No abstract

No abstract

No abstract

No abstract

The Quantification of Functional Load 1 0 Introduction. Of the many pt:oblems in linguistics on which the work of A. Martinet has shed light, one of the most interesting is the notion of .functionalload (or yield or burden). 2 In simplest terms, the notion is this. The function of a phonemic system is to keep the utterances of a language apart. Sorne contrasts between the phonemes in a system apparently do more of this job than others. For instance, in English there are hundreds of pairs of words that differ only in that one has /P/ where the other has fb/ (pat:bat, nipp/e:nibble, cap:cab), but only a very few are kept apart by fs/ versus j'if (for sorne speakers mesher: measure; for sorne Asher: azure; for sorne Aleutian:allusion). Presumably, then, the contrast between /p/ and /b/ does more work even in complete utterances than does that between fs/ and fi/. At [east, it is easier to coin a pair of wh ole utterances such as Don't take that cap: Don't take that cab than it is to find one for fs/ and /z/, sim ply because there are more minimally different words of the first type. Martinet's concern with functional Joad has been with its possible relevance in linguistic change. Suppose, for example, that in a particular community the random drift of sound change 3 threatens to wipe out a contrast that carries a certain functional Joad. If that Joad is sufficiently high, is it possible that exigencies of communication would prevent the impending coalescence? How high must the Joad be for this effect? Or,