By 1960 structural linguists could look back on thirty or more years of progress. The theory of the phoneme was now widely accepted; and, although there were disagreements in detail between most Americans and most Europeans, few disputed that it was one central unit of language. For many linguists the morpheme or a unit like it was another, and the identification of these units in ‘-eme’, as the elements of articulation on two different levels, was increasingly a basis for the practical description of languages. Such findings had begun to work their way from technical publications into textbooks. I have referred to Hockett's general introduction to linguistics (Hockett, 1958). But it had been preceded by at least one specialist textbook in descriptive linguistics, by H. A. Gleason, whose table of contents gives a clear view of how students, in at least part of the English-speaking world, were taught (Gleason, 1961 [1955]). Martinet's introduction, to which I have also referred, was to have a similarly wide success in continental Europe (Martinet, 1970 [1960]).

These achievements were ostensibly founded in the thought of Saussure, in lectures that had last been given in 1911, and of Bloomfield, formulated in its mature phase in the 1920s. A linguist was therefore describing either an underlying system whose reality was supra-individual (2.1), or a set of potential utterances (2.2). But both accounts had naturally reflected the preoccupations of their own day and, with hindsight, though the structure of a language system had been worked out in some detail, it was time to ask again exactly what reality such systems had.