The Galilean Challenge: Architecture and Evolution of Language To cite this article: Noam

Chomsky 2017 J. Phys.: Conf. Ser. 880 012015 View the article online for updates and
enhancements https://doi.org/10.1088/1742-6596/880/1/012015

it becomes possible to formulate what we may call the Basic Property of human
language: the language faculty of the human brain provides the means to construct a
digitally infinite array of structured expressions, each of which has a semantic
interpretation expressing a thought, and each of which can be externalized by means
of some sensory modality. The infinite set of semantically interpreted objects
constitutes what has sometimes been called a language of “thought”: the system of
thoughts that receive linguistic expression and that enter into reflection, inference,
planning, and other mental processes, and when externalized can be used for
communication and other social interactions. We may fairly assume that the language
faculty is shared among humans. There are no known group differences in language
capacity, and individual variation is at the margins.
The fundamental task of inquiry into language is to determine the nature of the Basic
Property. To the extent that its properties are understood, we can seek to investigate
particular internal languages, each an instantiation of the Basic Property, much as each
individual visual system is an instantiation of the human faculty of vision. We can
investigate how the internal languages are acquired and used, how the language
faculty itself evolved, its basis in human genetics and the ways it functions in the
human brain. This general program of research has been called the Biolinguistic
Program. The theory of the genetically based language faculty is called Universal
Grammar; the theory of each individual language is called its Generative Grammar.
The Basic Property takes language to be a computational system, which we therefore
expect to observe general conditions on computational efficiency. A computational
system consists of a set of atomic elements and rules to construct more complex ones.
For generation of the language of thought, the atomic elements are word-like, though
not words; for each language, the set of these elements is its lexicon. The lexical items
are commonly regarded as cultural products, varying widely with experience and
linked to extra-mental entities – an assumption expressed in the titles of standard
works, such as W.V. O. Quine’s influential study Word and Object. Closer examination
reveals a very different picture, one that poses many mysteries. Let’s put that aside for
now, turning to the computational procedure.
Linguistics has an additional motive of its own for seeking the simplest theory: it must
face the problem of evolvability. Not a great deal is known about evolution of modern
humans, but the few facts that are well established, and others that have recently
been coming to light, are rather suggestive, and conform well to the conclusion that
the language faculty is near optimal for a computational system, the goal we should
seek on purely methodological grounds. One fact that does appear to be well
established I have already mentioned: that the faculty of language is a true species
property, invariant among human groups – and, furthermore, unique to humans in its
essential properties. It follows that there has been little or no evolution of the faculty
since human groups separated from one another. Recent genomic studies place this
date not very long after the appearance of anatomically modern humans about
200,000 years ago, perhaps some 50,000 years later, when the San group in Africa
separated from other humans. There is no evidence of anything like human language,
or symbolic activities altogether, before the emergence of modern humans. That leads
us to expect that the faculty of language emerged along with modern humans or not
long after, a very brief moment in evolutionary time. It follows, then, that the Basic
Property should indeed be very simple. The conclusion conforms to what has been
discovered in recent years about the nature of language, a welcome convergence. The
discoveries about early separation of the San people are highly suggestive. They
appear to share the general human language capacity, but have significantly different
externalized languages. With irrelevant exceptions, their languages are all and only the
languages with phonetic clicks, with corresponding adaptations in the vocal tract. The
most likely explanation for these facts, developed in detail in current work by Dutch
linguist Riny Huijbregts,4 is that possession of internal language preceded separation,
which in turn preceded externalization, the latter in somewhat different ways in
separated groups. Externalization seems to be associated with the first signs of
symbolic behavior in the archaeological record, after the separation. Putting these
observations together, it seems that we are reaching a stage in understanding where
the account of evolution of language can perhaps be fleshed out in ways that were
unimaginable until quite recently.
Returning to the Basic Property, as we have seen we have reason to believe that it may
be quite simple. The challenge for research, then, is to show how the facts of language
are accounted for in terms of the Basic Property: more fully, by the interaction of the
Basic Property, specific experience, and language-independent principles, including
principles of computational efficiency. The challenge is of particular interest and
significance when it is clear that the child’s experience provides little or no relevant
evidence – a situation far more prevalent than commonly realized, so careful
examination reveals, from acquisition of word meaning on to syntactic structures and
the semantic properties of the generated language of thought. Also of particular
interest are the universal properties of the language faculty that began to come to
light as soon as serious efforts were undertaken to construct generative grammars,
including quite simple principles that had never been noticed, and that are quite
puzzling. One crucial and puzzling principle is structure-dependence: the rules that
yield the language of thought attend solely to structural properties, ignoring properties
of the externalized signal, even such simple properties as linear order.
The property is illustrated by elementary examples. Consider the sentence “John and
his father are tall” – not is tall, though the bigram frequency of father-is is much
greater than of father-are, and the computation using linear order (adjacency) is far
simpler than the computation that has to analyze the sentence into phrases and use
phrasal locality. To take an example of semantic construal, consider the sentence birds
that fly instinctively swim. It is ambiguous: the adverb “instinctively” can be associated
with the preceding verb (fly instinctively) or the following one (instinctively swim).
Suppose now that we extract the adverb from the sentence, forming instinctively,
birds that fly swim. Now the ambiguity is resolved: the adverb is construed only with
the linearly more remote but structurally closer verb swim, not the linearly closer but
structurally more remote verb fly. The only possible interpretation – birds swim – is
the unnatural one, but that doesn’t matter: the rules apply rigidly, independent of
meaning and fact. What is puzzling, again, is that the rules ignore the simple
computation of linear distance and keep to the far more complex computation of
structural distance. The principle of structure dependence holds for all constructions in
all languages, and it is indeed puzzling. Furthermore, it is known without relevant
evidence, as is clear in cases like the one I just gave and innumerable others.
Experiment shows that children understand that rules are structure-dependent as
early as they can be tested, by about age 3 – and are of course not instructed. We can
be quite confident, then, that structure-dependence follows from principles of
universal grammar that are deeply rooted in the human language faculty.
The only plausible conclusion, then is that structure-dependence is an innate property
of the language faculty. That raises the question why this should be so? There is only
one known answer, and fortunately, it is the answer we seek for general reasons: the
computational operations of language are the simplest possible ones. Again, that is the
outcome that we hope to reach on methodological grounds, and that is to be expected
in the light of the evidence about evolution of language already mentioned. The
simplest recursive operation, embedded in one or another way in all others, takes two
objects already constructed, say X and Y , and forms a new object Z, without modifying
either X or Y or adding any further structure. Accordingly, Z can be taken to be just the
set {X, Y }. In current work, the operation is called Merge. Since Merge imposes no
order, the objects constructed, however complex, will be hierarchically structured but
unordered, and operations on them will necessarily keep to structural distance,
ignoring linear distance. It follows that the linguistic operations yielding the language
of thought will be structure-dependent, as indeed is the case, resolving the puzzle.
Externalization of language maps internal structures into some sensorimotor modality,
usually speech. The sensorimotor system requires linear order; we cannot speak in
parallel. But none of this enters the generation of the language of thought, which
keeps to structural relations. More generally, externalization of language seems to be a
peripheral aspect of language, not entering into its core function of providing a
language of thought, contrary to a long tradition, including the formulation of the
Galilean challenge.
Perception yields further evidence in support of this conclusion. The auditory systems
of apes are quite similar to those of humans, even attuned to the phonetic features
that are used in language. But the shared auditory-perceptual systems leave apes
without anything remotely like the human faculty of language. Some have attributed
this lack to deficiency of articulatory apparatus and vocal learning, but apes can
gesture quite easily and as has been known for many years, sign language is virtually
identical to spoken language in its basic properties and acquisition, and human groups
(including children), under the right conditions, even invent normal sign languages
with no linguistic input at all.
These results support the conclusion that internal language is independent of
externalization, and that internal language evolved quite independently of the process
of externalization.
The most powerful evidence, however, is what we learn from investigation of language
design. One crucial example is the one just mentioned: the explanation for the puzzle
posed by the linguistic universal of structure-dependence, which follows from the null
hypothesis: that the computational system is optimal, and hence ignores linear order,
the most elementary feature of externalization.
Not long ago it would have seemed absurd to propose that the operations of human
language could be reduced to Merge, along with language-independent principles of
computational efficiency. But work of the past few years has shown that quite intricate
properties of language follow directly from this assumption, along with a few other
quite simple ones. One important result has to do with the property of displacement, a
ubiquitous and also quite puzzling property of language: phrases are heard in one
position but interpreted both there and in some other position that is silent but where
they could have occurred – a puzzling property, which is never built into artificial
symbolic systems for metamathematics, programming, or other purposes.
For example, the sentence “which book will you read?” is interpreted as meaning
roughly: “for which book x, you will read the book x,” with the nominal phrase book
heard in one position but interpreted in two positions. I will not go into the details, but
it is easy to show that Merge-based computation automatically yields displacement
with copies; in this case, two copies of which book, yielding the correct semantic
interpretation directly. The same process yields quite intricate semantic
interpretations, and also has significant implications about the nature of language. To
see why, consider for example the sentence “the boys expect to see each other” and
the same sentence preceded by “which girls”: “which girls do the boys expect to see
each other.” In the latter sentence, “each other” does not refer back to the closest
antecedent, “the boys,” as such phrases universally do, but rather to the more remote
antecedent “which girls.” The sentence means “for which girls the boys expected those
girls to see each other.” That is what reaches the mind, under Merge-based
computation with automatic copies, though not what reaches the ear. What reaches
the ear violates the locality condition of referential dependency. Deletion of the copy
in externalization causes processing problems: such filler-gap problems, as they are
called, can be become quite severe, and are among the major problems of automatic
parsing and perception. If the copy were not deleted, the problem would not arise.
Why then is it deleted? Again, because of principles of efficient computation that
reduce what is computed to the minimum: at least one copy must appear or there is
no evidence that displacement took place at all, so only the structurally most
prominent one remains (with important qualifications strengthening the conclusion,
which I will put aside), leaving a gap that must be filled by the hearer – a matter that
can become quite intricate.
These examples illustrate a general phenomenon of some significance. Language
design appears to maximize computational efficiency but disregards communicative
efficiency. In fact, in every known case in which computational and communicative
efficiency conflict, communicative efficiency is ignored. These facts argue against the
common belief that communication is the basic function of language. They also further
undermine continuity assumptions about language evolving from animal
communication. And they provide further evidence that externalization, which is
necessary for communication, is a peripheral aspect of language.
As I mentioned, there are methodological reasons and also some evolutionary reasons
to expect that the basic design of language will be quite simple, perhaps even close to
optimal. With regard to externalization of language, the same methodological
arguments hold, but the evolutionary arguments do not apply. In fact, externalization
of language may not involve evolution at all. The sensorimotor systems were in place
long before the appearance of language. Mapping the internal language to some
sensorimotor system for externalization is a hard cognitive problem, relating two
systems that are unrelated: an internal system that may be highly efficient
computationally, and a sensory modality unrelated to it. That would lead us to expect
that the variety, complexity, and easy mutability of observed languages might lie
primarily in externalization. Increasingly, it seems clear that that is the case. And in fact
it should be expected, since the principles of the internal language are largely known
by children without evidence, as, indeed, is a great deal more about language,
including almost all semantic and most syntactic properties – a matter of contention,
but solidly established, I think.
Let us return finally to the second component of a computational system, the atomic
elements: for language, the lexical items. As I mentioned, the conventional view is that
these are cultural products, and that the basic ones – those used for referring to the
world – are associated with extra-mental entities. This representationalist doctrine has
been almost universally adopted in the modern period. The doctrine does appear to
hold for animal communication: a monkey’s calls, for example, are associated with
specific physical events. But the doctrine is radically false for human language, as was
recognized as far back as classical Greece.