Constraints

Abstract

Abstract The preceding two chapters provide linguistic evidence for the position that a learning mechanism for language must be able to accommodate a substantial range of variation in syntactic generality. Many linguistic phenomena are very general, perhaps even falling under the designation ‘Affect a’, and these have formed the basis of ‘core grammar’. Many phenomena are not general, as I have argued, but are no less part of the language and constitute knowledge of language as much as the more general phenomena do. It is of course natural to try to account for the apparent exceptionality of apparently idiosyncratic phenomena as following from the interaction of general principles, thereby requiring the minimal special stipulations in the grammar of a particular language. However, I have argued that it is impossible to eliminate entirely the need for special stipulations. Moreover, I have suggested that the phenomena that require special stipulations do not fall into a special class for which a special learning mechanism is required, but are merely instances of the kinds of sound/meaning correspondences that are found in the lexicon and in syntax for all degrees of generality.