Testing Two Scenarios

What causes the other? When cause and effect are confusing, a knowledge of origins can clarify paradoxes.

Which came first, the word or the sentence? Although the answer may seem obvious, I have been thinking about last week’s post and it occurs to me that the answer to my basic question can help judge the correctness of Chomsky’s theory of generative grammar versus the theory of shared perception favored by this blog.

Children today speak single words first, which might seem to indicate that the word preceded the sentence, but generative theorists have a reply. According to the shared perception theory, when a baby says doggie the word draws attention to the dog. According to generative theory, however, the baby's speech is "holophrastic," expressive of an entire phrase, and can be properly represented as meaning, “There is a dog,” or some such. To date, the choice between the two interpretations has always seemed a matter of taste, but now I wonder.

The issue comes up because of a couple of paragraphs in the article titled , “A Biolinguistic Agenda” by Mark D. Hauser and Thomas Bever in the November 14 Science magazine (abstract here). I discussed the paper last week (see: A Biolinguistic Agenda?) and wasn’t too favorable, but the paper made some theoretically interesting remarks about the division of the brain into distinct right and left halves, or as it is called technically: the lateralization of the brain.

The authors report that lateralization varies from person to person, with the deepest ambiguities involving members of families that include left-handed people. Words can come from either hemisphere of the brain. However,

all subjects [investigated] show left-hemisphere activation for syntactic processes. … The observed variability in how the lexicon is accessed and represented suggests that it is indeed a biologically separable component on linguistic knowledge. [p. 1058]

Their point is the classic Chomskyan one that syntax is an independent element of language that can function without reference to the meaning of the words. The point was made in the famous sentence Colorless, green ideas sleep furiously which is meaningless, but grammatical. Although the sentence is likely to puzzle unsophisticated or uneducated people, the kind of person found at MIT can easily see that it follows a recognizable pattern and that somehow colorless modifies the subject green ideas even while contradicting it.

The observation that syntax is produced by a distinct, specialized region of the brain does not prove Chomsky’s theory but it stiffens the spine of generative thinkers. And it provokes the question I posed, in slightly different form, at the start of this post: which came first, the generally distributed lexicon or the syntactic lateralization?

Different theories point to different answers. The shared-perception theory of language makes sense of the common assumption that words came first, sentences later. Single words (doggie) piloted attention, later groups of words (biggie doggie) focus attention on a detail, and then sentences create full perceptions (biggie doggie chase kitty). So in this  view, the syntactic lateralization came after the brain was already using a more generally distributed lexicon.

Generative grammar, especially in its take-no-prisoners, Chomskyan form, points in the opposite direction. We evolved syntactic structure (and, therefore, lateralization) first. It gave us the capacity to think more clearly. Only as syntactic thought spread did speech and words evolve. (See: Chomsky's Theory of Language Origins). So the generally distributed lexicon is more recent.

Today we cannot settle which processes came first. We cannot tell from fossil skulls when lateralization occurred. But that’s today. DNA and fossil evidence may eventually settle the matter. And it is a relief to think that, in principle, the dispute may not have to always turn on matters of taste.

Comments

[JanetK]

I do not think we can associate the origin of lateralization with the origin of syntactic structure. It appears that lateralization is fairly wide-spread in animals and therefore likely to pre-date the first Homo species. I do not think that we have a good handle on the what, how and why of lateralization yet. However, I would put my money on syntactic structure developing by taking advantage of some already existing facility in the left hemisphere having little or nothing to do with language.
Also, I can see how syntactic structure can be independent of the meaning of words, but I have difficulty imagining a sentence without any words at all in it.

[María]

“all subjects [investigated] show left-hemisphere activation for syntactic processes”. This observation that is emphasized by generativists is right. But the left hemisphere is also the responsible for manual pantomimes. (Funnell, Ferry, Gerry & Gazzaniga, 2005, who have researched this question in split brain patients -i.e. patients who lack the corpus callosum connecting the hemispheres of the brain-, put forward an even more significant piece of information: the left brain predominates for manual pantomime even in left-handed people. In addition, it has been proven that “this effect is not attributable to differences at the conceptual level, as the left and right hemispheres are equally and highly competent at associating tools with observed pantomimes”). Manual pantomimes adapt to a learned model.
Movements in the whole body can occur in two different ways: either by adapting to the environment, or by adapting to the model. These two ways may have to do with specific human hemispheric specialisation. Let us remember that both technical actions and articulatory-phonetic patterns adapt to the model, and that intonation (right hemisphere: Peretz & Hyde, 2003) is not copied. In my view, cultural learning may be the key of lateralisation of language.

[JoseAngel]

The generativist bias in giving preeminence to syntax is wrong-headed from the start. Syntax is grammaticalized semantics, just like semantics is grammaticalized pragmatics. Find the logic of perception, agency attribution, and action, and you will find the deep-buried roots of syntax. But do not look for an implanted "syntactic module" in the brain because you won't find any.

[watercat]

The ethnocentric myth of phoneme—morphme—word—sentence impedes progress, marginalizing languages such as Coast Salish which lacks free morphemes altogether. Sentences in agglutinating languages are indistinguishable from words, or even syllables in the case of signed languages. Since an adequate cross-linguistic definition of 'word' does not exist, for the question to make sense it needs framed in terms of morphemes rather than words.

Chomskyans critically, and wrongly, assume that syntactic categories exist without reference to morphemes. To determine the meaning of a 'holophrastic' expression, they must take into account non-vocal production, as a disturbance in air molecules cannot by itself focus attention on a detail. Once you factor in pointing gestures such as mutual eyegaze, you are no longer dealing with one morpheme, but a 'sentence'.
Children's so-called one-word utterances like [“awgon” (while pointing at an empty bowl)] consist of two clear morphemes expressing two semantic roles, Act and Agent respectively, and two 'syntactic' categories: referential pointing, which indicate agents; and vocables, which indicate Acts. This quasi-syntactic categorization is present and typically organizes into consistent, syntactic, word order even in pantomime.
This organization appears before 'words' in: the manual productions of pre-verbal children; in Home Sign; with Adults who never develop language; in the earliest stage of language creation, as in Al-Sayyid, a third-generation language still deficient in stable form-meaning pairs. Uniformitarianism predicts it also appeared first in proto-humans.