Thursday, January 31, 2013

Is Cartography anti-minimalist?

In my Syntax II course yesterday, I discussed the motivations for the move to Minimalism.  Namely, why would we want to abandon an effective framework like GB for a model with initially less empirical coverage. The obvious answer is that why GB is great for explaining the data, we have a greater challenge connecting it to the biological realities associated with linguistics.  This got me thinking about Minimalism 400-lb gorilla-- Cartography.

As some of you well know, I have serious reservations about the Cartographic program.

My problems with Cartography are as such:

  • In a minimalist approach we must assume that the language specific mental faculties (and by extension the genetic endowment) must be, well, minimal.

  • Cartography needs to assume a richly specified and language specific mental faculty (at least in my opinion).

However, the results of a lot of Cartographic work have been highly effective. More troubling, I'm not certain what the possible alternatives could be. (See the excellent volume edited by van Craenenbroeck Alternatives to Cartography which ends with the foreboding paper by Williams "There is no alternative to Cartography".

Though that title is somewhat misleading, and Williams paper does allow room for alternatives, I worry that that title might yet be right.

In my dissertation, I sketched a difference between what I termed "strong Cartography" and "weak Cartography".  However, I've never been particularly statisfied with that idea.  Namely, if we allow a "weak" version, then there would not seem to be much of a logical reason of excluding a strong version.

Even analyses that argue against aspects of the Cartographic project need to at least partial Cartography.  This is certainly true for me.  And also illustrated by the Ritter and Wiltschko (2009, from the aforementioned van Craenenbroeck volume) "...we assume that UG provides a template of fixed functional positions".

So how do we proceed?

My hope that is at least in part Cartographic facts can be explained outside of the narrow syntax.  Let's take the C over v over V case.  I think this is one of the clearest cases of an empirically robust Cartographic fact. My speculation (and hope) that this organization could be tied to a more general system not specific to language (i.e., our conceptual organization).  I am certainly not alone here. If this were anything but a blog post, I'd include a more compelling literature review.

I know some of you work specifically withing Cartographic perspectives.  What are your thoughts on this?


  1. Nice post. As you know, my own dissertation work also connects with cartographic work. If anything, I have an even stronger bias to want the specificity of the structures posited by cartography to have some principled explanation.
    One avenue that's been pursued, in part by critics of Cinque who want to explain adverb ordering in terms of exogenous considerations, is that the hierarchy is dictated by "natural scope" - the logical structure, external to language and humans, inherent in the semantic categories involved. This would then be a Universal grammar that belonged to the universe, not to the human faculty of language as such. (I should mention Tom Bever has an old (but very smart - it's Tom) paper suggesting similar ideas about language on a broader scale, that the structures of language are a Platonic object that humans are susceptible to by virtue of having a sufficiently complex brain - sort of like numbers and mathematics.)
    Cinque makes a nice reply (if I weren't drowning in work I could track down the reference), pointing out that while some such 'natural scope' explanations seem plausible for some adverbial orderings, not all can be so explained. Much more seriously, such an account cannot in principle say anything about the hierarchy of adjectives in the nominal domain. This is so because the semantics of adjectival modification is intersective: big red things and red big things are the intersection of the set of things that are red, and are big, and it doesn't matter what order you do intersection in.
    So, insofar as the cartographers are right about these facts, there is a considerable residue that is not forced as such by the mere logical structure. So, it seems it has to be something about us; it could be specific to syntax, or it could be inherited from the cognitive systems that language interacts with. Certainly there's some indication that at least rudiments of theta structure (agent-action-patient schemas) are there already in the great apes, but obviously the cartography is way more articulated than anything discovered in other species, as far as I'm aware. That doesn't rule out the idea. When you look at the organization of the visual system, neurons specific to certain angular orientations, certain areas of the visual field, certain kinds of motion, maybe it's not that crazy to think that our conceptual structure is hard-wired in that specific a way.
    On another note, Rizzi (I think in the intro to the second volume of the 'cartography of syntactic structures' series) argues that articulated structure as such is not, in itself, inimical to Minimalist aims. As he puts it, it just shows that concerns are balanced ina particular way in the organization of the syntactic system, such that "recursion is cheap", and specific elements of meaning are spread out into dedicated positions, quite plausibly one position per element of meaning.
    More should be said about this, but I'll leave it here for now.
    --Dave Medeiros

    1. Thanks for the comment. There are other flavors of adjectives as well (subjective, for instance). And there is an order between subjective and intersective adjectives. That certainly is something that could fall out of other cognitive systems. But as your said the "red" and "big" cases are more problematic...

      Thanks again for your very insightful comments!

    2. Very interesting. Lots of questions, but just one observation for now. The visual system is (certainly?) much, much older than whatever language-specific hardware we might possess. Meaning we might expect the visual system to be far more specialized/complex than something like narrow syntax. So it might be misleading to suggest that narrow syntax of this kind is plausible based on an analogy to the visual system. Not trying to start a big argument here, but curious what people think.

    3. Thanks for the comment, Alan!

      I wrote a longer response but then Google ate it. Anyway, I think you are spot on. Vision has been around a lot longer, it should be more complex. In a pure minimalist view, the complexity from language should be borrowed from other systems.

      Interestingly, Tom Bever has argued that movement may be borrowed from the visual system.

    4. Interesting. Do you have a citation handy? I'd like to take a look at it. Not a big hurry.

    5. I don't on hand. But I will try to track it down. I've heard him speak about it.

  2. Just to clarify, what I had in mind in referring to the specifity of the visual system was that syntax was connecting with older conceptual architecture, "inheriting" structure from it. I agree it's implausible that something like that should arise de novo just in and for language, in the syntactic system. So, that suggests that we might find correlates of these structures in, say, chimps, if we had any idea how to identify them.
    Probably worthwhile to insert a reference to Randy Gallistel's work here, finding remarkable complexity in the cognitive architecture of insects... and giving us reasons to worry about just how much more complex things might be in humans, and how much more work there is to do before we really understand human cognition.
    I've seen some of Tom's stuff on movement. The basic idea is rooted in the observation that the areas in human brains that seem to be involved in processing syntactic movement correspond to areas in the brains of apes that are part of the visual system, relating to fine motor control and recognition of the motion of conspecifics. Tom has some ideas about why these relate that he's asked not to be widely distributed, but for the basic architectural facts have a look at Wilson et al (2010) and Poeppel & Hickock.

  3. Yeah, though it need not necessarily be in chimps. But I think there's a certain degree of plausibility of a step-by-step evolution of a conceptual system that is not found with Language.

    I honestly don't think we are within generations of having very much concrete to say about human cognition. Maybe that's an argument for keeping up with Cartography...

    1. Indeed. That's the great appeal of studying language: it gives us some of the clearest evidence readily available of the intricate inner workings of the human mind. Though I think we have to be more careful in thinking about the properties we seem to find, and whether they're really 'in' the mind or merely emergent from the evolution of languages themselves - Alan no doubt has more informed things to say about the latter point.