Rhodes University Logo
hello@example.com
  • hello@example.com
  • info@example.com
  • addAdd another account...
Rhodes > English Language and Linguistics > Latest News

The evolution of natural language syntax by natural selection

Date Released: Tue, 11 September 2012 09:25 +0200

Rhodes linguist Mark de Vos delivered a talk at North-West University on Friday 24 August entitled "The evolution of natural language syntax by natural selection: optimization of the mental lexicon yields syntax for free".  His talk presented an account of the evolution of human language by natural selection that focuses on the evolution of the lexicon.  Here is a brief summary of what he said:

By developing an evolutionary narrative around the acquisition of the lexicon, I show that the same mechanisms used in lexicon organization are critical for recursion.  In other words, the communicative capacities imparted by natural language syntax were not the original locus of natural selection (cf Pinker and Jackendoff 2005) developed as a side-effect of lexical development.

One might caricature ideas about the evolution of language into two camps. On the one hand, there are those who believe that around 40 000 years ago, a sudden 'Great Leap' occurred: a mutation in the brain of previously mute Homo sapiens sapiens caused modern language to spring into existence.  The other position is that language evolved slowly and gradually with a smooth improvement in linguistic ability.  Unfortunately, most theories of synchronic syntax were never designed to explain language evolution (witness the statements by Chomsky and Gould that language is too-complex a system to have evolved by natural selection) and so seem to obscure more than to elucidate.  In the absence of adequate syntactic theory, the available paleontological evidence has been interpreted in both directions. Thus interdisciplinary work is a necessity -- as has been repeatedly called for (e.g. Hauser, Chomsky and Fitch 2002, Christiansen 2002, Tobias 2005, Kirby 2007 etc).

Some questions that arise are (a) was the evolution of language gradual or rapid; (b) what was the exact grammatical locus of natural selection (as opposed to vague statements like 'communication' or 'deception' which fall dangerously close to a teleological position); (c) do models have sufficient predictive and explanatory value to assist in interpretation of paleontological and comparative-behavioural data?

I argue that focusing on the evolution of syntax and recursion is misguided (Tinkoff and Hauser 2006) and is a case of asking the wrong questions.  As pointed out by many authors (Hauser 1996, Tinkoff and Hauser 2006, Kirby 2007, De Vos 2008), it is the size, speed of acquisition and flexibility of the human lexicon that is the true elephant in the room – and unfortunately, most research misses it.  Clearly, the gradual expansion of the lexicon has adaptive value but I argue that in order to acquire a lexicon, it must be structured and optimized. 

A theoretical basis of the structure and optimization of knowledge representation systems is, fortunately, provided by Normalization Theory (Codd 1973), a branch of mathematics.  These optimization formalisms – called Normal Forms – are clear, mathematically defined, universal and cognitively independent constraints applying to knowledge representations. Following De Vos (2008), it can be shown that syntax and normalization are equivalent. Consequently, the ability to normalize a knowledge representation system is equivalent to the ability to apply syntactic operations – and the evolution of these levels of normalization is presumably, therefore, equivalent to the evolution of syntax.

The advantage of this framework, is that, because each level of normalization is independently defined, it is possible to predict the likely linguistic output – and possibly cognitive ability – for an animal at each level.  The prediction that emerges is that lexical facility increased gradually and that this was mirrored by syntactic development which was severely constrained at lower levels of optimization.  However, once the normalization level called 4NF was reached, syntactic ability improved immediately and dramatically. 

Thus, this model predicts a position intermediate between the 'Great Leap' and the gradualist positions.  I also attempt to show that this is most consistent with the available paleontological evidence. 

Source: