Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives


SRB Archives

This review appeared in Volume 3 (2) of The Semiotic Review of Books.

The Quest For A Protolanguage

by William R Thurston

Language and Species. By Derek Bickerton. Chicago: University of Chicago Press. 1990 Pp. 297 ISBN 0-226-04610-9.

Bickerton is best known for his controversial work in pidgin-creole studies, particularly Roots of Language (1981) in which he advances the hypothesis that human children are biologically programmed to create real languages (creoles) in social contexts where they are exposed only to pidgin, a code that he considers pregrammatical. Furthermore, according to Bickerton, since the language bioprogram is both genetically determined and identical in all human populations, it follows that the structures of all true creoles should be identical, wherever they emerge. Although his claims have been received as state of the art by the general public, they have been subject to severe criticism among academics (see Goodman 1985). In Language and Species, he furthers the tradition of treating subjects on a grander scale than individual experience usually permits and is bound to generate controversy once again.

Written for a general reader with little background knowledge of either linguistics or biology, the current volume poses the following questions: "How did we get so much more powerful than anything else, and how at the same time did we get our peculiar kind of consciousness (3)"? In answer, Bickerton lays out a recapitulationist model of language origins and speculates about the role of language in the evolution of human mind, consciousness, knowledge, technology, social inequality and violence. Bickerton suggests that the emergence of a unique, omnipotent and omniscient species such as ourselves is the inevitable result of teleological processes of evolution.

For as long as linguistics has been a discipline, discussion on the origin of language has been problematic. The oldest specimens of language are comparable with modern languages; there are no extant languages anywhere in the world with primitive structures that might be considered antecedent to languages of a modern type; and the methods of comparative-historical linguistics result in the reconstruction of ancestral languages that are also recognizably modern in structure. Consequently, at least in the ideology of the scientific paradigms in which linguists have been operating for the last century, speculation on the origin of language is supposed to be taboo.

Nevertheless, as the evidence bearing on the problem has been accumulating, professional dabbling in the question has continued sporadically over the last century, and, once the embellishing is stripped away, Bickerton's recapitulationist model is a fairly standard view.

Bickerton starts with an examination of what he calls 'the continuity paradox'. That is, if language is a product of the same evolutionary processes that have gradually moulded the physical forms of plants and animals, then why is there such a gap between the systems of communication used by animals, including our closest pongid relatives, and human language? The first four chapters deal with the various features that constitute the gulf between human language and animal communication systems. The next three chapters present Bickerton's model of language origins, a recapitulationist model in that it gleans insight on language origins, guided by the principle that the developmental stages of the individual from conception to adulthood might be similar to the stages that the species as a whole has gone through in the course of evolution. To this end, Bickerton compares data from a range of animal communication systems, experiments in teaching American Sign Language (ASL) to pongids, early primary language acquisition in humans, the 'language' of socially isolated children, pidgins and aphasia. His analysis yields an important distinction between protolanguage and true language, the latter possessed only by modern humans. He then maps the stages of language evolution onto the stages of hominid evolution. The final two chapters of the volume are almost totally speculation. Here, Bickerton argues that "... three of the phenomena supposedly unique to humans (mind, consciousness, and our search for knowledge) may derive directly from the possession of a language-based (secondary representational system) with a syntactic processing unit (200)".

As a prelude to the narrative, Bickerton contrasts the structure of language with that of animal communication. He points out that animal communication is holophrastic (which he calls 'holistic'), that is, consisting of individual symbols that have no internal structure and which constitute entire messages that cannot be recombined with other signs to convey more complex messages. In contrast language is characterized by propositionality (which he calls 'predicability', the ability to freely attribute various qualities and behaviours to various entities.

Reality presents us with instances of entities inextricably bound to their attributes. When we encounter cattle, for instance, we see specific cattle of particular colours grazing in a meadow or tied up in a barn. They occur in combination with their various qualities, activities and contexts. One of the things that language follows us to do is abstract CATTLE from BROWN CATTLE GRAZING IN A MEADOW and from WHITE CATTLE TIED UP IN A BARN. Furthermore, it permits us to abstract the qualities and activities from specific entities - BROWN from BROWN CATTLE and BROWN HORSES, and RUN from CATTLE RUNNING and BOYS RUNNING. The ability to analyze reality into entities and attributes underlies the propositionality of language.

Holophrastic systems used by animals permit only single-symbol messages like SNAKE, which is simultaneously a declaration (I PERCEIVE A SNAKE), a warning (WATCH OUT), a statement of emotion (I AM AFRAID), and a command (HEAD FOR THE TREES). Holophrastic communication does not permit one to ask questions (DID SOMEONE SEE A SNAKE? or WHAT DID YOU SEE?), communicate about a snake not physically present (I SAW A SNAKE HERE YESTERDAY), or discuss the attributes of snakes (SNAKES HAVE NO LEGS). In contrast, provided the concepts have been labelled, language makes it possible to predicate any attribute of any entity, even to the point of creating propositions that are impossible in reality (THE UNICORNS ARE BOWLING). In short, language allows us to say something about something else.

The gulf between these two systems of communication is so enormous that many linguists have been led to deny that language could have evolved continuously out of animal communication systems. To escape this paradox, Bickerton breaks the distance between the two systems into a series of smaller steps, each requiring a smaller evolutionary leap.

First, he agues that language should be recognized primarily as a representational system and not primarily as a system of communication (16). He distinguishes two levels of representational systems -- the primary representational system (PRS) and the secondary representational system (SRS). A PRS is characteristic of all animals and provides a neurological model of reality filtered through the sensory and neurological apparatus peculiar to each species. The SRS, possessed only by humans, but latent in a primitive form among pongids, supplies a model of a model of reality one further step removed from reality, the SRS is capable of making greater abstractions and generalizations about reality, because it can conceptualize both from the PRS and from itself. A large, structured lexicon and syntax to organize lexical items into propositions make it possible to think and to acquire knowledge without the equipment of conditioning stimuli that limits the acquisition of knowledge among animals.

The modern human SRS has a large lexicon that is structured such that entries are contrasted with other entries and arranged hierarchically into classes by virtue of shared semantic features. For example, hawks and sparrows are types of birds while hawks and leopards are types of predators. This exhaustive categorization of the known universe results not only in a large vocabulary, but one in which lexical items are conceptually related to one another along various parameters. Assuming that a larger and more-structured lexicon would provide its possessor with a selective survival advantage, Bickerton suggests that the lexical capacity of modern humans could have been achieved by gradual additions to the lexical store. The selective pressure favouring a larger lexicon would place a premium on the memory capacity to store it and, in turn, would act as pressure selecting for increased brain size. At some point, a critical mass would have been reached such that a quantitative difference in brain size would be de facto an improvement in quality.

In conjunction with this sophisticated lexicon, the modern human SRS is also organized around syntax, but the leap from holophrastic communication without grammar to language, which has both propositionality and syntax is too great to have been accomplished in a single leap. To bridge this particularly vast gap, Bickerton suggests protolanguage as the intermediary step.

At this point, further discussion of Bickerton's book requires a basic understanding of propostional structures. In a sentence such as WALLY KILLED THE DRAGON WITH A SWORD, the central event is coded in the verb KILL and the arguments of the verb, concepts coded by the nouns, participate in the event in different thematic roles. For instance, WALLY is the agent (the doer), DRAGON is the patient (the undergoer) and SWORD is the instrument (the tool used for the action). The remaining parts of the sentence (-ED, THE, WITH and A) are functors. In addition to word order, they glue the proposition together grammatically and pragmatically. That is, they tell the hearer that the event took place in the past, that the dragon is the same dragon that the speaker expected the hearer to have in mind at the time of the utterance, and that the identity of the instrument was expected to be unknown or unimportant to the hearer. One additional feature of language noted by Bickerton is that verbs are subcategorized by the number of arguments that must accompany them. For instance, while DANCE requires only an agent, GIVE requires an agent (the donor), a patient (the thing given) and a goal (recipient). Translation and second language learning are possible because all languages have words coding the same events with the same obligatory arguments. Propositional structure is among the most important features distinguishing language from other systems of communication.

Whereas holophrastic communication permits only individual symbols, one at a time, protolanguage permits a primitive kind of propostionality by conjoining a maximum of two symbols, like beads on a string, into a single utterance. Without functors, one is left to infer from the non-linguistic context of an utterance who did what with what and to whom. For example, lexical items such as DOG, MARY and BITE, one could construct utterances in protolanguage such as DOG BITE or BITE MARY, only two items being allowed, but one would be left to deduce that once women do not normally bite dogs, DOG is the agent (the biter) and MARY is the patient (the bitee). In contrast with protolanguage, the grammar of true language would allow one to relate both the usual situation (THE DOG BIT MARY) and the unusual situation (MARY BIT THE DOG) without ambiguity. Furthermore, in languages with passive constructions (MARY WAS BITTEN BY A DOG), one can alter the order of constituent lexemes from agent-verb-patient to patient-verb-agent without altering the propositional structure. Even more remarkable is the fact that syntax and functions permit one to build a proposition with more than two lexical items (THE DOG BIT MARY ON THE LEG YESTERDAY) and to embed a large number of propositions into a single utterance (THE MANGY OLD DOG THAT WALLY BOUGHT WHILE HE WAS SELLING PIGS IN TOWN LAST WEEK BIT MARY ON THE LEG YESTERDAY, SO RALPH DROVE HER OVER TO THAT NEW HOSPITAL ON THE OTHER SIDE OF THE RIVER, AND THEY GAVE HER A SHOT). The embedded propositions in this last example include: THE DOG IS MANGY; THE DOG IS OLD; WALLY BOUGHT THE DOG; WALLY SOLD PIGS IN TOWN LAST WEEK; and so on. While DOG features in several of the component propositions, it occurs only once in the grammaticalized utterance. Bickerton points out that syntax and functors, including pronouns, allow one to recover the propositonal structure from such complex sentences unambiguously and effortlessly in most cases.

Bickerton illustrates the distinction between protolanguage and language with a telling comparison of utterances recorded from a child in the process of acquiring language with utterances recorded from a chimpanzee in the most advanced stages of its training in ASL. Before acquiring true language, children pass through two fairly discrete non-linguistic stages: holophrases (the 1-word stage characterized by single word utterances) and protolanguage (the 2-word stage). From there, even though they make errors, especially in the irregular morphology peculiar to individual languages, they leap, without recourse to an intermediary stage, directly to fullblown language with its unambiguous propositionality and embedding. This is possible because the syntactic structures that allow one to build a noun phrase, for instance, are identical to the structures that allow one to build any other kind of phrase, including a complex sentence. Although he undervalues the accomplishments of trained pongids somewhat, Bickerton quite properly recognizes that their 'linguistic' capacities are virtually identical to those of children under two -- that is, both are making use of protolanguage rather than true language.

Bickerton suggests that both protolanguage and language are evidence of secondary representational systems, and that both allow predication (saying something about something else but, whereas protolanguage permits only two items to be joined in an utterance and has no device for marking the thematic roles of the constituents, the SRS underlying language is much more powerful because its syntax unambiguously (in most cases) indicates who does what with what and to whom. At this point in the narrative, Bickerton's analysis has resulted in three types of communication: (l)holophrasis, used by all animals, by human infants in the easiest stages of language acquisition, and by adult humans alongside language, (2) protolanguage, used by human children under two, by pongids trained in ASL, and by human adults under severe stress; and (3) true language, used only by modern humans over two. The leap from protolanguage to language requires only the acquisition of syntax, and since syntax operates according to only a single principle, once that principle is acquired for any grammatical construction, syntax has been acquired in its entirety. This diminishes the distance between protolanguage and language to a single, though crucial, step.

The next stage of Bickerton's argument involves correlating holophrasis, protolanguage and language with the various stages of hominid evolution. The fact that pongids have a latent capacity for protolanguage suggests that it was also latent in our common ancestors. The ratio of brain size to body size plotted over time, the advances in toolkits, and the fact that H. habilis was limited to Africa while H. erectus ranged across the entire Old World and built fire, all lead him to associate a fully-developed protolanguage with H. erectus rather H. habilis. The major force of the argument however seems to rest on there being no intermediate stage between protolanguage and language. Since protolanguage precedes language and H. erectus precedes H. sapiens, then protolanguage belongs to H. erectus. The logic is reasonable, but not particularly secure because there are no data to further support or refute the hypothesis.

In the case of H. erectus, there is little change in physical morphology and artifact production over a long period of time; then, with the emergence of H. sapiens sapiens, there is a sudden explosion, not only in toolkit diversity, but also in the arts. Bickerton takes this as evidence for a qualitative improvement in the conceptual power of the SRS of H. sapiens over that of H. erectus and, consequently, as evidence for the use of language. Since the earliest H. sapiens finds are in Africa and date from around 200 kya (thousand years ago) Bickerton places the beginning of language at this point. He attributes the lack of sophisticated tools from this earlier period to the use of perishable materials such as wood for tool manufacture. Again, although this seems reasonable, the assertion cannot be supported with tangible data.

To soften the abrupt transition from protolanguage to language somewhat, Bickerton suggests that, in the later H. erectus period "... protolanguage did develop a set of protogrammatical terms, that is, meaningful if somewhat abstract units that may have included some or all of the following: negators, question words, pronouns, relative-time markers, quantifers, modal auxilaries, and particles indicating location (185)". Support for this assertion is suggested by finding these same items in the late protolanguage of children, signing pongids and speakers of pidgin (as he defines them).

Having established a possible sequence of steps in the evolution of language, Bickerton continues his narrative by speculating about the other unique characteristics of humans made possible by language -- mind, consciousness and the search for knowledge. For him, the mind is roughly the subjective experience of autonomous identity. It is modular with syntax as the central organizing module to structure the lexicon, concepts, sensory input, thinking and speech. While non-humans have only a PRS, which is constrained to some extent by reality, humans have a particularly-powerful SAS which makes it possible to represent whatever "...seems to suit us, at any given moment, to choose to believe is in the world (226)". In not exactly these words, Bickerton suggests that by being at one more remove from reality, the SRS makes imagination, religion and science possible and that these are what enable human omniscience and omnipotence. It would not be bending his words too much to say that in answer to the questions he poses at the beginning of the book, he asserts that the SRS and syntax have made humans into gods.

This is not an academic work. Bickerton frames his narrative within a naive and archaic concept of biological evolution, unsupported assertions, and fanciful speculations presented as facts. Supposedly written for a broad general audience, proper rules of citation are not followed because they "would only distract the reader and disrupt the flow of the argument (6)." For those who are more interested in his sources than distracted by citations, this is partly remedied by a collection of notes at the back of the volume. At the same time, I expect that a reader lacking specialized background would balk at some passages. For instance, without previous exposure to recent linguistic theory, I doubt that most educated readers would be able to make sense out of his discussion of X-bar theory, where he asserts that all syntax operates according to a single principle, an assertion that is crucial to narrowing the gap between protolanguage and language.

Other passages, not even relevant to the topic, are designed to allienate some of his more likely readers, namely some linguists and most anthropologists, readers for whom the subject is of considerable professional interest. For instance, he asserts that "most of what we know about language has been learned in the last three decades (5)." This amounts to nothing more than a gratuitous and undeserved dismissal of the pre-1957 contributions to linguistics made by Leonard Bloomfield, Franz Boas, Joseph Greenberg, Charles Hocke, Antoine Meillet, Kenneth Pike, John Reinecke, Edward Sapir, Ferdinand de Saussure and Nikolai Trubetzkoy, to name but a few. His revisionist history of the discipline, in which there is nothing of linguistic worth before Chomsky, "arguably the Newton of our field (5)", is a plain misrepresentation of the facts. Expanding on the insult he states as fact that: "Some linguists will tell you that the formal structure of language is very important. Others will tell you that it is relatively unimportant (8)." The latter class of linguists, dubbed "antiformalists", seems to be a category that Bickerton's SRS has chosen to believe into existence, but why it suits him to set up this contrast is never made clear in the text.

Bickerton's model of evolution is naively teleological. The Victorian picture that he presents is one of the inevitable progress toward omniscient, omnipresent, omnipotent humans as the outcome of evolution. Taking his model to the level of speculation about extraterrestial life, he states that: " ...wherever life evolves, it will sooner or later produce creatures that think and communicate...more or less as we do (253)". This contrasts markedly with Gould's (1989) model, in which the chances of surviving extinction are extremely small and the evolution of a creature like Homo sapiens is a major fiuke. Whereas Gould urges us to cherish this life because there will likely never be another opportunity for it to evolve, Bickerton's logic urges us not to worry about devastating the planet, because evolution inevitably leads to humans, and so, if we destroy ourselves this time around we will emerge from the ashes once again. This view is entirely consistent with the quasi-theological overtones of the book.

If one has the background to see beyond the polemic nastiness and the general naiveté on detail that form the matrix of Bickerton's narrative, it is not really a bad story, but it is not entirely original either. At least in my experience, introductory classes in linguistics and anthropology have been using similar recapitulationist models for the origin of language since the early 1970's.

If you read everything printed on language origins, then, of course, Bickerton (1990) is already on your list but if you plan to read selectively on the subject there are other works you should definitely consult first. Philip Lieberman (1984,1991) has two readily-available volumes in print.The earlier work is larger, detailed, rigorously argued, and, depending on one's experience, difficult to plough through in places. The 1991 contribution is smaller and aimed at a general audience, but it is properly referenced. It contains much of the same information as the 1984 text, but condensed a great deal. Unfortunately, the 1991 book, like Bickerton, but to a lesser degree, devolves into a somewhat shaky speculation on altruism that detracts from an otherwise excellent work.

Both Bickerton (1990) and Lieberman (1984, 1991) however, fail to account for another important factor in the debate on language origins, the human capacity to express language through modalities other than speech, especially sign language, but also writing, braille and hand-to-hand finger spelling. Although speech is undeniably the dominant modality in modern humans, no account of language origins can be complete without also addressing the multiple possible media through which language can be expressed.

If language were tied specifically to speech, then neither sign language nor writing would be possible, and without writing, the accumulation of knowledge would be limited to what can be preserved within an oral tradition. In his 1973 work, Hewes revives the hypothesis that, in the course of hominid evolution, language passed through a stage during which the dominant modality was sign rather than speech. Although neurolinguistic support for this hypothesis continues to accumulate (Dingwall 1988:290-291), both Bickerton and Lieberman dismiss it far too lightly, Bickerton because it would complicate his unilinear narrative, and Lieberman because he is married to the conviction that the evolution of language was constrained by the evolution of the modern vocal tract. On this point, Bickerton's obsession with language as an SRS balances Lieberman nicely.

As either a starting point for reading on language origins or for a good overview of the subject I would recommend Dingwall (1988). He gives a list of recent authors, lays out various models of language origins and generally cites other studies for following up on points that he treats in a cursory manner. Although his emphasis is on the biological aspects of language, he gives fairer treatment of Lieberman and Hewes within a sophisticated context of evolutionary theory than does Bickerton.


Bickedon, Derek (1981> Roots of Language. Ann Arbor: Karoma.

-- (1990). Language and the Species. Chicago: University of Chicago Press.

Dingwall, William Orr (1988). "The evolution of human communicative behavior." In Linguistics: The Cambridge Survey Volume III Language, Psychological and Biological Aspects. Newmeyer, Frederick J. (ed.). Cambridge: Cambridge University Press: 274-313.

Goodman, Mords (1985) "Review of Bickerton (1981)" International Journal of American Linguistics 51.1:109-137.

Gould, Stephen Jay (1989) Wonderful Life: the Burgess Shale and the nature of History. New York: W.W.Norton.

Hewes, Gordon W. (1973) "Primate communication and the cultural origin of language." Current Anthropology 14 1-2:5-24.

Lieberman, Philip (1984) The Biology of Language. Cambridge, Massachusetts: Harvard University Press.

--- (1991). Uniquely Human; The Evolution of Speech, Thought and Selfless Behaviour. Cambridge, Massachusetts: Harvard University Press.

William Thurston was University Research Fellow and Assistant Professor of Anthropology at McMaster University from 1987 to 1992, and is currently Professor of Anthropology at Okanagan University College. He is author of Processes of Change in the Languages of North-Western New Britian (1987), "How exoteric languages build a lexicon esoterogeny in West New Britain"(1989), and several descriptive works on the languages and belief systems of New Britain (forthcoming).

Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives