There might be something there here

If you're reading these words out of your own free will, I'm sure I don't have to explain to you that locative expressions such as here and there play a central role in the type of linguistic racket that I'm in, especially when they're used like the first instance in the title of this post. Not surprisingly, then, the time has come for me to venture into this red ocean and add some thoughts on the subject. I'd already dabbled in it—not in 'Nam of course—in a CGWS-talk in 2011 which I still have to write up as a paper—story of my professional life—but in this new paper I take a different approach. I focus on expletive data from one dialect of Dutch (my own) and try to make two points:

  1. Expletives behave exactly like regular subject pronouns in this dialect, in two ways. On the one hand, they show the same distinction between strong and deficient forms that subject pronouns do, and on the other, expletives can undergo pronominal doubling (and even tripling), like subjects, but unlike any other elements (including locative pronouns).
  2. Unlike what is often assumed and sometimes explicitly stated, it is not only the distal locative adverb (i.e. there) that can be used as an expletive. In this dialect, its proximate counterpart here can also display expletive-like behavior.

The paper is still very much work in progress (it's currently under review) and so any comments/thoughts/questions you might have would be most appreciated, but I think there might something there here. Plus, it has allowed me to continue my hobby of devising example sentences that, when judged by their glosses, look like there is no way they can be part of natural language. The winner this time is example number (64), which can be glossed as 'Here has there here here no-one with Jef talked.'

  1. As an aside, yes, I know that it's been ages since my last post. Turns out it's not easy to combine a heavy teaching semester with a vibrant blog.

  2. Obviously, the post-nominal element (modifier? suffix?) -like is pretty important here: it turns out that even in its use as an expletive, here doesn't lose it's locative meaning (unlike there); see the paper for details.

  3. Can you even say that in English?

  4. A previous favorite of mine can be found in my OUP-book, p.255, fn6: 'I think that that that that that must replace.'

amoebae are the new dinosaurs

Looks like I'll be going to Göttingen in early April, as the abstract Marjo van Koppen and I sent in for this year’s GLOW was accepted for a poster presentation. The abstract builds on earlier work Marjo and I did both separately and jointly, but (a) ties it together into a unified account, (b) provides a quantitative analysis of the data (along the lines of what I did with verb clusters), and (c) discovers some surprising correlations and anti-correlations among the phenomena we discuss. The title of the abstract, which—credit where credit is due—was Marjo's invention, basically says it all: a microparameter in a nanoparametric world. It's a play on words on the title of this excellent paper by Mark Baker, where he argues that it's not microparameters all the way down (or up?), but that some parameters by design have a broader reach or impact (and these are his macroparameters). Similarly, we claim that when you start looking at how microvariational phenomena pattern together, it's not just nanoparameters freely combining with one another; there too, some points of variation seem to target a higher level. The scale is quite different from Baker's paper, of course, but the principle is the same: he's saying that a diplodocus is bigger than a compsognathus, while we're comparing Chaos carolinense with Dicty the amoeba.

As an aside, it was funny how the reviews we got for this abstract reaffirmed my continued belief that no matter how good the intentions of the conference organizers or the quality of the material, having your abstract (or paper or project proposal or ...) accepted always remains something of a crap shoot. We received a total of 9 reviews, of which I suspect—on the grounds of having been a part of this committee before—that the final two were from the selection committee. Most of the reviews we got were informative and clear (though not always positive), one was even quite long and extremely helpful (and not hard to figure out who it was from). Reviewer number 5, however, merely wrote: "I find the paper weak." Now, just to be clear: I have absolutely no problem with people disagreeing with my work. On the contrary, some of the most informative and helpful exchanges I've had came from people who were very critical. It's just that this review is unhelpful on all fronts: it's of no use to the authors, who can't use the review as a basis for addressing the weaknesses in their paper, and it is also maximally uninformative for the selection committee, who don't know if this reviewer actually has a valid, critical point to make, or whether s/he hasn't even read the abstract and was just in a particularly bad mood or simply doesn't like the authors. All in all, I'd say that reviewing, in all of its shapes and forms, is another area that’s ripe for disruption.

  1. This is what you get when your four-year-old is obsessed with dinosaurs.

  2. This is what you get when you start googling for amoebae of different sizes.

  3. The abstract was anonymous of course, but it wasn't hard to guess who it was from.


Here's another pickle: we've all been taught in Syntax 101 that nominal constituents and pronouns have something in common that sets them apart from (embedded) clauses. In more technical terms: nominal constituents and pronouns are DPs, while clauses are CPs. Some languages, like French, even wear this likeness on their sleeve, by using the same lexical items for their determiners (la table) and their pronouns (Je la vois.)

So far so good. Now let's take a look at subcategorization, in the simple, straight-from-Wikipedia sense of "the ability/necessity for lexical items (usually verbs) to require/allow the presence and types of the syntactic arguments with which they co-occur". If the bipartition between CPs and DPs sketched above is on the right track, we'd expect the subcategorization schemas (schemata?) of verbs to group pronouns and nominal constituents together and distinct from clauses. More generally, all things being equal—but are they ever?—we'd expect there to be four types of verbs:

Types 2 and 4 should allow for both nominal constituents and pronouns as direct objects, while types 1 and 3 should allow for neither. As it turns out, English is the poster child for this hypothetical state of affairs:

Summing up, life is hunky-dory in the English-speaking parts the Subcategorizational Empire: pronouns and nominal constituents always work in tandem (either both present or both absent), and independently of sentential complements (all four cells of the logical space are filled). As it turns out, however, not everyone is so well-behaved. Let's now turn our attention to Dutch. At first sight, Dutch yields exactly the same four cells as English:

In addition to these four types, however, there is another:

These verbs—and others like them; this is a pretty common pattern—can be combined with pronouns and sentential complements, but not nominal constituents. Under the simple DP/CP-split sketched at the beginning of this post, this is quite mysterious. The generalization seems to be that there's no verb in Dutch—none that I could find anyway—that can select a clausal complement, but not a pronoun (which is why in order illustrate Type 3 (CPs but not DPs) before, I had to take recourse to a copular verb with an adjectival complement). At present I have no idea what is going on here: What does it mean for pronouns to pattern with clauses (in some contexts) in Dutch? Why does English differ from Dutch in this respect? Is this a difference in their pronominal or their verbal system (or both)? I also don't yet know if/how this pattern generalizes across the rest of Germanic: is the English pattern the default one or is the Dutch one, and are there other options? As you can see, there's plenty of stuff here to keep me occupied during the upcoming Christmas break. Should you want to give me a hand (and save me some time), feel free to let me know if/how your native language fits into this picture.

  1. Ok, I've been taught this, and I hope the same holds for you.

  2. And there's of course also the classic 1966 Postal-analysis of pronouns as nominal DPs that have undergone NP-ellipsis.

  3. Yes, I got a little carried away with the nominal constituent in Type 2.

  4. I realize that blij zijn 'to be happy' in Type 3 is technically not a verb, but bear with me; this will turn out to be important. Also, in case you're wondering about the word order here (or in example (6) below): given that blij 'happy' occupies roughly the same position as a clause-final verb in an embedded clause, we see the well-known fact kick in that Dutch is OV in embedded clauses with non-clausal objects, but VO with clausal objects. In all the other examples I can avoid this complication by V2'ing the verb out of the way.

  5. My linguistic spidey sense tells me that this is case-related: somehow these pronouns seem less susceptible to the case filter, which allows them to occur in contexts where full DPs are not allowed. Let's assume there are not two, but three possible contexts when it comes to case assignment: (1) typical adjectives like blij 'happy' do not assign case at all and as a result only allow sentential complements, (2) verbs like hopen 'to hope' typically require a prepositional complement (like they do in English), but also have some residual case assigning capabilities and as a result are compatible with both pronouns and sentential complements, (3) fully case-assigning verbs like vertellen 'to tell' (or case-assigning adjectives like beu 'tired') are compatible with the whole gamut of complements.

  6. Well, that and preparing the ultimate Christmas-Thanksgiving crossover dinner.

sushi chefs doing patisserie

Move over Whitehead and Russell, the era of Van Craenenbroeck and Vanden Wyngaerd has begun:

Ok, admittedly "Van Craenenbroeck and Vanden Wyngaerd" doesn't have quite the same resonance or mouth feel to it as "Whitehead and Russell"—let alone the scope, breadth, and impact on the world of science as a whole—but I'm still pretty happy about and proud of this. You see, together with my colleague Guido Vanden Wyngaerd I wrote an intro textbook on formal semantics. A not unimportant tidbit of information in all of this is the fact that neither Guido nor I are, in fact, semanticists; we're both full-blooded syntacticians. The idea of two syntacticians writing a textbook on semantics might sound a lot like two sushi chefs instructing you in the fine art of patisserie: sure, you expect them to bring a high level of skill and dedication to the job, and they'll definitely be good at slicing and dicing stuff, but they do seem to lack the tools, materials, know-how, and even basic vocabulary to help you get the most out of your Schichttorte. Like any good disadvantage, though, this one too can be turned into an asset. The book Guido and I wrote is the semantics handbook for syntacticians: you have a basic grasp of how tree structures work, and you would like to know how to complement your insight into hierarchical structure with an understanding of what it means to merge two syntactic nodes into a single constituent. The book doesn't eschew formal semantic notation, but at the same time doesn't go overboard on it: you get what you need to be able to give a complete, explicit compositional semantic analysis of a basic syntactic structure. No more, no less.

As with everything in life, though, there's a catch: the book is in Dutch, thus rendering it difficult to read for all but roughly 0.00379% of the world's population. What I'd like suggest, is that you all contact Amsterdam University Press and request a version of this important book in your own native language. Maybe that way we'll be able to give Whitehead and Russell a run for their money after all.

  1. H/t to Freek Van de Velde for coming up with the comparison.

  2. Tried to Right Node Raise a demonstrative out of an AP-contained PP there. I'll let the native speakers among you be the judge as to how successful that was.