Abstract

John Perry is known for his work on personal identity, and Jon Barwise, Director of the Center for the Study of Language and Information at Stanford, is a logician who works on model theory. Their cryptically titled collaboration is a first book-length foray into the semantics of natural language. Situations and Attitudes is programmatic even where detailed, and a sequel is promised which will carry through the program extensively and with full-dress technical detail. The program, for which the basic theoretical machinery, the central categories, and the broad rationale are provided in this book, is an "innocent" model-theoretic semantics of natural language, in the spirit of realism and with debts to the likes of Putnam, Kripke, Donnellan, Montague, and psychologist J. J. Gibson. The authors describe their realism as "ecological" because they view meaning as a ubiquitous phenomenon of sentient organisms' natural and necessary attunement to constraints on the flow of information throughout their surrounds. They see an adequate semantic theory accounting for how language fits into the general flow of information. Their view is a realism not only in stressing the external significance of language, the primacy of its function of conveying information about the external world, but in seeing meaning itself as an objective uniformity in that world, a systematic relation of a special sort between different types of situations. The "innocence" of this semantics emerges in its leading concern--a correct treatment of the famously troublesome verbs of attitude which are used to embed sentences and so to report perception and cognition. Barwise and Perry refuse to sacrifice the view that sentences embedded in attitude reports "work just as they do elsewhere," i.e., are about situations, parts of the real world. The broader innocence is that of non-collaboration in the "Fregean tradition" of philosophical semantics which is "seriously flawed" by its inappropriate emphases, due to preoccupation with mathematics and symbolic logic, on univocal expressions, eternal sentences, entailments between sentences, arbitrariness and uniqueness of the name-named mapping, and a general tendency to think that all information available from an utterance must be part of its interpretation. The book counters that tradition on these points and others.