New connections between the cognitive and computational sciences: some avenues for exploring the possibility of language acquisition
Speaker: Geoffrey Pullum
Institution: University of Edinburgh
Abstract:
I will call a clause in English a "give-away clause" iff its subject (i) contains an auxiliary verb and (ii) is preceded in the clause by an auxiliary verb. Give-away clauses have an interesting role in the history of theorizing about human language acquisition. If it were not for give-away clauses, one could defend a statement about English that I will call the Wrong Generalization: that interrogative (and other auxiliary-initial) clauses differ from their related declarative clauses in that the first auxiliary in the declarative counterpart is placed at the beginning of the clause in the auxiliary-initial counterpart. Give-away clauses crucially refute this, because the extra auxiliary contained in the subject of the declarative version is not the one that appears clause-initially.
It has been proposed that give-away clauses are so rare in ordinary language use that many speakers must have learned English without ever hearing one (the Rarity Claim), and hence that some universal principle barring the Wrong Generalization must be innate. This raises a number of interesting lines of research, some well explored and others not.
One might try to formulate the right innate universal principle and show that it does really help to make language learning feasible (nobody really knows). If this can be done, it might be due to some deep-seated formal property of syntactic structure or some computational property of human processing mechanisms. Alternatively, one might try to find out whether the Rarity Claim is true (nobody really knows this either). If it is not, we would need to know whether give-away clauses are frequent enough that learners actually do make use of them. If the Rarity Claim is true, one could investigate computationally whether English can be learned from an input in which give-away clauses are entirely lacking (there are some indications that the answer is yes), and if so, to understand why. This talk discusses developments in a number of such lines of investigation.
