It begins,
What we have learned from Saussure is that, taken singly, signs do not signify anything, and that each one of them does not so much express a meaning as mark a divergence of meaning between itself and other signs. Since the same can be said of all other signs, we may conclude that language is made of differences without terms; or more exactly, that the terms of language are engendered only by the differences which appear among them. This is a difficult idea, because common sense tells us that if term A and term B do not have any meaning at all, it is hard to see how there could be a difference of meaning between them; and that if communication really did go from the whole of the speaker's language to the whole of the hearer's language, one would have to know the language in order to learn it. But the objection is of the same kind as Zeno's paradoxes; and as they are overcome by the act of movement, it is overcome by the use of speech. And this sort of circle, according to which language, in the presence of those who are learning it, precedes itself, teaches itself, and suggests its own deciphering, is perhaps the marvel which defines language.Language is learned, and in this sense one is certainly obliged to go from the parts to the whole. The prior whole of which Saussure is talking about cannot be the explicit and articulated whole of complete language as it is recorded in grammars and dictionaries. Nor does he have in mind a logical totality like that of a philosophical system, all of whose elements can (in principle) be deduced from a single idea. Since what he is doing is rejecting anything other than a "diacritical" meaning of signs, he cannot base a language upon a system of positive ideas. The unity he is talking about is a unity of coexistence, like that of the sections of an arch which shoulder one another.
In a unified whole of this kind, the learned parts of a language have an immediate value as a whole, and progress is made less by addition and juxtaposition than by the internal articulation of a function which is in its own way already complete. It has long been known that for a child the word first functions as a sentence, and perhaps even certain phonemes as words. ... The important point is that the phonemes are from the beginning variations of a unique speech apparatus, and that with them the child seems to have "caught" the principle of a mutual differentiation of signs and at the same time to have acquired the meaning of the sign. ... The whole of the spoken language surrounding the child snaps him up like a whirlwind, tempts him by its internal articulations, and brings him almost up to the moment when all this noise begins to mean something. The untiring way in which the train of words crosses and recrosses itself, and the emergence one unimpeachable day of a certain phonemic scale according to which discourse is visibly composed, finally sways the child over to those who speak. Only a language taken as an integral whole enables one to understand how language draws the child to itself and how he comes to enter that domain whose doors, one might think, open only from within. It is because the sign is diacritical from the outset, because it is composed and organized in terms of itself, that it has an interior and ends up laying claim to a meaning.
...
That is only one fortieth of the essay, and I am only partly through reading it. But the immediate insight is that words - signs - are not the active elements in language; meaning is not explicit symbolism but rather implicit in their arrangement. Merleau-Ponty is an exciting thinker.