Morris Halle and the bases of generative phonology


 Morris Halle, born in Latvia in 1923, had a rather different background from Chomsky¡¦s. After emigrating to the United States in 1940, he studied engineering in New York until he was drafted in 1943. After the war, he attended the University of Chicago, where he received a degree in linguistics; in 1948, he went to Columbia to study with Jakobson, and came with Jakobson to Harvard the next year. He was a part of the circle of talented young Slavicists around Jakobson during these years, and collaborated with Jakobson on a number influential works. His Harvard Ph. D. thesis in 1955 was entitled ¡§The Russian Consonants: A Phonemic and Acoustical Study¡¨.

On the basis in part of his engineering background, Halle held an appointment in the Research Laboratory of Electronics at MIT while he was a student at Harvard. He was subsequently hired by the MIT department of modern languages to teach German and Russian . A part of his research on acoustics at MIT while he was still a student involved collaboration with Jakobson and Gunnar Fant, and resulted in Preliminaries to Speech Analysis, a formation of the Jakobson distinctive feature framework in simultaneous acoustic and articulatory terms. His initial reputation was earned as an acoustic phonetician, and only secondarily(through his association with Jakobson) as a phonologist.

 Although Jakobson was particularly interested during the early 1950s in providing an explicit acoustic basis for the distinctive features, he had also done some particular elegant work on the morphophoonemics of Slavic languages ¡Vwork which was virtually ignored by American linguists other than Slavicists at the time. Halle was greatly impressed and attracted by this work, but also struck by the difficulty of relating anything like an adequate morphophonemic representation of utterances directly to the acoustic facts. The morphophonemic analysis had such coherence and explanatory value that it seemed inconceivable to deny its importance in the structure of language, as most American phonemicists attempted to do; but on the other hand, if this analysis did indeed have a real role in linguistic structure, its relation to the physical signal must be much less direct than was posited for a structuralist phonemic representation.

In 1959,Halle published The sound Pattern of Russian, in which the several aspects of this problem are addressed. Indeed, the many distinct areas which this book deals with in combination make it rather strange reading today: we are unused to finding a combination of general linguistic theory and detailed analysis of the phonology of a specific language together in a single volume with an introduction to the surface segments of the language. All of these are central to Halle¡¦s project. He had decided that morphophonemic representation was extremely important to the structure of language: but he believed that a representation which was perceptually recoverable from acoustic data was important as well. His goal was to show how these two levels of description were related to each other, and what characteristics each level had. Since linguists were not generally acquainted with the results of the rapidly developing theory of the acoustics of speech, it was necessary to give an introductory sketch of this area in order to make his results comprehensible.

Though it is not the most readable part of this book, the section for which it is primarily remembered is the introduction to the first chapter, ¡§A Theory of Phonology¡¨. Here Halle lays out a number of assumptions claimed to be necessary to an adequate phonological theory, but he argues a level of representation meeting the specific conditions of structuralists phonemics cannot naturally be incorporated into such a theory.

In order to account for ¡§ambiguities due to homophony¡¨ a morphophonemic representation is indispensable and a universal phonetic representation is similarly necessary to express the facts of speech. The latter is, for Halle, one in which all of the features in a universal inventory- for example, the Jaobsonian system are specified in a way that has a direct acoustic and articulary interpretation. In contrast to both of these, a structuralist phonemic representation has the property of specifying all and only the distinctive sound properties of an utterance, while still being recoverable from phonetic data alone. Halle¡¦s argument was that such a representation can not in general be incorporated in an analysis without resulting in ¡§an unwarranted complication which has no place in a scientific description of language¡¨.                          

~ Adopted from Stephen R. Anderson.1985.Phonology in the Twentieth Century. UP: Chicago.