Since Gramsci's prison notes are only fragmentary, there is no direct evidence that he developed a general theory (cf. Paggi (1979)). Nevertheless, his fragmentary style cannot be taken to cancel every claim to 'organic' systematicity. Remarks on language are dispersed throughout Gramsci's writings; arranged in a certain order, together they form a coherent linguistic theory. Such an ordering is acceptable in the light of Gramsci's proposed central theme: the demand to develop one's language in an 'organic' direction. Organicity, i.e. the organizing power of a language, is measured by Gramsci by the degree of coherence of this language. Therefore, if his remarks on language aim for organicity and unity of language, they must also be directed towards coherence, as the limit towards which his theorizing tends.
Linguists try to discover coherences both between language and reality (in short: reference), and within the language system itself (linguistic 'rules'). In this way, they attempt to establish an a priori unity of language. Often, such a unity is taken as a precondition for the possibility of communication; for this reason, many linguists have tried to base the unity of language communities on some determining 'scientific' datum. Rationalism considers the rational structure of the mind as given (Descartes, Chomsky, most of modern psycholinguistics, speech act theory, and empirical sociolinguistics). Empiricism considers the social or cultural structure of individual experience as given (Locke, Croce, Sapir, Basil Bernstein). The former regard misunderstanding and dialectic variation as exceptions derived from universal rules (or from a mysterious creative freedom within the boundaries of those rules). The latter explain unity (the very possibility of communication) from diversity (or they deny it). Both positions give rise to doubts, because they look for conditions of understanding as if these were given before their practical realization.
This is not the kind of coherence that Gramsci points to (see Gramsci (1971), from now on referred to as PN; esp. pp. 323-377): for him, coherence has to be developed historically. It is expressed by solidarity among the speakers of a language, and measured by the extent to which their practices and world views are interrelated. In linguistics, concepts like 'language community', 'convention', and 'agreement' often imply a concept of history in which the unity between speakers is premised. But for Gramsci, history is basically a process in which antagonistic groups strive for political and cultural domination, or, in other words, for hegemony (cf. the influence that is attributed to socially prestigious language forms by historical linguistics, Lo Piparo (1979: 103 ff.)). In the contest of history, agreement is never given. There is no pre-established consensus, but an ongoing pursuit of recognition, as a necessary precondition for hegemony. In this search for points of agreement, linguistic problems can easily arise. But the 'question of language' that thus emerges is alien to traditional linguistics, in that it subordinates [p. 553:] referentiality and systematicity to the practical realization of a new language community.
In the historical struggle for hegemony, ideological and cultural identifications are always taking place. All the time, judgments are adopted unconsciously from leading groups. The self-evident truths of language, common sense, and popular religion are promoted continuously (PN 323; CW 183). They inflict all kinds of contradictions on the thinking of non-hegemonic groups. This is a reason to try and go beyond everyday language use, which is instrumental in an uncritical imitation of the ruling groups.  It is noteworthy that Gramsci defines the uncritical knowledge implied in the everyday practices of social groups as explicit or verbal knowledge. Although this verbal knowledge holds together a specific group, it also influences its moral conduct and the direction of its will, "often powerfully enough to produce a situation in which the contradictory state of consciousness does not permit of any action, any decision or any choice, and produces a condition of moral and political passivity" (PN 333). Such a passive attitude can only be directed towards a new unity by a critical relationship towards one's language. Putting one's verbal knowledge on a par with one's daily practical experience, is possible only through a common struggle for independence from that which is said within hegemonic groups. In this way, Gramsci politicizes the concepts of language and knowledge (cf. Buci-Glucksmann (1980: 327-397)) in a philosophy of praxis that is not a philosophy of the 'pure' act (as worshipped by fascism), but rather of the real, 'impure' act, in the most profane and worldly sense of the word (PN 372).
In the next three sections, I shall compare the way Gramsci politically ordered the linguistic confusion of his day by a 'principle of coherence', with the philosophies of language of Saussure, Wittgenstein, and Vološinov respectively.