I summarize the major conclusions of Carstairs-McCarthy's 1999 book, and show how it differs from other evolutionary accounts of the human language faculty. On that basis I recommend the book. I then raise several arguments (commonality vs. influence; languages without onsets; nature of phonological vs. syntactic representations) that deserve further discussion in the context of Carstairs-McCarthy's proposals.
2. Carstairs- McCarthy's concern for details are made explicit right from the start, as he stresses that the inquiry will be limited to only certain aspects of human syntax (viz. those aspects that according to him are reducible to phonological aspects). Thus, the book does not set out to answer all the whys and hows of language evolution. Rather, the analysis focuses on three problems, taken to be core properties of human language: (I) vocabulary size (why do we have as many words as we do?); (II) duality of patterning (typically, linguistic expressions are analyzable on two levels, composed of meaningless elements (the 'sound' side), and meaningful ones (the 'meaning') side; put differently: syntax relates sound and meaning); (III) the distinction between Sentences and Noun Phrases (why do we have both? Why can't we speak 'Monocategorial' to borrow Cartstairs-McCarthy's term: a language consisting of only verbal or nominal elements, but not both?). Defining the inquiry the way Cartstairs-McCarthy does not only makes it more tractable (and, one is tempted to add, more 'falsifiable'); it also suggests that if a coherent answer can be given to these questions without affecting other aspects of language evolution, the origin of language has more than one source. It is not an all-at-once catastrophic event.
3. As Cartstairs-McCarthy correctly observes, the above questions are typically not addressed, as the characteristics they pertain to are taken for granted by everybody. Cartstairs-McCarthy's merit is then to convince us that the issues are deep, and the answers might not be what common sense dictates. Indeed, as will be detailed below, his solution is that the source of the three core characteristics listed above can be traced back to properties of phonological organizations, and ultimately to the way the vocal tract is structured in humans. Claiming, as Cartstairs-McCarthy does, that (for instance) the time-honored distinction between truth and reference dear to many a philosopher might be the result of the way sounds are arranged into syllables is, to say the least, somewhat unconventional, which makes it all the more interesting.
4. As can already be seen, Cartstairs-McCarthy's solution comes close to giving substance to the claim repeatedly made by Stephen Jay Gould or Noam Chomsky (among others) that major characteristics of language might be 'exaptations,' by-products of changes that took place for reasons quite unrelated to language itself. Thus is not the place to re- open the vices and virtues of exaptations. Dennett (1995), for one, addresses the issue explicitly, but in a way that I find unsatisfactory (see Rose 1997 for general remarks against Dennett's rejection of 'exaptive' explanations). For comments specific to language evolution, see Jenkins 2000 and Uriagereka 1998. Once again, the virtue of Cartstairs-McCarthy's account is that it supplements the basic idea with a precise proposal.
5. As physicist Frank Close (2000) has made clear in "Lucifer's Legacy," asymmetry (alternatively, broken, hidden, or lost symmetry) pervades in our world, so much so that only symmetry seems to catch the eye. It is therefore harder (but no less rewarding) to concentrate on asymmetries, and try to account for them. Researchers currently engaged in the quest for supersymmetry (see Gribbin 1998 for overview) are doing just that: spotting asymmetries, and working their way back to their origins: Supersymmetry, or the breaking of symmetry. As any physicist would be ready to concede, the task is by no means trivial.
6. The task Cartstairs-McCarthy undertook is, I believe, of roughly the same order of complexity. As he amply shows, the Sentence-Noun Phrase (hereafter, S-NP) distinction, or its semantic correlate, the Truth-Reference distinction, are common parlance in linguistics, philosophy, and related fields. While the NP-S distinction has characteristically gone unnoticed, the Truth-Reference distinction has attracted the attention of many famous philosophers of language. But, as Cartstairs-McCarthy is at pains to show, no one has succeeded in justifying it in non-linguistic terms (i.e., finding an extra-linguistic motivation for it) (see Baker 2000, where the same point is made with equal emphasis). Much as the S-NP distinction, it remains intra- linguistic, and in need of explanation.
7. The other two asymmetries Cartstairs-McCarthy chose to explore (vocabulary size and duality of patterning), unlike the previous one, has figured in many evolutionary hypotheses, but typically the answers offered do not stand up to closer scrutiny (once linguistic details are taken into consideration). In chapter 4, he reviews some of the major proposals to account for the above asymmetries. In a nutshell, Cartstairs-McCarthy's solution to the three asymmetries is as follows. Humans share with closely related primates what he dubs the avoidance of synonymity. As the human vocal tract evolved (mainly) as a result of bipedalism, humans were able to produce increasingly differentiated sounds. Memory limitations and synonymy avoidance led them to organize sounds in a more abstract, economical, but dual way. To keep things economical, sounds were organized according to sonority to maximize contrast, which, according to Cartstairs-McCarthy, led to the birth of the organized syllable as we know it today, with a crucial distinction between Onset and Nucleus (Consonants and Vowels, to use a gross oversimplification). The parallelism between syllable structure and syntactic phrase structure has often been noted (see FIGURE 1).
a. Syllable b. Phrase / \ / \ / (Rhyme) / ( ) Onset / \ Specifier / \ Nucleus Coda Head Complement
8. Cartstairs-McCarthy's central idea is that the reconfiguration of the vocal tract in human led to (1a), and (1a) was co-opted ('exapted') by syntax, yielding (1b). Taking as a clue the fact that many languages do not have codas, (1a) is reducible to the Onset-Nucleus distinction. Its syntactic correlate, according to Cartstairs-McCarthy, is the S-NP distinction, which in turn gave rise to the Truth-Reference distinction. In chapter 6, Cartstairs-McCarthy shows how his account is compatible with neurological and anthropological findings, sometimes deepening and complementing alternative evolutionary accounts of language. As stated at the beginning, the virtue of Cartstairs-McCarthy's accounts lies in (I) his attention to linguistic details to motivate his general claims, (II) his focus on previously unnoticed, yet central, asymmetries, and (III) the originality of his account. Any future attempt to find the origins of language will have to wrestle with the above asymmetries, and at the very least take the relationship between syntax and phonology seriously, if only for the reasons given in chapter 6.
9. I would now like to raise some issues that I think deserve further discussion in the context of Cartstairs-McCarthy's proposal. The first point is that it is now believed that categories like NP and S (/V(erb)P(hrase)) are projections of features like Case or Agreement (not categorial features per se) which do not seem to me to be amenable to the same kind of reduction Cartstairs-McCarthy attempts. Phrase structure might look identical to Syllable structure, but the underlying substance (features) might be different. Typically, syntactic features are kept distinct from (morpho-)phonological features.
10. How syntactic features could arise without appealing to duality of patterning and thereby verging on circularity) is far from clear. Uriagereka (1998) proposes that syllable structure (and possibly syntactic structure) should be regarded as the linguistic reflex of a more general pattern of self-organization: the Fibonacci series. Cartstairs-McCarthy briefly mentions evolutionary explanations relying on laws of Self-organization. See: http://www.inform.umd.edu/ARHU/Depts/Linguistics/People/Faculty/JUriagereka/
11. It is conceivable that syntactic and phonological organizations are alike not because one derives from the other, but because a common pattern (the way features project) underlies both. The fact that the substance of syntactic patterns is hardly reducible to phonology and that current syntactic theorizing views this substance as the essence of projection suggests that a more abstract underlying schema of self-organization might be at play. Pursuing this line, one would not be surpised to discover that the NP-S distinction is at bottom a 'chemical' distinction. Like chemistry, syntax aims at combining units. For units to be combined, there must minimally exist 'matching' (binary) features (elements). This might well have given rise to the fundamental distinction of the NP-S type. Why precisely this one is unclear, but, if current research is on the right track, the answer is bound to lie within the set of features syntax operates on, but whose origins Cartstairs-McCarthy does not account for. So, ultimately, the NP-S distinction remains unexplained.
12. There have recently been attempts (see Chomsky 1995 and references therein) to unify the basic syntactic operation leading to phrase structure with other syntactic rules (like displacement/permutation operations). If Cartstairs-McCarthy is right, one would have to claim that these other syntactic operations also have a phonological correlate. This seems harder to prove. There is now ample evidence (see Bromberger and Halle 1989) that syntax relies on principles that are ontologically different from phonological rules. In addition, Raimy 1999 argues at length that despite superficial similarities, syntactic and phonological representations encode fundamentally different relations. Likewise, Uriagereka (1999) tackles the issue of recursivity -- a core property of syntax, but not (obviously) one of phonology.
13. Cartstairs-McCarthy's proposal requires a more comprehensive discussion of Sign Language than he provides. The role of the vocal tract is crucial to his account. It is therefore of utmost importance to address the issue of the emergence of sign language. Current research has it that spoken and signed languages share the same syntactic principles. This is surprising in an account which (ultimately) derives basic syntactic principles from the architecture of the vocal tract. Cartstairs-McCarthy addresses the issue briefly, and claims that signed languages may lack the crucial NP-S distinction. This can only be accepted once the vast array of facts suggesting the contrary has been reanalyzed.
14. Finally, the universality of the onset-nucleus distinction has recently been challenged in a convincing way by Breen and Pensalfini (1999). They show that the Central Australian language Arrernte lacks onsets. This might be dismissed as a 'fluke' of nature, or related to the existence of verb- initial languages (which might be said to lack NP-'onsets'). The problem is that Arrernte, and presumably all human languages, make the NP-S distinction, have recursivity, etc. This is surprising if the postulated link between phonology and syntax is so tight. It is less surprising under a self-organization account, which allows syntax and phonology to operate on different substances.
15. The Arrernte case suggests that although Cartstairs-McCarthy's adopts the widely accepted functionalist account of syllable structure (motivated by the architecture of the vocal tract), it is not obviously correct. Language is plastic in that it allows for cross-linguistic variation, but maybe not so plastic as to have its form dictated by its alleged function (see Gazzanigga 1998 concerning the limitations of 'plasticity' in a more general context). The debate about form versus function has a long history. The adequacy of Cartstairs-McCarthy's account will ultimately lie in the resolution of this deep issue.
16. Whether the above considerations ultimately cast doubt on the enterprise will have to await further research, but hopefully they will generate fruitful debate.
Baker, M. (2000) Categories and category systems. Ms., Rutgers University.
Breen, G. and Penslfini, R. (1999) Arrernte: a language with no syllable onsets. Linguistic Inquiry 30, 1-25.
Bromberger, S. and Morris H. (1989) Why phonology is different. Linguistic Inquiry 20, 51-70.
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
Chomsky, Noam. 1995. The minimalist program. Cambridge, Mass.: MIT Press.
Close, Frank. 2000. Lucifer's legacy. Oxford: Oxford University Press.
Dennett, Daniel. 1995. Darwin's dangerous idea. New York: Simon and Schuster.
Gazzaniga, Michael. 1998. The mind's past. Berkeley: University of California Press.
Gribbin, John. 1998. The search for superstrings, symmetry, and the theory of everything. Boston, Mass.: Little, Brown, and Co.
Jenkins, Lyle. 2000. Biolinguistics. Cambridge: Cambridge University Press.
Raimy, Eric. 1999. Representing reduplication. Doctoral dissertation, University of Delaware.
Rose, Steven. 1997. Lifelines. Oxford: Oxford University Press.
Uriagereka, Juan. 1998. Rhyme and reason. Cambridge, Mass.: MIT Press.
Uriagereka, Juan. 1999. What's in a word. Ms., University of Maryland. http://www.inform.umd.edu/ARHU/Depts/Linguistics/People/Faculty/JUriagereka/home.html