Philosophy-Neuroscience-Psychology Program
Washington University, St. Louis, MO, USA
tad@twinearth.wustl.edu

This paper examines the relative merits of two competing paradigms for explaining the evolution of natural language. According to the standard account, competence in natural language evolved when a grammar, capable of mapping propositional thought structures onto a serial medium of communication, was selected for in human pre-history.(Pinker & Bloom 1990) According to the 'neo-Vygotskyan' alternative, competence in natural language co-evolved with a capacity for propositional thought, as the result of "cumulative cultural evolution" (Boyd and Richerson 1996) in populations capable of a non-propositional form of cultural learning.(see especially Tomasello et. al. 1993; Tomasello & Call 1997) The discussion proceeds in two parts. In the first part, I summarize the standard account, and motivate the neo-Vygotskyan alternative by discussing two problems with it. In the second part, I attempt to make the neo-Vygotskyan alternative palatable by defusing some obvious objections to it.

According to Pinker and Bloom (1990), natural language is a clear example of complex, adaptive design, and, as such, its evolution is best explained by the mechanism of natural selection. They argue that the complex design apparent in the grammar of natural language is ideally suited for the function of mapping propositional thought structures onto a serial medium, for the purposes of communication: grammars for spoken languages must map propositional structures onto a serial channel, minimizing ambiguity in context, under the further constraints that the encoding and decoding be done rapidly, by creatures with limited short-term memories, according to a code that is shared by an entire community of potential communicants . . . language is a complex system of many parts, each tailored to mapping a characteristic kind of semantic or pragmatic function onto a characteristic kind of symbol sequence.(Pinker & Bloom 1990, 713)

Clearly, this explanation of the evolution of natural language implies that some non-language-using, evolutionary ancestor of humans had propositional thought structures to communicate, without the capacity to communicate them. If the function that natural language competence was selected for was the communication of propositional thought structures, then propositional thought must have pre-dated natural language.

There are two major problems with this picture. First, it is far from established that our non-language-using precursors wielded a language of thought. If the cognitive accomplishments of our nearest primate relatives can be accounted for without attributing a system of propositional thought to them, then this is prima facie reason not to attribute propositional thought to our common evolutionary ancestor. Currently, there exist numerous research programs based on the premise that we do not even need to attribute propositional thought to humans, in order to explain most human behavior, let alone other primates.(for a nice survey, see Clark 1997) Thus, it is far from clear that explanations of the evolution of natural language can help themselves to the assumption that our non-language-using precursors wielded a language of thought.

The second problem is apparent in an uncontroversial empirical fact. Our closest primate cousins do not use natural languages. But our most recent, non-language-using precursors were descendants of an ancestor that we share with our closest primate cousins. Thus, if our most recent, non-language-using precursors wielded a system of propositional thought, then it is likely that the ancestor that we share with our closest primate cousins did. If this is granted, then the standard account seems to imply the following: whereas, in the 5 to 7 million years since proto-chimpanzees and proto-hominids diverged from the common evolutionary ancestor (Pinker & Bloom 1990, 726), hominids have managed to develop a system to communicate their propositional thought, chimpanzees have not. Pinker & Bloom (Ibid., 724-725), as well as other defenders of the standard account, try to come up with ecological scenarios to explain this, often focusing on the unique challenges of a hunter-gatherer ecology. However, it is obvious to any student of chimpanzee ecology, that the adaptive benefits of a system for communicating propositional thought, if chimpanzees had it, would be considerable. Therefore, the question remains, if chimpanzees and their precursors have had propositional thought for just as long as humans and their precursors, why did a system for communicating propositional thought only evolve in humans?

There are two possible responses to this question. First, one might question the assumption that the evolution of natural language required the pre-existence of a system of propositional thought. However, if the function that competence in natural language was initially selected for consisted in the communication of thought, then surely we must follow Pinker & Bloom (1990) in defending this assumption. Natural language is not a good tool for communicating other kinds of cognitive state, like emotions, or representations encoding Euclidean relations.(Ibid., 715) Thus, if one questions the assumption that the evolution of natural language required the pre-existence of a system of propositional thought, one must also reject the assumption that natural language was initially selected to communicate thought. As I argue in the longer version of this paper, this option is most consistent with the neo-Vygotskyan alternative.

Second, one might argue that the brief period of evolutionary time, separating humans from the ancestor we share with chimpanzees, saw the evolution of propositional thought first, followed by the evolution of language. However, given the relatively brief period of time involved, it is exceedingly hard to construct ecological scenarios that can explain this progression. The only plausible scenario seems to call for some kind of self-amplifying, co-evolutionary dynamic between language and propositional thought, in which more language-like forms of communication led to more proposition-like forms of thought, and vice versa. However, as I argue in the longer version of this paper, this is also more compatible with the neo-Vygotskyan alternative, than with the standard account.

According to the neo-Vygotskyan alternative, the evolution of language should be understood in the context of cultural evolution. For example, Tomasello, et. al. (1993), and Tomasello & Call (1997), suggest that the key cognitive divergence between the evolutionary precursors of hominids and the ancestor we share with chimpanzees, consisted in a capacity to create, transmit, and elaborate cultural practices. This led to a kind of 'cultural evolution,' among the products of which was natural language. By learning to use this product, humans learn to wield a propositional cognitive system.

I want to address two obvious problems with this alternative. First, on some understandings of what 'culture' and 'cultural learning' consist in, this is not really an alternative to the standard account. Many theorists would argue that the capacity to create, transmit, and elaborate cultural practices requires the kind of propositional thought that Pinker and Bloom claim natural language was selected to communicate. Therefore, in order to constitute a credible alternative to the standard account, the neo-Vygotskyan account must provide a definition of 'culture' and 'cultural learning' that does not presuppose the kind of propositional thought presupposed by the standard account.

Second, even if an acceptable definition of 'culture' and 'cultural learning' that did not presuppose propositional thought were provided, the connection between such phenomena and natural language would still appear mysterious. The great advantage of the standard account is that it explains why natural language has precisely the structure that it does: its main function is to communicate thought with analogous structure. Why would culture and cultural learning based on non-propositional forms of thought lead to the evolution of a system of communication with the structure of natural language?

In the long version of this paper, I suggest some strategies for defusing these two serious problems. With regard to the first problem, I turn to the animal culture literature, in search of definitions of 'culture' and 'cultural learning' that do not presuppose a capacity for propositional thought. Following Tomasello, et. al. (1993), I suggest that any population that displays the "ratchet-effect" (495), or what Boyd & Richerson (1996) call "cumulative cultural evolution" (79), should count as displaying culture and cultural learning. Such phenomena do not require propositional thought. They merely require mechanisms of "social canalization" (Boesch 1996, 257), like fairly rudimentary capacities to imitate models. There is ethological evidence that certain chimpanzee populations display such phenomena.(Ibid., 255-265) Furthermore, there is neurobiological evidence that many primates have neural mechanisms capable of implementing imitative learning.(Arbib &Rizzolatti 1996)

Given this understanding of 'culture' and 'cultural learning', the neo-Vygotskyan alternative seems threatened by the same sorts of objections as the standard account. If some chimpanzee populations display evidence of culture and cultural learning, and if many primates have neural mechanisms capable of implementing cognitive capacities necessary for culture and cultural learning, then why is there no evidence for the cultural evolution of natural language in non-human, primate species? In response to this worry, I draw on Boyd & Richerson's (1996) formal, evolutionary argument for the claim that, while culture may be common, cumulative cultural evolution is inevitably rare.(82-88)

Finally, I conclude the paper by offering some speculative suggestions for defusing the second serious problem with the neo-Vygotskyan alternative: if language is the product of cumulative cultural evolution based on non-propositional forms of cultural cognition, then why does it have the structure that it has? I suggest that the phylogenetically earliest function of language-like systems of communication consisted in supplementing imitation, in the transmission of ecologically crucial, cultural practices from parents to offspring. Proto-language may have consisted in a gestural form with a mimetic function: by miming hierarchically organized sequences of gestures involved in tool use, parents could enhance the transmission of ecologically crucial, tool-using practices to offspring. Such communicative behaviors would inherit the combinatorial structure of the tool use that they mimicked, and would thereby constitute an early form of a combinatorial, communicative system, like natural language.

 Conference site: http://www.infres.enst.fr/confs/evolang/