Nahuatl Dominant Word Order

[macswan at asu.edu]

On Fri, 29 Dec 2000, Davius Sanctex wrote:
> Dear Jeff MacSwan:
>
> I have read carefully your paper on polysunthesis
> parameter. I thing it is a good work, but I dislike
> general tone it is too controversial and it seems a
> a personal attack.

Sorry you got that impression, David. In fact there is not the slightest hint
of a personal attack in my paper. I say Baker makes "ingenius" proposals, and
show great respect for his work. But I argue that his basic proposals are
wrong, so naturally there is controversy abounding in the text. If you don't
like controversy, then you should stand clear of academic writing quite
generally.

> One way or another, it is a valuable
> paper; although it is too collateral to the question of
> diachronic change in dominant word order.

The paper is not about diachronic change in Nahuatl. You'll find it equally
peripheral to any number of topics you care to enumerate.

> I don't know the work of Baker, and I will silent about
> the matter. One of my criticism to the paper are:
>
> 1) The examples ob langages having properties 2c-2n without
> being polysinthetic are irrelevant as yourself say and don't
> disclaim Baker's work. If we discuss the validity of statement
> "A implies B" it is sophistic mention examples with "B but not
> A" (these examples are irrelevant).


This is a good point, which I do point out in the paper. Baker says that all
languages which have properties a and b also have c-n. To show this is false,
it is sufficient to exhibit one example of a language which has a and b but
lacks any one of c-n.  I point this out explicitly in the paper, in the
context of implicational universals, but want to make a stronger point which
guards against quick and easy revisions of the thesis: I wanted to show that
for any property c-n, there are languages which have a-b and not one of these,
and languages which do not have a-b but do have one of these. In other words,
there is no relationship between the cluster of properties Baker outlines.

I said all of this in the article, on page 111, where I write:

"It should be emphasized that Baker's claims about polysynthetic languages are
implicational in nature. If a language has 2a and 2b, then it will also have
2c-2n. To refute this claim, one must simply exhibit a language which has 2a
and 2b but lacks one or more of 2c-2n, as I have done here. In additional,
however, by pointing out that the linguistic features 2c-2n are in fact quite
common across a wide range of languages, I have tried to suggest that the
particular cluster of features Baker observes for Mohawk and attributes to
other polysynthetic languages is a coincidence and not a consequence of 2a and
2b..."


> 2)I have carried out a chi-2 stadistical test of the samples you
> give in page 105 and the result is that there are by no means
> homogeneous (X2 = 53,91 muy lejos del valor límite que es 21,1
> dispongo de los cálculos detallados!). Probabily the samples are
> excesively reduced to be statiscally significant. In fact, with a confidence
> of 0,95 the samples don't represent the same population
> and therefore they correspond to diferent styles. Comparaision by
> pairs are:
>
.....

> Text 1 is (see values of X2) strongly different from others.
> Text 2 and text 4 are similar, and finally text 5 is similar
> to all others.

Interesting observations, but clearly not relevant to any point I was
making. My interest was in showing what is grammatically permissible in
Southeast Puebla Nahuatl, not in analyzing stylistic variation amoung
storytelling in SE Puebla Nahuatl. Also, because text invites characteristic
variation, I caution against exclusive reliance on this sort of evidence, and
focus instead on grammaticality judgments collected from native
speakers. Taken together, all the evidence suggests that SE Puebla Nahuatl is
an SVO language which allows postverbal subjects and occasionally preverbal
objects for purposes of focus and contrast.

> Claims that Southeast Puebla Nahuatl is otherwise valuable,
> as exemple (5) in page 104 shows, but I don't see valuable
> the evidence from text because they are too different and
> reduced.

I think the textual evidence is extremely valuable, and while writers may have
tended to use one permissible word order more than another, collectively they
defined the same range of permissible word orders that the native speakers
(example 5, page 104) did. Thus, with respect to grammaticality -- the topic
of the paper -- they are not more varied than the native speaker judgments.

The full text of the article is available at
http://www.public.asu.edu/~macswan/swjl.pdf

Jeff MacSwan
Arizona State University