Natural Language Processing and Computational Linguistics 2

3The Sphere of Discourse and Text

3.1. Discourse analysis and pragmatics

3.1.1. Fundamental concepts

As noted by [VAN 85], the practice of analyzing speech, literature or simply language goes back more than 2000 years. The first works had a marked normative dimension and were located in the domain of rhetoric. They were intended to formulate rules for planning, organizing and delivering spoken communications in a legal or political context.

In the context of modern works in linguistics, notably since Zellig Harris [HAR 52], it is generally accepted that the sentence cannot be the maximal unit of linguistics studies. Consequently, many linguistic works consider that linguistic productions are made up of a set of interconnected utterances whose interpretation depends on the situation of communication. Some use the term discourse to designate such a set. Unfortunately, this term is one of the most polysemic terms that exists. The most precise definitions that have been proposed for it are those that have been formulated negatively in opposition to other linguistic entities.

One of the reasons behind this divergence is the multitude of movements and disciplines that have an interest in extra-sentential phenomena: functional linguistics, cognitive linguistics, sociolinguistics, textual linguistics, discourse analysis, etc. The common point between all of these approaches is the rejection of the Chomskyan idea expressed in his Standard Theory [CHO 57] according to which the sentence ...