Background: A comprehensive assessment of language ability addresses an individual’s language skills in form, content, and use. The analysis of language performance in discourse production provides information from a functional perspective, that is not examined in standardised, norm-referenced assessment tools. Language sample analysis offers the speech pathologist an ecologically valid tool for identifying strengths and weaknesses in an individual’s text-level language. In making clinical decisions regarding the most relevant and appropriate discourses to evaluate, the speech pathologist needs to consider two factors: (a) the type of elicitation task and (b)the measures used to assess language performance. Discourse production hasbeen studied extensively in relation to the school-aged child but has primarily been viewed through an educational lens. This needs to be extended to describe and evaluate the discourse production of adolescents in genres other than those with an education bias. Specifically, are there discourses that could provide relevant and meaningful information regarding adolescent language, for the speech pathologist in a mental health context? The overall aim of this research is to consider the effectiveness of four elicitation tasks in eliciting spontaneous language samples from typically developing adolescents in order to evaluate language use at word, sentence, and text level for the mental health context. Method: Forty-five, typically developing adolescents (25 younger adolescents: 12-13 years and 20 older adolescents: 16-17 years) participated in an assessment protocol consisting of four discourse elicitation tasks. These included: story generation to a wordless picture book, fable retell, telling personal narratives, and a monologic response to stories that contained a moral dilemma. First, discourse production was examined at (a) word level (lexical diversity, lexical complexity, three semantic domains –affective, social, and cognitive, and verbal facility) and (b) sentence level (verbalproductivity and syntactic complexity) across the four tasks. Second, the problem stories from the personal narratives were studied by analysing text level narrative coherence. Three holistic approaches for evaluating coherence: (a) high-point coherence, (b) story grammar coherence, and (c) Gricean coherence were used to rate the adolescents’ problem stories. Third, adolescents’ monologic responses on the moral dilemma task, which used stories from Kohlberg’s work on the stages of moral development (1976), were examined using a coding schema developed from Bloom’s taxonomy of thinking (1975) to evaluate the language used in critical thinking and reasoning. Results: Descriptive statistics were calculated across eight word and sentence level performance measures, demonstrating sensitivity to context but not age. Results on the moral dilemma task generally followed a normal distribution and could potentially provide benchmark measures for this context. Performance on the personal narratives was much more varied, reflecting highly individualistic samples elicited by this discourse task. Results from rating the problem stories for narrative coherence revealed that adolescents performed at an acceptable level on all three measures and that there was moderate to strong correlation between them. Finally, the coding schema developed to examine the language used for critical thinking clearly demonstrated the progression in skills between the two age groups, highlighting the shift from processes involved in data gathering to drawing conclusions in reasoning. Clinical Implications: Language performance in discourse production is an essential component of language assessment to provide a comprehensive profile of an individual’s language ability. The two elicitation tasks: personal narratives and the moral dilemma task show strong potential as clinical tools that are age-appropriate for adolescent populations and provide relevant and meaningful information for the speech pathology clinician, particularly in the mental health setting. Conclusions: The four elicitation tasks successfully yielded language samples from the adolescent participants, that could be analysed at word, sentence, and text level. The moral dilemma task shows promise for norming purposes.