Recent research on the early development of knowledge about emotions shows that young children's use and comprehension of emotion language develops from an initial emphasis on expressive/behavioural referents to situational terms towards referents emphasising the sub] ective/experiential nature of emotions. Gender, the type of emotion, the discourse context of the emotion talk and individual differences in strategies to regulate negative emotions are some factors that are shown to moderate the development of emotion language abilities. However, as most of the data comes from early language users there are significant limitations to our knowledge of emotion language development and its implications for emotion regulation. This thesis examines emotion language in early to middle adolescence. It develops a theoretically derived classification model to study the representational and causal structure of emotions evident in the emotion language of 13 to 17 yearolds. Study 1 uses a group format to sample descriptive accounts of emotions and their causes from a normative sample of 303 adolescents in response to emotionally relevant vignettes prototypical of anger and fear. Study 2 compares the lepresentational structure and quality of emotion language between 21 adolescents diagnosed with extemalising disorders (Conduct disorder, Oppositional Defiant disorder), 18 with internalising disorders (Depressive disorders, Anxiety disorders) and 16 without a disorder. It broadens the types of emotion eliciting material by including autobiographical events and an actual emotional challenge as well as the vignette stimuli. In addition, the second study uses an individual participant-interviewer procedure. Results of Study 1 indicated increase in the range and complexity of emotion referents and causal accounts of emotions from early to middle adolescents. Despite an increase in internalist/subjective causal accounts of emotions with age, there was a move towards a more externalised or situational focus in the representation of emotions for the older adolescents in response to the anger material. The findings showed that the ability to distinguish between sadness and anger and appropriately use anger and sad referents develops relatively late with some younger adolescents continuing to have difficulties with this distinction. There were a number of specific gender related differences in emotion language consistent with gender differences in display rules for emotions. In particular, boys showed a preference for expressive/behavioural emotion referents while girls preferred referents with a cognitive focus and use more inner-focused referents. Study 1 also provided initial data about differences between adolescents with extemalising problems, those with intemalising problems and non-problem adolescents. Results indicated more use of non-specific referents by adolescents with extemalising behavioural problems as well as less intensity and involvement in their emotion referents. Adolescents with extemalising problems were more likely to use non-specific referents in responses to anger material than those with intemalising problems. The results of Study 2 showed that adolescents with oppositional and conduct problems show deficits in the fluency, complexity and degree of specification of their emotion language and their causal accounts of emotions compared to non- problem youth and those with depression and anxiety problems. In addition, adolescents with intcmalising problems were less fluent in the production of causal accounts of emotions and used less specific emotion referents to fear events compared to non-problem youth. The results highlight the finding that emotion language is affected differentially for extemalising and internalising adolescents depending on the nature of the emotion-eliciting event. In particular, intemalising youth's language responses to anger events are characterised by inner-directed referents, and reduced intensity and involvement while their conceptualisation of salient fear material is dominated by cognitively focused terms and accounts. Extemalising adolescents language responses to anger events are more outer-directed and intense, and their emotion construals in a fear situation less cognitive and more affect orientated. The data from these studies highlight the need to study emotion language for specific emotion domains, and suggest that the most interesting theoretical questions are in respect of emotion understanding and emotion language abilities for specific behavioural and emotional disorders. The results also support the utility of an approach that combines knowledge about emotion language from the psychological and linguistic literature. It argues for an expansion of our knowledge about the development of the lexicon for emotions and other syntactic and pragmatic linguistic competencies that are important for conceptualising emotions in language. Such an expansion is crucial to investigating associations between early emotional competencies assessed through language and later outcomes in terms of behavioural, emotional and social difficulties.