Explorations in abstract phonology By Edmund Gussmann (review)

REVIEWS431 Metcalf, George J. 1974. The Indo-European hypothesis in the sixteenth and seventeenth centuries. Studies in the history of linguistics, Traditions and paradigms, ed. by Dell Hymes, 233-57. Bloomington: Indiana University Press. Piron, Maurice (ed.) 1961. Etymologie, by Anne Robert Jacques Turgot. Brugge: De Tempel. Sebeok, Thomas A. (ed.) 1966. Portraits of linguists: A biographical source book for the history of Western linguistics, 1746-1963. 2 vols. Bloomington: Indiana University Press. [Received 5 June 1979.] Explorations in abstract phonology. By Edmund Gussmann. Lublin: Uniwersytet Marii Curie-Sklodowskiej, 1978. Pp. 190. Reviewed by Lee A. Becker, University of Alberta, and Christina Y. Bethin, University of Virginia This Habilitationsschrift addresses the question of abstractness in phonology, mainly on the basis of Polish data (Chapters I-IV; Chapter V is on English). The discussion centers on Polish vocalic phonology—which, because of its complexity, 'should be particularly useful in deciding questions of general interest' (p. 5). This review will focus primarily on the theoretical aims of the book. The general outlines of Gussmann's description of Polish do not differ greatly from those of previous generative treatments (e.g. Lightner 1963, Steele 1973, Laskowski 1975); however, some of G's most significant innovations are discussed below. In chapter I, G argues against the True Generalization Condition (TGC) of Hooper 1976. As Jensen 1978 has pointed out, Hooper's formulation of TGC is not precise; however, for purposes of this discussion, the pertinent aspect of TGC is that 'rules express generalizations that are true for all surface forms' (Hooper, 13); this would be equivalent to disallowing Kiparsky's first type of 'opacity' (1971 : 621-2). G presents a non-automatic alternation of stem final [k g] with [k' g'], the latter occurring before some (but not all) desinences beginning with [e] ; by TGC, this would have to be handled with morphological (or morphophonemic) rules.1 G points out that foreign or nonsense stems which end in velars exhibit palatalization before those inflectional endings where palatalized velars occur in native forms; and he concludes (13) that 'a consequence then of adopting the TGC is the claim that speakers are familiar with the ex definitione unpredictable morphophonemic alternation of words they do not know.' Although some would agree that the behavior of foreign and nonsense stems indicates that velar palatalization is rule-governed, this does not in any way indicate what the rule in question is. It is important to bear in mind that no means of representation of non-automatic (non-Prule ) generalizations has yet found general acceptance in Natural Generative Phonology (NGP). In Hooper's approach, no distinction is made between morphophonemic alternations occurring in wholly phonologically definable environments, but affecting a restricted class of morphemes (e.g. the e ~ 0 alternation mentioned below), and those occurring in (at least partially) morphosyntactically defined environments (e.g. the velar alternation). The former can be described by a suppletive or diacritic notation (cf. Harris 1978); the latter can be considered an 'exponent' 1 Hooper makes a primary distinction between derivation and inflection. In derivation, alternate forms of a morpheme are related by means of viA-rules (cf. Vennemann 1972). In inflection, alternate forms of morphemes which are not the result of phonologically transparent P-rules are treated by means of suppletive representation (cf. Hudson 1974, 1975). For the alternation in question, Hooper would probably adopt the latter analysis. 432LANGUAGE, VOLUME 56, NUMBER 2 (1980) (Matthews 1972) of the given morphosyntactic category.2 In Chapter II, G discusses evidence for absolute neutralization. He presents an argument against Kiparsky's (1973) Weak Alternation Condition, which claims that non-automatic neutralizing rules can apply only to derived forms. In Polish, some [e]'s alternate with zero (others do not); in addition, alternating [e]'s (as well as non-alternating ones) can occur after both palatalized and non-palatalized consonants, e.g. nom.sg. dzien 'day', gen.sg. dnia; nom.sg. sen 'dream', gen.sg. sna. G's analysis of the e~0 alternation is similar to that of Lightner 1963, in that the alternating vowels are represented underlyingly as [ + high]; they are distinguished from non-alternating high vowels by the feature [—tense]. The palatalization of the preceding...