Code-switching comes in three major sub-types: insertion, alternation, and congruent lexicalization. Turkish-Dutch code-switching is supposed to feature the first two types but not the third, because when the languages in contact are typologically distinct, there is not enough shared lexicon and structure to make the fine-grained integration of the two languages possible that is the basic requirement for congruent lexicalization. Recent data of third generation Dutch Turks, however, shows some instances of code-switching that do not rely as much on keeping the two languages separate as the prototypical instances of insertion and alternation do. This empirical development lends support to a theoretical approach to linguistics that makes use of an expanded definition of the lexicon. Many of the complex instantiations of code-switching qualify as both insertion and alternation, because the inserted unit is so large that it involves a complete switch to the other language at the same time.