Sophie's Story looks at the strengths and limitations of Specific Language Impairment.
The story of language and its origins that has been emerging on this blog is fairly simple:
- Members of the human lineage began using words when a population became communal enough to trust one another with shared knowledge.
- Those first language users differed from their ancestors in the nature of their community, not in the acquisition of some new verbal skill.
- Once populations of language users became competitive, selection pressures to enrich language functions grew stronger and new verbal abilities did evolve.
- The competition à enrichment cycle persisted and continued to produce expanded verbal abilities.
How might such a story be tested? Points 1 and 2 imply that, intellectually, some parts of language should require no special human abilities, but point 3 implies that some parts of language do depend on special adaptations. So I have been keeping an eye open for any findings that would support or demolish these expectations.
A review article by Heather K.J. van der Lely and Steven Pinker titled "The biological basis of language," seems to be the paper I have been looking for. Basing their work on the study of children with Grammatical Specific Language Impairment (G-SLI), the authors sort syntax and phonology under two headings: basic and extended.
An utterance falls under this heading if it is concrete and the meaning of a word is inherent or can be understood by referring to an adjacent word; ball for example is concrete and directly associated with a type of physical object; red ball is slightly more complicated but the two words stand side-by-side and the two together present a clear, concrete image. These conditions are sufficient to enable language users to distinguish between a red ball and a blue one, or a gray ball and a gray hat. Chimpanzees can be trained to make these kind of distinctions in sign language, and I would not be surprised to see a dog turn up on the David Letterman show that could make the same distinctions.
Verbs are also found in basic syntax, but they are concrete and not conjugated. Basic syntax will get you I want milk when I wanted milk would be more appropriate.
Basic syntax is supported by two ventral pathways in the brain. (Ventral pathways are on the lower side of the brain.) One supports "local phrase structure," that is statements like red ball while the other pathway supports use of the correct word.
An utterance goes beyond basic syntax by (a) not being concrete, or (b) if understanding requires more than use of an adjacent word, or (c) the words used require some regular process of marking.
Suppose I ask, Who did Joe see? The ventral pathways cannot handle the question. Who is not a concrete word. It asks for a concrete answer, but by itself it evokes no image. Then look at the verb: did…see. First, Joe splits the verb phrase into separate sections. Second, did serves no associative purpose. It is not like red, telling us something about its neighbor ball. The last part of the sentence might evoke an image of Joe with wide open eyes, but the first two words cannot be managed by the ventral pathways. We need to appeal to some other part of the brain.
Or consider, John played and John ran quickly. These examples are concrete and use phrases that depend upon adjacent relationships, but the words play+ed and quick+ly cannot be just memorized. They must be constructed. Again, the ventral pathways cannot serve.
The authors report on a dorsal pathway in the brain linking Broca's area with Wernicke's area. (Dorsal pathways are in the upper side of the brain. What's the importance of dorsal vs ventral? For one thing, they make clear that the routes are separate and have separate histories.) The dorsal paths, "unlike the ventral ones, do not mature until approximately 7 years of age…. Moreover, the dorsal pathways in human brains differ substantially from those in other primates, suggesting that phylogenetic changes to the dorsal pathway may have been a key driver of the evolution of language." [p. 5]
The primary evidence for the existence of these two distinct syntactic systems is a developmental disorder called Grammatical Specific Language Impairment (G-SLI), which is a highly hereditable condition first identified by Dr. van der Lely. The condition depends on two genes seen before on this blog: FOXP2 and CNTNAP2 (see my post).
So there we have it: as expected, a base line intelligence allows for language as soon as the lineage has the proper motivation, but more complex language required further evolutionary events. The evidence, however, rests on a poorly understood developmental condition, so the paper should be considered supportive, but not yet conclusive.
I cannot refrain from remarking that this evolutionary process is not at all like the one favored by the authors of the paper discussed in my last post. That paper imagined a one-step process in which the brain develops the ability to attach two syntactic structures—e.g., red+ball, play+ed—but basic syntax allows some merges while disallowing others.
Specific language impairment can also disturb the ability of a listener to understand the rhythms of speech. Basic phonological skills, however, are not impaired. Basic phonology includes the ability segment sounds sothatthelistenercanbreakstreamsintowords. It also includes the ability to distinguish one word from another, based on a single phoneme, e.g., fan versus van. Listeners can also distinguish real words—e.g., brown—from nonwords—e.g., browm. The authors do not say so, but these fundamental abilities are probably sufficient to manage basic syntax.
When speech sounds enter the ears they travel onto the brain where they are split "into two streams" . One stream heads for a ventral pathway "which relates words' sounds to their meanings" .
Language, especially language dependent on extended syntax, requires more than the ability to manage phonemes. Take the sentence: Joe wore mismatched red and black and pink socks. In print, you cannot tell whether one sock is red or red and black. In speech, however, the rhythm will indicate either [red and black] [and pink] or [red] [and black and pink].
There was an old Johnny Carson joke, "It has been a muggy day here in New York. Muggy, as in I is de mugger, you is de muggee." Besides depending upon the morphology of extended syntax, it requires a working memory to recall that the joke is a play on the word muggy. Basic phonology cannot handle it.
Some words have to be sounded out. Just this past week Stephen Colbert had a joke using the word supercalifragilisticexpialidohshit. Basic phonology cannot handle that either.
A dorsal pathway in the brain's left hemisphere handles the demands of extended phonology. "The dorsal [phonology] network appears to overlap the dorsal network proposed for syntax and morphology"  and the ventral networks proposed for basic syntax and phonology also seem to overlap.
Let me interrupt right here to say that this pattern is not what would be expected if the language faculty had evolved independently of the 'externalization' of language. On the other hand, the matter is not settled. Phonology and phonological memory appear to depend on two other genes, ATP2C2 and CMIP.
By the way, I consider this article important enough to justify a second post, so I will have more to say on basic and extended syntax tomorrow.