In this paper, we investigate the claim that although the same lexical units are involved in speech production and comprehension, there are separate input and output phoneme layers (Foygel and Dell, 2000). Data from a case series of aphasic patients are used to test this claim by examining the relationship between performance on a test of picture naming and performance on tests of phonological input. Estimates of each patient's semantic-lexical and phonological impairments in speech production were derived from Foygel and Dell's computational model of picture naming. It was found that the strength of the semantic-lexical impairments in speech production was significantly correlated with performance on auditory comprehension tests. This finding is consistent with the claim that the same lexical units are involved in speech comprehension and production. Conversely, the correlations between the strength of the phonological lesions in speech production and performance on tests of phonological input were non-significant, consistent with Foygel and Dell's claim that there are distinct input and output phoneme layers.