Idioms-and-Compositionality.pdf
《Idioms-and-Compositionality.pdf》由会员分享,可在线阅读,更多相关《Idioms-and-Compositionality.pdf(14页珍藏版)》请在三一文库上搜索。
1、Idioms and Compositionality Dag Westerst ahl March 31, 1999 Abstract A standard view is that idioms present problems for compositionality. The ques- tion of compositionality, however, should be posed for a semantics, not for individual phrases. The paper focuses on the idiom extension problem: Suppo
2、se in a given lan- guage a certain phrase acquires the status of an idiom.How can the syntax and semantics be extended to accommodate the idiom, while preserving desirable proper- ties such as compositionality? Various ways to achieve such extensions are discussed within an abstract algebraic framew
3、ork for language due to Wilfrid Hodges. Contents 1Introduction2 2A methodological point2 3Real issues about idioms and compositionality4 4The idiom extension problem4 5An algebraic framework5 6Atomic extensions7 7Paraphrase semantics8 8How to handle new idiomatic meanings8 9Idioms with syntactic str
4、ucture10 10 Discussion11 11 Further directions13 1 1Introduction If there is one piece of received wisdom in the linguistic community about id- ioms, it probably is that they spell trouble for the principle of compositionality. After all, the meaning of a typical complex idiom is not determined by t
5、he meaning of its parts and the way they are composed, right? If you know En- glish, and in particular the meaning of kick and the bucket, but are unfamiliar with the idiom, there is no way you can compute the (idiomatic) meaning of kick the bucket. Like lexical items, idioms have to be learned one
6、by one. Yet they appear to have syntactic structure, so compositionality is in trouble. Or is it? In the present paper I want to do two things. First, I claim that the received view as commonly expressed is misleadingly put, and may confuse the issues. Second, a handy algebraic framework (due to Wil
7、frid Hodges) will be used for a fi rst stab at approaching some of the formal semantic issues that do exist concerning idioms and compositionality. Details and proofs must be deferred to another occasion; here I try to state the main ideas. There is no way I could do justice to the vast linguistic l
8、iterature about idioms. Instead I will take one paper, Nunberg, Sag and Wasow 4, as back- ground, because it surveys a fair amount of that literature, and furthermore, it propounds a general view of idioms that seems reasonably widespread, at least in its broad outlines, these days. More precisely,
9、Nunberg et. al amply exemplify with many quotations what I called the received view about idioms and compositionality, and for some idioms such as kick the bucket, they too subscribe to that view. On the other hand, for many other idioms, of which we can take pull strings as a prototype, they claim
10、that compositionality does hold, provided one realizes that the parts or chunks of such idioms also have idiomatic meanings. 2A methodological point A simple methodological point needs to be made, I think, in connection with the received view about idioms: While it makes good sense to ask if a seman
11、tics is compositional or not, it makes no sense to ask the same question about a particular phrase. Consider how the received view might be put forward, indeed how it might seem self-evident: The meaning of kick the dog, say (kick the dog), is a function, say r, of the meanings of kick and the dog:
12、(1) (kick the dog) = r(kick),(the dog). But, for the idiomatic reading, (2) (kick the bucket) 6= r(kick),(the bucket). Yes, but so what? Compositionality only asks that some function determine the meaning, not that this function be r. Simply redefi ning r for the idiomatic case will do the trick. 2
13、This is trivial, but ask any semanticist what the principle of compositionality says, and you will most likely get something like (C) The meaning of a complex expression is determined by the meanings of its parts and the mode of composition. And here is determined is taken as is a function of, possi
14、bly with some added requirement of computability of that function.But that extra requirement makes no diff erence to the present issue. One idiomatic exception (hence any fi nite number) can always be taken care of, it seems. Perhaps it is retorted that the claim about kick the bucket is in fact int
15、ended as a counterexample to the compositionality of a particular semantics. This could be spelled out as follows. Consider the relation of synonymy: p q iff (p) = (q). Assume that bucket pail. Then (3) lift the bucket lift the pail, as compositionality would have it, but (4) kick the bucket 6kick t
16、he pail. contradicting functionality. This might look better, but actually it only reveals another trivial point. Meanings cannot be assigned to surface manifestations, because then kick the bucket would be ambiguous. Now, ambiguity is not our problem here. Any state- ment of the compositionality pr
17、inciple already presupposes single-valuedness of meaning. And indeed, as writers about idioms generally acknowledge, on the level where meaning is assigned the idiomatic version of kick the bucket must be distinct from the ordinary version, if only by the presence of an idiomatic marker. But once th
18、is is acknowledged, the argument loses all force. For then, the modes of composition in (1) and (2) are distinct, so there is no reason to expect the same function to operate in both cases. Similarly, on both sides of the synonymy sign in (3), and on the right hand side in (4), we have one mode of c
19、omposition, but on the left hand side in (4) we have another, so no violation of compositionality occurs. Perhaps what the received view really comes down to, in the case at hand, is rather something like this: There is a familiar way to compose a transitive verb meaning with a NP meaning to form a
20、VP meaning, but (a) that way is not used to derive the idiomatic meaning of kick the bucket, and (b) no other familiar way to compose the meanings of kick and the bucket gives the correct result. The familiarity of the function or rule is required to explain how we can fi gure out, or know, or under
21、stand the meaning of a complex phrase. This question is worth discussing, but it is hardly one about compositionality. Likewise, when Nunberg et al. (appear to) claim that for idioms like pull strings we actually can use the familiar meaning function, given that in the idiom, pull and string have no
22、n-standard meanings, this is an interesting issue, but it is still misleading to call those idioms compositional and contrast them with non-compositional ones. The question is rather: Is there a reasonable compositional semantics for a language containing both kinds of idioms? 3 3Real issues about i
23、dioms and compositionality Assume again that bucket pail, and suppose we have agreed on a particular mode of composition, or rule, or marking, that is at work in the idiomatic version of kick the bucket. Can we then avoid, except by ad hoc stipulation, that the same mode of composition is applied to
24、 yield also an idiomatic version of kick the pail? If not, then, by compositionality, the latter has to mean die as well. This is a real issue, an instance of the overgeneration problem. Note the diff erence from the alleged argument above. There we (falsely) claimed to have a counter-instance to co
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- Idioms and Compositionality
链接地址:https://www.31doc.com/p-3769091.html