Pete: I agree with you that the DLM paper doesn't prove much.
It certainly doesn't show that UG (in the Chomsky sense) exists.
But it does say something, however weak, about the nature of linguistic
cognition. Maybe only that we (humans) have difficulty remembering
syntactic relationships of words if they are far apart. But that still says
something, although not new.

Something happens in the human brain that allows us to produce and
understand language. But what, exactly? When I read papers like this I
don't think about UG, I think about biology. And then I wonder how
conlanging can help explore the fringes of the biologically-constrained or
-mediated aspects of linguistic cognition.


On Tue, Aug 11, 2015 at 12:05 AM, Pete Bleackley <
[log in to unmask]> wrote:

> The point of the paper I referred to was that the diversity of languages
> is more interesting and a more important topic of study than any search for
> "universal" features or tendencies.
> Let's look at the DLM claim in more detail.
> Related words tend to occur more closely together than they would if words
> were arranged randomly.
> Therefore the order of words is generally not random.
> Therefore there is information contained in word order.
> Therefore syntax exists.
> So, based on a statistical analysis of a small sample of languages, the
> authors conclude something that should surprise nobody. Then they claim
> that this is evidence of Universal Grammar. But what would be different
> without Universal Grammar? Why would they predict that, without Universal
> Grammar, syntax would not exist?
> Also, the effect varies in strength between languages. Latin syntax
> carries less information than English syntax. However, suppose we were to
> count the frequency distribution of inflected forms in the sample, we could
> calculate how much information was carried by morphology. I would expect
> that the more information there is in morphology, the less there will be in
> syntax. But that is just basic information theory, and will apply equally
> well to any model of language.
> Pete Bleackley
> The Fantastical Devices of Pete The Mad Scientist -
> Emily Semantic Recommendation -
> -----Original Message-----
> From: Jeffrey Brown <[log in to unmask]>
> To: [log in to unmask]
> Sent: Mon, 10 Aug 2015 10:58 pm
> Subject: Re: DLM (dependency length minimization)
> It seems that there is a misunderstanding about what a "universal" is, in
> terms of linguistics. As far back as Greenberg's "Universals of Language"
> (1963) -- and probably earlier -- the term has meant a universal *tendency*
> -- not a mathematical axiom nor a firm constraint.
> As a trivial, non-linguistic, illustration of this terminological
> confusion, consider this:
> *It is observed that most people scratch their left hips with their left
> hands and their right hips with their right hands. A hippy-guist states
> that this is a universal, namely, "same-side-hip-scratchiness." Other
> hippy-guists deride this, by pointing out that it's not a constraint nor
> restriction, and in fact it is possible for people to scratch their left
> hip with the right hand, and vice versa. The first hippy-guist simply
> reiterates his claim of the universality of same-side-hip-scratchiness, and
> goes on to say that this phenomenon is caused by the particular anatomy of
> the human skeleton. The second group of hippy-guists say that this is
> obvious, if not trivial, and moreover not part of the esteemed science of
> hippy-guistics.*
> (That was supposed to be "humerus" -- in case you were wondering. A little
> "brachial" from the serious discussion, as it were.)
> <ahem> And so, when people, as in the referenced article, are talking
> about linguistic universals, please do not bring up counter-examples from
> formal logic, mathematics, or computer science, where the term "universal"
> has a decidedly different meaning. Also, please do not state that
> something is trivial or boring. That is rather a dismissive comment. What
> is boring to one may be interesting to others. (As, for example,
> an acquaintance of mine who recently took a position crafting life
> insurance policies, and chewed my ear off for near an hour about the
> nuances of annuities -- and he wasn't trying to sell me one; he indeed
> found the subject fascinating!)
> Jeffrey
> On Mon, Aug 10, 2015 at 9:33 AM, Pete Bleackley <
> [log in to unmask]> wrote:
> > I saw a paper a while back arguing that anything truly universal to all
> > human languages would be trivial and boring.
> >
> > Pete Bleackley
> > The Fantastical Devices of Pete The Mad Scientist -
> >
> > Emily Semantic Recommendation -
> >
> > -----Original Message-----
> > From: Matthew George <[log in to unmask]>
> > To: [log in to unmask]
> > Sent: Mon, 10 Aug 2015 5:31 pm
> > Subject: Re: DLM (dependency length minimization)
> >
> > It seems pretty clear that there are no 'universals' restricting human
> > communication.
> >
> > Matt G.
> >