An introduction to natural generative phonology By Joan B. Hooper (review)

REVIEWS667 Kuno, SusuMU. 1972. Pronominalization, reflexivization and direct discourse. LI 3.16195 . ------. 1974. Super Equi-NP Deletion is a pseudo-transformation. Papers from the 5th Annual Meeting, North Eastern Linguistic Society, 29-44. ??e, Norman H., et al. 1975. Statistical package for the social sciences. 2nd ed. New York: McGraw-Hill. Thomason, Richmond. 1976. Some methodological remarks on semantics. University of Pittsburgh, mimeo. [Received 29 June 1977.] An introduction to natural generative phonology. By Joan B. Hooper. New York: Academic Press, 1976. Pp. xviii, 254. $18.50. Reviewed by John T. Jensen, University of Ottawa This book presents a thoroughgoing challenge to the standard theory of generative phonology, which Hooper refers to as transformational generative phonology (TGP), in the tradition of Chomsky & Halle 1968. As such it merits the careful attention of phonologists of all persuasions. It represents a thorough revision of a 1973 dissertation, written under the direction of Theo Vennemann, who may be considered the originator of natural generative phonology (NGP). Vennemann 1971, 1972, 1974a,b are some of his articles on the subject, reversing his own earlier abstract position (1968, 1970). The work under review is the first full-length treatment of NGP. Hooper has tightened and refined some of the arguments, weakened some of the more radical claims (she has dropped the strong naturalness condition), and improved the organization of her earlier work.1 The book is divided into two parts. Parti, 'Concreteness in morphophonology', comprises the bulk of the work and consists of eight chapters. Part II, 'Natural phonological structure', containing Chapters 9-13, discusses phonotactic constraints in the light of a theory of phonology which includes the syllable. 1. Part I is devoted to the abstractness problem and to a point-by-point comparison of TGP with NGP with respect to questions of rule order, cyclic application of rules, and underlying representations. As Hooper notes, the underlying forms of morphemes in TGP are not entirely abstract, but 'have intrinsic phonetic content', and are related to phonetic forms 'in a non-arbitrary way' (4; cf. Postal's 1968 naturalness condition). However, this requirement still allows rather abstract analyses. Citing data from Spanish, she criticizes in particular the derivation of [léce] 'milk' from /lakte/ in Harris 1969, on the grounds that the rules required are not productive for Spanish speakers. Various attempts have been made to constrain TGP, in particular Kiparsky's 1968 alternation condition; but such conditions are insufficient to rule out analyses like Harris", because of the existence of alternations like [léce] ~ [laktár] 'to lactate'. Hooper concludes that all attempts to build constraints such as the alternation condition into TGP are futile efforts to plug the leaks in a sinkingboat, and decides that a new theory is needed, based on the premise that 'it is theoretically interesting to formulate the strongest possible constraints on the theory and 1 My sincerest thanks go to Douglas C. Walker for many hours of insightful discussion of the issues raised here, and for critical comments on earlier drafts of this review. 668LANGUAGE, VOLUME 54, NUMBER 3 (1978) then to investigate the consequences of these constraints' (5). I will investigate the consequences of her constraints in the course of this review. One of them is 'to allow only a small subset of the grammars allowed by the unconstrained theory' (xi). However, her characterization of TGP as ' unconstrained ' seems incorrect. Implicit in TGP is at least the fundamental constraint that each morpheme has a unique underlying form (except in cases of suppletion). Hooper rejects this constraint , as we shall see, replacing it with quite different constraints. In this respect NGP, which she offers as a replacement for TGP, allows an entirely different set of grammars than TGP, and the two sets may not even intersect. 1.1. Hooper's great bugbear is abstractness, and her attack is frontal. Anything which even remotely smacks of abstractness must go. Accordingly, in pursuit of the 'strongest possible constraints' on generative phonological theory, she brings out two constraints on rules: the true generalization condition (TGC) to relate rules to phonetic forms, and the no ordering condition (NOC) to define the relations between rules. It is a pity that Hooper offers no formal...