Carstairs-McCarthy's (1999, 2000) novel proposal to account for the evolution of syntax contains too many flaws to win acceptance. The attempt to derive a presyntactic language from a primate call system ignores the most crucial distinctions between the two, while trivializing others. The comparison between syllabic and syntactic structure hardly works even as an analogy; it fails as an explanation because it fails to deal with the most basic properties of syntax, in particular recursion. The supposed problems that motivated the work large vocabulary size, duality of patterning, and the NP-S distinction--are not really problems at all.
(a) human language developed from animal call systems.
(b) hominid ancestors created a variety of novel calls.
(c) the larynx lowered, giving rise to the syllable.
(d) syllable structure imposed syntax on the call system, leading to a (perhaps simpler)form of modern human language.
I shall have nothing to say about (c), which may be relatively unproblematic. Each of the other three, however, poses some serious difficulties.
2. Carstairs-McCarthy's motivation derives from three things he regards as problems that are not problems at all. Why are human language vocabularies so large? (Because given a primary representation system, a secondary system, such as language, cannot be stopped from representing everything that is in the primary system, and even things like ghouls and goblins that aren't, see Bickerton 1990). Why is there duality of patterning? (Because there is a limit to the number of distinguishable speech sounds that can be made, so unless you went to a two-level system in which those sounds themselves had no meaning, you could never get enough words to represent everything in the primary representational system.) Why does language distinguish between sentences and NPs? (Because, given the social utility of verbs, it is absurd to imagine hominids capable of speech yet unable to say 'Give!', 'Come!', 'Stop!', 'Look out!', 'Duck!', or 'Run!', and because once there are verbs, and nouns to go with them, sentences or at least, imperative utterances containing both nouns and verbs, such as 'Give meat!' or 'Look out lion!'--can hardly be far behind.) Surely there are enough problems in language evolution without inventing new ones.
3. Carstairs-McCarthy takes an idiosyncratic approach to the continuist-discontinuist issue. He believes that exaptation of syllabic structure to create syntax constitutes a genuine discontinuity, yet holds that, prior to this, call systems had developed seamlessly into some form of protolanguage. He brushes aside the well-known fact that language and call systems are operated by different parts of the brain after all, the neural infrastructure of calls could have moved, in all those years, couldn't it?. He similarly brushes aside the profound differences between call systems and language, in part by failing to mention the most significant of these.
4. Those he lists (p. 211) are: closed versus open; affective versus referential; innate versus learned; iconic versus symbolic; and involuntary versus voluntary. But he seriously distorts the nature of these oppositions. For instance, the third contrast should be indexical, not iconic, versus symbolic. Further, contrary to what he suggests, the work of Cheney and Seyfarth (1990) does not blur the fourth contrast: the calls of vervets are entirely innate in both acoustic form and generalized content (e.g. 'something potentially dangerous in the air'). All that young vervets learn is to limit the extension of a call, e.g. to refrain from giving it for owls, falling leaves etc, and save it for birds capable of seizing vervets.
5. But Carstairs-McCarthy fails to mention two much sharper and clearer distinctions between call systems and language. The first involves predication, which at its simplest level consists of taking something and saying something about it: 'Dogs bark', 'John left', and so forth. Even imperatives are predications: 'Go!' is predicated of an unspoken but unambiguous 'you'. Yet while the vast majority of language utterances are predications, predication is impossible in any call system. The second involves displacement. Language utterances have meaning whether the things they refer to are physically or temporally present or not (for obvious reasons, most things we talk about are not present). Calls can have meaning only when referring to something in the immediate environment (calls used deceptively do not mean anything, except that the animal concerned has the smarts to dupe his or her group members). You cannot use a leopard call to mean that there was a leopard here yesterday, or that leopards are dangerous, or anything other than 'There's a leopard!'.
6. These two distinctions are not graded but absolute, and it should be mandatory for anyone who claims continuity between calls and any kind of (proto)language to show how and under what circumstances a call system could have acquired predication and displacement. Consider next the claim that hominids developed a variety of calls. Carstairs-McCarthy talks about a 'strategy to create more calls', 'a larger call vocabulary', 'the learning of more calls' (p. 131). But calls are neither created nor learned; they develop as part of the genetic inheritance of a species, and, as pointed out above, only the precise range of their extension is learnable. If these calls could be created and learned in quantities sufficient to stretch the memory capacity, as Carstairs-McCarthy suggests, then they were not calls at all, they were words.
7. Moreover, if these were truly calls, what would they have meant and what would they have been used for? The fact that no species repertoire contains more than about forty or fifty calls may reflect not a memory limitation but the fact that the calls any species needs or can make use of are strictly limited. At no point in Carstairs-McCarthy's book is there any discussion of what these additional calls might have meant, or how they might have been used, or why they would have been needed. Carstairs-McCarthy might suggest that such issues lie outside his terms of reference. I disagree. It is all too easy to spin abstract theoretical models of how language might have developed, that was what got the field into disrepute in the first place. No model worth taking seriously in the twenty-first century can afford to ignore what we know (knowledge that every year increases) about the behavior and ecology of our remote ancestors.
8. But Carstairs-McCarthy's lack of explicitness is understandable. If he had tried to describe in detail his intermediate stage between a primate call system and a (simple) language, he would have been forced to choose between two equally distasteful alternatives. One is that the calls of this immediately pre-linguistic system were like the calls he describes on pp. 21-22, potentially translatable by a phrase ('To the bushes!) or a sentence ('Take cover in the bushes'). If this were so, then at the syntactic stage these calls would have had to be decomposed and analyzed into a series of words ('take- cover', 'bushes', some kind of directional). How could this have been done, if the calls were as ambiguous as Carstairs-McCarthy himself suggests (he offers 'An eagle!', 'There's an eagle overhead!', and 'Run from the eagle!' as further alternative translations of the vervet martial- eagle warning)?. It would be hard enough to figure out which bit of the call meant 'take-cover' and which bit meant 'bushes', but when you have to ask yourself whether the bit that you thought meant 'take-cover' might not just as easily mean 'run', or even 'eagle', the task becomes impossible.
9. But the other alternative is equally unattractive. That is, that the calls were not really calls in the sense of vervet calls, after all they were just words in disguise. Such a choice, instead of explaining the origins of language, would merely push those origins back into the mists of time, leaving them totally mysterious, while undermining any claims of continuity with true call systems. Carstairs-McCarthy's lack of explicitness is such that one cannot be sure which alternative he intends. But his talk of 'creating' and 'learning' calls makes one lean towards the second. So does his suggestion that hypothetical calls 'tipu' and 'naka' could be joined to form 'tipunaka' with 'a meaning derived from or related to those of [tipu] and [naka] separately' (p. 133). This is exactly how words behave and exactly how calls don't and can't behave. But anyway, why would you have to wait for the larynx to drop in order to concatenate things like that? If a call system had just two units, one that meant 'give' and one that meant 'meat', what would stop any organism from producing 'Give meat!'?.
10. Finally, there is the role of the syllable as a template for syntax. Carstairs-McCarthy (p.144) discusses four possible reactions to the similarity between syllabic and syntactic structure; it's a coincidence; it's two different expressions of a more fundamental factor; syntax influences phonology; phonology influences syntax. There is, alas, a fifth: the similarity is illusory. According to Carstairs-McCarthy, the tripartite division of the syllable onset, nucleus, coda corresponds to the tripartite division of the sentence-- subject, verb, object. But what about the majority of the world's languages, which have orders other than SVO?. Carstairs-McCarthy tries to account for verb-final languages by appealing to onset maximization. But all that onset maximization does is determine where syllable boundaries occur in the minority of languages that permit consonant clusters. The only possible syllabic model for SOV sentences would be a language that did not allow codas but permitted two or more consonants in the onset. But no such language exists. Moreover, while subjects and objects would both inhabit the onset equivalent in verb-final languages, they would both inhabit the coda equivalent in verb-initial languages.
11. Again, there are no pairs of languages such that one puts all non-nuclear constituents of the syllable in the onset while the other puts them all in the coda. Carstairs-McCarthy's attempt to incorporate VSO languages into his schema by pointing out they often have function words in the onset (pre-verbal) position is a red herring it's subjects and objects we're talking about here. But even if the analogy held, it would explain nothing. Word order is the most trivial part of syntax, if indeed it is part of syntax at all it simply falls out from the order in which arguments with different thematic roles are mapped onto phrase structure. The real elements of syntax; hierarchical theme-to-phrase- structure mapping, command, control, binding, scope, movement, constraints on movement, the reference of empty categories, not to mention the biggest of all, recursion are left entirely unexplained by the structure of syllables. As for X-bar structure, to which Carstairs-McCarthy also finds a syllabic analogy, this has been abandoned as unnecessary in more recent minimalist work (e.g. Chomsky 1995--Carstairs-McCarthy actually cites this paper, but apparently without having appreciated its significance).
12. The only element Carstairs-McCarthy comes clean on is recursion; he could hardly avoid it, since it is perhaps the most basic property of syntax, and syllabic structure is devoid of anything resembling recursion. Recursion, he suggests, 'might reasonably be expected to have arisen during the course of syntactic evolution after it had been kick-started by syllable structure' (p. 173). When? Why? How? There already exists at least one evolutionary explanation of syntax which does not require this Micawberish approach; in the model of Calvin and Bickerton (2000), recursion is not merely explained, but unavoidable.
13. Furthermore, there's the basic fact that an analogy isn't an explanation. To say that 'the neural organization underlying syllable structure was co-opted to provide a syntax' (p. 148) tells us very little. Why and how would a neural organization for motor articulatory processes become involved in the sequencing of lexical items? Nor does Carstairs-McCarthy show any way in which syllable structure could have factored out from monocategoric calls those nouns and verbs without which any correspondence with nucleus and margins is impossible.
14. In his book, Carstairs-McCarthy has expounded a novel and ingenious approach to language evolution. Several times during the course of the book he suggests that the only obstacle to the ideas he advances is the inability of critics to see past conventional assumptions about language. Alas, novelty is no guarantee of quality, and there are just too many holes in his proposals for them to stand as viable alternatives to more established views.
Bickerton, D. 1990. Language and species. Chicago: University of Chicago Press.
Calvin, W., and Bickerton, D. 2000. Lingua ex machina. Cambridge, Mass.: MIT Press.
Carstairs-McCarthy, A. (1999) The Origins of Complex Language: Oxford: Oxford University Press.
Carstairs-McCarthy, A. (2000) The origins of complex language. PSYCOLOQUY 11(082) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/2000.volume.11/ psyc.00.11.082.language-origins.1.carstairs-mcCarthy http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?11.082
Cheney, D., and Seyfarth, R.M. 1990. How monkeys see the world. Chicago: University of Chicago Press.
Chomsky, N. 1995. Bare phrase structure. In G. Webelhuth, ed., Government and binding theory and the minimalist program, 383- 439. Oxford: Blackwell.