Translation scholars have, in recent decades, sought to distance their theories from the influence of structuralist linguistics. Yet, they have continued to rely on concepts such as source language and target language, which assume languages to be mutually exclusive, normalised entities. Little attention has thus been paid to varieties form mixed socio-cultural contexts which do not meet the criteria of linguistic norms. This article argues that the reliance on linguistic norms makes translation theory structuralist in orientation. By contrasting the language context of normalised languages with that of non-stnadard varieties, it sheds light on the challenges the latter pose to conventional translation theories, and proposes a view of translation as text transformation within language rather than between languages.