Introduction

1Categorization is a basic cognitive process. Humans use language to categorize, primarily through the nominal category (Mignot [2017]). More particularly, categorization enables them to better organize the world around them in order to facilitate learning and decision-making (Lakoff [1987]). Metaphorically, humans resort to cognitive boundaries to define their everyday experiences and bring them together according to the properties they have in common, or the “family resemblances” (Familienähnlichkeiten) that they share, i.e. networks of more or less significant similarities (Wittgenstein [1953]).

2The classical view of categorization considers categories as clear-cut and homogeneous – no element can belong to two distinct categories, and all the elements within a category have the same status. However, Rosch’s [1973] research in psychology posits that categories are organized around a central element called “prototype”, with other elements moving progressively further from it. Thus, categories include marginal elements – for example, a chair is considered a more prototypical piece of furniture than a telephone (Rosch [1975]). In linguistics, various authors have taken up the notion of prototype. In his work on notional domains, Culioli [1990] puts forward the notions of “inside” and “outside”, separated by a boundary, while Lakoff [1987] defines cognitive models as categories built around a membership gradient, a generator (i.e. a specific, central member of the class) and common features.

3Prototype theory is used more or less explicitly in the study of many lexical units, whether it is kinship (Lounsbury [1964]), color (Berlin & Kay [1969]) or taxonomy (Wierzbicka [1992]).

4From a metalinguistic perspective, Taylor [1998] applies categorial gradience to linguistic description. He posits that a specific lexical unit should not be categorized according to strict selection criteria (which would imply a clear boundary and equality of the members within the class), but according to tests of prototypicality which make it possible to distinguish central members from marginal ones. This point of view is taken up by Aarts [2007], who puts forward the notion of subsective gradience: lexical categories are organized around prototypes, and within these classes some elements are closer to the center or to the periphery (e.g. utter is a less typical adjective than happy).

5Subsective gradience can be the source of intersective gradience: a peripheral element within a linguistic category can also be at the margin of another category, and therefore be a kind of hybrid, such as near (both an adjective and a preposition). This more flexible view of parts of speech also calls into question some traditional cases of conversion: in real good or come quick, are the units real and quick adverbs derived from adjectives, adjectives used as adverbs, or units belonging to both categories?

6This thematic issue of Lexis tackles both linguistic and metalinguistic categorization and the problems it poses, with all authors showing how difficult it is to work with pre-established, clear-cut classes when dealing with specific linguistic phenomena.

7To open this issue, Vincent Hugou and Mathilde Pinson focus on the categories of adjectives and past participles, through the study of the sequence very Xed by. They posit the existence of a continuum featuring an adjectival end and a verbal end, on which the construction may be placed. From a diachronic perspective, the authors show that the construction has progressively opened to “affect verbs”, thus licensing more and more participles (as opposed to adjectives). This study leads them to the conclusion that there is no clear-cut distinction between these two traditional syntactic categories, and that the construction actually transcends strict categorization.

8Marie Turlais then explores the nominal category though the case of deverbal phrasal nouns in –ing. She shows that the subcategory of compound nouns is gradient, with the nouns under study lying at the margins of this class. Those deverbal phrasal nouns are also original in that they exhibit great variation as to where the plural suffix may fall, which suggests a difference in structure and lexical integration.

9In the third paper of this issue, Martin Konvicka undertakes a study of the causal connector because when it is used in the sequence because X. This connector can be used as a conjunction, a preposition or as a member of a third category. The author argues that the connector does not belong to any of these categories prior to its use: it has an abstract category potential that is only realized in discourse. This shows that linguistic categories should be seen as emerging from language use, and not as an independent, essentialist property of language.

10Finally, Angelina Aleksandrova tackles linguistic categorization in sports terminology. Her contrastive study of CrossFit terms in English and French notably shows that lexical items denoting movement terms in sports do not refer to the same realities. Although the terms used in specialized domains have to be unequivocal, “naïve” speakers using sports terms will rather use familiar words denoting pre-established categories to translate terms they have never encountered before.

11Far from being isolated cases, all the issues tackled here tend to show how complex language is. Even though categorization is necessary in order to describe the world and language itself, the process has limitations when it comes to classifying marginal phenomena. The question of category membership for some linguistic units, as well as the definition of central features and boundaries for linguistic categories, remain major issues that must be accounted for when describing and analyzing language. Through the study of intermediate cases, the papers from this thematic issue prove that categorization and category membership must be conceived as emerging from a radial cognitive process, rather than from a dualistic perspective – or, as Ross [1981] writes, that “things in syntax (and elsewhere in linguistics, for that matter) are rarely black/white, but are rather usually of varying shades of grey”.