Causatives and accomplishments: The case of Chinese

Causatives and accomplishments: The case of Chinese

Reviews 292 tut es nun doch, und zwar in der Uberzeugung, dal3 es sich bei dieser Arbeit urn ein Paradebeispiel fur ein in sich geschlossenes dedukt...

509KB Sizes 2 Downloads 58 Views

Reviews

292

tut es nun doch, und zwar in der Uberzeugung, dal3 es sich bei dieser Arbeit urn ein Paradebeispiel fur ein in sich geschlossenes deduktivistisches und zirkulares System handelt, das - obwohl es mit hohem Anspruch und durchaus mit gewissen stilistischen Qualitaten daherkommt - in absoluter Empiriefeme verharrt und damit aus wissenschaftstheoretischer Sicht als interessanter Fall gelten mug.

Literatur Guillaume, G., 1965 [1929]. Temps et verbe. ThCorie des aspects, des modes et des temps. Paris: Champion. Helbig,

G., 1987. Zur Klassifizierung

Zustandspassiv). schen:

M., 1990. Zeitaspekte

tionen. Jakobson,

Wiesbaden:

(Hrsg.),

1990, Bd. 1, 97-104. Lobner,

tlber des Passiv

Writings

215233. Aspekt

II (Was

Das Passiv Tiibingen:

ist ein

im DeutNiemeyer.

und temporalen

Konjunk-

Tubingen: zu einer

Mouton.

aus linguistischer

- Akten

Sicht.

des 25. Linguistischen

In: R. Pogarell, Kolloquiums,

C. WeiR Paderborn

Niemeyer. integralen

Aktionsart. In: U. Ehrich, H. Vater (Hrsg.), Zeitreferenz. Tiibingen: Niemeyer.

SSDI 0024-384

im Deutschen, von Tempus,

II. Den Haag/Paris: Kants

der Linguistik

S., 1988. Ansltze

Germanique,

Universitatsverlag.

1991. Die Urteilsarten

Neue Fragen

mit sein + Partizip

en Linguistique

- Die Bedeutung

Deutscher

R., 1971. Selected K.-P.,

der Konstruktionen

de Recherche

Akten des Kolloquiums

Herweg,

Lange,

In: Centre

semantischen

Theorie

Temporalsemantik

von Tempus,

- Beitrlge

Aspekt

und

zur Linguistik

der

1(93)E0026-4

Rint Sybesma, Causatives and Accomplishments: The Case qf’ Chinese ba. Holland Institute of General Linguistics, 1992. xii + 205 pp. Reviewed by Grant Goodall, Department of Languages and Linguistics, University of Texas at El Paso, El Paso, TX 79968, USA. The ba-construction is probably one of the most well-known features of Mandarin Chinese syntax. It involves the positioning of the direct object, together with the word ha, to the left of the verb, as in (1). (1) Wo ba Zhang San da-si-le. hit-die-LE I BA ‘I killed Zhang San.’ This is in contrast

to the canonical

SVO order (without

ha) seen in (2).

293

Reviews

(2)

Wo da-si-le Zhang San. I hit-die-LE ‘I killed Zhang San.’

The reason for this construction’s fame is that it has been one of the most popular topics within Chinese linguistics for several decades now, within traditional and structuralist grammars as well as within generative and functionalist accounts. Given this background, Rint Sybesma’s new book might seem like just one more tiresome treatise on an already overworked subject, but in fact it is a refreshingly original and provocative contribution to an area of study which has been badly in need of new ideas. I. Background

The previous there sometimes

literature on the ha-construction is particularly frustrating because appears to be little agreement even on what the basic facts are, let

alone on the analysis. One prominent source (Wang 1970), for instance, gives (3) as the canonical example of the ha-construction, while others (and many native-speaker informants) take this very same sentence to be ungrammatical! (3) Wo ba Zhang San da-le. I BA hit-LE ‘I hit Zhang San.’ Despite many inconsistencies such as these, there is general agreement in the literature that patterns (1) and (2) are not simply in free variation; (1) is much more restricted in its occurrence than (2) is. For example, there is some sort of definiteness requirement on the ha-NP (the NP following ha), as seen in the contrast between (4) and (5) (data from Teng 1975). (4) Ni qu ba nei-zhi ji sha-le. you go BA that-M chicken kill-LE ‘Go and kill that chicken.’ (5) *Ni qu ba yi-zhi ji sha-le. you go BA one-M chicken kill-LE ‘Go and kill a chicken.’ There is quite a lot of disagreement in the literature on exactly what this consists of, but its existence is widely acknowledged. Sybesma adopts the ha NP must be definite, generic, or specific (i.e., ‘strong’ in the sense of Cooper 1981). Another restriction on the ha-construction is that the ha-NP must be the verb. Thus sentences like (6) are disallowed (from Li and Thompson

requirement view that the Barwise and ‘affected’ 1981).

by

294

Reviews

(6) *Ta ba ni xiang. he BA you miss ‘He misses you.’ Even though ni ‘you’ is the direct object of the verb here, it is not ‘affected’ enough by the verb to be allowed in the ha-construction. Finally, there appears to be a strong preference for the verb in the ba-construction to be complex,

as seen in the contrast

(6) *Wo I

ba ta BA he ‘I hated him.’ (7) Wo ba ta I BA he ‘I thoroughly A locative

between

(6) and (7) (from Y.-H. A. Li 1990).

hen-le. hate-LE hen-tou-le. hate-thorough-LE hated him.’

complement

can also satisfy

this complexity

requirement,

as seen in (8).

(8) Zhuge Liang

ba Lusu qing dao chuan-li BA invite to boat-in ‘Zhuge Liang invited Lusu into the boat.’

Trying to pin down this notion of verbal complexity

has been notoriously

difficult,

but

it seems clear that some such requirement, or at least the tendency, exists. Within the generative literature on the ha-construction (see Huang 1982, Travis 1984, Koopman 1984, Goodall 1989, and Y.-H. A. Li 1990 for some GB examples), much emphasis has been placed on the relationship between (1) and (2) and on whether the ba-NP moves into its surface position or is base-generated there. These studies have had very little insightful to say about the restrictions noted above in (4j(8). 2. Theoretical

framework

Sybesma’s work stands out in that first of all he makes a sincere attempt to sort out the very complex data, based both on his own field work and on a comprehensive survey of the literature. Second, he makes use of a theoretical framework which allows him to address seriously the issue of the restrictions on the ha-construction. This framework, which is based on recent work by Hoekstra (e.g., Hoekstra 1988) and is related in spirit to current work by a number of others, essentially eliminates Cl-theory as such and rejects the idea that syntactic structure is projected from lexical structure. Instead, syntactic structure is taken to be primary. Lexical insertion takes place more or less freely, and the resulting sentence is interpreted based on both the lexical items

Reviews

295

and the aspectual properties of the syntactic configuration. This model predicts a certain fluidity or ‘dynamism’ in the syntactic behavior of lexical items which is not expected under the more standard view. For instance, verbs such as laugh which describe an event requiring only one participant are thus typically intransitive, as in (9a), but they may also appear with a small clause complement, as in (9b), because this configuration is associated with results. (9a) John laughed. (9b) John laughed himself silly. Since (9b) has a coherent interpretation (‘John laughed to such an extent that he became silly’), the sentence is allowed. This dynamism in the syntactic environment in which laugh may appear is not expected if the structure is wholly projected from the O-grid of laugh. The structure in (9b) with a resultative small clause is taken to be the usual structure for accomplishments (accomplishments like John knit the sweater are assumed to have a covert small clause). The subject John is not thematically related to laugh (since there is no B-theory in this view), but is interpreted as the causer of the following VP. The fact that John is interpreted as the one doing the laughing is referred to as a ‘shadow interpretation’, i.e., our knowledge of how the world works tells us that if John is the causer of laughter resulting in his becoming silly, then John is probably the one doing the laughing. Direct evidence for this comes from similar structures in Chinese: (10a) Zhang San ku-de lei-le. (ch. 6, (Sb)) cry-DE tired-LE ‘Zhang San got tired from crying.’ (lit. ‘Zhang San cried (himself) tired.‘) (lob) Zhei-jian shi ku-de Zhang San lei-le. (ch. 6, (4b)) this-M tired-LE case cry-DE ‘This thing got Zhang San tired from crying.’ (lit. ‘This thing cried Zhang San tired.‘) (lOa) is similar in structure and interpretation to English (9b), but in (lob) the matrix subject is very clearly thematically unrelated to the verb. The possibility of sentences like (lob) is just what is predicted by Sybesma’s analysis, in that zhei-jian shi ‘this thing’ is interpreted as the cause of the following VP. Since it is inanimate, our realworld knowledge tells us that it cannot be doing the crying, zo Zhang San is probably doing it instead. Notice that (lob) is extremely problematic under a O-theoretic view; we would expect that ku ‘cry’ would assign its external B-role to the subject in (lob) just as it does in (lOa), but it does not.

Relsiews

296 3. Accounting for the ba-construction The structure

which Sybesma

NP

thus assigns

to accomplishments

is given in (11).

CausP Caus

VP V

result clause

The verb incorporates into Caus. For Chinese, however, Sybesma claims that there is another possibility. Caus can be filled by ha, and the subject of the result clause adjoins to VP, where it then receives Case from ha. Sybesma thus follows the tradition of movement analyses of the ha-construction, but unlike these previous analyses, his can shed some light on the restrictions discussed above in (4)-(8). Sentences like (5) are disallowed in his account, for example, because ha occurs with accomplishments. Accomplishments are bounded, and bounded events, he argues, require a ‘strong’ object. This effect can be seen independently of the haconstruction with any accomplishment predicate: (12)

Wo he-guang-le tang. (ch. 6, (45a)) I drink-up-LE soup ‘I finished the/*some soup.’

As for the affectedness requirement, this follows from the fact that an accomplishment always has a result clause, which essentially tells how the NP is affected by the action of the main verb. Again since ha-sentences are always accomplishments, the baNP will always be affected in some way by the matrix verb. The tendency for the verb to be complex in the ba-construction also receives an explanation. The verb in the obligatory result clause may incorporate into the matrix verb. Since this is a common process, most ba-sentences will have a complex verb. It is possible for the result clause to not be overtly realized, however, so in these cases a ‘bare’ verb will show up. 4. Remaining problems To a much larger extent than other analyses, then, Sybesma succeeds in accounting not just for the basic syntactic facts of the ha-construction, but for some of its more subtle, seemingly intractable aspects as well. The analysis is fully laid out only in the final chapter, though, and although many of its major implications are discussed, there

297

Reviews

are several important loose ends which don’t get the treatment they deserve. For example, Sybesma argues that with ‘bare’ ha-sentences such as (13), there is a covert result clause which is what makes the sentence possible. (13)

Ta ba zhu mai-le. he BA pig sell-LE ‘He sold the pigs.’

(ch. 6, (49a))

He points out, using facts from Smith (1990), that Chinese do not have an accomplishment reading, as seen in (14). (14)

verbs in and of themselves

mai-dao. Wo mai-le yi-ben shu keshi mei not.have buy-succeed I buy-LE one-M book but ‘I tried to buy a book, but I did not succeed.’

That (13) is interpreted

as an accomplishment

thus must be due to the result clause in

his analysis. But what determines when a null result clause is possible? apparently not possible to have one in (1.5) (from Y.-H. A. Li 1990)? (15)

Why is it

*Wo ba ta hen-le. I BA he hate-LE ‘I hated him.’

That this verb can take an overt result clause is shown in (16) (assuming Sybesma’s analysis in which elements such as tou have been raised from a lower clause). (16)

Wo ba ta hen-tou-le. I BA he hate-thorough-LE ‘I hated him thoroughly.’

Thus once Sybesma allows for the possibility of null resultatives in sentences like (13), it is not clear how he can prevent them from occurring in sentences like (15). Even with resultatives which are overt, in Sybesma’s sense, it is sometimes unclear whether he has considered all the consequences of their structure. He gives examples such as (17) and argues, as seems plausible, that Z/rang San and fuqin ‘father’ form a single constituent underlyingly. (17)

Lao Li ba Zhang San sha-le fuqin. kill-LE father BA ‘Lao Li killed Zhang San’s father.’

Now suppose we use the complex verb da-si ‘hit-die’ (i.e. ‘kill’) in place of sha, as is possible. Under Sybesma’s analysis si ‘die’ has been raised up from the embedded

298

Reviews

result clause. The only potential structure for this clause which makes sense is one in which si ‘die’ is the predicate and Zhang San de fuqin ‘Zhang San’s father’ is the subject (or object, if si is ergative). Si ‘die’ then incoporates into the higher verb and Zhang San is raised also. This latter raising is deserving of comment, at least, by Sybesma, because it involves movement of something like the subject of the embedded clause, which leads one to wonder, among other things, how the trace of this movement is properly governed. Another potentially troublesome area which requires some discussion is the hei-construction, or passive, as in (18). (18) Zhang San bei wo da-si-le BE1 I hit-die-LE ‘Zhang San was killed by me.’ There is good evidence that this involves movement of an NP (Zhang San in (18)) into subject position (see Y.-H. A. Li 1990, for example), analogous to the movement of Zhang San in (1) (repeated here as (19)). (19) Wo ba Zhang San da-si-le. I BA hit-die-LE ‘I killed Zhang San.’ Moreover, this movement appears to occur under essentially the same conditions as what we saw in the ba-construction (see Li and Thompson 1981), thus strongly implying for Sybesma’s analysis that the hei-construction requires the same syntactic structure as the ba-construction (i.e., (11)). The main difference between the two constructions then would be that the subject of the result clause moves to matrix subject position in the hei-construction and adjoins to VP in the ha-construction. This seems quite plausible, but given that the result clause presumably has only one subject, we would then not expect to find a sentence with both hei and ha, in which two NP’s have moved. Such sentences do exist, however, as shown in (20) (from Li and Thompson 1981). da-po-le. (20) Wo bei ta ba wode daziji BE1 he BA my typewriter hit-broken-LE I ‘What happened to me was that my typwriter was broken by him.’ Under Sybesma’s analysis, this sentence would presumably have a result clause with po ‘broken’ as the predicate and wode daziji ‘my typewriter’ as the subject, but where does wo ‘I’ originate? Perhaps the most glaring problem for Sybesma’s analysis is that it claims that ba and the ha-NP do not form a constituent. Ba occupies the Caus head position and ba-NP is adjoined to VP. Almost all previous analyses have assumed that these two elements do

Reviews

299

form a constituent, although they have apparently taken the point to be so obvious that they in general have not provided explicit motivation for it. Evidence for constituenthood does exist, however, as shown by Y.-H. A. Li (1990). The examples are difficult to construct, but Li demonstrates that ba and the following NP may be conjoined with a PP with marginally acceptable results, as shown in (21) (the less-than-perfect acceptability is not surprising, given that the first of the purported PP’s is an adjunct and the second is more complement-like). (21)

‘!Ni you wei ta you ba ta qiang qian, shi sheme yisi? you and for him and BA him steal money, is what meaning ‘You stole money for him and from him; what does this mean?’

Under standard assumptions, this shows that the ba plus ba-NP sequence is a PP. Most of the recent literature making extensive use of functional categories pays very little attention to traditional constituenthood arguments, but partly this is because it is not clear how they could be applied to the proposed structures. In the case of Sybesma’s analysis, however, the categories in question are overtly realized, so traditional tests can be applied, and the non-constituent analysis proposed is sufficiently counterintuitive that some discussion is clearly needed. Other points that are left hanging at the end of the book include details concerning the definiteness requirement on the ha-NP. Sybesma discusses the difference in interpretation between (22a) and (22b), but as he himself points out, it is not obvious how the difference follows from his analysis. (22a)

(22b)

Li laoshi gai-le ji-fen zuoye. (ch. 5, (24)) teacher correct-LE few-M homework ‘Teacher LI has corrected a few pieces of homework.’ Li laoshi ba ji-fen zuoye gai-le. teacher BA few-M homework correct-LE ‘Teacher Li has corrected the few pieces of homework.’

Also, given that much of Sybesma’s data are sentences in which one verb has incorporated into another, it is surprising that there is no mention of Y. Li’s (1990) work on V-V compounds in Chinese. This is all the more surprising since Sybesma’s impressive coverage of the previous relevant literature has obviously been meticulously researched. In any event, since Y. Li (1990) explicitly adopts B-theory and Sybesma explicitly rejects it, a careful comparison of the two accounts would clearly be instructive. 5. Conclusion Despite these and other unresolved questions, there is no doubt that Sybesma’s book is an important contribution both to the study of Chinese syntax and to general syntac-

300

Reviews

tic theory. On the Chinese side, the book provides, arguably for the first time, an explanation for some of the puzzling restrictions on the ha-construction which have proven elusive to many previous researchers. Although I have not discussed it in this review, there is also a valuable discussion of word order and complement structure in Chinese. There is little question, then, that this book should be required reading for anyone working on the syntax of Chinese. On the general theoretical side, the book also has much to offer. Its basic premise, that syntactic structure is not projected from @structure, is provocative, perhaps even unsettling, but it is remarkable to what extent it allows one to come to grips with facts from Chinese which could previously only be described as bizarre (sentence (lob) is a good example). In fact, these phenomena in Chinese seem so natural within the framework adopted that one is left wondering, as Sybesma points out, why such facts are not more prevalent or apparent in Western languages. Much work thus obviously remains to be done on this question, but the efforts by Sybesma and others to deny that syntactic structure is projected from lexical structure seriously challenge one of the guiding assumptions of generative syntactic theory over the last I5 years or so (and not just in GB; see Wasow 1985). Whether these efforts are ultimately

sucessful

or not, we are likely to learn much in the process.

Barwise, J. and R. Cooper, 1981. Generalized Philosophy Goodall,

quantifiers

and natural

language.

Linguistics

and

4, 159-219.

G., 1989. Evidence

for an asymmetry

in argument

structure.

Linguistic

Inquiry

20,

669-674. Hoekstra, Huang,

T., 1988. Small clause results. C.-T. J., 1982. Logical

tation,

MIT, Cambridge,

Koopman,

relations

CA: University Li, Y., 1990. 117-201.

74, lOlll39. and the theory

of verbs. Dordrecht:

1981. Mandarin

of California

A functional

On V-V compounds

in Chinese.

Natural

in Mandarin Linguistics

Travis,

1975. A semantic in Linguistics

L., 1984. Parameters

study of transitivity 80). Berkeley, and effects

Cambridge, Massachusetts. Wang, P. C.-T., 1970. A transformational

CA: Center

SSDI 0024-3841(93)E0032-3

reference

grammar.

Berkeley,

of word approach

for the Study of Language

Language

Chinese.

and Linguistic

Dordrecht:

Theory

8,

Kluwer.

28, 3099336.

relations

CA: University

tion, University of Texas at Austin. Wasow, T., 1985. Postscript. In: P. Sells, Lectures Stanford,

disser-

Press.

Smith, C., 1990. Event types in Mandarin. Publications

Doctoral

Foris.

Chinese:

Li, Y.-H. A. 1990. Order and constituency Teng, S.-H.,

of grammar.

Massachusetts.

H. 1984. The syntax

Li, C. and S. Thompson,

Lingua

in Chinese

order

in Chinese of California variation.

to Chinese

(University

of California

Press.

Doctoral

dissertation,

BA and BEI. Doctoral

on contemporary and Information.

syntactic

theories,

MIT, disserta-

193-205.