This chapter attempts to delineate the debate concerning the main tenets of the American descriptivist or 'mechanist' approach to language from the 1940s to the 1950s, and well into the 1960s, and to evaluate central proposals concerning the nature of language and the methods of analysis maintained by the majority of its adherents. The main focus of the account is the issue of 'meaning' in linguistics during the years preceding the publication of Chomsky's Aspects in 1965, which moved the attention away from the study of linguistic meaning as previously attempted. While it is not argued that Bloomfield himself neglected the content side of language in his work, it is nevertheless held that his teachings and writings have laid much of the foundation for the theoretical bias so characteristic of the arguments of many eminent linguists of the post-Bloomfieldian era. Their general data-orientation, with a strong emphasis on the description of speech in conjunction with the rejection of any 'mentalistic' speculation about semantic implications in linguistic analysis, made them argue against the use of 'meaning' however defined. As semantic matters could not be readily expressed in certain 'rigorous' terms, phonology, which soon came to be called phonemics, became the main focus of attention, in the belief that one could bypass semantic considerations by simply determining whether a particular form was the 'same' or 'different'-and could be done by asking a native informant. 1
It is perhaps possible to agree with Karl V. Teeter's (b.1929) assessment in 1964 that most American linguists of the period were subject to what he called the 'post-Bloomfieldian fallacy' of taking rigorous data-manipulation for theory. But by that time transformational-generative linguistics had begun to sweep the American linguistic community. 2 The Chomskyan emphasis on syntax as against the levels of morphology and phonology as well as its general
1 Sapir's 1933 paper on "La réalité psychologique des phonèmes" (English version first published in Sapir [1949:46-60]) is probably the most classic example of this procedure.
2 Looking back some 35 years later, Teeter (1998:253) is able to give a more balanced analysis of what the situation in American linguistics was at that time.