American structuralist phonology was concerned primarily with formalizing the notion of the PHONEME, and with elaborating a framework for the phoneme analysis of languages. By the early 20th century, phonetic science had revealed that languages often had more sound in their phonetic inventories than had originally been suspected. For example, English p[l]ease has the voiceless liquid also found in Welsh; the final sound in ma[t]resembles the glottalized stop of many Amerindian languages. While phonetically accurate, these transcriptions are linguistically misleading: [l] and [t] are of quite different status in English than in these other languages. Understanding the nature of this difference was taken to define the field of phonology.
In the 1920s, Edward Sapir and Leonard Bloomfield initiated two different approaches to the problem. For Sapir, a language’s phonetic inventory is but a distorted manifestation of its inner system of phonemes. The phonemic structure is a conceptual system—an inventory of ‘ideal sounds’, in terms of which the phonetic segments of speech are perceived and articulated. Sapir 1925 shows how the same phonetic sound can have a quite different status in languages with different phonemic systems. Sapir believed that a language’s inner phonemic system can often be intuited by the native speaker.
Bloomfield was suspicious of such ‘mentalism’. He felt it more prudent to approach the phoneme exclusively through objectively observable data. For Bloomfield, the phoneme was not an ‘ideal sound’ or ‘mental image’, but rather a bundle of distinctive features which are present in the overt, physical manifestation of speech. The task of phonemic analysis is to isolate these distinctive features—principally through the discovery of minimal pairs—and to state the distribution of redundant, non-distinctive features.
Because of the positivistic climate of the 1930s, Bloomfield’s approach attracted much more attention than did that of Sapir. At the theoretical level, American Structuralist Phonology in the next two decades occupied itself principally with elaborating Bloomfield’s surface-oriented, inductive view of the phoneme; explicit criteria were developed to justify given phonemic groupings. Bloch 1941, posed one of themost serious challenges to the Bloomfieldian view. For example, in most American English dialects, the /t/ phoneme of betting, butter is realized a as a flap [D]; in some of these dialects, the /r/ phoneme is also realized as [D] after the interdentals of three, throw. On this analysis, there is no invariant bundle of features that distinguishes among all occurrences of /t/ and /r/. To maintain the invariance condition, [D] would have to be set up as a separate phoneme—despite its limited distribution and redundant status.
Another much-discussed problem is the minimal pair writer [wayDr] vs. [ra:yDr]. The surface contrast is in the length of the vowel. But most analysis felt that the correct phonemicization registers the contrast in the consonant as [t] vs. [d]. Chomsky 1964 seized on this example from Harris in a devastating critique of Bloomfieldian phonology. In the 1950s, Chomsky and his collaborator Morris Halle concluded that the best solution to such problems was simply to abandon the strong behaviorist stance of the Bloomfieldians, and to explore instead the mentalist approach to phonology advocated by Sapir. In the resultant generative model, the only constraint on underlying (systematic phonemic) representations was their effectiveness in maximizing the overall simplicity of the grammar. The focus of phonological research shifted to rules—their discovery, formulation, and ordering. However, Generative Phonology also carried over several features of American Structuralism.
STEIN JEFFEREY P. & MCGARY JANE etc. EDS. 1992. INTERNATIONAL ENCYCLOPEDIA OF LINGUISTICS.
E OXFORD UP.