It will certainly be objected that cognitive semantics, with
authors like Langacker,1 broke away with the dualistic
conception of the sign. Nothing is less sure. Indeed, the signified is continuously located in another
order of reality as the signifier. It is simply that cognitive semantics divorced the theory of direct
denotation, and watered it down without giving up the principle of compositionality. But when
mentalism strikes back and it defines the space where states of affairs happen, it links back signifieds,
become concepts once again, to cognitive fields with no defining principle, in a mental space which
is a new version of transcendental space (Rastier 1993).
All things considered, none of the great competing paradigms of cognitive research on
signification acknowledges that the semiotic is a level which would allow to consider it as an area of
objectivity that can be studied in itself. The semiotic unit as such is surreptitiously replaced with the
percept on the one hand, with the concept on the other. Certainly, the percept/concept distance tends
to be reduced, insofar as cognitive schemes have retained the outline of forms in space from their
Kantian ancestors __ while losing their mediating function, for want of pure
concepts of understanding.
The objective common to all of the cognitive paradigms, which might lead one to say that
they are pursuing the programme fulfilled by general grammars before the advent of comparative
linguistics, consists in going upward from language to thought, and from the expression to the
concept. Dummett summarized the attempt in the same way: "By examining the mechanisms of
language, it is of course thought we are trying to analyze. Language in itself would hardly have any
importance at all if it did not make it possible for us to access the workings of thought"
(iv) The function of symbols in the symbolic paradigm.
Their eponymous function allows to guess that symbols are entrusted with missions of
a) They are the means of any scientific description. According to Minsky and Papert:
What do we mean by "description"? We do not mean to suggest that our descriptions must
be made of strings of ordinary language words (although they might be). The simplest kind of description is a structure in which some features of a situation are
represented by single (primitive) symbols [...] (1974, p.223).
This is the cognitivist road to perpetuate the thousand-year-old search for primitives. The
underlying idea is that symbols are simpler and more primitive than words, as we will see further in
relation to the language of the thought.2
b) As we know, they are the basis for Artificial Intelligence. We have to credit Newell and
Simon for formulating the Strong Physical Symbol System Hypothesis: a physical system
that can handle symbols has the necessary and sufficient characteristics to produce or at least simulate
This functionalist thesis rests in particular on a quite peculiar mode of relationship between
types and occurrences. Symbols in their form, and as types, are computed, whereas in their material
format, and as tokens, they are implemented. It is then enough to state that matter is related to spirit
in the same way implementation relates to computation, and the problem of their articulation is
apparently solved. However, the relationship between types and tokens is a crucial problem in
semiotics and semantics: it governs for example the questions of polysemy and typicality, and more
generally categorization. Characterizing this relationship, qualifying how tokens differ from types,
is a major objective of interpretation. When one postulates that they do not differ, one implies that
interpretation works on its own (Rastier 1995).
1. Langacker is the only author in that current to mention
Saussure: he borrows from the latter's iconic figuration of the signified. [RETURN]
2. Symbols then act as what the gentlemen of Port-Royal
called words of ideas: they represent such simple ideas that they cannot be defined, and
are used to define all the others. The elementary does not have to be interpreted as soon as it is
decreed that elementariness provides meaning: it is a founding principle. Greimas's
indefinables, and Fodor's primitives are actually the same notion. [RETURN]