Note: This paper was written for P717 - "Evolution and Learning" - a class taught by William Timberlake at Indiana University, Spring Semester, 2000. The entire text is available here in PDF format. Below is the introduction and the bibliography. If you prefer another format or have any questions or comments, feel free to E-mail me.

Evolution and Learning of Language: Insights from Modeling

by Sean McLennan

April, 2000




Introduction

How can we characterize the dual contributions of biology and environment to language? This is one of the most hotly contested topics in linguistics, psychology, and related disciplines. Even the staunchest adherents to both ends of the “nature / nurture” spectrum must acknowledge that there are both components of genetics and learning to language: no one would claim that specific languages are encoded in genes, nor would anyone propose that humans learn to have brains. The question becomes where in that spectrum we draw the line.

Strong nativists (Pinker & Bloom (1990) for example) hold that even aspects of grammar are genetically determined and thus can be selected for via natural selection. This accounts for the kind of cross-linguistic regularity that characterizes “Universal Grammar” – the linguistic knowledge that all humans possess. On the other end, strong developmentalists would maintain that the creation of neurons is as far as biology contributes; before the contribution of environment, the brain is just a ball of “connectoplasm”. Any regularity in language is the result of regularity in the environment and other physiological regularity (for example, we all have eyes and ears that respond to stimuli in highly structured and similar ways). Thelen & Smith (1993) is representative of the approach taken by researchers on this end of the spectrum.

Nativists refer to reports like Gopnick (1990) which describes the KE family who show a language deficit that appears to be caused by a recessive gene. Developmentalists point to the profound cortical plasticity of the brain and its ability to self-organize; for example, within the visual system which is relatively well understood (Gazzaniga et al. 1998, pg 123-134, 484-493). As with most arguments, the truth likely lies somewhere in the middle; from the preface of Elman et al. (1996: xi-xii):

“The obvious conclusion is that the real answer to the question, Where does knowledge come from, is that is comes from the interaction between nature and nurture, or what has been called ‘epigenesis.’ Genetic constraints interact with the internal and external environmental influences, and they jointly give rise to the phenotype.”

However, the authors go on to rightly point out: “Unfortunately, as compelling and sensible as this claim seems, it is less a conclusion than a starting point. The problem does not go away, it is simply rephrased.” (Elman et al. 1996: xii). No line has been drawn.

Particularly in studying the evolution of language and other cognitive capabilities, researchers must address the question of whether their pursuit is viable in the face of a profound lack of information. We have no way of obtaining direct information about the cognition of our ancestors. Clearly, Lewontin (1998: 128-129) has little respect for the study:

“Wherever cognition came from, one would like to make arguments about the forces of natural selection that established it. On the face of it, generalized problem solving and linguistic competence might seem obviously to give a selective advantage to their possessors. ... stories about how the species as a whole would be benefited by speech are not to the point. ... The problem is that we will never know [the factors involved] and never will. There is no end to plausible storytelling.”

Lewontin’s attack on the study of evolution of cognition is so vehement as to border on the belligerent. He continues to conclude (pg 130):

“Finally, I must say that the best lesson our readers can learn is to give up the childish notion that everything that is interesting about nature can be understood. History, and evolution is a form of history, simply does not leave sufficient traces, especially when it is the forces that are at issue. Form and even behavior may leave fossil remains, but forces like natural selection do not. It might be interesting to know how cognition (whatever that is) arose and spread and changed, but we cannot know. Tough luck.”

Despite Lewontin’s overly emotional, and perhaps condescending attitude, his point is valid. Without evidence that is more conclusive than speculative, studying the evolution of cognition is no more than telling “just-so” stories.

Botha (1997a, 1997b, 1998a, 1998b., 2000), in a series of articles that discusses Neo-Darwinian accounts of language, particularly focusing on Pinker & Bloom (1990), systematically criticizes both Pinker & Bloom (1990) but also their detractors. Botha’s point is not to take a stance on either side of the issue, but to point out some of the fundamental flaws that hamper the proper exposition of the arguments. Perhaps the most damning comes from Botha (1997a) in which he systematically shows that no one involved in the argument is discussing the same ontological entity. Even within Pinker and Bloom (1990), the evolving linguistic entity under scrutiny varies wildly (Botha, 1997a: 253).

With such criticisms being brought to bear against the entire domain of study, the question moves away from “where do we draw the line between nature and nurture?” to “can examining the evolution of language lead to a better understanding of language?” There is one area of language evolution research that seems to adequately acknowledge evidential restrictions, refrain from “just-so” storytelling, and provide profound insights into the nature of the interaction between language evolution and language learning. That area is modeling, the primary focus of this discussion.

It is worth quoting the introduction to Batali (1998) at length for it addresses (at least indirectly) many of the issues brought forth above:

The attitude extant in this passage is one that is common among modelers of all varieties, and one that seems poorly understood by other disciplines. That is, that models are to be used as a tool for exploring assumptions, consequences, and the mapping of input to results. Indeed, it is a fundamental caveat of modeling that regardless of how well the model matches the data, never make the leap from “model as description” to “model as explanation”. It is clear that this is a vital assumption in Pinker and Bloom’s (1990) argument and could possibly account for a great deal of the controversy surrounding it.

What follows is a review of three models of the evolution and learning of language and what they have brought to our understanding of the roles of biology and environment in language acquisition: Werner and Dyer’s (1991) model of the evolution of signaling systems, Batali’s (1998) model of the emergence of combinatorial grammar, and Kirby and Hurford’s (1997) model of the emergence of a “Language Acquisition Device”. I hope to make it clear that although discussions of the evolution of language are rightly suspect, it is not, as Lewontin would suggest, a “childish” pursuit.


References:

Botha, R. (1997a). Neo-Darwinian accounts of the evolution of language: 1. Questions about their explanatory focus. Language and Communication 17:3, 249-267.

Botha, R. (1997b). Neo-Darwinian accounts of the evolution of language: 2. Questions about complex design. Language and Communication 17:4, 319-340.

Botha, R. (1998a). Neo-Darwinian accounts of the evolution of language: 3. Questions about their evidential bases, logic and rhetoric. Language and Communication 18, 17-47.

Botha, R. (1998b). Neo-Darwinian accounts of the evolution of language: 4. Questions about their comparative merit. Language and Communication 18, 227-249.

Botha, R. (2000). Discussing the evolution of the assorted beasts called language. Language and Communication 20, 149-160.

Deacon, T. (1997). The Symbolic Species. Norton: New York, NY.

Donald, M. (1993). Précis of Origins of the modern mind: Three stages in the evolution of culture and cognition. Behavioral and Brain Sciences 16, 737-791.

Elman, J., E. Bates, M. Johnson, A. Karmiloff-Smith, D. Parisi, and K. Plunkett. (1996). Rethinking Innateness. MIT Press: Cambridge, MA.

Chomsky, N. (1986). Knowledge of Language: Its Nature, Origin and Use. Praeger: New York, NY.

Cook, V. and M. Newson. (1996). Chomsky’s Universal Grammar. Blackwell: Cambridge, MA.

Gazzaniga, M., R. Ivry, and G. Magnun. (1998). Cognitive Neuroscience. W.W. Norton & Company Inc: New York, NY.

Gopnick, M. (1990). Feature-blind grammar and dysphasia. Nature, 344(6268), 715.

Kirby, S., and J. Hurford. (1997). Learning, culture, and evolution in the origin of linguistic constraints. in P. Husbands and H. Harvey (eds.) Fourth European Conference in Artificial Life. 493-502. MIT Press: Cambridge, MA.

Lewontin, R. (1998). The evolution of cognition: questions we will never answer. In D. Scarborough, and S. Sternberg (eds.) An Invitation of Cognitive Science Vol 4: Methods, Models, and Conceptual Issues. MIT Press: Cambridge, MA.

Pinker, S., and P. Bloom. (1990). Natural language and natural selection. Behavioral and Brain Sciences 13, 707-784.

Thelen, E., and L. Smith (1994). A Dynamic Systems Approach to the Development of Cognition and Action. MIT Press: Cambridge, MA.