1. What is the syntax component responsible for?

• order of words
• lexical syntax: making sure that the obligatory arguments of major class items are given, e.g.
• intransitive verbs: no objects, subject (the dog runs)
• intransitive verbs with no subject: (it rains)
• transitive verbs: eat a pie
• ditransitives: put the book on the table, give someone a book, promise someone to do something
• agreement
• works to assist semantic interpretation (disambiguation etc.)

See Hagoort, Brown and Osterhout (1999) for an overview.

N400. Kutas and Hillyard (1980) discovered that semantically anomalous combinations of verb and object causes a negative peak to occur in the electroencephalograph signal at about 400 ms after the anomaly occurs (the N400 component).

The pizza was too hot to eat: small N400
The pizza was too hot to drink: larger N400
The pizza was too hot to cry: big N400

Ganis et al. (1996) found that semantic violations in the form of pictures elicit N400 effects that are no unlike the N400 effects found with linguistic input, support the view that the N400 reflects semantic anomalies.

Left anterior negativities (see Hagoort et al. section 9.3.1.1): more frontal, left-lateralized, in same timescale as N400 (300-500 ms post-stimulus). Observed if the syntactic context requires the next word to be of a particular class (e.g. a noun), but a different class is presented (e.g. a verb). Similar early negativities are observed with number, case, gender and tense mismatches. As if the N400 is a response to "surprises" or "anomalies" rather generally, including linguistic anomalies.

P600/Syntactic positive shift: response to a syntactic constraint violation, e.g. agreement violation, phrase-structure violations, subcategorization violations, subjacency violations, empty-category principle violations. More posterior/parietal location.

Lesion data supports a role for the left anterior superior temporal gyrus in the neural circuitry for parsing (Dronkers). Intercranial electrode recordings also support this. N400 is probably generated in anterior medial-temporal lobe.

3. Associationist models of syntax

Say word 1 (a word that an utterance can begin with): the
then
say word 2 (a word associated with the previous word): dog
then
say word 3 (associated with the previous word): stole
then
say word 4 (associated with the previous word): a
then
say word 5 (associated with the previous word): pie
then
finish.

Usually modelled as a finite-state transition network. Some animal calls might also be modelled reasonably well using finite-state transition networks (e.g. Okanoya 2004 on Bengalese finch song) or not so well (e.g. Holy and Guo 2005 on male mice songs). For human languages, though, there are problems with finite-state modelling (Chomsky): languages requiring centre-embedding, or matching up of an indefinite number of earlier and later parts, cannot be modelled as a finite-state transition network:
 

Incomplete clause beginnings ...						Completions ...
[the malt										lay in the house that Jack built]
	[that the rat								ate]
		[that the cat						ate]
			[that the dog				chased]
				[that the cow		tossed]
					etc.

Processing such sentences requires working memory in which to store the incomplete constituents. For discussion of the workability of associationist models of such data (e.g. using connectionist techniques), see Allen and Seidenberg (1999). Recall that in lecture 1, recursiveness of syntactic structure, rather than the mere existence of structure, seemed to be one of the few characteristics of language that is apparently unique to humans (a claim of Hauser, Chomsky and Fitch 2002, especially, but refuted by e.g. Parker

Just et al. (1996) conducted a study in which subjects read written sentences aloud, and then answered a question about them:

1. Control: consonant strings, e.g. Pws ntkgqrfm zjkjrng kwtdc sbfght swn mjrdbxd kgt mxbtq
2. Active conjoined sentences, with no embedded clause, e.g. The reporter attacked the senator and admitted the error.
3. Subject relative clause, e.g. The reporter that attacked the senator admitted the error.
4. Object relative clause, e.g. The reporter that the senator attacked admitted the error.

4. Module conclusion

Although studies of the peripheral parts of the language organ are fairly advanced, our understanding of the central processes of language (i.e. in the brain) are at an early stage. The future is promising. Nevertheless, enough has already been learned to gain a general picture of brain function in various linguistic processes. The results of newer methods (e.g. "brain imaging") are proving remarkably consistent with inferences about the language organ made previously on the basis of studies of brain-injured patients.

References

Allen, J. and M. S. Seidenberg (1999) The emergence of grammaticality in connectionist networks. In B. MacWhinney, ed. The Emergence of Language. Mahwah, New Jersey: Lawrence Erlbaum Associates. 115-151.

Hagoort, P., C. M. Brown and L. Osterhout (1999) The neurocognition of syntactic processing. In C. M. Brown and P. Hagoort, eds. The Neurocognition of Language. Oxford University Press. 273-316.

Holy, T. E. and Z. Guo (2005) Ultrasonic songs of male mice. PLoS Biology, 3 (12), e386. 2177-2186.

Keller, T. A., Carpenter, P. A., & Just, M. A. (2001). The neural bases of sentence comprehension: a fMRI examination of syntactic and lexical processing. Cerebral Cortex 11, 223-237.

Just, M. A., P. A. Carpenter, T. A. Keller, W. F. Eddy, and K. R. Thulborn (1996) Brain activation modulated by sentence comprehension. Science 274, 114-116.

Okanoya, K. (2004) The Bengalese Finch: A Window on the Behavioral Neurobiology of Birdsong Syntax. Annals of the New York Academy of Sciences 1016, 724-735.
 
Parker, A. R. (2006) Evolving the narrow language faculty: was recursion the pivotal step? Paper presented at Evolution of Language, Sixth International Conference, Rome, 12-15 April 2006. http://www.tech.plymouth.ac.uk/socce/evolang6/parker.doc