Contact languages: A wider perspective ed. by Sarah G. Thomason (review)

REVIEWS631 and if in the syntax it can function as a complementizer. Jerrold M. Sadock's contribution is the most engaging and enjoyable. After pointing out that many elements of the practice of ALS can be considered a matter of taste, he suggests some improvements and addresses the recurring question of the number of modules, arguing cogently for a discourse-functional module (dealing especially with the notions of topic and comment), which would be separate from the semantic module. Finally, two papers, while insightful, original, and interesting, must be considered somewhat peripheral to the main thrust of ALS in that they do not treat the problems at hand in terms of interface of modules but rather deal with elaborations within a particular module. In 'Autostructural analysis of semantic roles', Jan Terje Faarlund argues that within the semantic module there is a tier of semantic roles that associates to noun phrases in the same way that tones spread to segments in autosegmental phonology (Goldsmith 1976). The range of linguistic structures discussed in this paper is impressive: case marking, prepositions, locative alternations, causatives, passives, infinitives, gerunds, topicalization, among others; and insightful comparisons are made to other theories such as relational grammar. However, I find the idea of semantic roles behaving like autosegments somewhat forced since it suggests that semantic roles have a fixed underlying order, which does not accord well with what we expect from a semantic module. Faarlund is of course quite aware of the fact that semantic roles do not associate in a fixed order, and he has to resort to some stipulations to get them to associate in the desired way, which results in many cases of crossing association lines. Since crossing association lines are not allowed in autosegmental phonological theory, I consider it an unfortunate weakening ofFaarlund's proposal to allow them in the association of semantic roles. Gregory T. Stump's paper deals with very interesting morphological problems in Breton and in Tigre, a Semitic language ofEthiopia, and attempts to solve them in a paradigm-based approach to morphology. As Stamp himselfadmits, his paper deals with the problem within the morphological module rather than with the interaction of modules. As with Faarlund's paper, the similarity to ALS resides in the postulation of association lines; and whereas Faarlund allows too many crossings of association lines, Stump appears to allow no crossings at all, which is probably too restrictive in the framework of ALS as a whole. To conclude, this is a challenging and high quality collection which will be an outstanding resource for anyone interested in linguistic theory. Since ALS is a theory of interactions offormal systems, it should be taken very seriously by formal linguists interested in constraint interaction (McCarthy & Prince 1993, Prince & Smolensky 1993). I hope that this collection will stimulate a healthy and much needed exchange of data and ideas between practitioners of ALS and of optimality theory. REFERENCES Goldsmith, John. 1976. An overview of autosegmental phonology. Linguistic Analysis 2.23-68. McCarthy, John, and Alan Prince. 1993. Prosodie morphology I: Constraint interaction and satisfaction. Technical report 3. New Brunswick, NJ: Rutgers University Center for Cognitive Science. Prince, Alan, and Paul Smolensky. 1993. Optimality theory: Constraint interaction in generative grammar. New Brunswick, NJ & Boulder, CO: Rutgers University and University of Colorado, ms. Sadock, Jerrold. 1985. Autolexical syntax. Natural Language and Linguistic Theory 3.379-439. -----. 1991. Autolexical syntax: A theory of parallel grammatical representations. Chicago: University of Chicago Press. 422 Magnolia Street Denton, TX 76201 Contact languages: A wider perspective. Ed. by Sarah G. Thomason. (Creole language library, 17.) Amsterdam: John Benjamins. 1996. Pp. xii, 506. Cloth $160.00. Reviewed by Anthony P. Grant, University of St Andrews This long-awaited book is an essential contribution to the small but growing literature on what most linguists readily call 'mixed languages', even if they cannot always provide a watertight 632LANGUAGE, VOLUME 74, NUMBER 3 (1998) definition of them. 'Mixed language' is a term which over the past centuries has been used to refer not only to pidgins and créoles but to languages which show a high degree of (especially lexical) influence from other languages, (English being the most famous example), and latterly to languages...