Following recent efforts by Acton (2021) and Eckert (2019) to bridge gaps between pragmatics and sociolinguistics, this study looks at the interplay between pragmatic-functional factors and social indexicalities lying behind discourse marker selection in multilingual settings. The case study it proposes is non-English discourse markers in Namibian English, a postcolonial English variety set in a multilingual context. The study's methodological approach proposes to make Schneider's (2021) variational pragmatic framework more compatible with studying variation in multilingual settings by looking at multilingual speech data elicited from informants observed across contexts differentiated according to ethnolinguistic background distribution. The study finds that some discourse markers are overtly or covertly transferred along with their pragmatic functions across indigenous languages, Afrikaans, and English. It also finds that social indexicalities mobilized for social persona construction constitute a potent if not overarching factor in discourse marker selection: As it turns out, Coloured Afrikaans discourse markers rather than indigenous or English ones constitute the common core of non-English discourse markers that symbolically mark specific Namibian English varieties as simultaneously ‘Black’ and urban rather than as ‘traditional ethnic’ or ‘White’.