blogpost: Post-minimalism?

 Posted by on 02/16/2020  Add comments
Feb 162020
 

I was recently invited to contribute a short piece about Agree to the This Or That Publisher’s Handbook of Minimalism, and it made me wonder to what extent I, or really most other generative syntacticians who got their PhDs after about 2005 or so, can be considered to still be doing “minimalism.”

In one sense: this doesn’t really matter, who cares about labels, we all contain multitudes, yada yada yada.

But in this case, I actually think there is something substantive to this beyond the label. In particular, I think we are no longer really in the “minimalist” era anymore. It’s just that the demarcation of generative-syntax epochs has traditionally been led by Chomsky himself. And Chomsky himself is off doing… well, more on that below.

Let’s start with some clarifications. There are at least four ways the terms “minimalism” or “minimalist” get thrown around as it pertains to syntactic theory. One, the original one really, which traces its origin to Chomsky’s 1993 paper / 1995 book, is a substantive hypothesis about natural language. Roughly: that natural language is nothing but Merge, plus whatever the non-language-specific influences are of general cognition, sensory-motor systems, conceptual-intentional systems, and even laws of physics (a.k.a. the Strong Minimalist Thesis). In the service of trying to make this more than a completely-empty speculation, Chomsky then laid out (in chapter 4 of the 1995 book, but more than that, in the two subsequent papers Minimalist Inquiries, 2000, and Derivation by Phase, 2001), the general contours of what a theory that adhered to this hypothesis might look like. It was only a sketch, of course, buried in the usual menagerie of plausible-deniability hedges (“It’s not a theory, it’s a program!”). In going from speculation to a sketch of substantive theory, Chomsky was forced to make certain choices/commitments. Some turned out really well (e.g. probe-goal). Others, not so much (e.g. uninterpretable features). Regardless of how well they turned out, though, this collection of substantive choices leads to the second way the term “minimalist” is now used, namely to refer to syntactic theories that adopt most or all of these substantive choices. So, for example, anyone doing some version of a probe-goal syntactic theory nowadays (myself included) will usually get lumped in as “minimalist,” though as I will discuss shortly, it is pretty straightforward to imagine a probe-goal approach to syntax that eschews all but the most trivial minimalist commitments.

The third way the term “minimalist” gets bandied about is as a very particular computational-linguistic construct, namely the Minimalist Grammars formalism developed by Ed Stabler and his collaborators and students. This is a welcome effort to computationally formalize the collection of substantive theoretical choices that goes under “minimalism (sense 2)”; but for those (many) of us who think the object so formalized contains as many bad choices (e.g. a checking-based feature calculus, rather than valuation-based one) as good ones, this computational object is of limited interest to the Practicing Theoretical Linguist, Non-Computational Division.

The fourth (and as far as I can tell, final) way the terms “minimalism” or “minimalist” get used is the most general and, consequently, the most trivial. It essentially boils down to the methodological heuristic of “less is better,” a.k.a. Occam’s razor. Chomsky’s Minimalist Program has put a new spin on this age-old methodological heuristic, in the form of a concern for evolutionary plausibility (viz. the less we put into UG, the easier it is to envision it evolving in the species and being phenotypically more or less uniform; a.k.a. Darwin’s Problem). But in practice, I see no effective difference between this new variant and Occam’s more venerable version, at least for how I go about doing theoretical linguistics. More is worse, less is better. This is old news.

Let’s summarize so far. The third, Stablerian sense of “minimalist” (as in Minimalist Grammars) is of limited relevance beyond comp-ling. The fourth sense is so bland that it amounts to little more than “Do good science.” This leaves us with sense 1 (as in the Strong Minimalist Thesis), and sense 2 (as in, a shorthand for adopting a significant subset of the substantive choices made by Chomsky in his first few minimalist papers).

So, let’s see: is someone like me a “minimalist” in either of these two remaining senses? (Of course, if this were really about me, it would be of extremely limited interest. As I’ll discuss shortly, I don’t really think this is unique to me at all. But I am the one whose views I know with the greatest certainty, so I’ll start there.) I certainly don’t think the Strong Minimalist Thesis is right, and am on record with an argument to that effect. So that’s out. And I’m far from unique on that front: I was speaking recently with a very prominent morphosyntactician about that linked-to paper (someone of my generation, trained in one of the very top departments for generative syntax), and their response was something like, “Yeah, I’ve never really cared what Chomsky has to say about uninterpretability and ‘crashes’. I’ve basically always assumed things work like [what this paper says], so I really don’t see what the fuss is about.”

And what of the substantive choices that go under “minimalism (sense 2)”? Let me take inventory of my own views. The success of the probe-goal model, in my opinion, has been spectacular. As for Phase Theory, it is mostly Subjacency/Barriers leftovers, thrown in the microwave and reheated. And the evidence for it has been seriously overstated. (That’s not to say that there doesn’t remain some serious evidence for phases, as the comments to the linked-to post make clear.) Uninterpretability, as anyone vaguely familiar with my work knows, was a nice try but ultimately failed. It was influenced, I think, by ideas like the Case Filter, which themselves have turned out to be wrong. Feature-inheritance? Again a nice try, again doesn’t work (see Haegeman & van Koppen 2012, among many others). And as for the recent “labeling” stuff (Chomsky’s Problems of Projection and related papers), I’ll quote Jason Merchant’s comment on FoL:

[…] But no C-I requirement can plausibly tell us that “angry at”, “proud of”, “interested in”, etc pair they way they do. We need l-selection [category-dependent lexical-selection; O.P.], which means we need labels that are at least as fine-tuned as distinguishing “at” from “of”, “in” etc requires. These relations are fundamentally *syntactic*, and so any theory of syntax that claims that these relations can be captured by or at the C-I interface has an uphill battle for the hearts and minds. […]

Jason is being kind/gentle here. Pesetsky and others established way back in the 80s (and beyond any reasonable doubt, I think) that c-selection cannot be reduced to semantics, and so an approach to labeling that situates it at the “C(onceptual)-I(ntentional) interface” is nothing short of laughable, imo.

So the question is: does one really count as “minimalist” because one uses probe-goal and, for lack of a better available proposal, some version of Phase Theory, but eschews all other minimalist technology and doesn’t really believe in the Strong Minimalist Thesis?

And here’s the thing: I think something roughly similar can be said of the overwhelming majority of allegedly “minimalist” syntacticians of the current generation (which I have arbitrarily and capriciously demarcated as getting your PhD post-2005). There are exceptions: Dan Milway and Marc Richards spring to mind, among others. But really, I think I can count the people that are exceptions on one hand. (Marc’s PhD is actually from 2004, if I remember correctly.)

If we track the development of this particular branch of generative syntax from the various “((Revised) Extended) Standard Theory” periods, through the “Government & Binding” period and the “Minimalist” period, isn’t it fairly clear that we are now in a distinct period? This isn’t an issue of labels per se, but of a difference in the substantive set of assumptions that currently lie at the core of our endeavor. (To be clear, I think that “post-minimalism” is a rather shitty label; I consider it more of a placeholder than a real candidate for what this should be called.)


Postscript

When I say things like “{most generative syntacticians, the overwhelming majority of allegedly ‘minimalist’ syntacticians} of the current generation,” I’m really talking about people doing morphosyntax. This is partly a matter of familiarity, as this is the literature that I read the most. I can’t say with confidence that the same things hold of what’s going on on the LF side. But this may not be just a matter of familiarity. Chomsky used to pay close attention to morphosyntactic matters (look no further than the treatment of the English auxiliary system in LSLT / Syntactic Structures, but also, e.g., Remarks on Nominalization). But for my money, his work starting from the Minimalist Program (if not earlier) is characterized by a cavalier, almost dismissive attitude towards even the most basic issues of morphosyntax, and a nearly monomaniacal tendency to equate syntax with LF. This is coupled with dismissive remarks about “externalization” as a cover-all for morphosyntax (and/or an attempt to sweep the severe morphosyntactic problems with his proposals under the rug). It’s probably no coincidence, then, that the late 90s / early 2000s were, at least in some circles, the heyday of LF syntax. (Disclaimer: if you thought the discussion above was parochial, I’m about to delve into some extremely localist opinionizing. Equip yourself with a gigantic grain of salt, please.) From my vantage point as a late-aughts graduate student in the MIT syntax program, for example, the generation before us was dominated by people doing LF syntax (Fox, Hackl, Nissenbaum). There were exceptions, as there are bound to be (Wurmbrand, Legate), but one can still identify something of a sea change, I think, occurring right after that time. I was strictly in the riding-the-coattails category, mind you. To the extent that there were specific individuals whose influence on the syntax grad-student body led to this sea change, I would single out Jessica Coon and Claire Halpert. So, in addition to Coon and Halpert (and myself), you suddenly have Bjorkman, Levin, van Urk, Yuan, and many other morphosyntax theses. (This impression may fall apart under careful quantitative scrutiny; I’m expressing what was my subjective impression living through this period.) I’ve certainly brought with me to UMD my belief that morphosyntax is the “main event” when it comes to syntactic theory, and I feel like my colleague Maria Polinsky shares this view (though please don’t hold her responsible for what I write here). At UMD, this is well-balanced by the views of faculty members like Alexander Williams, Valentine Hacquard, Howard Lasnik, and Norbert Hornstein, who can all be described as LF-centric to varying degrees. But regardless, I think it’s noteworthy that as “core minimalism” drifts ever more towards what is essentially a theory of meaning, with fewer and fewer commitments to form (“externalization”), its ability to contend with even the most basic facts about morphosyntax has degraded to the point that I, for one, find it hard to still consider myself a “minimalist” at all.

avatar
6 Comment authors
PhilipJohannesStanlyThomas GrafJason Merchant Recent comment authors

  Subscribe  
Notify of
Thomas Graf
Guest
Thomas Graf

I’m really curious what you think MGs lack in the faithfulness department, because it’s certainly not feature valuation. That can be added without altering the essential properties of the formalism. We can also remove the distinction between Move and Merge features, or make the grammar entirely feature free. And we can add various types of movement, constraints, DM, even economy conditions and grafts. There’s very few things that greatly affect the computational properties, one example… Read more »

Stanly
Guest

Well, what about phonology, semantics, and pragmatics and language-to-thought and thought-to-language interfaces. And gestures and sign languages? Duuhhh.

Johannes
Guest
Johannes

Interesting blog post! But could you explain in a bit more detail what you mean by Chomsky’s ‘LF-syntax’? For all (the very few things) I know, Chomsky nowadays is busy talking about workspaces, and how search for external merge is more computationally costly than internal merge etc. pp… that seems to be very syntax-syntax to me (actually, it doesn’t sound like anything to me, but anyway…), which seems different from the late-merging type (e,t) predicates… Read more »

Philip
Guest

‘The third, Stablerian sense of “minimalist” (as in Minimalist Grammars) is of limited relevance beyond comp-ling.‘ This strikes me only as being the case if you are committed to the idea that explanations at Marr’s computational level need never (or at least for the foreseeable future) link up with explanations at the algorithmic/representational level. Which is a perfectly viable position to take, of course, but if that’s the case what is the term “computationally costly”… Read more »