Adjectives and comparison in English: A semantic study By Jan Rusiecki (review)

716 LANGUAGE, VOLUME 62, NUMBER 3 (1986) separating morphological from lexical operations facilitate such a line of research, enabling us to eliminate the concept of null morphology altogether. Though K does not exhaust her topic, she provides intriguing and highly suggestive insights in this important monograph. [Robert Beard, Bucknell University.] Adjectives and comparison in English: A semantic study. By Jan Rusiecki. (Longman linguistics library, 31.) London & New York: Longman, 1985. Pp. xiv, 206. The Journal ofsemantics 3:1/2 (1984) was a special issue supposedly on the comparative construction; in fact it was devoted almost exclusively to X > Y comparatives (X = primum, Y = secundum comparationis), and there was surprisingly little insight into the meaning of even that comparative. R offers a great deal more in almost exactly the same space: 'a unified approach to the semantics of gradable adjectives in English, in all their forms: the positive degree, the comparative degree, and the superlative degree; and in all their uses: both as predicates ofsentences and as attributes in noun phrases ... A theoretical interpretation of the semantics of gradable adjectives, done in settheoretical terms, is checked against ... the Survey of English Usage corpus, and the results of elicitation tests ...' (xiii). Adjectives are given a crosscut classification as either 'unary scale' (e.g. red) or 'binary scale' (e.g. tall-short), and concomitantly absolute (e.g. wet) or relative (e.g. red, tall) Where necessary, distinction is made between a broad interpretation (Ed is six foot tall 'Ed is 5?0" - 6'2"') and a narrow one (Ed is five foot eleven [ = five feet eleven inches tall] D-» 'Ed is 5' 11"'). Because of the different implications oftheir comparatives, the NP arguments of the adjectives are distinguished depending on whether they refer to individuals or generics; cf. Ed is taller than Jo vs. A giraffe is taller than a gerenuk. For many comparatives, a statement is made about the required or preferred range of values of the measure function for one or both of the argument NP's, expressed in terms of membership of the fuzzy sets many and few: thus John is tall translates as 'Mhj (John) e many feet and inches'; John is short as 'Mh5 (John) e few feet and inches.' Both interpretations have the following conditions: Pragmatic assumption: the speaker assumes the hearer will identify correctly the set S of which the referent of the argument is a member. Empirical normality condition: ? units e |Hs|. (pp. 36, 73) Mhs is the measure function for the height of the set S, and |Hs] is the fuzzy set of numbers expressing the possible values for set S on the scale for the dimension H. R claims that his 'measure function' is not subject to the objections raised against 'average' or 'median', found in other semantic descriptions of these adjectives ; but I can see no substantive difference. R reports experiments where, given sentences like X is tall or X isn't short, subjects judged how tall X was. The results are interesting for the definition of fuzzy sets such as |tall men|, (old women| etc. R also carried out many experiments in which he got people tojudge, e.g., the respective heights of X and Y, given sentences like X is taller than Y or X is shorter than Y. I cannot see how such tests have any value. It is relative dimensions (height or whatever) which are compared in X Y, X < Y etc. and not the actual dimensions of X and Y. The motivation for such pointless tests is a mistaken belief(also held by many others) that comparisons should be reduced to a standard scale ofdiscrete measurements. But this is rarely necessary . In X is not so tall as Y, we infer that 'X < Y in height'; but there is no mention of a standard metric (meters, feet etc.), and I see no reason to invoke one just because it is readily available. In Ed waxed the car more carefully than Jo washed it, we readily understand the comparative degrees of care mentioned without having recourse to a standard scale of measurement for care. Why invoke such a standard for some comparatives if not for all? Notwithstanding my gripes about certain aspects of R's analysis, anyone seriously interested in the semantics of gradable adjectives and their comparative degrees will find this book very useful. [Keith Allan, Monash University.] Epistemische Bedeutung. By Monika Doherty. (Studia grammatica, 23.) Berlin: Akademie-Verlag, 1985. Pp. 160. M 15.00. ...