Inst, för lingvistik, Stockholms universltet, S-10691 Stockholm, Sweden oesten@ling.su.se Haspe Ima th's account of the role of teleological explanations in linguistics is clear and illuminating, as is also his analysis of the not too explicit relationship between OT constraints and functional motivations. But the main point of OT lies, as far as I can understand, not in the constraints themselves but rather in the claim that variation between and within languages can be reduced to differences

more »

... n the ranking of constraints. It ought to follow from this that language change has essentially to be modification of the weightings of one or more constraints. Haspelmath's paper does not really address the issue of the viability of such a view on language change. Instead, he brings in adaptation as an explanatory concept from evolutionary biology. The application of adaptation to the theory of linguistic change is not unproblematic. Typically, biological populations change their genetic make-up as a response to changes in the environment. Such situations are also found in language change, as when new words are introduced to refer to cultural innovations, but attempts to explain grammatical and phonological change in a similar way have not been convincing. Adaptation does constrain linguistic change in the sense that new structures have to meet minimal demands on functionality to be accepted. But this will not, in the normal case, lead to a transition from a sub-optimal to an optimal stage in the development of the language. A case in point is Haspelmath's paradigm example, the assumed transition from [kxtz] to [kaets]. As he himself acknowledges in fn. 3, this example is "not ideal" since [kxtz] is "very difficult to pronounce" and "was presumably eliminated very soon". Indeed, it can be doubted whether there has ever been a stage of English where [kaetz] was a normal pronunciation. Rather than language change proper we seem here to be dealing with "instant adaptation". Another key concept in Haspelmath's account is variation. Its role in the story of language change, however, is also problematic. Darwin's notion of natural selection presupposes pre-existing variation in the population. But since natural selection inevitably eliminates some of the variation, there must be something that constantly "feeds" the pool of variation if the whole process is not going to