Of Minds and Language (3 page)

Ever since Chomsky stressed the importance of attaining “explanatory adequacy” for any linguistic theory, all hypotheses on processes, mechanisms, constraints, and computations that are not supposed to be innately available have had to be answerable to the possibility of acquisition by the child on the basis of normal linguistic input. For instance, it is a true descriptive generalization about English that all verbs derived from Latin are regular (form the past tense by adding the suffix -
ed
). But since this is patently a generalization that the monolingual child acquiring English has no access to, a theory based on such a generalization would have no explanatory adequacy whatsoever. This part of the book offers several interesting approaches to theories and data by researchers who are highly sensitive to explanatory adequacy, from various angles.

Rochel Gelman deals with the issues of similarity, causality, and core or “skeletal” (innate) versus non-core (acquired) domains. She insists that appeal to universal innate principles does not exclude learning; rather, it forces us to
ask what kind of theory of learning is needed to account for early learnings and the extent to which they help, redirect, or hinder later learnings. Taking up the hard case of counting and natural numbers, and subtraction, Gelman concludes that core domains provide structure to the learning process, because they provide a mental skeletal structure that helps search the environment for relevant data and move readily onto relevant learning paths. The difficulty about non-core domains is that both the structure and the data have to be found. In her words: “It is like having to get to the middle of a lake without a rowboat.”

Instead of marveling at how fast children acquire their mother language, Lila Gleitman invites us to wonder why it takes so long. Although prelinguistic infants discriminate kinds of relations, such as containment versus support or force and causation, they tend to understand and talk about objects first. Since objects surface as nouns, these overpopulate the infant vocabulary as compared to verbs and adjectives, which characteristically express events, states, properties, and relations. Why are verbs “hard words” for the infant? Explaining the acquisition of “perspective verbs” (
chase/flee, buy/sell
) and “unobservables” (
know, think, believe
) leads us into a circle: the transition from the word to the world must be made to a world that is observed
in the right way
, that is, under the characterization that fits the word being used. The central datum is that syntax, in itself, is not only a powerful cue, but the
strongest
of all.

Janet Fodor explores plausible linguistic inputs (“triggers”) that allow the child to fix syntactic parameters. If ambiguous, such triggers do not solve the acquisition process; in that hypothetical situation, the acquisition mechanism must evaluate (as in Chomsky's original 1965 formulation) competing grammar hypotheses. How this could be done by a learner is not obvious, and the possibility is explored here of building on “partial decoding” of competing grammar hypotheses. The approach is based on organizing grammars (vectors of parametric values) in terms of a lattice that learners must tacitly assume for the orderly setting of parameters. As learning proceeds, the smallest grammars are tried out on input sentences and some fail, then being erased from the learner's mental representation of the language domain. In effect this “keeps track” of disconfirmed grammars, by erasing them from the presumably innate lattice. The paper ends by puzzling over the nature of such a lattice.

Thomas Bever was unable to attend the conference, although his approach to the EPP (Extended Projection Principle) had been discussed at the meeting. In light of exchanges with Chomsky, and after reading relevant sections of the transcripts, Bever offered the present paper. The odd requirement that sentences must “sound” as though they have subjects, even when there is no semantic
motivation for this (cf.
It
rained,
There
are problems,
It
seems that he left
, etc.) is still an anomaly within the minimalist program. The condition was initially proposed as a syntactic universal, but while it is roughly correct for English, its presence in other languages is less obvious. Bever takes the EPP out of syntax and explains the vagaries of its generalization by means of a Canonical Form Constraint (CFC). His contribution also explores the implications of this constraint for language comprehension, language acquisition, and Broca's aphasia.

The final section of the proceedings is based on more open-ended talks, some of which were delivered to a more general audience, after the end of the ordinary sessions. In these, broader speculations are often attempted, although, once again, occasional disparity exists between the normally non-technical character of the presentations and the tone of some of the ensuing discussions, as different participants eagerly engage the speakers in lively discussion.

Marc Hauser anticipated some of the issues that were to appear in his recent book on “Moral Minds.” His point of departure, methodologically and conceptually, is Chomsky's insistence on universal innate constraints on humanly possible mental procedures and contents, and the notion of generativity. These are tentatively expanded by Hauser to the domains of ethics (via the work of John Rawls) and aesthetics, with special reference to musical tastes in humans and non-human primates. Universal minimalism is, in his own words, what he is arguing for. Connecting his considerations with other presentations at the conference (especially those by Chomsky, Gallistel, and Cherniak), he offers an interesting panoply of novel experimental data to support his hypotheses. In the discussion, several of Hauser's hypotheses are sympathetically, but also rigorously, challenged by other participants.

Itziar Laka retraces the early steps of the innatist hypothesis for language, probing its limits and suggesting the hardest tests. Thus she takes up a challenge launched by the organizers in the invitation document: thinking about what we know and what we would like to know about minds and language. She examines innate mechanisms disclosed by the study of the perceptual salience of rhythmic/prosodic properties of speech, some specific to humans, some also found in other species. The acquisition of phonemes across different languages suggests that the peculiar thing about human babies is that they are very quickly able to construct something new, using largely an old perceptual mechanism. At the end of her exploration of the conceptual and empirical development of the field of generative linguistics, connecting with several other issues freshly
discussed at the conference, Laka cannot help but wonder about the nature of parameters.

Nuria Sebastián-Gallés explores the reasons why some individuals are better than others at acquiring a second language (L2). After discussing the issues the literature has raised with regards to possible causes for this disparity, she presents several data showing differences in brain structure and function in relevant groups tested (of poor versus good L2 learners). Importantly, in general these differences are not in language-related areas. This leads her to conclude that it is probably not the language faculty as such that is involved in proficient L2 learning, but other, perhaps general, cognitive capacities. Inasmuch as such differences are not at all important for the acquisition of a first language, these results suggest that the two processes may be quite distinct.

Angela Friederici examines the different computations carried out by the two hemispheres of the brain and tests the prediction that there are separate, and sequential, phases in processing syntactic and semantic information. She also reports on data suggesting that the right hemisphere is responsible for the processing of prosodic information. The focus of her presentation is intonational phrasing and the hypothesis that it tracks syntactic phrasing. Processing structural hierarchies activates Broca's area, parametrically as a function of the number of syntactic movements involved. A judicious insertion of morphological markers in German allowed her also to conclude that local structure-building processes precede lexical-semantic processes. Curious data on sex differences in the interactions of semantic-emotional and prosodic-emotional processes during language comprehension show women using prosodic-emotional information earlier than men.

In Chomsky's
concluding remarks
, virtually all of the different threads spun during the conference finally come together. Sharing with us his unique impressions, perplexities, excitements, and after-thoughts – and merging some of the issues discussed during the conference, while suggesting disparities between others – Chomsky retraces the main lines of development of the generative enterprise. With his vast knowledge and perspective, after reconstructing historical antecedents, he insists on the strangeness of the amnesia that has struck the cognitive sciences in the last couple of decades. Many of the fundamental problems that still (should) define the agenda for our understanding of mind at work, how it evolved and develops, and how it is embodied in brains, were openly discussed from the eighteenth century on, but appear to have been partially forgotten in our times. Perhaps Chomsky's most lasting message in this book, in our view full of both humility and insight, is that a look into the future must be accompanied by a rediscovery of the intellectually relevant past.

Noam Chomsky

I have been thinking about various ways to approach this opportunity, and on balance, it seemed that the most constructive tack would be to review, and rethink, a few leading themes of the biolinguistic program since its inception in the early 1950s, at each stage influenced by developments in the biological sciences. And to try to indicate how the questions now entering the research agenda develop in a natural way from some of the earliest concerns of these inquiries. Needless to say, this is from a personal perspective. The term “biolinguistics” itself was coined by Massimo as the topic for an international conference in 1974
¹
that brought together evolutionary biologists, neuroscientists, linguists, and others concerned with language and biology, one of many such initiatives, including the Royaumont conference that Massimo brought up.
²

As you know, the 1950s was the heyday of the behavioral sciences. B. F. Skinner's William James lectures, which later appeared as
Verbal Behavior
(1957), were widely circulated by 1950, at least in Cambridge, Mass., and soon became close to orthodoxy, particularly as the ideas were taken up by W. V. Quine in his classes and work that appeared a decade later in his
Word and Object
(1960). Much the same was assumed for human capacity and cultural variety generally. Zellig Harris's (1951)
Methods of Structural Linguistics
appeared at the same time, outlining procedures for the analysis of a corpus of materials from sound to sentence, reducing data to organized form, and particularly within American linguistics, was generally assumed to have gone about as far as theoretical linguistics could or should reach. The fact that the study was called Methods reflected the prevailing assumption that there could be nothing much in the way of a theory of language, because languages can “differ from each other without limit and in unpredictable ways,” so that the
study of each language must be approached “without any preexistent scheme of what a language must be,” the formulation of Martin Joos, summarizing the reigning “Boasian tradition,” as he plausibly called it. The dominant picture in general biology was in some ways similar, captured in Gunther Stent's (much later) observation that the variability of organisms is so free as to constitute “a near infinitude of particulars which have to be sorted out case by case.”

European structuralism was a little different, but not much: Trubetzkoy's
Anleitung
, a classic introduction of phonological analysis,
³
was similar in conception to the American procedural approaches, and in fact there was very little beyond phonology and morphology, the areas in which languages do appear to differ very widely and in complex ways, a matter of some more general interest, so recent work suggests.

Computers were on the horizon, and it was also commonly assumed that statistical analysis of vast corpora should reveal everything there is to learn about language and its acquisition, a severe misunderstanding of the fundamental issue that has been the primary concern of generative grammar from its origins at about the same time: to determine the structures that underlie semantic and phonetic interpretation of expressions and the principles that enter into growth and development of attainable languages. It was, of course, understood from the early 1950s that as computing power grows, it should ultimately be possible for analysis of vast corpora to produce material that would resemble the data analyzed. Similarly, it would be possible to do the same with videotapes of bees seeking nourishment. The latter might well give better approximations to what bees do than the work of bee scientists, a matter of zero interest to them; they want to discover how bee communication and foraging actually work, what the mechanisms are, resorting to elaborate and ingenious experiments. The former is even more absurd, since it ignores the core problems of the study of language.

A quite separate question is whether various characterizations of the entities and processes of language, and steps in acquisition, might involve statistical analysis and procedural algorithms. That they do was taken for granted in the earliest work in generative grammar, my
Logical Structure of Linguistic Theory
(LSLT) in 1955, for example. I assumed that identification of chunked wordlike elements in phonologically analyzed strings was based on analysis of transitional probabilities – which, surprisingly, turns out to be false, as Thomas Gambell and Charles Yang discovered, unless a simple UG prosodic principle is presupposed. LSLT also proposed methods to assign chunked elements to categories, some with an information-theoretic flavor; hand calculations in that
pre-computer age had suggestive results in very simple cases, but to my knowledge, the topic has not been further pursued.