NELS 35, Connecticut University, October 22-24,2004
dou-(dis)harmony in Chinese
Liping Chen, Rutgers University
Introduction It is a well-known fact that dou co-occurs with quantifiers in Chinese, a phenomenon I call dou-harmony. This talk is concerned with dou-disharmony, by which I mean the incompatibility of dou co-occurring certain quantifiers such as henshaode ‘few’ and yixie ‘some’.
1. dou (dis)harmony in Chinese
· Facts
(1) dou harmony: dou co-occurs with quantifiers such as the following:
a. womenxi meige laoshi *(dou) maile fangzi Öevery--dou
our dept mei-CL teacher all buy-ASP house
‘Each teacher in our dept. bought a house’
b. women xi (yi) da bufen laoshi (dou) maile fangzi Ömost--dou
‘Most teachers in our dept. bought a house.’
c. women xi henduo laoshi (dou) maile fangzi Ömany--dou
‘Many teachers in our dept. bought a house’
d. womenxi suoyou de laoshi *(dou) maile fangzi Ö all--dou
‘All teachers in our dept. bought a house.’
(2) dou disharmony: dou does not co-occur with the sort of quantifiers below
*a. (you) henshao de laoshi dou maile fangzi * few--dou
(exist) very-few DE teacher all buy-ASP house
‘(There are) very few teachers who bought their houses’
*b. (you) yixie laoshi dou maile fangzi * some--dou
(exist) one-CL teacher
· dou non-compatible NPs in (2) are not weak NPs (Wu 1999) or DE NPs
Ø Not weak NPs:
(3)
Quantifiers / Weak / dou compatiblehenduode ‘many’ / Ö / Ö
henshaode ‘few’ / Ö / ´
(4) you henshaode / henduode laoshi zai fangjian li
exist very few DE/many DE teacher at house inside
‘There are very few/many teachers in the house.’
Ø Not DE NPs: yixie ‘some’ doesn’t show any set-subset relation as is typical for a DE quantifier:
(5) yixie laoshi you fangzi ¹ yixie waiguo laoshi you fangzi.
some teacher have house some foreign teacher have house
‘Some teachers have houses’ ¹ ‘Some foreign teachers have houses’
2. Problems posed to the previous analyses (Lin 1998, Yang 2000)
Ø To deal with the co-occurring facts in (1), they propose that a quantifier NP in Chinese introduce a plural variable.
Ø dou is an overt realization of the (generalized) D(istributivity) operator, exerting universal quantification over a plural argument, following Dowty and Brodie 1984, Link 1987, and Schwarzschild 1991,1996:
(6) dou = dflP lX "y [yÎX®P (y)]
Ø Schematically, the composition of NP dou VP is as follows:
(7) $X [X is many/most/all NP] & "y [yÎ X®P (y)]
Problem: insufficient to rule out the dou-disharmony quantifiers in (2) because presumably all the NPs in (1) and (2) should be able to introduce a plurality and thus be quantifiable by dou, but they are not:
(8) $X [X is many/few NP] & "y [yÎ X®P (y)] Ömany… dou ; *few… dou
Ø In (8), The size of X matters!
3. Accounting for dou-disharmony
· Assumptions:
Ø dou is context dependent relative to one’s expectation, just like many. It is licensed only when the cardinality of the plural argument involved in a context is big enough to be considered as many.
Ø The context sensitivity of dou is presupposition.
· Proposal: modify the meaning of dou s.t the presupposition of dou is encoded. This is done by adding one more conjunct (underlined) to the meaning of dou, which guarantees that the size of X involved in a context must be above the contextually specified n.
(9) dou = dflP lX "y [yÎ X®P (y) & ½X½³n ], where n is contextually specified
Ø My intention of this addition here is presupposition.
· Illustrations with many and few
(10) My definitions of many and few a la Lin (1998), who gave credit to Yabushita (1989):
a. henduode ‘many’ = lPlQ$Z$X[P(X)&"Y(P(Y)àYÍX)&ZÍX&Q(Z)& ½Z½ ³ n]
b. henshaode ‘few’=lPlQ½Max-Z: [$X[P (X) &"Y(P(Y)àYÍX)] & ZÍX&Q (Z)]½<n (where n is contextually specified)
Ø Both many and few introduce a plural variable Z except that the cardinality of Z is bigger than n for many and smaller than n for few.
Ø For few, the plurality Z must be the maximal set of X in a context (see Krifka 1989 for discussion of decreasing quantifiers)
(11) *few…dou * henshao de laishi dou maile fang ‘Few teachers bought their houses.’
IP: |Max Z: $X [teacher’ (X) &"Y (teacher’ (Y)àYÍX)] & ZÍX
"y [yÎ Z®buy’ (y, house’) & ½Z½ ³ n] ½<n
VP: lXe "y [yÎ X®buy’ (y, house’) & ½ X½³n]
DP: lQ | MaxZ: $X[teacher’(X) &"Y (teacher’(Y)àYÍX)] VP : lx buy’(x, house)
& ZÍX &Q (Z) ½ <n
dou:
lPet lXe "y [yÎ X®P (y) & ½X½³n]
Few teachers bought their houses
Ø As shown in IP, the presupposition of dou requires the cardinality of Z should be equal or bigger than the context specified number n, but the semantics of few requires that the cardinality of the maximal Z should be smaller than n. Thus *few… dou is ruled out due to their conflicting expectation requirements.
This sort of conflict doesn’t arise in many…dou:
(12) Many teachers dou bought their houses => $Z$X [teacher’ (X)&"Y (teacher’ (Y)àYÍX) & ZÍX &"y [yÎ Z®buy’ (y, house’) & ½Z½³n] & ½Z½³n]
4. Supporting arguments
· Context dependent of dou vs. many/ few
(13) Controlling Situation 1 (CS-1): We need 6 students to open a graduate course.
If 10 registered, then dou and henduo ‘many’ can be used, but not henshaode ‘few’:
Ø I am concerned here with cardinal interpretations of many and few (see Partee 1988 for cardinal and proportional distinction)
(14) a. you shi ge xuesheng dou xuanle zhemen ke n=6, 10> n 6 Ö10… dou
exist 10 CL student all sign up ASP this CL course
‘There are ten students who signed up for the course.’
b. henduo de xuesheng xuanle zhemen ke Ömany
‘Many students signed up for the course’
c. henduo de xuesheng dou xuanle zhemen ke Ömany…dou
‘Many students signed up for the course’
d. henshao de xuesheng xuanle zhemen ke few(false)
‘Few students signed up for the course’
Under CS-1, if 10 registered, but the professor was expecting 20 students, then the reverse situation is expected:
(15) a. you shi ge xuesheng dou xuanle zhemen ke n=20, 10< n20, #10… dou
exist 10 CL student all sign up ASP this CL course
‘There are ten students who signed up for the course.’
b. henduo de xuesheng xuanle zhemen ke many (false)
‘Many students signed up for the course’
c. henduo de xuesheng dou xuanle zhemen ke many…dou (false)
‘Many students signed up for the course’
d. henshao de xuesheng xuanle zhemen ke Ö few
‘Few students signed up for the course’
Ø many and dou are used when the number of the plural argument involved meets or exceeds the context specified n and few is used when the number involved is below one’s expectation.
Ø What is understood as few and many in a context has nothing to do with the absolute value of a number. Rather, it is related to how the speaker will perceive it in a context.
Ø dou, like many and few, is context dependent relative to one’s expectation
· dou is a presupposition trigger
Under CS-1 and 10 registered, (16a) and (16b) are both felicitous and fine:
(16) a. you shi ge xuesheng dou xuanle zhemen ke with dou
exist 10 CL student all sign up this CL course
b. you 10 ge xuesheng xuanle zhemen ke without dou
(17) Presupposition of dou:
The number involved meets or exceeds the expectation of the speaker.
· Evidence for the presupposition
Ø If the presupposition is denied, dou results in infelicity as in (18a).
(18)# a. you 10 ge xuesheng dou xuanle zhemen ke, dan hai yuanyuan bugou.
exist 10 CL student all sign up this CL course but still far not enough
b. you 10 ge xuesheng xuanle zhemen ke, dan hai yuanyuan bugou.
‘There are 10 students who signed up for the course, but that is far from enough’
Ø The presupposition is projected in the following embedded contexts
(19) Negation:
you 10 ge xuesheng dou meiyou xuan zhemen ke
exist 10 CL student all not sign up this CL course
‘There are 10 students who didn’t sign up for the course’
(20) Modal
keneng you 10 ge xuesheng dou xuanle zhemen ke
possible exist 10 CL student all sign up ASP this CL course
‘It’s possible that 10 students have signed up for the course’
(21) Conditional antecedent:
ruguo you 10 ge xuesheng dou xuanle zhemen ke, women jiu
if exist 10 CL student all signup ASP this CL course, we then
zai 305 jiaoshi shangke
at 305 classroom have class
‘If there are 10 students who registered for this course, we will have class at room 305.’
5. Other quantifiers
How about quantifiers that are not context dependent? We will look at most, every/all and some.
· most
(22) Controlling Situation 2 (CS-2):
There are 20 students in this class and 18 passed the exam.
In this situation, most can be used regardless of one’s expectations:
(23) dabufen xuesheng tongguo le kaoshi most
big CLportion student pass ASP exam
“Most students have passed the exam.”
However, dou can also be used in this case:
(24) dabufen xuesheng dou tongguo le kaoshi
“Most students have passed the exam.” most…dou
The predicted difference between (23) and (24) is that when dou is used (24), 18 students passing the exam must be considered satisfactory by the speaker. This is captured as follows:
(25) dabufen = lPlQ $Z$X[P(X) &"Y(P (Y)àYÍ X) & ZÍX &Q (Z) ½Z½½X½-½Z½]
(Yabushita 1989 and Lin 1998)
(26) Most students dou passed the exam=> $Z$X [student’ (X) &"Y (student’ (Y)àYÍX)& ZÍX &"y [yÎ Z®pass’ (y, exam’) & ½Z½³n]&½Z½½X½- ½Z½]
Ø Crucially, in (26), the condition ½Z½³n must be satisfied.
· Evidence
Under the CS-2 above, if the professor had expected all 20 students to pass the exam, then 18 was still below his expectation. In this case, dou can’t be used, instead, zhi ‘only’ can be used to indicate that the number/percentage involved is below one’s expectation:
(27) Professor: a. zhi dabufen xuesheng tongguole kaoshi, dan bushi quanbu. Öonly… most
only big CLportion student pass ASP exam, but not all
“Only most students passed the exam, but not all.”
b. zhi 90% de xuesheng tongguole kaoshi, dan bushi quanbu. Öonly… 90%
“Only 90% of the students passed the exam, but not all.”
In relation to this, because many also requires that the number involved to be beyond one’s expectation. It predicts that many cannot co-occur with only. This is borne out.
(28) Professor: # zhi henduo xuesheng tongguole kaoshi, dan bushi quanbu. #only… many
only many student pass ASP exam, but not all
“Only many students passed the exam, but not all.”
· every/all
Ø mei ‘every’ and suoyoude ‘all’ are obligatory with dou as in (1a,d). It shows that dou is relevant to the proportional characters of the quantifiers as seen below:
(29) womenban zhiyou wuge ren dele A, dan meige dele A de tongxue
we class only 5-CL person get-ASP A, but every-CL get- ASP A DE classmate
dou hen gaoxing
all very happy only…every…dou
‘There are only 5 who got A in my class, but each of those who got A is happy.’
Ø The fact that 5 got A is not satisfactory, but what is satisfying is that 100% of them are happy.
Ø zhi ‘only’ is not allowed to co-occur with mei and suoyoude to express dissatisfaction:
(30) zhi *suoyoude/*meige / *100% de xuesheng dou tongguole kaoshi
only all/each CL /100% DE student all pass ASP exam
“Only all/every/100% students passed the exam” *only..all/every/100%
· some
yixie ‘some’ patterns with henshao ‘few’ in terms of dou-quantifiability. Given this, it indicates that yixie is not simply an existential quantifier like some. If we think it along the line of conversational implicature scale (Horn 1972):
(31) some<many<most<every, all
for a sentence in the form of D N V, the sentence with the D on the right of the scale is informationally stronger than the one with the D on the left of the scale, then dou-harmony suggests that anything that is as strong as many on the scale is quantifiable by dou. This seems to suggest that yixie means not many as in (32).
(32) a. * meiyou henduo ren dou lai. *not many… dou
not many person all comeASP
‘Not many people came.’
Cf b. meiyoy henduo ren lai.
not many person come
‘Not many people came.’
6. Conclusions
· dou carries a presupposition, relative to one’s expectation
· The meaning of dou is modified to encode the presupposition
· The modified definition of dou is successful to account for the dou-(dis)harmony facts
References
Brisson, Christine (1998) Distributivity, Maximality and Floating Quantifiers. PhD Dissertation. Rutgers University
Chierchia, G (1995) The Pluralities of Mass Nouns and the Notion of Semantic Parameter. ms. University of Milan.
Chierchia, G. (1998) “Reference to Kinds across Languages”, Natural Language Semantics 6, pp.339-405
von Fintel, Kai (1997) “Bare Plurals, Bare Conditionals, and Only,” Journal of Semantics 14 1-56
Horn, Laurence (1972) On the semantic properties of the logical operators in English. Bloomington, IN: Indiana University Linguistics Club
Huang, Shi-zhe (1996) Quantification and Predication in Mandarin Chinese: A case study of Dou. PhD Dissertation, Pennsylvania University
Krifka, M (1989) “Nominal Reference, Temporal Constitution and Quantification in Event Semantics,” In R. Bartsch et al (eds), Semantics and Contextual Expression. Dordrecht: Foris. 75-115.
Landman, Fred (1989) ‘Groups I’, Linguistics and Philosophy 12,559-605
Landman, Fred (1996)” Plurality” in The Handbook of Contemporary Semantic Theory, ed. S. Lappin, Blackwell, Cambridge.
Lasersohn, P. (1998) “Generalized Distributivity Operators”, Linguistics and Philosophy 21, pp 83-93
Lee, T. (1986) Studies on Quantification in Chinese, PhD dissertation, University of of California, Los Angeles