The Evolution of the Evolution of Language
 
Matthew Bronson



I. Introduction and Overview of Paper
As long as there have been people, someone somewhere has been asking, Where does language come from? Indeed, this question is like a "strange attractor," an archetypal mystery whose gravity captivates the imagination of each generation, whether their tribe is Maidu or M.I.T. This is not to imply that each generation or researcher exactly agrees with anyone else; indeed, the space between competing narratives about the origin and evolution of language shows it to be a kind of cultural Rohrschach test, a site better suited for an investigation of the social construction of inquiry in the Human Sciences than for research in a "normal science" modality. The eternally contentious nature of the debate about the simplest terms is attested by the move of the Royal Scientific Society of Paris in 1866 to ban all further discussion on the topic (Pinker 1994).

This same contentious atmosphere characterizes our own time precisely because of the pre-paradigmatic character of our current understanding of these issues (which qualifies this line of investigation as "dubious science" in Foucault's estimation (viz. Dreyfus & Rabinow, 1986:11) as well as the multiplicity of data sources and theoretical orientations invited by such an inherently transdisciplinary focus of inquiry. Additional complexities include the lack of tangible artifacts in which to discern transitional stages in the development of full-blown human language. The words of "proto-language" posited by many researchers (but especially, Bickerton, 1990) allegedly spoken by our long lost ancestors (c. 100,000 years or earlier) must have become "full-blown language" by the time their descendants began to paint the startlingly beautiful bison in the caves at Lascaux and Alta Mira some 30,000 years ago; it is hard to imagine otherwise when confronted by the unmistakably and eerily modem sensibility of Paleolithic Cave Art. Unfortunately, the words of this putative "simpler language" no longer echo in our valleys. We also do not have direct access to the brains of those who came before us from one hundred years ago, not to mention those who walked the earth some 2,000,000 years at the time when apes and humans parted ways evolutionarily; along with the brains and soft tissues of all those who came after, these have long since vanished into dust.

The lack of a clear record has not proven to be a substantive deterrent just as the 1866 prohibition against further inquiry into the evolution of language didn't hold long. This makes sense once it is clear that this question is a crucible in which any theory of language must be tested; that is to say, any stand as to the language must necessarily explain how that nature came to be.

In a time when Darwinian evolution is the accepted paradigm in biology, an explanation about how language came to be must optimally conform with the generally understood tenets of that paradigm in order to gain the widest legitimacy. When people like Stephen Pinker and Terrence Deacon, who supposedly disagree on fundamental principles, show up at the party having wrapped themselves in the same flag of Darwinism, it takes some careful reading to see who really deserves to wear it. I guess (to give away my punch line early) if it were my party they both could wear it, though Deacon's colors would be truer to the Master's.

The purpose of this paper is to situate current inquiry concerning the evolution of language by examining how two very important thinkers in this area, Stephen Pinker (1994), a psycholinguist, and Terrence Deacon (1997), a biological anthropologist, have approached it in their recent books on this topic. Since this is a small paper and these are both very ambitious books, I have chosen two key areas in which a critical reader might discern the commonalities and differences between these perspectives (chimp language and creoles), and ultimately what each might contribute to a more encompassing view.

I have discovered in the process that it is good to resist the temptation to take what the writers themselves articulate as fundamentally different points of view at face value. For every place where the "co-evolutionary" view of Deacon and the "nativist" or "innatist" perspective seemingly diverge in a meaningful way, there are any number of corresponding points of overlap. For example, Deacon's "cognitive biases" could be a notational variant of Pinker's "language instinct". In a curious way, the "continuities" and "discontinuities" between their discourses mirror the tension between incremental and punctuated equilibrium (as in Bates's "Big Bang" attributed sarcastically to the Chomskyans in Pinker (1994:342)) models of evolution.

There is plenty of evidence both that language is special and that it isn't really so special; it all depends on how you look at it. Many of the cases that Pinker takes as clear evidence of his view (that it IS special) can be shown to be consistent with alternate explanations. In the common parlance of the social scientist: your point of view depends on whether you are a splitter (A la Pinker) or a "lumper" (A la Deacon, Bates. et al.) In a more perfect world, there would be more general recognition of the validity of both views and more concern with what the productive questions are that need to be asked and answered by all concerned.

The way that these gentlemen make use of the chimpanzee language debate, and pidgin/creole language data to build their respective cases is emblematic of the overall timbre of their arguments. I will begin with a brief prima facie articulation of how these thinkers apparently differ in their orientations. I will then consider very briefly how their positions contrast in the areas mentioned and even how much more alike than different they really are. Finally, I will consider the wider significance of what I have learned in the process of this query.

II. Deacon vs. Pinker. Paradigm Difference or Notational Variants?
In an attempt to create a matrix in which to compare and contrast what I saw as three contrasting paradigms in the materials I was exposed to in this class, I think I began to think more clearly about the competing claims and orientations. I realized that many of the root issues were more philosophical and metaphorical than methodological inasmuch as Pinker and Deacon draw from virtually the same bodies of neuroscience and other research to support their positions. There is enough here to explain their differing interpretations of the same data and to set the background for a deeper consideration of their commonalities. (Please consider this as a "work in progress" meant to help clarify thinking and raise questions rather than as a final taxonomy). In a sense, my raw data were the statements of the men themselves, and as such, I leave open to the conclusion whether all of these distinctions are valid or non-trivial. I have also included my own certainly defective version of a cognitivist position since it figures as a player in the work of both men, though I don't actually discuss it in the paper as I did in the presentation.

The following table was carefully constructed by examining the entire work of both men. I can quote Deacon directly on Pinker to give a flavor for the debate, but unfortunately not vice versa. Deacon (1997) references Pinker's work directly no less than five times and if one were to count the numerous side jabs and innuendoes directed at the nativist camp, the number would surely be three times that. I don't think it would be an overstatement to say that Pinker was a principle rhetorical foil for Deacon. Given two major books on the same theme in three years perhaps this is to be expected. Here's Pinker to start:

Knowing a language then is knowing how to translate mentalese into strings of words and vice versa. People without language would still have mentalese, and babies and non-human animals presumably have simpler dialects. Indeed if babies did not have a mentalese, to translate to and from English, it is not clear how learning English could take place, or even what learning English would mean." (Pinker, 1994:82)
Deacon (1997:27) takes this to be a statement of extreme "innatism" (called "nativism" in my chart). He uses much of chapter one of his book to show that Pinker's ideas are not consistent with evolutionary biology. His "language" organ can only be understood in metaphorical terms according to Deacon; he goes so far as to reject Pinker's comparison of language to wings and eyes as a "spurious analogy." Deacon shows himself a respectful adversary when he states, later in the same chapter:
... Pinker, a proponent of the Universal Grammar view of language abilities and an articulate champion of many of Chomsky's original insights about the uniqueness of language, argues ... that innate grammatical knowledge is not at all incompatible with an adaptationist interpretation of its origins. He argues that a language instinct could have gradually evolved through the action of natural selection. This is a far more plausible alternative to miraculous accidents and it challenges us to face some of the more difficult challenges ignored by theories relying on miraculous accidents to fill in the gaps. On the other hand, an adequate formal account of language competence does not provide an adequate account of how it arose through natural selection and the search for new structures in the human brain to fulfill this theoretical vacuum, like the search for phlogiston, has no end point. (Deacon 1997: 38)
I don't think it would be an unfair summary that Deacon has taken up Pinker's challenge to begin constructing a plausible narrative about the origin and evolution of language based on natural selection but without any a priori commitment to preserve a separate language module with an associated "instinct" (admittedly chosen by a Pinker (1994) somewhat mindful of its unhelpful and, according to Deacon, "dated" connotations within biology). The two men thus believe in pursuing the same overall project ("explaining language by natural selection"), many of the same data sources (neuroscience, linguistics, evolutionary biology) but with different starting points and different prime orientations (innatist and adaptationist/co-evolutionary, respectively). Following the summarizing chart, I will pick up on how the two researchers differentially sample the data from chimpanzee language research within the context of their different research projects.
 
Contrasting Nativist, Cognitivist and Co-evolutionary Orientations
Paradigm Nativist 
(Chomsky, Pinker)
Cognitivist 
(Bates, Rumelhart, etc.)
Co-evolutionary 
(Greenfield, Deacon)
Image of Language Separate, autonomous formal. Part of general cognition Adaptive symbiote of cognition
Origin of Language Language Instinct. Scaled up associative learning? Co-articulation of lang. and brain
Architecture Binary, a priori symbolic, global rules PDP,  graded, experience over time as prime Icon, index, symbol as primes.
Project Introspection converges with observation Neural net models human learning and behavior Diachronic account consistent with natural selection.
Paradigm Case Novel utterances. Output as function of input (congruent with behaviorism.) Evolutionary adaptation.
Foregrounded Design Principle Generativity, productivity. "There is no longest sentence." -N. Chomsky (1957). Language and the world contain enough structure for the organism to discover the critical factors 
in time. Speech formulas and "scripts" foregrounded.
Language is a multi-tiered process where each type of reference "tokenizes" into the next highest 
level: (icon, index, symbol).
Process of Discovery 1.Examine parameters of given language to discern 'minimal' representation of complexity. 
2.Relate to 'principles" of UG 
3.Confirm formally derived model in biology and cognitive neuroscience
PDP models are instantiated as neural nets without a prior bias (in one version), exposed to 
massive input; ”learnability" is problematized by comparison with human behavior 
("comparative psychology")
Observation of diachronic phylogenetic evolutionary adaptation sets absolute horizon for formal model accounts. Language NOT assumed a priori to operate as rationalist computer. Neural 
substrate co-articulates w/ language form/ structure. No a priori commitment to a model of "language".
 
III. The Chimpanzee Language Debate
It should perhaps be no surprise that both of these works weigh in on the now decades long controversy concerning whether chimpanzees are capable of learning human language. Needless to say, nearly every term in that embedded question is contested, from which kind of chimpanzees are best to use to what actually constitutes "learning" and "language". In this section I will outline some of the key literature and then show what Deacon and Pinker make of it.

Leaving aside some of the earliest work, Gardner and Gardner (1974) brought the issue of whether chimps can learn language to national attention in their work with Lana who allegedly was able to manipulate lexigrams to accomplish symbolic communication. Terrace (1979a, 1979b) started out very excited at the prospect of interspecies communication but became a skeptic after an unsuccessful attempt to teach a chimp named Nim Chimsky what could reasonably be called language (defined largely in terms of the categories of Chomskyan syntax). Petito & Seidenberg (1979) delivered a careful summary of this phase of the work which criticized the methodology and assumptions of researchers and didn't find much of lasting theoretical significance.

Greenfield & Savage-Rumbaugh (1991) continued the pro-chimp language research charge with new data and subjects, though not subject to peer review. Wallman (1992) can be summed up by saying: "It's all a big hoax, i.e., chimp language is neither homolog nor analog of human language!" Savage-Rumbaugh & Lewin's (1994) book on Kanzi brought the world's attention to this Bonobo chimp with an apparently remarkable aptitude for signing in human-like fashion. This book was also not pre-reviewed carefully by experts and the over-ambitious claims of the Kanzi research are still debated (though not by an adoring public who continue to make a sympathetic NOVA special a favorite repeat).

Pinker (1994) in classic iconoclastic style presents an uncompromisingly hostile posture to the value of this line of research. His argument (pp. 333-342) essentially boils down to the following key points:

Chimp research is seriously and universally flawed by lack of interrater reliability and basic design lapses.

Most proponents are animated by a misguided "animal rights" agenda which seeks to break down the artificial barrier between "animal" and "human" rather than figuring out what is going on.

Even if true, what chimps do has little inherent bearing on what people do (the 2 percent difference in DNA between chimps and humans means an unpredictably large variation in expression because DNA is a binary code).

There is no evidence that chimps can learn syntax which (for an innatist) is the defining characteristic of human language.

Deacon (1997: 92), along with virtually everyone else in the field, shares many of Pinker's critiques of the earlier research, while taking a more generous tone concerning the "third wave" embodied in the Kanzi work: "...more recent studies by these same experimenters (Rumbaugh & Rumbaugh) with a bonobo named Kanzi have demonstrated far more effortless and sophisticated symbolic abilities." He takes the case of Kanzi, who learned sign from very young as most small children do as an example which "...throws a monkey wrench (sic) into the whole critical period argument as it applies to language acquisition in chimps (p. 126). If, as Deacon surmises, Kanzi's unique accomplishments are largely a function of simple immaturity at first exposure with a concomitant lack of processing bandwidth ("less is more"), then her acquisition of "symbol" as opposed to "syntax") and her status as a chimp presumably devoid of the "language instinct" constitute a refutation of the construct of the "critical window". By extension the entire innatist agenda is suspect.

From Deacon's point of view, Pinker's vehement denial that there might be something like language going on in the case of Kanzi is a position motivated more by ideology than based on evidence. We can not so easily dismiss Deacon as an ideologue on this matter; he has a substantial reputation in comparative primate neurology. Indeed, Pinker himself quotes Deacon as an authority in this area (Pinker, 1994: 350, 443)). At the very least, Deacon's careful review of this entire area of research which is easily subject to, in many cases, deserved ridicule, ("Susan, come out of your cage, your 2:00 appointment is here!") will probably serve as a better guide to future scholars than Pinker's unabashedly partisan diatribe. If one admits a Vygotskyan (Vygotsky, 1978) "zone of proximal development" as a legitimate experimental space for animal research, than some limited conclusions about behavior evidencing symbolization in animals coached by humans may be recoverable from the attempts to teach chimps human language.

An ultimately more promising approach, authentically grounded in comparative ethology and psychology, would draw upon the careful investigations of "emic" systems of animal communication in the wild currently taking place. We may not be as tempted to personify ants, prairie dogs or elephants (Disney notwithstanding), so the ability to draw genuinely useful comparisons may be enhanced. When such creatures' behavior is taken as the starting point, rather than that of cage-bound chimps pushing buttons to get a bunch of M&Ms, the conclusions of innatists and adaptationists alike are likely to be quite different from those they have drawn based on three decades of such monkey business.

Pinker on Creoles: Bad Faith or Bad Homework?
Someone like Pinker is easy to pick apart because he so generously doles out scathing critiques of everyone he doesn't agree with. He does a total hatchet job on Whorf (which deserves a paper to itself to rebut properly) in which he draws on secondary sources, misquotes Whorf and makes several factual errors (e.g., p.60, the example with the canoe is actually from Nootka, a language of the Northwest, not Apache as stated; moreover, the Apaches live in the desert and don't have canoes!).

He misses the boat completely on the creole data (p. 32-39 especially highlighted here) as well and demonstrates the dangers of reading outside of one's field in order to disseminate "facts about language" in a popular book. This is particularly egregious in one who is so painstakingly deconstructing the "received wisdom" about language and culture ("The Standard Social Science Model") since he ends up propagating a widely discredited theory without ever really "looking under the hood." The net result is an increase in confusion about these issues and more "soundbyte linguistics" (ala Eskimo snow terms) of the type he so shrilly decries. This book has made Bickerton's theory virtually the only thing that the educated populace "knows" about creoles. This is Pinker's fault since virtually all of the opposing information cited here would have been available through a cursory review of the literature available to him at the time he wrote his book.

Let's start with Pinker's (p.32-35) accounting of Bickerton's (1990) version of his theory. First he accepts that we can "see how people create a complex language from scratch." This is an appealing notion, but far from the more nuanced realities of creole genesis (McWhorter, 1997). What follows is basically Pinker's recitation of Bickerton's now discredited (within creolistics) Language Bio-Program Hypothesis ("LBH") based on the evidence of grammaticalization that characterized the transition from "pidgin" to "creole" as the first generation of Hawaiian immigrants acquired the vernacular as their native tongue. Bickerton is a staunch Chomskyan, and the LBH has had wide-ranging impact as a signal example of the empirical verifiability of the generative paradigm.

In fact,

... the theoretical foundations of the LBH depend on exactly one keystone: Bickerton's interpretation of data he collected on Hawaiian Creole English (henceforth HCE) (1976,1977,1981:6-17, Bickerton & Wilson 1987). It is the discontinuity between what B analyzed as Hawaiian Pidgin English and HCE, and only this discontinuity which led B to propose creoles are formulated by children from sharply reduced pidgin input. If this presumed discontinuity is shown to be a misinterpretation, then there is no reason to propose that today's creoles are descendants of radically distinct pidgins and thus no reason to attribute the final form of creole languages exclusively to children (McWhorter 1997: 57)
Bickerton himself admitted that "failure to find support for it would be fatal to the hypothesis." (1984a:174). The discontinuity between the speech of B's elderly informants and modern HCE McWhorter shows that, based on the work of four separate scholars, what B sees as the pidgin ancestor of HCE was actually just second language HCE. "This would mean that the discontinuity between this speech and that of his younger informants is thus of little application in genesis issues." (McWhorter, 1997:59) In addition there is ample evidence that there was a widely stabilized English pidgin developing alongside the Hawaiian pidgin before Bickerton's interviewees arrived, which is "fatal" for his hypothesis.

McWhorter (1997) also demolishes the LBH's reliance on serial verbs in Surinamese Creole grammar. Rather than emerging as an autonomous UG-inspired attempt to make up for the languages' lack of a P category as Bickerton infers, all instances in Saramaccan must be direct transfers from the African substrate, once one considers the type of serialization involved. Indeed, much of what Bickerton wants to attribute to UG and the LBH turns out, under closer analysis, to be entirely accountable for in terms of what is known about the substrate languages, internal developments and the demographics of early Surinamese plantation life.

Proponents of the LBH need not despair entirely. McWhorter does manage to salvage some substantial pieces that hold up under the closest scrutiny and which continue to be verified by independent research: the importance of perceptual salience in superstrate transfer, TMA markers ("a major contribution to linguistic theory" (p. 81)), and a suggestion that "creoles can reveal to us the beginnings of an ontological cline of structural accretion over time." However, "the data ... lead us to the conclusion that if children are indeed capable of transforming fragmentary jargon into fluent creole, then pidgins and creoles to date have not offered us a demonstration, and are not likely to in the future " (p. 82).

McWhorter, an acknowledged leader in the field of creolistics, does not imply here that there is nothing to be learned by students of psycholinguistics studying creoles and pidgins. His work accepts many descriptive facts that the UG/LBH approach appropriates for its program as well, but he is open to plausible alternative explanations which don't involve an a priori hard-wired language instinct. His view will square much more closely with the co-evolutionary perspective of Deacon as you'll see in a moment:

There is nothing in the present findings which rules out a more functional view of language acquisition which sees language acquisition as the construction of more general cognitive as the result of a maturational process which depends upon the inseparable co-action of human genes and environment (Cziko 1989:29, as quoted in McWhorter, 1997:75)
Deacon on Creoles
Deacon touches on this topic only briefly, but his characteristic epistemological humility gives him an advantage over Pinker in critically interpreting this data:
Not only does it appear that languages have evolved to take advantage of the learning biases of children, but to whatever extent prior language input is impoverished, it appears these learning biases tend to reshape language to fit. Universal (i.e., convergent) trends emerge either way in the absence of a preexisting plan, whether in the input or in the mind (p. 140).
Deacon, while not yet up to speed on the actual applicability of creolistics to child language acquisition, reveals here a stance which embodies a radical critique of the UG/LBH approach echoed by many people in many fields. McGroarty (1998), for example, in examining robust constructivist tendencies within SLA research inspired by the inapplicability of formal models to practical problems, lauds the careful descriptive work of the generativists, but chides them for elevating their structural descriptions "from explanans to explanandum" (p. 153).
IV. Conclusion
In a pre-paradigmatic field, indeed in all fields, there is plenty of room for true believers, who take strong stands that crystallize key issues (while sometimes playing a bit fast and loose with the "facts") and more cautious and open-minded thinkers; it is for the critical reader to take the time and trouble to discern the actual contributions of each. In this slice of the debate on the evolution of language, I have attempted to establish a dialogue between what appear on the surface to be two current and competing narratives about how language might have evolved. In the course of this project, I have become very aware that much of the debate is as motivated by a rhetorical need for differentiation as much as by substantive intellectual difference.

At the end of the day, both Pinker and Deacon agree that any account of language evolution must square with ontogeny, the fossil record, cognitive neuroscience and the small, but robust set of features that in the (generative and non-generative) consensus deserve the title of "Universal Grammar". Both ascribe to Darwinist ideas about natural selection as naturally as they breathe and shave each morning (judging by their press photos).

Many of the apparent differences between them can be understood as differences in intended audience (popular versus more scholarly), in disciplinary training (cognitive psychology and psycholinguistics vs. biological anthropology), in rhetorical style (bold, polemic, certain vs. lyrical, erudite, expansive) and in emphasis (language as a formal system mapped into neural substrate and co-evolution of language and brain). By considering how the work of these two men makes differential use of two key areas of research data, I have worked through for myself many of the important paradigmatic and methodological issues of this highly contested area.

This will serve as important groundwork for my own continuing original research on animacy distinctions and the evolutionary and neurological basis for their role as catalyst in grammaticalization in creole languages (and universally). As far as I can tell, this approach is consistent with Pinker's delineation of a kind of "naive physics" (p. 420) as well as with Deacon's understanding of "cognitive bias" and his acceptance of specific regions in the brain to deal with "people" and "animals". The question of how animacy, as a fundamentally semantic category can serve as the progentor for the articulation of syntactic structure may not be so easily squared with both models. That question will, with so many others, alas, have to be left for another day.


References

Bickerton, D. (1990) Language and species. Chicago; U. of Chicago.

--, (1988) Creole languages and the bioprogram. In Linguistic theory: extensions and implications, ed. by Frederick J. Newmeyer, 268-282. Cambridge: Cambridge University Press.

--, (1984a) The Language bioprogram hypothesis. The Behavioral and Brain Sciences, 7:173-88.

--, (1984b) The language bioprogram hypothesis and second language acquisition. Universals of second language acquisition, ed. By William Rutherford, 141-161. New York: Academic Press

Cziko, Gary. (1986) Testing the language bioprogram hypothesis: a review of children's acquisition of articles. Language, 62:878-898.

Deacon, T. (1997). The symbolic species: The co-evolution of language and the brain. New York: Norton.

Dreyfus, B. & Rabinow, P. (1983) Michel Foucault: Beyond structuralism and hermeneutics. Chicago: U. of Chicago.

Gardner, B.T. & R.A. Gardner (1969 ) Teaching sign language to a chimpanzee. Science, 165: 664-672.

Givón, T. (1998) Toward a neurology of grammar. The Behavioral and Brain Sciences, 21:1, 154-155.

Greenfield, P.M. (1991) Language, tools and the brain: The ontogeny and phylogeny of hierarchically organized sequential behavior. The Behavioral and Brain Sciences, 14:531-595.

--, (1998) Language, tools and brain revisited. The Behavioral and Brain Sciences, 21:1, 159-162.

Greenfield, P.M. and Savage-Rumbaugh, E.S. 1991. Imitation, grammatical development and the initiation of proto-grammar by an ape. In Krasnegor, N.A. ed. (1991) Biological and behavioral determinants of language development. Hillsdale, NJ: Erlbaum.

McGroarty, S. (1998) Constructive and constructivist challenges in applied linguistics. Language Learning, 14: 151-168.

McWhorter, J. (1997) Toward a new theory of creole genesis. New York: Peter Lang.

Pinker, S. (1994). The language instinct. New York: HarperCollins.

Savage-Rumbaugh, E.S. & Roger Lewin. (1994) Kanzi: The ape at the brink of the human mind. New York: John Wiley.

Swann, P. (1998) Greenfield on language, tools, and brain. Behavioral and Brain Sciences, 21:1, 155-159.

Terrace, H.S. (1979). Nim. New York: Knopf.

Vygotsky, L. S. (1978) Mind in Society. Cambridge, MA: Harvard University.

Wallman, J. (1992) Aping language. New York: Cambridge University.