La liaison avec et sans enchaînement: Phonologie tridimensionelle et usages du français (review)

REVIEWS La liaison avec et sans enchaînement: Phonologie tridimensionelle et usages du français. By Pierre Encrevé. Paris: Éditions du Seuil, 1988. Pp. 321. F 200. Reviewed by Annie Rialland, C.N.R.S., Paris Encrevé's book innovates on two fronts: it brings to light new facts about French phonology, and it proposes a new theoretical treatment of French liaison. Until recently, it was widely believed that the liaison consonant occurring in expressions like/avfl/i[z] un rêve is always phonetically syllabified with the following vowel. E shows that this is not the case, at least in political speeches, which he has chosen to study because they favor occurrences of stylisticallygoverned optional liaisons. E studied 10,816 instances of potential liaisons in the recorded speeches of 21 politicians, representing two thirds of those who regularly expressed themselves in public in the period 1978-1981. Of these potential liaisons, 7,842 were actually realized, and, of these, 5,029 were obligatory liaisons and 2,815 were optional. Within this last group (and only here) E counts 316 liaisons non enchaînées, that is, liaisons in which a brief period of silence, a glottal stop, or even a schwa is inserted between the consonant and the vowel. Although their overall percentage is not large (11.2%), their distribution is of interest, varying from one speaker to another with some consistency , from 0% for Rocard to 18.7% for Debré (the record-holder). The variation is even greater from one speech to another, ranging between 0% (evidently Rocard) to 35.7% (Chirac). Liaison non enchaînée occurred only rarely if at all in the speeches of earlier politicians (Pétain, Blum, De Gaulle, Pompidou), and as a phenomenon that seems to be on the rise it deserves particular attention. After surveying the data, E offers a rapid historical overview (written with a good deal of humor) of the problem of liaison and its treatment within various theoretical frameworks. He follows the birth of theories of liaison, their subsequent development, and the distortions to which they were occasionally subjected . The contribution of sociological factors internal to linguistics (e.g. the weight of certain institutions and certain linguists) is brought into focus and observed with some amusement. It should be emphasized that this part of the book (comprising most of chs. 1 and 3) goes well beyond a simple historical review of the treatment of liaison; in fact E takes up a part of the modern history of phonology here. E then proposes his own theory of the two types of liaison, liaison enchaînée and liaison non enchaînée. I will explore the logic of his treatment by examining the analysis of the example that he develops at considerable length: j'avais un rêve, realized with enchaînement &sj'avai[z] un rêve (in which [?] syllabifies with the following vowel) and without enchaînement as j'avai[z] I un rêve (in which [z] syllabifies with the preceding vowel). E develops his treatment of liaison within the context of a general theoretical framework belonging to the autosegmental tradition, and inspired more pre134 REVIEWS 135 cisely by the 'no-rules' model. His basic assumptions are that there are no rules (in the sense of standard generative grammar), and that all regularities observed in a language can be accounted for in terms of highly specific lexical representations and certain general principles which are applicable to them. More specifically, he assumes that underlying (lexical) representations provide not only phonemes, but also linked and unlinked skeletal points and empty and filled syllable constituent positions (A = attack, R = rime, N = nucleus, C = coda). In French, these points and positions contain all the information necessary for determining fully syllabified surface representations. They serve as a sort of scaffolding which is elaborated on in the course of a derivation through a variety of association processes that build up the surface form. Let us now follow the theoretical treatment of the example. The proposed lexical representation ofj'avais un rêve is the following: (1) ARAR ARAR N NC N NC 3 a ? e ? e r e ? (Note that this representation does...