Theories of speech production suggest that phonetic encoding involves an access to stored syllable-sized articulatory plans. Both neurolinguistic and psycholinguistic investigations have reported an effect of the frequency of use of syllabic units, respectively on accuracy and errors in brain-damaged (BD) speakers and on production latencies in nonbrain-damaged speakers. Beyond these convergent results, the fact that the same effects have been reported with BD patients with and without impairment ascribed at the level of phonetic encoding challenges the architecture of speech production models and the interpretation of patients' behaviour. Here we carry out a fully parallel neurolinguistic and psycholinguistic investigation to address whether previous diverging results can be accounted for by methodological differences between neurolinguistic and psycholinguistic studies. We analysed production accuracy in 14 BD speakers and production latencies in 24 non brain-damaged speakers using the same pseudo-word stimuli, same reading and repetition tasks, and same multiple regression approach. Results replicate evidence from previous neurolinguistic and psycholinguistic studies on an influence of syllable frequency independently of other sublexical variables in both populations. In addition, the effect of syllable frequency on production accuracy was not limited to BD patients with impaired phonetic encoding. We suggest that these results are best accounted for by postulating interaction between phonological and phonetic encoding.