Clause subordination Conclusions
It is hoped that we have shown that it is possible to find an answer to what Bickerton (2005) calls the ‘‘last question’’ on language evolution. Discussion was confined to very few structures of clause combining; however these structures illustrate the two main mechanisms by which patterns of clause subordination may arise: Either via the reinterpretation of thing-like concepts as propositional concepts, leading from nominal to clausal participants (i.e. expansion; “Clause subordination Expansion”) or, alternatively, two juxtaposed sentences [S1 + S2] that grow together into one sentence [S1 [S2]], where one sentence assumes the function and subsequently also the morphosyntactic structure of a subordinate clause (integration).
Thus, wherever there is appropriate data it turns out that clause subordination can be traced back to non-subordinated sentences. This does not mean that in the languages concerned there was no earlier subordination. In accordance with the bridge hypothesis that we adopted, however, we hypothesize that at some stage in the development of human languages there was a situation where the processes sketched happened for the first time, that is, where there were no complex sentences but where the means that were available in simple sentences were rearranged in novel ways to create complex sentences.
The findings made are at variance with claims that have been made that there was no intermediate stage in the rise of clause combining. For example, Berwick (1998: 338–9) rejects the idea that language could ever have been dramatically different from what it is now, since ‘‘there is no possibility of an ‘intermediate’ syntax between a non-combinatorial one and a full natural language’’ (see also Newmeyer 2002: 363 for discussion). According to the findings on grammaticalization, combinatorial syntax does not belong to the earliest layers of language evolution; formal means of clause subordination arise only at a more advanced stage of evolution, namely at layer IV.
And these findings also cast doubts on scenarios proposing a two-stage process, such as those of Bickerton (1990, 1995) or Carstairs-McCarthy (1999). Both assume that there are linguistic correlates to the distinction between Homo erectus and Homo sapiens, in that the former had no genuine syntax and that it is only the latter that acquired ‘‘modern’’ syntax with its recursive structure and everything else that goes with it. As our findings suggest, there was not one intermediate stage but rather an entire series of stages in the evolution of combinatorial syntax. However, we are restricted here to the analysis of morphosyntactic processes.
