This study covers an area of sociolinguistic methodology which so far has received relatively little theoretical attention: the composition and analysis of orthographic transcriptions of non-standard speech. The following aspects will be addressed: first, the linguistic/semiotic properties of orthographic transcriptions, i.e. the relationship between graphemic representation and phonology, morphology and syntax. Secondly, the different functions of orthographic writing and its ‘ideological’ (popular and linguistic) dimensions (e.g. social stereotypes, linguistic distance). The discussion of these in a database of sample transcriptions of earlier African-American English illustrates that the interplay between the functional, linguistic and ideological dimensions of orthographic transcripts is highly complex, and that their analysis and interpretation have implications for (socio)linguistic methodology in general.