Our contribution compares the lexical characteristics and combinatorial properties of nouns in two kinds of entries from the Encyclopædia Universalis (2005). We split the Encyclopædia into two big subcorpora, one for texts dealing with the humanities, the other for texts representing the natural sciences. We compared these subcorpora in terms of lexical overlaps, lexically distinctive features and combinatorial properties of certain nouns. On the basis our tools to determine the combinatory profile of words and to compute degrees of similarity between them, profound lexical and semantic differences between both “academic cultures” emerged.

In this paper, we propose a first step leading to the description of the lexicon of scientific language by identifying a transdisciplinary scientific lexicon (TSL). The TSL is domain independent and forms a central lexical core of all domains; it is at the center of argumentation we find in scientific discourse as well as its structuring. In order to gather the transdisciplinary lexicon, we use natural language processing (NLP) tools and statistical techniques; central to our method is the calcul des spécificités (specificity measure) put forward by Lafon (1980). By using NLP tools, we want to verify if it is possible to quickly and simply complement existing lexical resources without much manual intervention. We conclude our study by exploring an observation made by Phal (1971) about collocations and scientific discourse. We focus here on V-N collocations revolving around nouns taken from our TSL.

This study aims at developing a method for investigating the invariants across different scientific discourses. The paper illustrates a model analysis for processing general scientific vocabulary by applying three complementary approaches: cross-disciplinary, onomasiological and phraseological. The combination of these approaches allows us to bypass traditional methods generally based on a lexical analysis of a single domain. We first define the borderline between the specialised scientific language features and those features that are common to various scientific discourses, then we present a corpus-based study of a combinatorial profile of lexical resources used for formulating general scientific ideas. This method thus provides a framework for the modelling of general scientific language resources.

This article presents the qualitative analysis of a corpus of published linguistics articles. Two types of lexical markers were studied: those signaling convergence (e.g. conformément à nos attentes) or signaling divergence (e.g. contre toute attente), in relation to the expectations of the writer-researcher or of the larger scientific community. The objective of this study is to examine to what extent these markers reveal the way the writer-researcher constructs and validates knowledge in their article. The corpus analysis found several functions for these markers, thus highlighting a wide variety of possible expectations. As a result, we argue that such an analysis must not only see these markers as linked to other terms but must also to take into account their relation to the entire text. Therefore, markers of convergence and divergence may not necessarily be the locus of an implied scientific reasoning. However, they potentially express various epistemological styles.