !two-CL!
‘He bought two pens; I bought two pens, too.’!
!
The relevant structure is illustrated in (4).!
!
(4)!
!
!
!
!
!
!
In Mandarin, there are no data showing that a classifier without a specifier does
not allow NPE. If there were such data, the support for the analysis in (4) would be
much stronger. Cantonese provides such evidence.!
!
!
!
!
!
!
!
!
The material presented here are part of the following paper:!
Yang, Jing. To appear. The Interaction between Cantonese ge and Classifiers. Monograph on the Attributive Particle in Chinese. Beijing: Peking University Press. !
!
In this paper, some properties of Cantonese classifiers and numerals are
examined. It is observed that, contrary to the common view that
classifiers always license NP-ellipsis in Chinese, Cantonese classifiers
do not allow NP-ellipsis when they encode definiteness. We propose that
Cantonese classifiers (Cls) have two types of projections: ClP may or
may not project a specifier position. Only when there is Spec, ClP (which
is filled by a numeral) does the Cl head license NP-ellipsis. The numeral
‘one’ may be covert, and this is possible only when the noun phrase is
nonspecific. !