Ebenen der Textstruktur: Sprachliche und kommunikative Prinzipien Ed. by Wolfgang Motsch (review)

460 LANGUAGE, VOLUME 74, NUMBER 2 (1998) grammatical functions. Clearly, in such cases the reflexive clitic does not function as an a-structure binder as described above. A therefore ascribes a second function to the reflexive clitic in Catalan: It may also optionally identify the 'referential indices' oftwo arguments in a-structure. However, both functions ultimately lead to the same effect, namely the colinking of different arguments to the same semantic participant. Thus one might wonder whether a more uniformanalysis couldbeprovidedtreatingthis kind of colinking as the basic function of the reflexive clitic and trying to derive other effects such as a-structure binding from independent (mapping) principles. A's book is highly recommended to linguists interested in syntactic theory as it covers central syntactic topics such as reflexive clitics and causative constructions . It offers new perspectives on the formal architecture of LFG but will also be of interest to linguists from different theoretical backgrounds. Scholars working on Catalan or on Romance in general will be provided with interesting analyses and new insights into the structure of these languages. [Lutz Gunkel, Freie Universität Berlin.] Explaining culture: A naturalistic approach . By Dan Sperber. Oxford: Blackwell, 1996. Pp. 175. Paper $18.95. The way one tries to explain a cultural fact follows from what one takes the nature of cultural facts to be. The same is true of linguistic facts; for Saussure, for example, linguistic facts were entities in a realm of their own, connected but not reducible to individual psychological facts. The cognitive revolution has given us a naturalistic reduction of linguistic facts. These are now taken to be configurations in the mind/brain ofan individual speaker, and so theirevolution is best explained as the interaction of that speaker's inherent cognitive mechanisms and the data to which he is exposed. In a large body of work over the last fifteen years, Sperberhas been attempting a naturalistic interpretation for the facts of culture. The six essays brought together in Explaining culture represent an overview of this work. Cultural facts, for S, are not Platonic entities but representations in the mind/brain of human individuals. This definition raises a problem which Noam Chomsky's parallel conception of Ilanguage does not face. That is, whereas linguistic representations remain linguistic even if found in only one speaker, cultural representations are only cultural by virtue of being shared by a community. S must therefore expand his definition to state that cultural representations are that subset of all mental representations which come to be widely shared in a particular community. On this definition, there is no discontinuity, oreven clear dividing line, between cultural representations and individual ones. S's approach to cultural facts is epidemiological. To give an epidemiological explanation is to show how and why a particular representation came to be distributed as it did across the minds of a human group. The factors involved will be ecological —some aspect of the group's lifestyle or social structure helped spread the representation in question —and also psychological. As S argues, human cognition systematically transforms representations by processing them. An example here is that of an oral folk tale which, to persist culturaUy, has to be remembered between tellings and also to remain attractive to audiences. These factors wiU cause it to evolve towards a form which is most memorable and relevant to the human mind. Note that this evolution is not a process of random mutation and selection. Rather, cognition exerts definite directional transformations on the representations it deals with. There are obvious parallels here with functionalist work in hnguistics, and, overall, there are plenty of ideas to interest the linguist in this volume. The essays are, however, speculative and poorly backed up with examples. Despite some brave rhetoric about the possibility of a science of culture, the approach is totally programmatic. I have not seen a single weUworked empirical explanation using the epidemiological approach. Nor is it clear how this would be done; we might explain the persistence ofa religious belief by asserting that it was specially relevant to some group of people, but how would we know that this was so? Presumably by observing how persistent it was. Until this circle is broken...