This thesis describes a longitudinal assessment of explanation and behaviour shift in alcohol users. It involves applying two 'process' models of addiction to the same sample of alcohol using subjects. Prochaska and DiClemente's Transtheoretical Model of Change (1979) is one of the more influential and widely used models in assessing and treating problem use of a wide range of substances. The model is based on philosophies of psychological measurement which assume literal meaning and veridical status of verbal report. The methodology employed by the Trantheoretical Model to measure an individual's 'Stage of Change' is that of forced choice questionnaire scores, which allow allocation into one of four stages of change. This method therefore assumes an individual's stage of change allocation reflects their current internal state with regard to their substance use. In contrast, the Functional Discursive model (Davies 1997) is based on philosophies of discourse and emphasises the predictive value of discourse. The model regards discourse as functional and indicative of the individual's current motivational state in relation to their substance use. Therefore the model postulates language is functional in context, and performative rather than informative. This model was developed in response to a growing dissatisfaction with many traditional methods of data collection in the field of substance use. The Functional Discursive Model is also very much a social model because the explanations surrounding the substance using behaviour occur within differing social contexts. The current study examines the efficacy and applicability of the two models which employ opposing methodological approaches. Using different treatment and non-treatment groups of alcohol users the study also assesses whether progression through the stages of either model differs according to the type of treatment intervention. In addition, two independent measures were chosen to provide additional data against which to assess anThis prompted the selection of Rosenberg's Self-esteem Inventory to measure levels of self-esteem across differing contexts (i. e. time, treatment approach and stage of alcohol use). The second independent measure included in the study was the Alcohol Use Disorders Identification Test (AUDIT), which was chosen to measure levels of problematic alcohol use across the differing contexts, to enable further comparisons to be made between the models and agencies. The general hypothesis states that the discourse based model which postulates that language is performative and context dependent will perform better across a range of contexts. The general hypothesis generates a number of more specific hypotheses which are detailed in the introduction. The following study aims to assess how the two different models perform in terms of these hypotheses. Differences and relationships between levels of self-esteem and levels of problematic alcohol use were found according to stage position in each model. For example, low self-esteem was found to be a feature of the stages (for each model) associated with problem alcohol use. Conversely, higher self-esteem was associated with non-problem stages. Use of the AUDIT confirmed patterns of alcohol use through the stages as proposed by the authors of each model. In addition, AUDIT scores closely mirrored self-esteem scores providing further confirmation that low self-esteem is associated with more problematic alcohol use. A clear relationship was also found between stage position in each model and type of treatment agency with which contact was made. Agencies supporting an abstinence-based approach to treatment were found to be treating only those individuals at the most problematic stages of each model. In contrast, agencies supporting a return to controlled drinking were found to treat individuals across a range of stages. Comparisons made between the two models suggest that while some overlap regarding the defining characteristics of individual stages was present, the Discursive Model can better accommodate a broader spectrum of substance use than is measured by the Stages of Change. Thus, the two models cannot be mapped directly onto each other. In terms of clinical application, the two models were found to be better suited to different treatment approaches. The Discursive Model appears better equipped to accommodate harm reduction and controlled drinking treatment approaches and the Stages of Change appear more pertinent to an abstinence-based approach.