Essentials of Functional Grammar: A structure-neutral theory of movement, control, and anaphora By George M. Horn (review)

228 LANGUAGE, VOLUME 68, NUMBER 1 (1992) between discourse sequences climaxes in the remarkably imprecise comment that '[a] sequence is as long as it needs to be in order to achieve its function' (176)—this without any reference to the difficulty that discourse analysts have in defining just what such a claim means in concrete terms. This work is best read as an essay in the classic sense: a thought-provoking, mildly polemical exercise where clear-cut conclusions are difficult or impossible to reach but stimulating insights abound. As such, it is very definitely a success. It offers a remarkably wide-ranging and fair survey ofthe terrain, and is bound to initiate profitable discussion among both discourse analysts and translators. [John Sivell, Brock University.] Essentials of Functional Grammar: A structure-neutral theory of movement , control, and anaphora. By George M. Horn. (Trends in linguistics : Studies and monographs, 38.) Berlin & New York: Mouton de Gruyter, 1988. Pp. iv, 404. DM 154.00. This book attempts to develop a functional theory of universal grammar along the same lines outlined in Horn (Lexical-functional grammar , Berlin: Mouton, 1983). In the proposed theory, the syntactic component consists offour rule systems (in addition to categorial and lexical : (i) the S-structure/F-structure Interface System (rules and principles for indexing Sstructure constituents and F-structure arguments ); (ii) the Generalized Co-Indexing System (e.g. reflexive and reciprocal rules); (iii) the Generalized Argument Reduction System (e.g. passivization); and (iv) the E-A System (responsible for variable binding). Grammatical processes are defined in terms of operations on F-structures ('pred/arg structure'). Ch. 1 ('Preliminary remarks', 1-33) introduces the basic theoretical assumptions and reviews GB and LFG. H's theory owes much to both, though it differs from them in important respects. Ch. 2 ('Basic concepts', 35-164) deals with the basic characteristics of the F-structure Component, while Ch. 3 (The Co-Indexing System : A reanalysis of control and bound anaphors ', 165-242) characterizes the control relation in terms of the configurational properties of the pred/arg structure ('prepositional clause'). While H's analysis of the control of the subject of the gerundive complement is interesting , that of the reflexives and reciprocals is less convincing. In Ch. 4 (The Argument Reduction System: A reanalysis of bounded movement and deletion ', 243-328) the notion 'movement' is redefined in terms of the set of FS[tructure]i/ FS[tructure]2 pairs viewed as either a lexical or a nonlexical argument reduction process. Since some of the argument reduction processes (e.g. passivization) must be lexically specified, a more thorough discussion of lexical structure and semantic relations would have been appropriate . In Ch. 5 (The EA System: A reanalysis of Wh-movement', 329-92) H discusses the EAnaphora (EA) rule, which, applying at the derived F-structure level, binds a variable to a relevant operator/antecedent (a WH-element) generated directly in comp. The analyses of whmovement , topicalization, and cleft sentences are then unified in terms of the EA rule (cf. Noam Chomsky, 'On wh-movement' , in Formal syntax, ed. by Peter Culicover et al., 71-132, New York: Academic Press, 1977). Most of H's data come from English and. to lesser extent, from Polish (although nineteen languages are represented in the corpus). H's most interesting proposal is his treatment of coindexing strategy in terms of the distance in the (derived) F-structure rather than structural configuration (such as c-command). He can thus unify the accounts of the choice of the reference of reflexives and the interpretation of the subject of picture nouns and embedded (gerundive ) VP complements. The proposed notion 'possible asymmetric FS pair' accounts for diverse phenomena which may or may not involve 'NP movement' in earlier (and some of current) theories, such as English active/passive, Turkish causative/noncausative, and Polish (and Italian ) transitive/reflexive-intransitive sentence pairs. However, the alleged universality of the overall theory depends on how the languagespecific correlation statements (e.g. languageparticular case-marking or word-order conventions ) can be generalized at a higher level to form a universal set (contra H's conception). There is, nevertheless, no doubt that the book deserves a close reading by syntacticians...