LINGUIST List 4.1016

Tue 30 Nov 1993

Review: Syntax of Scope

Editor for this issue: <>

Directory

Message 1: Review of The Syntax of Scope

Date: Mon, 29 Nov 93 10:22 CST
From: <bcjtamuts.tamu.edu>
Subject: Review of The Syntax of Scope

Book Editor's note:

Book reviews on Linguist are intended to initiate discussion of books and/or of topics dealt with in them. We hope the review will make it clear enough what the book is about that people who have not yet had a chance to read it will be able to join in along with people who have. Please play by the normal scholarly rules; book discussions are occasions for exchanges of and about ideas rather than for carrying out personal vendettas, etc. We look forward to some lively discussion.

If you are interested in writing reviews, stay tuned for our announcements of books available for review.

********************** Joseph Aoun and Yen-hui Audrey Li. 1993. SYNTAX OF SCOPE. Linguistic Inquiry Monograph No. 21. Cambridge: MIT Press.

Reviewed by A. Kroch:krochchange.ling.upenn.edu

This book is a comparative study of the syntax of quantifier and wh- operator scope in Chinese and English, written in the framework of transformational grammar as formulated in the late 1980's. The book's purpose is to develop, through the comparative analysis of English and Chinese, the theory of Logical Form (LF) as a syntactic level of representation. It follows a line of research initiated by Robert May in the middle 1970's and developed further by many others. As such, it should be of interest to a wide range of linguists, especially those working on the syntax-semantics interface; but the discussion on Linguist will benefit particularly from the comments of specialists in the syntax of Chinese and native speakers of the language who feel comfortable giving quantifier scope judgments. The analysis depends to a considerable extent on the accuracy of Aoun and Li's description of quantifier and wh- operator scope interaction in Chinese, which many readers, me among them, will be unable to evaluate. Also, a descriptive question of considerable complexity and interest that only experts in Chinese can hope to shed light on is the effect on scope interactions of the of the presence or absence of the mysterious adverb "dou". Beyond the domain of Chinese and English, it would be interesting to know how much variation in scope ambiguity behavior there is across languages generally; that is, what happens when the crucial examples discussed by A&L are translated into the languages that Linguist subscribers work on or speak natively.

The SYNTAX OF SCOPE is an elaboration of earlier comparative studies of Chinese and English by the authors and others. The first chapter discusses the relative scope of quantifier phrases (QP's) in Chinese and English sentences with two such elements and the second the interaction between QP's and wh- phrases. The third chapter gives a unified account of the phenomena analyzed in the first two. Chapters 4 and 5 investigate the interaction between operators occurring within simple and complex noun phrases, respectively, and draw conclusions for the character of LF movement and the binding of variables. Finally, chapter 6 discusses the special properties of the interaction of wh- adjuncts and QP's, while chapter 7 extends the analysis to Japanese, a language superficially quite different from English or Chinese. The book covers a number of different, though related, phenomena; and I won't try to summarize it. Instead, I will discuss one of these phenomena, scope interactions between QP's, hoping that other readers will find it worthy of further comment. I also hope that others will bring up additional aspects of A&L's book in the discussion.

A&L approach the analysis of quantifier scope interactions through the comparative study of similarities and differences in scope ambiguity in English and Chinese. Apparently, Chinese sentences with two quantifiers, like (1) below, are unambiguous in the scope of the universal and existential quantifiers; and in this regard they differ from their direct English translations, which are always ambiguous. [All of the examples I give below are from the book unless otherwise noted.]

(1) Meige nanren dou xihuan yige nuren every man all like one woman "Every man loves a woman."

This difference between English and Chinese, if it is real and stable across speakers, poses an important theoretical problem. Since the principles governing the construction of LF must be universal and the semantics of quantification must also be, it is not clear how the difference between the languages can be accommodated. One of A&L's central goals is to account for it in a way consistent with a strong and restrictive theory of Universal Grammar. They propose that quantifier scope is determined by two basic principles:

(I) The Minimal Binding Requirement (MBR): Variables must be bound by the most local potential antecedent (A-bar binder).

(II) The Scope Principle: A quantifier A may have scope over a quantifier B iff A c-commands a member of the chain containing B.

I give here the version of these principles stated in chapter 1. Later in the book, A&L revise the Scope Principle and also their assumptions regarding the nature of LF movement in order to give a unified account of QP/QP scope interactions and QP/wh- scope interactions; but to avoid complicating and lengthening my introductory remarks, I am assuming the initial formulation. I think that what I have to say translates easily to the revised one.

Given A&L's principles, the ambiguity of the English version of (1) depends on adopting the VP-internal subject hypothesis. The derived structure needed to support the two scope orders arises as follows: The subject starts out adjoined to VP and moves to [Spec, IP] position, creating a chain. Then at LF the QP's move to A-bar positions to create operator variable structures. The MBR requires that the object QP be adjoined no higher than VP, so as not to interfere with the binding of the subject position by its QP, which will adjoin to IP. The LF representation of the sentence, given in (2) below, is ambiguous under the Scope Principle because at the same time that the subject QP c-commands the object QP, the latter c-commands the trace of the former.

(2) [IP everyone-i [IP x-i [I' I [VP-1 a woman-j [VP-1 t-i [VP-2 loves x-j ]]]]]]

Why then is the Chinese sentence in (1) unambiguous? The answer, according to A&L, is not that the MBR or Scope Principle vary in their formulation across languages but rather that the phrase structure of the Chinese sentence differs from that of its English translation in a simple but crucial way. Chinese, they say, has a defective system of verbal inflection, entirely lacking agreement (and perhaps tense as well - AK). This defective inflection will not support V-to-I raising, and the absence of V-to-I raising prevents subjects from raising out of VP. As a result, no syntactic chain is formed by subject raising and the only LF structure consistent with the MBR has the subject QP adjoined to IP (or the higher VP) and the object QP adjoined to the lower VP. This structure is unambiguous in its interpretation under the Scope Principle.

An interesting consequence of A&L's analysis is that can explain an otherwise surprising fact. Even though simple active sentences are unambiguous in scope interpretation in Chinese, passives like (3), whose LF is as in (4), are ambiguous, just as they are in English:

(3) Meigeren dou bei yige nuren zhuazoule. everyone all by one woman arrested

(4) [ meigeren-i [ x-i yige nuren-j [ dou bei x-j [ zhuazoule t-i ]]]]

This ambiguity is, however, expected because the trace of the passive subject creates the syntactic chain needed to induce the subject/object scope ambiguity.

Not surprisingly, many facts appear recalcitrant to A&L's analysis and deserve further discussion. For example, sentences with the Chinese equivalent of 'seem', which should induce ambiguity through subject raising in the same way that passives do, do not. A&L claim that reanalysis occurs between 'seem' and its complement predicate, removing or deactivating the relevant trace. We might ask how plausible this is. Problems also arise in English. Thus, A&L, following others, note that while a sentence like (5) is ambiguous in the scope of the direct and indirect object QP's, its near equivalent (6) is not:

(5) John assigned every problem to one of the students. (6) John assigned one student every problem.

In order to account for this difference, A&L adopt a version of Larson's analysis of the double object and dative constructions. This analysis is quite complex and would take too much space to discuss here; but if adopted, it correctly predicts the ambiguities in (5)/(6) and related sentences. The technical devices needed are, however, open to question. For instance, A&L assume that QP's can adjoin to non-maximal projections and also that the reanalysis of verb and indirect object required in the double object construction is undone at LF. Hence, such reanalysis is not equivalent to NP incorporation into verbs, which it superficially resembles. Furthermore, while LF movement out of the reanalyzed structure is allowed, they must require that the structure itself not be a possible adjunction site for the moved QP. This requirement strikes me as odd. Either the reanalyzed structure is really lexical, in which case extraction should be blocked, as in other cases of NP incorporation, or it is phrasal, in which case adjunction to it ought to be possible. Finally, we can note a fact that A&L do not mention, which may raise serious problems for an account of scope ambiguity in double object sentences based on Larson's analysis. Consider the sentences in (7) and (8):

(7) Mary assigned one problem to every student. (8) Mary assigned every student one problem.

Here my judgment is that both (7) and (8) are ambiguous. In particular, (8) can mean that there was a problem which Mary assigned to every student. That reading becomes more prominent if we give the indefinite added descriptive content, as in (9):

(9) Mary assigned every student one problem that she particularly liked.

Under A&L's analysis the wide scope reading of the indefinite should not be available. One way around the problem would be to say, as has been proposed, that indefinites have both quantificational and and referential readings. On their referential interpretation they behave like definites and aren't interpreted as being inside the scope of a quantifier. This solution deserves careful exploration but it raises difficult questions regarding A&L's analysis of these cases, since now the ambiguity in a sentence like (5) may no longer be a quantifier scope ambiguity. However, I myself doubt that this solution is correct because the scope ambiguity in (8)/(9) seems to persist in (10):

(10) John may assign every student one problem.

Here one reading of the sentence is that it may happen that there is a problem that John assigns to every student. Under this reading the indefinite is not referential, since it is interpreted as being within the scope of the modal operator. A definite NP would not be taken to be within the scope of the modal, as (11) makes clear:

(11) John may assign every student the problem.

We are thus left with the question of why (8)/(9) should be ambiguous, given that its phrase structure is the same as that of (5), which is unambiguous. The contrast threatens somewhat A&L's attempt to account for all scope ambiguity differences across sentence types and languages in terms of phrase structure differences.

The remarks I have made here are only the briefest sketch of a discussion of one part of a very interesting book. I hope that they will stimulate others to read the book and share their reactions with the Linguist community.