Inspired by people such as Harris, and experiencing an increasing intellectual fascination with the kind of work that he had undertaken for his B.A. thesis, Chomsky decided to enter graduate school at Penn. He began in the fall semester of 1949, and within a short period completed a master's thesis (his degree was granted in 1951), which was a revision of his B.A. thesis. The work underwent further editing in 1951, and was finally published in 1979 as Morphophonemics of Modern Hebrew.
His friendship with Harris was growing and it took on what could now be described as mythic proportions. Chomsky seemed to have been elected to follow up on, and expand, Harris's work, and Harris became for Chomsky a figure with whom, and ultimately against whom, he could measure his own achievement. Much has been written about this relationship, and much conjecture has been published as to the influence that each man had upon the other. But what was the real nature of the discussions that Chomsky had during this period with Harris and others? What effect did these exchanges have upon Chomsky's intellectual development?
Chomsky notes that Harris thought of linguistics as a set of procedures for organizing texts, and was strongly opposed to the idea that there might be anything real to discover. He did think that the methods of linguistic analysis could be used for analysis of ideology, and most of my actual graduate courses were devoted to that; you can see some of the fruits in his articles on discourse analysis in Language in the early '50s, though he kind of downplayed the political side that was everyone's main interest. (13 Dec. 1994)His attempts to make Harris's methods work constituted Chomsky's early linguistic research. Out of these endeavors came his first published article, which appeared in The Journal of Symbolic Logic. His undergraduate thesis also applied some of Harris's ideas, but he had by then totally abandoned all of his methods and adopted a "completely non-procedural, holistic (in that the evaluation measure proposed was a measure applied to the whole system), and realist" approach (31 Mar. 1995):
Phrase structure rules can generate representations of syntactic structure quite successfully . . . for quite a range of expressions, and were introduced for this purpose in the earliest work on generative grammar. It was at once apparent, however, that phrase structure rules . . . are insufficient in themselves to account properly for the variety of sentence structures. The earliest approach to this problem, which has a number of subsequent and current variants, was to enrich the system of rules by introducing complex categories with features that can "percolate down" to the categories contained within them, expressing global dependencies not captured in a simple system of phrase structure rules. . . . I adopted this approach in an undergraduate thesis of 1949, modifying ideas of Zellig Harris from a somewhat different framework. (Knowledge 64)So even at this early stage, Chomsky was producing highly original work, which diverged fundamentally from Harris's. In his B.A. thesis he was doing things that were, in his own words, "radically at odds with everything in structural linguistics . . . which is why [it, and Logical Structure of Linguistic Theory] were published only 30 years later." The thesis was "as different from structural linguistics as anything could be," which was why "Harris never looked at it and no one in the field reacted to it." In fact, Morphophonemics of Modern Hebrew remains "the only text in existence, to my knowledge, that seeks to apply an evaluation measure in anything remotely like that detail" (31 Mar. 1995).