This thesis provides an interactional account of some multilingual usage patterns found in Kuala Lumpur, Malaysia. Kuala Lumpur, the capital city and major commercial center in this internationally-focused country, can be described as a ‘high contact’ area, where several languages (the major ones being Malay, Sinitic (especially Cantonese, Southern Min and Mandarin) and Tamil) are in intensive and extensive contact. The approach taken here argues that an important aspect of exploring language usage patterns, especially in understudied locales such as Kuala Lumpur, necessarily involves situating the investigation in the wider local context, since emergent patterns of language use also reflect reflexivity between speakers as social actors and salient macro-level conditions (Agha 2007, van Dijk 2008a,b, Stroud 2004, Gumperz 1992, 1982, amongst others). I first elaborate on two salient social conditions in Kuala Lumpur: (1) pluralistic organization of Malaysian society through three ethnic constructs – Malays, Chinese and Indians, and (2) expression of cosmopolitanism of mainstream Kuala Lumpur. Next, using procedures from interactional linguistics developed to account for code-switching elsewhere (Gafaranga 2005, Auer 1985), I demonstrate how such social conditions as schematic representations can organize language selection patterns. I then apply these procedures to the exploration of usage patterns of specific language units. Some of them, e.g., discourse particles lah and ah24, person reference forms, and topic prominence, are well-researched, but typically from the context of particular languages. Others, such as the entry of English discourse marker like into KL speech, have not been researched in any depth. What these usage patterns share in common is that they regularly occur across more than one language. Departing from a language-particular approach, I demonstrate that we can analyze usage patterns through another dimension - a multilingual usage perspective. I propose that alongside other considerations, there are also interactional reasons for such convergences. Interactional reasons can emerge from local context of the text or from situational aspects, such as participant backgrounds and discourse topic. I show that in these multilingual usage patterns, traces of social conditions are present, via indexation of social schemas, or indirectly, through the action of actually selecting multilingual patterns.