Written chatspeak is said to be marked by two basic principles: (1) write like you speak and (2) write as fast as possible. As for Flemish chat language, the first principle seems to result in a multilayered mixed code, in which dialectical, substandard Flemish and standard Dutch features interact in an eclectic way. In addition, most of the chatters insert English words in their chat discourse as well. This intensive code mixing is assumed to be – at least to a considerable extent – a reflection of the daily speech of these Flemish chatters. But what about the validity of this assumption? Can chatspeak function as an alternative dataset for the study of (spoken) language variation and change and thus as a research tool for e.g. the study of Flemish teenage talk and the representation of non-standard speech in spoken interaction? The dependent variables for the present test case are two substandard Flemish (or ‘tussentaal’) features that urge the chatters to violate the second principle, since their use implies an extension of the utterance. The central question is whether the second principle prevents the use of these substandard forms in Flemish chatspeak. In other words, do the analyses undermine the validity of using written chat corpora as a graduator for speech variation? We finish with a small excursion on the use of English by the Flemish chatters: can we separate English insertions that are triggered by the chat medium from English insertions that are not?