Friday, June 16, 2017

Vapid and vacuous

It’s hard to be both vapid and vacuous (V&V), but some papers succeed. Here is an example. It is, of course, a paper on the evolution of language (evolang) and it is, of course, critical of the Chomsky-Berwick (and many others) approach to the problem. But the latter is not what makes it V&V. No, the combination of banality and emptiness starts from the main failing of many (most? all?) these evolang papers. It fails to specify the capacity the evolution of which it aims to explain. And this necessarily leads to a bad end. Fail to specify the question and nothing you say can be an answer. Or, if you have no idea what properties of what capacity you aim to explain, it should be no surprise that you fail to add anything of cognitive (vs phatic) content to the ongoing conversation.

This point is not a new one, even for me (see, for example, here). Nor should it be a controversial one. Nor, to repeat, does it require that you endorse Chomsky’s claims. It simply observes the bare minimum required to offer an evo account of anything. If you want to explain how X evolved then you need to specify X. And if X is “complex” then you need to specify each property whose evolution you are interested in. For example, if you are interested in the evolution of language, and by this I mean the capacity for language in humans, then you need to specify some properties of the capacity. And a good place to start is  to look at what linguists have been doing for about 60 years.

Why? Because we know a non trivial thing or two about human natural language. We know many things about the Gs (rules) that humans can acquire and something about the properties required to acquire such Gs (UG). We have discovered a large number of non-trivial “laws” of grammar. And given this, we can ask how a system with these laws, generating these Gs (might have) evolved. So, we can ask, as Chomsky does, how a capacity to acquire recursive Gs of the kind characteristic of natural language Gs (might have) evolved. Or we can ask how a G with these properties hooked up to articulation systems (which we can also describe in some detail) might have evolved. Or we can ask how the categorization system we find in natural language Gs (might have) evolved. We can ask these question in a non trivial, non vacuous non vapid way because we can specify (some of) the properties whose evolution we are interested in. We might not give satisfactory answers mind you. By and large the answers are less interesting than the questions right now. But we can at least frame a question. Absent a specification of the capacity of interest there is no question, only the appearance of one.

Given this, the first thing one does in reading an evolang paper is to looks for a specification of the capacity of interest. Note saying that one is interested in explaining the evolution of “language” without further specification of what “language” is and what capacities are implicated is not to give a specification. Unfortunately this is what generally happens in the evolang world. As evidence, witness the recent paper by Michael Corballis linked to above. 

It fails to specify a single property of language (more exactly the capacity for language for it is this, not language, whose evolution everyone is interested in) yet spends four pages talking about how it must have evolved gradually. What’s the it that has so evolved? Who knows! The paper is mum. We are told that whatever it is is communicatively efficacious (without saying what this means or might mean). We are told that language structure is a reflection of thought and not something with its own distinctive properties but we are not given a single example of what this might mean in concrete terms. We are told that “language derives” from mental properties like the “generative capacities to travel mentally in space and time and into the minds of others” without having a specification of the either the relevant generative procedures of these two purported cognitive faculties nor a discussion of how linguistic structures, whose properties we know a fair bit about, are simple reflections of these more general capacities. In other words, we are given nothing at all but windy assertions with nary a dollop of content. 

Let me fess up: I for one would love to see how theory of mind generates the structure of polar questions or island effects or structure dependency or c-command or anything at all of linguistic specificity. Ditto for the capacity for mental time travel. Actually, I’d love to see a specification of what these two capacities consists in. We know that people can think counterfactually (which is what this seems to amount to more or less) but we have no idea how this is done. It is a mystery how it is that people entertain counterfactual thoughts (i.e. what cognitive powers undergird this capacity) though it cannot be doubted that humans (and maybe other animals) do this. Of course unless we can specify what this capacity consists in (at least in part) we cannot ask if linguistic properties are simple reflections of these. So, virtually all of the claims to the effect that theory of mind (not much of a theory by the way as we have no idea how people travel into other minds either!) and time travel suffice to get us linguistic structures is empty verbiage. Let me repeat this: the claims are not false, they are EMPTY, VACUOUS, CONTENTLESS.   

And sadly, this is quite characteristic of the genre. Say what you will about Chomsky’s proposal it does have the virtue of specifying the capacity of interest. What he is interested in is how the generative capacity that give rise to certain kinds of structured arose and argues that given its formal properties it could not have arisen gradually. Recursion is an all or nothing property. You either got it or you don’t. So whenever it arose it did not do so in small steps, first 2-item structures, then 3, then 4, then unboundedly many. That’s not sensible, as I’ve mentioned more than a few times before (see, e.g. here and here). So Chomsky may be wrong about many things, but at least he can be wrong for he has a hypothesis which starts with a specified capacity. This is a very rare thing in the evolang world, it appears.

Actually, it’s worse than this. So rare is it that journals do not realize that absent such specifications papers purportedly dealing with the topic are empty. The Corballis paper appears in TiCS. Do the editors know that it is contentless? I doubt it. They think there is a raging “debate” and they want to be the venue where those interested in the “debate” go to be titillated (and maybe informed). But these is no debate because at least the majority of the discussants don’t say anything. The most that one can say of many contributions (the Corballis paper being one) is that they strongly express the opinion that Chomsky is wrong. That there is nothing behind this opinion, that it is merely phatic expression, is not something the editors have likely noticed.

The Corballis paper is worth looking at as an object lesson. For those that want more handholding through the vices, there is also a joint reply (here) by a gang of seven (overkill IMO) showing how there is no there there, and pointing out that, in addition, the paper seems unaware of much of modern evolutionary biology.  I cannot comment on the last point competently.[1] I can say that the reply is right in noting the Corballis paper “leave[s] the problem [regarding evolang, NH] exactly where it was, adding nothing” precisely because it fails to specify “the mechanisms of recursvie thought” in time travel or theory of mind and “how might lead to the feat that has to be explained” [i.e. how language with its distinctive properties might have arisen NH].

So can a paper be both vapid and vacuous? It appears that it can. For those interested in writing one, the Corballis paper provides a perfect model. If only it were an outlier!

[1] Though I can believe it. The paper cites Evans and Levinson, Tomasello and Everett as providing solid critiques of modern GG. This is sufficient evidence that the Corballis paper is not serious. As I’ve beaten all of these horses upside the head repeatedly, I will refrain from doing so again here. Suffice it to say, that approving citations of this work suffice by themselves to cast doubt on the seriousness of the paper citing it.

No comments:

Post a Comment