The Glasgow Review Issue 3
A Grammar of Narrativity: Algirdas Julien Greimas
The search for precise concepts and coherent systems in analyses of art, which under various headings has been the dominant tendency of the past quarter of our century, has brought as one of its main achievements the emergence of text grammars and, within their scope, studies on the phenomenon of narrativity.
The importance of these studies appears still greater if we keep in mind present-day meanings of the word 'text'. Just as logic does not deal with sentences but with propositions, text grammars operate on the semantic-plane. A text is a semantically coherent whole which may be presented as a surface structure either in natural language (as was the case with the traditional notion), or in any of the sign systems used by mankind. Thus the notion of text is extended in art studies from literature to visual arts, while the greatest benefit accrues perhaps to studies of theatre, where there is a semiotic polyphony of linguistics and paralinguistics, of kinesics, proxemics, and iconics, often supported by music.
Narrative grammars are only part of text grammars, narrative structures not being the necessary condition for the existence of a text; let us mention here, by way of example, a large part of modernist lyrical poetry, or the theatre of Robert Wilson. Nevertheless, these grammars furnish an instrument to analyse the great part of our literary and theatrical heritage which is based on the principle of narration, including 'characters' and their actions.
Purely pragmatic research was started long ago and in a limited field by Vladimir Propp (1928), and later on by Étienne Souriau (1950); further cam from French structuralists in the nineteen- sixties (Roland Barthes, Tzvetan Todorov); but the man who elaborated a complete theory of narrative structures was Algirdas Julien Greimas. His theory, propounded and modified by several steps (Sémantique Structurale 1966, Du sens 1970, Les actants, les acteurs et les figures 1973, Entretien avec A J Greimas sur les structures élémentaires de la signification by Frédéric Nef ,1976) has now obtained fairly wide acceptance, at least among writers on drama and theatre (e.g. in Anne Ubersfeld's two-volume manual Lire le théâtre 1977 and L'école du spectateur 1981). What is perhaps even more important is the fact that Greimas' ideas are now being expanded in the light of modern logic; the leading figure here is Teun van Dijk. These continuators of Greimas drop certain rather intricate formulations of the French structuralist idiom of the nineteen-sixties and instead take up the more universal equipment of predicate logic and intensional logic. As yet, we still do not possess any other complete system for investigating narrative structures than the one elaborated by Greimas. This seems quite sufficient reason for presenting it here in brief outline.
In spite of some loose ends and terminological complications, the system works in practice: a Polish example is the study of the contemporary playwright - Edward Bond - undertaken according to principles of analyse actantielle by the author of the present article.
Let us begin with a few remarks on text grammars in general in which (as has been said) the studies of Greimas take a crucial position with regard to problems of narratives.
The idea which anchors text grammars in the achievements of generative-transformational grammars in general is, in Teun van Dijk's phrase, the hypothesis from which he evolves the basic notion of macrostructure . He says that it is possible to establish a homology between the deep structure of a sentence and that of a text. What corresponds to the structural coherence of a sentence (which enables us to distinguish between a sentence and a non-sentence) is the macrostructure of the text; in this respect a textual whole corresponds to one great sentence. Van Dijk calls the presence of macrostructures (including narrative macrostructures) the central hypothesis of his text grammar. The presence and the actions of 'characters' (or 'superficial constraints' in the terminology of GT grammar) are not occasional, but ruled by deep textual structures arising from the presence of actants. The notion of actant will be explained later on; what requires stressing at this moment is that van Dijk makes a clear distinction of two levels of analysis of texts: the superficial level subordinated to the kind of language used to express the meaning, and the deep level. Narrative structures are analysed at the deep level in their pure form, free of the shaping that is given them on the superficial level. Narrative structures are, then, primary to their manifestations. The grammar of the immanent level generates all the récits through the superficial level.
According to van Dijk 'the theory of narrative structures is part of the theory of man's symbolic actions and may be the object of anthropology as well as semiotics, of linguistics as well as poetics' . Broad vistas open here for studying the historically changing ideologies which model the symbolic activities of man (and the contributions of Soviet semiotics are of particular importance), but what seems to be the most universal point is the thesis that the deep, immanent level of narration combines not only universal structures of the narrative, but also universal virtual structures of human thinking.
The universal structure we have in mind here and which Greimas calls the semiotic square was first formalised by Apuleius of Madaura (De philosophia rationali) on the basis of Aristotle's theory of opposition. Taken over by scholastics it is known in manuals of classical logic as the 'logical square' . The square has the following shape (the symbols S are Greimas' replacement for the symbols of earlier logicians):
The perforated lines mean the relation of contraries, the dotted lines mean the relations of implication, and the solid lines mean the relations between contradictory terms. All the empty places may be filled by concrete or abstract notions. In classical manuals we find the following substitution:
s1 - This solid is hot
s2 - This solid is cold
s-1 - This solid is cool
s-2 - This solid is lukewarm
Axis s1 - s2 in this case could be called fundamental temperature assertion, and s-1 - s-2 non-fundamental temperature assertion, being the negative of the former axis.
With regard to abstract notions we may give the example of fundamental assertions of ideology:
s1 - He is a Marxist
s2 - He is an ardent Catholic
s-1 - He is a Catholic, but he does not deny the Marxist doctrine of social relations.
s-2 - He is a Marxist, but he does not deny the existence of a god.
Axis s1 - s2 in this case could be called fundamental religious ideological assertion, and its negation (s-1 - s-2) non-fundamental religious-ideological assertion.
The semiotic square may be treated as a dynamic model, as a series of transformations, following one another. Each logical relation between the categories of the model may be interpreted as a certain transformation. Since the dominant model covers two elementary structures which are logically connected with each other (s2, s1 and s-1, s-2) in each elementary structure the following relations are possible:
opposition - s1 - s2, s-1 - s-2,
contradiction - s1 - s-1, s2- s-2,
implication - s1 - s-2, s-1- s2.
The number of transformations is determined by the very construction of the semiotic square.
Greimas' contribution is the application of the universal model (which, as we said, he now calls the semiotic square) to narrative structures. He postulates that narrative syntax is based on the transformations of the four elements of the model. The operations are ordered into series and thus the first transformation may also account for the transformations that follow. The grammar of the immanent level generates all the récits through the superficial level.
If syntactic operation is one of the basic notions of fundamental grammar, it corresponds on the superficial level to syntactic acting (faire). The problem here is that of passing from logical operations expressed by universal symbols to principles of narrativity, still treated on an abstract level. We shall try to present this operation step by step while introducing at the same time the basic notions of Greimas' system.
The passage from logical operations to syntactic faire and from a sentence to narration means at first one thing: the subject of analysis becomes 'who does what', not yet in terms of the content of a given narration, but in terms of a syntax that may generate all possible narrations. In this analysis the categories of the imminent level become actants while the logical relations appear as functions. An actant is a class of 'characters' (in the broadest meaning of the term) which in their different manifestations in a narrative have the same function. Actants appear as certain forces in a given text, and are by no means equivalent with concrete characters of a story or with the dramatis personae of a play. The reasons are as follows:
1) An actant may be abstraction (God, Liberty), or a collective character (the chorus in ancient tragedy, a group of characters fulfilling the same tasks - like soldiers of some army), or it may be represented by different characters that all act in a definite way.
2) One character may simultaneously or successively assume different actantial functions.
3) An actant may or may not be present on the stage, it does not have to appear in the utterances of the characters - it may simply be the general abstract notion which is presented on the ideological level of the play.
Basing his considerations on the oppositional value of functions Greimas distinguishes three pairs of actants:
subject - object
giver - receiver
helper - opponent
In some cases, as will be shown later, two actants may perform the same function, but generally we have to do with one force conducted by its own action, e.g. the giver = the subject looking for the object for the receiver. Those who help the subject in his search are actant-helper, those who provide obstacles on his way - opponent.
The abstract and collective character of the actants is clearly presented in an example given by Greimas, which illustrates in terms of actants the story of the Holy Grail. Here the subject appears as knights of the Round Table; the object as the Grail, the sender as God, the receiver as Humanity, the helper as Saints and Angels, and finally the Devil and his acolytes constitute the opponent. Sender and receiver may appear - and they usually do - as abstract notions, and they most often express the motivation of the subject to perform a certain action.
It may also be remarked that the couple sender-receiver determines the ideology of the subject. Greimas presents as examples two philosophical models. One is the classical model in which the subject is Philosophy, the object - the World, the sender - History, the receiver - Humanity, the helper - Spirit, the opponent - Matter. His other model is the Marxist one, in which the subject is Mankind, the object Classless Society, the sender - History, the receiver Humanity, the opponent - the Bourgeoisie, the helper - the Working Class . In both these cases sender and receiver appear as abstract notions and point to ideological concepts, or values to which the subject aspires. The pair sender - receiver is the most ambiguous in Greimas' actantial model, but on the other hand, since actants should be defined a posteriori, after having got acquainted with the whole of the utterance, the sender may be interpreted as the source of knowledge of the subject, and the receiver as the group of people or humanity in general which receives the message - the object looked for by the subject.
The main feature of the actantial model of Greimas is the differentiation between actants - certain forces permeating the utterance, and the characters presented in the utterance. Actants belong to the syntactical order of the utterance and characters to the sematic order, being manifestations of the actants. For the sake of clarity and coherence in this presentation of Greimas' theory, let us defer its further elaboration, which introduces the differentiation: actant - actor - stage character, to the final part of this essay. But we may note that when trying to isolate actants in a text one has to remember Greimas' basic assumption about the syntactic character of the actantial structure and the syntactic analogy between the text and the sentence. Just as we discover the relations and forces by reading a whole sentence, we must determine the actants a posteriori.
Having dealt with functions and actants let us now take up the notion of syntactic faire. It implies the introduction of an anthropological dimension in the grammar; the introduction may be specified in the following way:
1) The faire (action) in question is not 'real' action, situated in the semiotics of the world, but linguistic action (and it is of no moment just what the kind of 'language' is used - natural or visual). Action in this sense forms the message-object situated in the process of communication, implying the existence of sender and receiver . Action is coded in the message, is objectified in it and implies the transmission between sender and receiver. What results is a process of communication which, in its turn, implies the anthropomorphic character of the actants.
2) If, instead of performing a logical operation which is an autonomous metalinguistic process, we introduce the subject of action, one who 'does' something, we imply a human subject. In other words, the acting, or action, is specified by the addition of the marker 'human'.
Thus the action (faire) is anthropomorphic in two ways: either as an action presupposing its subject, or as a message, where it is objectified and implies the existence of the axis of transmission between the sender and the receiver. The passage from the fundamental to the superficial grammatical level may be presented in the case of simple narrative utterance (Énoncé narratif, EN) by the following formula: EN=F(A), where the action (faire) in the process of actualization is called function (F) and the subject of the action is called actant (A). The operations of the fundamental grammar may be converted into EN, whose minimal form is F/A) .
At this stage, while discussing the notion of action, we must introduce the notion of modalities. They are certain expressions attached to propositions (to operators in logic) such as e.g. possibility, impossibility, and necessity. In narratives the modalities are necessary constituents of a story by dominating a whole sequence of actions as a global constituent: a modal category takes charge of the constituent and the message and organizes it by establishing a certain type of relations between the constituent linguistic objects.
To give an example: if a certain class of functions is specified by the marker 'want' (vouloir), the actants will appear as actants-subjects. The 'want' is the anthropomorphic marker (classème) which puts the actant in the position of the subject, or, in other words, as the operator of the action (faire). In this way, apart from descriptive utterances (Énoncé descriptif, ED) one may have a modal utterance (Énoncé modal, EM).
E.g. 'John wants Peter to leave' - is modal, while 'Peter wants to leave' - is descriptive. In the first of the utterances John is the actant-subject and Peter an actant-object, while in the second Peter is the actant-subject.
Thus a modal utterance will appear as EM = F: want (S, O), or we may say that a modal utterance is a function of the marker 'want', pointing to the subject and the object and at the same time distinguishing them. If we add one more restriction, postulating that the subject of the modal utterance and descriptive utterance should be the same, one can say that the syntactic faire is the transformation of a virtual programme into actualized programme.
The modality of 'want' allows the construction of modal utterances with two actants: the subject and the object. The axis of 'wanting' or desire which joins them permits subsequently to interpret them as virtual acting subject and object which may achieve a modal value. To the basic modality of desire (vouloir) Greimas adds such epistemic and alethic modalities as 'knowledge' (savoir) and possibility (pouvoir)
The fundamental sequence of Greimas' modal values may be presented as want - know, i.e., how - be able - do
(vouloir -- savoir - pouvoir - faire) .
These values, as has been said, organize certain types of relations on the syntactic level. On the superficial level we have to deal with the equivalent of the transformations - the actual 'doing' (faire), which Greimas calls performance (using the term in the sense of generative-transformational grammar as realization of the modal values, the acquisition of which is presupposed in any performance). If a narrative utterance consists of a number of performances, it means that it also presents the introductory attempts to achieve modal values according to the above-presented sequence.
The hierarchy of modal values in the basis for organizing syntagmatic sequence of performances. The actant-subject must at first want or desire to achieve an aim, but this very fact is not the only performance necessary for achieving it. He must know the way to achieve it in order to be able to act accordingly. In a situation in which the subject possesses the knowledge about the necessity of acting, he is called a 'knowing subject' - sujet savant. In a situation in which the knowing subjects possess at the same time the modal value of 'being able' they are called 'capable subjects'- sujets puissants, having the possibility of achieving the aim.
The acquisition of a modal value (desire, knowledge, capacity) may be manifested, as has been indicated, by obtaining of a magical object or some information which enables the subject to arrive at the right decision. The object or the information appears then as a helper.
The problem of modal utterances is of particular interest in the study of drama. After the first overall reading of a play, we may observe the actions of characters and the motivations of these actions. The next step is to abstract from the actions the performances and the modal values as set out in what has been covered so far. The subjects of the performances will be actants - the main forces of the drama - which we may classify as willing, knowing, able (or not) to act. Analysis now becomes objective, based both on the data of the fundamental and of superficial grammar.
Motivations may be evident from the very beginning of the text and then we know that we have to do with a knowing subject, but it may happen that motivations remain unknown till the very end and thus the syntagm of performances of the actant qualifies him as the actant who is willing to act, but who does not posess the knowledge or capacity of acting. In this case the performance of the actant would be limited to willingness (vouloir) to achieve an aim, e.g. to change his way of life without any particular knowledge what the change is to bring.
Further important contributions which render Greimas' system both more precise and comprehensive are included in his essay Les actants, les acteurs et les figures and his Entretien sur les structures Élémentaires de signification - a conversation withe Frédéric Nef . It should be kept in mind all the time that the interpretation of narrative utterances depends on structures which are both unchanging and previous to verbalization. It is verbalization which allows, as a step that follows, to articulate the sense.
To begin with, Greimas refines his earlier notion of the actants by turning once more to the the semiotic square. For the sake of clarity and convenience we reproduce it once again:
Let us pay attention to the relations of implication represented by the dotted lines visualizing the deixis s1 - s-2 and s2 - s-1. If, at the end of the axis of contraries (the perforated lines), we assert the subject and the object, or the sender and receiver of s1 and s2, we obtain at the end of the axis of implication the notions of anti-subject and negative object, or negative sender (anti-sender) and negative receiver (anti-receiver).
A generalized visual scheme of the above reasoning is presented in the Entretien as the actantial square of J. C. Picard:
The anti-actants may be distinguished especially in texts in which there exists a rigid division into good versus bad group of actants and in which the notions: opponent and helper are not sufficient to present the whole network of relationships. An example for the actant-subject and anti-actant-subject (anti-subject) might be Shakespeare's Hamlet, where originally
the father is sender
Hamlet is subject
Claudius is object,
while - Mother and Polonius are opponent.
Because of the development of the situation, Polonius' son, Laertes, becomes unwillingly not only opponent, but anti-subject, being at the same time in a parallel situation to the initial one of Hamlet. At this state of the development of the narrative, those who were the helpers of Hamlet become negative anti-actants and those who were his friends become negative actants .
The next elaboration deals with the concept of performance. As has been said above, this notion is connected with the subject of an action (sujet-du-faire) who, in order to perform it, must acquire the prerequisite modalities. Now the reasoning can become more precise thanks to a more consistent application of the terminology of generative-transformational grammar.
To the notion of performance corresponds the twinned notion of competence - of being in possession of the generating system (in the terms of de Saussure we might speak of langue and parole). In narratives, competence is the sequence of modalities which the subject must acquire as the presupposition of the performance.
If the competent subject happens to be different from the performing subject, they are not constituted around two different actants, but are two different instances of the same actant. Let us take an example from John Osborne's Look Back in Anger, where Jimmy Porter and his friend Cliff are the only actant-subject, Jimmy Porter being the active and competent subject, while Cliff is merely someone who accepts the situation, unwillingly helping Jimmy, but possessing no knowledge whatsoever about the actual situation. Mrs Warren's Profession by G. B. Shaw provides another such example; there again there are two instances to the same actant: Vivie is the competent subject, while Mrs Warren herself is the one who unwillingly performs actions that make Vivie arrive at the final decision.
Between the abstract syntactic actantial and its concrete surface realization in a narrative (in our case a play), Greimas now introduces the intermediate notion of actor (acteur) and role. Although these notions appeared in his earlier writings (Sémantique structurale), we have thus far omitted them in our presentation as they take their final shape only in the essay Les actants, les acteurs et les figures .
The starting point for the new definition of Greimas is his previous observation that one actant may be represented by many characters and that one character may take on several actantial roles. The simple notion of character (personnage) proved insufficient to analyze this phenomenon and involved the necessity of introducing the notion of actor - a semiotic term which has nothing to do with the meaning of the word player or comédien.
To explain the notion it may be convenient to start with a linguistic analogy used by Anne Übersfeld in her explanation of Greimas' system . A lexical item (lexème nominal) is a manifestation of the deep semantic structure on the superficial level. It is a bundle of semantic markers (sèmes), but such a bundle (a 'word' in terms of everyday language) does not possess individuality. Passing from linguistics to narrative theory we may replace the notion of semantic markers by the notion of characteristic features based on the type of action performed by our hypothetical unit - the actor. It is an element with the marker (+human) which is characterized by functioning identically in different situations, even if it functions under different names.
The suitors for the hand of a princess in a folk-tale, although individualized as concrete characters, have one function in common: pretending to the hand of the princess. Thus, they are one and the same actor, although not the same actant, only one of them, the winner, is the actant- subject, the rest form the actant-opponent. On the other hand, in Molière's Le Misanthrope all the petits marquis who surround Célimène are one actor (they are endowed with the identical constant function of wooing) and at the same time one actant (the opponent to the subject - Alceste). Scapin in Molière's Les Fourberies has different actantial roles according to the different angles of analysis and the different sequences of this play, yet he is always the same actor - the 'master of roguish tricks', this being his repeated action. Also two actants - subject and anti-subject - may be united in one actor. Such is the case if we consider the inner struggle of Goethe's Faustus. So, different actants and different actantial roles may be manifested in one actor and inversely, actantial roles may be divided among many actors, thus accounting for more complex patterns than the 'bare' actantial structures.
To sum up: actants are abstract elements of the syntactic deep structure which may serve as base for a number of different texts. Actors are also abstract elements, but of a semantic character; they are semantic concepts to be deduced from an individual story, from a particular text in which they form actorial structures. Actantial structures are patterns of general narrative syntax connected (via the semiotic square) with the general patterns of human significative process; actorial structures are semantic patterns underlying one given story with definite meaning. They mediate between the general rules and their surface meanings or manifestations in concrete actions of the characters.
The last of the factors on the semantic level which Greimas introduces in his system are figures . The notion is taken from Hjelmslev's glossematics: a figure is the minimal category of sense on the context plane, just as a phoneme is the minimal unit of natural language on the level of expression. Figures of content are rendered apparent by arranging minimal semantic oppositions. In pair girl - boy we have the figure 'male' - 'female' (sex); whereas in the pair girl -- heifer we have the figure 'human' - 'animal' (shape). The figure of the first pair is part of the content words woman, girl, mother, cow, mare, ewe, hen, as well as of the words man, boy, father, bull, stallion, ram, cock. The figure of the second pair is part of the content of the words boy, girl, child, chicken, lamb, foal, as well as of the words man, woman, grown up, cock, horse, stallion . Hjelmslev's early approach became several decades later the basis for compositional analysis with its notion of semantic markers or semes.
What is basic for the taking up of Hjelmslev's notion of figures by Greimas is his thesis that narrative forms are particular instances of the semiotic content form. Figures are, then, the organizing principle of the sense of a discourse. They create discursive configurations which exist apart from the narrative configurations of texts. The narrative programme is deliberately chosen from the frame of narrative grammar; the discourse programme is related to the discursive dictionary of which the figures are minimal units as 'forms of content'; they create configurations which really come from the universe of man.
For instance, the figure 'sun' is the point of departure for a net of relationships which allow it to be placed in various contexts and to appear in various configurations in discourses which may use only part of its virtual meanings, such as rays, light, heat, air, transparence, opacity, clouds, or even gods. The figure 'sacred' may be represented, for instance, in a folk-tale by priest, sacristan, or beadle.
These configurations organize the text on the discursive plane, just as actantial roles organize the narrative plane. In a discourse, such configurations form a syntagmatic chain, and what underlies such a syntagm is a thematic role. This can be exposed by reducing the syntagm of configurations to one figurative course. As a second step, we distinguish in such a course a simple thematic role which manifests itself in different configurations. As example Greimas gives the chain of culinary figures which is maintained throughout the Bororo myth of the origin of life .
Thematic roles are semantic elements; in turn, they are taken charge of by the syntax of narrative grammar - the actantial system - and they assume actantial roles as we saw at the beginning of this presentation. The process of semantic investment of the syntactic system has been established by Greimas as the basis for his notion of actor. The figures of the thematic roles become equal then to actors.
On the syntactic level the actor possesses at least one actantial role (a role in the structure of forces acting in the narrative); on the semantic level the actor possesses on thematic role (the role of carrying meanings in the structure of content of the narrative). Actantial roles and thematic roles converge in the actor, making possible their transformations, operating narrative objects, and taking part in gaining and losing of modal values.
The introduction of the notions of figures and thematic roles serves several purposes: within the framework of Greimas' system. It accounts, as a mediating concept, for the passage from discursive to narrative structures, for the linking together of the grammatical and the semantic components of the system.
In the analysis of narrative texts (non-dramatic, dramatic, or posysemiotic theatrical texts- representations) Greimas' theory gives a precise well-formed meaning to the traditional intuitive notion of theme. Particularly in the study of drama the description of thematic roles makes possible to introcduce into the discussion psychological, social, and political problems which may be manifested in actors since actors in his system belong to the semantic plane of the narrative.
Generally speaking, Greimas' narrative grammar in its fully elaborated form gives due place to the semantic component. It gives us the means to study the artistry of narrative composition in adequate, precise terms, but does not preclude the interest in what the artist has to tell us. It serves as an important intratextual support for any extratextual material that may be at our disposal.
APRESJAN J. (1971) Idei i metody sovremennoi strukturalnoi lingvistiki, Polish transl. Z. Saloni, Koncepcje i metody wspó_czesnej lingwistyki strukturalnej, PIW, Warszawa.
DIJK VAN T. (1974) 'Grammaires textuelles et structures narratives,' in: Sémiotique narrative et textuelle, ed. C. Chabrol, Larousse, Paris.
GREIMAS A. J. (1966) Sémantique structurale, Larousse, Paris.
GREIMAS A. J. (1970) Du sens, Editions du Seuil, Paris.
GREIMAS A. J. (1974) Les actants, les acteurs et les figures, in: Sémiotique narrative et textuelle, ed. C. Chabrol, Larousse, Paris.
GREIMAS A. J. (1976) Entretien sur les structures élémentaires de signification, ed. Frédéric Nef, Editions Complexes, Bruxelles.
ÜBERSFELD A. (1977) Lire le théâtre, Editions sociales, Paris.
ÜBERSFELD A. (1981) L'école du spectateur, Editions sociales, Paris.
T. van Dijk (1974), p.204.
For the history of the square see: A. de Libera, La sémiotique d'Aristote, in: Structures élémentaires de signification, ed. Frédéric Nef, Editions Complexes, Bruxelles 1976. pp. 28-28.
The explanations of Greimas' theory are based on: A. J. Greimas, Sémantique structurale, Larousse, Paris 1966, pp. 30-182.
A. J. Greimas, Du sens, (1970), pp. 166-168.
A. J. Greimas (1974), Les actants, les acteurs et les figures, in: Sémiotique narrative et textuelle.
A. J. Greimas (1976), Entretien sur les structures élémentaires de signification, in: Structures élémentaires de signification, pp. 18-26.
In his earlier writings the actor was treated as the immediate particularization of the actant on the surface level - the actor then corresponded to the traditional notion of character (personnage).
A. Ubersfeld, Lire le théâtre, (1977,) pp. 108-109.
A. J. Greimas (1974), Les actants, les acteurs, pp. 166-169.
Hjelmslev's argument is discussed further by Jurij Apresjan, in Koncepcje I metody wspólczesnej lingwistyki strukturalnej. Polish trans. Z. Saloni, PIW, Warszawa 1971, pp. 84-85.
A. J. Greimas (1970), Du sens , pp. 138-140.