Is dense codeswitching complex?

- Dense codeswitching as a new variety.
- Dense codeswitching does not involve switching.
- Dense codeswitching has own, new rules that aim for optimal simplicity.
- Dense codeswitching has innovative features not present in contributing languages.

In this paper the question is raised to what extent dense code switching can be considered complex. Psycholinguistic experiments indicate that code switching involves cognitive costs, both in production and comprehension, a conclusion that could indicate that code switching is indeed complex. In recent immigrant communities, code switching would potentially be even more complex, because of the instability of the sociolinguistic situation. However, speakers of the Dutch-Turkish immigrant community perceive of code switching as the 'most comfortable' way of talking. In this paper it is argued that in dense code switching there is no actual switching between languages involved. Some neurolinguistics models have been proposed in this line (e.g. Green and Li, 2014). It is assumed there that speakers draw from two separate linguistic systems. However, if dense code switching would really involve drawing from two complete linguistic systems, it would indeed be complex, not only in terms of processing costs, but also in terms of structure. In this paper it is argued that dense code switching may be better conceived of as a new 'code' in its own right, to which elements of both languages contribute. This new code may also contain innovative features that cannot be traced back to the contributing languages. This code may not necessarily be conventionalised (as is e.g. the case with so called Mixed Languages), but it may have some implicit norms or rules that make the code predictable. In this paper it is explored what such a code may look like. To this end digital code switch data from a bilingual Dutch-Turkish internet forum is discussed. It is argued that naturalistic data continue to be an important source of information about underlying processes of language production, and that digital data has certain advantages compared to oral data when it comes to uncovering patterns. The data reveal predictable patterns, both in a structural as well as a socio-psychological, and a discourse organisational sense. It is proposed that these patterns are, apart from contact effects, also shaped by universal mechanisms and sociolinguistic factors.