The Work of Zellig Harris
The work of Zellig Harris in language, grammar, and information, and in the methodology of linguistics, is remarkable for its consistency and integrity over a span of almost 60 years, culminating in an elegant and comprehensive theory of language and information.
In his own retrospective survey (published as Harris 2002b, and in translation as Harris 1990) he said "it was possible to describe the entire program from the outset". Nevertheless, working out its consequences and demonstrating its results required many years of painstaking work, with tests in many languages.
A survey of his major publications suggests five stages of development:
|Harris worked a subject deeply before bringing it to publication. The appearance of discrete "stages" reflects how long it took to work the data of language into confirmations worthy of presentation.|
|For example, in the course of his investigations as a Semitist, he laid the foundations for the distributional methodology (Stage 2), summarized in (1946/1951), which seems suddenly to have sprung into view in a series of papers beginning with the (1940) review of Gray’s Foundations of Language. Similarly, transformations and the correlation of linguistic form to linguistic information were evident from the beginning (Harris 1990/2002). When Leigh Lisker became his student in 1939 he was already teaching transformational analysis and discourse analysis (Stage 3), including the work on Hidatsa (1939c) and Koto (1945f) texts, more than a decade before it was first published (1952).|
|Harrisian distributionalism is usually represented as a prime exemplar of the alleged striving of 'Bloomfieldians' to eliminate meaning from linguistics.(*) In fact, it explicates Leonard Bloomfield’s affirmation that the form of an utterance and the meaning that it conveys are two aspects of the same thing. It has a deep connection with the search for configuration and pattern in language data exemplified by Edward Sapir, who regarded Harris as his intellectual heir. Harris was more open than most of his contemporaries to developments in the Prague school and elsewhere in Europe, such as the work of Roman Jakobson (e.g. Harris 1951a: 68n18, 79n1, 125n4, 146n48, etc.), who he assisted in getting established in the U.S.|
|The work on linguistic information that is made most explicit in Stage 5 is thus the formative theme underlying all of Harris’s work from the beginning. The correlation of form and information is the motivation for distributionalism, discourse analysis, transformational analysis, and the identification of elementary transformations (1964, 1969). The analysis of elementary sentence-differences led directly to Operator Grammar (Stage 4), the "least grammar" (1988:57) with which we may characterize the informational capacities of language. Carrying this forward into sublanguage analysis discloses the information "in" or "carried by" a given utterance. The form of an utterance, taken within the patterning of the language, is itself the "semantic representation" of the information "in" the utterance, as distinct from additional kinds of meaning brought to it by the recipient.|
|André Lentin (1990) clarified the mathematical basis of Harris's work—a question of "how could a little mathematics transmute itself into linguistics?"—as Harris acknowledged in a 1991 letter to him. The English translation of Lentin's paper (Lentin 2002) serves as the introduction to vol. 2 of The Legacy of Zellig Harris.|
Notes* Bloomfield wrote, in a letter to Kenneth Pike on January 19, 1945 (Barsky 2011:134), “… refuting the notion that I, or rather a whole group of language students of which I am one, pay no attention to meaning or neglect it…. It is … something which, if allowed to develop, will injure the progress of our science by setting up a fictitious contrast between students who consider meaning and students who neglect it. The latter class, so far as I know, does not exist”.