The author addresses the need for a new acoustic recognition strategy, extending the position that any adequate grammar of a language must distinguish between auditory and articulatory knowledge. Reviewing the existing literature and theories of language and its acquisition, the author discusses their limitations and inadequacies in accounting for natural acoustic patterns. As a solution to the acoustic recognition problem he proposes differentiation of speech signals based on time and intensity. This would provide unique representations for input syllables according to their linguistic differences. (Author/LG)