La r\'eduction de termes complexes dans les langues de sp\'ecialit\'e

TL;DR

This study investigates how complex terms are reduced in specialty languages like French and Italian, analyzing the phenomena of anaphoric and lexical reduction through statistical methods across different discourse types and domains.

Contribution

It introduces a statistical analysis of term reduction phenomena, defining anaphoric trees and proposing the concept of term lifecycle in specialized language contexts.

Findings

Anaphoric reduction depends on discourse type, not domain or language.

Lexical reduction depends on domain and is common in fast-evolving technical fields.

Anaphoric reduction tends to follow full terms rather than precede them.

Abstract

Our study applies statistical methods to French and Italian corpora to examine the phenomenon of multi-word term reduction in specialty languages. There are two kinds of reduction: anaphoric and lexical. We show that anaphoric reduction depends on the discourse type (vulgarization, pedagogical, specialized) but is independent of both domain and language; that lexical reduction depends on domain and is more frequent in technical, rapidly evolving domains; and that anaphoric reductions tend to follow full terms rather than precede them. We define the notion of the anaphoric tree of the term and study its properties. Concerning lexical reduction, we attempt to prove statistically that there is a notion of term lifecycle, where the full form is progressively replaced by a lexical reduction. ----- Nous \'etudions par des m\'ethodes statistiques sur des corpus fran\c{c}ais et italiens, le ph\'enom\`ene de r\'eduction des termes complexes dans les langues de sp\'ecialit\'e. Il existe deux types de r\'eductions : anaphorique et lexicale. Nous montrons que la r\'eduction anaphorique d\'epend du type de discours (de vulgarisation, p\'edagogique, sp\'ecialis\'e) mais ne d\'epend ni du domaine, ni de la langue, alors que la r\'eduction lexicale d\'epend du domaine et est plus fr\'equente dans les domaines techniques \`a \'evolution rapide. D'autre part, nous montrons que la r\'eduction anaphorique a tendance \`a suivre la forme pleine du terme, nous d\'efinissons une notion d'arbre anaphorique de terme et nous \'etudions ses propri\'et\'es. Concernant la r\'eduction lexicale, nous tentons de d\'emontrer statistiquement qu'il existe une notion de cycle de vie de terme, o\`u la forme pleine est progressivement remplac\'ee par une r\'eduction lexicale.