Monkey syntax

This talk will offer a tender and sensitive exploration of the language and thoughts of arboreal primates, with lots of color photos of them in the wild... No, it won't; that's a wicked lie. The title is a scandalous example of false advertising, and the speaker, who does not even own a monkey, should be thoroughly ashamed of himself. What he does offer, though, is a broadly-targeted introduction --- designed for linguists, logicians, mathematical psychologists, developmental psycholinguists, and computer scientists --- to some not very well known mathematical results which appear to be highly relevant to ongoing experimental work on precursors to syntax in non-human primates.

Recent experimental work in Marc Hauser's lab at Harvard has explored cottontop tamarins' abilities to learn regularities in vocal stimuli (Hauser, Chomsky & Fitch, Science 298, 5598 [22 November 2002], p.1578; Fitch & Hauser, Science 303, 5656 [16 January 2004], 377-380; O'Donnell, Hauser & Fitch, Trends in Cognitive Sciences 9, 6 [2005], 284--289). This work refers to the distinction between the finite-state (FS) and context-free (CF) classes of stringsets. Infinitely many classes of hierarchically stratified stringsets fall below (are proper subclasses of) FS. The talk will survey their properties. Some of these classes may plausibly be correlated with abstract cognitive abilities. Fitch & Hauser unfortunately contrasted a stringset that belongs to ALL of these classes with a stringset that belongs to NONE of them, so their positive result (distinguishing tamarins from undergraduates on learning the latter) is less surprising than it might have been --- their exemplar languages are not representative. Experiments involving more minimal language-theoretic contrasts can be envisaged. The aim of this talk is to open a discussion about how to clarify and focus the possibilities for such research.

Those who like to get a head start on things may want to consult the Pullum & Rogers paper at <http://people.ucsc.edu/~pullum/MonkeyMath.pdf> before the talk, as well as the above-cited papers by Hauser and colleagues. But others will prefer to just put their feet up and listen to an informal talk about the results before they decide whether to look at denser presentations; and that's OK too.

Geoff Pullum's work is supported by the Radcliffe Institute for Advanced Study at Harvard University, where he is the Constance E. Smith fellow this year. In real life he is Professor of Linguistics at the University of California, Santa Cruz. This lecture is based on joint research with a former IRCS postdoc, Professor James Rogers (Radcliffe Institute and Earlham College). Useful comments and substantive contributions by Barbara Scholz, Stuart Shieber, Dmitri Tymoczko, and several others are gratefully acknowledged.