A convergence of the last three types of research is still possible. That coming together
supposes, on the one hand, a shared rejection of objectivism and immanentism, and correlatively
moving from logico-grammatical questions on to rhetoric-hermeneutic questions. It does not matter
whether research becomes more or less scientistic or scientific, according to preference: as it is taking
textual complexity into account, it is already showing higher descriptive capacity, which in the mid-term will allow us to side with effectiveness.
There is still a state of the art to be made, a synthesis to be
sought. At the same time, it is necessary to confront, evaluate, and combine the various analytical
methods used to describe scientific and technical texts, and literary or mythical texts. A textual
typology must be built in terms of concepts borrowed from semantics (since textual structures are
primarily semantic). The linguistic description of scientific and technical genres has not yet been
systematically attempted, at least not in France, unlike in Denmark or Germany
(Fachtextlinguistik) for instance.1
2. Rebuilding Semiotics on Semantic Foundations
In all likelihood, contemporary semiotics derives its hermeneutic deficiencies from the
language sciences, which privilege the sign (where reference is located) and propositions (where truth
is located), and show reluctance in fact to treat texts. Benveniste was aware enough to phrase the
problem: "semantics is 'meaning', insofar as it results from the way signs are linked together, how
appropriate they are to such and such a situation and how compatible they are with one another.
That's an absolutely unforeseeable vulgar fact. It's an opening to the world. Whereas semiotics is
meaning closed back on itself and to some extent self-contained" (1974, p.21). The proposition is
the boundary between the two disciplines, or at least it is the border between their respective objects. However pervasive it may be, this
separation merely expresses a state of affairs: logico-grammatical problems aim for signs and their
syntactic combination, which does not span beyond the proposition. On the other hand, rhetoric-hermeneutic problems aim for the text and all its levels, down to that of the word. The meaning of
a word constantly depends on the text that includes it (this is truistic for the semantics of literary
texts, for example).
The paradigm of sign, which falls to the lot of logic and the philosophy of language to study
because of its content, and to the grammatical tradition, which culminates in contemporary morpho-syntactic research, because of its expression, is an attribute of semiotics, whereas semantics is
justifiably associated with the paradigm of text.
While taking care not to generalize from the perspective of linguistics, it should be
recognized that texts are among the most complex semiotic performances, and are exemplary for this
reason. In addition, texts (oral and written ones) are plurisemiotic constructs that implement, besides
natural languages, genres and styles, graphic and typographical systems (a punctuation mark does not
function the way a morpheme does), prosodic, gestural (some sort of kinesics is always associated
with oral performances). All these aspects are neglected by semiotics, and also by the linguistic
schools that work on morphosyntax alone.
Finally and above all, the semantic relationships established or recognized between the
various parts of a text, as an interpretive process, are of such complexity and variety that they cannot
be boiled down to logical compositionality, but more often than not, on top of everything else, they
involve interpretants lifted from other semiotic systems than natural languages.
1. In France, researchers still have to do without a
sufficiently diverse digitized corpus of reference. Admittedly, the Frantext base of the National
Institute for the French Language fortunately collects literary texts as well as technical and scientific
ones (in an insufficient but growing number). It is the only one, in the French field, to allow a
uniform access to these texts, so they can be contrasted. But the Frantext corpus has not been
systematically exploited in its generic variety (which does not account for the lexicographical studies
in progress). Novels and short-story collections, for instance, are indexed together. A new
classification is a good idea. In a thematic study of feelings, we have thus been able to verify that
feelings in the novel are different from feelings in the essay. [RETURN]