A Short Polemical and Orientational History of Linguistics*
Ferdinand de Saussure was on the right track, except on one crucial matter.
He was essentially right that signs get their meanings in contrastiveness, and that they have no intrinsic meanings already.
He was right that there are two dimensions, the syntagmatic and the paradigmatic.
He was wrong to suggest that the Language is a single synchronic system, and that words, (or sounds, at the phonemic level) at any one moment mean what they mean by contrast to all the others in the system. Saussure himself recognized this move as a tactical one, merely a heuristic. Unfortunately, it became the defining characteristic of Structuralism as it came to be remembered.
Roman Jakobson was on the right track as a Structuralist, except for one big mistake.
He was right to develop the original ideas of Structuralism, particularly the theory of the phoneme and the morpheme, and the idea of the "emic."
But, seduced by the thought of putting linguistics on a scientific footing, he thought he could give physical validity to meaning by defining contrastiveness in binary markedness terms which referenced universal phonetic facts. In effect, by moving from dynamic contrast to binary calculation, he abandoned the emic in favor of the etic, and established linguistics as a computational science.
Chomsky, by taking the erroneous idea of computational systematicity in Structuralism as an ontological ideal, has through successive versions of Transformational Grammar, Government and Binding, Optimality Theory, etc., created a monolithic linguistic establishment, dominating discourse about language. Its recent manifestation in Cognitive Linguistics, (though it has also spawned significant opposition in west-coast Functionalism), has created an orthodoxy based on a modular, innate, universal grammar; a notion of sign-meaning based in platonic realms; and an expectation that the goal of linguistic science is to write the formal rules which will generate the unique logical form of the syntax of every sentence.
I believe that we must go back to about 1910, to revive the spirit of Saussurian Structuralism. I propose two moves to accomplish this:
Contrast, or, more accurately, molecular dynamics (there are several kinds), indeed determine the meanings of signs. Words mean what they mean by contrast to other words, but, prototypically, only the other words in a single discourse. Thus I propose the concept of the molecular sememe, which is the small and dynamic set of signifiers (or signifieds) which, chosen by discourse-context, exist as potentials at any decision-point in the creation of the next move of a discourse. Thus, it is a single discourse which can be said to be a coherent synchronic system, not the Language as a whole. (Language as a whole gets what consistency it has from a process of conventionalization.)
Seeing where the decision-points are requires a concept of discourse salience. Discourse salience is a kind of markedness fundamental to discourse. Some have called it Focus Structure, and argued that it is an aspect of syntax. I believe it belongs to discourse and underlies syntax. It is a covert structure in English, though it is overt in some languages.
Therefore I offer the model of the molecular sememe as the unit of meaning, and Molecule Selection-and-Execution (MSES) as the structure of discourse. I would not be embarrassed to be called a micro-structuralist, in honor of Saussure, except that when I have acknowledged Structuralist roots people have taken me to be a Chomskian. Nothing could be further from the truth.
*For a fairer, more detailed and nuanced account, go here.