You are here : Home Page / Theories / Rastier / Interpretive Semantics

Available Languages

Follow us on

SignoSemio's Facebook SignoSemio's Twitter

Interpretive Semantics

By Louis Hébert

Professor, Université du Québec à Rimouski

1. Abstract


François Rastier

We will present here the principal notions of François Rastier's interpretive semantics: morpheme/lexia, sememe/semia, actualized/virtualized seme, inherent /afferent seme, specific/micro-, meso-, macrogeneric seme, specific/generic isotopy, semic molecule, topos, metaphorical/symbolic connection, rewriting, micro-/meso-/macrosemantic levels of analysis, semantic graph, dialogics, and so on.

This text can be found in extended version in this book:
Louis Hébert, Dispositifs pour l'analyse des textes et des images, Limoges, Presses de l'Université de Limoges, 2007.

Click here to obtain the English translation of this book.

This text may be reproduced for non-commercial purposes, provided the complete reference is given:
Louis Hébert (2006), « Interpretive Semantics » with Lucie Arsenault’s collaboration, in Louis Hébert (dir.), Signo [online], Rimouski (Quebec),



Interpretive semantics was founded by François Rastier (see Rastier, 1997 [1989], 1991, 1994, 1996 [1987] and 2001; Hébert, 2001), a student of Greimas and Pottier. It is a "second-generation" synthesis of European structural semantics, developed in the wake of Bréal and Saussure, then Hjelmslev, Greimas, Coseriu and Pottier.


According to interpretive semantics, four components make up the semantic plane of texts (the plane of content, or signifieds, as opposed to the plane of expression, or signifiers): thematics (the invested content), dialectics (states and processes and the actors they involve), dialogics (modal evaluations, for example, veridictory (true/false) and thymic (positive/negative) categories) and tactics (the linear sequencing of content).


Microsemantics is associated with the lower levels of the text (from the morpheme to the lexia), mesosemantics with the intermediate levels (from the functional syntagm to the period; this level can go beyond the sentence) and macrosemantics with the higher levels of the text (beyond the period and up to the text level). As a simplification, we will say that these three levels correspond to the word, the sentence and the text, respectively.


The conventional markers used in interpretive semantics allow us to distinguish between "sign", signifier, 'signified', /seme/ and/isotopy/, //semantic class//, → |rewriting|.


Semantic units operate on two levels. The type is a unit that is manifested to varying degrees through its tokens. For example, the context-independent content of the morphemes "water" and "love" are types, and their actual content tends to vary depending on their occurrence in different expressions and sentences. The minimal linguistic sign is called a morpheme. A lexia is a functional unit that includes more than one morpheme. Sometimes a lexia corresponds to only one linear morpheme position: For example, "water" is a lexia, since "zero" morphemes are "superimposed" in it. A lexia may correspond to one word or more than one word ("water", "brussel-s sprout-s", "walk-ing"). The word is a unit that is quite easily definable by its graphic signifiers - it is preceded and followed by a space - and for this reason, we favour it over the lexia. The sememe is the signified of a morpheme and the semia is the signified of a lexia.


In order to simplify the representation of sememe analyses, a sememe is generally designated by the word in which it occurs (for example, 'fuse' and 'spirit' stand for the sememes 'fus-' (cf. "fusion") and 'spir' (cf. "inspire")). The generic term signified includes the sememe (the signified of a morpheme) and the semia (the signified of a lexia), but also encompasses other semic groups that act on the same level as the morpheme and the lexia or on higher levels: the syntagm, the period (a group of sentences, more or less), the section (a chapter, for instance), and the text. Semic molecules, for instance, are found at all the levels of analysis.


The signified of any semantic unit is composed of semes, or features of content. A generic seme marks the fact that the sememe belongs to a semantic class (a semantic paradigm, made up of sememes). A specific seme distinguishes a sememe from all other sememes of the same class. The specific semes of a sememe constitute its semanteme; its generic semes make up its classeme. There are three kinds of generic semes: microgeneric, mesogeneric and macrogeneric. These correspond to three kinds of semantic classes: taxemes (the minimal classes where sememes are interdefined), domains (which are linked to the social context and correspond to spheres of human activity; dictionary field labels exemplify this, as in chem., phys.) and dimensions (the most general of classes, grouped into oppositions, such as //animate// vs. //inanimate//, //concrete// vs. //abstract//, //human// vs. //animal//, //animal// vs. //vegetable//, etc.).

For example, the taxeme //tableware// (eating utensils) includes three sememes. Each one contains the microgeneric seme /tableware/ and is distinguished from the other sememes of the same taxeme by a specific seme: /for piercing/ in 'fork', /for cutting/ in 'knife' and /for containing/ in 'spoon'. Since this taxeme comes under the domain //food//, the three sememes also contain the mesogeneric seme /food/. And finally, the three sememes are also members of the common dimensions that define macrogeneric semes, like /inanimate/. (We use "inanimate" not to designate things that are dead or do not move, but things that cannot be alive, such as a rock or freedom.)


A seme belonging to a sememe's type is called an inherent seme, and in context it is actualized by default, unless there is some instruction to virtualize it (neutralize it). Afferent semes are semes that are present only in the sememe's token, that is, only by contextual indication.


There are two possible kinds of connections between sememes (or groups of sememes). A metaphorical connection links two sememes that are present in a linguistic chain (as in a comparison). A symbolic connection (for example, the in absentia metaphor, where the compared term is absent) links two sememes, one of which is present in the chain, the other of which is present in the reading: In the political utterance "The Eagle overcame the Bear", |'United States'| and |'USSR'| are rewritings that are present only in the reading. The two sememes linked in a connection have at least one incompatible (generic) seme and at least one identical (specific) seme. For instance, in the expression "this woman is a flower", the metaphorical connection involves the incompatible semes /human/ and/vegetable/, while a seme like /beauty/ is present in both sememes.


The iteration of a single seme in context — either inherent or afferent — forms an isotopy (see the chapter on isotopic analysis). Isotopies are distinguished not only by the name of the seme on which they are based (for instance, /inanimate/, /religion/), but also by the type of seme involved (specific/micro-, meso-, or macrogeneric). For example, the sentence "I only use a knife for picking up peas" contains the (mesogeneric) isotopy /food/, which indexes the sememes 'knife' and 'peas'. In addition, it virtualizes the inherent specific seme /for cutting/ in 'knife' and actualizes the afferent seme /for taking/. Allotopy is the oppositional relationship between two sememes (or groups of sememes, such as a lexia, for example) that have incompatibles semes (as in 'black snow', for example).


All of the semes actualized in a context, whether inherent or afferent, define the meaning of that particular unit. The signification is the set of (inherent) semes of a given unit abstracted from its context and its communicational situation. Interpretation is the process of assigning meaning to a linguistic chain. Interpretation produces a reading. Intrinsic interpretation reveals the semes present in a linguistic chain and yields either a descriptive reading or a methodological reductive reading (purposely restricted). Extrinsic interpretation adds semes (a productive reading) or mistakenly leaves semes out (a reductive reading).


An interpretant is an element of a text or its surroundings (non-linguistic context) that allows one to establish a semic relationship, that is, to definitively actualize or virtualize at least one seme. For example, when the phonic signifiers are identical (homophonic), as in Hamlet's response to King Claudius: "Not so, my lord: I am too much in the sun [son]", then the mesogeneric semes /weather/ and /filiation/ can be realized simultaneously. (For an analysis of homonymy, cf. Hébert 1998). A topos (topoï in the plural) is a socially-normed interpretant that can often be expressed as an axiom (for instance, the countryside is preferable to the city in rural legend).

A rewriting is an interpretive operation of the type X → |Y|, by which one or more signs, signifiers, or signifieds is transposed into one or more different signs, signifiers, or signifieds. The source unit (X) is part of the object-text, and the target unit (Y) is part of its reading (although it can have correspondences in the source text).


A semic molecule is a grouping of at least two co-recurrent (appearing together) semes (usually specific) whose tokens are lexicalized in various ways, with the constituent semes of one type reappearing in varying combinations. We can study the formation, maintenance (full or partial) and possible dissolution of a semic molecule by following the thread of its occurrences. The variations in typicality of the tokens can be interpreted as variations in the prominence of the molecule (the intensity of its presence or actualization). An isotopic bundle is a group of isotopies that index more or less the same units (the same sememes, in the most exacting analysis). A molecule produces or gives rise to a bundle of isotopies (usually specific). Let us give an example of a molecule. In the poem "The Golden Ship", by Nelligan, there is a semic molecule made up of the semes /precious/ + /dispersion/. It appears at least three times: in the "streaming hair" ("cheveux épars") of the beautiful Goddess of Love who was "spread-eagled" ("s'étalait") at the ship's prow, and also in the sinking of the ship's treasure, which the sailors "disputed amongst themselves" ("entre eux ont disputés"). This molecule is in opposition to another, with which it overlaps partially, composed of /precious/ + /concentration/, and which is found in the "solid gold" ("or massif") of the ship, and even the "blazing sun" ("soleil excessif").


Semantic cases, which are limited in number, are semantic primitives or universals of method (and not of fact) that allow us to portray relationships between semes, and thereby integrate the semes of a unit into a semantic structure. Interpretive semantics uses primarily (although not exclusively) the following semantic cases: ACC (the accusative): a patient of an action; ATT (the attributive): a characteristic; BEN (the benefactive): an entity benefiting from something; COMP (the comparative): a comparison; DAT (the dative): an entity receiving a transmission; ERG (the ergative): the agent of an action; FIN (the final): a goal; INST (the instrumental): a means used; LOC (the locative): spatial location (LOC S) or temporal location (LOC T); and RES (the resultative): result, consequence. For example, saying that a woman is beautiful implies the following semantic structure: the semes /woman/ and/beauty/ linked by the attributive case (ATT). In another example, the type for the sememe 'kill' admittedly covers a process involving the semes /inanimate/ and /animate/, but these semes vary in the ergative (the thing doing the killing can be animate or inanimate), whereas only the second seme is found in the accusative (whatever is killed is, by definition, animate; although /animate/ could be virtualized in the context and replaced by /inanimate/, as in the expressions "killing time" and "Kill your television", for example). Semes and cases are the two components of signifieds (not to be confused with the four semantic components).


Semantic graphs (based on Sowa, 1984) (see the chapter on semantic graphs) are a convention used to visually represent semantic structures (semes and the cases that link them together). Cases are the links between the semes (actors, for example), which are constituted in nodes. Two formats are used for semantic graphs: the proposition and the straight graph. The proposition is a textual format, with the links shown in brackets and the nodes in parentheses. For example, the structure mentioned previously could be represented thusly: [animate] or [inanimate] ← (ERG) ← [KILL] → (ACC) → [animate]. In the strictly graphic format, we use ellipses and rectangles, respectively. In both formats of a semantic graph, the arrows indicate the direction of the relationships between nodes.


It would be impossible to give a complete presentation of the possible applications of interpretive semantics in the assigned space. In other chapters of the site you will find applications for isotopic analysis, dialogics and semantic graphs.

Here we will give a brief semic and isotopic analysis of the title of a novel: Black Snow, by the prominent French-Canadian author, Hubert Aquin (1978).

Semic analysis of the title of a novel by Hubert Aquin
Semic analysis of the title of a novel by Hubert Aquin

A few explanations are in order.

Independent of context (at the level of langue), the sememe 'Snow' contains the inherent microgeneric seme /precipitation/ (which refers to the taxeme //precipitation//, which includes the sememes 'snow', 'rain', etc.) and the specific inherent seme /whiteness/ (which distinguishes between 'snow' and 'rain', for example, within the taxeme). In context, the second seme is virtualized by the effect of the qualifier: It says that the snow is black, and by correlation, the afferent seme /blackness/ is actualized for the same reason.

Independent of context, the sememe 'black' contains the microgeneric inherent seme /colour/ (which refers to the taxeme of //colours//, which includes sememes like black, white, etc.) and the inherent specific seme /blackness/ (which distinguishes between 'black' and 'white', for example, within the taxeme). Both of the semes are actualized in context. Since the seme /blackness/ is actualized in two different signifieds, 'snow' and 'black', the isotopy /blackness/ is created.

The title refers to a topos, or literary (and non-literary) common place that renders black as a dysphoric, harmful element (for instance, we find it in Nerval: "black sun", "black spot"). For this reason, the macrogeneric afferent seme /dysphoric/ is actualized in 'black' (and since it is actualized through a topos, we can call it a sociolectal afferent). Since the snow is said to be black, the same seme is actualized in 'snow'; but the sememe 'snow' is itself a potential carrier of the afferent seme /dysphoric/, also by virtue of a topos. Therefore, both semes reinforce each other mutually and become salient as a result; the two sememes serve as interpretants for each other. However, the effect of salience, or prominence, is achieved primarily to the noun's advantage, due to the direction in which qualification operates (actualizations generally go from the adjective to the noun). As the seme /dysphoric/ is actualized in two different signifieds, the isotopy /dysphoric/ is formed. This macrogeneric seme refers to the dimension //dysphoria//, which is in opposition to the dimension //euphoria//.

Moreover, since the semes /blackness/ and /dysphoric/ are co-recurrent in two different signifieds, the title thus contains the semic molecule /blackness/ + /dysphoric/.

Obviously, other semes are featured in the title, specifically, /woman/ (cf. in Aquin's diary (cited in Aquin, 1995, XXXIV): "la nuit féminoïde", "la femme obscure", "la femme est noire", etc.) and /literature/ (cf. the "roman noir", which, like Black Snow, is characterized by eroticism and religious references). We do not claim to have exhausted the meaning of this title, but let us finish with a very probable symbolic connection. The sign "snow", as we know, can refer to "cocaine powder" (Le Petit Robert and The American Heritage Dictionary). If homonymy is recognized, the symbolic connection is based on the opposition between the mesogeneric inherent semes /weather/ and /addiction/ (and/or the microgeneric inherent semes /precipitation/ and /drugs in powder form/) of 'snow' and |’cocaine’| on the one hand, and their identical specific inherent semes /whiteness/ on the other. (There are undoubtedly other identical specific semes). According to some who were close to him, "Aquin was addicted to medications (particularly amphetamines), which he had used liberally since his youth [...], to maintain his "dynamism" " (Aquin, 1995, 175). In 1963 he went through a detox program during the three months of hospitalization following a suicide attempt (Aquin, 1995, 202). There is a significant isotopy /medication-drugs/ in several of his works (for example, Next Episode and "De retour le 11 avril"), and in Black Snow (1978, 262), Linda says: "C'est comme si j'étais intoxiquée par un divin poison..." ("I felt like I had been drugged with a divine poison.")


AQUIN, H., Neige Noire, Montreal: Le Cercle du livre de France, 1978.
AQUIN, H., Prochain épisode, Édition critique de l'oeuvre d'Hubert Aquin, Vol. III, 3, ed. J. Allard, Montreal: Bibliothèque québécoise, I-LXXXIV, 1995.
HÉBERT, "Onomastique, homonymie, paronymie et polyglossie. L’exemple de Prochain épisode", Degrés, Brussels, 1998, 26-94, summer, d1-d21.
HÉBERT, L., Introduction à la sémantique des textes, Paris: Honoré Champion, 2001.
RASTIER, F., Sémantique interprétative, Paris: Presse universitaire de France, 1987.
RASTIER, F., Meaning and Textuality, trans. Frank Collins and Paul Perron, Toronto: University of Toronto Press, 1997.
RASTIER, F., Sémantique et recherches cognitives, Paris: Paris: Presse universitaire de France, 1991.
RASTIER, F., Sémantique interprétative, Paris: Paris: Presse universitaire de France, 1996 [1987].
RASTIER, F., Arts et sciences du texte, Paris: Paris: Presse universitaire de France, 2001.
RASTIER, F., M. CAVAZZA and A. ABEILLÉ, Sémantique pour l'analyse, Paris: Masson, 1994.
SOWA, J., Conceptual Structures, Reading (Massachusetts): Addison-Wesley, 1984.


A. Answer the following questions:
  1. Find the common mesogeneric seme: 'table', 'chair', 'desk'.
  2. Find the common microgeneric seme: 'cigarette', 'cigar', 'pipe'.
  3. Find the common macrogeneric semes: 'man', 'fox', 'rock'.
  4. Find the common macrogeneric semes: 'glory', 'friendship', 'brotherhood'.
  5. Find the sememes in the taxeme //breakfast citrus//. For each sememe, find a specific seme that distinguishes it from the others.
  6. Find one semic molecule present in each of the following words: "Died, croaked, perfected, closed".
B. Quote a short sentence from a well-known literary text (for example, Rimbaud's "Sleeper in the Valley") where, in the transition from type (langue) to token (parole), at least one seme not found in the type is actualized and one seme found in the type is virtualized. Justify your answer.

Actualized seme: / __________ / in the signified '__________'

Virtualized seme: / __________ / in the signified '__________'

C. For each of the two following metaphorical connections, find the incompatible generic semes and the common specific semes: "Where is my heart, that empty ship, oh where? // Alas, in Dream's abyss sunk without trace." (Nelligan, "The Golden Ship")
D. Give semes that exemplify the four types or classes of semes defined in interpretive semantics (inherent/afferent, actualized/virtualized, specific/ micro-, meso-, macrogeneric) and justify your answers. Model: The seme / __________ / in the signified '__________' is a __________ seme.

Share this page