Saturday, January 19, 2013

The Nub of the Matter

I follow several other blogs but I don’t often follow the comment threads.  On the assumption that you might be like me I have decided to highlight a very interesting interchange on the last post “Effects, Phenomena and Unification” (here).  Alex Clark pens (can you use this verb for an e-post? ‘authors’?) an interesting comment. He clarifies the discussion immensely by identifying precisely where he disagrees with the thrust of my comments.  Here’s a taste of what he has to say, but I encourage you to read the whole thing:

So I think the root of our disagreement is in your final sentence of your post. What is the central fact of the matter? What is the central phenomenon that linguistics should explain?

I am in a minority here because I think the fundamental empirical problem of linguistics is to account for language acquisition, and not to account for "island effects, principle A, B and C effects, weak and strong crossover effects, the PRO theorem, Superiority effects etc. "

Cederic Boeckx has a reaction that is also worth reading in full. Here is a taste:

You are not in a minority, Alex. The central problem is still language acquisition (in fact, some of us have been trying hard to relate Darwin's problem and Plato's problem), but I think it's wrong to say that the whole point is "to account for language acquisition, and not to account for island effects, principle A, B and C effects, weak and strong crossover effects, the PRO theorem, Superiority effects". You can't care about one without caring about the other….

…If you don't care about island effects, etc. you can't be claiming to care about language acquisition, because you'd be ignoring the conditions that make learning possible in the first place. (The principles guide the child: Don't do this, don't do that. That's why most of them have a negative format; cf. Chomsky 1973, which you mention: "No rule can relate X and Y ...")

To repeat, read both posts for they are very enlightening and represent two ways of thinking in the field.  I bet you know that my view of things corresponds quite closely to Cederic’s. However, I would like to add a point to the one he makes for in a sense I believe that he is being too generous to Alex’s stated position.

From where I sit, Alex’s position is actually incoherent. Here’s why. If GB is a reasonable compendium of the properties of UG (as I’ve assumed it is) then it is a good description of the kinds of properties the particular Gs have. Gs are what people acquire and what makes them linguistically competent (I assume that this is common ground as nobody believes that what is acquired is a list of sentences but some kind of generative procedure, i.e. set of rules, a grammar).  So, if you want to describe how someone acquires linguistic competence then you need to explain how/why particular Gs have the shapes that UG describes them as having (i.e. why they obey the GB laws of grammar). Thus, particular Gs (French, English, your G, mine, Chomsky’s, Geithner’s, Madonna’s etc.) respect principles A, B and C of the binding theory, forbid movement across islands, control exclusively into the subject positions of non-finite clauses, respect cross over restrictions etc. In other words, particular Gs (the things acquired in the course of language acquisition) conform to the strictures of UG as described by GB. So if you want to explain language acquisition then you want to explain that. In other words Alex’s sentence “…I think the fundamental empirical problem of linguistics is to account for language acquisition, and not to account for "island effects, principle A, B and C effects, weak and strong crossover effects, the PRO theorem, Superiority effects etc.” is literally incoherent for it is saying that the program is to account for language acquisition but not to account for the properties of what is acquired. It’s sort of like saying that you are interested in the dynamics of atoms at high temperatures but not in the gas laws, or in the properties of gravity but not in Kepler's laws.  If GB is a roughly accurate description of UG then particular Gs will have the properties GB outlines and so learning/acquiring/osmosing/ingesting/tatooing language implies ending up with a system of rules aka a grammar, a G with those properties.

This said, I believe that Cederic response addresses Alex’s real doubts abut the generative enterprise. I suspect that what Alex doubts is that the properties that GB outlines are innate parts of FL. He thinks that either this assumption doesn’t help with the acquisition problem or that you can do without it even if it might help (see the thread here for discussion on this point).  Cederic and I believe that this is wrong.  But say that Alex is right (he isn’t, but as everyone who read Superman comics as a kid (before graduating to Spiderman and the Fantastic Four) bizzarro worlds are fun to consider), then it’s still the case that he should be interested in deriving the properties of GB as boundary conditions on Gs. Why? Because as a factual matter Gs have these properties.  That’s what the tons of descriptive cross linguistic work of the last 60 years has empirically demonstrated. That’s why I say that from where I sit Alex’s position is either incoherent or involves a position akin to climate change denial . Either it amounts to saying that you are not interested in deriving the features of the actual systems you are interested in or it amounts to denying that the descriptive work of the last 60 years is (roughly) accurate.

In sum, the only way of avoiding incoherence is to deny that we know much about UG, i.e. to become the linguistic equivalent of a climate change denier.  The descriptions we have of UG may not be perfect, but they are pretty good. Actually, in my view very good. Given this, any account of language acquisition that wants to be taken seriously must explain why Gs have the GBish properties I alluded to.  In addition, I happen to believe that there are good reasons for thinking that the reason that Gs have these properties is because large chunks of UG as specified by GB are innate (i.e. accurately describe FL) when suitably reanalyzed in minimalist terms. Thus: Why do Gs have GB features? Because the language acquisition device is constrained to produce Gs with these properties. Are there gaps in this story? You bet. That’s why we keep doing research on these topics. But the argument towards innate structure starts from recognizing certain facts, viz. that the descriptions provided by grammarians over the last 60 years is roughly correct. This is where I part company with Alex, not over what the central questions are but over what the facts are.

Addendum: after posting this I saw that Alex added another comment to the earlier thread where he in fact identifies the source of his skepticism. This too is very much worth reading. I will likely post a comment on this later on today, but first the gym!


  1. (Too long for one comment so I split it into two)

    I think we need to be a little more precise here: you are taking a rather delicate position in that you
    accept that GB is false, but you want to keep parts of it alive as valid phenomenological descriptions,
    but not all (e.g. deep structure has gone?).
    And you keep shifting your use of UG between meaning '1:initial state of the language faculty' and '2:universal properties of human grammars'. Now these are different things, and a claim that Principle B is part of 1 is a different claim from the claim that Principle B is part of 2.
    Below I will use 'UG' to mean the first, and universal properties to mean the second.

    And Principle B now seems just to be a 'phenomenon' which is something which is to be explained, rather than a theory that explains anything; which is the opposite to how Cedric is using it (I think).
    So this may seem either too pedantic or maybe too confused, but I hope it helps.

    I like your way of thinking as Principle B as a phenomenon; something along the lines of, say, retrograde motion of a planet. So you want to save Principle B as a phenomenon, while jettisoning it as an explanation of the phenomenon. Indeed, the final explanation might reduce Principle B to something else, perhaps unifying it with some other phenomenon that seems unrelated -- this is generally a very good thing in science, we agree (?).
    So presumably you accept that a final explanation may entirely dispense with Principle B and just have the Principle B effects being explained by some more general Principle X.

    That seems fine: I accept the existence of retrograde motion, but I don't like the explanation of epicycles.

    For me I think there is something more fundamental than retrograde motion. I think the fundamental question is whether the earth goes round the sun or whether the sun goes round the earth. I think, if you solve *that* problem, then the solution to retrograde motion will magically pop out, and indeed we will find that there is no concept of *retrograde motion* in the final explanatory theory.
    It's not that I am not interested in retrograde motion, or that I am denying the existence of retrograde motion, I just don't think it's fundamental.


  2. "So, if you want to describe how someone acquires linguistic competence then you need to explain how/why particular Gs have the shapes that UG describes them as having (i.e. why they obey the GB laws of grammar). "

    So here you are using UG to mean universal properties (I think?)

    All universal properties need to be explained eventually, and UG is one explanation -- they are baked into the genome. I just don't think it is a very good explanation, in the absence of some details (like, where exactly in the genome?, and how did it evolve? and why isn't there any variation? is it adaptive and how ? If not why is it conserved? and so on). But there are other non UG explanations for Universal properties of the set of observed human languages. Among which are (a) functional explanations, (b) historical relatedness if they all descend from a single proto language (c) effects of millenia of cultural evolution (d) random chance -- its just an accident (e) the supposed universals are not in fact universal and the next language in the Amazon will refute them, and so on and on.

    So yes I *am* exactly "saying that the program is to account for language acquisition but not to account for the properties of what is acquired.". It may be that the explanation of language acquisition does account for *some* of the properties of what is acquired: I hope and think it will. I think Principle B is a good candidate for the sort of phenomenon that might be partially explained by some bias towards types of hierarchical representation. But that is not my primary goal.

    Indeed I think focusing on trying to explain these secondary phenomena first is a mistake -- it leads to epicycles. Trying to find a clean explanation by idealising the fundamental problem is the truely Galilean path ....
    Ignoring Principle B is just like ignoring air resistance. (Maybe I am laying on the Chomsky rhetoric a little thick here, but this is a blog post..).

    So my view is entirely coherent, once we are clear about the phenomenon/theory distinction.
    I accept that Principle B is an interesting but approximate generalisation about the patterns of acceptability of some sentences in many languages but I do not think that ultimately we will find an innate Principle B (whatever innate means?).
    Does this make me a climate change denier? I have been called worse.

  3. Now we are on more or less the same page, I believe. So let me see:

    We both agree that there is a Principle B EFFECT (excuse the caps, it's just that I can't get italics to work on this damn thing). The question is what explains it. One theory, the GB theory is that Princ B is a fundamental feature of UG, i.e. the fundamental account of FL will have principle B as a stated axiom. I don't believe this to be so, nor do you. Ok, we are on the same page here.

    Second, we both agree that what we need is a fundamental theory that has principle B EFFECTS as a consequence, just like a good theory of planetary motion has retrograde motion as an EFFECT. Why? Becuase that's the way the world looks. Those are the facts. So we are still in agreement.

    Third, I believe that trying to develop theories that have the GB effects as consequences is the right way to proceed. You seem to demur. You seem to think that approaching matters more grab-like is a better bet. On this I have no real opinion. Your strategy is your strategy (I like throwing in a truism every now and then to grease discussion and to sound sage). So long as we agree that a story is wanting if it does not in the end get the effects and that a story is prized if it does. We are talking the context of justification, not the context of discovery.

    Where I part company from you regards research strategy and how you treat the binding principles as "epicycles." I regard these as pretty deep generalizations, with a lot of cross linguistic evidence to back them up. My analogy would be to the gas laws. They are not isolated one off phenomena regarding the motions of a recalcitrant planet or two (mars) (this would be more like demanding the theory abstract away from subject inversion in Romanian). Binding effects are pretty deep and so abstracting away from them hoping to get to them in a round about way strikes me as an odd strategy. But, like I said, your plan of attack is yours.

    Last point, I still take it as an empirical issue whether GB is wrong. We actually know very little about how brains change and how genes affect this. Thus, though I like the minimalist "problematic" I think that we judge replacements by how well they do covering ground that we know needs covering. This is how Einstein replaced Newton (the latter's laws are derived in the former when speed of velocity is small fraction of c), and how stat mechanics explained thermodynamics, and how Newton explained Kepler. The successful cases have generally recognized that the target of explanation are the earlier established laws, flawed as they might be.

    So, I absolve you of climate denial (if I got you right) and think that how we explain these effects is an open question that we don't fully understand yet. So long as we agree, that without getting these data we have not engaged the problem. I'm all for fundamental theory, but it's interesting just in case we agree on what effective theory we are trying to explain.