Andrew Carstairs-McCarthy (2000) The Origins of Complex Language. Psycoloquy: 11(082) Language Origins (1)

Volume: 11 (next, prev) Issue: 082 (next, prev) Article: 1 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 11(082): The Origins of Complex Language

[Oxford University Press 1999, ISBN 0-19-823822-3, 0-19-823821-5]
Precis of Carstairs-McCarthy on Language-Origins

Andrew Carstairs-McCarthy
University of Canterbury
Department of Linguistics
Private Bag 4800
New Zealand


Some puzzling characteristics of grammar, such as the sentence/NP distinction and the organization of inflection classes, may provide clues about its prehistory. When bipedalism led to changes in the vocal tract that favoured syllabically organized vocalization, this made possible an increase in vocabulary which in turn rendered advantageous a reliable syntax, whose source was the neural mechanism for controlling syllable structure. Several features of syntax make sense as byproducts of characteristics of the syllable (for example, grammatical 'subjects' may be byproducts of onset margins). This scenario is consistent with evidence from biological anthropology, ape language studies, and brain neurophysiology.


ape, aphasia, brain development, evolution of language, grammar, language, larynx, noun phrase, predication, principle of contrast, reference, sentence, sign language, speech, syllable, truth

    Below is the Precis of "The Origins of Complex Language" by Andrew
    Carstairs-McCarthy (740 lines). This book has been selected for
    multiple review in Psycoloquy. If you wish to submit a formal book
    review please write to indicating what
    expertise you would bring to bear on reviewing the book if you were
    selected to review it.

    (If you have never reviewed for PSYCOLOQUY or Behavioral & Brain
    Sciences before, it would be helpful if you could also append a
    copy of your CV to your inquiry.) If you are selected as one of the
    reviewers and do not have a copy of the book, you will be sent a
    copy of the book directly by the publisher (please let us know if
    you have a copy already). Reviews may also be submitted without
    invitation, but all reviews will be refereed. The author will reply
    to all accepted reviews.


    Psycoloquy reviews are of the book, not the Precis. Length should
    be about 200 lines [c. 1800 words], with a short abstract (about 50
    words), an indexable title, and reviewer's full name and
    institutional address, email and Home Page URL. All references that
    are electronically accessible should also have URLs.


    Most recent investigators assume that the brain has always been the
    most important part of human anatomy for the evolution of language,
    and do not seriously examine other conceivable directions in which
    grammatical evolution might have proceeded. In "The Origins of
    Complex Language," it is suggested that certain central features of
    language-as-it-is, notably the distinction between sentences and
    noun phrases, are by no means inevitable outcomes of linguistic or
    cognitive evolution, so that where they come from constitutes a
    genuine puzzle. The solution that is proposed is that
    grammar-as-it-is was, in fundamental respects, exapted from, or
    tinkered out of, the neural mechanisms that arose for the control
    of syllabically organized vocalization, made possible by (among
    other things) the descent of the larynx. This proposal turns upside
    down mainstream views about the relationship between language
    development and vocal tract development, and also challenges the
    logical and epistemological basis of notions closely tied to the
    distinction between sentences and noun phrases, such as
    'reference', 'predication' and 'assertion'. It should therefore be
    of interest to anthropologists, psychologists, cognitive
    scientists, linguists and philosophers of language.


1. A difficulty that affects human researchers on the origin of language is that we cannot observe language 'from the outside', as members of another equally intelligent species would be able to do. However, we can overcome this deficiency with imagination, thereby recognising how much of language-as-it-is cannot be taken for granted, and hence needs explanation.


2. This inquiry focusses on three peculiarities of human language: (i) vocabulary size; (ii) duality of patterning (i.e. analysability into both meaningless elements [sounds] and meaningful ones [morphemes and words]); and (iii) the distinction between sentences and noun phrases (NPs). These three characteristics seem superficially unrelated, so a scenario for language evolution that accounts for them in a unified fashion will be prefer able (other things being equal) to one that does not.

3. One can imagine a kind of language, like Bickerton's (1990) 'protolanguage', that would lack any kind of syntax, so would necessarily lack a sentence/NP distinction. But two kinds of syntactically structured language are conceivable also, lacking the sentence/NP distinction, and representing two out of the many alternative directions in which language might conceivably have evolved. A task for any reasonably complete account of language evolution, therefore, is to explain why language has evolved in the direction it has, rather than in either of these two alternative directions, which I will call 'Spatiotemporal' and 'Monocategoric'.

4. In Spatiotemporal language, each well formed, complete expression consists of three constituents, expressing (i) a location in space (e.g. 'in Christchurch'), (ii) a location in time, (e.g. 'yesterday') and (iii) an object or event (e.g. 'the Dalai Lama' or 'man bites dog'). Spatiotemporal syntax complies with Pinker's suggestion (1994:117) that an expression needs to refer to time in order to be true or false, and to that extent represents a plausible outcome for language evolution; also, if complete expressions can be embedded within the object-or-event constituent, it provides scope for recursion. However, it contains no counterpart of the sentence/NP distinction.

5. In Monocategoric language, there are simple expressions and complex ones, the latter being formed by combining an apppropriate number of expressions (themselves simple or complex) with an operator of the appropriate valence (one-place, two-place, etc.). For example, if you and snake are simple expressions, SEE is a two-place operator, and YESTERDAY is a one-place operator, then a conceivable complex expression is you snake SEE YESTERDAY, glossable in English as 'You saw a snake yesterday' or 'the snake that you saw yesterday'. This seeming ambiguity reflects the fact that there are no distinctions in grammatical category among Monocategoric expressions comparable to the distinction between verbs, nouns and sentences. In its categorial uniformity (though not in its syntactic elaboration), Monocategoric language resembles the alarm call system of vervet monkeys (Cheney & Seyfarth 1990), concerning which it makes no sense to ask whether an individual call is nominal or verbal.

6. In lacking the sentence/NP distinction, Spatiotemporal and Monocategoric lack any direct method for encoding the distinction between assertion (whether true or false) and identifying reference (whether successful or unsuccessful). But an intelligent nonhuman observer might well be puzzled as to why we should think there need to be these two distinct ways in which expressions in human language can fit the world (or fail to fit it). Why not three or seventeen ways, or just one? To such an observer, Monocategoric or Spatiotemporal might seem a more natural outcome of linguistic evolution than actual human language, especially if the starting-point was something like the vervet monkey call system. So the sentence/NP distinction, as a central feature of syntax-as-it-is, demands explanation.


7. A possible explanation for the sentence/NP distinction may lie in the truth/reference distinction, provided that that distinction is motivated independently of grammar; that is, provided that that distinction does not turn out to be a mere logical and semantic Doppelganger of the grammatical one. A grammar-independent motivation for the truth/reference distinction is sought, unsuccessfully, in the work of the philosophers Frege, Wittgenstein and Strawson.

8. According to Frege (1980), the referring expression 'the Morning Star' and the judgement 'Archimedes perished at the capture of Syracuse' both have a 'reference' ('Bedeutung'), namely Venus and truth respectively. So finding the motivation that we are seeking is a matter of finding the criteria for identifying those expressions that have 'judgeable content', and hence can have truth as their reference. Frege claims that these criteria are independent of grammar, so that an expression that refers to truth need not be a sentence, and a sentential or clausal expression may refer to (for example) an instant or period in time, rather than truth. But in practice the criterion does seem to be grammatical, in that the only expressions that turn out invariably to lack 'judgeable content' are expressions that are not paraphrasable by means of sentences in German or English, such as house and Julius Caesar.

9. For Wittgenstein, in his Tractatus Logico-Philosophicus (1961), the truth/reference distinction reflects the linguistic distinction between propositions and names, which in turn reflects the distinction in the world between 'states of affairs' and 'objects', a state of affairs being a 'combination of objects' or something produced by 'the configuration of objects'. This seems to leave open the possibility of a state of affairs involving just one object just as, in logic, a set may have just one member. But if a proposition reflects a single-object state of affairs it is not obviously distinguishable from the name that constitutes it. If Wittgenstein had recognized and discussed this possibility, he might have been less confident that the proposition/name distinction reflects a fundamental distinction in the world.

10. In his Philosophical Investigations (1958), Wittgenstein abandons the notion that propositions are direct reflections, or pictures, of reality. However, he still adheres to the view that 'propositions' are not only distinct from names but also an essential element in language, because (he claims) naming is 'not a move in the language-game - any more than putting a piece in its place on the board is a move in chess.' But this chess analogy leads to a fatal equivocation on 'naming'. Putting a piece in its place on the board is equivalent to 'naming' in the sense of defining a term; but we can also 'name' in another sense, when we use a term that has already been defined. From the fact that one cannot 'make a move' linguistically by 'naming' in the first sense, it does not follow that one cannot 'make a move' by 'naming' in the second sense. Therefore Wittgenstein's argument for the linguistic insufficiency of 'naming', and hence for the necessity of 'propositions', collapses and, with it, any nonlinguistic support that he might provide for the truth/reference distinction.

11. Strawson (1959; 1974) seeks to supply a basis for the grammatical subject/predicate distinction (and hence a basis for the kind of expression that is potentially 'true') in terms of the ontological distinction between 'particulars' and 'universals'. This ontological distinction is held to underlie the distinction between expressions that identify particulars, such as the Empire State Building, and expressions that denote universals, such as generous or generosity. But he underestimates the significance of sentences that contain more than one particular-identifying expression, such as Jean hates the Empire State Building. For Strawson's argument to succeed, he must explain why just one of these particular-identifying expressions (here, Jean) deserves to be classified as 'subject', while the others are relegated to a position within the 'predicate', and this explanation must crucially be nonlinguistic. However, he supplies no such explanation. He appeals also to a parallel between particulars and particular-identifying expressions, both being supposedly in some sense non-negatable; but this parallel turns out to break down, because NPs can in some languages be negated.

12. These philosophers' failure to supply a nonlinguistic motivation for the truth/reference distinction, suitable to underpin the sentence/NP distinction, confirms that the sentence/NP distinction may be a stranger feature of language than at first appears, and reinforces the attractiveness of any language evolution scenario that accounts for it, such as the one offered in chapter 5.


13. No serious explanation has ever been offered for the human ability to acquire a far more elaborate vocabulary than is necessary to subserve a hunter-gatherer life-style. As for duality of patterning, it has been suggested that individual sounds were originally meaningful, but lost their semantic motivation when combined into larger units ('words') which became subject to phonological change and semantic drift. However, no one can agree what the original meanings of individual sounds may have been. Besides, supposed evidence for this view, drawn from the small vowel inventories reconstructed for the proto-languages of certain language families, is weak, because even if in some language there are few vowel phonemes, that language may still exploit a large part of the phonetic vowel space.

14. The sentence/NP distinction has attracted more attention. Some scholars (Pinker and Bloom 1990; Newmeyer 1991) see sentences, or perhaps 'propositions', as a natural method of encoding predicate-argument structures, i.e. semantic complexes denoting an action-or-state and its participants (who does what to whom). Apes and even monkeys today seem to represent their world mentally in terms of such structures, so it is likely that our prelinguistic ancestors did so too. But it does not follow that any linguistic encoding of predicate-argument structures would lead to the sentence/NP distinction. This is because predicate-argument structures can be encoded nominally (e.g. the Romans' capture of the city) just as well as sententially (The Romans captured the city).

15. It has also been argued that, for a proposition to be asserted and not merely entertained, its expression must locate it in time, as verbal tense does (Pinker 1994). But this too is insufficient to account for the sentence/NP distinction, because in some languages (e.g. Chinese) most or all sentences lack grammatical tense, while in some languages the nominal encodings of predicate-argument structures regularly display it, through tense contrasts in infinitives or participles.

16. Some psychologists and philosophers posit a fundamental binary distinction between things and events, and this distinction has been claimed (e.g. by Langacker 1990) to underlie the linguistic distinction between NPs and sentences. However, it is at least conceivable that the boot is on the other foot: we humans think in terms of a bipolar thing/event distinction (or continuum) only because we speak languages that distinguish NPs and sentences. To exclude that possibility it would be necessary to demonstrate that, quite independently of language, our experience is structured in terms of a one-dimensional continuum rather than, say, a multidimensional 'thing-event space'. But such a demonstration is not obviously feasible. Our collective experience includes, for example, phenomena known as 'the Niagara Falls' (indefinite as to time and participants but spatially circumscribed) and 'the Second World War' (definite as to time and participants but spatially vague); and it is not clear that we do justice to either phenomenon by pigeonholing it as either an 'event' or a 'thing'.

17. In many languages, the grammatical structure of utterances often seems to serve the function of distinguishing 'given' from 'new' information (or 'topic' from 'comment'). This is one factor in the choice between active and passive voices in English, as in (What did the Romans do?) They defeated the Gauls versus (What happened to the Gauls?) They were defeated by the Romans. But, again, even if the topic/comment distinction influences grammatical structure, it does not suffice to motivate the sentence/NP distinction, because similar differences in structure are regularly available in NPs too (e.g. the Romans' defeat of the Gauls versus the Gauls' defeat by the Romans). Also, the fact that sentences whose entire content is 'new' (sometimes called 'thetic judgements') generally lack special grammatical characteristics militates against any special role for the topic/comment distinction in grammatical evolution.

18. The argument that all complex systems display hierarchical structure (Simon 1962) may explain why syntax involves the embedding of constituents inside larger constituents, but it does not explain why syntactic constituents should differ grammatically in the way they do.

19. Some proponents of the view that sign language (or gestural language) predated spoken language see certain Deaf signs as providing a model for the subject-verb-object structure of sentences (Armstrong, Stokoe and Wilcox 1995). However, many Deaf signs are structured quite differently, so this suggestion creates the problem of explaining why these other signs have not left their mark on syntax too. One reason why the gestural-origin theory appeals to some scholars is that they see the arbitrary, or symbolic, nature of modern language as a major hurdle to be surmounted by hominids whose signing had until then been predominantly representational or iconic, for which gesture would have been better adapted than speech. But this argument overlooks the fact that, even among vervet monkeys, vocal communication is symbolic rather than iconic (vervets' alarm calls do not resemble the predators that they warn of), so symbolic communication is not a peculiarly human achievement.


20. Two principles that appear to guide concept formation and vocabulary acquisition in early childhood are Mutual Exclusivity (the extensions of word meanings are mutually exclusive) and Contrast (every two words contrast in meaning) (Markman 1989). Children learn that Mutual Exclusivity can be overridden, for example, when one word is a hyponym of another (as plum is a hyponym of fruit); still, its influence lingers in that word meanings rarely contain incompatible disjuncts (for example, because a fruit cannot be simultaneously a plum and a banana or a plum and an apple, Mutual Exclusivity predicts correctly, that no language should possess terms meaning 'plum-or-banana' and 'plum-or-apple' denoting items at the same level of generality but with overlapping extensions). As for Contrast, its effects linger in that pairs of exact synonyms such as zucchini and courgettes are rare, and mainly limited words learned in adulthood.

21. A puzzle, then, is the apparent exact synonymy of some inflectional affixes, e.g. the plural suffixes of Tag-e 'days', Bild-er 'pictures', Park-s 'parks' and Held-en 'heroes' in German. However, this apparent synonymy disappears once we notice that, among such sets of rival affixes, either all or all but one are reliably diagnostic of the inflection class that a noun carrying it belongs to (i.e. of the noun's inflectional behaviour in all relevant case-number combinations). It is as if -e in German, for example, means not just 'plural' but rather 'plural, inflection class A'. Two theoretical possibilities seem not to occur: (a) two affixes are exact synonyms in that both mean just 'plural', neither of them conveying any information about inflection class; and (b) an affix conveys information that is disjunctive in nature, such as 'class A or class C'. These observations suggest that, despite appearances, the principles guiding word learning apply to inflectional affixes too; for (a) is excluded by Contrast, and (b) illustrates the kind of disjunctive meaning that is inhibited by Mutual Exclusivity. It follows, rather surprisingly, that communicatively pointless, purely intragrammatical 'meanings' such as 'inflection class A' may exist whose sole function is to ensure that the bans on exact synonymy and on incompatible disjunctive meanings are maintained by affixes that would otherwise risk violating them.

22. If synonymy-avoidance principles are so deeply entrenched as to sustain "meaning" differences, even quite pointless ones, wherever potentially synonymous forms exist, what does that imply for language evolution? A synonymy-avoidance assumption seems well adapted to facilitate the learning of a small finite vocabulary of a few score vocalizations (not much more extensive than the call vocabulary of contemporary chimpanzees): the learner can assume confidently that any new vocalization means something new. But if some physiological change had the effect of vastly expanding the repertoire of distinct vocalizations, such synonymy-avoidance principles could have been powerful enough by themselves to instigate the emergence of new meanings, quite independently of cognitive factors that might give some meanings higher priority for expression over others. Precisely such a physiological change occurred during early human evolution: the lowering of the larynx, which, by rendering the pharynx and the oral cavity largely separate as resonating chambers, fulfilled the acoustic preconditions for distinguishing a wide range of vowel sounds, and hence for a new kind of vocalization organized into syllables with contrasting vocalic nuclei.

23. The lowered larynx, combined with synonymy-avoidance principles already in place, could thus have provoked a vocabulary explosion. But the new vocal resources would make potentially available a far larger vocabulary than limited memory resources could cope with. An advantage would therefore accrue to individuals who could square this circle by reliably interpreting some actual or possible vocalizations as strings of shorter vocabulary items; that is, who could develop a reliable syntax. Moreover, such multi-item strings would directly account for duality of patterning, because they would now be analysable not only into individually meaningless consonantal and vocalic elements, as before, but also into meaningful elements ('words'). But what was the source for the kind of syntax that actually emerged, given that this is not the only conceivable kind, and that evolutionary accounts so far have not explained the sentence/NP distinction?

24. Evolution operates not by designing ideal solutions to problems, but by tinkering with what is available so as to arrive at solutions that may be messy but are good enough to survive in a population's gene pool. What was available at the point when an advantage would accrue from possession of a reliable syntax included a set of neural mechanisms for controlling syllabically organized vocalization. It is therefore of interest to consider what sort of syntax might emerge by evolutionary tinkering from those neural mechanisms, and compare it to the sort of syntax that exists.

25. A maximal syllable can be divided into a central nucleus (generally vocalic) and two margins (generally consonantal), namely an onset preceding the nucleus and a coda following it. These elements, and the syllable itself, have the following characteristics:

    (i) Only the nucleus is obligatory in all languages.

    (ii)  Marginal material is assigned to onsets in preference to
    codas ('onset maximization'), and in some languages codas are not
    permitted at all.

    (iii) In many languages, a distinction in 'weight' between 'heavy'
    and 'light' syllables affects the distribution of stress and shows
    up in the metrical conventions of verse. Only the nucleus and coda
    (often grouped together as the 'rhyme') contribute to syllable
    weight; the onset is irrelevant to it.

    (iv)  Syllables cannot be embedded in syllables (i.e. a syllable
    cannot function as a nucleus or margin of a larger syllable).

In sum, syllables have the structure [(Onset) [Nucleus (Coda)]].

26. Let us consider the likelihood that the syllable, with these characteristics, might be tinkered into a syntax on Spatiotemporal or Monocategoric lines, as described in chapter 2. The likelihood seems low. A Spatiotemporal expression has three obligatory elements, not just one, as (i) above would lead us to expect, and there is nothing in Spatiotemporal corresponding to the distinction between onset and coda margins, the privileged status of the onset margin, or the role of the rhyme constituent at (ii)-(iv). In Monocategoric, the well-formedness of simple expressions (such as snake, you)in isolation makes them at first seem analogous to syllable nuclei, but this analogy disappears when one notes that a complex expression can contain two or more simple expressions in conjunction with an operator (e.g. you snake SEE), whereas a syllable cannot contain two or more nuclei. Also, as in Spatiotemporal, there is nothing in a complex Monocategoric expression corresponding to the rhyme or to the privileged status of the onset.

27. How likely is it that some characteristics of syntax-as-it-is may be reflections of an origin in syllable structure? This possibility seems more promising. A candidate for the sentential counterpart of the nucleus, the sole obligatory element in the syllable, is the verb, the sole obligatory element in the sentence (that is, in free-standing sentences, ignoring the kind of ellipsis that occurs in dialogue). Candidates to be counterparts of the privileged status of the syllable onset are various ways in which some nonverbal material is privileged, namely:

    (a) by placing all nonverbal material before the verb, as in
    consistently verb-final languages such as Japanese (which mimic
    syntactically the syllable structure of languages that disallow
    codas entirely);

    (b) by according special status to a nonverbal constituent
    immediately before the verb, e.g. by allowing only one such
    constituent (as in 'verb-second' languages such as German), or by
    making this position an informational 'focus' (as in Hungarian);

    (c) by according a special syntactic status, such as 'subject' or
    'topic', to just one nonverbal (nominal) constituent, usually but
    not always before the verb, or by requiring that all sentences
    should contain a subject, even if semantically empty (e.g. 'it' in
    'It is raining').

Although there are some verb-initial languages, by contrast to (a), they are a small minority. As for (b) and (c), it is striking that there are no mirror-image languages that give analogous special status to a constituent immediately after the verb, and there are no languages that require every verb to have an object. Finally, the clearcut difference in status between syllables and syllable margins may be reflected in the difference in status, not explained by any of the approaches discussed in chapters 3 and 4, between sentences and NPs.

28. A requirement that every sentence should contain a counterpart of the syllable nucleus will be most easily fulfilled if the items that can appear in this position include some that are readily combinable with any other semantic content. This may account for the prevalence of auxiliaries as a kind of semantically bleached verb in many languages. It may also help to explain why the expression of negation tends to migrate to the verbal or auxiliary constituent, even if the semantic scope of the negator does not include the verb. For example, Peter hasn't lost his bicycle may, according to intonation and context, have a variety of interpretations, including 'Somebody has lost his bicycle, but it isn't Peter', or 'Peter has lost a bicycle, but it isn't his'.

29. The analysis of many sentences in many languages into two constituents, a NP and a verb phrase, functioning as so-called 'subject' and 'predicate', may reflect the division of the syllable into an onset and a rhyme.

30. One characteristic shared by Spatiotemporal, Monocategoric and actual syntax is the possibility of recursion or embedding. This characteristic is not derivable from any aspect of syllable structure. But that is not an embarrassment for the syllabic scenario proposed here. The syllabic scenario accommodates well certain aspects of syntax that, on close examination, turn out not to be good design features (despite what we as speakers of human languages may be inclined to assume), and that are for that very reason most likely to yield clues about its origin. By contrast, recursion is obviously functional, in that a syntax that permits it is clearly superior, both for knowledge representation and for communication, to one that does not. Therefore, it is easy to understand why syntactic recursion should have taken root, once it became possible for whatever reason, and it is not necessary to suppose that it should have been present at the earliest stage of syntactic evolution.


31. This chapter explores whether the scenario presented in chapter 5 is consistent with evidence from biological anthropology, brain neurophysiology, and primate behaviour. The outcome is positive, and the syllabic scenario also suggests new ways of interpreting certain evidence from research on aphasia and on Deaf sign language.

32. The syllabic scenario presupposes that vocal tract changes should have led fairly directly to vocabulary enlargement and a syntax incorporating a sentence/NP distinction, with no long intervening period. If we believe Noble and Davidson (1996), then that must be wrong, because (they argue) language developed only quite recently, at the same time as the manufacture of the first artefacts with 'symbolic' significance. But Noble and Davidson define language in an idiosyncratic way, involving awareness of symbolism; so their scenario is consistent with the possibility that syntax arose long before what they call 'language'. Besides, their emphasis on symbolism accompanies an assumption that only humans use 'symbols', an assumption that I argued in chapter 4 to be mistaken.

33. The syllabic scenario also presupposes that vocal tract changes, especially the lowering of the larynx, should have got under way originally for reasons quite independent of language. Such a reason is suppled by the 90-degree reorientation of the skull consequent on bipedalism (Schepartz 1993; Aiello 1996b).

34. As a possible precursor of the hierarchical organization of syntax, Greenfield (1991) has cited the hierarchical organization of behaviour. However, this leaves it quite unclear why syntax should have evolved the particular kind of hierarchical structure that it has, involving especially the sentence/NP distinction and the asymmetries discussed in chapter 5. Calvin (1993) has cited more specifically the neural mechanisms necessary for accurate throwing at a small target as possible precursors for the syntactic relationship between a verb and its object. But such mechanisms could just as well have given rise to the relationship between an operator and its argument in Monocategoric, and they supply no obvious precursor for subject NPs in syntax. However, Calvin's suggestion complements the syllabic scenario in that the agility and precision needed for accurate throwing could have contributed to the agility of the tongue and other parts of the vocal apparatus needed to articulate syllables in rapid succession.

35. The enlargement of the human brain is claimed by some (e.g. Aiello 1996a) to have taken place mainly in two spurts, one about 1.5-2 million years ago and the other within the last half-million years. This is consistent with the possibility that the most primitive syllable-derived form of syntax came into use around the transition from australopithecines and Homo Habilis to Homo Erectus, but that most of the later elaboration (including, in particular, recursive embedding) did not emerge until the transition to Homo Sapiens.

36. The syllabic scenario has much to say about how syntax is structured, but nothing to say about what language was originally used for. Theories of language evolution that emphasis its role in fostering social cohesion through gossip (Dunbar 1996) therefore complement the syllabic scenario rather than conflict with it.

37. The brain areas that underlie language include in particular Wernicke's area (damage to which disrupts access to vocabulary), and Broca's area (damage to which typically affects articulation and some, but not all, aspects of grammar). The association between articulatory and grammatical disruption is broadly consistent with the syllabic scenario, as is the proximity of Broca's area to the motor control areas for the tongue and lips rather than (say) the hands. Moreover, the fact that some aspects of grammar are affected more than others in Broca's aphasia (Linebarger et al. 1983) fits in with the syllabic scenario, in that it seems plausible that it is precisely the syllable- derived aspects of syntax that suffer (the basic architecture of the sentence, including the identification of the verbal nucleus and of privileged nonverbal material such as the subject), not other aspects (including complementizer choice, relative clause formation strategies, and the like).

38. Deacon's (1997) view of brain-language coevolution emphasizes the gradually increasing role of the prefrontal cortex in controlling vocalization, which is consistent with a view of language as evolutionarily continuous with a primate vocal call system. Deacon also argues that 'symbolic reference' presupposes 'some form of grammar and syntax'. However, he does not explain why the form of grammar and syntax that emerged should have been precisely one with the characteristics noted in chapter 5, and the relationship that he posits between grammar and 'symbolic reference' seems circular, and hence empty of explanatory force.

39. Researchers who deny any continuity between primate vocal call systems and human language tend to emphasize the closed, nonreferential, innate and involuntary character of call systems, as opposed to the open, referential, learned and voluntary character of language. But in respect of referentiality, learning and voluntary control, increasing evidence tends to show that other primates are more human-like than used to be thought. Moreover, as already mentioned, the view that only humans have made the transition from iconic to symbolic referential communication is shown to be wrong by Cheney and Seyfarth's (1990) studies of vervets.

40. Deaf sign languages are now recognized to be fully-fledged languages quite independent of, and just as versatile as, spoken languages. However, the syllabic scenario for syntactic evolution suggests an intriguing possibility. If the sentence/NP distinction derives from syllable structure in phylogeny, could it be that in ontogeny also a form of syntax incorporating a clearcut syllable/NP distinction depends on exposure to syllabically structured speech? If so, we might expect that Deaf sign languages, while being grammatically elaborate in many ways and permitting recursion (for example), should not display any clear syntactic evidence of a sentence/NP distinction. The reports by Liddell (1980) on relative clauses, and by MacLaughlin (1997) on structural parallels between sentences and 'determiner phrases' in American Sign Language, suggest that this possibility may be worth investigating.

41. Much controversy has surrounded the attempts to teach American Sign Language to chimpanzees and gorillas. In the context of the present inquiry, what is of interest is to determine whether any syntax that they inquire, either through instruction as in the earlier experiments or spontaneously as in the case of the bonobo Kanzi (Savage-Rumbaugh and Lewin 1994), shows clear evidence of a sentence/NP distinction; for, if there is such evidence, then the syllabic scenario is clearly wrong. However, careful examination of the linguistic achievements of chimpanzees and bonobos shows that, remarkable though these achievements are, they display no clear evidence of that distinction, either in comprehension or production. The various experiments do tend to confirm, however, that chimpanzees share with humans an expectation that distinct signs should have distinct referents, which makes it at least likely that a no-synonymy expectation was already in place in time for vocal tract changes to drive vocabulary enlargement in early humans.


42. The syllabic scenario for the origin of syntax presents an at first sight bizarrely topsy-turvy picture of language evolution, in that cognition and brain development are demoted from the central instigating role that is usually ascribed to them. However, this scenario has crucial advantages over any other that has been recently proposed, both in providing a unified account of vocabulary size, duality of patterning and the sentence/NP distinction, and in motivating otherwise mysterious characteristics of syntax. Moreover, there is still scope to examine how cognitive factors interacted with linguistic development, for example how social cognition sheds light on what language was first used for and what sort of reproductive advantage it originally bestowed.

43. If sentences, with the syntactic characteristics that they have, are not an inevitable feature of any sort of complex language but merely an accidental byproduct of the route by which language happened to develop in humans, then the contrast between knowledge that is most naturally expressed in sentential form ('declarative', 'propositional' or 'explicit' knowledge) and knowledge that is not so expressed ('procedural' or 'implicit' knowledge) recedes in importance. This in turn reduces in importance one of the last remaining respects in which humans are often held to differ radically from other animals, namely access to 'propositions' and their status as 'true' or 'false' representations of the world.

44. A final advantage of this scenario over those that emphasize cognitive or cerebral factors is that it supplies a straightforward reason why it should have been humans, rather than other apes, that have progressed furthest linguistically. The more we come to appreciate the sophistication of other primates' mental representation of reality, the more mysterious their failure to develop language becomes, if mental sophistication is indeed what was important for language development. On the other hand, if what got language going was something peculiar to humans, namely habitual bipedalism and its anatomical consequences, then that puzzle disappears.


Aiello, Leslie C. 1996a. Hominine preadaptations for language and cognition. Modelling the Early Human Mind, ed. by Paul Mellars and Kathleen Gibson, 80-99. Cambridge: McDonald Institute Monographs.

Aiello, Leslie C. 1996b. Terrestriality, bipedalism and the origin of language. Evolution of Social Behaviour Patterns in Primates and Man, ed. by W.G. Runciman, John Maynard-Smith, and R.I.M. Dunbar, 269-89. Oxford: Oxford University Press.

Armstrong, David F., Stokoe, William C., and Wilcox, Sherman E. 1995. Gesture and the Nature of Language. Cambridge: Cambridge University Press.

Bickerton, Derek. 1990. Language and Species. Chicago: University of Chicago Press.

Calvin, William H. 1993. The unitary hypothesis: a common neural circuitry for novel manipulations, language, plan-ahead, and throwing? Tools, Language and Cognition in Human Evolution, ed. by Kathleen R. Gibson and Tim Ingold, 230-50. Cambridge: Cambridge University Press.

Carstairs-McCarthy, A. (1999) The Origins of Complex Language: Oxford: Oxford University Press.

Cheney, Dorothy L., and Seyfarth, Robert M. 1990. How Monkeys See the World: Inside the Mind of Another Species. Chicago: University of Chicago Press.

Deacon, Terrence. 1997. The Symbolic Species: The Co-Evolution of Language and the Human Brain. London: Allen Lane.

Dunbar, Robin. 1996. Grooming, Gossip and the Evolution of Language. London: Faber & Faber.

Frege, Gottlob. 1980. Translations from the Philosophical Writings of Gottlob Frege, ed. by Peter Geach and Max Black (3rd edn.). Oxford: Blackwell.

Greenfield, Patricia M. 1991. Language, tools and the brain: the ontogeny and phylogeny of hierarchically organized sequential behavior. Behavioral and Brain Sciences 14:531-51.

Langacker, Ronald W. 1990. Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin: Mouton de Gruyter.

Liddell, Scott K. 1980. American Sign Language Syntax. The Hague: Mouton.

Linebarger, Marcia C., Schwartz, Myrna F., and Saffran, Eleanor M. 1983. Sensitivity to grammatical structure in so-called agrammatic aphasics. Cognition 13:361-92.

MacLaughlin. Dawn. 1997. The structure of determiner phrases: evidence from American Sign Language. PhD dissertation, Boston University.

Markman, Ellen M. 1989. Categorization and Naming in Children: Problems of Induction. Cambridge, MA: MIT Press.

Newmeyer, Frederick J. 1991. Functional explanation in linguistics and the origins of language [with peer commentary and response]. Language and Communication 11:3-114.

Noble, William and Davidson, Iain. 1996. Human Evolution, Language and Mind: A Psychological and Archaeological Inquiry. Cambridge: Cambridge University Press.

Pinker, Steven. 1994. The Language Instinct. New York: William Morrow.

Pinker, Steven and Bloom, Paul. 1990. Natural language and natural selection [with peer commentary and response]. Behavioral and Brain Sciences 13:707-84.

Savage-Rumbaugh, E. Sue, and Lewin, Roger. 1994. Kanzi: The Ape at the Brink of the Human Mind. New York: Wiley.

Schepartz, L.A. 1993. Language and modern human origins. Yearbook of Physical Anthropology 36:91-126.

Simon, Herbert A. 1996. The Sciences of the Artificial (3rd edn.). Cambridge, MA: MIT Press.

Strawson, P.F. 1959. Individuals: An Essay in Descriptive Metaphysics. London: Methuen.

Strawson, P.F. 1974. Subject and Predicate in Logic and Grammar. London: Methuen.

Wittgenstein, Ludwig. 1958. Philosophical Investigations, trans. G.E.M. Anscombe (2nd edn.). Oxford: Blackwell.

Wittgenstein, Ludwig. 1961. Tractatus Logico-Philosophicus (the German text of Logisch-philosophische Abhandlung with a new translation by D.F. Pears and B.F. McGuinness). London: Routledge & Kegan Paul.

Volume: 11 (next, prev) Issue: 082 (next, prev) Article: 1 (next prev first) Alternate versions: ASCII Summary