James Hurford, “The Origins of Meaning”.

Oxford University Press, 2007

Reviewed by Derek Bickerton

 

     James Hurford has been working on the evolution of language for the last two decades.  His first paper on this topic appeared in the 1980s (Hurford 1989), antedating the work of Pinker, myself, and just about anyone else currently active in the field.  He has written numerous stimulating and insightful, if often controversial, papers and has co-edited two collections of conference proceedings (Hurford et al. 1998, Knight et al. 2000).  However, until now he has not attempted any full-length treatment of the subject, so the current volume has aroused considerable interest and high expectations.

 

     This volume is the first of two on language evolution; the second, on formal and structural aspects of language, will be published at a later date.   The present volume devotes itself entirely to the question of how the formal units of language came into existence, and how obstacles to using such units communicatively were overcome.  The book is divided into two parts, “Meaning before communication” and “Communication: what and why?”  These parts, each of which contains five chapters, are of almost identical length, accurately reflecting a balance between the cognitive developments that led up to language and the ways in which, and purposes for which, that cognition was expressed in interactive ways.

 

Chapter 1 performs the ground-breaking operation, vital in any interdisciplinary inquiry, of removing ambiguities from expressions that may have different meanings in different disciplines, or between lay and expert users: “reference”, “intention”, and of course “concept” are just some of these.  In the second chapter, Hurford picks up the notion of “concept”, sticking closely (as he does almost throughout) to the literature on animal experimentation, to discuss properties that may form necessary pre-requisites for a semantics adequate to support language, such as transitive inference, object permanence, animate-inanimate distinctions and cross-modal links.  While he overtly rejects any kind of cognitive scala naturae leading ineluctably to humans, it is not easy to see how such a ladder would differ greatly from his stated belief that “our animal ancestors approached modern human cognition to various degrees” (20).

 

In Chapter 3, he deals with the evolution of episodic memory, essential in humans first for the organization of experience and second for retailing those experiences with the help of a syntax based upon argument structure.   He analyzes behavior that suggests the presence of some form of episodic memory in animals, from scrub-jays caching food to rats interrupted in searching mazes, and appeals to Kant’s philosophical notions to support a clear distinction between semantic and episodic memory, likening their content to, respectively, Kant’s “analytic” and “synthetic” propositions.    He sees the ability to distinguish between the two forms of memory as being shared by animals as disparate as rats, jays and chimpanzees, and as an essential pre-adaptation for language.

 

Chapter 4 returns to the theme of an earlier article (Hurford 2003) in which Hurford hypothesized that the dorsal and ventral pathways in the brain are what underlie the most basic distinction in formal logic, that between predicates and arguments—the dorsal pathway (the “where” channel) merely predicating the existence and location of something, the ventral pathway (the “what” channel) analyzing sensory data more finely to determine what that something is.  Since these pathways are fully operative in many animals, he concludes that the latter are able to process predicate-argument structures. Since he regards the logical predicate/argument distinction as prior to, and in some sense underlying, the most basic distinction in linguistics (that between subject and predicate), he sees a capacity to make the logical distinction as a pre-adaptation for basic syntax.  To put it more simply, animals can, potentially at least, form mental propositions

 

In the final and fifth chapter of Part 1, and the only one that does not draw extensively on the psychological literature, Hurford seeks to “capture the private mental representations of animals at an evolutionary stage just prior to the emergence of public language” (125)—in other words, to show how the mental propositions of Chapter 4 might be represented in animal brains.  Since propositions normally involve ascribing some behavior (a predicate, an action or event) to some entity (a subject, an entity of some kind), he then faces the difficult task of determining how, in terms of mental operations, animals might have gone about making distinctions that are not obvious from the stream of raw sensory data that constitutes the input to cognitive processes—the “blooming, buzzing confusion” that William James described.  In analyzing a scene, the “global” and “local” distinctions derived from dorsal and ventral pathways here furnish the basis of a novel formalism for expressing what, according to Hurford, animals can see and know but cannot yet express.

 

Although he never uses the expression “language of thought”, the notion of something similar to this seems implicit in much of what Hurford writes in this and other chapters.  He cites Fodor approvingly, and concludes the first half of the book with the following advice to seekers after the pre-requisites for language: “[S]tart your thinking…by imagining yourself in a non-linguistic state, but otherwise cognitively pretty much as you are” (164, emphasis added).  Summarizing that first half in a final sentence, he says, “We’ve seen what [animals] know; in Part II…we’ll examine why they don’t tell.”

 

The second part of the book kicks off with a chapter (#6) on the origins of communicative behavior.  Hurford notes that “communicative acts evolved out of non-communicative acts” (169) and that common to all of them should be some benefit to the actor.  He sees early communication as merely dyadic, involving a sender and a receiver but no third party, no referent that is meant to focus the attention of both, and hypothesizes that more complex forms of communication developed out of, rather than alongside, this more primitive form.  The remainder of the chapter examines dyadic communication in other species, the factors that promote it, and the extent (a limited one, he admits) to which one could say that signals used by animals were learned rather than innately specified,

 

In Chapter 7, Hurford turns to an examination of how objective reference might have entered animal communication systems.  He discusses the mixed evidence with regard to what is perhaps the most basic referential act—pointing—in animals, and continues with a discussion of types of animals signal (food calls and alarm calls) that contain what has been termed “functional reference”.  He concludes that at least some signals, for example the alarm calls of vervets (who must be tired by now of appearing in so many language-evolution books) not only refer but denote, and like words have a particular class as their extension.

 

Chapter 8 focuses on motivation.  To communicate information to others seems to run counter to the self-interest of the informant, since it is the receiver not the sender who principally benefits from such acts; for this reason, language constitutes a long-standing puzzle for evolutionary biologists,  Hurford surveys some of the major attempts at solving that puzzle, including theories of kin selection, inclusive fitness and reciprocal altruism, as well as theories that try to relate language to conspicuous display such as Zahavi’s handicap principle and Miller’s sexual selection scenario.  Since none of these attempts seems to solve all the motivational problems, he then gives qualified approval to the notion that here the group rather than the individual may be the unit of selection.  Groups where members communicate more with one another collect a greater store of useful information than groups where they do so to a lesser extent; the former should therefore out-compete the latter under most circumstances.

 

However, such communicative groups must first establish high levels of cooperativeness and mutual trust; words are cheap and cheaters have much to gain.

 

Chapter 9 suggests how such levels might be obtained; one pre-requisite is some sort of theory of mind, an ability to perceive the intentions of others, and to know that these are not necessarily the same as one’s own.  Co-operation depends on shared intentions, and Hurford endorses the claim of a recent article (Tomasello et al. 2005) that shared intentionality necessarily preceded language.  As for mutual trust, he notes some experiments with the neuro-peptide oxytocin, raised levels of which may have provided the level of trust necessary to offset the cheapness of words.  The tenth and final chapter, a mere three pages, briefly summarizes his view that the cognitive and communicative capacities of great apes had already reached a stage of imminent language-readiness, and concludes with a capsule summary of the companion volume that will detail the “cascade of consequences” (333) following the “breakthrough” into language.

 

As the foregoing suggests, Hurford brings to his task an almost encyclopedic knowledge of the literature in a half-dozen fields.  This knowledge is displayed in a superbly-organized argument into which hundreds of disparate pieces of information are neatly slotted, giving rise to a smoothly-flowing and cohesive whole.  Only rarely does his erudition lead him into superfluous detours: his occasional nitpicking to satisfy philosophers, or the discussion of “simultanagnosia” on pages 107-109.  For the most part, the reader is swept along effortlessly, despite the inevitable density and complexity of much of the material.  For a work aimed both at experts and educated laypersons, his tone is pitch-perfect; he never condescends to nor patronizes the reader, nor does he oversimplify complex material, yet he makes difficult ideas easy to digest and even, quite often, entertaining.  He balances opposing arguments fairly, yet seldom equivocates or fails to indicate where his own sympathies lie.  For once, book-jacket encomia—“this major intellectual endeavor,” “the major publication dealing with language evolution to date”—do not seem over-strained.

 

I sincerely wish it were possible to close my review on this note.  I genuinely admire what Hurford has done here, and doubt whether anyone else in the field could have synthesized so much information in so palatable and user-friendly a manner.  Yet I would be remiss in my duty if I stopped there.  Precisely because the book is so good, it succeeds in concentrating and focusing as never before an approach to language evolution that is very widely shared--a view some aspects of which I myself have shared--but one that over the last few years I have come to regard as an obstacle rather than an aid in the search for the evolutionary origins of language.  What follows should therefore be regarded not so much as criticism of Hurford and his book as criticism of the tradition that it represents.  Indeed, I am in Hurford’s debt, since the clarity and thoroughness of his work has helped to clarify my own thinking on the vitally important issues the book raises.

 

Hurford mentions two or three times—unlike many writers in the field, he doesn’t hide things—that the fact that humans alone, but no other species, not even those closest to us, have acquired language poses a serious problem for explanations of how language evolved.  He frankly admits he has no solution to this problem, but since nobody has ever even suggested one (and most people don’t seem to recognize the problem) he then drops the subject.  At the same time, although he makes occasional disclaimers (“We will also see the still large gap that exists between us and other species,” p. 2), his statement (cited above) that the primates who pioneered language were “cognitively pretty much as you are”, plus his view that “animal concepts provide a foundation for the meaning of words” (19), suggest that he regards the cognitive capacities of other great apes as little inferior to those of humans.  The only thing he doesn’t realize is the strange place that the conjunction of these two lines of thinking leads to.

 

In his Ulyssean voyage across the tangled landscapes and manifold perils of the field, Hurford must, sooner or later, meet with a Scylla and a Charybdis:

     Scylla: the lower you rate the cognitive capacities of apes, the harder it

                 becomes to explain how we got language.

     Charybdis:  the higher you rate the cognitive capacities of apes, the

                 harder it becomes to explain why they didn’t get language.

It was possible, with the original Scylla and Charybdis, to somehow navigate between them.  With this pair, it isn’t.  Evade one and you are trapped by the other.  Maybe, instead of trying to go between them, you must seek out some entirely different path.

 

Hurford is very much aware of Scylla, indeed he describes it quite explicitly: “If concepts only appear with language, the origin of language itself is that much more of a mystery” (10).  So let’s look a little more closely at concepts, but first let’s put them in a broader framework.

 

Hurford’s book appeared, unfortunately, too late to include reference to a seminal but sure-to-be-controversial article (Penn, Holyoak & Povinelli, in press) that takes a diametrically opposite view on the cognitive capacities of apes.  Citing many of the same sources Hurford cites, the authors re-analyze the findings of these, and find a massive discontinuity between ape and human forms of cognition; in particular, apes and other non-human species lack the ability to perceive higher-order relationships such as transitive inference and hierarchical structuring.  But equally compelling, and quite uncontroversial, is the point from which their article starts: “Human animals—and no other—build fires and wheels, diagnose each others’ illnesses, communicate using symbols, navigate with maps, risk their lives for ideals, collaborate with each other, explain their world in terms of hypothetical causes, punish strangers for breaking rules, imagine impossible scenarios, and teach each other how to do all of the above” (p. 1).  This list is a random selection from one that could be multiplied a millionfold.  True, many of the things in that longer list require language, but many do not; representational drawing and tap-dancing are examples, as is the invention of nooses, deadfalls and spiked pits for catching the colobus monkeys chimpanzees are so fond of snacking on.  If apes are “cognitively pretty much as [we] are”, why don’t they do ANY of these things?

 

One possibility is that we have concepts but they don’t.  Since there are so many definitions of concept around, let me make clear what I mean: on the analogy of “news you can use”, I mean mental representations, formed from neurons and their connections, that bear some kind of referential relationship (we needn’t be too specific here about exactly what) to entities and events in the real world, but (and this is the really important point) can be concatenated with one another and purposely manipulated to produce novel structures of thought.  What animals have in their place is “news you can’t use”—categories, rather than concepts, that come together in animal minds only when triggered by some external circumstance; otherwise, they will only appear sporadically in random brain activity, and cannot be joined or used voluntarily for computation in the absence of stimuli.  This distinction would have a purely physical basis. 

 

Animal categories would consist of separate loci, in the visual, auditory etc, areas of the brain, where, for example, the sounds, smells and appearances that characterize, say, a leopard would be separately stored, but there would be no locus in the brain where all these representations came together—where they linked to a single, all-embracing representation of “leopard”.  Human concepts would be based on a similar set of distributed representations, but these would on addition be linked with just such a central locus, which could be voluntarily activated without necessarily evoking all or any of the partial, particular representations, and would be linked with similar central loci.  Such loci would often, but not necessarily always, be tightly linked to representations of their corresponding words.  This is NOT a claim that we can’t think about something if we don’t have a word for it; words were probably the means by which the first of such conceptual loci were set up, but we obviously can think about things without using words and about things we have no words for.  The point is that, however they were originally formed, we would have permanently accessible instantiations of abstract class concepts while non-humans would have scattered representations of physical properties that could be brought together only by external stimuli.  Note that the empirical status of this category/concept distinction can in principle (and perhaps already in practice) be tested by means of modern brain-imaging techniques. 

 

Hurford, perhaps wisely in his case, hedges his bets: “My choice of the terms ‘category’ and ‘concept’ allows the possibility (indeed the probability) that these mental entities are fuzzy” (17).  However, the distinction made above is not fuzzy, and the distinctions between the things apes spontaneously do and the things humans spontaneously do (in terms of the Penn et al. list, above) is not fuzzy either.  The idea that these two discontinuities (sharp ones, no matter how hard we try to blunt them) are linked—indeed, that the behavioral discontinuity is a direct result of the representational discontinuity--deserves, at the very least, closer scrutiny.

 

This distinction between being able to perform mental computations only when external stimuli are present and being able to perform them voluntarily in the absence of such stimuli—what I loosely described as the distinction between “on-line” and “off-line” thinking (Bickerton 1995)—becomes crucial in Chapter 3, where the outputs of dorsal and ventral streams are characterized as combining to produce proto-propositions.  From Hurford’s viewpoint, if an animal can compute, from sensory input, “X exists” and “X is a Y”, then it is ready to acquire linguistic labels for X and Y and (once it overcomes the purely communicative problems set forth in Part II) deploy them in some form of language.  (Just exactly what kind of language, paleo- or proto- or the real McCoy, Hurford’s second volume will presumably clarify.)

 

But we’re talking apples and oranges here.  Given the distinction between categories and concepts described above, the fact that an animal reacts to certain types of external stimuli in ways that can be compared with the components of propositions in no way guarantees, or even implies, that those proto-propositions then take on a mental life of their own, become subject to voluntary control, and (once words are available) can without more ado be brought forth in linguistic dress.  Hurford is so convinced that the “first propositions, then language” position is correct that he never seriously considers the “first language, then propositions” possibility.   He mentions niche construction theory, but only in passing.  He should perhaps consider some of its implications; if, as the theory suggests, a major driving force in evolution is a feedback process between what animals do (at first with very limited genetic equipment for it) and the genetic variations that their activities select, the second possibility (or at least some kind of beneficent spiral) begins to look the more plausible of the two.

 

Wherever animal cognition is concerned, Hurford always gives animals the benefit of any doubt there is.  Nowhere is this more evident than in his discussion of vervet alarm calls.  Maybe it’s just his wording, but to me the following passage seems almost to suggest that global concepts are epistemologically prior to distributed, sensory-based representations:”[W]e can expect the vervet monkey to have, not only leopardvervet but also leopard-motionvervet and leopard-smellvervet.  None of these prelinguistic predicates is necessarily simply translatable into the words of any human language, but leopardvervet is pretty close to an English speaker’s leopard” (97, emphasis added).  Little evidence is offered for the latter claim.  Hurford grants that English speakers can think of leopards in their absence, but extends a similar privilege to vervets: “[W]e have seen that rats probably dream of mazes, so might not a vervet dream of a leopard?” (ibid.).  Possibly, but so what?  Dreams are not subject to the dreamer’s control, and even with regard to our own dreams, we don’t know whether these trigger core concepts or merely the distributed representations on which categories are based, so how could we know what semantic level is being tapped in animal dreams?

 

One thing vervets (and other non-humans) cannot do is communicate any information about past, future and physically non-present leopards.  It seems likely that this inability merely reflects a similar inability to think about leopards in their absence, and that both deficiencies are functions of the lack of any of the class concepts that underlie human thought and language.

 

But to Hurford, monkeys’ alarm calls are as close to words as monkeys’ internal representations are to human concepts, and bear the same (or at least  a comparable) relation to one another as human words and concepts do.  He treats an experimental finding (that Diana monkeys show less alarm on hearing an eagle’s shriek five minutes after an eagle warning than on hearing a leopard’s cough five minutes after an eagle warning) as evidence that the monkeys have maintained a focal concept of “eagle” in mind and therefore are less startled when the same threat is repeated than when a new one appears.  He does not consider simpler explanations; for instance, after an alarm call monkeys will continue for some time to employ the appropriate strategy (“Scan the sky, make sure you’re near cover” after an eagle call, “Scan at eye-level, get near a tall tree” after a leopard call).  Having to suddenly switch strategies, or becoming unsure which strategy to pursue, is far more alarming than simply having to focus the original strategy.

 

The problems with alarm calls are, however, merely symptoms of a much broader problem, which involves the relative properties of animal signals and language.  For Hurford, language and animal signals fall under the same broad umbrella: both involve “the vital illocutionary doing-things-to-each-other property that is common to primate calls and all human utterances” (184).  Well, yes, this is a property they do share, but aren’t there some equally important ones that they don’t?   As mentioned above, Hurford grants that communicative acts arise out of non-communicative acts, which is certainly not true of any of the units that compose language.  He does not pursue this issue far enough to ask what these original non-communicative acts (or the present communicative ones, for that matter) were for.  What they were for, why they were there at all, of course, was maximization of animals’ (inclusive) fitness; postures that indicate subservience, for example, refined by natural selection into stereotyped signals, save animals from being trapped into confrontations where their procreative potential might be damaged or altogether destroyed.  Obviously such signals won’t work unless they communicate to the receiver that this dog won’t fight.  But communicating information is not their primary function.  Maximizing fitness comes first; communicating information is incidental to that, and the fitness enhanced is the sender’s, not the receiver’s.  However, while language as a whole may be adaptive, there is no sense in which the units that compose it (words, phrases etc.) in and of themselves contribute in any way to the sender’s fitness.  Indeed, the reverse is often closer to the truth: if I give you information via language, that information may increase your fitness, but it’s hard to see how (bar a pious hope for reciprocity) it could increase mine, and if it gives you an advantage over me it could actually diminish my fitness.

 

By what metric does one decide that the “doing-things-to-each-other” property outweighs the properties of animal signals that distinguish them from language?  I don’t know, but inevitably one suspects that the property in question has been chosen simply because it is the only one that yokes language and animal signals together, and consequently the only one that accords with the belief that the one is the true ancestor of the other.

 

The belief that cognitive and communicative capacities of great apes brought them to the brink of language inevitably suggests that such capacities must have developed gradually and progressively across the phyla, starting from humble beginnings among simple organisms and reaching some kind of climax among bonobos and chimpanzees—in Hurford’s own words (cited above), that “our animal ancestors approached modern human cognition to various degrees”.  In fact, communication systems of other animals show no sign of approaching anything, least of all richer and more complex cognitive and communicative systems.  Such local peaks as there are in the fitness landscape do not increase in height or number as we get nearer to humans.  “Functionally referential” alarm calls, supposedly as close as other animals get to words, are found among monkeys and even chickens, but Hurford cannot adduce a single example from our closest relatives.  Number of signals per system falls into the lower range of double digits, regardless of whether primates or fish are involved.  There are no clear cases of calls concatenating or modifying one another, and none of the tiny and highly controversial handful of cases that have been claimed involves any of the great apes.  The state of affairs in communication is mirrored by other advanced capacities; Caledonian crows show at least as much dexterity in making tools as any ape, and the episodic memory of scrub jays at least equals that of any mammal.  Far from a steady march towards heightened powers, what one sees looks more as if someone had shaken up a jarful of varied abilities and then scattered them randomly across the phyla. 

 

But what might look random from a progressive perspective is anything but random when seen from the viewpoint of evolutionary biology.  Here there is no “approaching” of anything; species get the particular cognitive and communicative skills they need in order to exploit the particular niches that they have chosen or constructed, and once they have those skills, Nature, wholly uninterested in perfecting the type, simply stops selecting. Hurford gives a favorable mention to niche construction theory, and I kept hoping he was going to draw the conclusion I drew from it: that the key to the origin of language must be sought somewhere among the niches constructed by human ancestors between the date of the last common ancestor of humans and other apes and the present—niches very different from any occupied by other apes.  But he did not.

 

He did not, for a very interesting reason.  I spoke earlier of how well Hurford covers evidence from a variety of fields.  But there is one startling omission—at least, one that would be startling, were it not found among so many other writers on language evolution.  He virtually ignores anything to do with human evolution.  Only two of the numerous species ancestral (or possibly ancestral) to humans, habilis and erectus, are even mentioned, and each species is mentioned only once in the entire volume.  Moreover, these mentions are not even Hurford’s, but merely brief and tangential references in passages from other authors that he cites!  He seems unaware that in ignoring the overall pattern of human evolution he is not only closing the most promising avenue to the origins of language but also falsifying the nature of the human ancestors who first developed language, for whom forest-dwelling, at least partially arboreal apes make very poor ecological models. 

 

In consequence, it is hardly surprising that when it comes to actually specifying what was the tipping point from call-system into language and what pressure drove our ancestors past that point, Hurford has nothing substantive to offer.  Granted, his case is that minor improvements in ape cognition and communication gradually accumulated until some progenitor of humans became “language-ready”, so that the actual transition to language was no big deal.  But any such claim sends its maker spinning into the vortex of Charybdis.  If these minor improvements accumulated in one ape species, why not in all, or at least one or two others?  Why are these improvements not continuing in modern apes, so that we can observe them in action?  Why is it that while we have thousands of complex languages with convoluted structures and tens or hundreds of thousands of words in each, they have communication systems resembling those of birds and fish?  Why, while we are making moon-landings and sonatas, cathedrals and concentration camps, are they still fishing for termites and cracking palm-nuts?

 

Nobody as intelligent as Hurford could fail to be aware that, without at least some account of the tipping-point, his book is seriously incomplete.  Indeed, in at least one place  he is commendably frank about this; to a reader who complained that “[Chapter 8] needs a more powerful conclusion that matches the great expectations created in the introductory pages” (305), he responds “I’m sorry, but we have to be honest…Humans, with the emergence of language, appear to have squeezed through an evolutionary loophole, and it is hard to specify exactly which combination of the factors discussed in this chapter provided that loophole” (ibid.).  The most he does is to suggest the emergence of shared intentionality, plus elevated levels of oxytocin—a chemical that appears to induce trust, in this case trust in the honesty of cheap words.  Such things, of course, raise more questions than they answer.  How did shared intentionality evolve?  What selected for it?  Why wasn’t it selected for in other species?  What caused oxytocin levels to rise in humans?  Why didn’t they rise in other species, or if they did, why didn’t they have the same results as they did in humans?  These aren’t solutions, they’re just additional problems.  Yet the penultimate paragraph of the book commences: “We have reached a watershed.  As soon as the breakthrough was made for animals to communicate their thoughts freely to others, a cascade of other innovations were (sic) selected…” (333).

 

Breakthrough?  What breakthrough?  Early last century there was a genre of serial magazine adventure stories for boys in which at the end of each episode the hero was caught in some trap from which there seemed no way of escape.  The reader bit his nails for a week, wondering by what incredible miracle of ingenuity the hero could survive for the next week’s adventure, only to find the new installment beginning with some such sentence as, “With one bound, Jack was free.”  Well, having made the breakthrough sound as impossible as any of Jack’s predicaments, Hurford transitions to his next volume in exactly the same way.  With one bound, Jim is free to continue the story of language in his second volume.

 

I hope you’ll read that volume, as I certainly will, and I hope that nothing I have said will prevent you from reading the present one.  Whether you agree with it or not, it’s a must-read for anyone who’s ever had the slightest interest in language evolution, or in how humans came to be human, if it comes to that.  Hurford is always a stimulating and thought-provoking writer, and never more so than here; proof of that stimulation can be found in this very review, where his book has forced me to sharpen and express my own ideas in ways I had not imagined before.  I doubt if anyone will produce a more competent and thoroughgoing defense of what Irene Pepperburg (2005) called the “primate-centric” position.  Those who would challenge that position, from whatever perspective, must be able as a minimum to answer all of Hurford’s arguments, or abandon their enterprise altogether.



References

 

Bickerton, D.  1995.  Language and human behavior.  Seattle: University

     of Washington Press.

Hurford, J.R.  1989.  Biological evolution of the Saussurean sign as a

     component of the language acquisition process.  Lingua 77.187-222.

________.  2003.   The neural basis of predicate-argument structure.

     Behavioral and Brain Sciences 28 (3).261-283.

________, J.R., Studdert-Kennedy, M. & Knight, C. (eds.)  1998. 

     Approaches to the evolution of language; social and cognitive bases.

     Cambridge: Cambridge University Press. 

Knight, C, Studdert-Kennedy, M. & Hurford, J.R. (eds.)  2000.  The

     evolutionary emergence of language. Cambridge: Cambridge University

     Press. 

Penn, D.C, Holyoak, K.J., & Povinelli, D.J.  (in press)  Darwin’s mistake:

     explaining the discontinuity between human and nonhuman minds.  To

     appear in Behavioral and Brain Sciences

Pepperberg, I.M.  2005.  An avian perspective on language evolution:

     implications of simultaneous development of vocal and physical object

     combinations by a Grey parrot (Pittacus erithacus),  In Yallerman, M.

     (ed.),  Language origins, 239-61.  Oxford: Oxford University Press.

Tomasello, M., Carpenter, M., Call, J., Brehne, T., & Moll, H.  2005,

     Understanding and shared intentions: the origins of cultural cognition.

     Behavioral and Brain Sciences 28.675-735.