… It is … not a question of discovering the meaning of symbolic representation but, on the contrary, of inventing a relevance and a place in memory for them …
This is from Rethinking Symbolism by Dan Sperber (1974):
… Symbols are not signs. They are not paired with their interpretations in a code structure. Their interpretations are not meanings.
… Symbolism is, in large part, individual, which is doubly incomprehensible from the semiological point of view. Firstly, a system of communication works only to the extent that the underlying code is essentially the same for all; secondly, a code exhaustively defines all its messages. Symbolism, which is a non-semiological cognitive system, is not subject to these restrictions.
A corollary of this cognitive nature is that there is no multi-symbolism analogous to multi-lingualism. An individual who learns a second language internalizes a second grammar, and if some interference takes place, it is on a remarkably small scale. Conversely, symbolic data, no matter what their origin, integrate themselves into a single system within a given individual.
… One of the first things that the anthropologist learns in the field is strictly to observe the local forms of politeness. For at least a time, he has the impression that he is acting, rather than expressing himself normally; the symbolic values of these forms of politeness escape him, or he only apprehends them upon reflection. The longer he is there, the more easily he can pass from one ‘code’ of politeness to another, like an actor who changes roles. But if he should become sensitive to the grace of a gesture of offering; to the nuances by which the warmth or coolness of a welcome are expressed; to the perfidiousness of a disguised insult; if at the death of a friend he comes to feel comforted rather than troubled, by reproducing the traditional gestures of mourning; in short, if he should internalize these forms instead of imitating them: then, on returning home, he will catch himself following rules that are not in force, and resenting it when others break them.
[ ... ]
… The symbolic value of ‘fox’ owes nothing to the sense of the word, and everything to what we know or believe about foxes: to their skill as predators, their look, their coat, etc. What matters, symbolically speaking, is neither how foxes are semantically defined nor what foxes actually are, but what is known of them, what is said of them, what is believed about them.
… the conceptual mechanism never works in vain; when a conceptual representation fails to establish the relevance of its object, it becomes itself the object of a second representation. This second representation is not constructed by the conceptual mechanism which turned out to be powerless, but by the symbolic mechanism that then takes over. The symbolic mechanism tries to establish by its own means the relevance of the defective conceptual representation.
To return to Lévi-Strauss’ image, the symbolic mechanism is the bricoleur of the mind. It starts from the principle that waste-products of the conceptual industry deserve to be saved because something can always be made of them. But the symbolic mechanism does not try to decode the information it processes. It is precisely because this information has partly escaped the conceptual code, the most powerful of the codes available to humans, that it is, in the final analysis, submitted to it. It is therefore not a question of discovering the meaning of symbolic representation but, on the contrary, of inventing a relevance and a place in memory for them despite the failure in this respect of the conceptual categories of meaning. A representation is symbolic precisely to the extent that it is not entirely explicable, that is to say, expressible by semantic means. Semiological views are therefore not merely inadequate; they hide, from the outset, the defining features of symbolism.