Rastier, François: "On Signs and Texts"


And yet, neither contemporary semiotics nor cognitive research has brought forward a theory of text compatible with rhetoric-hermeneutic problems.
(i) Certainly, the descriptive practices of contemporary semioticians often go beyond the logico-grammatical theories they advocate. They created the semiotics of discourse, developed narratology to exceed the confined framework of linguistics.1 And yet their theories are still premised on signification (which pertains to the sign), not on meaning (which pertains to the text). Hjelmslev, when he chose commutation as the test to identify linguistic units at all levels, unified the definition of content on the paradigm of sign __ signification or denotation being defined as the relationship between a unit on the plane of content and the corresponding unit on the plane of expression. Greimas's theory, more complex on this point, distinguishes signification and meaning, but makes the one proceed of the other. Specifically, Greimas's generative trajectory, through a series of conversions, attempts to derive textual meaning from the elementary structure of signification, emblematically summarized with a weak Boolean square called "semiotic square" __ which still testifies to the logical origin of the concept of signification.2
Semantics and semiotics have entertained dubious relationships.3 But it matters more here that semiotics (insofar as it is limited to signs) has produced theories of signification only, whereas semantics (when it treats texts, that is) is bound to produce theories of meaning.
(ii) In the field of cognitive research, the most prominent theories of text remain attached to the logico-grammatical branch in two principal fashions.
__ Textuality is summarized as sentence-level phenomena which extend on to neighbouring sentences (agreement in tense, anaphora), which are as many local isotopies. Albeit interesting, the research on macrosyntax and paragraph- or period-level semantics remains within the limits of text and textuality.
__ Another approach is given as a reply: that of reducing the text to a set of propositions. Its principle is well known, and Van Dijk gave a noteworthy illustration of it: after the sentences have been coded into propositions, the propositions considered to be secondary are removed, so that one proposition, called macroproposition, is kept, which supposedly represents the text.4

2. Problems of the Text and the ALP Breakthrough

The situation is changing, however, and as it is often the case more quickly in practice than in theory. Let us consider for one moment what happened with automated language processing (ALP). Far from being an unspecified province of linguistics or data processing, ALP, along with the Chomskyian current, has provided some of the major conceptions behind cognitivism.
In retrospect, in computational linguistics, data processing seems to have fallen victim to the charge of computationalism, while linguistics was confronted with the dire limitations of the logico-grammatical paradigm __ such limitations are quite obvious, though, as soon as we agree that a text is not merely a series of sentences. Logico-grammatical theories have made the construction of morphosyntactic analyzers, and of formal proposition representation, such as by conceptual graphs, possible. In spite of their practical merits, they are on the wane, at least as global explanatory and descriptive theories.


1. Generative discourse analysis and narratology, drafted recently in opposition to linguistic theories exclusively reduced to morphosyntax, are the best known semiotic theories and regard the linguistic level as a surface variable (Greimas). As the field of linguistic studies expands, the principle of autonomous discoursive semiotics becomes increasingly difficult to defend. [RETURN]

2. The generative trajectory produces textual meaning, which belongs to "surface" linguistic structures, by starting out from the elementary structure of signification, set up into a constitutional model of all semiotics, by definition the "deepest" model one can imagine. As for myself, I merely consider this elementary structure (presented in Greimas and Rastier 1968) as one of the attested structures among basic lexical classes. [RETURN]

3. For example, Hjelmslev's semiotics (1957) came before Greimas's structural semantics. But Greimas's semiotics, presented in On Meaning in 1970, originates from his Structural Semantics (1966). [RETURN]

4. In the early Sixties, Nicolas Ruwet showed the way and offered the macroproposition "I love you" as a summary for a sonnet by Louise Labé (1964). The characterizing value of this kind of description seems low, because thousands of other sonnets written at the time would of course have produced the same result. [RETURN]

Page - 1     Page + 1

AS/SA Nº5, Article 4 : Page 19 / 27

© 1998, AS/SA

E-mail to the editors
Pour écrire à la rédaction