Early language Conclusions
We observed the use of old means for novel purposes, and we showed how people constantly create novel usage patterns and functional categories by using existing linguistic material, extending it to new contexts and endowing it with new grammatical functions. We showed how this general strategy can be held responsible for human language developing into an increasingly more refined tool of communication and conceptualization. It is hoped that our discussion has demonstrated that there is nothing mysterious about complex structural phenomena such as syntactic recursion or displacement, which fall out naturally from a long sequence of applications of that very strategy.
A number of the hypotheses discussed here have been proposed in some form by other authors (especially Givo ´n 2002a, 2005; see also Jackendoff 2002; Tallerman 2007); what distinguishes our presentation from those works is that it is based strictly on the application of grammaticalization theory. This restriction has the advantage that our hypotheses can be falsified with reference to this theory; at the same time it also had enormous disadvantages. Language evolution has become a paradigm Weld of interdisciplinary research, where new findings made in areas such as evolutionary biology, psychology, psycholinguistics, neurology, genetics, palaeo-anthropology, the computer sciences, etc. have been or can be combined in order to reconstruct what a few decades ago still seemed to be beyond the scope of empirical research. In an attempt to be consistent in our methodology, we had to ignore all this work except for occasional cross-references.
For example, recent studies in computer simulation show independently that complex syntactic rules can emerge out of quite simple systems, such as neural nets, which have a small number of initial assumptions and learn from imperfect inputs, with numerous words acquired by observational learning (e.g. Batali 1998; Kirby 2000; Steels et al. 2002; Tonkes and Wiles 2002; Davidson 2003). Some of these simulations exhibit striking similarities to our grammaticalization scenario, and it would be tempting to relate the two to one another. However, as long as it remains unclear how exactly these similarities are to be defined, we refrain from pursuing this issue—hoping that future research will be able to draw on a more comprehensive basis of analysis.
In the preceding section we tried to find answers to questions that we consider to be relevant for understanding language genesis and evolution. But there were many other questions in addition that have been raised by students of language evolution. Suffice it, in concluding, to mention one question that was ignored here, namely to what extent language is, or is part of, an innate human faculty, and whether language genesis should not be accounted for with reference to such a faculty. This is in fact a much-discussed issue, but it is also an unresolved one. For our purposes it was of secondary import since our account of the genesis of grammar did not require any assumptions on innateness. Accordingly, irrespective of whether future research will establish that language evolution is, at least to some extent, determined by innate mechanisms, this would not seem to affect the findings presented here in any dramatic way.
