pub bickerton on chomsky

Derek Bickerton. Chomsky: Between a Stony Brook and Hard Place. A report on Chomsky at the 2005 Stony Brook symposium.

http://www.derekbickerton.com/blog/_archives/2005/10/24/1320752.html

CHOMSKY: BETWEEN A STONY BROOK AND A HARD PLACE

by Derek Bickerton on Mon 24 Oct 2005 10:10 PM PDT

On October 14, 2005, Chomsky disembarked on Long Island for one of the few conferences he has attended in

the last several decades: the Morris Symposium on the Evolution of Language at S.U.N.Y Stony Brook. He

arrived too late for any of the presentations given by other scholars on that date, gave his public lecture, [see

page 6 below for the abstract] gave his conference presentation at the commencement of the next morning’s

session, and, despite the fact that all of the morning’s speakers and commentators were expected to show up for a

general discussion at the end of that session, left immediately for the ferry back without having attended a single

talk by another speaker. For me, and for numerous others who attended the symposium, this showed a lack of

respect for everyone involved. It spelled out in unmistakable terms his indifference to anything anyone else

might say or think and his unshakable certainty that, since he was manifestly right, it would be a waste of time to

interact with any of the hoi polloi in the muddy trenches of language evolution.

Some may say, “Oh, he’s such an important man, he has no many important things to do, we should be grateful

that he could spare even a little time to be with us.” That is not so. If he’s so busy, he shouldn’t have come to the

symposium at all. A symposium is not a forum for making ex cathedra pronouncements. Originally “a convivial

meeting for drinking, music, and intellectual discussion among the ancient Greeks”, it is currently defined as “a

meeting or conference for discussion of a topic, especially one in which the participants form an audience and

make presentations” (emphasis added). In other words, rather than somewhere to talk and run it’s a place where

you interact with other researchers, discuss topics and exchange ideas.

Since Chomsky missed the opportunity of learning something from his colleagues, I will show how he might

have benefited from a fuller attendance. Among other things, he might have learned the reasons why his current

approach to language evolution is not merely wrong, but logically impossible.

He summarized that approach in the following paragraph (I quote the written version on the symposium

website):

Putting these thoughts together, we can suggest what seems to be the simplest speculation about

the evolution of language. In some small group from which we all descend, a rewiring of the

brain took place yielding the operation of unbounded Merge, applying to concepts with

properties of the kind I mentioned. Third factor principles enter to yield an infinite array of

structured expressions with interpretations of the kind illustrated: duality of semantics,

operator-variable constructions, unpronounced elements with substantial consequences for

interpretation and thought, etc. The individual so rewired had many advantages: capacities for

complex thought, planning, interpretation, and so on. The capacity is transmitted to offspring,

coming to predominate. At that stage, there would be an advantage to externalization, so the

capacity might come to be linked as a secondary process to the sensorimotor system for

externalization and interaction, including communication – a special case, at least if we invest



the term “communication” with some meaning. It is not easy to imagine an account of human

evolution that does not assume at least this much, in one or another form.

Let’s look at this in detail:

In some small group from which we all descend…

This is typical of the abstract approach common to many, mostly linguists, beside Chomsky. No evolutionary

biologist would dream of approaching anything in this way. The evolution of language is a real process that

actually happened at some specific time(s) and place(s) in the past. The idea that one could tackle the problem

without even considering the where, when, why and how of this process strikes most non-linguists (and at least

this linguist) as simply bizarre.

What Chomsky is doing is simply transferring, from generative acquisition studies, the idealization of

instantaneity. Now this is legitimate in acquisition, precisely because the language mechanism already forms part

of our biological heritage and is thus, in some real sense, “there” during the babbling, one-word, two-word and

“telegraphic speech” stages, consequently there is no sense in assuming, as a behaviorist might, that these stages

somehow ‘drive’ the acquisition process and form essential pre-requisites for its completion (indeed, anecdotal

evidence suggests that a few children skip these stages entirely, remaining mute until they can utter complete

sentences). But the circumstances of language evolution are totally different from those of language acquisition.

At this stage of development, the language mechanism did NOT form part of our biological heritage, therefore it

is at least a reasonable assumption that the precise stages through which language evolution occurred DID drive

the process and DID influence the nature of the final product.

…a rewiring of the brain took place yielding the operation of unbounded Merge…

Abracadabra! This is a piece of what Dan Dennett calls “figment” (as in “figment of the imagination”). Makes it

sound like a single event (“a wedding took place”) or like something you might get done at Radio Shack. What

rewiring? How? When? Why? Chomsky would doubtless respond that we have no idea what the answers might

be, so it is useless to speculate. If so, he’s simply wrong. Speculation is the horse that drags the chariot of theory.

If we don’t speculate, we’ll never get a hypothesis to test, and thus never be able to rule out any of the large

number of possible answers that presently face us. In fact, Chomsky's claim is not a scientific proposition, but

rather a mantra to be uttered by true believers when their faith is challenged. Scientific propositions can be

decomposed into units about which constructive thought is possible; mantras can’t.

…applying to concepts with properties of the kind I mentioned.

So let’s look at what kind he mentioned (all emphasis is mine):

Comparative work on the second interface, systems of thought, is of course much harder.

There do, however, seem to be some critical differences between human conceptual systems

and symbolic systems of other animals. Even the simplest words and concepts of human

language and thought lack the relation to mind-independent entities that has been reported

for animal communication: representational systems based on a one-one relation between



mind/brain processes and "an aspect of the environment to which these processes adapt the

animal's behavior," to quote Randy Gallistel’s introduction to a volume of papers on animal

communication. The symbols of human language and thought are quite different…What we

take to be a river, or a person, or a tree, or water turns out not to be identifiable as a physical

construct of some kind. Rather, these are creations of what 17th century investigators called

the “cognoscitive powers” that provide us with rich means to refer to the outside world from

certain perspectives, but are individuated by mental operations that cannot be reduced to a

“peculiar nature belonging” to the thing we are talking about, as Hume summarized a

century of inquiry...In this regard, internal conceptual symbols are like the phonetic units of

mental representations, such as the syllable [ba]; every particular act externalizing this

mental entity yields a mind-independent entity, but there is no mind-independent construct

that corresponds to the syllable. Words and concepts appear to be similar in this regard, even

the simplest of them. These properties seem to be unique to human language and thought.

This is where Chomsky gets into deep trouble. .

He has claimed that “concepts with the properties of the kind I mentioned” were what recursion (“unbounded

Merge”) originally applied to. Those properties, as he quite correctly states, are precisely the properties that

distinguish human concepts from the concepts of other species—they refer to mental constructs rather than

natural objects. But if concepts with such properties are unique to human language, HOW COULD

RECURSION HAVE APPLIED TO THEM WHEN LANGUAGE DID NOT YET EXIST?

Either those concepts (and probably the words with which they were linked) already existed, implying some kind

of system intermediate between animal communication and true language, or recursion could not (on Chomsky’s

own admission) have applied to anything. Since syntactic language now exists, it is a logically unavoidable

conclusion that there must have been some kind of protolanguage before recursion.

Third factor principles…

Say WHAT?!

…enter to yield an infinite array of structured expressions with interpretations of the kind

illustrated: duality of semantics, operator-variable constructions, unpronounced elements with

substantial consequences for interpretation and thought, etc.

Since these things now exist, it is trivially obvious they must have come into existence somehow. Chomsky’s

observation merely rephrases this simple fact in a more oracular form.

The individual so rewired...

Bionic Man?

…had many advantages: capacities for complex thought, planning, interpretation, and so on.

The capacity is transmitted to offspring, coming to predominate. At that stage, there would be an

advantage to externalization, so the capacity might come to be linked as a secondary process to



the sensorimotor system for externalization and interaction, including communication – a

special case, at least if we invest the term “communication” with some meaning.

Now Chomsky isn’t the first to suggest that language couldn’t have arisen from prior communication systems,

but could only have come about through the development of the human mind. I did this fifteen years ago in a

book called Language and Species (cited as Species and Language in Hauser, Chomsky and Fitch 2002, and,

adding injury to insult, attributing to me a view I never held and would at any time have flatly rejected--great

scholarship, guys!). However there is a twist to this. Although language couldn’t have come into existence

without the pre-existence of a rich conceptual system, it had to be kick-started by something, and the most

obvious candidate for that “something” is a pressure to communicate in a novel way when failure to

communicate in such a way would have prejudiced survival. Only some strong force like this could have forced

our ancestors out of the typical animal communication mould, which is exactly as Chomsky described it in the

paragraph cited above, but which has been adequate for the needs of virtually every species that has ever existed.

What’s the alternative? Chomsky’s is to suppose that structured language came into existence as a form of

mental computation well in advance of any “externalization”. What selective pressure would have brought this

about? If none, how plausible is it that such a system, complexified beyond measure by all those “third factor

principles”, would have come into existence in the hominid mind by mutation, ‘laws of form”, or any other non-

selectional agency? What use would planning capacity have been if it wasn’t possible to involve anyone else in

your plan? What use would interpretation have been if there were no utterances to interpret? And are we to

suppose that human ancestors began from ground zero speaking perfectly well-formed sentences like the alleged

first words of the three-year-old Lord Macaulay (“Thank you, madam, the agony has sensibly abated”)?

Of course this is absurd, because--supposing for the moment that recursion could indeed apply to symbolic

concepts--how would the “externalization” of internalized sentences ensure that everyone would have chosen the

same word for the same concept? Of course it could have done nothing of the kind, so we are left with the puzzle

of how a species equipped with all the bells and whistles of syntactic language could not know how to refer to

concepts as basic as “dog”, “rock”, “tree” or “go” in such a way as to make the simplest utterance

comprehensible, whether for “communicative” or other unnamed purposes.

Note that if we assume a protolanguage we are saved from all these absurdities, since language and thought

would have co-evolved in a beneficent spiral, each driving the other.

It is not easy to imagine an account of human evolution that does not assume at least this much,

in one or another form.

Well, actually it’s quite easy, as the foregoing would suggest, to imagine an account that, while inevitably

speculative to some degree (what account isn’t, at this stage?) not only differs from Chomsky’s but avoids the

logical flaw that effectively invalidates it.

It will be interesting to see how, if at all, Chomsky will respond to this. His presence at Stony Brook was due to

his awareness that language evolution has become central to cognitive and behavioral studies: it is the high

ground and if he is to maintain his intellectual eminence he has to take it. However, language evolution is a



multi-disciplinary field with plenty of give and take; anyone who fails, as he so signally did at Stony Brook, to

engage in such give and take is quickly going to be marginalized and treated as irrelevant.

It’s his call now.



ABSTRACT OF CHOMSKY’S PRESENTATION TO THE STONY BROOK SYMPOSIUM

Some simple evo-devo theses: how true might they be for language?

Study of evolution of some system is feasible only to the extent that its nature is understood. That seems close

to truism. A sensible approach is to begin with properties of the system that are understood with some

confidence and seek to determine how they may have evolved, then turning to less obvious properties and

asking what additional problems they pose for inquiry into evolution. I’ll try to outline such a course for

language, keeping to a sketch of general directions.

I will also mention some analogies between what some biologists call “the Evo Devo revolution” and ideas that

have been lurking in the background of “biolinguistics” since its origins about half a century ago, and that have

been pursued more intensively in recent years. The analogies have been suggestive in the past, and might

turn out to be more than that in the years ahead.

The term “biolinguistics” was proposed in 1974 by Massimo Piattelli as the topic for an international

conference he organized, held at MIT, bringing together evolutionary biologists, neuroscientists, linguists, and

others concerned with language and biology, one of many such initiatives. A primary focus of the discussions

was the extent to which apparent principles of language are unique to this cognitive system, plainly one of

“the basic questions to be asked from the biological point of view” and crucial for the study of development of

language in the individual and its evolution in the species.

The term “language” as used in this context means internal language, sometimes called “I-language,” the

computational system of the mind/brain that generates structured expressions, each of which can be taken to

be a set of instructions for the interface systems within which the faculty of language is embedded. There are

at least two such interfaces: the conceptual/intensional or semantic systems that use linguistic expressions for

thought and for organizing action, and the sensorimotor systems that externalize expressions in production

and assign them to sensory data in perception. Languages so construed are particular instantiations of some

genetically-determined format, which we can call the language faculty, adapting a traditional term to this

framework. Certain configurations are possible human languages, others are not, and a primary concern of

any theory of human language is to establish the distinction between the two categories.

At the time of the 1974 conference, it seemed that the language faculty must be rich, highly structured, and

substantially unique to this cognitive system. In particular, that conclusion followed from considerations of

language acquisition. The only plausible idea seemed to be that language acquisition is rather like theory

construction. Somehow, the child reflexively categorizes certain sensory data as linguistic, not a trivial

achievement in itself, and then uses the constructed linguistic experience as evidence for a theory that

generates an infinite variety of expressions, each of which contains the information about sound, meaning, and

structure that is relevant for the myriad varieties of language use.

To give a few of the early illustrations for concreteneness, the internal theory that those of us in this room more

or less share determines that the sentence “Mary saw the man walking to the bus station” is three-ways

ambiguous, but the ambiguities are resolved if we ask “which bus station did Mary see the man walking to?”

The explanation appears to rely on computationally plausible principles of minimal search, for which there is a

good deal of independent evidence. The phrase “which bus station” raises from the position in which its

semantic role is determined and is reinterpreted as an operator taking scope over a variable in its original

position, so the sentence means, roughly, “for which x, x a bus station, Mary saw the man walking to x”; the

variable is silent in the phonetic output, but must be there for interpretation. Only one of the underlying

interpretations permits the operation, by virtue of the minimal search conditions, so the ambiguity is resolved

in the interrogative.



To take a second example, consider the sentence “John ate an apple.” We can omit “an apple,” yielding “John

ate,” which we understand to mean “John ate something or other, unspecified. Now consider “John was too

angry to talk to Bill.” We can omit “Bill,” yielding “John was too angry to talk to,” which, by analogy to “John

ate,” would be expected to mean that John was so angry that he wouldn’t talk to someone or other. But it

doesn’t mean that: rather, that John is so angry that someone or other won’t talk to him, John. In this case,

the explanation lies in the fact that the phrase “too angry to talk to,” with the object missing, actually has an

operator-variable structure based on movement of a phonetically invisible operator meeting the same

conditions as in the “bus station” example. In this case, the operator has no content and is silent, yielding an

open sentence with a free variable, hence a predicate. Again, there is substantial independent evidence

supporting the conclusion, for a variety of constructions.

In both cases, then, general computational principles yield the required interpretations as an operator-variable

construction, with the variable unpronounced in both cases and the operator unpronounced in one. The surface

forms in themselves tell us little about the interpretations. That is a common situation. For such reasons, it has

been understood from the earliest work in generative grammar that determination of the grammatical status of

a sentence, or efforts to approximate what appears in a corpus, are of only marginal interest. The language

that every person quickly masters relies on inner resources to generate internal expressions that yield

information of the kind just illustrated, only very partially revealed in a corpus of data, no matter how huge.

Even the most elementary considerations yield the same conclusions. In the earliest work in generative

grammar 50 years ago, it was assumed that phonetic units can be identified in a corpus, and that words can be

detected by study of transitional probabilities (which, surprisingly, turns out to be false, recent work has

shown). We also proposed methods with an information-theoretic flavor to assign such words to categories. But

it was evident that even the simplest lexical items raise fundamental problems for analytic procedures of

segmentation, classification, statistical analysis, and the like. A lexical item is identified by phonological

elements that determine its sound along with morphological elements that determine its meaning. But neither

the phonological nor morphological elements have the “beads-on-a-string” property required for computational

analysis of a corpus. Furthermore, rather as in the case of the sentences I gave as examples, even the simplest

words in many languages have phonological and morphological elements that are silent. The elements that

constitute lexical items find their place in the generative procedures that yield the expressions, but cannot be

detected in the physical signal. For that reason, it seemed – and seems

– that the language acquired must have the basic properties of an internalized explanatory theory. These

are elementary and quite general properties that any account of evolution of language must deal with.

Quite generally, construction of theories must be guided by what Charles Sanders Peirce a century ago called

an “abductive principle,” genetically determined, which “puts a limit upon admissible hypotheses,” so that the

mind is capable of “imagining correct theories of some kind” and discarding infinitely many others consistent

with the evidence. For language development, the format that limits admissible hypotheses must, furthermore,

be highly restrictive, given the empirical facts of rapidity of acquisition and convergence among individuals. The

conclusions about the specificity and richness of the language faculty seemed to follow directly. Plainly such

conclusions pose serious barriers to inquiry into how the faculty might have evolved, matters discussed

repeatedly, and inconclusively, at the 1974 conference.

A few years later, a new approach crystallized that suggested ways in which these barriers might be overcome.

This “Principles and Parameters” (P&P) approach was based on the idea that the format consists of invariant

principles and a “switch-box” of parameters that can be set to one or another value on the basis of fairly

elementary experience. A choice of parameter settings determines a language. The approach largely emerged

from intensive study of a range of languages, but it was also suggested by an analogy to early evo-devo

discoveries, specifically François Jacob’s account of how slight changes in regulatory mechanisms can yield

great superficial differences – a butterfly or an elephant, and so on. The model seemed natural for language as



well: slight changes in parameter settings might yield superficial variety, through interaction of invariant

principles with parameter choices. The approach has been pursued with considerable success, with many

improvements and revisions along the way. One illustration is Mark Baker’s demonstration, in his book Atoms of

Language, that languages that appear on the surface to be radically different, such as Mohawk and English,

turn out to be remarkably similar when we abstract from the effects of a few choices of values for parameters

with a hierarchic organization that he argues to be universal, hence the outcome of evolution of language.

The approach stimulated highly productive investigation of languages of great typological variety, and also

reinvigorated neighboring fields, particularly the study of language acquisition, reframed as inquiry into setting

of parameters in the early years of life, with very fruitful results. The approach also provided a new

perspective to undermine the long-standing though implausible belief that languages can “differ from each

other without limit and in unpredictable ways,” in the words of a prominent theoretician summarizing received

opinion in the mid-1950s, with some exaggeration but not too much. Similar views were familiar in biology as

well. Thus until quite recently it appeared that variability of organisms is so great as to constitute “a near

infinitude of particulars which have to be sorted out case by case” (molecular biologist Gunther Stent),

conceptions now significantly modified by evo-devo discoveries about organizing principles, deep homologies,

and conservation of fundamental mechanisms of development, perhaps most famously hox genes.

The P&P approach also removed a major conceptual barrier to the study of evolution of language. With the

divorce of principles of language from acquisition, it no longer follows that the format that “limits admissible

hypotheses” must be rich and highly structured to satisfy the empirical conditions of language acquisition. That

might turn out to be the case, but it is no longer an apparent conceptual necessity.

Here too research programs within linguistics had certain analogies to the evo-devo revolution, including the

discovery, quoting Jacob and others, that “the rules controlling embryonic development” interact with other

physical conditions “to restrict possible changes of structures and functions” in evolutionary development,

providing “architectural constraints” that “limit adaptive scope and channel evolutionary patterns.” Evidently,

development of language in the individual must involve three factors: First, genetic endowment, which sets

limits on the languages attained; second, external data, which selects one or another language within a narrow

range; and third, principles not specific to the language faculty. The third factor principles have the flavor of the

architectural and developmental constraints that enter into all facets of growth and evolution. Among these are

principles of efficient computation, such as those I mentioned. These would be expected to be of particular

significance for generative systems such as the internalized language. Insofar as the third factor can be shown

to be operative in the design of the language faculty, the task of accounting for its evolution is correspondingly

eased.

Recent inquiry into these topics has come to be called “the minimalist program,” but it should be noted that the

program is both traditional and pretty much theory neutral. The serious study of language has always sought to

discover what is distinctive to this cognitive faculty, hence implicitly abstracting from third factor effects. And

whatever one’s beliefs about design of language may be, the questions of the minimalist research program

arise.

Let’s turn to the approach I suggested at the outset: beginning with the properties of language that are

understood with some confidence. The internal language, again, is a computational system that generates

infinitely many internal expressions, each an array of instructions to the interface systems, sensorimotor and

semantic. To the extent that third factor conditions function, the language will be efficiently designed to

satisfy conditions imposed at the interface. We can regard an account of some linguistic phenomena as

principled insofar as it derives them by efficient computation satisfying interface conditions.

Any generative system, natural or invented, is based on an operation that takes structures already formed and



combines them into a new structure. Call it Merge. Operating without bounds, Merge yields a discrete infinity of

structured expressions. The only alternatives are, effectively, notational variants. Hence unbounded Merge

must be part of the genetic component of the language faculty, a product of the evolution of this “cognitive

organ.”

Notice that the conclusion holds whether such recursive generation is unique to the language faculty or found

elsewhere. If the latter, there still must be a genetic instruction to use unbounded Merge to form structured

linguistic expressions satisfying the interface conditions. Nonetheless, it is interesting to ask whether this

operation is language-specific. We know that it is not. The classic illustration is the system of natural numbers.

That brings up a problem posed by Alfred Russell Wallace 125 years ago: in his words, the “gigantic

development of the mathematical capacity is wholly unexplained by the theory of natural selection, and must

be due to some altogether distinct cause,” if only because it remained unused. One possibility is that it is

derivative from language. It is not hard to show that if the lexicon is reduced to a single element, then

unbounded Merge will yield arithmetic. Speculations about the origin of the mathematical capacity as an

abstraction from linguistic operations are familiar, as are criticisms, including apparent dissociation with lesions

and diversity of localization. The significance of such phenomena, however, is far from clear; they relate to use

of the capacity, not its possession. For similar reasons, dissociations do not show that the capacity to read is

not parasitic on the language faculty.

Suppose the single item in the lexicon is a complex object, say some visual array. Then Merge will yield a

discrete infinity of visual patterns, but this is simply a special case of arithmetic. The same would be true if we

add a recursive operation – another instance of unbounded Merge – to form an infinite lexicon, on the model of

some actual (if rather trivial) lexical rules of natural language. This is still just a more elaborate form of

arithmetic, raising no new issue. Similar questions might be asked about the planning systems investigated by

George Miller and associates 45 years ago. If these and other cases fall under the same general rubric, then

unbounded Merge is not only a genetically determined property of language, but also unique to it.

Either way, evolution of language required some innovation to provide instructions for unbounded Merge,

forming structured expressions accessible to the two interface systems. There are many proposals involving

precursors with Merge bounded: an operation to form two-word expressions from single words to reduce

memory load for the lexicon, then another operation to form three-word expressions, etc. There is no evidence

for this, and no obvious rationale either, since it is still necessary to assume that at some point unbounded

Merge appears. Hence the assumption of earlier stages seems superfluous. The same issue arises in language

acquisition. The modern study of the topic began with the assumption that the child passes through a two-word

state, etc. Again the assumption lacks a rationale, because at some point unbounded Merge must appear.

Hence the capacity must have been there all along even if it only comes to function at some later stage. There

does appear to be evidence for that conclusion: namely, observation of what children produce. But that carries

little weight. It has been shown long ago that what children understand at the early stages far exceeds what

they produce, and is quite different in character as well. At the telegraphic speech stage of production, for

example, children understand normal speech with the function words in the right places but are baffled by

telegraphic speech, as shown by experimental work of Lila Gleitman and associates 40 years ago. Hence for

both evolution and development, there seems to be little reason to suppose that there were precursors to

unbounded Merge.

Suppose X is merged to Y. Maximally efficient computation will leave X and Y unchanged (the No-Tampering

Condition). Plainly, either X is external to Y or is part of Y: external and internal Merge, respectively, the latter

sometimes called Move. A well-designed language, lacking arbitrary stipulations, will have both cases. Internal

Merge yields the familiar phenomenon of displacement, as in the cases I gave earlier: say a question of the

form “what did John see,” with two occurrences of “what,” one pronounced in sentence-initial position, the

other deleted by phonological rules mapping to the sensorimotor interface. The full internal expression is

interpreted at the semantic interface as an operator-variable construction, with “what” given the same



semantic role it has when it is not displaced, as in “who saw what.” In a well-designed language, the two kinds

of Merge will have different interface properties. That appears to be true. They correlate with the well-known

duality of semantics that has been studied from various points of view. External Merge yields argument

structure: agent, patient, goal, predicate, etc. Internal Merge yields discourse-related properties such as topic

and old information, scope, etc.

Notice that all of these are elementary properties of optimal Merge, and with quite broad empirical support.

They are, therefore, properties to be explained by an account of the evolution of language. They follow from the

fact that Merge become available at some point in the evolutionary process, and the assumption that third

factor properties of efficient computation enter into language growth (“acquisition”).

Another question is whether the relation of language to the interface systems is symmetrical. At the 1974

symposium, a number of participants suggested that it is not; rather that the primary relation is to the

semantic interface, to systems of thought. Salvador Luria was the most forceful advocate of the view that

communicative needs would not have provided “any great selective pressure to produce a system such as

language,” with its crucial relation to “development of abstract or productive thinking.” The same idea was

taken up by his fellow Nobel laureate François Jacob, who suggested that “the role of language as a

communication system between individuals would have come about only secondarily.... The quality of language

that makes it unique does not seem to be so much its role in communicating directives for action” or other

common features of animal communication, but rather “its role in symbolizing, in evoking cognitive images,” in

“molding” our notion of reality and yielding our capacity for thought and planning, through its unique property

of allowing “infinite combinations of symbols” and therefore “mental creation of possible worlds.” These ideas

trace back to the cognitive revolution of the 17th

century, which in many ways foreshadows developments from

the 1950s.

Generation of expressions to satisfy the semantic interface yields a “language of thought.” If the assumption

of asymmetry is correct, then the earliest stage of language would have been just that: a language of

thought, used internally. It has been argued that an independent language of thought must be postulated. I

think there are reasons for skepticism, but that would take us too far afield.

The empirical question of asymmetry can be approached from the study of existing languages. We can seek

evidence to determine whether they are optimized to satisfy one or the other interface system. There is, I

think, mounting evidence that the thought systems are indeed primary in this respect, as Luria and Jacob

speculated. We have just seen one illustration, in fact: the properties of Internal Merge. The No-Tampering

Condition entails that the outcome should include the initial and final occurrences, and all intermediate

occurrences. This is correct at the semantic interface; I mentioned a simple case, but it is true far more

generally, in quite interesting ways, a phenomenon called “reconstruction.” It is, however, not true at the

sensorimotor interface, where all but the final position are deleted (with marginal exceptions not relevant here).

Why should this be? Here conditions of computational efficiency and of ease of communication are in conflict.

Computational efficiency yields the universally attested facts: only the final position of Internal Merge is

pronounced. But that leads to comprehension problems. For parsing programs, and perception, major problems

are to locate the “gaps” associated with the element that is pronounced, problems that would largely be

overcome if all occurrences were pronounced. The issue does not arise at the semantic interface. The conflict

between computational efficiency and ease of communication appears to be resolved, universally, in favor of

computational efficiency to satisfy the semantic interface, lending support to speculations about its primacy in

language design.

Perhaps related are discoveries about sign languages in recent years, which provide substantial evidence that

externalization of language is modality-independent. There are striking cases of invention of sign languages by

deaf children exposed to no signing and by a community of deaf people brought together very recently, who



spontaneously developed a sign language. In the known cases, sign languages are structurally very much like

spoken languages, and follow the same developmental patterns from the babbling stage to full competence.

They are also distinguished sharply from the gestural systems of the signers, even when the same gesture is

used both iconically and symbolically. Laura Pettito and her colleagues have studied children raised in bimodal

(signing-speaking) homes, and have found no preferences or basic differences. Her own conclusion is that even

“sensitivity to phonetic-syllabic contrasts is a fundamentally linguistic (not acoustic) process and part of the

baby’s biological endowment,” and that the same holds at higher levels of structure. Imaging studies lend

further support to the hypothesis that “there exists tissue in the human brain dedicated to a function of human

language structure independent of speech and sound,” in her words. Studies of brain damage among signers

has led to similar conclusions, as has comparative work by Tecumseh Fitch and Marc Hauser indicating, they

suggest, that the sensiromotor systems of earlier hominids were recruited for language but perhaps with no

special adaptation.

Comparative work on the second interface, systems of thought, is of course much harder. There do, however,

seem to be some critical differences between human conceptual systems and symbolic systems of other

animals. Even the simplest words and concepts of human language and thought lack the relation to mind-

independent entities that has been reported for animal communication: representational systems based on a

one-one relation between mind/brain processes and "an aspect of the environment to which these processes

adapt the animal'

s behavior," to quote Randy Gallistel’s introduction to a volume of papers on

animal communication. The symbols of human language and thought are quite different, matters explored in

interesting ways by 17th

-18th

century British philosophers, developing ideas that trace back to Aristotle. There

appears to be no reference relation in human language and thought, no relation between an internal symbol

and a mind-independent object. What we take to be a river, or a person, or a tree, or water turns out not to be

identifiable as a physical construct of some kind. Rather, these are creations of what 17th

century investigators

called the “cognoscitive powers” that provide us with rich means to refer to the outside world from certain

perspectives, but are individuated by mental operations that cannot be reduced to a “peculiar nature belonging”

to the thing we are talking about, as Hume summarized a century of inquiry. In this regard, internal conceptual

symbols are like the phonetic units of mental representations, such as the syllable [ba]; every particular act

externalizing this mental entity yields a mind-independent entity, but there is no mind-independent construct

that corresponds to the syllable. Words and concepts appear to be similar in this regard, even the simplest of

them. These properties seem to be unique to human language and thought.

If I understand the professional literature correctly, it is reasonable to suppose that fairly recently, not too long

before about 50,000 years ago, there was an emergence of creative imagination, language and symbolism

generally, mathematics, interpretation and recording of natural phenomena, intricate social practices and the

like, yielding what Wallace called “man’s intellectual and moral nature,” now sometimes called “the human

capacity.” It is commonly assumed that the faculty of language is essential to the human capacity. In a review

of current thinking about these matters, Ian Tattersall writes that he is “almost sure that it was the invention of

language” that was the “sudden and emergent” event that was the “releasing stimulus” for the appearance of

the human capacity in the evolutionary record --the “great leap forward” as Jared Diamond called it, the result

of some genetic event that rewired the brain, allowing for the origin of modern language with the rich syntax

that provides modes of expression of thought, a prerequisite for social development and the sharp changes of

behavior that are revealed in the archaeological record, and presumably the occasion for the trek from Africa,

where otherwise modern humans had apparently been present for hundreds of thousands of years. The

dispersion of humans over the world must post-date the evolution of language, since there is no detectable

difference in basic language capacity among contemporary humans. Like Luria and Jacob, Tatersall takes

language to be “virtually synonymous with symbolic thought,” implying that externalization is a secondary

phenomenon, ideas that I think are supported by internal linguistic evidence, as I mentioned.



Putting these thoughts together, we can suggest what seems to be the simplest speculation about the evolution

of language. In some small group from which we all descend, a rewiring of the brain took place yielding the

operation of unbounded Merge, applying to concepts with properties of the kind I mentioned. Third factor

principles enter to yield an infinite array of structured expressions with interpretations of the kind illustrated:

duality of semantics, operator-variable constructions, unpronounced elements with substantial consequences

for interpretation and thought, etc. The individual so rewired had many advantages: capacities for complex

thought, planning, interpretation, and so on. The capacity is transmitted to offspring, coming to predominate.

At that stage, there would be an advantage to externalization, so the capacity might come to be linked as a

secondary process to the sensorimotor system for externalization and interaction, including communication – a

special case, at least if we invest the term “communication” with some meaning. It is not easy to imagine an

account of human evolution that does not assume at least this much, in one or another form.

Assuming so, what further properties of language require an evolutionary account? That depends on how far

one can proceed in giving a principled account of properties of language, in the sense mentioned earlier:

showing that they derive from interface conditions, primarily the semantic interface, by third factor properties

of efficient computation and the like. If all properties of language could be given principled explanation, then we

would conclude that language is perfectly designed to satisfy semantic conditions, and that the mapping to the

sensorimotor interface – phonology and morphology and probably more – is a maximally efficient means to

convert syntactically generated expressions to a form accessible to the interface. That is too much to expect,

but recent work seems to me to show that the ideal is considerably less remote than would have been imagined

not long ago. If so, we may be able to gain new insights into evolution and development of language by inquiry

into its fundamental design.

pub bickerton on chomsky

Documents