The Origins of Grammar: Language in the Light of Evolution 
by James Hurford.
Oxford, 791 pp., £37, September 2011, 978 0 19 920787 9
Show More
Show More

How did​ human beings develop the capacity to speak, and to speak such complex languages? That is the question that James Hurford, emeritus professor of general linguistics at Edinburgh, has sought to answer. The Origins of Meaning, published in 2007, looked at animals’ cognitive representations of the world, their part in the evolution of abstract thought, and the limits of their communication with each other. The Origins of Grammar tells how human beings broke away from other animals to develop that capacity.

In the first section he deals with the pre-conditions of syntax or ‘pre-grammar’. He notes, among other examples of animal behaviour, that the honeybee indicates the direction and distance of food by varying the angle of its posture and the length of its waggle-dance; that the calls of male chaffinches combine territorial challenges addressed to other males with invitations addressed to females, and that those of Campbell’s monkeys have components with specific meanings that can be combined (always in the same order) to express new meanings. But although certain messages can be conveyed and intelligently responded to, there is a limit on both content and complexity. The findings are tested against the Formal Language Theory based on the hierarchy of language types drawn up in Chomsky’s early papers: the animals’ communication systems are found wanting for their inability to generate complex meanings, but Formal Language Theory too is found wanting for its neglect of quantitative constraints and semantics.

The suggestion that syntax and semantics may have evolved independently at first then combined leads to the study of the first shared lexicon, the commonalty of unitary learned symbols that resulted from the convergence of individual expressive choices into a socially accepted pattern, and that preceded combination ‘in meaningful expressions with grammatical shape’. Hurford’s investigation of the origins of the shared lexicon takes in theories of synaesthesia, which claim that, for example, high front vowels, particularly i-sounds, are associated with proximity or femaleness, owing to the higher pitch at which such vowels tend to be uttered. He also considers conventional sound symbolism theory, according to which new words are created with phonetic resemblance to old ones of similar meaning. But even at their strongest, such theories can account for only a few words within the total vocabulary of individual languages. Even in antiquity it was debated whether words had their meanings by nature or convention, as when Nigidius Figulus, in the first century bc, attempted to correlate first and second person pronouns with their mode of articulation. Hurford touches on this topic, noting that the association of particular sounds with those pronouns is fairly well marked within language families but not uniform across the world; he also remarks on the tendency for words meaning ‘mother’ to begin with m or n, and those meaning ‘father’ to begin with p or t, unfortunately without noting that in Georgian ‘father’ is mama and ‘mother’ deda, or that these associations do not protect a consonant against phonetic change: to Latin pater correspond English father, Armenian hayr, and Irish athair, all in accordance with the transformation of Indo-European p.

In the second section of the book, Hurford sets about establishing what our ability to learn languages consists in. Human beings have the capacity for massive storage of vocabulary; for combining words into phrases and phrases into sentences; for morphologically combining units of meaning into single words (from the simple inflections of English to the multiply affixed single words of Inuit or Turkish); for categorising words and phrases into such classes as noun and verb, noun phrases and verb phrases; for identifying grammatical relations; for recognising dependences between words some distance apart; for forming constructional templates whose variables may be given specific values (X was Y’d by Z, for instance, which is treated as a construction in its own right rather than derived from Z Y’d X); and for observing the so-called ‘island constraints’ that limit the possibility of separating words that semantically belong together.

Particular attention is paid to languages thought to be outside the norm, such as Pirahã, an Amazonian tongue that lacks numerals (even ‘one’) and cannot or will not formulate such recursions as John’s brother’s house (allegedly, too, it lacks sentence-embedding, although it often achieves the same results through verbal suffixes), and Riau Indonesian, in which the lack of grammatical markers and the flexibility of word order reportedly allow both ayam makan and makan ayam to express any relation between ‘chicken’ and ‘eating’ without even specifying whether the chicken is (or was, or will be) eating or being eaten.

The book’s final third concerns the evolution of syntax. Hurford takes us back through the mists of human evolution to try to understand why human beings developed language and apes did not. Apes have only a primitive ‘neural specialisation for meaningful signals’: they lack the storage capacity required for language, and cannot understand the hierarchical structures that are essential to it. Even individual apes that have been taught to understand spoken words can follow instructions only that relate to things and persons present, and even then at no great level of complexity. An immense difference in degree of language capacity developed between human beings and apes ‘in the space of no more than four million years’. The former then, Hurford argues, made good their capacity through a process of gene-language co-evolution whereby evolution in the human brain and in language took place in conjunction. Hurford believes that syntax itself evolved in gradual continuity, rather than the sudden advances proposed by Chomskians. The disagreement is a special case of that between the saltationist ‘evolution by jerks’ and the gradualist ‘evolution by creeps’. As Hurford engagingly declares, ‘I am a creep, not a jerk.’ (One is relieved to be past the animus of the 1980s, when the creeps accused the jerks of being Marxists – a charge that would have surprised Chesterton’s Thinker Boulnois, and indeed Chesterton himself.) This is not the only controversy in which he takes sides, for he treats one-word sentences as prior to those with subject and predicate, and maintains that the first human speakers progressed by putting atomised words together rather than breaking up words that in themselves expressed complex propositions. Finally, he surveys the development from autonomous words to bound morphemes and inflections (a set of processes known as grammaticalisation), beginning with the evolution of noun-verb and subject-predicate distinctions, then passing to other categories.

I cannot pretend to competence in animal behaviour or human evolution, but this book is the work of a writer who has read widely and pondered much on his subject and whose account of it is a major contribution to scholarship. The style is too colloquial for my taste, but this is, if not of a piece, then at least consistent with linguists’ habit of presenting trivial example sentences about the daily doings of John and Mary. Even in an evolutionary context, which requires the privileging of spoken language over written as temporally prior to it, one might have hoped that early human beings occasionally had something more interesting to say. As it is, relief must be taken where it is found: in the nursery rhyme ‘This is the farmer sowing his corn,’ quoted to demonstrate the ‘less stringent constraints on multiple right-embedded clauses’, or the contrived example taken from a 2005 paper on syntax, ‘Who plays what instrument? Katie the Kazoo, and Robin the Rebec.’ (I cannot pretend to sympathy with Hurford’s populist question ‘whoever has heard of a rebec?’; even linguistically the word deserves attention for its irregular development from Arabic rabãb.)

One of Hurford’s themes is the importance of semantics, which has been neglected by linguists fixated on abstract structures. If even a child knows that the interrogative form of The girl we had met earlier was singing is Was the girl we had met earlier singing? and not Had the girl we met earlier was singing?, that is not because it has selected the correct rule for transforming statements into questions, but because it wishes to ask about the girl, and knows how to identify her through the previous encounter. Hurford also endorses the opinion that whereas I like the gears in that car may reasonably give rise to the question Which car do you like the gears in?, this is no longer the case if gears is replaced by girl because all cars have gears in them and not all have girls – but his objection fails if more than one car in fact has a girl in it. On the other hand, when the same passage from reasonable to unreasonable is alleged in the sequence Sam said: John likes the gears/girl in that car. Which is a lie – he never saw the car, I do not see why He never saw the car should be thought reasonable in either case; more plausible would be he never drove the car and he never saw the girl.

A few other passages invite disagreement or comment. Hurford is unsure whether My and Sue’s house is grammatical. It does sound awkward, but that is because we expect the first-person pronoun to come last (Sue’s and my house), not first as in Greek and Latin (we recall Wolsey’s alleged ego et rex meus); there is nothing wrong with Your and Sue’s house or His and Sue’s house. It follows that My and Sue’s house is bad style, but it is not bad grammar. On the next page, apropos of sentences with referential-indefinite noun subjects, Hurford writes that whereas in the colloquial Arabic of Egypt raagil figgineena is ungrammatical, its English equivalent A man is in the garden ‘is not felt to be strictly ungrammatical, although we may feel somewhat uneasy about it, and prefer There’s a man in the garden.’ Agreed, but the subject is still a man, just as in the Egyptian Arabic example raagil is still the subject in the grammatical sentence fiih raagil figgineena (where fiih, literally ‘in it’, corresponds neatly to our there). Rather there is a prohibition in Egyptian Arabic on placing an indefinite noun in the initial position that has become normal for the subject. This is plainly not the case in English (A man’s a man for a’ that, A man’s gotta do what a man’s gotta do). A man is in the garden may be awkward, but A man is standing in the garden conveys a more objective sense of menace than the panicky There’s a man standing in the garden, and A man was standing in the garden is not awkward in the least.

Linguists​ are subject to failure of the grammatical instinct, the intuitive perception of a sentence as grammatical or ungrammatical. More serious is a failure of analysis. When Chomsky pronounced They expected that each other would win and John expected that himself would win ungrammatical, others made similar assertions, only for the constructions to be found in actual usage. Since not even the most anti-prescriptive of linguists would seriously maintain that every speaker and writer of English always spoke and wrote grammatically, mere attestation proves nothing about grammaticality. So far as They expected that each other would win is concerned, Chomsky’s intuition is the immediate recognition (such as might have been expected of a far lesser syntactician) that each agrees with the subject and other is the object or other complement; They expected each other to win is correct, but with the noun-clause other has no function. The case stands quite otherwise with John expected that himself (meaning ‘he himself’) would win, for since this construction, though no longer current among good speakers or writers, is recorded by the OED as late as Tennyson, the question of its grammaticality turns on when an older usage, though still perfectly comprehensible, ceases to be felt as an archaism (whether grace or blemish) and becomes a grammatical fault.

But my confidence in linguists had already been shaken on encountering, in one of the theorists under review in Hurford’s book, a hair-raising absurdity that no undergraduate or even schoolchild should have been allowed to perpetrate, namely that all English declarative sentences contain an implicit I tell you that, because an antecedent is required for the reflexive pronoun myself in sentences such as The paper was written by Ann and myself. In this sentence myself is not a reflexive but an intensive pronoun used (wrongly, as many would say) instead of me; nor will it do to say that synchronically there is but one pronoun with two functions, for if intensive myself in the above example, merely by possessing a reflexive function, automatically exercised it, then the reflexive in I cut myself while shaving would also have to function intensively, as if I might have cut someone else instead.

Not all the evidence cited by Hurford is beyond doubt: faced with disagreement between researchers into the Australian language Warlpiri, Hurford is reduced to exclaiming ‘Help!’ (Are we meant to pack our bags and find out for ourselves?) We learn of a comparative comprehension test on an eight-year-old ape with some understanding of English and a two-year-old child; some researchers have argued that the findings exaggerate the ape’s abilities, but few readers can be expected to procure an ape (and at the right point in its training engender a child) in order to replicate the experiment. However, a greater burden is laid on our poor simian: ‘Imagine that Kanzi, somehow miraculously, could be furnished with full phonological representations of input stretches of about seven words. It is extremely unlikely that any grammatical machinery is ready in his head to do a better analysis of them than he can of the partly ordered bags of content word islands that he probably hears now.’ In other words, he probably could not take advantage of a miracle because he is probably not equipped to do so.

That is a venial fault of logic compared with a fault that Hurford exposes in linguists’ discussions of pidgins and creoles. As already noted, he denies that animal communications count as languages since even when they have syntactic organisation they lack semantic compositionality. For the opposite reason – their semantic composition lacks syntactic organisation – pidgins are dismissed as not ‘fully fledged human languages’, apparently meaning languages in which there is a right way and a wrong way, or more than one of either or both, to express one’s meaning, as opposed to a ‘semantic soup’ in which anything goes so long as it is understood.

Since the term ‘pidgin’ was coined not by linguists, but by English-speaking traders whose Chinese counterparts found it difficult to pronounce ‘business’, it is not for linguists to determine what it ought to mean, but only how they themselves will use it. The Neo-Melanesian of Papua New Guinea is called ‘Pidgin’ in local English and describes itself as Tok Pisin (‘Pisin’ is pronounced like the French piscine) even though linguists (including Hurford) call it a creole. But how are pidgins to be distinguished from creoles?

Terminology in this area is applied rather haphazardly. One reliable constant is that the three terms ‘jargon’, ‘pidgin’ and ‘creole’ identify forms of language ranked in that order on a continuum, jargons being the crudest form, with no conventional grammar, and creoles said to be ‘full’ languages. Pidgins are more systematic and conventionalised varieties than jargons, but less so than creoles. In the literature, especially where the term ‘jargon’ is not always used, the term ‘pidgin’ can also imply a lack of any conventional grammatical organisation. Conversely, some languages which clearly have grammatical organisation, and are spoken as native languages, are sometimes referred to as pidgins.

This does not so much open the can of worms as tip them out onto the table. Even if ‘no conventional grammar’ is taken literally, so that the slightest hint of organisation promotes a contact-language from jargon to pidgin, we still need to be told at what point on the continuum the difference in degree between pidgin and creole becomes a difference in kind. But if the term ‘jargon’ is sometimes not used, making the classification bipartite instead of tripartite, the ‘reliable constant’ is neither reliable nor a constant. Worse follows: if, as the absence of a contrasting specification suggests, the ambit of ‘in the literature’ extends from the penultimate sentence of the passage quoted above to the final one, then even some linguists use ‘pidgin’ in the broad sense that covers Tok Pisin.

We learn of an argument between David Gil and J.H. McWhorter about whether Riau Indonesian is a creole: the former demonstrated that the language did not come into being through a creolising process, the latter retorted that since it shared its simplicity of structure with undoubted creoles, it was a creole itself. As Hurford remarks, all too mildly, ‘Without a definition of “creole” independent of the properties creoles are hypothesised to have, any hypothesis runs the risk of circularity’; more precisely, we need a clear distinction between the defining property of a creole and accidents that neither make a language a creole if present or prevent it from being one if absent. Defining properties, in fact, have been variously sought in origin, structure and function: Gil, in declining to call Riau Indonesian a creole because it did not result from creolisation, privileges origin; McWhorter’s claim that even so it is a creole because it behaves like a creole privileges structure. The ‘reliable constant’ defines the genus, contact-language, by origin but the species creole, pidgin and jargon by structure. I have also seen a creole defined as a contact-language that has become some community’s first language, and a pidgin as one that has not, a functional definition that does not imply any greater complexity in the one than in the other. How, in the face of such inconsistency, are we to understand passing references to pidgins and creoles? This is not meant as a criticism of Hurford, who shows himself aware of the problem, but the fact remains that, in the colloquial idiom he affects, linguists need to get their act together.

Send Letters To:

The Editor
London Review of Books,
28 Little Russell Street
London, WC1A 2HN

letters@lrb.co.uk

Please include name, address, and a telephone number.

Read anywhere with the London Review of Books app, available now from the App Store for Apple devices, Google Play for Android devices and Amazon for your Kindle Fire.

Sign up to our newsletter

For highlights from the latest issue, our archive and the blog, as well as news, events and exclusive promotions.

Newsletter Preferences