In lieu of an abstract, here is a brief excerpt of the content:

Reviewed by:
  • Syntactic theory: A formal introduction by Ivan A. Sag, Thomas Wasow
  • Patrick Farrell
Syntactic theory: A formal introduction. By Ivan A. Sag and Thomas Wasow. Stanford: CSLI Publications, 1999. Pp. xiii, 481.

Finding an appropriate textbook for an introductory course on syntactic theory has become difficult. Since most attempt to portray a more or less current version of generative transformational theory (TG), they tend to become obsolete fairly rapidly and the trace-per-morpheme average in phrase structure representations rises dramatically as a function of up-to-dateness. Given that the trace-heavy popular analyses that recent TG textbooks rely on often seem to be [End Page 158] motivated as much by theory-internal assumptions as by linguistic facts,1 it can be tricky to provide novice students of syntax with a satisfactory account of the rationale underlying the analyses and to get students to have much faith in their trustworthiness or usefulness. Against this background, Ivan Sag and Thomas Wasow’s Syntactic theory is a welcome new choice. Focusing on the grammar of English, it introduces syntactic analysis from the general perspective of constraint-based lexicalist theories, drawing heavily on head-driven phrase structure grammar (HPSG) (Pollard & Sag 1994). At the same time it manages to provide a thoroughly coherent and relatively comprehensive account of many of the phenomena typically covered in an introductory textbook (agreement, anaphora, auxiliaries, wh-movement, passive, control, raising, etc.) without allowing anything but complete overt words and genitive ’s to appear under the nonphrasal nodes in phrase-structure trees and making only minimal and reasonable use of what I take to be theoretical analogs of traces and other empty categories.2

One of the things that makes a surface-oriented phrase-structure grammar approach possible, of course, is the use of elaborate feature structures both within lexical entries for words and within associated phrase structures. Two brief introductory chapters treat such matters as the rule-governed nature of language, the distinction between descriptive and prescriptive grammar, the generative methodology of syntactic inquiry, the basic notions of phrase structure and phrasestructure rules, and simple constituency diagnostics; then the reader is quickly plunged into the business of (1) figuring out the feature structures for lexical items, based on valence considerations, case and agreement requirements, etc., and (2) modeling syntactic categories such as VP and PP as feature structures that appropriately encode the dependency relations between heads and their arguments and allow for important cross-categorial generalizations to be expressed. Before returning to constraints on pronouns and reflexives, a topic touched on in the opening general discussion of linguistic rule systems, a generalized phrase-structure schema similar to that of standard X-bar theory is presented and justified; a set of mechanisms and principles for feature-passing and unification is developed with enough precision and detail to allow analyses of agreement and case marking in English, Icelandic, and Wambaya to be fleshed out; and lexical semantic feature structures and principles of semantic composition are explained.

Ch. 7, ‘Binding theory’ (147–70), presents an appropriately simplified version of the o-command- based binding principles of HPSG (Pollard & Sag 1992, 1994). The key ideas are:

  • • Among the features that can be associated with a word is an argument-structure (ARG-ST) feature whose value is a list consisting of the specifier and complements of that word.

  • • [+ ANA] elements, including reflexives, must be outranked by a coindexed member of the same ARG-ST list.

  • • Pronominal elements must not be outranked by a coindexed member of the same ARG-ST list. [End Page 159]

In essence, since a subject would outrank a direct object co-argument of the same verb, the latter, if it were a reflexive, would have to be coindexed with the former, but if it were a pronoun, it could not be so coindexed. A reflexive could not ordinarily have its antecedent in a higher clause, while a pronoun could, since constituents of an embedded clause (raising cases aside) would not be on the ARG-ST list of the predicate of the higher clause. Two specific issues concerning the distribution of reflexives and pronouns are dealt with in some depth: the interaction of the imperative rule...

pdf

Share