-syntax proposals, the vaguely adumbrated post-Deep Structure semantic proposals, the peculiar notational variants position—all looked feeble. And that was just the start of his troubles.Unsurprisingly, I have no explanations to offer here.
—Haj Ross (1973c:243n40)
The transcendent truth of my position has been buttressed time and again, most recently by the splendid work being accomplished in progressive (as opposed to regressive, although of course no directionality is implied) semantics. I refer here not so much to the writings of McQuarrelly, whose thought is not always sound dogmatically, but rather to the output of Coughlake, that prolific exponent of generative power (see, inter alia, Coughlake to appear a, to appear b, to appear c, to appear N). Coughlake’s irrefutable, nay absolutely crushing, indications of the necessity for wholly novel forms of grammatical apparatus—approximately one revolution in theory each week, . . .—quite boggle the mind. Nothing could prove the correctness of the ESP [Erector Set (British Meccano) Proposal] more convincingly than these repeated demonstrations that the required pulleys must be larger and stronger than we were inclined to believe.
—Ebbing Craft aka Arnold Zwicky, in a multipronged parody; 1992 [1971]:148)
Judith Levi set to work on The Syntax and Semantics of Complex Nominals in 1974, an extensive reworking of her doctoral dissertation under McCawley that does for Generative Semantics what Lees’s The Grammar of English Nominalizations, a reworking of his doctoral dissertation under Chomsky, does for early Transformational Grammar. Levi’s book proves—within all the variations, uncertainties, and controversies that characterize any scientific attempt to model reality, especially a reality as mucky as language—that Generative Semantics works.1 But when she finished, and the book was in press (published 1978), Levi “looked around and found that nobody was interested. There was no community.”
On the other side of the battlefield, Howard Lasnik, an emerging scholar who was shortly to become the most important post-bellum Chomskyan, noticed that the burning issues of the last five years had disappeared overnight from the MIT hallways. “People just stopped talking about them,” he recalls. “I have this distinct impression. It was like someone put his head out the window and it wasn’t raining anymore, and that was that.”
The Wars were over.
Chomsky, it very shortly became clear, had won.
The victory was crushing, and Chomsky deserves a good deal of praise or blame, depending on your perspective, for the total eclipse of Generative Semantics. But far from all. Some of his counterargument bullets were very well placed, but Generative Semanticists also shot themselves in the foot, the knee, the hand, any and all extruding limbs or appendages, frequently and publicly, just as Chomsky’s model began making impressive headway.
That headway was the absolute deciding factor. It was Chomsky’s positive proposals—providing an alternative to a model undergoing a crisis of confidence—far more than his negative attacks on Generative Semantics, that pulled his Interpretive bacon from the fire, however badly singed.
Let’s face it. The publication of Reflections on Language [Chomsky 1975b] leaves little doubt that transformational-generative grammar has become an intellectual fraud.
—Bennison Gray (1978:70)
By the mid-1970s, Chomsky looked to be, in Zwicky’s biting parodic pseudonym, Ebbing Craft. He had been outflanked by his progeny and had retrenched to a Bloomfieldian anti-meaning position rather than embracing their advances. His counteroffensives and attacks—the misunderstood Lexicalist and -syntax proposals, the vaguely adumbrated post-Deep Structure semantic proposals, the peculiar notational variants position—all looked feeble. And that was just the start of his troubles.
His reputation in psychology had taken a beating. In the early 1960s, he was hailed as a founding father of the cognitive revolution; by the early 1970s, even as his previous accomplishments were making their way into textbooks and popularizations, working psychologists were rapidly losing faith in his program. For one thing, it wasn’t panning out experimentally. Miller and Chomsky had suggested in the early 1960s that the
psychological plausibility of a transformational model of the language user would be strengthened, of course, if it could be shown that our performance on tasks requiring an appreciation of the structure of transformed sentences is some function of the nature, number and complexity of the grammatical transformations involved. (Miller and Chomsky 1963:481)
Psychologists had extrapolated this into a very pretty process theory of Transformational Grammar with clear empirical consequences: “that the ‘transformations’ underlying variations in linguistic structure correspond to ‘mental operations’ that have to be carried out in a certain order and in a definite amount of time in order to achieve certain results” (Carroll 1964:122). The more transformations involved in a sentence, that is, the longer it should take for someone to understand it. A passive sentence should take longer than its active counterpart. A negative should take longer than a positive. A negative-passive should take proportionally longer than a positive-active. You get the picture.
At first, this model—known as the Derivational Theory of Complexity—seemed spectacularly successful, giving a psychological boost to Transformational Grammar, a grammatical boost to cognitive psychology, and an empirical boost to the hybrid fledgling, psycholinguistics. It did take people longer to understand sentences with more transformations in their derivation. But all too soon, when sentence length and meaning were factored in (passives are longer than actives, for instance, and have subtle differences in meaning; negatives are slightly longer than positives, and very different in meaning), transformations receded in importance; at best, they now seemed untestable.
Worse, transformations which resulted in no differences of length or meaning (relating sentences like Norman called up Eddie and Norman called Eddie up) also had no appreciable impact on comprehension time. Whatever the early indications from Chomsky, Miller, and other early generativists—all of that literature is rife with the language of transformational processes—it was clear that transformations were not real-time psychological operations that took one neural signal and yielded a different neural signal on the way to physical articulation or inscription. A similar story unfolded for experimental attempts to confirm the psychological reality of Deep Structure: initial success, followed by reinterpretations of that success considering other factors, and then outright failure, by which time Chomsky had rejigged Deep Structure anyway. The competence/performance distinction looked like it was developed to inoculate his formal model from testing. The “processes,” the derivational “levels,” the “directionality” and “cyclicity” of the various relations among representations—were all revealed as metaphorical in a way that offered no empirical handles for an experimentalist to grab hold.
It didn’t help that increasingly abstruse, counterintuitive notions began showing up in Chomsky’s work—filters, constraints, traces, -syntax, many of them working in ways that seemed to undermine or oppose or compromise the storied transformation, and all of which had to be sorted out with respect to the Syntactic-Structures-cum-Aspects legacy—steepening the learning curve almost to verticality for many outsiders. What was retained from the celebrated early work? What was dropped? What was modified? What was old? What was new? What was borrowed? What was blue? What was the resulting theoretical picture of all the ongoing and various revisions? These questions had to be answered annually, if not monthly; certainly after every major publication by Chomsky. Books and dissertations had to be emended as they were being written. Including grammatical theory in psychology textbooks seemed pointless. A book like Miller, Galanter, and Pribram’s (1960) Plans and the Structure of Behavior, with its central dependence on Chomskyan linguistics, was now unthinkable. “The adventure into transformational grammar” for many psychologists, recalled Eric Wanner, looked to have “reached a dead end” (1988:150).
For those psychologists who managed to maintain their interest in formal linguistics, Generative Semantics seemed to offer the same attractive and inevitable outgrowth of Chomsky’s early work that most linguists initially found, and they uniformly preferred the more elegant, more intuitive, mediational thought-to-speech picture of language it offered. But it proved no more amenable to experimentation. Nor did the squabbling help. “Linguists between MIT and Berkeley,” as Walter Kintsch complained, “kept changing their minds and contradicting each other” (Kintsch 1984:112). Instability was everywhere psychologists looked, and most just found the dispute arcane and ill-mannered. Chomsky’s reputation suffered the most, in exactly the way it suffered in the media accounts of the dispute. Ebbing Craft indeed.
George Miller jumped ship. His affiliation with Chomsky through the late 1950s had been extremely important for both of them, and their collaborations were foundational for psycholinguistics. By the mid 1960s, with kernels and Generalized Transformations disappearing, recursion moving into the base, and so on, Miller had lost his affection. “I had given up on syntax,” Miller said; “In that area, the theory changed too rapidly” (Baars 1986:216). He also seems to have lost some patience with Chomsky. “At the time,” Miller said of the derivational theory of complexity when it was first proposed and explored, Chomsky “thought it was an interesting thing to do. Later he said, ‘Nobody would do anything that stupid! That’s not what I was talking about. I was talking about competence, not performance.’ But at that time it wasn’t all so clear. So we tried it.” (Baars 1986:208).
The single biggest point of disaffection for psychologists, and many others, was undoubtedly the “broken promise” of early Chomskyan theory, as Jackendoff put it, that “Deep Structure would be the key to the mind” and, following up the ladder of familiar claims, that it would provide the long-held secret of humanity: “If Deep Structure is innate—being dictated by Universal Grammar—then linguistic theory gives us unparalleled access to the essence of human nature” (Jackendoff 2003a:654). Discarding the magic key (in letter if not in spirit by the Generative Semanticists; in spirit if not in letter by the Interpretive Semanticists) violated that promise.
The more linguists I see, psychologists were beginning to feel, the more I like my rats, and someone finally published a response to Chomsky’s Verbal Behavior review (MacCorquodale 1970). Reports of the death of behaviorism began to look somewhat exaggerated.
Things were just as bad for Chomsky in English studies, also because of unrequited optimism. The hoped-for panaceas in composition and poetics failed to cure any ills and made some of them worse; the Linguistics Wars looked arcane and ill-mannered; and English scholars could see, like everyone else, that the damn theories wouldn’t keep still. English folk went from delight to disillusionment at least as rapidly as psychologists, and by the mid-1970s articles like “Why Transformational Grammar Fails in the Classroom” were common—hysterically drawing attention to the negative correlation between national ACT scores and “the rise of the linguistic revolution,” and now using analogies to the New Math (once recommendations) derisively (Luthy 1977).
Things were, if anything, worse yet in AI, the first field to embrace Chomsky’s work, before psychologists or even linguists. This split, as so often with Chomsky, was punctuated with acerbity and lasting bitterness. From the late 1960s on, he made no secret of his skepticism about AI and became positively contemptuous with the rise of connectionist approaches. One of his favored targets was Roger Schank, director of Yale’s AI lab. Daniel Dennett remembers one particular MIT seminar conducted by Chomsky and Fodor which zeroed in on Schank. “If you went by Chomsky’s version,” Dennett says,
Schank had to be some kind of flaming idiot. I knew Roger and his work pretty well, and though I had disagreements of my own with it, I thought that Noam’s version was hardly recognizable, so I raised my hand and suggested that perhaps he didn’t appreciate some of the subtleties of Roger’s position. “Oh no,” Noam insisted, chuckling. “This is what he holds!” And he went back to his demolition job, to the great amusement of those in the room. (Dennett 2014:29–30)
Schank, two decades removed, could still barely contain his bitterness:
The MIT linguist Noam Chomsky represents everything that’s bad about academics. He was my serious enemy. It was such an emotional topic for me twenty years ago that at one point I couldn’t even talk about it without getting angry. I’m not sure I’m over that. I don’t like his intolerant attitude or what I consider tactics that are nothing less than intellectual dirty tricks. (Schank 1996:174)
Computer scientists could surely have lived with the contempt, if only Chomsky’s work had continued to be fruitful, but he had moved away from the mathematical modeling that made it so attractive to them in the first place.
Even the moral implications of Chomsky’s linguistics—or, rather, the lack of them—were under attack, many people noting the irony that one of America’s premier social critics avoided the social aspects of language like a case of the hives. American linguistics prior to Chomsky had long been a field with a social conscience—in a real sense, it was born out of a social conscience—most notably in its approach to indigenous languages and cultures. In the late Bloomfieldian period, this conscience had atrophied somewhat. With Chomsky, it withered away to nothing: if linguistics is a branch of cognitive psychology, its relevance for social issues is negligible at best. But the 1960s and 1970s saw a renewed interest in the sociopolitical dimensions of language, tending to focus more on the urban poor and with that renewed interest came what Newmeyer calls “the moral critique” of Chomskyan linguistics (1986b:120–26). The criticism had a range of concerns, but prominently included the program’s emphasis on English data, gathered through introspection, and on the resultant ivory tower into which it locked the field. Here is Dell Hymes:
The almost exclusive study of their own language, English, by so large a proportion of the world’s linguists, has seemed to the participants a source of deepening insight into the underlying structure of all languages. Leaving aside the methodological difficulties that have become increasingly apparent, we must consider that to many other communities, including those of American Indians, such a concentration may seem an expression of ethnocentrism at best, a hostile turning of the back at worst . . . Many participants in formal linguistics are liberal or radical in social views, and yet their methodological commitments prevent them from dealing with part of the problems of the communities of concern to them. (Hymes 1974b:21–22 [Hymes, not coincidentally, was an ethnological linguist with some pre-Chomskyan allegiances]).
The social climate of the 1960s and 1970s was highly receptive to this style of attack on Chomsky, and many people found it very telling, seeing additional support for it in the overwhelmingly predominant funding source for Transformational Grammar, the U.S. military (Newmeyer & Emonds 1971, Knight 2016:15ff, 2018a).
There were more attacks. Aarsleff pummelled Chomsky’s claims to Cartesian roots, and his scholarship, and his honesty (1970; 1971).2 Bennison and Gray denounced the “Alice in Wonderland state of affairs” in Chomsky’s idealizations and formalisms (Gray 1974:5), and his “inversion of priorities” (1976:38), and the “intellectual fraud” of his work (1977:70). Maher railed against Chomsky’s school—MITniks, MITnik myths, MITnik methods, general MITology, and sundry MITnikia (1973a: passim). Derwing launched a book-length assault on Chomsky’s rhetoric, puzzling in particular over his success “despite key arguments which include the fully specious, the mainly irrelevant, and even the out-and-out false” (1973:222). Other books joined in: Koerner (1975), Robinson (1975), Hagège (1981 [1976]), and Anttila, in an all-purpose, anti-Chomskyan harangue, celebrated these proliferating assaults:
So much of the current criticism is directed against transformational-generative grammar and its various offshoots. It does not mean that an innocent is brutally drawn under gang attack, but that they are indeed real offenders who have brought scholarly discourse to an all-time low. (Antilla 1975:172)
There was, however, little in these attacks to comfort Generative Semanticists. So far from being seen as a welcome alternative to Chomsky’s program, Generative Semantics was usually viewed in this quarter as exemplifying its worst excesses. As Anttila’s celebration of the abuse torrent indicates, it was not only Chomsky, but all things Chomskyan, coming under attack in the 1970s, especially the squabble that had broken out among his formerly close-ranked followers. Maher complained that “every MITnik is a revisionist” (1973b:30), and Talmy Givón muttered from the wings about a discipline going to hell in a handcart, “rife with fads, factionalism, and fratricide” (1979:xiv). Searle noticed some of the Bloomfieldians “rubbing their hands in glee at the sight of their adversaries fighting each other” (1972:20).
Only the philosophers were unmoved by all this agitation around Chomsky, theoretical, rhetorical, and sociological. They never had the same optimism about his program, for one thing, and their own progress, if progress it be, is glacial. Nor are they unfamiliar with wranglings and treason. And Chomsky never fails to give them a good argument. For what more could they ask? He is more of a philosopher than almost any linguist in the history of the discipline. Certainly, he is a closer intellectual kin to the scattered philosophers like Zeno and Humboldt who have had an active, informed interest in linguistics, than to the linguists like Whitney and Bloomfield who have had an interest in philosophy. So, philosophers have welcomed him, in their rough and contentious way, to their hearth. They have attacked him, celebrated him, quibbled with him, and endorsed him. One philosopher, reviewing a book in which Chomsky confronts Dummett, Quine, Putnam, Searle, and assorted other formidable members of the profession, likens the spectacle to “watching the grandmaster play, blindfolded, 36 simultaneous chess matches against the local worthies” (Hacking 1980:47). The interactions between Chomskyans, especially the titular one, and philosophers have been continuous and fruitful. Chomsky’s place in philosophy seems assured.
One outcome of the Interpretive/Generative Semantics squabble is clear. The bickering did not help the reputation of linguistics. Disputes are often healthy for a field, and this dispute makes that case as well as might be imagined. It churned up a great deal of knowledge, raised a diverse spectrum of perspectives, and reorganized the field intellectually in a way that spawned highly productive research programs in the twenty-first century. But bickering, a decidedly pettier activity which can accompany disputes, is much less healthy. Once again the Interpretive–Generative dispute makes that case as well as might be imagined. Bickering has two general effects: (1) It maintains a state of bitterness, the bickerers accumulating more and more personal animus, which virtually guarantees they will never reconcile; and (2) it annoys the neighbors. They move away.
[Generative Semanticists] have said a great many potentially interesting and illuminating things . . . [But] even the most illuminating suggestions are bound to lose a great deal of their light if put forward in a theoretical vacuum, and, furthermore, . . . without the constraining influence of a coherent theory there is bound to be a mixture of proposals, going from the brilliantly illuminating to the downright silly.
Generative Semantics faltered, hemorrhaging confidence. Its practitioners lost their collective direction. The movement fractured into a number of loosely aligned interests. Generative Semantics expired.
Once it became clear that Transformational Grammar could not handle meaning as elegantly as Homogeneous I had promised—that the strong form of the Katz-Postal Principle could not stand—the writing was on the wall. The movement went in several directions at once, and by the mid-1970s it was inescapably clear, to friend and foe alike, that there was no longer a single beast called Generative Semantics. It was legion. Initially, there were promises for definitive texts. First, Ross and Lakoff were to write an authoritative monograph, Abstract Syntax. Then Lakoff was to write another text on his own, Generative Semantics. But new data and new theoretical devices kept derailing their plans. In the end, the closest thing to a definitive text was Donald Frantz’s quasi-publication from the Indiana University Linguistics Club, Generative Semantics: An Introduction, which carries the qualification in its preface that it doesn’t quite meet the demands of its generic title, that a more localized title would be more accurate, Generative Semantics According to Frantz (1974:2).
We can see the fractionation of Generative Semantics in the early-to-mid-1970s most concisely by a quick trace of the paths taken by the four horsemen.
Postal dismounted quite early (around 1972–73, though some later published work still participates in Generative Semantics), to develop an alternate formal model with Perlmutter, Relational Grammar. This framework was in partial competition, partial alliance, with Generative Semantics, but its effects on the fortunes of Generative Semantics were uniformly detrimental. This is what it looked like: Postal was deserting a sinking ship. By his own reckoning, he was the last to come aboard. He was certainly the first to leave, Defector Zero.
Ross drifted into murkier and murkier realms of data and began to argue that the distinctions among grammatical categories and constructions were not, as virtually all grammatical theory had hitherto treated them, distinctions of kind, but only of degree. So, for instance, he proposed that the following continuum (he calls it a squish) more realistically captured the differences among word categories than did the discrete approaches typical heretofore of Transformational Grammar, and of virtually all other forms of linguistic analysis (Ross 1972c:316):
The terms here function more as concentration points of properties and behaviors than as category labels: some words in this view are sort-of, kind-of verbs; others are sort-of, kind-of nouns, and in between them are verby nouns, usually called prepositions, and nouny verbs, usually called participles, and a bunch of androgynes called adjectives. These new sort-of classifications all came embedded in a mesh of very subtle arguments, depending heavily on personal intuition and loose surveys, and involving the interaction of controversial or poorly explored processes, and leaning heavily on imagery:
To pass from left to right [on the verb-to-noun continuum] is to move in the direction of syntactic inertness, and to move away from syntactic freedom and volatility. To wax metaphorical, proceeding along the hierarchy is like descending into lower and lower temperatures, where the cold freezes up the productivity of syntactic rules, until at last nouns, the absolute zero of this space, are reached. (1972c:317)
We pass over these reason-meshes of observation and simile here, fascinating as they are, for the dénouement: squishiness was not a hit. The biggest problem was a familiar one: Ross argued repeatedly that a discrete grammar (like all contemporary transformational work, Interpretive and Generative Semantics alike) faces intractable problems which a nondiscrete grammar would not face. But he never offered a nondiscrete grammar. Gazdar and Klein (1978:666) invoke Pirandello to describe one of his squishy papers, calling it “little more than a collection of data in search of a theory.” Ross’s good buddy Lakoff went even further, saying “no current theory of grammar can even begin to accommodate the facts that Ross has observed” (1973b:271). Of course Lakoff meant this as a high compliment, but it contributed to a distinct lack of confidence that Ross knew where he was going, and really, who wants to follow someone groping his way out into the dark when there is a candle nearby, even a flickering and besmirched candle. There was such a candle held out by the steady and confident hand of Chomsky.
George Lakoff, you might recall, does not lack in confidence either. He went looking for candles, launching a campaign to retool Generative Semantics into something capable of handling squishes; no, of embracing squishes.
The defining notion in Ross’s arguments is the notion of degree, and Lakoff began to splice into his program ideas from parallel disciplines that evoked degrees of variability. He found them down the hall at his new institution, Berkeley. The work of two colleagues in particular inspired him: psychologist Eleanor Rosch and polymath Lotfi Zadeh, with the emphasis on math—a mathematician, engineer, and computer scientist who worked in AI. Rosch and Zadeh both investigated nondiscrete representation, but from different angles. Rosch researched mental representation, around the notion of prototypes; Zadeh, formal representation, in a notation he called fuzzy logic. Zadeh’s work also contributed to the branding of Lakoff’s new approach, which he called Fuzzy Grammar.
Rosch’s research shows that people often categorize by imagistic or feature-density notions rather than by sharp definitions of the [+male, -married] variety. So, for instance, a robin is more quickly and readily identified as a bird than a penguin is, despite people knowing that they both equally fit the dictionary definition of bird (perhaps because it corresponds to more frequent bird-percepts or has richer concentration of features typically associated with birds, or both). Zadeh’s logic operates not on the traditional binaries, true and false, but degrees of truth. A term that comedian Stephen Colbert later coined for much different purposes, truthiness, is actually a pretty good label for Zadeh’s focus.
Lakoff published a number of investigations blending Rosch and Zadeh’s work, but the principal paper on this approach that was still more or less under the Generative-Semantics umbrella is his analysis of linguistic hedges, like sort of and pretty much (1973d [1972]). He claims that the truth of such statements as those in 1, is graduated, that their truth values move from 1a, which is absolutely true, to 1e, which is absolutely false, through several intermediate degrees:
Then, he charts the interaction of certain hedges and intensifiers (sometimes called degree words) on the perceived truth values of such sentences:
So, 1a is the truthiest of all (getting the highest value in Zadeh’s calculus, 1), and certainly truthier than 1b, in turn truthier than 1c, and so on, until we get to the absolute zero of truthiness (with, appropriately, a fuzzy-logic value of 0), 1e.
The end game of importing these notions into linguistics is to put them into the service of grammatical theory, in part to help make sense out of Ross’s squishes. But Lakoff doesn’t get very far. He proposes a rough sketch of some logical apparatus for treating the semantics of hedges, and, appropriately, he hedges it substantially. “I don’t want to give the impression that I take the proposals . . . to be correct in all or even most details,” he says. “In fact, it is easy to show that far more sophisticated apparatus will be needed merely to handle the hedges discussed so far” (G. Lakoff 1973d [1972]:246).
And fuzziness was not all Lakoff was up to in the mid-1970s. He was also exploring ways to incorporate Grice’s informal work on conversations into formal grammar, looking at syntactic amalgams, and linguistic gestalts, and generally probing pragmatic phenomena, and expanding his investigations of logic, under the slogan of “natural logic,” to the point where he was calling for “hundreds, perhaps thousands more” concepts (Parret 1974 [1972]:162).3
George Lakoff was very, very busy.
McCawley was busy, too, but in a much more low-key and sharply angled way. Early in the debates, he advanced some of the strongest theoretical arguments, largely around the theme of reducing complexity by treating syntax and semantics as a unitary phenomenon. But McCawley dropped his meta-theoretical approach in the early 1970s to pursue more specific work. That shift in approach came from his unease about the lack of reflection in his own argumentation (for instance, around the notion of grammaticality), and his perception of the sloppiness in linguistic argumentation all round him. He made a conscious decision to be more responsible: to concentrate on trenchant critiques of the Interpretivists and on the application of Generative Semantics principles to specific phenomena. His work in the 1970s largely assumed Generative Semantics, rather than championing it (like Postal) or celebrating it (like Lakoff), though he didn’t make his motives clear until the end of the decade:
I do not mean to suggest by [assuming, rather than explicitly marketing, Generative Semantics] that I am so arrogant as to regard all the controversies over claims I and other “Generative Semanticists” have made about the unity of syntax and semantics as having been settled in our favor. Rather, I simply think that for further discussion of these questions to be productive, the disputants need a much broader and deeper understanding of the relevant factual areas than any of them (myself included) had around 1970. (1979:viii)
He worked fastidiously to replace “the sweeping and often rash generalizations in my earlier work about the relationship of grammar to logical structure and the lexicon by more detailed proposals whose backing is less anecdotal” (1979: viii).
McCawley, in short, spent most of the 1970s buttering syntactico-semantic parsnips. He followed Postal’s reductionist program to find the atomic categories; arguing, for instance, that tense (1976b [1971]:257–72) and not (1976b [1971]:277—84) were underlying verbs, and defending Ross’s analysis of modals (1979 [1975]:96–100). Eventually, this line followed the path of the modern atom very closely, and McCawley finally argued that there were no syntactic categories at the deepest level after all, that they were composites of smaller particles yet (1982b [1979]:200)—something that was in principle compatible with Ross’s squishes, but no merger was ever pursued. He also proposed that the underlying word order of English was not, as had been assumed since Harris, a reflection of the canonical surface order, subject-verb-object, but more like Polish-notation symbolic logic, verb-subject-object (1976b [1970]:211–28). The argument, regarded as something of an unwitting reductio by many anti-abstractionists, is a clear, well-reasoned application of Occam’s razor to the organization of Transformational Grammar: Adopting a deep verb-subject-object order complicates no transformations and simplifies fully a third of them. The VSO hypothesis was adopted by most Generative Semanticists, rejected by all Interpretivists; the expected pattern.
But here’s the thing about that pattern, McCawley’s VSO hypothesis has no necessary connection to the architecture, or any unique mechanisms, of Generative Semantics. While it complicates some of Chomsky’s more cherished working assumptions of the period (in particular, that there is an underlying verb phrase), it is fully compatible with an Interpretive Semantic component—as, of course, is the position that not is a verb, or that tense is a verb, or that syntactic categories are not primitive. Much of McCawley’s analytical work in the 1970s was conducted under the flag of Generative Semantics (he didn’t officially fold that flag until the end of the decade, and one suspects he kept it lovingly in a hope chest thereafter). But that work, more properly, was Abstract Syntax.
Abstract Syntax, by name or otherwise, was out of favor. It had been rejected virtually wholesale by the Interpretivists, since it was temperamentally, if not logically, incompatible with Lexicalism, but other Generative Semanticists appeared to have little use for it either. George Lakoff was working on everything but. Ross was wading in squishy data. Robin Lakoff, an early pioneer with Abstract Syntax and Latin Complementation (1968), had turned away from formal modeling to pursue her pioneering work on gendered language practices (1973c 1975). More broadly, as a diversity of models began to characterize linguistics, concreteness became a byword for responsible theorizing, sometimes deliberately counterposed against Generative Semantics. “The most noticeable characteristic of daughter-dependency syntax,” Richard Hudson wrote in advocacy of his model, “compared with transformational syntax, is probably its concreteness.” First among the virtues of concreteness, he added, is that “it rules out the kind of highly abstract syntactic structures typical of generative semantics” (Hudson 1976:23). Aside from McCawley and a few of his students (most notably, Judith Levi and Georgia Green), abstract transformational analyses were shunned.
So, ten years after the “Remarks” lectures: Postal was gone; Ross was hip-deep in murky data; Robin Lakoff was off inventing feminist sociolinguistics; George Lakoff was as inventive and daring as ever, but wouldn’t sit still long enough (figuratively or literally) for other linguists to get a fix on his work; and McCawley was out of step with the times.
I have never made but one prayer to God, and a very short one: ‘O Lord, make my enemies ridiculous.’ And God granted it.
Without a defining center, and with the times a-changin’ (protests were out, discotheques were in), many of Generative Semantics’ identifying traits became liabilities, and Chomsky’s reputational ebb did not benefit them. A receding tide lowers all boats. The Generative/Interpretive squabble certainly hurt both sides, but the boisterous Generative Semanticists were a much easier target. For Gray, “Chomsky sounds quite Trotskyish compared to the logico-linguistic Stalinists of the younger generation [specifically citing McCawley]” (1974:4), and Hagège complained that “the promoters of Generative Semantics have only prolonged, extending them to the point of caricature, [the] already existing procedures” that he found repugnant in Chomsky’s work (1981 [1976]:83). Maher, giving full vent to his fetish for capital letters, identifies the worst MITniks as the authors of “innumerable facetious pieces of juvenilia presented as scholarly papers at LSA, CLS, and other TG club meetings,” particularly “QPhD’s [Quang Phuc Dong’s] pupils Binnick, Morgan, and Green” (1973b:30).
The juvenilia in Maher’s complaint had mostly to do with the loony humor infusing Generative Semantic work. He was not alone in entertaining the view that the lack of seriousness ran deeper than style. O’Donnell, for instance, sniffs “serious grammatical studies may, as [G. Lakoff] claims, be in their infancy: serious grammarians, however, are not” (1974:75). But there was another problem with the humor. It was mostly directed, rather narrowly, toward other Generative Semanticists. It was not invitational; Robin Lakoff calls the characteristic style of her community “a kind of secret handshake” (1989:977). One example of this invitational failing, in a major article by her husband (1971b), was especially disastrous. It was written as a reply to Chomsky’s “Deep Structure, Surface Structure, and Semantic Interpretation” under very tight publication circumstances, and Lakoff decided he would parody Chomsky:
Chomsky wrote this paper [“Deep Structure”] attacking Generative Semantics, for a collection by Steinberg and Jakobovits [1971]; Danny Steinberg called me up and asked me if I wanted to write a reply. I said “Sure.” And he said “Well, you have a month.”
I looked at this thing. It was this huge paper. It was full of quotes taken out of context, and characterizations of our position that were really wrong. I had a month to reply to it. I didn’t know how to deal with it, so I decided to write a parody. I took [Chomsky’s] own style, and tried to turn the style against him. Apparently, he never realized that it was a parody.
This quotation is somewhat misleading in that “On Generative Semantics” is not a flat-out parody of Chomsky, though it certainly contains a good many parodic elements, particularly in the first few pages. It is the mixture of imitative mocking and straight-ahead argumentation that mostly sinks the paper, but the immediate lesson here is that satire always misses some people—if rhetoricians had such things, it would be one of their Ironic Laws—and there is no question but that Chomsky, for this satire at least, was one of those people.
Chomsky completely failed to recognize the parody, welcoming Lakoff’s paper graciously for having adopted, “with only a few changes, the general framework and terminology of [“Deep Structure”], so that the differences between [Interpretive] and Generative Semantics, as so conceived, can be identified with relative ease” (1972b [1969]:134). Chomsky’s missing the joke, in fact, became the source of yet more back-room knee-slapping for some Generative Semanticists, showing just how irredeemably humorless their opponent was. But it is an easy joke to miss.
Once the paper is identified as a parody of “Deep Structure,” it is very effective. Lakoff sets up a “basic theory” in parallel to Chomsky’s “standard theory,” and defines it so broadly as to encompass virtually any conceivable theory of language, thereby making every theory a notational variant of the basic theory. Much of the style is recognizably mock-Chomskyan, with rather tortuous syntax, numerous uppercase variables, subscripts, and superscripts, and the tendency to traffic for a long while in abstractions before offering specific proposals or analyses.
But parody needs some other signal than a backstage elbow in the ribs of a confederate if it is to serve some purpose beyond closed-off, insider ridicule, and Lakoff utterly fails to flag his paper as a parody for the broader community. The title—“On Generative Semantics”—is serious, and, in fact, promises a definitive treatment of the movement Lakoff very prominently represents. Nothing in the text is so outrageous as to provide an unequivocal clue. 4 Lakoff apparently regarded the following definition of “the semantic representation SR of a sentence” as patently absurd (attacking the sort of work Jackendoff published in Semantic Interpretation):
SR = (P1, PR, Top, F, . . .), where [P1, is the first phrase marker in a derivation,] PR is a conjunction of presuppositions, Top is an indication of the topic of the sentence, and F is the indication of the focus of the sentence. We leave open the question of whether there are other elements or semantic representations that need to be accounted for [hence, the ellipses at the end of the formula]. (1971b:234–35)
But without knowing Lakoff’s intention, the definition just makes him look like a wild-eyed semantic imperialist, a conception most Interpretive Semanticists already held of him, and one that many spectators held of Transformational Grammarians in general. Nor is this proposal much stranger than the one that Lakoff was seriously advocating only the following year:
The abstract objects generated [in Generative Semantics] are not sentences but quadruples of the form (S, LS, C, CM) where S is a sentence, LS is a logical structure associated with S by a derivation, C is a finite set of logical structures (characterizing the conceptual context of the utterance), and CM is a sequence of logical structures, representing the conveyed meanings of the sentences in the infinite class of possible situations in which the logical structures of C are true. (Parret 1974 [1972]:163)
There is even a direct parallel with the ellipsis of the “parodic” 1971 formulation, as Lakoff adds “even this is inadequate,” since “one must take into account much more than conceptual contexts” (Parret. 1974 [1972]:163).
Much of “On Generative Semantics” is also written utterly deadpan, particularly the opening pages, and it is not until late in the article, when Lakoff starts proposing rule features like [± pedro] that there is a fairly clear hint the author is not simply drab by constitution, but drab by choice; until that point, it is more reasonable to regard the author stylistically as a clone of Chomsky, whatever their differences, than a parodist of Chomsky. Even the sample sentences are very restrained, prominently featuring Chomsky’s old friend John.
More problematically, Lakoff’s paper contains the one ingredient that muffles irony better than anything, a good deal of serious discussion, which is frequently impossible to disentangle from the smirking.
Almost no one beyond the inner Generative Semantics circle, in any case, appears to have got the joke, certainly not the book’s editors. Howard Maclay’s introductory discussion of Chomsky’s and Lakoff’s papers, for instance, takes Lakoff’s at face value (pausing to remark that “the extent to which the structure of argumentation in [Lakoffs paper] is modeled on Chomsky is quite striking,” 1971:178), and it is. Here is Chomsky:
I will refer to any elaboration of this theory of grammar as a ‘standard theory,’ merely for convenience of discussion and with no intention of implying that it has some unique conceptual or empirical status” (1972b [1968]:66)
Here is Lakoff:
I will refer to the above theory of grammar as a ‘basic theory,’ simply for convenience and with no intention of suggesting there is anything ontologically, psychologically, or conceptually basic about this theory. (1971b:236)
“On Generative Semantics” entered the literature as just another article, with even Lakoff, apparently determined to occlude his joke completely, regularly citing it in subsequent work as a straightforward account of the theory of Generative Semantics (1970b:627, 635n7; 1970e:542n2; 1972d:290n4). Its effect was uniformly deleterious, contributing as much as any individual paper could to the communication breakdown that characterized the dispute in the early 1970s. Spectators to the dispute generally took the essay as signaling the departure of coherence—illustrated most strikingly by Dubois-Charlier’s (1972:43) comments on “the paradoxical aspect of the matter . . . that both Chomsky and Lakoff end up by saying at the same time that the two theories are identical, but that the other’s is wrong!” (Hagège 1981 [1976]:33n29).
The worst consequence was one that Lakoff really should have expected, given the title, the venue, and his own reputation: “On Generative Semantics” was soon taken to be “an important manifesto of the approach” (Raskin 1975:462). It was the only publication from any of the leading figures in the movement to include Generative Semantics in the title and it appeared in a very prestigious anthology. An earlier version of the paper (1969b) was given at a CLS session, stirred along by a live presentation, where it very likely went over splendidly. Most of the audience would have been sympathetic to both its aims and its style, and irony works much better with visual and tonal, nudge-nudge, wink-wink signals. Everyone in the audience would have been familiar with Lakoff’s personality and Chomsky’s tone, as well as with the issues surrounding the definition of the standard theory. Even the published version of that talk, in the proceedings of a conference becoming famous for its sense of goofiness, may not have caused much confusion. But, in a thick, staid volume, alongside papers by Grice, Strawson, Searle, Quine, Katz, and Chomsky, and a whack of others, all of them very serious in tone and intent, it is extremely difficult to take the essay as anything but a single-minded promotion of Generative Semantics, and a befuddled one at that. McCawley’s rebuke is rather mild in these circumstances:
Lakoff is guilty of . . . a failure to distinguish adequately between what he would seriously propose as a correct theory of grammar and what he offered (partly in jest) as a general framework for the discussion of theories both correct and incorrect. (1982b [1973]:75)
Field Marshal Radetzky is reported to have told his troops, retaking Sardinia for the Hapsburgs in 1848, “Spare the enemy generals—they are too useful to our side” (Robertson 1952:354). Lakoff was, as a general, extraordinarily useful to the Interpretivist cause.
By the time Katz and Bever put together what Newmeyer (1980a:169; 1986a:134) calls “the consummate critique of the philosophical implications of late Generative Semantics,” a few years later, they were sociologically and rhetorically justified in zeroing in on “Generative Semantics, Lakoff Style” (1976 [1974]:30). By the mid-1970s, it was natural to consider the Generative Semanticists as “Lakoff and those associated with him” (O’Donnell 1974:54). With Postal gone, Robin and Ross on different fringes, and McCawley avoiding strong theoretical claims—the role was George’s pretty much by default, and, well, George Lakoff rather likes the spotlight.
One might even say that Lakoff campaigned for the role as the sole and singular leader of Generative Semantics. As early as 1972 he was defining Generative Semantics very clearly in terms of his own (broad) personal interests, making little reference to the work of Postal, Ross, or McCawley, and citing his stillborn 1963 paper as the starting point (see especially Parret 1974 [1972]:151–78), a claim he has made or assumed repeatedly since. For instance, in a discussion of logical form—in a festschrift for McCawley!—he mentions McCawley in only the most cursory way, represents his own work as definitive, and suggests the logical enlightenment begins in 1963 (1992:175–76).
The late stages of the schism, accordingly, are often characterized as Chomsky versus Lakoff (or Lakoff versus Chomsky; most linguists, then as now, have a preferred home team). The contrast could not have been stronger. Chomsky defined his mid-1970s program in terms diametric to Lakoff’s most notorious theoretical proposal, global rules.
With Lakoff at the helm, promoting his very wide conception of linguistic theory, and a growing Interpretivist attack based on the virtues of restrictiveness, the general perception came to be that Generative Semantics was theoretically promiscuous, incapable of saying no. It opened its doors to any and all phenomena impinging upon language. It adopted increasingly powerful formal mechanisms: global derivational constraints, transderivational constraints, even meta-transderivational constraints. It was given to frighteningly naïve and sweeping claims (“What we have done is to largely, if not entirely, eliminate pragmatics, reducing it to garden variety semantics,” G. Lakoff 1972b:655). Generative Semantics concerned
not just syntax-semantics, phonetics-phonology, historical linguistics, anthropological linguistics, etc., which form the core of most academic programs in this country, but also the role of language in social interaction, in literature, in ritual, and in propaganda, and as well the study of the relationship between language and thought, speech production and perception, linguistic disorders, etc. (G. Lakoff, in Parrett 1974 [1972]:151)
One gets the impression that Lakoff stopped the list more because he ran out of breath than because he ran out of vision, carefully remembering to throw in that etc. before gulping for air. Generative Semantics under Lakoff tried to do too much, this line of argument goes, and it burst at the seams. There is much to recommend this version of events. Certainly, Generative Semantics tried to do a great deal, and certainly Lakoff was one of the prime forces stuffing the bag, stressing the seams; and, certainly, it burst. It burst very publicly. As a matter of course, the great majority of Generative Semantics papers included comments like:
We are forced to conclude that, awkward though it may seem, the similar properties of both and each cannot be accounted for by the same formal mechanisms in our existing theory. An explanation of whatever underlying regularity there may be will have to wait for a cleverer linguist. (Carden 1970:189n 10)
And:
This paper was undertaken as an attempt to shed light on some very mysterious problems. I fear I have done little more than show which lamps have cords too short to reach the outlets, but hopefully this information will be helpful eventually in finding explanations for these mysterious distributions. (Green 1972:93)
And, most tellingly:
It is not a very satisfactory experience to write an entire paper without being able to offer any decent analyses or explanations for the phenomena I have discovered. It is, however, an enlightening one, and I believe, a necessary one. (Lawler 1972:255)
These were not just declarations of the inadequacy of Generative Semantics, of course: it was the entire program of formal syntactic theory-building, back to Syntactic Structures and beyond, that Carden, Green, Lawler, Lakoff, et al. were seeking to expose as woefully inadequate.
In fact, a new genre arose that pushed this data-mongering tendency to the limit (the Green and Lawler quotations, in fact, come from exemplars of the genre). Gene Gragg calls such articles “creature features” and defines the genre’s intentions as “ point[ing] out some oddities which a theory of . . . speech will eventually have to come to grips with” (1972:75), but which nothing on the horizon appears capable of treating. The seminal creature features—Postal’s highly corrosive underground classics, “Linguistic Anarchy Notes”—make it very clear that bursting the seams of linguistic theory is the primary motivation. The first note begins:
This is the first in a random, possibly nonfinite series of communications designed to show beyond any doubt that there exists no linguistic theory whatever. There are apparently endless numbers of fact types not incorporable within any known or imaginable framework. In particular, what has been called the theory of transformational grammar, seems to have only the most partial relation to linguistic reality. (1976 [1967–70]:203)
The notes turned out to be finite in the end, but they sparked an increasing number of similarly dissensual efforts, papers whose sole aim was bringing to light data that would give any and all pretenders to theoryhood the heebie-jeebies. The high point (or low point; take your pick) of the cycle came in the first few years of the 1970s, with creature features like “Semi-Indirect Discourse and Related Nightmares” (Gragg 1972), and “Read at Your Own Risk: Syntactic and Semantic Horrors You Can Find in Your Medicine Chest” (Sadock 1974b).5
These papers were only just the most overt symptoms of a mood pervading Generative Semantics of the period, one which shows up in the nooks and crannies of the overwhelming majority of papers. Even Postal’s public persona, which, whatever he said in private or circulated underground, was normally cocksure and authoritative, cracked in one straightforward Linguistic Inquiry paper, which ends “There is, of course, an explanation of these [very strange and mysterious properties] but, believe it or not, the present writer does not know what it is” (1972d:400).
Postal’s Anarchy Notes raised something of a stink. In particular, it is said that Halle thought them very ill-advised, probably fearing that they would spark exactly the kind of attitude that they did in fact spark—though fan may be more appropriate—and S.-Y. Kuroda chided him gently with his “Linguistic Harmony Notes” (the first number being “Charms of Identity”—1976 [1967]). But we shouldn’t miss the important scientific motives behind this traffic with embarrassing data. It is not to deal a nihilistic blow to the heart of formal linguistics. One of Postal’s avowed short-term aims in the Notes, true, is to demonstrate in the most graphic way possible that the enterprise is “not just slightly in error and rather incomplete”—the attitude that characterized much of the interpretivist camp—“but in deep ways hopelessly far from linguistic reality” (1976 [1967—70]:215); his ultimate aims, though, are much higher. His motives, he says, “are entirely positive.” He is trying to save grammatical theory from itself:
Many people today are engaged in the attempt to construct linguistic theories. My view is that an important difficulty with all such attempts is that there is not a good a priori statement of the full range of known facts which a theory must handle. To the extent that theories are formulated in the absence of explicit awareness of this range of facts, they are dreamlike. (1976 [1967–70]:205).
This desire to save grammatical theory from its own short-sightedness was the single most powerful impulse behind the Generative Semanticists’ preoccupation with recalcitrant data.
Even in the face of such data, however, many Generative Semanticists kept an essentially Chomskyan formalist framework, prominently including Postal, rejecting or modifying only peripheral assumptions. A strong case can be made, in fact, that Chomsky deviated further from Syntactic Structures than Postal or McCawley, not just in his overall career arc, but in the program he was pursuing in the 1970s.
Whatever our metric for deviation, however, one thing is clear: the Generative Semantics ethos, in its goofiness, in its insularity, and in its creature-feature, data-loving anarchism, backfired.
Generative Semanticists dispersed like the crowd after Woodstock, everybody wandering off, individually or in clumps, to pursue other interests. Some straggled off under the force of Chomsky’s restrictiveness arguments, and the concomitant unwillingness of any prominent Generative Semanticist—in particular, the unwillingness of G. Lakoff, who had issued the call for global grammar—to meet the challenge of those arguments. A few Generative Semanticists, most notably Frederick Newmeyer and David Lightfoot, felt that the best hopes for the field lay with Chomsky’s program. But most went in other directions. While Postal publicly departed early, our other horsefolk might be said never really to have ‘left’ the framework at all, in the sense that they retained various constellations of the commitments on which they built that framework.6 But all of them abandoned the name.
G. Lakoff’s and Ross’s work mutated in directions that bore less and less resemblance to their original starting point, adopting new labels along the way. First Lakoff began talking about Global Grammar, which was a clear continuation of Generative Semantics, but the name implied a new direction, the crossing of a threshold (in fact, the threshold Postal marked between Homogeneous I and Homogeneous II). Then Ross became associated with Squishy Grammar, Lakoff with Fuzzy Grammar, approaches that had shared perspectives, but which were different enough to have separate names. Lakoff also embraced names like Cognitive Grammar and Experiential Linguistics, which identified different elements of one evolving program, but which made it look as if he was hopping from theory to theory depending on what example sentences he saw on a billboard on the way into work that morning.7 The constant renewal of names for his work was largely true of Ross as well, who eventually rounded to an “almost terminal distrust” for labels, noting in the 1990s that “I have already tried on modifiers for linguistics like human, holistic, ecological—and each time they tended to sound like some increasingly more horrific concoction of the trendy, the buzzword, the big deal” (1991:2), so he resolved on this title for a late credo: “Toward a Linguistics” (1991). McCawley kept doing what he always did, producing perceptive and challenging syntactic, semantic, lexical, and phonological analyses, far closer to the mode of classic Generative Semantics than anyone else, but he followed Ross’s suit in abandoning not just the label Generative Semantics, but the practice of labeling altogether. By a 1979 conference on syntactic approaches, he one-upped Ross’s fill-in-the-blank credo with a negation, Unsyntax, to label the work he was doing (Kac 1980; Moravcsik and Wirth 1980).
The label Generative Semantics was not entirely dead. Time-delayed surveys written for beginners and exogenous scholars were still hailing Generative Semantics, usually in its Homogeneous I incarnation, as the “latest development,” and it continued to surface in polemics from Interpretivists. But linguists everywhere knew that the movement was finished. The tag was on the toe. The drawer was shut. The light was out.
At the same conference where McCawley was discussing Unsyntax, John Lawler lamented being a fluent speaker of a “dead metalanguage” (1980:54, 59n12). Gerald Gazdar, reworking his 1976 thesis for publication (1979a), put all the present-tense references to Generative Semantics into past tense (and, in another sign of the times, also changed the bathroom-haunting Nixon of his example sentences to Dixon). Suzette Haden Elgin’s second edition of her What Is Linguistics? (1979), excised her invocation of Generative Semantics. I cited a review a few chapters back saying that the true title of Stockwell’s Foundations of Syntactic Theory textbook should have been Introduction to Generative Semantics. Since Stockwell’s text came out in 1977, this was not a recommendation. The review continues:
There is a sad irony in the fact that the first, and possibly only, [mainstream] Generative Semantics textbook should appear at a time when it must have become apparent to even the greenest postgraduate that the Generative Semantics paradigm has been quietly abandoned by virtually all its major proponents (abandoned for Montague Grammar, Radical Pragmatics, Relational Grammar, Chinese cooking, or some combination thereof). (Gazdar 1979b:179)
The string of post-Generative-Semantics pursuits Gazdar invokes signals the fractionation wrought in the last years of the dispute (well, except for Chinese cooking, which is just a tip of the hat to McCawley, who hadn’t really abandoned anything but who was a known gourmand of numerous cuisines; see McCawley 1984). The end of hostilities brought a measure of harmony, but no unity. There was no Pax Linguistica. The Generative enterprise, bifurcated in the late 1960s, splintered in the late 1970s.
Generative grammar was initially defined by Chomsky as “simply a system of rules that in some explicit and well-defined way assigns structural descriptions to sentences” (Chomsky 1957a:48). Explicit, well-defined, and a host of other synonyms evoking clarity and precision, were routine in the vocabulary of the early movement, promising to bring the rigor and purity of axiomatic logic, and the glow of the natural sciences, to linguistics. Although some of our heroes tried valiantly to hold them back over the course of the dispute—most notably, Jackendoff and McCawley—looseness and vagueries abounded on both sides, and many linguists saw precision as the ultimate casualty of the Linguistics Wars. Chomsky continued to issue promises of precision and rigor (“we want to formulate precise principles and precise rules within a formalized system”—Chomsky 1979 [1976]:125), but they were no longer seen always to match his own practice. Pullum’s “Formal Linguistics Meets the Boojum” (1989; 1991:48–55) was satiric, but all satires have a target, and Chomsky was in its cross hairs. The Boojum, you recall, is that variety of snark which causes one to “softly and suddenly vanish away” (Carroll [1876]); explicit and well-defined formal linguistics, softly and suddenly, was gone from the work of its revolutionary promoter and most famous advocate.
Many linguists who had entered the field to help achieve the Chomskyan goals of rigor and precision were betraying disillusionment with the Ur-generativist. “A sizeable number of grammarians had begun to express reservations about the direction in which Chomsky . . . was headed,” Newmeyer remembers, “and began to develop . . . alternative models whose common feature was an even further diminished role for transformational rules” (Newmeyer 1996:156).
“We were not idiots,” Postal says, “we knew we had lost” (Huck & Goldsmith 1995:136). Generative Semanticists didn’t cross the ring to raise Chomsky’s arm in victory—or, only a very few did (Lightfoot, Newmeyer)—but other approaches were developing (most notably, Montague Grammar and Relational Grammar), along with other interests, especially around pragmatics and sociolinguistics.
Obituaries for the transformation, though, proved premature.
Where then does the expression “Generative Semantics” come from? It is from a general attitude or point of view which was expressed, for example, by Lakoff in an article entitled “[On] Generative Semantics,” or by Postal in his 1969 article “The Best Theory.” But nobody—at least not to my knowledge—has actually accepted this theory, which in the form presented was virtually empty. What the theory asserted was that there exist representations of meaning, representations of form, and relations between the two. Furthermore, these relations between the two representations were virtually arbitrary. . . .
A theory that permits global rules has immense descriptive potential. . . . To approach an “explanatory” linguistic theory, or—which is the same thing—to account for the possibility of language acquisition, it is necessary to reduce severely the class of accessible grammars. Postulating global rules has just the opposite effect, and therefore constitutes a highly undesirable move. . . .
[My own work has led to hypotheses which] restrict very severely the expressive power of transformational rules, thereby limiting the class of possible transformational grammars.
—Noam Chomsky (1979 [1976]:150, 152, 180)
Chomsky turned his back on Generative Semantics in his formal publications by 1970. He figured his “Some Empirical Issues” (1972 [1969]) settled the matter. He made no secret of his contempt for Generative Semantics to his students and his colleagues, and in the occasional interview, where he played variations on his worst-imaginable-theory theme, but there was nothing more public from him than that. He had other causes to pursue, political as well as linguistic, and his fame continued to grow well beyond the academy. But the lines of the attack had been drawn up and the Interpretive Troops waged war.
Chomsky’s technical work followed the rhetorical arc he adopted in opposition to the “Transformationalist” position he pinned on G. Lakoff’s ass in 1967. He followed this arc, a campaign against the specificity of transformations, as far as it could go, almost. All specific transformations were eliminated. Some were axed outright by various theoretical tradeoffs, the process that had begun with “Remarks,” where transformations were swapped out for lexical redundancy rules. Deletions, for instance, were eliminated in favor of base-generated, phonologically null proforms. So, where the Aspects model derived sentences like 2a from Deep Structures like 2b by deleting someone and the second occurrence of Leila, the new model borrowed from the dummy-symbol generation playbook of the Aspects period and just plugged empty pronouns like PRO and Op (for empty Operator) directly into the Deep Structure (or whatever its current facsimile happened to be), as in 2c. As always, there’s a bit more going on (some movement, its trace, co-referential indices, maybe other stuff), but that was effectively the new easy-to-please story.
Talk about your notational variants: instead of putting things in at the base and then taking them out in the course of a derivation, we just put in some holes where they would have been if we hadn’t taken them out. But somehow an increasingly elaborate network of filters, constraints, pulleys, levers, and so on (none of them global!), ensured that the overall grammar was more “restrictive,” the new God term.
In the most interesting (and, for a while, the most controversial) policy change, all the movement rules were collapsed under one very general transformation. The beloved Passive was out; Topicalization, out; Wh-fronting, out; the beloved Affix-hopping, out. In a very real sense, all of the 1957–1965 work of developing, modifying, and calibrating specific transformations to describe and explain syntactic phenomena was out. A single indiscriminate rule, called simply Move, was in.8 This rule could move any constituent anywhere and therefore, on its own, would just churn up all kinds of constituent mixtures, every conceivable kind of word, empty category, and phrase scrambles—transformational frittatas. But it wasn’t on its own. Now there was a powerful battery of filters and constraints to ensure that (in principle; many details remained to be worked out) only the grammatical sequences actually made it to the surface.
The resulting model came to be known as the Y Model but also the T Model (consistency is for the timid!), as well as the Inverted Y Model, because of its resemblance to an upside down Y, as in Figure 7.1. The model, like Homogeneous I, is deceptively simple, with the emphasis on deceptively. The Roman numerals all represent distinct rule systems: I represents Phrase Structure Rules; II represents Transformations; III represents Phonological Rules; and IV, “rules of the LF component,” essentially Semantic Interpretation Rules concerned with quantifier scope and anaphoric relations. The other elements of the model are, as follows: PF is “phonetic form”; LF is “logical form” (sound familiar?); D-structure and S-structure I’m sure you can work out without any assistance, but I guess I forgot to mention that Chomsky repudiated the term Deep Structure, sort of, and Surface Structure, sort of. In his 1975 Whidden Lectures at Canada’s McMaster University, Chomsky pledged to avoid Deep Structure especially, because, for some reason—after years of associating it with thought, Universal Grammar, and the ultimate goals of linguistics—people unfortunately had the impression that “their properties . . . are truly ‘deep’ in the non-technical sense” (1975b:82). Chomsky tried initial phrase marker and even (foreshadowing “The Artist Formerly Known as Prince”) “the former ‘deep structures’” (1975b:82), and he regularly used quotation marks around the phrase, before finally settling on the coy D-Structure, with its companion, S-Structure.9
Figure 7.1 An outline of Chomsky’s Y-model, from Chomsky 1986c:68.
Chomsky stopped concerning himself very directly with semantic argumentation, to the extent he ever did. Semantics was a carrot, attracting a good deal of attention to his model in the late 1950s and early 1960s. He endorsed the work of Katz and Fodor, and helped them bend the syntax to the will of their limited semantics, but he did not engage directly in work on the semantic component. When semantics became a stick with which to beat the Generative Semanticists in the late 1960s and early 1970s, the same was true. He raised problems for the Katz-Postal Principle—by attending to focus and presupposition phenomena, and by discussing quantifier-scope difficulties— and he endorsed solutions to these problems when it suited him, but he never proposed any solutions on his own beyond waving at Surface Structure (with Jackendoff taking notes). Once Generative Semantics was pretty much dispatched, he stopped mentioning these issues. His interest in explicit semantics has always been modest. The only aspects of meaning that Chomsky feels any responsibility for in his work are the ones that directly and exclusively result from syntactic structures.
Chomsky has been called, rightly, a syntactic animal (Passmore 1985:39), and Jackendoff characterizes the attitude around MIT in the mid-1970s as “Let’s do some syntax again, now that this whole schmeer is over.” The whole schmeer wasn’t quite over at that point—in fact, the Interpretivist assault was moving through the streets, building to building, mopping up perceived pockets of resistance—but the threat was gone, and the theoretical questions in Chomsky’s framework took a decided turn back toward syntax.
Through the 1970s, Chomsky’s program had its own nominal complications. As it continued to develop, it was alternatively but uneasily known at the time by a couple of different and overlapping labels—the Revised Extended Standard Theory, signaling increased modification, and Trace Theory, taking the name of its single most prominent modification, though neither is particularly useful. In fact, it became clear at about this point that putting labels on Chomsky’s work at all is an iffy business at best, even the ones he himself endorses (Government-Binding Theory for a while, giving way to Principles-and-Parameters, then to The Minimalist Program, which is the late-Chomskyan preference, in some kind of overlap with Biolinguistics).10 A more productive approach might be to treat him like a great and restless artist—like Picasso, with his post-impressionist period, his blue period, his rose period, his cubist period . . .
Just over the course of our story, we have seen Chomsky’s Syntactic Structures period, his anti-Bloomfield period, his Aspects period, his anti-Generative Semantics period; as with Picasso, one blurs into the next, positive moods alternate with negative ones, elements and themes connect through them all. The next period, when he broke rather dramatically from his anti-Generative Semantics focus to concentrate on the more positive work, is marked most clearly by “Conditions on Transformations” (1973a [1971]), the paper in which, Sadock tells us, Chomsky began speaking a new lingo. The majority of Chomsky’s work in formal linguistics between 1966 and 1971, that is, was directed against Generative Semantics, rather than for Interpretive Semantics. Even positive and massively influential suggestions—x-syntax, for instance, and lexicalism—seemed to interest him at the time primarily for the obstacles they placed in front of Generative Semantics. With the Conditions period, propelled by its restrictiveness rhetoric, he began again to articulate an appealing, workable new linguistic model.
At exactly the time that it became impossible for linguists to find a unified and coherent program under the label of Generative Semantics, Interpretive Semantics was getting its act very much together. With Jackendoff (1972) and Chomsky (1973a [1971]), offering a genuine alternative, rather than a few sketchy and barbed suggestions, increasing numbers of students, junior professors, and neutrals found the new package persuasive. The package included not only increased attention to lexical items, an elegant new treatment of phrase structure, and a simultaneously stripped down (the rules) and souped up (the conditions) transformational component, but also definitionaly entailed the utter and complete rejection of Generative Semantics. Chomsky and his allies defined his more “restrictive” theory largely against the Bogey Man of Generative Semantics.
The feeling was mutual. The one common denominator among the overwhelming majority of scattering Generative Semanticists is that, whatever else they were up to, they continued, until the very last minute (in fact some of them are still at it), rejecting Chomsky. Even the gracious and gentle McCawley couldn’t have thumbed his nose at the syntaxicentric Chomskyan vision any more explicitly than by choosing to call his work Unsyntax.
This man, Noam Chomsky, obviously, provides the mass. He gave Interpretive Semantics an irresistible center of gravity. Anyone in his immediate framework who begins working on a strand that is uncongenial to him, or even just uninteresting, rapidly becomes, by definition, out of the program—not necessarily for reasons of pique or malice, perhaps not even by design; simply because Chomsky’s gaze automatically fixes the prospect for the community. If he supports research that involves his framework, as he supported Katz and Postal’s pre-Aspects proposals about kernel semantics, it is in the program. If he stops talking about someone’s work, as he largely stopped talking about Katz’s semantics in the post-Aspects period, or if he rejects someone’s work, as he rejected Postal’s Abstract Syntax in the post-Aspects period, it is out of the program.
Katz certainly promoted an interpretive brand of semantics and expended a good deal of energy doing battle against Chomsky’s principal enemies, but he was not really, in the usage that best suits the pattern of the debate, an Interpretivist. Linguists soon realized that Katz no more spoke for Chomsky than did Lakoff or Ross. With Postal, the case is even clearer. He was (and remains) a Chomskyan linguist in the most pristine, originary sense of the term, interested primarily in building “precisely constructed models for linguistic structure” (Chomsky 1957a:5), and he was the driving force behind research that almost everyone thought would have been dear to Chomsky’s heart, seeking out deep lexical regularities and explaining surface diversity via transformations. But Chomsky rejected the work and Postal rapidly moved into the role of opponent.
There was some of this in-the-movement, out-of-the-movement categorization to Generative Semantics as well; Newmeyer, in particular, felt ostracized when he rejected McCawley’s VSO proposal, and George Lakoff sullenly insists that “Newmeyer was never a Generative Semanticist.” But it was much less prevalent, and much less obvious. The major reason for this difference is probably that, as much as Lakoff may have lobbied for the role, Generative Semantics was not organized around a single individual. All five of the horsefolk had their own interests, and collectively their range was vastly broader than Chomsky’s. But the movement was also inherently more open and anarchic than Interpretivism (in fact, than any of Chomsky’s frameworks). If someone with Generative Semantics sympathies began working on a particular data set or formal mechanism, it was usually assumed that they took the theory with them. Sadock examined the syntax, semantics, and pragmatics of medicine bottle labels (1974b). Green looked at proverbs (1975). Ross looked at his squishes (1972d). Lakoff looked at Rosch’s category work, Zadeh’s fuzzy set theory, Grice’s conversational maxims. And, automatically, the phenomena or principles were under the Generative Semantics tent.
I think the argument will be most fairly tested, if we take the “if” out of it.
—Socrates (Plato, Protagoras 331c)
Chomsky regards the death of Generative Semantics, in its sociological dimensions, as largely the implosion of ethos. “They destroyed themselves by insisting on taking on the mannerisms, the style, of one of the crazier sixties cults,” he says. There is truth here, and it is difficult to avoid analogies with the youth-culture-cum-counterculture-cum-Woodstock-Nation, which changed its T-shirt slogans recurrently. The issues of hippiedom were real, and many people were committed, and worked very hard to bring about social and political changes of considerable magnitude. But the majority of people were committed, if at all, for only a short period. The issues were real, but they were also faddish, falling out of favor with long hair and beads. I certainly have no particular insights as to why they fell out of favor, although much of their transience surely had to do with the complete lack of focus in (and, in many cases, outright lack of) positive alternatives to the institutions and practices under attack; the counterculture was, says Roszak in the thick of the flight, “much more a flight from than toward” (1969:34). It’s in the name. Counterculture. Even what was being rejected grew rapidly, from a fairly tight set of concerns about civil rights inequities and the invasion of Vietnam, into a great amorphous stew of beliefs, objects, people, culture. The same trends are apparent in Generative Semantics: the flight from Chomsky; the rapid expansion of dissent from a few specific technical questions; the ever-widening range of phenomena to be embraced. But these are very deep reasons for failure, utterly distinct from a few epithets, some absurdist humor, and a political jab or two. All of these traits contributed in some measure to the decline of Generative Semantics, but it was the Generative Semanticists’ remarkable frankness about the shortcomings of the theory, and their tendency to celebrate rather than marginalize anomalies, both reflections of deep methodological and philosophical positions, that truly marked the movement [+doom].
Much of the Generative Semantics style is, unfortunately, exclusionary. Rubbing hostile scholars the wrong way is not, or should not be, much of a concern. Their predispositions would be satisfied no matter what style they encountered, and, in any case, rubbing them the wrong way is always part of the point. The problem is that neutral scholars were simply discouraged from reading material they found offensive for reasons other than its linguistic claims—because of general irreverence, or allusions to drugs, or aspersions on public figures. The Venerable Quine, for instance, seems only to recall that G. Lakoff and McCawley “were over-eager to amuse . . . with whimsical examples” at a 1969 conference on logic and language, complaining of their “chatter and clowning” (1985:358). Scholars come in all persuasions, but a trait common to many is pricklishness. Aggravation is not the road to adherence.
Even these traits, however—humor, drug allusions, political aspersions—might not have done much damage in isolation. In fact, they contributed to the movement’s attractiveness for enough young scholars that the annoyance of others might have been counterbalanced. The real damage was done by the closely related traits that George Lakoff calls honesty and data-love.
Although the term honesty is obviously self-serving, and perhaps should be replaced with one like frankness, the Generative Semanticists were explicit about their shortcomings in a way very few scientific collectives are.11 They were, in their own self-image, at pains to avoid what they took to be Chomsky’s rhetorical underhandedness.
This frankness had deleterious effects in many quarters. Frequent reminders that “the author is fully aware of the fact that there are many uncertainties, unclarities, and errors in the text as presented here” (Seuren 1975:84) undermines the audience’s confidence that the rhetor is reliable about anything. No one, on any side of the dispute, would deny that every transformational work of the period was full of uncertainties, unclarities, and errors, and most linguists acknowledged them in some way. Here is Chomsky at the end of “Some Empirical Issues”:
Very roughly, this seems to me a fair assessment of the state of the theory of transformational generative grammar—at the moment. Obviously, any such assessment must be quite tentative and imprecise at crucial points. I will be very surprised if in a similar review several years from now, or perhaps next week, I will not want to present a rather different picture—surprised, and not a little disappointed as well. (1972b [1969]:199)
But the incidence of such admissions, and the general tone of discussion, are vastly different in Chomsky and in the Generative Semanticists; neither he nor his followers ever penned a creature feature.
Chomsky has the impressive rhetorical talent of offering ideas which are at once tentative and endorsed, of appearing to take the if out of his argument while nevertheless keeping it safely around should it prove necessary later on. Few of the Generative Semanticists had this talent, with Ross’s work, especially in the 1970s, marking the antithetical inclination. It contains the “embarrassing candor and intense emotional involvement” that Holton says marginalized Kepler in the history of science (1988 [1973]:54). Ross’s arguments are not as wild as Kepler’s, but they are remarkable all the same—multilayered, vinous, and subtle, very sensitive to fluctuations in the data. They contain dozens of threads, looping around one another in the main text and annotated with lengthy, contorted discussions, some of which offer counterexamples, some of which offer alternative analyses, few of which offer any support for the point that triggered the digression. They contain long catalogs of subarguments—some strong, some weak, the weak ones always painstakingly flagged as such—interspersed with declarations of mystification and awe.
And, of course, there was always work in the wings, apparently containing more such claims, calibrations, and contra-considerations; the current tangle of data and instruments was just the tip of the iceberg. Here, for instance, are the last five entries under “Ross, John Robert” in the works-cited section of one paper:
These promissory notes seem to refer to various jottings for conference presentations, or just claims made in the course of a talk, ideas or arguments or provocative data-clusters that intrigued him for a while but apparently not enough to ever reach publication (I can find no evidence that they were ever published, at least under these titles). Ross is, in Parmenides’s phrase, one of “the men with two heads” (DK28B6), someone who has taken dialectic to almost pathological levels; Terry Langendoen is said to have summarized one of Ross’s talks with “It’s fifteen arguments for us and nine for them; so I guess we win.”
If honest is the appropriate descriptor for Ross’s style, and it does seem right, it is painfully honest. He is an adventure to read, an earnest and good-natured tone leading us into a wilderness, but the overall effect is confessional, in a way that perhaps makes one empathetic to the difficulties of his program, but hardly motivates one to join him. For all his fraternal good will, the effect can also be discordant and chaotic, the rhetorical analog of Brownian motion, argument jostling argument, data jamming into data, until the challenge to find a clear line of thought rising above the bustle becomes defeating. “It must be seriously open to doubt whether there is a coherent point of view to communicate,” W. R. O’Donnell said of the Generative Semanticists’ work overall:
Add . . . that some of them appear to change their minds almost continuously and that they are addicted to somewhat tendentious publication of views they no longer hold at the time of publication and you have a recipe for an intellectual confusion which might daunt even the most committed seeker-after-truth. (O’Donnell 1974:74)
O’Donnell’s disgruntlement was not uncommon; nor was his theoretical and methodological response to the data-celebration papers issuing from the Generative Semantics camp, silence. What could a working linguist do with a paper full of facts whose raison d’etre was that there was no conceivable explanation for them in current theory?
It’s not like they were alone, either, in their sense that linguistic mysteries abounded. Here’s Chomsky:
I would like to distinguish roughly between two kinds of issues that arise in the study of language and mind: those that appear to be within the reach of approaches and concepts that are moderately well understood—what I will call “problems”; and others that remain as obscure to us today as when they were originally formulated—what I will call “mysteries.” . . . [About mysteries:] although there is much that we can say as human beings with intuition and insight, there is little, I believe, that we can say as scientists. . . . Some would reject this evaluation of the state of our understanding. I do not propose to argue the point here, but rather to turn to the problems that do seem to me amenable to [scientific] inquiry. (1975b:137, 138–39)
The contrast with leading Generative Semanticists could not be more dramatic:
Her [Borkin’s] examples are dazzling and remain deep mysteries to this day. Such mysteries are central to our vision of what problems should be addressed by the linguistics of the 1980s and beyond. We are grateful for the gift of mysteries so worthy of the attention of those who would understand how language works. (G. Lakoff & Ross, in Borkin 1984:viii)
The Generative Semanticists celebrated mysteries, Chomsky avoided them, and the implications of these two strategies for working linguists are clear. Lakoff and Ross’s program offers them a chance to work on dazzling data that promises to remain a deep mystery, perhaps for centuries to come: Chomsky’s offers them a chance to work on amenable problems. The choice, for most of the field, was clear.
It’s one thing putting away the past, and quite another to tape its mouth shut.
—Liam Lacey (1991:C1)
“Will [the victorious] group ever say that the result of its victory has been something less than progress?” Thomas Kuhn asks (2012 [1962]:165). Perhaps Michael Brame, whose list of Generative Semantics’ Fatal Flaws is lengthy, can address the question for us:
1. Generative Semantics is not sufficiently explicit (no rules, no explicit underlying structures, no formalization of global rules, no characterization of semantic predicates) to provide an unambiguous basis for comparison with current explicit theories. . . .
2. Generative Semantics is a breeding ground for syntactic irregularity as a consequence of its exception devices, unnaturally remote underlying representations, its abolition of deep structure and incorporation of semantic predicates and global devices and its blurring of the distinction between the lexicon and transformations as well as between syntax and semantics. . . .
3. Generative Semantics is prescientific as a consequence, in part, of its incorporation of semantic predicates and global mechanisms. . . .
4. Generative Semantics is anti-abstractionist inasmuch as it refuses to stray from what it alleges to be the phonetic or semantic given . . .
5. Generative Semantics gives up the quest for universals . . .
So, yeah. Victory is progress. Oh, whoops. Sorry. I interrupted. He’s not quite through:
In addition to these points, I think it can be accurately said that Generative Semantics fails on almost every single proposal or suggestion for the analysis of a fragment of English grammar that it has advanced. (Brame 1976:67)
Nor does he quite end there—Brame still has another Four Fatal Faws (68) to enumerate—but, thanks Michael, we get the picture. The victory of Interpretive Semantics was clear-headed and righteous progress, wresting linguistics from the hands of a deluded and pre-/non-/unscientific mob of data-fetishists.
Brame’s list, in the present tense, reads like a diagnosis of some terminal disease, but really it’s an autopsy. By the time Brame’s book came out, “Generative Semantics was essentially dead and largely discredited” (Gazdar 1982:465).
All of this took place a long way from the pristine, disinterested thoughts of Noam Chomsky, in his telling. “I never paid attention to the ‘linguistic wars,’ ” he says recurrently. It is certainly true, in one central premise of his story, that “while the battle raged” over Interpretive and Generative Semantics, “Chomsky reached ‘the peak’ of his antiwar activity,” and therefore had only little bursts time to concern himself with these issues (Barsky 1997a:151). Chomsky concluded one letter to McCawley from the period,
I had expected to come to Chicago, but haven’t got the energy for it. I’ve been spending an enormous amount of time speaking, writing, getting arrested, etc., and can’t work up much enthusiasm for linguistics, I’m sorry to say. Maybe someday the world will return to relative sanity, and we’ll have a chance to talk. (Chomsky to McCawley, December 20, 1967)
He was busy. There were more important things. But those premises do not support the conclusion that Chomsky was aloof from the battles, and it is curious that Barsky, who rightly lauds Chomsky’s extraordinary energy, stamina, and intellectual capacity throughout his biography—his ability to accomplish tremendous amounts in compartmentalized bursts of energy—would think Chomsky incapable of making the best of his time for attacking a movement he so clearly loathes.
Chomsky’s account just does not stand up. A few people endorse it, confederates all. Many others, including the people who have looked at the clash most fully, offer accounts rather more out of sync with Chomsky’s.12 Your humble narrator has read the primary and secondary literature in some detail and has had the privilege of talking and corresponding to many linguists, including several who were present at Chomsky’s lectures though the mid-to-late 1960s and early 1970s, and a few who worked closely with Chomsky before and after the clash: none of them endorses Chomsky’s version. Of the published accounts that support Chomsky’s version, none of the authors had a closer alliance with Chomsky over the relevant period than Ray Jackendoff, whose version does not find Chomsky sitting disinterested in his office or giving temperate and neutral lectures, away from the fray. Here is Jackendoff’s précis of the Generative/Interpretive Semantics clash:
All the people who admired Aspects for what it said about meaning loved Generative Semantics, and it swept the country. But Chomsky himself reacted negatively, and with the aid of his then-current students (full disclosure: present author included), argued vigorously against Generative Semantics. When the dust of the ensuing “Linguistics Wars” cleared around 1973, Chomsky had won—but with a twist: he no longer claimed that Deep Structure was the sole level that determines meaning. Then, having won the battle, he turned his attention not to meaning, but to relatively technical constraints on movement transformations. (2003:654)
Chomsky may not have been hurling obscenities from the dais along with the frontline forces, but he fashioned the ammunition, inspirited the troops, and if he did not issue marching orders, he certainly insinuated them.
Victory for the generation of Chomskyan linguists coming on stream in the 1970s was not only necessary progress for linguistics. It was a holy mission. Brame’s belief in the utter and irremediable vacuity of Generative Semantics was common, encapsulated in a remark Brame quotes from the eponymous Chomskyan that a Generative Semantics victory in the dispute would mean “the virtual demise of syntax (Chomsky, personal communication, August 13, 1975)” (Brame 1976:26)—nothing being more precious to Chomsky and his students than syntax.
By the mid-1970s Generative Semanticists were pretty much imaginary beasts—snarks, let’s call them—but they were hunted all the same. Generative-Semantics-bashing articles were the order of the day. The arguments all ran from the specific to the general. They were not “proposal x from linguist y is wrong” so much as “Generative Semantics is totally bankrupt, unscientific, and vapid, which is illustrated by the failures of proposal x, from the misguided y.” Categorical dismissals of this sort are direct reflections of Chomsky’s style, especially in lecture and conversational formats, where he sweepingly dismisses entire lines of thought, even entire disciplines, with a few contemptuous adjectives. Many of these articles were self-conscious proclamations of the death of Generative Semantics, and they didn’t stop with counterarguments to one or two or three Generative Semantic positions. They reeled off long catalogs of failures—driving nails into the coffin, chucking it into a grave, shoveling dirt over it, and erecting tombstones with inscriptions like Brame’s “Final Verdict” (1976:67).
Brame’s (1976) Conjectures and Refutations in Syntax and Semantics is the acme of the apogee of the trend. It rarely matches the slathering hostility of Dougherty’s diatribes (which mostly date from this period), but it is irredeemably snide. One section, for instance, begins with a quotation contrasting Einstein to an amoeba, “which cannot be critical vis-à-vis its expectations and hypotheses,” followed by one from Lakoff which Brame intimates exhibits this amoeba-like incapacity (1976:3).13 In other places he resorts to academic priggeries, like slipping a sic snidely into a Generative Semanticist quotation (“if we adopt this explanation [sic]”—1976:58) and calling global rules a “theoretical prophylactic” (1976:45). Elsewhere, he simply insults the work directly, as when he calls one of Lakoff’s analyses “no more than a tortuous description of uninteresting facts” (1976:15). Green’s appraisal of Brame’s book—“so unreasoned and prejudicial a treatise (I refrain from calling it a tract)” (1981:704)—is positively tame in the face of its systematic, symptomatic contempt.
The attacks and obituaries from the new Interpretivist crop were bolstered by several assaults from older loyalists, some with allegiances to Chomsky generally, others with specific allegiances to his 1965 position. Katz, of course, was at the head of this contingent (1970, 1971, 1972a, 1976), but he was joined by Bever (Katz & Bever 1976 [1974]) and Stockwell (1977:131), among others. Ostensibly nonaligned linguists also began to chime in (Kuiper 1975; O’Donnell 1974; Sinha 1977b). Chomsky did not address Generative Semantics in his theoretical work, except rather obliquely (for instance, in a footnote which does not so much as name one of its proponents; 1975b:238n2), but continued for decades to dismiss it in more informal settings, particularly in interviews. The fountainhead of Chomskyan linguistics, Logical Structure of Linguistic Theory, was finally published in this period (1976 [1955–1956]). The book—pre-dating Deep Structure and any overt attention to semantics—articulates a program that meshed fairly well with his post-Aspects Interpretivism, so the publication date nicely cements his triumph over Generative Semantics, with Chomsky’s introduction including a few choice and pointed remarks against the framework.14 Some disillusioned Generative Semanticists even threw in a few handfuls of dirt, as in Sadock’s “The Soft, Interpretive Underbelly of Generative Semantics” (1975).
Among the most consistent aspects of these publications is the recurrent targeting of George Lakoff, clearly propagating an attitude fostered in Chomsky’s classes. It’s not that Chomsky spared others, but he reserved a special animus for Lakoff.
The scapegoating of Lakoff in the 1970s offensive against Generative Semantics had two quite contrary lines of attack. On the one hand, it was often implied that Lakoff was not responsible in any significant way for the beginnings of Generative Semantics. It was “really” Postal’s theory, who had developed some pregnant suggestions from Chomsky. On the other hand, Lakoff was made to shoulder virtually all the blame for its excesses, largely on the basis of the sin of proposing global rules. Insofar as Generative Semantics had been a natural development exploring a promising hypothesis, this was due to Postal and to Katz and ultimately to Chomsky; insofar as it was wrong, unscientific, profligate, sloppy, and generally bonkers, Lakoff was to blame. The implication that Lakoff was little more than a loudmouthed opportunist, who perverted a potentially useful research program, has a number of underlying motives, but much of it surely stems from Chomsky’s apparently personal disregard for Lakoff.
The declaration of Lakoff as Public Enemy Number One has its roots in Chomsky’s “Remarks,” where he is set up as the chief proponent of the Transformationalist position, and strapping the sins of Generative Semantics to his back continues as a prominent theme through all of Chomsky’s anti-Generative-Semantics papers; by “Some Empirical Issues” Lakoff is a complete whipping boy, and “Lakoff’s revised Generative Semantics” (Chomsky 1972b [1969]:140)—that is, Homogeneous II—is the nadir of linguistics, the worst possible theory. Even after the shelling had pretty much stopped, for instance, Chomsky took up some of Lakoff’s criticisms in a catalogue of ad personams spotlighting his clear and utter failure of intellect (1980a:46):
[Lakoff’s] remarks betray a very serious misunderstanding. [He] shows no awareness [of important issues].
Lakoff seems totally unaware of the actual character of the technical work to which he refers.
[The semantic work of Interpretive Semantics is] a matter Lakoff has never understood.
Lakoff’s misunderstanding of the technical work is so far-reaching that his comments on it are completely irrelevant.
Lakoff shows no awareness of these issues.
Chomsky’s attitude toward Lakoff—like his demise-of-syntax characterization of the whole movement—surely permeated his discussions of Generative Semantics at MIT; certainly, it permeated many of his students.
For his part—don’t be too shocked—Lakoff is less than fond of Chomsky. In conversation, this disaffection takes the form of concerns about his politics, his honesty, and his ego. In print, Lakoff rarely foregoes the chance to attack a Chomskyan position, with enthusiasm. For instance, the anti-Lakovian ad personams I just quoted from Chomsky are in direct response to a critique in which Lakoff tears into Chomsky’s most sensitive area, calling him all talk and no science. “We are in the realm of rhetoric,” he says of Chomsky’s positions, “not science.” Much of Chomsky’s work, he goes on, is useless for practical purposes, “but as rhetoric, it is effective—at least so far as academic politics is concerned.” Chomsky is particularly guilty for having “artfully chosen” some of his terms, an accusation Chomsky finds deeply repugnant (G. Lakoff 1980:243).15
There is certainly no question—whatever Chomsky’s loathing for the observation might be—that he is a tremendously gifted, if less than scrupulous, rhetor. He is not a uniformly impressive prose stylist or an especially incendiary orator. His writing can be as dense, gnarled, and forbidding as a blackberry patch, full of fruit you can see but you just can’t get to—though Chomsky can also reach moments of great persuasive lucidity. But he is very effective orally, in lecture or in interview, and many of his books are from interviews or conversations that have been transcribed and edited. Even Syntactic Structures was to some degree originally intended for oral delivery; Aspects, too, was drawn from lecture notes. And of course, most of the anti-Generative-Semanticist troops that beset the discipline in the mid-to-late 1970s had heard the gospel first-hand at MIT. So it is no coincidence that Chomsky’s version of the Generative Semantics story—that it was disconfirmed early, but its practitioners absurdly clung to it anyway by changing it into the worst possible theory, whereupon they were driven from the field for irrationality and error, and let’s not forget that George Lakoff was biggest ringleader but also not very smart—became the received view for that generation. His opinions tend to stick fast in the minds of his audience.
Chomsky, in short, cannot be let off the hook for the war itself or for the venomous quality of the anti-Generative-Semantics campaign. That he neither accepts responsibility for the schism, nor even acknowledges that he played any but the most cursory, one-day role in it (e.g., Barsky 1997a:151), like much of his self-representation, does not look good on him. He was the principal architect of the Aspects framework that was the launching pad for the dispute. He was, just ten years after Syntactic Structures and not yet forty, the most famous linguist in the world. He was, unquestionably, the one who set the MIT linguistics agenda, if not the American linguistics agenda. Virtually everyone else relevant to the program he engineered, with a few very limited exceptions (Bach, Fillmore, Halle, Klima, Lees), was a junior professor in 1967, or a student.
Chomsky had a rhetorical choice. David Raup is surely right that “new ideas breed disagreement” (1986:150), but disagreement does not require malice. If Chomsky couldn’t be encouraging, he might at least have been tolerant of the developing trends. His argumentation might have been open, if not fully welcoming, especially given the retrospectively visible tentativeness of a several key themes in Aspects, the newly issued scripture, and especially given many of the paths his own work has later taken. He might have let a hundred—even two or three—flowers bloom. But just as he chose to adopt a revolutionary stance to the Bloomfieldians in the late 1950s, he chose to adopt a divisive stance in the post-Aspects climate of the latter 1960s, open to some developments, closed, locked, and bolted against others, providing the boiling oil to be poured down on them from the parapets.
Chomsky set the battle lines. He likes it that way. He works best, apparently, when there is an enemy abroad. George Lakoff had a choice, too, of course; if Chomsky’s dismissals framed one set of attitudes in the clash, Lakoff’s brash responses and counterattacks, in print but also in Chomsky’s own lectures, certainly framed another set of attitudes. Robin Lakoff had a choice, and Postal, and Ross, and McCawley; and they all made those choices, in different ways, to different effects. Individually, some of them provoked; collectively, they certainly fought back. But it was Chomsky’s war, from the beginning, and whenever I hear his account of the Linguistics Wars, it recalls that iconic moment from the first season of The Simpsons, when Bart, eyes shifting guiltily, says “I didn’t do it. Nobody saw me do it. You can’t prove anything” (Jean & Reiss 1990:6:04).