There has been limited investigation into the nature of specific language impairment (SLI) in adolescence, and even less investigation into speech production in SLI, particularly beyond the preschool years. Phonological output is crucial to oral expressive language, one of the characteristic areas of deficit in SLI. It was hypothesized that adolescents with SLI present with residual speech errors. Accordingly, this study investigated the presence, nature and cause of speech deficits in adolescents with persistent SLI. Adolescents aged 13.0-14.0 years with persistent SLI (n=31) were compared to typically developing (TD) peers (n=31) on an Oral Motor Evaluation (OME) and a Speech Assessment Battery (SAB), consisting of: Multisyllabic Word Repetition, Nonsense Word Repetition, Phrase Repetition, Sentence Repetition and Tongue Twisters. Tasks were analysed in terms of accuracy and the amount of effort required, to determine efficiency and automaticity of phonological output. All participants completed a battery of memory tasks and the language abilities of the SLI group were evaluated. Caregivers completed questionnaires investigating risk factors. Results indicated that adolescents with persistent SLI produced a higher number of inconsistent and context-dependent deviant speech behaviours than TD peers, involving syllable structure changes (phoneme additions/omissions), phoneme substitutions, fluency errors and repetitions/revisions. Participants with SLI also had a higher incidence of neuromotor deficits. Persistent SLI appears to relate to a multifactorial etiology, involving both genetic and early history factors. There was high comorbidity of attention deficits, with implications for resource allocation. Results were interpreted within a functional capacity viewpoint. Speech deficits are thought to contribute to a threshold effect, where even subtle deficits in one area can compromise processing in another area. Speech production in this population appears to be more resource-demanding than in TD peers, which may affect concurrent performance on linguistic and memory tasks. This hypothesis is supported by the high correlation between speech tasks and expressive syntax for SLI participants. The high incidence of speech deficits in adolescents with SLI ii has implications for research and clinical practice with individuals of all ages with SLI.