Linguists have habitually phrased their accounts of language knowledge as sound/meaning correspondences about which no mention need be made of differences that might exist between knowledge of how to analyze input versus knowledge of how to construct output . However, evidence from many sources increasingly indicates that the dissimilarities between language as comprehension versus language as production are so profound that they nullify attempts to describe language in a 'non-directional' manner, 'neutral' with respect to interpretive versus expressive functions. A two-component (...) model of idealized linguistic performance can be devised on the basis of much past and present linguistic scholarship. Elements needed for the comprehension component are derivable from 'taxonomic' distributional analyses of phonotactics, morphosyntax, and lexical semantics, as well as from recent cognitive psychological work on 'perceptual strategies', while the production component can be formulated in terms of 'generative-semantic' and 'generative-phonological' rules recast as meaning-to-sound transformational processes. Interaction between these two components may well be the crucial determinant of the 'creative aspect of language use'. (shrink)