Gramsci is usually best known because of his political thoughts. I shall introduce him in a different light: as a linguist and a philosopher of language. 
To Gramsci, all people are philosophers because they take part in 'sponta- [p. 550:] neous philosophies'. In the same way, all people are linguists because they utter language. Professional linguists, therefore, are on the wrong track if they hypostatize a unity that does not function socially. For Gramsci, the problem of the lack of unity of the Italian language of his time was closely interwoven with another socio-political problem, one which originated in the fall of the centralized Roman empire, when a split arose between the written language of the intellectuals (Middle Latin) and the countless dialects spoken by the people (Gramsci (1979: 184; CW 168-171)). This separation of language (and culture) had far-reaching consequences – especially in Italy itself. Because of its powerful feudal nobility and clergy, a new political unity arose rather slowly, compared with the rest of Europe. Attempts to restore the Italian unity of Antiquity (the 'Risorgimento') were founded mainly on the revolutionary efforts of the middle classes; even in Gramsci's time, there was no real ('organic') unity. Unable to form a united front, the peasants and labourers were exploited by powerful entrepreneurs. In particular Sardinia, where Gramsci grew up, was treated as a conquered territory (Fiori (1970); Gramsci (1978)).
The lack of unity in the language spoken in Italy indicated the presence of a 'passive revolution': whereas Tuscan, because of the literary tradition of Florence, became the linguistic norm of the bourgeoisie, a planned, unified language gained ground only laboriously against the obstinate persistance of dialects (Gramsci (1979: 187n2; CW 171-172)). The political situation directly involved Gramsci in a field of tension between the 'unified' language, grafted on bourgeois cultural ideals, and the cultural isolation of a dialect, which was his own background.
At the beginning of this century, the mainstream of European linguistics was dominated by the Neogrammarians with their strict conception of language change as obeying 'sound laws without exceptions'. Language was considered as mere sound; linguistics became phonology. The Neo-Hegelian philosopher Benedetto Croce rejects this positivist approach. He sees language as a subjective esthetic act; in other words, language is certainly sound, but interpreted as expression, as a work of art (Croce (1922); cf. De Simone (1967: 5-6)). Seen this way, language does not refer to reality. It exists in free creations of the Spirit. Speakers are involved in an eternal process of creation of strictly unique, individual expressions. Because of this continuous activity in practically shaping one's reality, Croce calls his philosophy a 'philosophy of praxis': it is not the sound laws that matter, but the individual differences in coping with them (Croce (1913)).
Croce rejects all sorts of rule-governed processes: rules would be in conflict with a free unfolding of the Spirit (which fascism replaced by the State). Yet, the creation of language is regulated – not by grammar, but by esthetics. To demonstrate the lack of regularity in grammar, Croce states that sentences such as 'this round table is square', however grammatical, are esthetically [p. 551:] wrong, because they do not create anything imaginable. Gramsci argues that this fixing of rules directly contradicts Croce's own concept of free creation; his use of the sentence 'this round table is square' proves that it expresses something (e.g. lack of logical coherence). Language can not be shackled by fixed esthetics. Linguistics, therefore, does not study language as an art, but as the 'material' of art, as a social expression (CW 179-180, 177; cf. Rosiello (1970: 350); De Simone (1967: 7-10)).
In this debate, Gramsci demonstrates a strong affinity to his teacher Matteo Bartoli, who "transformed linguistics, conceived narrowly as a natural science, into an historical science" (CW 174; quoted in Salamini (1981: 32)). While Bartoli agrees with Croce in the latter's rejection of mechanistic sound laws, he is himself far more oriented towards social regularities. Dialect maps (such as those composed at that time by the Swiss linguist Gilliéron) showed enormous diversity within a 'unified language', without thereby suggesting an arbitrary creativity in the spread of dialectic phenomena. This explains not only why Bartoli, while outlining linguistic norms, takes into account both differences and agreements, but also why such norms only offer broad and unsystematic criteria for fixing the most influential lines in historical linguistic transmission (cf. Hall (1946: 274); De Simone (1967: 15n)). Thus, the Neogrammarian concept of a 'primitive language' as a unity is undermined. Instead of mechanistic causal laws, nothing is left but statistical 'norms' without any validity beyond the time and area in which they appear. While Bartoli himself was not altogether successful in stressing this point of view within his linguistic theory (cf. Devoto (1947)), he apparently expected Gramsci to follow it up and carry it through.
In a letter from prison (1979: 79-81), Gramsci writes that he greatly regrets having disappointed Bartoli in the latter's conviction that he (Gramsci) was the Archangel sent to destroy the Neogrammarians once and for all. In the same letter, he states with some self-derision that he is plagued by the notion that he has to accomplish something für ewig. As an important future topic of research, he mentions "a study of comparative linguistics, nothing less! What could be more 'disinterested' and für ewig than that?" (p. 79). He ironically plays with the value-free image that linguistics had adopted under the influence of the Neogrammarians. But Gramsci's efforts go much further. All the time, practical considerations come into play on the point of the organization of national unity, which was exactly the point on which the bourgeoisie (and bourgeois science) had failed. In its extreme form, this failure manifested itself in the fascist attempt to save unity by violence (for example, by imprisoning those who 'speak another language'). For Gramsci, the study of language was a study of social and political relations, as well as of his own historical situation. His views are still able to cast new light on a blind spot of linguistics and social theory, more than fifty years after his death shortly after his release from Mussolini's prisons. [p. 552:]