The analysis of a literary corpus of West African and South East Asian prose (1954-2013) reveals that what appears, at least at first sight, to be a mimetic representation of non-standardized varieties of oral English in fact belongs to the symbolic, if not stereotypical order. The representation of character speech and thoughts functions in a tension between a commitment to represent linguistic alterity and the necessity to be understood by foreign readers who are not familiar with the socio-historical and linguistic context. Systematic annotation and quantitative analysis of the corpus show that the strategies used in the two regions differ at different levels: linguistic and metadiscursive. At the linguistic level, South East Asian texts largely resort to code-mixing whereas West-African texts predominantly have recourse to grammatical representations of local varieties of Engish. At the metadiscursive level, lexical foreignness is marked by italics in South East Asian texts; West-Afriacn texts resort to italics, translation and commentaries.