Statistik für Linguisten By Gabriel Altmann (review)

BOOK NOTICES 493 current research after an introductory course. This may have something to do with the fact that symbolic logic is an older subject than generative grammar, and has had more time to settle down. At any rate, students who have more than a casual interest in logic (or even semantics ) should be told that they will need to pursue the subject further than it is taken in this book. Exercises are provided here; but those who use it as a text will probably want to supplement them with exercises and problems of their own. Some quibbles: The gulf between sentences and propositions is greater than the authors make it out to be at the top of p. 22; different sentences can express the same proposition. In a couple of places, the exposition lapses; e.g., in the sentence spanning pp. 54-5, the authors seem to say that aformulaleads to contradiction and so is a tautology (they mean the antecedent of the formula). The book is vague (and occasionally even misleading) about the notion of a decision procedure: thus on p. 78, line 7, the authors seem to argue that, because infinitely many instances of a problem exist, the problem is undecidable; but explaining why predicate logic is undecidable is a much more delicate matter than this. Again, on p. 98, 'formal method' is used for 'mechanical method'. The definition of 'satisfaction' in §5.7 is marred by the fact that sequences are taken to be bounded (rather than eventually constant). Toward the bottom of p. 103, excluded middle is conflated with bivalence. Treatments of truthvalue gaps exist (cf. van Fraassen) that validate pv~p, but eschew bivalence. The account of D-* on p. 120 is oversimplified as regards David Lewis' theory ofthe conditional; it presupposes 'the limit assumption', which Lewis rejects. The authors place too much stress on possible worlds and problems of intensionality. These are certainly important, but it would be a shame to identify model-theoretic semantics with possible -worlds semantics in such a way that the infirmities of the latter (as a theory of meaning) are transferred to the former. The book also tends to conflate possible worlds with models. Not only would it be wrong to think of every model as corresponding to a possibility (as the authors are aware, but do not stress), but cases also exist in which the same model is associated with more than one possible world. The reader should beware of misprints. The worst I found were in item (I') on p. 51, where a disjunction sign is missing; in (14d) on p. 65, where the double underlining should be uninterrupted ; and in (11) on p. 141, where the topmost 'intention' should read 'extension'. In general, however, the book is sensible and well-informed. It would be a good text to recommend to graduate or undergraduate students in linguistics who wish to learn something about logic. But it is only an introduction; a student would have to be very gifted to deal successfully with the advanced literature in logical linguistics (e.g. Montague's papers) without further preparation . [Richmond W. Thomason, University ofPittsburgh.] Statistik für Linguisten. By Gabriel Altmann. (Quantitative linguistics, 8). Bochum: Brockmeyer, 1980. Pp. iii, 239. DM 29.80. This is an introductory text on statistical methods, intended for use by linguists. The first three chapters (combinatorial analysis, probability , and descriptive statistics), which occupy approximately half the book, deal with topics basic to any attempt to work with quantitative data. The exposition in this part of the text is technically straightforward and requires no previous knowledge ofcalculus; however, students lacking a solid background in mathematics will probably find themselves gasping for air under a stifling barrage of formulas. Unfortunately, A's dry style does not help: he provides only minimal intuitive introductions to the concepts formally defined, and he fails to show how they may be relevant for linguistic analysis. Thus, in discussing variance and standard deviation, he makes no attempt to justify the use of squared deviations from the mean in the definition—a point that always strikes beginning students as totally arbitrary until it is pointed out that, as a result of this...