Society, Cognition and Language. Understanding human thought and behaviour.



Dylan Glynn
University of Paris VIII

The aim of linguistics is to explain the phenomenon of language – how is it possible that a child learns perfectly a language with seeming ease and possesses an intuitive sense of right and wrong, where philologists have struggled (and thus far failed) for eons to write a grammar that is descriptively and predictively accurate. This is perhaps the only point that all linguists agree upon. Everything from what is language to what is right or wrong is debated and has been debated since the beginning of language science. This talk will argue that, perhaps, this is finally set to change.

Arguably, the fundamental theoretical questions of language science are:

Grammaticality: How do we judge grammatical from agrammatical, and how do we acquire that judgment?

Grammar: What is the structure of language and how to we acquire that structure?


Arguably, the fundamental methodological problems of language science are:

Observability: Neither grammar nor grammaticality are observable. How can we test hypotheses with no direct empirical evidence?

Variability: All the indirect evidence we have reveals significant variation in both grammar and grammaticality. How can we test hypotheses when exceptions are the rule?


The talk will present the usage-based model (Hopper 1987, Langacker 1987) of language and attempt to show how this model may unite the theoretical questions and, in part, overcome the methodological problems.

The usage-based model assumes that there is no grammar of a language sensu stricto, but that each individual in a speech community possess a competence. Langauge grammar is, therefore, merely a generalisation across individual competences. In this view of language, the structuring force of language is usage (between individuals of a speech community). In other words, the langue / competence is a result of the parole / performance and not the reverse, which is assumption sine qua non of structuralist models of language. This reversal of the relationship between structure and use allows us to answer the same fundamental theoretical questions but renders the methodological issues entirely non-problematic.

The talk will present an example of a study that shows how usage-based data can be used to identify conceptual structures relative to their social use. The corpus-driven study will examine the concept of ANGER in American and British English and compare its results with psycholinguistic results. The method, one of many usage-based methods, can be employed to describe any language phenomenon – identifying conceptual structure in behavioural usage. It follows that this method, mutatis mutandis, can be employed to test most predictive and explanatory theories of language structure, crossing the langue-parole divide.