Of Minds and Language (59 page)

H
AUSER
: Yes, it is a crude cut. And again, psychophysically there have been studies showing that animals exhibit octave discrimination and generalization.

In particular, Anthony Wright and coworkers have shown that if you train rhesus monkeys in a match-to-sample paradigm, using children's melodies, they can readily do the transpositions.
¹⁷

G
ELMAN
: The really interesting question is whether they will also generalize to the fifth; actually I have data I never published that show 5-year-olds will. But they're experienced.

H
AUSER
: Way experienced.

Itziar Laka

One of the most controversial and influential aspects of Chomsky's legacy is the hypothesis that there is an innate component to language, which he named universal grammar (UG) (Chomsky 1965), in homage to the rationalist thinkers in whose footsteps he was walking. I would like to present and discuss results obtained from a variety of sources, mostly in neighboring fields within cognitive science, that bear on this central issue.

Like Janet Fodor, who mentions in her paper Chomsky's
Aspects
(1965), I also want to go back, about fifty years, and start with a piece of literature that is crucial for understanding what the research program of generative linguistics is, and also what the current research program is in the interdisciplinary study of human language. I am referring to the review that Chomsky wrote of B. F. Skinner's
Verbal Behavior
(Chomsky 1959c).

The hypothesis that there are innate organism-internal factors that constrain the languages that humans know and use sounded preposterous to most scholars in the Humanities and Social Sciences back in 1959 when Chomsky published it. The UG hypothesis, as he later named it, has since generated
a great amount of research, discussion, and argument, a distinguished example of which is the Royaumont Debate between Piaget and Chomsky in 1975 (Piattelli-Palmarini 1980b). In other fields, however, such as biology, the claim that human languages are largely shaped by innate conditions not only did not encounter resistance, but was received with sympathy, because it naturally converged with a general view of living organisms and the importance of genetic factors in behavior (Piattelli-Palmarini 1994).
¹
I think it safe to say that, fifty years later, it is widely accepted that innate mechanisms have a relevant role to play in a full understanding of the human capacity for language. Current disagreements concern the nature and specificity of those mechanisms, regarding both our species and the cognitive domain(s) where they belong.

So the question I would like to pursue is: what are the contents of UG? That is to say, what has been discovered regarding the innate component of human language since it was argued, half a century ago, to be a significant part of a human's knowledge of language? I will not engage in an exhaustive review of the variety of linguistic arguments and evidences put forward during these years to substantiate the hypothesis within linguistic theory. Rather, I will look at a variety of mechanisms that stand the sharpest tests for innateness, and discuss which ones are good candidates for UG membership and why. This will constitute the part of “what we think we know” about innateness in language, but since the organizers of this event have also urged us to think of “what we would like to know”
²
I will also briefly mention at the end an aspect of language that still appears elusive, I think, but is crucial for a thorough understanding of its whole design.

A secondary goal of this talk is to bring to the attention of linguists results and findings from neighboring fields within cognitive science that bear on the issue of innateness and specificity in language. I happen to be a theoretical linguist who has become increasingly engaged in cooperative, experimental research with cognitive psychologists. I believe the benefits of this interdisciplinary way of working largely surpass the frustrations and communication difficulties that are inevitably encountered along the way. I will discuss
discoveries related to innateness and specificity relatively well known by language researchers within cognitive psychology, but perhaps not very well known in linguistics proper, and discuss their relevance both to the research program that took off some fifty years ago with the birth of generative grammar, and to our current concepts of grammar and language. I would also like to caution against the temptation to take it for granted that any innate properties found in language must necessarily be part of UG. As we shall see, innateness is a necessary condition for a given mechanism to belong in UG, but not a sufficient one: specificity is also required. UG should contain those properties of language, if any, that cannot be fully accounted for elsewhere, for example in the sensorimotor side of language or in the conceptual-semantic component, both of which predate grammar. This approach, this research strategy, takes the name of minimalism (Chomsky 1995b) and necessarily causes us to reflect on what UG is and to try to reduce it, trim it, and pare it down to its bare necessities.

The term universal grammar is not used in the 1959 review, but the hypothesis, though nameless, was already there, right at the start. In fact, the word “innate” appears three times in the review, once referring to imprinting in animals, and twice referring to human language in the context of language acquisition. One instance is this:

As far as acquisition of language is concerned, it seems clear that reinforcement, casual observation and inquisitiveness (coupled with a strong tendency to imitate) are important factors, as is the remarkable capacity of the child to generalize, hypothesize and process information in a variety of very special and apparently highly complex ways which we cannot yet describe or begin to understand, and which may be largely innate, or may develop through some sort of learning or maturation of the nervous system.

(Chomsky 1959c: 43)

We can see that Chomsky is not saying that imitation is irrelevant for the acquisition of language; he is simply making the point that it will not suffice to tell the whole story. In fact imitation is a crucial, rather distinctive property of humans, and our imitation is highly sophisticated (Meltzoff and Printz 2002). Despite this, language acquisition researchers have found abundant evidence that imitation alone does not account for language learning. The crucial issue in the quote is that it appeals to (then unknown) conditions that determine the process of language acquisition – that is, hypothetical acquisition mechanisms which were “complex ways which we cannot yet describe or begin to understand.” Today, although we still do not fully understand them, we have come a pretty long way. Some of the “special and apparently highly complex ways” in which infants process linguistic input have been discovered in recent years,
and have been discussed in other papers presented at this conference (see papers in Part III of this volume).

Moreover, in considering the acquisition of the lexicon, Chomsky (1959c: 42) says:

It is possible that ability to select out of the auditory input those features that are phonologically relevant may develop largely independently of reinforcement, through genetically determined maturation. To the extent that this is true, an account of the development and causation of behavior that fails to consider the structure of the organism will provide no understanding of the real processes involved.

Again, though at the time they stirred minds and thoughts, from a contemporary perspective these words do not say anything out of the ordinary; there is widespread agreement that, already at birth, infants do in fact select certain features from the auditory input. Today, few experts would disagree with the claim that it is crucial to know the structure of the human brain and its maturation in order to have a full picture of language acquisition. It is about the nature, specificity, and extent of these organism-internal conditions that the debate is taking place nowadays.

In 1959, however, none of this was so clear. In discussing Lashley's work on neurological processes, Chomsky (1959c: 55–56) proposed a research program for linguistics: “Although present-day linguistics cannot provide a precise account of these integrative processes, imposed patterns, and selective mechanisms, it can at least set itself the problem of characterizing these completely.” This research program should be of relevance to the study of the brain, and vice versa: “The results of such a study [of the characterization of the mechanisms of language] might, as Lashley suggests, be of independent interest for psychology and neurology (and conversely).” These statements, which sounded extremely foreign to people in linguistics and psychology at the time, paint a landscape that has become the dwelling space of contemporary linguistics and cognitive science. This expectation of mutual importance and increasing convergence is our present: there is a vast amount of research in human language where linguists listen to what other fields can contribute about human language, and conversely. In sum, the two main conceptual seeds in the review of
Verbal Behavior
have clearly stood the test of time and bloomed. The first such seed is that there are innate aspects to our knowledge of language, and the second one is that if we want to understand them, we first need to know what language is like. Finding this out is the natural research program for linguistics.

To answer the question of what language is like, we turn now to Chomsky's 1957 work,
Syntactic Structures
. This small book, which had a hard time finding a publisher, was very successful. It proposed an approach to the study
of language that set up most of the foundational issues still in the background of the discussion today, as I would like to show you. The goal of linguistics, according to
Syntactic Structures
(p. 11) is to determine

the fundamental underlying properties of successful grammars. The ultimate outcome of these investigations should be a theory of linguistic structure in which the descriptive devices utilized in particular grammars are presented and studied abstractly, with no specific reference to particular languages.

Whereas the review of
Verbal Behavior
is very much concerned with biological aspects of language,
Syntactic Structures
focuses on the formal architecture of grammar and its abstract properties, without mentioning biology or psychology. Years later, in the eighties, both sides of this research program, the biological/psychological side and the formal side, would appear hand in hand, as in this more recent quote from
Knowledge of Language
(Chomsky 1986b: 3):

The nature of this faculty is the subject matter of a general theory of linguistic structure that aims to discover the framework of principles and elements common to attainable human languages; this theory is now often called “universal grammar” (UG), adapting a traditional term to a new context of inquiry. UG may be regarded as a characterization of the genetically determined language faculty.

Universal grammar should therefore be the genetically determined part of language and would include those aspects of language that are not determined by experience. However, primitives and mechanisms involved in language that are not specific to language could (and should) be excluded from UG, because they belong to broader or related but independent cognitive domains. This naturally brings us to consider innateness and specificity in greater detail. These two properties are not synonymous, for a given trait might be innate in a species, but not specific to it, as is the case with fear of snake-like forms in mammalians. Also, there are increasingly restrictive degrees of specificity, relative to a species or relative to a cognitive domain. A given property could be human-specific, but not necessarily language-specific. This point was already discussed in the Royaumont debate in 1975, as this remark by Chomsky shows:

On this point I agree with [David] Premack. I think he is right in talking about two different problems that enter into this whole innateness controversy. The first is the question of the genetic determination of structures…the second problem concerns specificity. (Piattelli-Palmarini 1980a: 179)

There are phenomena that constitute necessary prerequisites for language, which are innate but which are clearly not specific, either to humans or to language. However, in the history of discovery, such phenomena have often
been thought (especially when noticed for the first time) to be specific to language. A lesson from history, therefore, is that when in our research path we find something characteristic of human language, we would be wise to check whether it is really specific to humans and specific to language. One of these, well known to psychologists but perhaps not equally well known among linguists, is categorical perception (CP).

Fig. 20.1. Categorical Perceptionof /b/ and /d/ (after Liberman et al. 1957).