[ CogSci Summaries home | UP | contact | email ]

Zurif, Edgar B. (1990). Language and the brain. In Language: An invitation to Cognitive Science, Vol 1. Osherson, D. N. & Lasnik, H. MIT Press, Cambridge, MA. (pp 177-198)

Author: Jim Davies 1998, jim@jimdavies.org

Cite this chapter for these claims:

This article presents a "Neural Theory of Language."
Dynamic model of brain functioning underlying language processing
A model of how linguistic computations are translated into Neural Action
Such a dynamic model is still far off. This article talks about language disorders as a result of brain damage. Surprisingly, damage to parts of the brain do not lead to accross-the-board reduction in language proficiency. Can this selective damage to specific aspects of language use be accounted for by theories of language?

p178: Aphasia

Broca's aphasia

damaged area: Left frontal lobe (including Broca's area), that portio of the cortex invloved with the muscles used to speak
clinical signs:

p: 179 Wernicke's aphasia

damaged area: Wernicke's area: posterior region of left hemisphere, adjacent to the area invloved with hearing.
clinical signs: speech is fine, marked comprehension deficit.

p:180 Wernicke's interpretation (1874):
That the brain had two anatomically distinct modules for listening and speaking. Much of the history of aphasia research has been reaction to this theory.

p:181 Jackson's interpretation (1884):
The unit of language is the proposition, not the word.
Aphasia is not being able to propositionalize.
What happens is that the brain damage frees primitive behaviors from higher-level control.
The linguistic symptoms are the result of a more general limitation, like a problem with using symbols.
Problems with this theory:
The relation of aphasia to other symbolic disruptions is inconsistent.
Other types of symbol-referencing behavior (such as pantomine, classification performance) can be individually impaired.
Maybe there is a system supporting reference, and a linguistic system for representing knowledge.
The symptoms of dementia supprt this idea, where symbol reference is impaired but grammar is not.

p: 182 Caramazza and Zurif (1976) theory
Broca's patients rely on semantics to interpret sentences.
Where they need to use syntactic cues, they fail.
E.g. they can understand The apple that the boy is eating is red because they know already that boys eat apples, not vice versa. They have trouble with The girl whom the boy is pushing is tall since either the boy or the girl could be doing the pushing; you need to use syntactic information of the sentence to determine which one is tall.
More specifically, they have trouble with object-relative construction. They are better with the subject relative construction:
The boy who is pushing the girl is tall.
They have trouble with the passive voice:
The girl was pushed by the boy.
and less trouble with the active voice:
The boy pushed the girl.
They have difficulty with object-cleft sentences:
It is the girl who the boy is pushing.
and less difficulty with subject-cleft sentences:
It is the boy who is pushing the girl
Here is a summary of the above:
Broca's Aphasia Comprehension
impaired better
object-relative: "The girl whom the boy is pushing is tall" subject-relative: "The boy who is pushing the girl is tall"
passive voice: "The girl was pushed by the boy" active voice: "The boy pushed the girl"
object-cleft sentences: "It is the girl who the boy is pushing." subject-cleft sentences: "It is the boy who is pushing the girl."

Also, interpretation of the word "the" in a sentence is impaired.

p: 184-186 Deep and Surface structure theory
These results can be simply explained using deep/surface structure theory (Grodzinsky 1986). When interpreting a passively stated sentence, you must leave a "trace" in the surface structure. The beavior of Broca's aphasics can be explained thus: they lack the ability to leave a trace. With no trace, interpretation is at chance. Indeed, they never do worse than chance.
It has been found that some of those with lesions in this place do not exhibit these missing-trace behaviors, so the theory doesn't tell us a whole lot about brain localization.

open class words
content words, words with meanings in the world

closed class words
functional words, articles, grammatical morphemes like the plural 's' in English

p: 188 A processing analysis
The idea is that open and closed class words ar processed differently.
Rosenberg (1985) did an experiment in which participants tried to cross out certain letters in a series of words while simultaneously doing another task (reading prose.. The effect went away when reading scrambled words). People are better at doing this with open class than closed class words. The closed class words seem to be "invisible" to conscious processing. Interestingly, those with Broca's aphasia perform better than normals with the closed class words. This seems to indicate that the closed class words are not invisible to the aphasics because they were not processing them in the same way. Wernicke's were not any better at it, though, but they had the same problems in both the prose and the scrambled word distractor task condition.

Here is a theory to explain it: Words are interpreted by a module before they are put into a sentential context. Wernicke's patients cannot do the second part.

It seems that Broca's patients have troube parsing closed class words and Wernicke's have trouble with later sentence processing.

p: 191 A priming finding: If you are presented (aurally) with a sentence that shows the insect meaning of "bug," such as He killed beetles and other bugs with pesticides , and shown ANT or SPY directly after it, both meanings of bug are equally primed. About a second and a half later, though, SPY is no longer primed as well. Normals and Wernicke's both behave this way.
Broca's patients show no priming for isolated words and in the cross-modal experiments they always take the most common meaning regardless of the context of the presented sentence. (Swinney et al 1985). Weird, since they seem to have no trouble understanding sentences in general. Why would this happen?

p: 192 Here's a theory: When normals disambiguate, they run through a list of all possible meanings, in order of frequency. This happens too fast for our instruments to pick up. Broca's aphasics search serially through this list slower, and we happen to be picking them up while they are still considering the first in the list, which is the most common.

Dynamic Analysis

Using ERP (Event Related Potentials) scans, you can measure brain activity. It has been found that after seeing a sentence one word at a time, if there are inappropriate words or words in a radically different font, then there will be a surprise peak in the processing.

The Author thinks that this line of research hasn't been very fruitful.

(Jim Davies note 1998: This information may be dated-- I suspect that recent fMRI scans can tell us a great deal.)

JimDavies ( CogSci Summaries Webmaster: jim@jimdavies.org )