This thesis is an examination of the effects of translation into French and of adaptation for the stage, in English, on the dialogue of Philip Pullman’s novel Northern Lights (published in North America as The Golden Compass). The study focuses on the speech of Lyra, the novel’s protagonist, in terms of both its linguistic qualities and the functions it supports within the novel and the trilogy of which it forms part, His Dark Materials. The study aims to identify the ways in which not just the linguistic surface of fictional speech is affected by translation and adaptation but also the degree to which the roles played by the dialogue in the source text are reflected or transformed in the different versions. The unusual research design, involving a comparison of the effects of interlingual translation and intermedial adaptation on the same text, consists of two main elements. In the first quantitative section, the relative incidence of three variables is measured for the purposes of identifying how features of spoken style and non-standard variation are treated. This analysis is followed by a detailed qualitative evaluation of a small number of dialogue passages that exemplify the key linguistic features and likely textual functions of Lyra’s speech in the novel. The passages concerned are compared with equivalent stretches of dialogue in the French translation and the theatrical script. The study finds evidence to suggest that Pullman uses dialogue in support of characterisation, plot, and also ideological and intertextual concerns. All of these aspects are affected, in subtle but significant ways, by the different decisions made by the translator and the dramatist in respect of Lyra’s speech. The study also finds that aspects of both user-related and situation-related variation in fictional speech may be worthy of further research.

Ce mémoire porte sur les transformations sociohistoriques du gayle en Afrique du Sud, une variété linguistique de l’afrikaans et de l'anglais, au répertoire lexical souvent improvisé. S’appuyant sur six mois de terrain ethnographique au Cap, ce travail se base sur des données (méta)linguistiques tirées d’observation participante ainsi que d’entretiens individuels et de groupe au sein de la communauté coloured. Je critique une perspective qui envisage le gayle comme une « anti-langue » (Halliday 1976), une méthode de communication secrète et exclusive à une communauté fermée. Plutôt, je suggère que le phénomène est mieux compris lorsqu’abordé en tant que registre d'une communauté de pratique (Eckert et McConnell-Ginet 2003). Mes entretiens révèlent son association historique et « enregistrée » avec la figure caractérologique (Agha 2007) de l’homosexuel efféminé coloured ou « moffie ». Toutefois, les participants expriment un malaise face à cette idéologie puisqu’ils sont conscients que cette construction « efface » (Gal et Irvine 2000) les locuteurs non canoniques du gayle. Au-delà de la description lexicale, mon analyse suggère des chevauchements entre les caractéristiques sémantiques et pragmatiques de la variété, toutes deux stratégiquement déployées pour invoquer des caractéristiques évanescentes et des attributs sociaux dans l’interaction (Butler 1990). Au même titre que les éléments du répertoire ne correspondent pas à des significations sémantiques fixes – mais plutôt à des réalisations discursives contingentes – l’utilisation du gayle ne se limite pas à des identités sociales statiques. L'appartenance à la communauté est plutôt marquée par l’utilisation légitime et la compréhension mutuelle de déviations lexicales créatives et contextuelles. / This thesis focuses on sociohistorical transformations of Gayle, a linguistic variety consisting of an often-improvised lexical repertoire superimposed upon either Afrikaans or English in South Africa. Based on six months of fieldwork in Cape Town, this thesis draws on rich (meta)linguistic data drawn from participant-observation, as well as individual and group interviews within the coloured community. I critique an 'anti-language' (Halliday 1976) perspective, wherein Gayle is seen as uniquely anchored to a closed community for which it serves as an exclusive, ingroup code. Rather, I suggest that Gayle is best understood when approached as the register of a community of practice (Eckert & McConnell-Ginet 2003). My interviews expose the historical and 'enregistered' association of Gayle with the racialized, gendered and sexualized ‘characterological figure’ (Agha 2007) of the effeminate coloured gay man or 'moffie'. This association serves as a backdrop to contemporary anxieties about the ways in which Gayle is mapped onto speakers’ social identities and perhaps shows the respondent’s consciousness that these formulations ‘erase’ (Irvine and Gal 2000) non-canonical users of Gayle. Moving beyond lexical description, I suggest overlaps between the semantic and pragmatic features of the variety, both strategically deployed to invoke evanescent characteristics and social attributes in interaction (cf. Butler 1990). Just as elements of the Gayle repertoire do not code for fixed semantic meanings – but are rather contingent, discursive achievements – Gayle does not 'code for' static social identities. Instead, it achieves group belonging in interaction through legitimate use and mutual comprehension of creative linguistic 'deviations' and 'subversions'