Event

The University of the West Indies, Open Lecture Committee will host a Distinguished Open Lecture on Thursday 9^th October, 2008 at the Learning Resource Centre (LRC) from 5:30 p.m. Professor Ewart A.C. Thomas , Professor of Psychology at Stanford University USA will deliver the lecture titled “Mathematical and Empirical Studies of English/Creole Language Variation”.

Abstract

A core issue in the analysis of language mixing concerns the number of Creole/English language varieties that are deemed ‘allowable’ in a speech community. Implicational scaling has been one of the most productive approaches to the study of these linguistic constraints ever since its initial application by D. DeCamp (1971) to the Jamaican Creole continuum. His model assumes that linguistic features can be ordered on the Creole-English continuum such that, if a speaker typically uses the ‘Creole’ variant of a given feature (e.g., nyam rather than eat, or dem rather than them), then the speaker will typically use the ‘Creole’ variant of all features that are less Creole on the continuum than the given feature. His model predicts a surprising regularity relating the number, an, of observable language varieties to the number, n, of linguistic features or variables characterizing a language variety, namely, An = n + 1.

In recent empirical and theoretical work, Hubert Devonish and myself take a different approach. We model the probability, P, that an ‘allowable’ sentence (i.e., a sentence with English or Creole variants that is judged as ‘acceptable’ in the speech community) of length n can be extended by the addition of one variant (English or Creole) to form an allowable sentence of length n + 1. We prove that An = n + 1 when the function, P, is linear and the ‘compatibility’ between English and Creole is neither too low nor too high. Further, computer simulations show that, when P is nonlinear and the compatibility is not too high, the number, An, of language varieties tends to 2 as n becomes very large. These results add to our understanding of language mixing, at least by offering a rigorous derivation of a family of data regularities, including the (n + 1)-rule. Whether other model-derived regularities will be discovered in future empirical studies remains to be seen. In the meantime, we need to better interpret the ‘linearity’ and ‘compatibility’ parameters of the model in linguistic or socio-psychological terms so as to increase the policy-relevance of our research.”