Concept and Form:
The Cahiers pour l'Analyse
and Contemporary French Thought

This project is funded by an Arts and Humanities Research Council (AHRC) research grant and is supported by the Centre for Research in Modern European Philosophy (CRMEP) and Kingston University's Faculty of Arts and Social Sciences.

In this essay, Antoine Culioli highlights the different problems that accompany any attempt to formalise linguistics, which leads him to suggest the construction of a model that would be adequate to the demands of rigour.

Culioli begins by posing the problem of formalisation of natural languages for the linguist, who is often trapped in a ‘naïve empiricism’ which obscures fundamental theoretical issues. Culioli then points out the dangers of having too great an infatuation with formalisation, which is harmful inasmuch as it may lead one to believe that a machine may be able to accomplish theoretical work in all its economy, or that ‘stenographic symbolisation’ will lead to greater clarity. Moreover, he points out that the desire for interdisciplinarity could lead to incoherence in the use of models, which are simply borrowed from mathematics, and badly accommodated for use in linguistics. This is due to an obfuscation of the theme of linguistics, which is ‘language [langage] apprehended through natural languages [langues naturelles]’ (CpA 9.7:106 ). Irresponsible formalisation leads to a ‘reduction of language’, and impedes the ‘marking of the dialectical relation between language [langage] and languages [langues]’. Thus, the methodological problem of linguistics is to find the ‘logico-mathematical instruments that would allow for an adequate description of linguistic activity [activité langagière] seized across languages’ (CpA 9.7:107).

Culioli then comes up with a preliminary list of problems with the formalisation of linguistics. He elaborates these problems under three headings (‘observables and models’, ‘concepts, terms and symbols’, ‘syntax and semantics’), deepening and complicating the concerns mentioned earlier.

‘Observables and Models’ deals with the relation between a model, the object (of linguistics), and the observer. It points out the challenge of a meta-language, whose terminology must be that of the language of usage,¹ which would lead to unanticipated traps for the linguist. It also poses the question of the relation between a model and its realisation, and subsequently, of the interior-exterior relation of a language to a subject, in a language considered as signifying [signifiante]. This problematises an instrumental conception of language which would not allow for any ‘lapsus’, and would consider style as inessential to the system. Culioli points out that ‘one of the properties of human language is that it lends itself both to Euclidian axiomatic and to poetic image’ (CpA 9.7:108). Finally he notes that it is impossible to pose the problem of observables without a theory of observation, or without questioning the position of the observer. Formal linguistics should not superficially ‘survey [survoler] languages in their generality’, but must consider the full range of what it finds, without exception. It should thus recognize and draw from the fact that no model can be exhaustive.

In the section on ‘Concept, Terms, and Symbols’, Cuiloli poses the problem of the double status of language, which, being both what is in question and the medium wherein it is discussed, leads to various confusions. On the one hand there is the problem of axiomatising a naïve linguistics (which Culioli deems impossible), and on the other, there is an absence of rigour, which produces only a closed re-writing, while strict formal demands disallow prevarications [faux-fuyants]. We thus have an ‘assimilation of the syntactic to the formal’, which is permitted only at the level of ‘first approximation’. Culioli points out that any unity of language is necessarily ‘engaged and ambivalent’ (CpA 9.7:110), and that any algebraic system of operators becomes intertwined in a system of variables on which it depends. This system, in turn, is incorporated in a ‘referential network which is in correspondence with lived situations’. To believe that we can find structures at the surface level of language is to underestimate the ‘parasitical surcharge’ (a concept drawn from Bachelard) of any ‘conceptualisation on language’ (CpA 9.7:111).

Finally, the section on ‘Syntax and Semantics’ questions the reduction of semantics to syntax, and the complete separation of syntax and semantics. It is held that every syntactic transformation leads to a semantic change. Culioli then posits the existence of a ‘lexical filter’ of rules, syntactics, semantics, and rhetorical modulations which are irreducible to syntax. A second filter gives a ‘lexis’ of not-yet ordered terms, which are however compatible with order. The passage from a lexis to an assertion implies a ‘rhetorical’ and a ‘stylistic’ modulation. These involve a partial ordering and subsequent weighing of terms, and give us a ‘pre-terminal sequence’ [séquence préterminale] (CpA 9.7:112-113). Culioli points out, that in a rejection of this model, one can posit the existence of two corresponding syntactic and semantic systems, or conceive of semantics as a ‘hyper-syntax’, thus raising the possibility of a formal semantics.

Culioli contends that the ‘central difficulty of formalisation in linguistics’ resides in discovering its object, constructing its heretofore untried logics, and adapting mathematical concepts to its ends. Mathematical techniques cannot simply be imported into linguistics, but must be invented and constructed [bricoler] (with the help of the mathematician). Culioli emphasises that it is not a question of ‘dipping in a stock of tools [outils]’ but rather of making use of what one finds, where one finds it (CpA 9.7:113).

In the last section of the essay, Culioli gives a quick illustration of a model that would try to avoid the problems listed above. This model would add a theory of predicates that would allow for formal analysis of languages, as well as for analyses of increasing complexity. A topological representation would allow for a better posing of problems concerning modality, and the construction of a complete set of operators would permit algebraic combinations. Finally, Culioli exemplifies his points with the help of a few ‘cam’ structures [structure en ‘came’], which roughly follow a cyclic or spiralling motion so that one returns to a point that is both similar to and different from the initial point (thus going from il, the third-person singular masculine pronoun, we pass through celui, ce and ça, to arrive back at il, which becomes an impersonal invariable neutral pronoun). Culioli maintains that these diagrams are not simply toys but tools with formal rules of usage. They allow us both to pose basic problems, and render diverse languages comparable (CpA 9.7:114-116).

Constructing such models, Culioli claims, is ‘to refuse to reduce language [langage] and to refuse the status of linguistics as nothing but a collection of individual phenomena; it is to permit the posing of theoretical problems, to be constrained to a common metalanguage and to rigorous modes of reasoning.’ (CpA 9.7:117).

In 1970, Culioli republished ‘La formalisation en linguistique’ in a volume co-authored with Michel Pêcheux and Catherine Fuchs, Considérations théoriques à propos du traitement formel du langage [Theoretical Considerations on the Formal Treatment of Language], with eleven annotations made by Culioli, Pêcheux and Fuchs. Also included in this volume was a paper on ‘Lexis et métalexis: Le problème des déterminants’ by Pêcheux and Fuchs, which further elaborated Culioli’s theories of enunciation and lexis.

Notes

1. Culioli uses the term ‘langue U’ to refer to the language of usage. In his 1967 text ‘U ou “Il n’y a pas de métalangage”’ Jacques-Alain Miller developed ideas about a U-language on different principles ↵