On the Absence of Nonrestrictive Relatives (in Chinese)
Niina Zhang Sept. 6, 2001
It has long been a controversial issue whether Chinese has nonrestrictive relatives or appositive relatives (Chao 1968, Tang 1979, Huang 1982, Tsao 1986: 34, etc.). The goal of this paper is to argue against the existence of nonrestrictive relatives in Chinese, and account for the absence in terms of the absence of relative pronouns in the language. More accurately, adopting Canac Marquis & Tremblay (1996, CT, hence), we argue that all relative clauses are restrictive and the so-called nonrestrictive relatives are DPs which have a coreferential relation with another DP and which contain a relative with a pro head. The presence of relative pronouns in the DPs licenses the pro syntactically to hold a coreferential relation with the antecedent DP. Since there is no relative pronoun in Chinese, the licensing is not available and thus DPs containing a relative with a pro head cannot have an apposition relation with other DPs in the language.
Although we show that certain constructions do not contain appositive expressions, we do not discuss their other properties in this paper.
Appositive expressions are divided into two types: those with an overt noun head, and those without, as shown in (1a) and (1b), respectively. In (1a) a woman is the noun head of
the relative clause whom Mary met at the party, and the whole DP a woman whom Mary met
at the party is an appositive expression, coreferential with the antecedent Jane. In (1b), the
appositive expression whom Mary met at the party does not have a noun head.
(1) a. Jane, a woman whom Mary met at the party, is brilliant.
b. Jane, whom Mary met at the party, is brilliant.
I will call the type of appositive expressions represented by (1a) Headed Appositive (HA), and the type represented by (1b) Null-Head Appositive (NHA).
CT claim that all relatives are restrictive, and those like the one in (1b) has a pro head. According to them, all appositive expressions are nominals, either in the form of a simple DP, like the underlined part in (2), or a DP which contains a relative clause, as in (1) .
(2) John, an absent-minded man, came late.
Obviously, simple nominal appositives like (2) and HAs like (1a) also occur in 1Chinese, as shown in (3a) and (3b), respectively:
(3) a. Bolin, zhe ge Ouzhou chengshi, you bashi duo jia bowuguan.
Berlin, this cl Europe city have eighty more cl museum
„There are over eighty museums in Berlin, the European city.‟
b. Bolin, zhe ge [ renren dou zhidao] de chengshi, you bashi duo jia bowuguan. CP
Berlin this cl everyone all know de city have eighty more cl museum
„There are over eighty museums in Berlin, the city that everyone knows.‟
Now the question is whether there are NHAs in Chinese. In next two sections we discuss in turn whether the Coda in the so-called Existential Coda Construction is NHA, and whether one of the two versions of relatives, one of which is nominal-initial and the other is
1 The abbreviations used in the Chinese examples are: EXP: experience aspect, PRF: perfect aspect, PRT: sentence-
final aspect particle, CL: classifier, DE: modification marker.
not, is NHA. We then look for an account for our conclusion in section 4. Section 5 is a short summary.
2. Are the Codas in ECCs NHAs?
The existential coda construction (ECC) refers to the construction in (4), where the NP must be indefinite and the XP (the underlined part in the data), called coda, is a clause modifying the NP.
(4) a. YP - V – NP – XP
b. Jie-shang lai-le yi ge xiaohair mei chuan xie.
street-on come-prf one cl child not wear shoe
„On the street has come a child who does not wear shoes.‟
b. Baoyu jiao-guo yi ge xuesheng hen wanpi.
Baoyu teach-prf one cl student very naughty
„Baoyu taught a student who was very naughty.‟
The construction has been discussed in Li & Thompson (1981: 611-619 “Realis
Descriptive Clause Construction”), Huang (1987), McCawley (1988: 451 cf. English "Pseudo-Relatives", 1989), and Tsai (1994/1999), among others. 2 The coda in the ECC is different from a NHA in at least four respects.
; 2.1 The Restrictiveness
The codas of ECC can restrict the meaning of their antecedents (Wilder 2000), whereas NHAs never do so. As it is well-known that removal of an appositive does not affect the content of the assertion. For instance, (5a) has an appositive DP na ben wo zuotian mai de shu „the book
that I bought yesterday‟, whereas (5b) does not. However, they make the same assertion, i.e.,
the book on the table is very interesting.‟ However, as shown below, the ECC in the a-
sentences and the corresponding b-sentences, where the coda is absent, make different assertions. For instance, (6a), an ECC, asserts that there is a blank page in the book, whereas (6b), which differs from (6a) in not containing the coda shi kongbai „be blank‟ and is thus not
an ECC, asserts that there is one page in the book. This is a semantically abnormal assertion, since a book cannot have only one page. The same assertion contrast is seen in the other pairs.
(5) a. Zhuo-shang de na ben shu, na ben wo zuotian mai de shu, hen youqu.
table-on de that cl book that cl I yesterday buy de book very interesting
'The book on the table, the one I bought yesterday, is very interesting.'
b. Zhuo-shang de na ben shu hen youqu.
table-on de that cl book very interesting
'The book on the table is very interesting.'
(6) a. Zhe ben shu you yi ye shi kongbai. ； b
this cl book have one page be blank
„There is one blank page in this book.‟
b. #Zhe ben shu you yi ye.
this cl book have onc page
„This book has one page.‟
(7) a. Zhe jia shangdian shang xingqi lai-le yi wei guke tebie tiaoti. ； b
this cl shop last week come-prf one cl guest very picky
„In this shop came a guest last week who was very picky.‟
2 I do not use the possibility of stacking as a test. Jackendoff (1977: 171) and McCawley (1988: 419) claim that HlessACs cannot stack. However, see Grosu and Landman (1998) for an opposite view. Vries (1999:4) also finds that appositives can stack in Dutch.
b. Zhe jia shangdian shang xingqi lai-le yi wei guke.
this cl shop last week come-prf one cl guest
„In this shop came a guest last week.‟
(8) a. Zhe ge senlin-li zhang-zhe yi ke shu hen qimiao. ； b
this cl forest-in grow-prg one cl tree very intriguing
„In this forest there is a tree which is very intriguing.‟
b. #Zhe ge senlin-li zhang-zhe yi ke shu.
this cl forest-in grow-prg one cl tree
„In this forest there is one tree.‟ 3 ； (9) a. Jia-li conglai mei lai-guo yi ge keren neng bodao fuqin.
home-in ever not come-exp one cl guest can convince father
„In my home there has never been a guest who can convince my father.‟
b. Jia-li conglai mei lai-guo yi ge keren.
home-in ever not come-exp one cl guest
„In my home there has never been a guest.‟
(10) a. Ta-mama mei sheng-guo yi ge haizi bu shi jinshiyan. ； b
he-mother not bear-exp one cl child not be short-sighted
„His mother never gives birth to a child who is not short-sighted.‟
b. #ta mama mei sheng-guo yi ge haizi.
he-mother not bear-exp one cl child
„His mother never gives birth to a child .‟
We thus conclude that codas of the ECC restrict the meaning of their antecedents and therefore cannot be NHAs.
; 2.2 Intonation
NHAs do not integrate into the intonation of the matrix clauses (Jackendoff 1971: 62, McCawley 1988: 418), while the coda of the ECC does. The ECC “is pronounced with one single intonation contour” (Li & Thompson 1981: 614, See also Huang 1987: 235).
; 2.3 Position in the matrix clause
NHAs can occur in the middle of the matrix clauses, as in (1b) above, while the coda of the
ECCs must occur at the right peripheral position of the whole construction.
(11) Ta song-le yi ben shu (*hen youqu) gei wo.
he give-ASP one CL book very interesting to I
„He gave a book to me (which is very interesting).‟
; 2.4 Adjacency to the antecedent
NHAs must be adjacent to the antecedents (McCawley 1988: 449, CT p. 7), whereas the coda
in the ECC does not need to (Huang 1987: 231). In (12b), the relative who Mary knows is not
adjacent to its head a man, and the sentence is acceptable. In contrast, in (13b), the NHA who
Mary knows is not adjacent to its antecedent Peter, but the sentence is not acceptable. Like (12b) and unlike (13b), when the coda hen youqu „very interesting‟ is not adjacent to its antecedent yi ben shu „a cl book‟ in (14), the sentence is acceptable.
3 Huang (1987: 249) and Tsai (1999: 173) claim that the matrix predicate in ECC cannot be negated. The data in
(9) and (10) and the example in McCawley (1989: 38) show that the claim is not true.
(12) a. A man Mary knows came.
b. A man came who Mary knows.
(13) a. Peter, who Mary knows, came.
b. *Peter came, who Mary knows.
(14) Ta song-le yi ben shu gei wo hen youqu.
ASP one CL book to I very interesting he give-
„He gave a book to me which is very interesting.‟
The above four points show us that the coda in the ECC does not exhibit properties of NHAs.
3. Are Non-Peripheral Relatives NHAs?
Chao (1968: 286) claims that nominal-initial relatives are restrictive, as in (15a), whereas non-peripheral relatives are descriptive, as in (15b).
(15) a. [dai yanjing de] na wei xiansheng (restrictive)
wear glasses de that cl gentleman
„the gentleman who wears glasses‟
b. na wei [dai yanjing de] xiansheng (descriptive)
that cl wear glasses de gentleman
The restrictive-descriptive contrast of modifiers with respect to the position of relatives, on the one hand, is denied in Tang (1979) and Tsao (1986: 15). On the other hand, it 4is accepted in Hashimoto (1971) and Huang (1982: 68). The general restrictive-descriptive
contrast, beyond the position of relatives, nevertheless, is discussed in other contexts (see Sproat & Shih 1988, Lu 1998). However, whether this position contrast is correspondent to the contrast between restrictive relatives and NHAs has never been clarified. Recently, Gobbo (2001) challenges such a correspondence, claiming that “in Chinese the distinction between
appositive and restrictive does not actually exist,” and “in Chinese relative clauses can only be restrictives.” (p. 2) In sections 3.1 through 3.4, I discuss some of the facts listed by Gobbo, and in sections 3.5 and 3.6 I add two new arguments, in addition to my arguments with respect to the ECC presented in the last section, to support her claim that relatives in Chinese, regardless of whether they take a nominal-peripheral position or not, are all restrictive.