Before the 19th cent., language was studied mainly as a field of philosophy. Among the philosophers interested in language was Wilhelm von Humboldt, who considered language an activity that arises spontaneously from the human spirit; thus, he felt, languages are different just as the characteristics of individuals are different. In 1786 the English scholar Sir William Jones suggested the possible affinity of Sanskrit and Persian with Greek and Latin, for the first time bringing to light genetic relations between languages. With Jones's revelation the school of comparative historical linguistics began. Through the comparison of language structures, such 19th-century European linguists as Jakob Grimm, Rasmus Rask, Karl Brugmann, and Antoine Meillet, as well as the American William Dwight Whitney, did much to establish the existence of the Indo-European family of languages.

In the 20th cent. the structural or descriptive linguistics school emerged. It dealt with languages at particular points in time (synchronic) rather than throughout their historical development (diachronic). The father of modern structural linguistics was Ferdinand de Saussure, who believed in language as a systematic structure serving as a link between thought and sound; he thought of language sounds as a series of linguistic signs that are purely arbitrary, as can be seen in the linguistic signs or words for horse: German Pferd, Turkish at, French cheval, and Russian loshad'. In America, a structural approach was continued through the efforts of Franz Boas and Edward Sapir, who worked primarily with Native American languages, and Leonard Bloomfield, whose methodology required that nonlinguistic criteria must not enter a structural description. Rigorous procedures for determining language structure were developed by Kenneth Pike, Bernard Bloch, Charles Hockett, and others.

See also structuralism.

In the 1950s the school of linguistic thought known as transformational-generative grammar received wide acclaim through the works of Noam Chomsky. Chomsky postulated a syntactic base of language (called deep structure), which consists of a series of phrase-structure rewrite rules, i.e., a series of (possibly universal) rules that generates the underlying phrase-structure of a sentence, and a series of rules (called transformations) that act upon the phrase-structure to form more complex sentences. The end result of a transformational-generative grammar is a surface structure that, after the addition of words and pronunciations, is identical to an actual sentence of a language. All languages have the same deep structure, but they differ from each other in surface structure because of the application of different rules for transformations, pronunciation, and word insertion. Another important distinction made in transformational-generative grammar is the difference between language competence (the subconscious control of a linguistic system) and language performance (the speaker's actual use of language). Although the first work done in transformational-generative grammar was syntactic, later studies have applied the theory to the phonological and semantic components of language.

In contrast to theoretical schools of linguistics, workers in applied linguistics in the latter part of the 20th cent. have produced much work in the areas of foreign-language teaching and of bilingual education in the public schools (in the United States this has primarily involved Spanish and, in the Southwest, some Native American languages in addition to English). In addition, such subfields as pragmatics, sociolinguistics, and psycholinguistics have gained importance.