Interlanguage and learnability: From Chinese to English By Virginia Yip (review)

BOOK NOTICES 227 and lexical semantics, Münster, Germany , September 13-15, 1994). Tübingen : Max Niemeyer Verlag, 1996. Vol. 1 : Plenary lectures and session papers , pp. xiv, 378; Vol. 2: Session papers , pp. xi, 411. These two volumes constitute the edited versions of 61 plenary and session papers (written in English, French, or German) out of about 90 papers. Vol. 1 contains the nine plenary papers and session papers in two categories: theory and methodology (thirteen papers) and the lexicon in computer linguistics and artificial intelligence (four papers). Vol. 2 contains the remaining seven categories of session papers: description of specific areas of vocabulary (thirteen papers ), metaphor (three papers), lexical aspects of word formation (five papers), contrastive word formation (four papers), contrastive lexical semantics (four papers), historical semantics (three papers), and lexical units in psycholinguistics (three papers). The papers are an interesting eclectic assortment, befitting a conference focused on words and their meanings in the context of language use. Because it is impossible to do justice to all the papers in this brief space, I will comment on a few and hope thereby to impart a flavor of the content of the two volumes. One disconcerting point should be made: the book references given at the end of each paper do not contain the publishers' names. Renate Bartsch's "The myth of literal meaning' (3-16) argues that semantics generally assumes there is something literally meant by a given word and that this literal meaning is shared among the word's multiple uses. This is, however, an unfounded assumption , it is argued, though one which has seemingly strong motivation. Taking Searle 1992 (The rediscovery of mind, Cambridge, MA: MIT Press) as a jumping-off point to discuss the principle of compositionality, the paper hearkens back to Frege's contexuality principle and Bedeutungen, or denotations , as the component of meaning to which the notion was originally directly applied, and only indirectly to Sinn, or meaning, as function composition of functions from world-time indices to denotations. Frege's principle does not require that expressions have literal meanings; instead, an expression itself can be viewed as a function which assigns a different meaning for each type of context, i.e. the different kinds of uses of the expression. By delineating 'context types and types of satisfaction situations' provided by perspectives, 'context-dependent meanings of parts of expressions are determined'. These are dependent on a theory ofconcept formation. The theory formally sketched here involves two stages: the first, based on experienced use of utterances of an expression with regard to satisfaction situations (a sequence of stabilized similarity sets), and the second , based on knowledge 'in terms of sets of general sentences held true'. 'Models of Lexical Decomposition' (169-83) by Dieter Wunderlich proposes a two-level theory of lexical semantics, lexical decomposition grammar, along the lines drawn by Manfred Bierwisch and Ewald Lang (Grammatische und konzeptuelle Aspekte von Dimenonsadjetkiven, Vols. XXVI + XXVII of Studia grammatica, Berlin: AkademieVerlag , 1987) counterposed against two other dominant theories, that of Ray S. Jackendoffs [lexical-] conceptual semantics (Semantic structures, Cambridge , MA: MIT Press, 1993) and James Pustejovsky 's lexical semantics ('The generative lexicon', Computational Linguistics 17.409-41, 1991). The two levels, conceptual structure and semantic form, which constitutes the interface between syntax/ morphology and conceptual structure, are argued to be necessary for a parsimonious and sufficient account of lexical semantics. Mechthild Rickheit's ? model of word concepts ' (275-84) attempts to position lexical semantics in a model which can be useful for philosophical, linguistic, psychological, and computational theories , drawing on distinctions made by the LILOG project (Text understanding in LILOG: Integrating computational linguistics and artificial intelligence, ed. by Otthein Herzog and Claus-Rainer Rollinger, Berlin: IBM, 1991). The model, based on a variant of discourse representation theory (Hans Kamp, ? theory of truth and semantic representation'. Formal methods in the study oflanguage: Proceedings ofthe thirdAmsterdam colloquium, ed. by Jeroen Groenendijk , T. M. V. Janssen, and Martin Stockhof, 277-322, Amsterdam, 1981) links semantics with world knowledge in the notion of 'word concept', a structure associating functor global and sortal categories with syntactic and semantic arguments. Those doing research in lexical semantics will find these two volumes of...