May 192022

As I promised over here, I will be blogging at some yet-to-be-determined frequency about some of the problems endemic to academic linguistics that I’ve seen during my time in the profession. These ills, I think, are arranged in something like concentric circles: some pertain specifically to generative syntacticians; others pertain to generative linguistics in general; others pertain to linguists of any stripe; others pertain to academics of any kind working at public/state institutions; and yet others pertain to academia in general. My tentative plan is to chart a path through these concentric circles, proceeding “from the inside outwards.”

Today, I’ll be talking about generative syntax in academia.

tl;dr — With (very) few exceptions, generative syntax in academia doesn’t exist, not even among those departments that profess to be generative or generative-friendly. There’s morphology, and there’s semantics-masquerading-as-syntax, and that’s about it.

This may seem like a radical or extreme claim, but it’s really not. Two trends in generative linguistics in the 90s combined to basically drive bona fide syntax into extinction. The first is Chomsky’s Minimalist Program. As detailed here, Chomsky’s explicit assumption was that there’s nothing interesting going on in syntax itself. There was only Merge, the putting together of two objects to form a new, complex object. Everything else that was of any interest was happening at the interface between syntax and semantics or at the interface between syntax and morpho-phonology.1 But even that interface-situated “everything else” could not be modeled using syntactic tools proper, since, by hypothesis, there was no such thing as syntactic tools. (Merge is conceptually monolithic; it cannot be parametrized, attenuated, or modified.) So any actual theoretical linguistic work of any interest was forced, again by hypothesis, to take place outside of syntax and within the modules that syntax had an interface with.

The second relevant trend is the rise of the Partee-Heim-Kratzer school of formal semantics. Among the distinguishing features of this school is that it took seriously (some of) the results of generative syntax, and worked towards a semantics that was compositional relative to a structure that an actual syntactician might recognize as plausible. (As opposed to some of the other branches of formal semantics, which were and are syntax-denialists to varying degrees, up to and including treating strings as an object of the grammatical ontology that semantics can see – a position which, at this point in linguistic history, can only be described as willful ignorance.) To what extent this project was or is successful is, I think, something reasonable people can disagree on; in my own personal view, the results are mixed. All too often, the resulting semantic theory is “compositional” only by virtue of a significantly warped syntactic theory, populated with elements and structures whose only motivation is semantic in the first place, a move that then renders the compositionality part of the endeavor quite circular. But even if you disagree, and find the Partee-Heim-Kratzer project to be very successful on its own terms, its admixture with the Minimalist Program-driven trend described above created a compound whose properties were particularly toxic towards syntax proper.

Plainly put—and I think anyone looking objectively at the last 25 years of generative syntax will find this pretty hard to deny—generative syntax split into three parts. The first part consisted of those who bought into minimalism but still took seriously the Autonomy of Syntax, i.e., the decades of results showing that natural language abounds in phenomena that are not reducible to structured meaning, not even structured linguistic meaning.2 These people turned their attention to morphology, and in particular, towards syntacticized theories of morphology, such as Distributed Morphology and Nanosyntax, since this was the one corner of the system where there was still license to work on linguistic structure without having to view it through the prism of meaning.3

The second part consisted of those whose primary interests, I suspect, were in linguistic meaning all along, but who for one idiosyncratic reason or another had found themselves practitioners of syntax rather than semantics. For these folks, this development was a godsend: you could now go about doing research into linguistic meaning but phrase your findings in terms of specs and heads and XPs (often, most or all of which are morpho-phonologically undetectable), and call the result “syntactic theory.” (After all: given minimalist assumptions, what else besides meaning could be responsible for syntactic structures being what they are, anyway?) For reasons that I don’t pretend to understand, this was and remains the largest subgroup. Not unrelatedly, though, it would also come to be the most influential subgroup.

The third part consisted of syntacticians who simply rejected minimalism outright. I’m not talking here about recent PhDs who have parted ways with minimalist orthodoxy; I’m talking about schools of syntactic theory that never bought into it in the first place. Most notably, constraint-based lexicalist frameworks, like HPSG and LFG.4 Whether you think these frameworks are superior to minimalism or not, the fact is that academic jobs for constraint-based lexicalists (who don’t have one foot in another field, e.g. computer science) are even harder to find than jobs for minimalist syntacticians are.

As you may have noticed, this leaves precisely no room for people who were both ostensibly syntacticians in the Government‑&-Binding / Principles‑&-Parameters tradition, and who were actually doing syntax (as opposed to morphology or semantics).

These developments had a self-reinforcing effect on the job market for generative syntacticians. Obviously, syntacticians were never going to constitute the majority of voting faculty members on any syntax hire (even if they were a majority on the usually-smaller search committee, something which itself was not always the case). And now there was a green light, from no lesser authority than Chomsky himself (see above), to do things like hire a de facto semanticist for your so-called “syntax” position.5 And this trend ends up creating a feedback loop because, the next time the same department searches for a “syntax” hire, even one of the ostensible syntacticians on faculty will be someone whose interests lie elsewhere. Thus, the chances of someone interested in syntax proper landing a job in that department will now be even lower than they were before. (I’m not suggesting that people are generally incapable of rising above their own personal scientific interests and agendas when they participate in hiring decisions, only that this dynamic tips the scales further and further in one direction. And given that linguistics, like all academic fields, features a glut of qualified candidates for every position, things very often come down to “tiebreakers,” and that’s where factors like this reign supreme.)

I don’t write any of this to absolve myself of personal responsibility for my part in it. Obviously, just like linguistics in general could do a better job in its attempt to convince the broader cognitive-science community of its importance and of the significance of its findings, so too could syntax proper (and the syntacticians who still do it) have done a better job of convincing the rest of the linguistic community of its importance and of the significance of its findings. But one can look at the very limited success of the former endeavor for clues about the prospects of the latter endeavor, I think.

The sad truth is that I don’t see a real way out of this downward spiral. The only thing that could change this dynamic, as far as I can tell, is if departments that spend “syntax” lines on linguists who are syntacticians-in-name-only would be publicly shamed for their conduct. But that would be hugely damaging to the hired individuals, and these people are our friends, classmates, and colleagues. Plus, the field is so very small that fracturing it along any lines is often treated as a third rail. (E.g. because it undermines the broader fight-for-its-life that linguistics is engaged in, in the context of the broader cognitive-science community.)

I imagine some readers might be tempted to dismiss these claims as being born out of some bitterness on my part. This is, after all, being written on the heels of my decision to leave academia.6 But I assure you, these observations have been kicking around in my head in one form or another for over a decade now, including my entire time on the tenure track (both before and after tenure), and I have voiced them privately. It’s just that I was not at liberty, until now, to be as explicit and public about these things as I currently am. Besides: the question of whether these claims are born of bitterness is less important than the question of whether these claims are true. All that should really matter is whether this post describes the reality of the profession. And I believe it does.



  1. This hypothesis turned out to be false, in ways that I believe are rather easy to demonstrate (see, e.g., here). But this is a post about the socio-history of generative linguistics, not about the results of linguistic inquiry. 
  2. Here‘s a paper that is ostensibly about the Autonomy of Syntax but which, quite ironically in my opinion, whiffs on this important point: syntax is, by definition, autonomous even from linguistic meaning. That’s exactly what sets it apart from semantics in the first place. This logical definition of what syntax is leaves room for the hypothesis that syntax is virtually empty (i.e., that language has no structural properties that are independent of linguistic meaning). That is precisely the Minimalist Program’s central hypothesis. It’s just that the evidence has shown that this particular logical possibility does not pan out. 
  3. This is not a critique of the syntacticized approach to morphology, which I believe is not only correct but pretty much inescapable, given what we now know about morphology and syntax. What I’m pointing out here is that working on this slice of the empirical pie became the refuge of those individuals who still believed that there was more to language than spelling out lambda-expressions using phonemes. 
  4. As longtime readers know, my blogging is a “Categorial Grammar”-free zone. 😉 More seriously, Categorial Grammar has no real handle on the kind of case and agreement patterns one finds cross-linguistically; and, as detailed on the front page of this here site, that means, in my book, that it can’t lay claim to being a serious theory of natural language syntax at all. 
  5. It would be poor form for me to list examples of such hires here, but there is also no need: I’m fairly sure one can count the exceptions to this trend, over the last 10-15 years, on two hands. A good friend points out that among these exceptions (i.e., recent “syntax” jobs that were filled by actual syntacticians), nearly all of them (us) work on phi-feature agreement. I think this is true, and is, in its own way, far from optimal. But now we’re really talking about the arrangement of deck chairs on the Titanic, aren’t we. 
  6. Though the proximate cause for my departure from academia was my personal circumstances, I’d be lying if I said that the ever-worsening fate of the kind of work I care about most didn’t contribute to my reluctance to “stick around for another year or two” and see if an opportunity to address the aforementioned personal circumstances didn’t happen to pop up. From where I sit, generative linguistics has, when it comes to syntax, more or less lost the plot.