SIL International Home

The SIL French/English Linguistic Glossary

English definitions of key terms in the Theory of Enunciative Operations

Théorie des opérations énonciatives : définitions, terminologie, explications


English translations

Coordination and proof-reading


The enunciator is the origin of subjective location in the Theory of Enunciative Operations and is one of the two co-ordinates of the SITUATION OF UTTERING, which is the ABSOLUTE ORIGIN OF LOCATING OPERATIONS. The enunciator is an integral part of the LOCATING SYSTEM, which is at the core of the theory. The enunciator is a theoretical construct elaborated from the role of the {LOCUTOR} or {SPEAKER} but it is not to be confused with the speaker: it embodies the enunciative position from which a linguistic representation is considered. Firstly it is LOCATED RELATIVE TO the speaker; secondly it is related to the CO-ENUNCIATOR, which is constructed relative to the speaker. The variable nature of these two relations makes it possible to represent the different enunciative positions of the speaker with respect to the utterance. For instance a locating relation of DISCONNECTION can be marked between the speaker and the enunciator as in the French “journalistic conditional” when something is said without being endorsed (Kyoto: Poutine aurait toujours l’intention de ratifier [Kyoto: Putin reportedly still intends to ratify]); rhetorical questions are an example of IDENTIFYING LOCATION between enunciator and co-enunciator, (Ne sont-ils pas magnifiques ? [Aren’t they magnificent?]).

The enunciator in the Theory of Enunciative Operations is different from the concept of speaker as used by E. Benveniste (who rarely used the term enunciative subject) and the difference is even more marked when we compare with the enunciator as this term is used by O. Ducrot or in pragmatics; in the Theory of Enunciative Operations, ENUNCIATION is not a question of discourse, but concerns primarily the construction of the utterance (énoncé), which is constructed in language before being constructed by any particular speaker.


Benveniste 1970, “L'appareil formel de l'énonciation”, Langages n°17, Larousse, in Problèmes de linguistique générale, II, Gallimard, Paris, 79-88.

Culioli 1986, “Stabilité et déformabilité en linguistique”, Etudes de Lettres, Langages et Connaissances, Université de Lausanne, reprinted in T. 1 : 127-134.

De Vogüé 1992, “Culioli après Benveniste : énonciation, langage, intégration”, LINX n° 26, Lectures d’Émile Benveniste. Paris 10, 77-108.

Franckel 1989, Étude de quelques marqueurs aspectuels du français. Genève : Droz. 20-24.

Simonin 1984, “De la nécessité de distinguer énonciateur et locuteur dans une théorie énonciative”, DRLAV n° 30, La Ronde des sujets, Paris 8 & CNRS, 55-62.


“A predicative relation (which is itself the result of a series of locating operations) is located relative to an event index, which draws its value from its own location relative to the time of speech, which is itself located relative to an origin locator (the absolute locator) [...]”
(Translated from Culioli 1986, T1:130).

“Thus we obtain the enunciative subject, reconstructed as a result of this analytical process leading back from the utterances to the mechanisms that underlie them. Culioli insists on the fact that these subjects are nothing but abstract parameters: the linguist goes back to the origin locator, not to the source of the determination of the enunciative construction and even less to the source of the act of speaking, i.e. the speaker-subject […] the enunciator-locator is, in the strictest sense, the origin of the operations of location involved in the enunciative process; it is not the origin of the enunciative process as such, whose only source is in language itself.”
(Translated from De Vogüé 1992:81-82).


That faculty of the human species which concerns the linguist in its dimension as a symbolic enunciative activity, as an activity producing meaningful representations. Unlike generative grammar, the Theory of Enunciative Operations distinguishes between language and LANGUAGES; a language is only the trace, the manifestation of {LANGUAGE ACTIVITY}, and not a state of the language ability; LANGUAGE MECHANISMS are reproduced (or reconstructed) from the data constituted by UTTERANCES.


Culioli 1974, “Comment tenter de construire un modèle logique adéquat à la description des langues naturelles”, in J. David & R. Martin eds., Actes du Colloque du Centre d’Analyse syntaxique de l’Université de Metz, reprinted in T. 2 : 53-66.

De Vogüé 1992, “Culioli après Benveniste : énonciation, langage, intégration”, LINX n° 26, Lectures d’Émile Benveniste. Paris 10, 77-108.


“It is because enunciation is conceived of as a process of constitution of meaning (and not as the act of a speaker) that language must be conceived of as an activity. Meaning is construed utterance by utterance. Consequently, language cannot be conceived of as an interpretative grid placed on the (semiotic) world, nor even as a discursive motor that human subjects avail themselves of to invest the world with a meaning: it is a machine suited to the construal of meaning […]. If language is an activity […] it can only be apprehended through what this activity produces: utterances. Culioli does not choose to study a language as a structured system […], even less the use of this language in discourse, rather he analyses utterances in a language and the interpretative value that these utterances construe.”
(Translated from De Vogüé 1992:99-100)

“In all these procedures, one should not confuse: language (adj.) [Fr. langagier] (which concerns language activity); linguistic (a term referring to complex operations that leave traces which are textual configurations) and metalinguistic (which refers to the activity of the linguist, to the extent that he or she describes, represents and possibly simulates the phenomena observed (production and products) in the field of language activity and linguistic operations).”
(Translated from Culioli 1974, T.2:53-54)

“I shall define linguistics as the science whose goal is to apprehend language through the diversity of natural languages. The definition is itself a research programme, since it sets a goal and therefore influences theorization and methodology […] Language, which is a meaningful representational activity, is only accessible through texts, that is, through patterns of markers which are themselves traces of underlying operations. Now even the most rudimentary attempts at observation and classification show that, beyond the diversity of categories and surface features, stable analogic features do exist. In short, languages are not irreducibly specific […] The goal is not to construct a universal grammar but to re-construct, by a theoretical and formal process of a foundational sort, the primitive notions, elementary operations, rules and schemata which generate grammatical categories and patterns specific to each language. In short, the goal is to find invariants which found and regulate language activity, as this activity manifests itself through the meaningful configurations of different languages.”
(Culioli 1978, T.1:72)

“The restriction (“apprehended…”) is, as we have seen, indispensable, since we reject the idea that linguistics can be the science of language; the plural (“through the diversity of natural languages”) prevents the linguist from falling into the naïve observation of one language (his or her own most often, or an “important” language in its written form) as the essence of all languages […] It is preferable to stress that the linguist cannot establish the theory of language, but that he or she has a role to play and that his or her field is the relation that obtains between language activity and the specific configurations that are constituted by the utterances in the various natural languages.”
(Translated from Culioli 1979, T.2:68-69)

“So linguistics can only be the search for what is generalizable, in the sense that a property, categories, operations, said to be general, must be able to be (re)constructed, by a process of reasoning, each time it is necessary.”
(Translated from Culioli 1979, T.2:72)

“Linguistics aims at the description of (individual) languages. But its primary task, which is the basis for its description of particular languages, is to characterize language activity.”
(Translated from De Vogüé 1992:100)


Languages are the sets of oral and written productions resulting from LANGUAGE ACTIVITY. NATURAL LANGUAGES are the visible and audible part of language activity. It is the relation between LANGUAGE and languages that is essential and the term “language” (“langue”) refers not to an abstract ability but to the diversity of languages.


Culioli 1974, “Comment tenter de construire un modèle logique adéquat à la description des langues naturelles”, in J. David & R. Martin eds., Actes du Colloque du Centre d’Analyse syntaxique de l’Université de Metz, reprinted in T. 2 : 53-66.

Culioli 1979, “Conditions d’utilisation des données issues de plusieurs langues naturelles”, Modèles Linguistiques, Tome I, fasc. 1, Université de Lille III, 89-103, reprinted in T. 2 : 67-82.


Linguistics is the scientific study of LANGUAGE ACTIVITY apprehended through the diversity of NATURAL LANGUAGES and the data from oral and written texts seen as ARRANGEMENTS OF MARKERS. Linguistics requires a METALANGUAGE, a SYSTEM OF METALINGUISTIC REPRESENTATION simulating the relation between natural languages and language.


Culioli 1979, “Conditions d’utilisation des données issues de plusieurs langues naturelles”, Modèles Linguistiques, Tome I, fasc. 1, Université de Lille III, 89-103, reprinted in T. 2 : 67-82.

Culioli 1982, “Rôle des représentations métalinguistiques en syntaxe”, Session plénière du XIIIe Congrès International des Linguistes, Tokyo, reprinted in T. 2 : 95-114.

Gilbert 1993, “La théorie des Opérations Énonciatives d’Antoine Culioli”, in P. Cotte et al., Les théories de la grammaire anglaise en France. Paris : Hachette. 63-66.


This is a fundamental relation in the Theory of Enunciative Operations, in which all terms (NOTIONS, RELATIONS, ENUNCIATIVE COORDINATES) are necessarily {LOCATED} relative to another term which functions as a {LOCATOR}. The concept of location is perhaps the most central concept in the theory of enunciative operations. It is a linguistic or LANGUAGE OPERATION which is involved in the production and recognition of utterances and constitutes the core of the construction of the {REFERENTIAL VALUES} of these utterances. The operation of location is the source of the determination of the different notions present in an utterance: it is involved in the very construction of these notions, in the determination of  nouns and verbs, in relations between notions, and finally in the construction of an utterance relative to an origin (see ENUNCIATOR and SITUATION OF UTTERING). If the locator is not itself an absolute origin locator (i.e. the situation of uttering), then it is located relative to another term which serves as a locator. No term is isolated; all terms can only acquire a referential value if they are part of a locating system. The operation of locating is symbolised by “” and can have three values: {IDENTIFICATION} (symbolised by “=”),{DIFFERENTIATION} (symbolised by “≠”) and {DISCONNECTION} (or DETACHMENT) (symbolised by “ω”).

In the case of enunciative location (relative to the situation of uttering, i.e. the absolute origin of locating operations), the main MARKERS of location relative to the SUBJECTIVE COORDINATE S are the personal pronouns (identification: je[I]; differentiation: tu[thou/you]; disconnection: il/elle[he/she/it]), and relative to the SPATIO-TEMPORAL COORDINATE T, the aspectuo-temporal forms:

The operation of location can have a fourth value, which is a mixed value, noted * (STAR or ASTERISK LOCATION), in those cases where there is a combination of several values: both disconnection and identification or disconnection and differentiation. This mixed value is to be found in fictive location, for example in hypothetical utterances or else in the so-called historical present or imperfect. The construction of a fictive origin locator also enters into the AORISTIC concept, based on disconnection and considered as a category that has aspectual properties and is associated with various operations of {LOCALISATION} and QUANTIFICATION/QUALIFICATION; this concept can be used to account for a great variety of markers in the field of aspect, modality, inter-sentential or inter-clausal location etc.

In cases where the relation of location concerns the terms of the utterance (i.e. predicative location), the locator is the term which introduces a form of determination that makes it possible to pose the existence of the located term or locatum (i.e. localisation) and / or to provide qualification of the located term (i.e. determination of a property of that term). Examples:


Bouscaren & Chuquet 1988, Grammaire et textes anglais : Guide pour l’analyse linguistique. Gap : Ophrys. 131-133.

Chuquet 1994, Le présent de narration en anglais et en français, Linguistique contrastive et traduction, N° spécial, Gap : Ophrys. 51-62.

Culioli 1978, “Valeurs aspectuelles et opérations énonciatives : l’aoristique”, in J. David & R. Martin eds., La notion d’aspect, Actes du Colloque de Metz, 181-193, reprinted in T. 2 : 127-144.

Franckel & Paillard 1998, “Aspects de la théorie d’Antoine Culioli”, Langages n° 129, Diversité de la (des) science(s) du langage aujourd’hui. Paris : Larousse, 52-63.


“Uttering means constructing a system of location, which implies a frame in which a web of referential values is set up, through orientation and determination.”
(Translated from Culioli 1973, T.2:49)

“The close study of different languages and the theoretical description of apparently unrelated phenomena has led me to posit a fundamental relation, constructed by a primitive operation of location. The concept of location (repérage) involves both relative location and determination. To say that x is located relative to (or with respect to) y means that y is the locator (or reference point) for x, whether y is located (repéré) relative to another locator (repère) or to an original locator, or whether it is itself the origin […] Within a referential system or frame, when a term x is located relative to a term y, the operation provides x with a referential value (i.e. it determines a property of x) which it did not have before […] The essential point is that an object only acquires a determined value through a system of location […] The relation of location is always binary; if three terms are involved, then a second binary relation is constructed on top of the first relation […] The locating operator is noted epsilon (), meaning “is located relative to”. So < x ( ) > describes an ongoing operation where x is the located term, from which a reference point (locator) — let us say y — is then constructed, leading to the relation: < x y >, i.e. “x is located relative to y” […] The concept of location makes it possible to develop a holistic theory and to account for cross-categorial questions (for instance, by linking up determination, modality and aspect).”
(Translated from Culioli 1982, T.2:97-98)

“The close study of a wide variety of languages has led me to posit as one of the primary operators an operator of location, noted , with its mirror operator ⊇, such as if x y = y x, there is a relation of identification (x = y); if x y = y ⊇ x, there is a relation of abstract location, where y is the locator (repère) and x the locatum (repéré); one may also easily derive from this: y ≠ x. A third value is disconnected from the first two, producing a relation of location that is neither identification nor differentiation. Within a referential plane P containing a locator r, let two terms a and b enter into the following relation: a = r, b ≠ r, therefore a ≠ b. A plane Q is then constructed, such as Q ω P (where ω signifies the value of disconnection taken on by the locating operator). This entails that for any term c within Q, the relations c ω a and c ω b will prevail. Finally there is a fourth value, produced by a combination of the first three; noted * (star), it takes on the value: neither identical nor different, or identical or different (i.e. ω or = or ≠). The operation of location can serve to construct such categories as person, modality, deictics etc ., depending on the notional domains involved.”
(Translated from Culioli 1978, T.2:129-130)


A marker is a linguistic representative of a LANGUAGE OPERATION. A marker can correspond to a morpheme, a periphrasis, a syntactic pattern etc.


Culioli 1986, “La frontière”, Lignes de partage, Cahiers Charles V n° 8, Université Paris VII, 161-170, reprinted in T. 1 : 83-90.

Culioli 1993, “Les modalités d’expression de la temporalité sont-elles révélatrices de spécificités culturelles ?”, Interfaces n° 5, CRDP de Paris, reprinted in T. 2 : 159-178.

Franckel 1989, Étude de quelques marqueurs aspectuels du français. Genève : Droz. 10-16.


“If we start from the theoretical principle, set out elsewhere, that there are three levels of representation (level I, language [notions, operations]; level II, languages [patterns of markers]; level III, the metalinguistic level), we shall posit that level II units are the markers of level I operations (the level to which we do not have access, except through the traces constituted by the markers). Metalinguistic work consists in reconstructing the operations and the chains of operations underlying a given empirical form, which is their marker.”
(Translated from Culioli 1986, T.1:129)

“Notions are inaccessible representations as such, but we can apprehend them through the material traces constituted by texts […]  These textual traces provide the second level […], namely level II representations; these may be described as arrangements or patterns of markers, where pattern indicates that we are not dealing with arbitrary forms (there are rules of well-formedness), while the term marker refers to the perceptible indication of mental operations, which lead from level I, of which we have only a trace, to level II, which is precisely the place where the traces are arranged in the form of utterances.”
(Translated from Culioli 1993, T.2:162)


A {SYSTEM OF METALINGUISTIC REPRESENTATION} to represent the relation between specific LANGUAGES and LANGUAGE as such. The terms of this system (symbols, categories, OPERATIONS etc.) are as few as possible, explicitly defined, generalisable and robust. They allow the linguist to account for phenomena and to carry out calculations going beyond mere semantic intuition.


Culioli 1974, “Comment tenter de construire un modèle logique adéquat à la description des langues naturelles”, in J. David & R. Martin eds., Actes du Colloque du Centre d’Analyse syntaxique de l’Université de Metz, reprinted in T. 2 : 53-66.

Culioli 1989, “Representation, referential processes, and regulation”, in J. Montangero & A. Tryphon, eds, Language and Cognition, Foundation Archives Jean Piaget, Cahier n° 10, Genève, reprinted in T. 1 : 177-213.


“Developing a system of metalinguistic representation in order to describe (that is, to represent by using a system of re-writing) and to calculate necessarily implies inquiring into the question of abstraction based on observations, of regularity and homogeneity as well as the stable, finite and closed nature of the metalinguistic discourse that one is constructing and using.”
(Translated from Culioli 1987, T.1:115)

“Saying that a model is metalinguistic means that anything that is relevant must be representable by the writing system chosen: obviously it is important not to have unanalysable symbols. All the operators, all the categories must either be posited as primitive or else constructed.
(Translated from Culioli 1974, T.2:55)

“We attempt to carry out our observation according to the three following rules:

  1. to construct metalanguages (necessarily outside languages) based on systems of representation elaborated inside each specific language and to use only metalinguistic items that have a status in the theory;
  2. not to separate the theory of language from the theory of the analysis of languages […];
  3. to always use formulae in which each occurrence of the metalinguistic symbols has an identifiable trace in a text and in which each constituent of the text acquires a metalinguistic status in the system of representation. This last rule is extremely demanding and fundamental.”

(Translated from Culioli 1979, T.2:81)


A notion is a complex system of representation organizing physico-cultural properties of a cognitive nature. The notion exists before words enter into categories; it is a generator of lexical units. It is generally represented between slashes, for example /cat/; the notion is potentially a predicate (“being [a] cat”, “having the property of being [a] cat”), defined exclusively in intension. Though possessing a certain degree of stability that is necessary for communication, notions are not fixed rigidly once and for all: on the contrary, they are dynamic and vary from one individual to another and from one situation to another. A notion can only be apprehended through its OCCURRENCES, whether they be phenomenal or enunciative.

This dynamic property is also found from one language to another, so that, on the whole, it is preferable to speak of the {BEHAVIOUR} of a notion (or the way a notion functions) when it is apprehended though ENUNCIATIVE OPERATIONS.

Thus we may speak of {DISCRETE} behaviour when an occurrence manifests a QUANTITATIVE AND QUALITATIVE DELIMITATION of a given notion (J’ai acheté un guide touristique de l’Inde [I bought a tourist guide to India]: here we have both a purchase — one occurrence of /acheter/ [buy] — and it bears on one exemplar of /guide/).

We may speak of {DENSE} behaviour when qualitative properties are determined in a way that is exterior to the notion in order to construct quantitative delimitation (Hier, j’ai lu toute la soirée [Yesterday I read all evening]/ Mélanger une livre de farine et cent grammes de sucre roux [/ Mix a pound of flour and a hundred grammes of brown sugar]): hier[yesterday], toute la soirée[all evening], une livre[ a pound], cent grammes[a hundred grammes] are MARKERS of QUANTIFICATION applied to a notion that has dense behaviour).

Lastly we may speak of {COMPACT} behaviour when the notion can only be apprehended insofar as it relates to a locator and has only QUALITATIVE determination (evaluated in terms of degree, if necessary): il a été très sage [he has been very good] / j’éprouve quelque lassitude à toujours me répéter[I feel some weariness at always repeating myself]. In these utterances, très [very] or quelque [some] are markers of degree and do not confer discrete behaviour to the notions /être sage/ [be good] or /lassitude/ [weariness]. It is the LOCATION of these notions relative to the subject (il[he], je[I]) which anchors them in a given situation.


Bouscaren & Chuquet 1988, Grammaire et textes anglais : Guide pour l’analyse linguistique. Gap : Ophrys. 145-147.

Culioli 1981, “Sur le concept de notion”, BULAG 8, Université de Besançon, 63-79, reprinted in T. 1 : 47-66.

Culioli 1982, “A propos de quelque”, Actes du premier Colloque franco-bulgare de Linguistique, Contrastive Linguistics 1-2, Sofia, reprinted in T. 3 : 49-58.

Franckel & Lebaud 1990, Les figures du sujet. A propos des verbes de perception, sentiment, connaissance.  Gap : Ophrys, collection HDL. 207-222.

Franckel, Paillard & De Vogüé 1989, “Extension de la distinction discret, dense, compact au domaine verbal”, Recherches Linguistiques Tome XIII, Termes massifs et termes comptables. Metz : Klincksieck, 239-247.


“[…] the notion [/renard/] [/fox/] exhibits a whole set of properties which boil down to “renardité” [“foxness”] or “vulpéité” [“vulpineness”], well, I don’t quite know how to express it. In a word, it boils down to “être renard” [“being fox”].

And this notion (with all that it implies: for example, you can make metaphors based on it […]) is a kind of crossroads; to go back to the definition that I generally give, it is “a bundle of physico-cultural properties” (I say “cultural” intentionally).

And it is part of a complex relation, part of a network with other notions, other properties and so it is always being remodelled. That is why I use the term “notion” and not “concept”, which refers to a scientific concept, with a certain stabilised meaning.

The notion can be summed up as an aggregate of events which have the same property, for example “être renard” [“being fox”]. And each occurrence of “être renard” [“being fox”] refers to a “renard” [fox] which is both a particular occurrence and at the same time any occurrence— any occurrence, because otherwise you could not say “un renard” [“a fox”] since each “renard” [“fox”] would be separate from any other “renard” [“fox”]. So you isolate a certain number of properties, thus defining the criteria for conformity which correspond to the representation that you have of what a real “renard” [“fox”] is, what really deserves the name “renard” [“fox”].

This procedure of abstraction is necessary and at one point there is a shift from one particular representation, whose properties you abstract, to a representation of any being that has those properties, and that you can henceforth designate as (a) “renard” [“fox”].”
(Translated from Culioli 2002:53)

“A notion can be defined as a complex bundle of structured physico-cultural properties and should not be equated with lexical labels or actual items. Notions are representations and should be treated as such; they epitomize properties (the term is used here in a very extensive and loose way) derived from interaction between persons and persons, persons and objects, biological constraints, technical activity, etc.

Notions are apprehended and established through occurrences (enunciative events) which involve: distinguishing phenomenal instances, identifying properties and assessing their distance from one another, gauging the degree and the kind of similarity, deciding whether to keep the occurrences qualitatively separate or to categorize them as equivalent relative to a type […] It should be obvious that notions have a status of predicable entities and could be described as unfragmented solid wholes; but they are apprehended through occurrences, i.e. distinguished through separate events, broken down into units (actually localized in the physical world, or imaginary) with variable properties. Through a process of abstraction, we thus pass from phenomenal occurrences to abstract occurrences.”
(Culioli 1978, T.1:70 [original quotation in English])

“When we talk about the notion we are talking about a set that we can express, for example by “lire; lecture; livre; lecteur; bibliothèque; etc.” [“read, reading, book, reader, library etc.”], which is to say that things cannot be reduced to a lexical unit: the unit merely serves as an introduction, a first step […]

The problem is that the notion is something virtual and productive. It is not given once and for all in all its meanings and that is why it cannot correspond to a lexical unit. It is a generator of lexical units; it defines a class of linked operations. It designates.”
(Translated from Culioli 1981, T.1:53-54)

 “As I said earlier there is no relation of labelling between words and concepts, but there is what I have called the “notion”, what might also be called a “structured representation”. The notion is to be distinguished from the concept, which has its history, for example an epistemological history (concepts are structured relative to one another in a technical field). When we are dealing with notions, we are in a domain that has ramifications; notions are organised relative to one another: this animal relative to another animal, and we inevitably create relations between them (relations of predation or of accompaniment or of identification) […] On the other hand there is proliferation, i.e. you have a whole set of properties that are organised relative to one another, physical, cultural, anthropological properties and as a result, one term does not refer to one meaning, but refers to what I would not call a field, for a field is already a certain type of organisation amongst terms, but to what I would call a notional domain, i.e. a whole set of virtual meanings. All metaphorical work is related to this fundamental property of symbolic activity through language activity which is characterized by plasticity (it has stability, which is why words can serve as labels, but it is also open to deformation).”
(Translated from Culioli 1986, T.1:85-86)


The notional domain is constructed from the NOTION; it enables the enunciator to structure the CLASS OF OCCURRENCES which is associated with that particular notion. The notional domain is thus organized around an OCCURRENCE which serves as a reference, a typical occurrence, which is called the ORGANIZING CENTRE and which is divided into three zones:

In this model, which is based on mathematical topology, the exterior stands for strongly CONTRASTING VALUES (“strong otherness” — not at all, in no way connected, completely different) and the boundary for weakly CONTRASTING VALUES (“weak otherness” — not really, scarcely, just enough).


Culioli 1982, “Rôle des représentations métalinguistiques en syntaxe”, Session plénière du XIIIe Congrès International des Linguistes, Tokyo, reprinted in T. 2 : 95-114.

Culioli 1991, “Structuration d’une notion et typologie lexicale. A propos de la distinction dense, discret, compact”, BULAG 17, Université de Besançon, 7-12, reprinted in T. 3 : 9-16.

Gilbert 1993, “La théorie des Opérations Énonciatives d’Antoine Culioli”, in P. Cotte et al., Les théories de la grammaire anglaise en France. Paris : Hachette. 69-74.


“From the notion, a notional domain will be constructed with, amongst others, the following properties: 1) all domains have a class of occurrences, which makes the notion quantifiable (such and such an occurrence of /( ) être chat/[( ) be a cat]) ; 2) from a qualitative point of view, all notional domains are made up of an interior, with an organising centre, a boundary, and an exterior: the interior gives occurrences that are both individuable and identifiable with one another, since they all have the same property. […] The organising centre appears clearly in the operation of self-identification or in “typing” (i.e. establishing “types”): “un chat est un chat [a cat is a cat], un chat est toujours un chat [a cat will always be a cat]; un vrai livre (pas une brochure) [a real book (not a brochure)], un livre livre (c’est-à-dire un livre qui est vraiment un livre) [an actual book (that is, a book that is really a book)]. The exterior will provide, according to the different possibilities, a void, absence, impossibility or radical difference or otherness. This last possibility is marked, in French, by revealing expressions such as: “c’est tout autre [it’s quite another thing/ it’s totally different]; ça n’est pas du tout un livre [it’s not a book at all]; ça n’a plus rien à voir” [it is nothing like it]. The boundary contains values which belong neither to the interior nor to the exterior, but which, according to the action of the different enunciators during an exchange, a discussion or a logical demonstration, may be included in either the interior or the exterior. […] Clearly, the boundary allows one to enter the class of non-null values […]; it also allows one to leave those values. What we are considering here is the construction of the linguistic complement, too often considered as being the same as the mathematical complement.”
(Translated from Culioli 1982, T3:54-55).


A concept  introduced within the Theory of Enunciative Operations  to signify that the object of study of linguistics is not given but constructed. The object only gains relevance through the aims that are set by the linguist and the SYSTEM OF METALINGUISTIC REPRESENTATION with which it is approached. Culioli insists on the fact that all observation of data must be founded on a theory of observables, which differentiates the Theory of Enunciative Operations from purely empirical approaches based on corpora (in particular corpus-driven studies).


Culioli 1968, “La formalisation en linguistique”, Cahiers Pour l’Analyse, 9, Paris: Seuil, 106-117, reprinted in T. 2 : 17-30.

Culioli 1982, “Rôle des représentations métalinguistiques en syntaxe”, Session plénière du XIIIe Congrès International des Linguistes, Tokyo, reprinted in T. 2 : 95-114.

Franckel & Paillard 1998, “Aspects de la théorie d’Antoine Culioli”, Langages n° 129, Diversité de la (des) science(s) du langage aujourd’hui. Paris> : Larousse, 52-63.


“It is impossible to examine the question of what is observable without setting up a theory of observation, in particular, without asking where the observers are to be situated. This is clearly marked by the distinction between surface structure and deep structure in Chomsky’s approach: the surface pattern is the trace of underlying operations. Studying the process of production requires that you leave the field of illusorily immediate observation in order to operate abstractly.”
(Translated from Culioli 1968, T.2:20-21)

“An observation can only be carried out within a theoretical framework (where theoretical means ‘belonging to a coherent set of hypotheses which can be tested’) […] Beginning with theorised observations based on a detailed description of various languages, one constructs a metalinguistic model. Then one studies the general properties (independent of local conditions) of the operative systems beyond the variety of the markers, whether one is concerned with predicative operations or the systems of enunciative coordinates used to constitute and locate the utterances.”
(Translated from Culioli 1974, T.2:54-55)

“We have to go beyond classifying properties and labelling, to free ourselves from purely intuitive remarks by constructing a system of metalinguistic representation (which will include everyday language), to construct a theory of observation (or of observables) and, from classes of phenomena (in particular by the construction of families of utterances related paraphrastically) to formulate problems and to construct procedures for reasoning.”
(Translated from Culioli 1982, T.2:96)


An exemplar, representative, embodiment of a NOTION. Occurrences are both referential and linguistic. “Sketch”, “drawing”, “image”, “diagram” etc. and also “draw”, “sketch”, “scribble”, “reproduce” etc. as well as “draughtsman”, “illustrator”, “caricaturist” etc. are all linguistic occurrences of the same notion, i.e. /drawing/. In: there is a fly in the soup, the existence of a referential occurrence of /fly/ and a referential occurrence of /soup/ are posited. An occurrence may then be considered as an enunciative event that carries out two forms of delimitation on the notion: a QUANTITATIVE DELIMITATION involving its spatio-temporal location, i.e. its existence, and a QUALITATIVE DELIMITATION which is related to its properties, its nature. The occurrences associated with a notion constitute a {CLASS OF OCCURRENCES} that corresponds to the extension of that notion.

This use of the term “occurrence” in the Theory of Enunciative Operations is distinct from that, also current in linguistics, used to designate the existence of a term (or an expression) in a given corpus (cf. “there are three occurrences of the word must in this text”).


Bouscaren & Chuquet 1988, Grammaire et textes anglais : Guide pour l’analyse linguistique. Gap : Ophrys. 145-147.

Culioli 1981, “Sur le concept de notion”, BULAG 8, Université de Besançon, 63-79, reprinted in T. 1 : 47-66.

Culioli 1991, “Structuration d’une notion et typologie lexicale. A propos de la distinction dense, discret, compact”, BULAG 17, Université de Besançon, 7-12, reprinted in T. 3 : 9-16.

Franckel, Paillard & De Vogüé 1989, “Extension de la distinction discret, dense, compact au domaine verbal”, Recherches Linguistiques Tome XIII, Termes massifs et termes comptables. Metz : Klincksieck, 239-247.


“One can distinguish two types of occurrences: phenomenal occurrences and linguistic (or metalinguistic) occurrences. Phenomenal occurrences do not concern the linguist. From the earliest age, one comes across objects that may be hard or soft, objects that fit into other objects, that can be placed side by side, matter that is liable to tear or is breakable or eatable, etc. […] So phenomenal occurrences can be manipulated in connection with the growing and learning process, thereby constructing complex systems of intra-cultural representations of phenomena. Linguistic occurrences refer, in the construction of the linguist, to each term that is introduced either directly or mediately (for example using “amongst others”).”
(Translated from Culioli 1981, T.1:55-56)

“The incarnation of a notion in the form of language represents the transition to materiality at the same time as the transition to a system of referentiation. We have access to the material existence (i.e. the traces constituted by the patterns of markers) but not to the transition itself, about which we know nothing. This transition from a mental representation with no tangible existence to an activity which allows one to construct reference corresponds to a “shaping” of the notion that I note (on a metalinguistic level) as QNT. The operation marked by QNT […] is based on a fundamental constructing operation linked to predicating existence, as marked by the existential expressions il y a [there is] or soit [let there be]. Existence […] can correspond to:

(Translated from Culioli, 1991, T.3:10-11)

“We shall use the word occurrence for any form of delimitation of a notion. There are two ways of delimiting a notion, on the basis of two possible criteria of difference (“otherness”):

Where nouns are concerned, this criterion can amongst other things, in the simplest cases, refer to the idea of an “exemplar” or “token”: one token of “table” can be distinguished from another token of “table”, independently of the fact that the “table” quality may be identical or different.”
(Translated from Franckel, Paillard & de Vogüé 1989:239)

“Take the following utterance: (9) Dans une rue où les chats sont des chats, on ne s’amuse guère. [Approximately: “In a place where spades are spades, you can’t have much fun” (literally: “in a street where cats are cats”)].

That such an utterance should be acceptable, and that it is interpretable (not being seen as tautological) shows that there is a discrepancy between what it designated by chat [spade] and that which has the property être un chat [be a spade]. For the content of the utterance is founded on the fact that les chats [“spades”] (i.e. what we call chat [spade]) can also not be des chats [spades](i.e. not possess the corresponding property). In other words, the situational occurrences of chat [spade] are not necessarily notional occurrences.
(Translated from de Vogüé 1987:55).


LANGUAGE ACTIVITY is analysed as the result of a series of operations (enunciative, predicative and determining operations) which the linguist attempts to reconstruct from the trace constituted by the arrangement of MARKERS in a text. The theory posits that there are three levels of representation:

The theory of enunciative operations supposes that on the mental level (level I), what the linguist establishes are generalisable operations and not fixed structures or systems.


Culioli 1978, “Valeurs modales et opérations énonciatives”, Le Français Moderne, Vol. 46 (4), reprinted in T. 1 : 135-156.

De Vogüé 1991, “La transitivité comme question théorique : querelle entre la Théorie des Positions de J.-C. Milner et la Théorie des Opérations Prédicatives et Énonciatives d’A. Culioli”,  LINX n° 24 , Sur la Transitivité dans les Langues. >Paris 10, 37-65.


“From a Culiolian point of view, all language objects are constructed. Hence the theory of operations, stating that one has to elaborate rules for the constitution of those objects instead of having objects which are already constituted […] As there are infinite (but not random) ways of constructing a language object, the units which constitute the object are never identical. We find a simple illustration of this in examples of (pseudo-)tautology: un chat est un chat [a spade is a spade (literally: a cat is a cat)], in which the two occurrences of chat [spade] are not interpreted in the same way and are therefore not the same object, since the operations that construct them are not the same.”
(Translated from de Vogüé 1991:41).

“The approach we have adopted explicitly refuses to allow the linguist to restrict himself to classifying procedures; the aim remains to apprehend language through the variety of languages and, in order to do so, to construct from systematic and meticulous observations a metalinguistic representation system which will allow the linguist to pose problems and to give them reasoned solutions. Thus one can hope to discover, behind the apparently banal form of everyday utterances, the hidden operations of enunciative work.”
(Translated from Culioli 1978, T.1:155).

“There is no cognition without sensation and no symbolic activity without operations. Consequently we are led to examine what the operations are on which we can construct.”
(Translated from Culioli 1986, T.1:85)

“The main effort of the linguist will be to construct a manipulable system of metalinguistic representations which will allow him or her to establish a correspondence between configurations (patterns of markers in oral or written texts) and operations. Thus, by using controlled manipulations, one can hope to construct a set of formal operations to apprehend, beyond the variety of languages and discursive phenomena, that set of fundamental operations that can be generalized and are the basis of enunciative activity (both producing and understanding utterances).”
(Translated from Culioli 1978, T.1:139, note 7).


There are two ways of defining this term.

  1. A non-technical way, which is to use the term to designate the syntactic-semantic structure consisting of <subject argument + verb + object argument (+ possibly adverbial elements)> before it becomes an utterance, i.e. before it is given the temporal, aspectual and modal determination which will make it interpretable relative to a particular SITUATION OF UTTERING. This definition is comparable to that of a “proposition” or “clause” in traditional grammar.
  2. A more precise definition showing the way the term fits into the different stages in the constitution of an utterance in the framework of the Theory of Enunciative Operations.

We can define this concept as a relation, organized by a series of LOCATING OPERATIONS between the terms which themselves represent NOTIONS. The term “predicative relation” is sometimes used instead of “LEXIS”, which is more technical but which does not refer to precisely the same concept. A “lexis” (from lekton, “speakable”, “utterable”) results from the instantiation of a LEXIS SCHEMA (or pattern) with notions whose properties partly determine their organization: the PRIMITIVE RELATIONS between notions (part-whole, INTERIOR/EXTERIOR, agent, animate, determinate, beneficial etc.) organize these notions into a SOURCE NOTION and a GOAL NOTION on the one hand, and a RELATOR NOTION on the other hand. Once the lexis is constituted, it can generate a family of predicative relations.

Optimally, a predicative relation is a relation between three terms (the lexis schema has three places), located relative to one another two by two by a series of binary OPERATIONS between LOCATOR and LOCATUM. If it is necessary to construct a relation concerning more than three terms, each extra term is in turn introduced by an extra binary relation which links it to the first relation and to each of its terms. So, for example, the addition of a notion indicating place to a transitive relation between an agent and a patient is specified in French by an operator which is a preposition.

Finally, a predicative relation is oriented from the first argument, which is also a {COMPLEMENT} (a “zero rank” complement, C0, or origin complement) of the relator-notion, i.e. the predicate in the strict sense of the word. This orientation determines the type of diathesis or voice, in that the choice of the first argument between the source notion and the goal notion of the primitive relation will entail, among other things, an active or passive orientation of the predicative relation.

The order of the operations which constitute the predicative relation is therefore, succinctly:

This is followed by other operations which are generally called “enunciative” operations (although the strict distinction between “predicative” operations and “enunciative” operations is better apprehended as seamless), such as:

The first argument, the starting term, and the constitutive locator may, in some instances, be the same term but they are distinguishable and distinguished from one another in different languages (see for instance the difference between the particles wa and ga in Japanese or the “cascades” of locating operations in spoken French, as in: Pierre, la cocotte- minute, il fait tout cuire dedans[literally : Pierre, the pressure cooker, he cooks everything in it]).

A predicative relation can also, like any occurrence of a notion, be represented as PRECONSTRUCTED, i.e. its existence, linguistically, is already defined; it may then be subjected to other operations:

This utterance contains a contradiction due to the fact that on the one hand the PRECONSTRUCTED predicative relation (the subordinate clause) is always located relative to an ENUNCIATIVE ORIGIN and, with the verb savoir [know], takes on one and only one value (“it is a fact that I locked the door”), but, on the other hand, the LOCATION relative to “je”[I], a pronominal marker of IDENTIFICATION with the same enunciative origin, takes on a negative value (i.e. non-location, DISCONNECTION: ne pas savoir [not to know], for je [I], is equivalent to saying that the subordinate clause no longer has a status for the enunciative origin). The solution of this contradiction may be found by changing the status of the subordinate clause (je ne sais pas si...[I don’t know if/whether…]) or by changing the time locator for je [I], (je ne savais pas que...[I didn’t know that…]): the je [I], at T-1 did not know...but the je [I] at To knows.


Bouscaren & Chuquet 1988, Grammaire et textes anglais : Guide pour l’analyse linguistique. Gap : Ophrys. 133-145.

Culioli 1982, “Rôle des représentations métalinguistiques en syntaxe”, Session plénière du XIIIe Congrès International des Linguistes, Tokyo, reprinted in T. 2 : 95-114.


“Given, on the one hand, a primitive relation specified by a predicate and, on the other hand, a schema, called a lexis schema and noted < ξ0 ξ1 π >, where ξ0 and ξ1 are variables for arguments and π a variable for operators of predication: from the primitive relation and the schema, we can construct the predicate and the arguments, distinguishing a first argument (order 0) and a second argument (order 1). In this way, a lexis is the result of the instantiation of a schema by terms which have themselves been constructed from notions. This operation produces a complex pattern which is not the product of a simple operation of assignment, by which categorized terms would be substituted for the variables of the schema […] Therefore, the construction of a lexis entails the formation of a bundle of relations among the constituent elements of the predicative relation.

A lexis is not an utterance (énoncé). It is neither asserted nor unasserted, for it has not yet been situated (or located) within an enunciative space defined by a referential network (a system of enunciative coordinates) […] A lexis is therefore both what is often called propositional content (in this sense it is close to the lekton of the Stoics) and a form which generates other, derived forms (a family of predicative relations, from which it will be possible to construct a paraphrastic family of énoncés). Any relation which has this property is a lexis, whether it becomes a syntagm or a sentence. Thus, le livre de Pierre [Pierre’s book]; Pierre, son livre [Pierre, his book]; Pierre, lui, son livre [Pierre, him, his book], etc., as well as Pierre a un livre [Pierre has a book]; Pierre, lui, a un livre [Pierre, as for him, has a book], etc. belong to the same family […] We might add that a lexis can be combined with another lexis and that we can construct a relationship of location between several lexes.”
(Culioli 1978, T.1:78-79 [original quotation in English]).


Quantification and qualification are two central concepts of the Theory of Enunciative Operations, used to represent the construction of an OCCURRENCE of a NOTION, which involves a {QUANTITATIVE DELIMITATION} (the OPERATION of quantification) and a {QUALITATIVE DELIMITATION} (the operation of qualification). The former, noted QNT, concerns the spatio-temporal determination of an occurrence, its LOCATION relative to the T (spatio-temporal) parameter of the SITUATION OF UTTERING, and consequently the existence of the occurrence; the second, noted QLT, relates to the nature, the properties, the qualities of an occurrence as given by a SUBJECT, and consequently to its location relative to the S (subjective) parameter of the situation of uttering. Both of these dimensions are at work in an utterance such as: Regarde! Un éléphant ! [Look! An elephant!]. Here the enunciator poses the existence (Qnt) of an occurrence which has the properties (Qlt) of the notion /elephant. In other cases, one of the two will predominate over the other, as in: L’éléphant a besoin de manger d’énormes quantités de nourriture pour alimenter son corps massif [the elephant needs to eat enormous quantities of food to nourish its massive body]. Here the qualitative delimitation is preponderant and the quantitative parameter does not directly intervene. Obviously, the quantitative delimitation is in no way related to quantification in the mathematical sense of the word: on the contrary, it is reminiscent of the “judgement of existence” of logicians, while qualitative delimitation echoes their “judgement of attribution”.

The operations leading to the construction of an occurrence from a notion also involve EXTRACTION (un livre vaut mieux que n’importe quel ordinateur [a book is better than any computer], il a eu un courage exemplaire [he showed an exemplary courage]), and PINPOINTING (le livre restera une manifestation culturelle incontournable [the book will continue to be an indispensable cultural artefact]), which is based on extraction. A further operation called SCANNING (cf. n’importe quel ordinateur [any computer]) links quantitative operations (the possible occurrences of the notion are taken into account) and qualitative operations (the same property is recognised in all these occurrences).

By taking into account the different possible combinations of the two dimensions, Qnt and Qlt, of the occurrence, it is possible to represent the system of noun determination and its different values. It is also possible to use the same linguistic tools to account for other, very different domains, such as aspect or modality; in that case, we are dealing with the occurrence of a relation, i.e. a combination of notions. Thus, by linking the Qnt and Qlt delimitations of the occurrence and by weighting them, it is possible to represent various aspectual values, those associated with “unaccomplished” aspect, which, roughly speaking, are essentially quantitative, or those associated with “accomplished” aspect, which are qualitatively fulfilled.

It is also possible to represent the range of modalities, from possibility to necessity, including orders and permission, which all, each in its own way, imply taking simultaneously into account the existential dimension of the occurrence of the relation (Qnt) and its subjective delineation by an enunciator (Qlt).


Bouscaren & Chuquet 1988, Grammaire et textes anglais : Guide pour l’analyse linguistique. Gap : Ophrys. 147-161.

Culioli 1999, “Des façons de qualifier”, in A. Deschamps & J. Guillemin-Flescher (eds), Les opérations de détermination. Quantification / Qualification. Gap : Ophrys, Collection HDL., 3-12, reprinted in T. 3 : 81-90.

Franckel & Lebaud 1990, Les figures du sujet. A propos des verbes de perception, sentiment, connaissance.  Gap : Ophrys, collection HDL. 207-222.

Gilbert 1993, “La théorie des Opérations Énonciatives d’Antoine Culioli”, in P. Cotte et al., Les théories de la grammaire anglaise en France. Paris : Hachette. 78-94.


“Let us start with a brief description of Quantification, Qualification. I use the term Quantification, not in the sense of logical quantification, but to refer to the operation whereby one constructs the representation of something that can be distinguished and located within a temporal and spatial frame (“referential space”). The vagueness of the terms is deliberate and temporary, inasmuch as one needs to start from elementary operations which are then enriched by complex series of supplementary operations. Thus something does not refer to an inanimate (as opposed to someone), but to a state, which can be described as distinguished from another state, thereby introducing discontinutiy, and can be located (in the abstract sense of the term) in a representational frame. In other words, something refers to an occurrence of whatever a subject may apprehend, perceive as an individual form in its environment, distinguish by eliminating indetermination and locate within a space-time frame (which may be imaginary).

Quantification enables two essential operations to be carried out:

  1. quantifiabilization (or to use a simpler term, fragmentation). Starting from a notion (P), characterized as mass (hence its symbolization by “being-P”, in an attempt to reflect the strictly qualitative, predicative character of notions), one then fragments the notion through an abstract operation of individuation, in order to construct occurrences of that notion when producing or interpreting utterances […]
  2. Secondly, quantification makes it possible to construct the existence of an occurrence (of the fragmented notion) by locating it in the space-time frame construed by an enunciator-subject in relation to a co-enunciator. Constructing existence thus consists in bringing an occurrence from nothing to something within the locating framework […]

Now as for Qualification, it is brought into play every time an operation of identification or differentiation is carried out on something (in the sense that was given to the term above).”
(Translated from Culioli 1999, T.3:82-84)

“An occurrence refers to an enunciative event which delimits the notion in two ways that combine variously with one another:

(Translated from Franckel & Lebaud 1990:207-209)


This is an absolutely central concept in the theory, in that PREDICATIVE and ENUNCIATIVE OPERATIONS are involved and in that, from a more general point of view, one can see how lexical items (and the properties of the NOTIONS concerned in the lexicon) construct the predicative relations and are indissociable from the syntactic relations.

The schematic form of a given lexical item can be defined on the one hand as an organizational relational schema (the term “scenario” is also used by some) imposed by the lexical item on its environment, i.e. on the other lexical items which may co-occur with it in a PREDICATIVE RELATION and, on the other hand, as a form or shape, an entity which may itself be reshaped since it enters into relations with other lexical items having their own schematic forms and consequently their own scenarios. For each lexical item a principle of INVARIANCE and modalities of VARIATION are defined. The INVARIANCE SCHEMA corresponds to EXTERNAL VARIATION of co-texts in which the lexical item occurs; the “form” corresponds to INTERNAL VARIATION, to different “facets” of the item itself. The schematic form crucially involves the concept of LOCATION but it cannot be reduced to one single OPERATION (any more than a lexical item can be considered as a simple operator, which would make its relation with other items rigid). Isolating the schematic form requires a description of what makes a lexical item specific, but a schematic form cannot be reduced to a “central value” from which peripheral or metaphorical values can be derived.

To give two examples studied in the literature, we can take the case of tirer+sur [draw on /shoot at etc.] which involves two schematic forms: isolating these SFs should allow us to account for uses such as tirer sur la ficelle [to go too far], tirer sur un lapin [to shoot a rabbit],tirer sur le bleu [to verge on blue], whereas choosing a “central value” of “traction” would inevitably lead to a dead end. We may also consider the case of lit [bed], for which, once again, the properties of “piece of furniture” could not enter into a definition of the SF, or else one would then have to resort to metaphors to account for contexts such as lit de la rivière [the bed of the river], sur un lit d’oignons émincés [on a bed of chopped onions], faire le lit du fascisme [to pave the way for fascism (lit. “to make the bed of fascism)] etc. In the case of lit [bed], one may consider that there is a limit to variation and that the expression lit à baldaquin [four poster bed] is an instance of lexical fossilization.

This gives an idea of the complexity of this approach, which involves several schematic forms, QUANTITATIVE AND QUALITATIVE DELIMITATION of the lexical items concerned, the NOTIONAL BEHAVIOUR (DISCRETE, DENSE, COMPACT) of the lexical items, the aspecto-temporal location etc.


Culioli 1986, “Stabilité et déformabilité en linguistique”, Etudes de Lettres, Langages et Connaissances, Université de Lausanne, reprinted in T. 1 : 127-134.

Culioli 1987, “Formes schématiques et domaine”, BULAG 13, Université de Besançon, 7-16, reprinted in T. 1 : 115-126.

Franckel & Paillard 1998, “Aspects de la théorie d’Antoine Culioli”, Langages n° 129, Diversité de la (des) science(s) du langage aujourd’hui. Paris : Larousse, 52-63.

Paillard 2000, “A propos des verbes ‘polysémiques’ : identité sémantique et principes de variation”, Syntaxe et Sémantique n° 2, Sémantique du lexique verbal. Caen : Presses Universitaires. 99-120.


“The analysis of a marker (i.e. not simply a label but a marker of an operation or possibly of a poly-operation) must result in a formal representation having stable and verifiable characteristics. From this formal representation, which I call the schematic form, supplementary forms are constituted that are deformations of the base form. The question is to understand how these patterns of deformation (“warps” or “shear” effects) [Culioli 1994, T2:85] are organised.”
(Translated from Culioli 1987, T1:115-116).

“I think it is important to insist on the concept of stability. Without stability, there would be no regulated adjustment, no communication […] but stability should not be confused with rigidity or immutability. Linguistic phenomena form dynamic systems which are regular but which have a margin of variation due to a great variety of factors: they are phenomena which are both stable and malleable. […] Deformation is a transformation that modifies a configuration so that some properties remain invariant throughout while others vary. In order for there to be deformability, there has to be a schematic form (such that there may be both modification and invariance), and deforming factors as well as a margin of flexibility, i.e. a space for adjustment endowed with topological properties.”
(Translated from Culioli 1986, T1:129-130).

“Our approach posits that the problem of the identity of a lexeme is a decisive issue. At the same time, we believe that it is necessary to move away from the question of polysemy (where lexical units are examined one by one) towards the study of the way certain fundamental principles of variation function, principles which are the basis of the conditions in which the lexicon contributes to the construction of the meaning of an utterance. The study of this variation is indissociably linked to establishing a pole of invariance defining the semantic identity of the word. […] We define the pole of invariance as a schematic form which in itself does not correspond to any of the values of the unit. The very notion of the schematic form means that the identity of the word is indissociable from its relation to the co-text: as a schema, it informs the co-text; as a form it gets its substance from the elements of the co-text that it activates. From this point of view, a schematic form can be likened to an abstract scenario.
(Translated from D. Paillard 2000:101).


This is the {ABSOLUTE ORIGIN OF LOCATING OPERATIONS}, noted Sit0: it comprises two {CO-ORDINATES}, S0 for the ENUNCIATOR (the uttering subject, i.e. the subjective parameter) and T0 for the time (moment) of uttering (the spatio-temporal parameter). The situation of uttering serves as the locator for the situation of locution, noted Sit1, which also has two co-ordinates, S1 for the LOCUTOR (or SPEAKER) and T1 for the moment of locution (of speech); this situation in turn is the locator for the index of events Sit2, which also has two co-ordinates, S2 et T2. A calculation based on the S parameter, involving the three values of the RELATION OF LOCATION, yields the category of person, and the tense system is obtained by a calculation based on the T parameter. It is important to note that the concept of situation of uttering is an operative abstraction and not an element of the real world.


Culioli 1987, “Formes schématiques et domaine”, BULAG 13, Université de Besançon, 7-16, reprinted in T. 1 : 115-126.

Gilbert 1993, “La théorie des Opérations Énonciatives d’Antoine Culioli”, in P. Cotte et al., Les théories de la grammaire anglaise en France. Paris : Hachette. 68-69.

Franckel 1989, Étude de quelques marqueurs aspectuels du français.Genève : Droz. 16-20.


“Any given term is located relative to Sit (S, T): one may take into account either S or T, or both S and T, or the relation to Sit as a whole […] As regards S, one should be careful to distinguish between enunciator and locutor (or speaker); T represents the space-time frame, more specifically the axis of instants.”
(Translated from Culioli 1987, T.1:116)

“This enunciative space is not given once and for all. It is brought into being by each act of uttering where every user of a natural language “constructs” his own spatial and temporal referential system; this system is not objectivized or universal, nor is it exterior to or independent of any language user.”
(Translated from Culioli et al. 1981:71)

“The articles in this volume show […] how essential it is to distinguish between the speech situation — where one is dealing with speakers (locutors) involved in a process of emission-reception implying physically located persons engaged in successsions of speech events that are necessarily located in time and space — and the original situation (Sit0), where the concept of enunciator (and of co-enunciator) refers both to an abstract entity in a specific linguistic framework and to the notion of a social and psychological subject.”
(Translated from Culioli 1984:11)


The concept of the subject in the theory of enunciative operations is limited strictly to that of the uttering subject (and its derivatives: the point of view or the ASSERTOR, according to different authors). In this framework, to speak of the “subject of the utterance” or “the subject of the verb” is a convenient shortcut but it is preferable to avoid them to eliminate possible ambiguities linked with these terms.

The uttering subject or enunciator is an abstract LOCATOR in the theory (see under ENUNCIATOR) and, as the origin of subjective LOCATING OPERATIONS, it is the basis for the construction and reconstruction of the REFERENTIAL VALUES of an utterance.

Instead of “subject of the verb”, it is preferable to use the term zero rank COMPLEMENT (C0), since it is a locator constructed to orient the PREDICATIVE RELATION (in a passive utterance, the C0 can in fact be the goal argument which would be in the C1 position if the orientation were in the active voice). The C0 can of course have the properties of a human animate and be constructed as an agent.

The notions “subject of the utterance” and “theme” are also redefined within the construction of the predicative relation and of the utterance. The STARTING TERM functions as a locator organizing the predicative relation in order to make it into an utterance. In the same way, the CONSTITUTIVE LOCATOR functions as a locator for the starting term and is itself the organizing domain of the utterance in the sense that it anchors the utterance in a SITUATION OF UTTERING.

Only the expression “subject in the utterance” could, in certain cases, have a specific role in the theory of enunciative operations to the extent that it might designate a MARKER whose referential value is a human animate, i.e. the locator of the operations of evaluation and (aspectual and modal) valuation in a text: in such cases there may be an asserting subject (reported in various ways), a point-of-view-subject etc.

For instance, in: il sortit de bonne heure le lendemain et se tint un moment sur le pas de sa porte : la mairie n’avait pas encore donné les instructions pour dégager la rue [he came out early the next day and paused a moment on the threshold: the town council had not yet given the instructions to clear the street], we can consider that the pronoun il [he], as an anaphora of a given character, constitutes the point-of-view-subject of the utterance in the second part of the quotation.


Bouscaren & Chuquet 1988, Grammaire et textes anglais : Guide pour l’analyse linguistique. Gap : Ophrys. 179-185.

Culioli 1973, “Sur quelques contradictions en linguistique”, Communications 20, 83-91, reprinted in T. 2 : 43-52.

Culioli 1968, “La formalisation en linguistique”, Cahiers Pour l’Analyse, 9, Paris: Seuil, 106-117, reprinted in T. 2 : 17-30.

Simonin 1984, “De la nécessité de distinguer énonciateur et locuteur dans une théorie énonciative”, DRLAV n° 30, La Ronde des sujets, Paris 8 & CNRS, 55-62.


“If we now turn to the concept of enunciator, we can say that there is a relationship of alterity, of otherness, between the enunciator and the co-enunciator, such that there can be coalescence or separation between them. On the other hand,  the speaker and the addressee are always separate and we should not confuse the intersubjective field (where “subject” refers to the enunciator) and the mechanics of dialogue.”
(Translated from Culioli 1986, T.1:130)

“As has already been indicated, S0 will symbolise the enunciative subject, which is the basis of the intersubjective space of all situations of uttering. S1 will be used to designate the first occurrence of a subject in the utterance (here we mean the subject of the utterance, as separate from the enunciative subject, which is different from the traditional syntactic meaning of this word). The second occurrence will be represented by S2 and so on […] The personal pronoun JE [I] indicates identification (represented by the operator =, or a double arrow or a loop), for example S1 = S0, or S2 = S1; the personal pronoun TU [YOU (singular)], on the contrary implies a displacement or shift.”
(Translated from Culioli 1973, T.2:49-50)

“In (3): Moi, je sais si Pierre est parti [I[contrastive stress on I]  know if Pierre has left],[…], “Moi, je sais” allows the enunciator to be a speaker but also a member of the class K of values which can instantiate the slot in < ( ) savoir si > [< ( ) know if>] (i.e. moi, toi, lui etc. [I, you, he etc.]). By dissociating the enunciative subject and the subject of the utterance (a member of the class K), the contradiction that one has in French in a non-contrastive utterance such as (1):*Je sais si Pierre est parti” is eliminated.”
(Translated from Culioli 1978, T.1:153-154)

“Some categories will be represented by vectors of properties […] Thus, we shall indicate that Subject (of a sentence) = (C0, Agent, Theme). C0 is to be read zero rank complement in a theory of complements, implying that the (surface) subject, in languages where it is obligatory, is necessary from the outset for the utterance to be canonically well-formed; the term Agent implies either that very frequently (the formulation is deliberately left approximate) the subject is an Agent at surface level or, looking at it from the opposite angle, that the Agent was the Subject at a deep level […] This is a “sliding” vector, i.e. each term except the C0 can be null. Consequently, one can have: (C0, Ag., Th.), (C0, Ag.), (C0, Th.), (C0). On the other hand, the Agent is represented by another vector (Agent, Animate, Determinate), which is also a sliding vector.”
(Translated from Culioli, 1968, T.2:26)

“The enunciator consequently has to be posited as an essential linguistic parameter, on the one hand because of the fundamental role that it plays in the construction of referential values, on the other hand as the source of modalisation operations. The enunciator as a parameter participating in the construction of referential values does not have the same status as the enunciator as a source of modalisation operations, and we shall henceforth call the latter the speaker (I should point out that Culioli does not always use the words enunciator and speaker in the ways I am using them here). The enunciator is a parameter of the situation of uttering, inseparable from the co-enunciator […]. On the other hand, to the extent that it is the speaker that endorses an assertion and not the speaker-addressee couple, one can justifiably say that the speaker parameter is heterogeneous with respect to the addressee […] Even though the speaker is generally identified with the enunciator, there are nevertheless forms whose use and meaning can only be explained by dissociating enunciator and speaker […] One can distinguish between texts which explicitly construct a term as a speaker and those in which the speaker is not explicitly constructed as such: explicit reported speech in the former case, free indirect speech in the latter.”
(Translated from Simonin 1984:56-57)

“… the distinctions that we can observe, in particular those that reveal the different planes of enunciation, lead us to believe that it is only possible to account for the complexity of the enunciative locating operations by positing metalinguistic situations constructed from the initial situation of uttering (Sit); all these constructed situations then serve as new locators which were themselves located in earlier locating operations […] In the utterance: ‘Il dit: “je lis”’ [He says/said, “I’m reading”] […], we find the trace of those earlier locating operations in the third person pronoun il [he] and the declarative form dit [says/said], which represent respectively a subject located relative to the initial enunciative subject S and a moment in time located relative to the moment of uttering T. The operations of aspectual location that are apparent in “je lis” can be glossed by ‘Il dit : “Pour moi au moment où je le dis, (il y a que) je lis” [‘He says: “For me at the time when I am saying it, (it is a fact that) I am reading”’], i.e. in a very general way, ‘je lis [I am reading] is located relative to a moment in time for a subject’. It is possible to specify the nature of this aspectual operation by considering that the locating situation constitutes a point of view on the located expression ‘je lis’ [‘I’m reading’], that is, essentially, on the action marked by the verb, and consequently it is possible to gloss this locating operation by the verb voir [see] — thus for the initial utterance: ‘Il dit: “Je vois que je lis”’ [‘He says : “I see that I am reading.”’] […] This shows that the utterance of the assertion (il dit [he says], noted E0) functions as a situation, i.e. it is in turn a metalinguistic locator relative to which another utterance is located (je vois [I see], noted E1). This is what we mean by a constructed metalinguistic situation. The expression Eo, which is made up of So and To, thus constitutes, on the metalinguistic level, a situation (noted Sito) made up of an asserting subject So and a corresponding moment in time To.”
(Translated from Fuchs & Léonard 1979:78-79)


The meaning of this term has changed over time. The most frequently used definition concerns the REFERENTIAL VALUE given to a PREDICATIVE RELATION. As a complex notion (i.e. a notion constructed by relating notions to one another), a predicative relation can be situated in a domain of validation: let (p,p’) be the representation of the domain of validation of a relation.

This relation is said to be either validated if, for a particular situation, it is given the value p (the positive value) or not validated if it is given the value p’ (negative value): an enunciative origin asserts (i.e. declares that such is the case) that this value exists for the relation.

A relation can also be “validatable” (waiting for an enunciative source to carry out the operation, as in the case of hypothetical relations); or the validation can be located by an operation called “VISÉE” (“prospective validation”), the assertion that “it is the case” (or “it is not the case”) being considered relative to a prospective (future) locating situation disconnected from the origin situation.

 Validation can be suspended, for example when the two values of the notional domain are considered and when, from a BRANCHING POSITION situated outside the domain (p,p’), the chances of validation of the predicative relation are evaluated (this is what one finds in epistemic modality, for example).

In studies on verbal determination and the delimitation of verbal notions in particular, the concept of validation is given a more technical meaning: it concerns the construction of the NOTIONAL DOMAIN in relation to the verbal NOTION. What are generally presented as operations of QUANTIFICATION and QUALIFICATION of the notion — construction of OCCURRENCES, comparison with reference to a type, delimitation of the INTERIOR of the domain (occurrences that have the properties of the notion in question) and the EXTERIOR, taking the non-null BOUNDARY into account (i.e. the occurrences that do not really have the properties of the notion in question) etc. — are brought into play in the delimitation of a particular verbal notion. Not only is the latter LOCATED relative to a SITUATION OF UTTERING that localises it (i.e. for which it is verified), but it is also evaluated with respect to the physico-cultural properties to which it refers, which involves aspectual determinations (accomplishment or not, resulting states etc.) as well as the role of the target argument (rank 1 COMPLEMENT) in delimiting (or not) the verbal notion.

From this point of view, the validation of a predicative relation is of a QUALITATIVE nature (for example in: il a écrit une lettre [he has written a letter]), producing a NOTIONAL OCCURRENCE of the verbal notion, corresponding to what is implied by the properties of that notion (in the example, écrire [write] with the meaning “produce a result”, la lettre [the letter], a delimited occurrence of writing). In contrast, there may be  no validation, only verification of the verbal notion, of a QUANTITATIVE type (for example in: ce matin, il a écrit, il a lu un peu, il est sorti faire des courses[this morning he wrote, read a little, went out to do some shopping…] and the occurrence is called a SITUATIONAL OCCURRENCE (for a given situation, there was “de l’écriture” [“some writing”]).

Of course, this distinction is linked to the way the verbal notion functions (its BEHAVIOUR), and the distinction between temporal LOCALISATION and notional validation can only operate in those cases where the particular way the verbal notion functions will allow it (i.e. when it is possible to distinguish OCCURRENCES of the verbal notion). Thus, when the verbal notion functions in the COMPACT mode, validation is naturally interpreted as the predication of a property, for example a property of the zero rank complement (it is impossible to delimit quantitative occurrences).

This second meaning is a more technical development of the concept of validation, which is, however, not in contradiction with the first meaning, and in fact can be shown to be related to it.


Chuquet 1993, “Perfect anglais et opérations de validation: quelques repères” in Danon-Boileau, L. & J.-L. Duchet, Opérations énonciatives et interprétation de l’énoncé. Mélanges offerts à Janine Bouscaren. Gap: Ophrys. 27-46.

Culioli 1988/90, “La négation : marqueurs et opérations”, in J. J. Franckel, ed., La Notion de Prédicat, Travaux de l’ERA 642, Université Paris VII, 17-38, reprinted in T. 1 : 91-114.

De Vogüé 1987, "Aspect : construction d'occurrences", T.A. Informations n° 1, Revue Internationale de l'Application de l'Automatique au Langage. Paris : Klincksieck, 47-61.

Franckel 1986, “Modes de construction de l’accompli en français”, in Aspects, modalité : problèmes de catégorisation grammaticale. Paris 7 – D.R.L. : collection ERA 642, 41-69.


“Let us consider a predicative relation in its simplest representation, i.e. before it has been asserted, which would imply a choice of validation. Connected to this predicative relation is a complex domain of possible values: disconnected from (p,p’); p (p is the case); p’ (p is not the case, or other than p) […] Let us now consider a case where tense-aspect plays a central role. Take a predicative relation (the exact term should be a lexis, but the distinction can be dispensed with here), and locate it on the axis of instants, thus taking into account the T-parameter of the enunciative origin Sit. We need to construct two occurrences of the predicative relation located relative to two locators, tm and tn, which are necessarily ordered in relation to each other. The predicative relation is noted <r>, and IE (read: reinforced E) represents the position disconnected from (p,p’), indicating that one is neither in the Interior nor the Exterior of the domain; I represents the Interior of the domain of validation, and E the exterior, construed from I. Whether <r>’s domain is IE, I or E, it is merely a matter of stating what is the case at the locating moments tm and tn; if stability prevails, there is no transition marked between instantaneous state tm and instantaneous state tn […]

If preponderance is given to the S-parameter […], from IE one constructs two paths, assessed as good or bad, one leading to I, the other to E, represented respectively as the good validation IE I (where the arrow indicates the gap that needs to be bridged in order to obtain the validation of <r>) and the bad validation IE E. For example, if an enunciator projects (from the IE position) the validation of the predicative relation <Luc venir> [<Luc come>], the result will be that ‘<Luc venir> validated’ is good (IE I) and ‘<Luc venir> not validated’ is bad (IE E), also implying that in the IE position, ‘<Luc venir> is not the case’, in other words ‘Luc ne pas venir> [Luc not come] is validated’.”
(Translated from Culioli 1988-90, T.1:103-105)

“The category of aspect results from the interaction of two domains that derive their structure from complex interconnections and modulations: on the one hand, the domain of instants, on the other, the notional domain related to a predicative notion. The source of these domains is the enunciative origin S0, T0 […] T0 is the origin for the temporal location of a process (P), S0 is the origin for the validation of a process. These two parameters can operate separately or jointly. The locating parameter T0 provides purely temporal location of occurrences of P, which does not imply any qualitative delimitation of P or any structuring of its notional domain […] At such and such a t, P occurs, and this is what we call factual […] The validating parameter S0 provides a qualitative determination of the process, constructing it independently from its temporal location. More specifically, it is the source of all modal constructions, defining the qualitative properties that characterize P for a given subject.”
(Translated from Franckel 1986:41-43)