Major findings are reported of a longitudinal, naturalistic study of grammatical morpheme development in an aphasic child from 5;5 to 6;1. The majority of the morphemes were not acquired in the same order nor at the same mean length of utterance (MLU) levels reported for normal children. As an alternative to the normal acquisition model, based on the theory of cumulative grammatical complexity, it is proposed that the forms developed in order of their semantic and phonological salience, as determined by syntactic, semantic, and phonological context. It is suggested, with respect to current remedial training procedures, that neither the MLU index nor level of grammatical morpheme production serves as a reliable measure of the aphasic child's grammatical knowledge. Given the operation of temporal sequencing constraints, abnormally low levels can result from a strategic tendency to reduce the length and phonological complexity of the sound string or phrase. It is argued that the more redundant and less sonorant items are particularly susceptible to these processing constraints and that their omission does not in itself constitute evidence of a syntactic or cognitive deficit. (Author)