Abstract

Recent grammatical theory begins with the notion that the sentences of any natural language are generated by phrase structure rules whose base structural outputs are filled in by lexical insertion rules, the result brought to surface structure through transformational rules. Phrase structure rules rewrite particular items; transformational rules have the greater power of rearranging items. The claim that there are deep structures may be thought equivalent to the claim that the system of rules which will most economically and accurately structurally-describe a language's sentences will also define a base, post-lexical and pre-transformational level. In addition Katz-Postal claimed that the deep structure level as defined above would also provide all the information needed for semantic interpretation. Obviously much will depend in all this on how one defines transformational rule. Bowers argues here that Chomsky's notion of transformational rule is an artifact of his commitment to syntactic deep structure, that it leads to inadequate, needlessly complicated and undecideable grammatical analyses; we cannot segregate phrase structural and lexical processes from purely syntactic transformational rules, and hence transformational rules state "cooccurrence relationships between sentences" rather than relating deep and surface structures. Bowers considers passivization and "causitivization," complement structure, topicalization, and some related matters. The sledding is heavy for those not familiar with these matters. There may have been a delay in publication of this book for the arguments, theories, and papers referred to belong to the late sixties-early seventies or earlier, though Bowers mentions more recent work in a prefatory paragraph: the trouble is that much of what Bowers criticizes has long since been criticized and repaired though not in Bowers's way by those in the Chomskian aegus. When he writes at a general level Bowers seems to conflate the claim that there is a syntactic deep structure with the Katz-Postal claim that this level also provides all that is needed for semantic interpretation: rationalism, nativism, deep structure, intensionality, and autonomous syntax are boated together and sink similarly. But syntax may be autonomous and largely unlearned, growing through exposure in a way sizably independent of the rest of cognition, without otherwise loading down the boat.--J.L.