autonomous syntax

[Frederick Newmeyer]

Along with some of the other contributors to the discussion of syntax,
grammar, and autonomy, I've decided to stop lurking in the woodwork. I
think that I can speak for a majority of 'orthodox' generative grammarians
when I assert that the question of the autonomy of syntax (AS) has nothing
whatever to do with the 'fit' between (surface) form and meaning. Thus
George's and Talmy's critiques of autonomy are not to the point.

The AS hypothesis is not one about the relationship between form and
meaning, but rather one about the relationship between 'form and form'. AS
holds that a central component of language is a *formal system*, that is,
a system whose principles refer only to formal elements and that this
system is responsible for capturing many (I stress 'many', not 'all')
profound generalizations about grammatical patterning.

Let's take the most extreme 'cognitive linguistics' position, held, I
think, by Anna Wierzbicka. According to this position, any particular
observable formal aspect of language (e.g. categories, constructions,
morphemes, etc.) can be characterized by necessary and sufficient semantic
conditions. Is this position compatible with AS? Certainly it is. One need
only go on to show that grammatical patterning is also to a large degree
governed by more abstract relationship among formal elements that are not
replaceable by statements whose primitive terms are semantic.

A brilliant demonstration to this effect (and hence support for AS) has
been provided by Matthew Dryer in an article in LANGUAGE. Dryer shows that
the underlying generalization governing the Greenbergian word order
correlations is not a semantic one (e.g. head-dependent relations or
whatever), but rather the *principal branching direction* of phrase
structure in the language. The two often are in accord, of course; where
they conflict it is the abstract structural relationships provided by
formal grammar that win out. In an in-preparation work, I argue that this
is the norm for language. Yes, the fit between surface form and meaning is
quite close. But yes, also, formal patterning has a 'life of its own', as
is asserted by AS.

Since we have also heard it claimed that the apparent nonlocalizability of
syntax in the brain refutes AS, I'd like to address that question too.
What precisely is implied by the claim that we are endowed with an innate
UG module? Among other things, presumably that there are innate neural
structures dedicated to this cognitive faculty. However, nothing whatever
is entailed about the *location* of these neural structures or their
degree of 'encapsulation' with respect to other neural structures. Perhaps
they are all localized in one contiguous area of the brain. On the other
hand, they might be distributed throughout the brain. It simply does not
matter. Yet any number of critiques of AS have attempted to refute the
idea of an innate UG when, in fact, they have done no more than refute a
localist basis for it.

One might object that Chomsky has invoked the image of the 'language
organ' on a number of occasions, which has the effect of implying that the
neural seat of UG must be localizable in some part of the brain. But it
seems clear that his use of that expression is based on his hypothesis
that it is determined by a genetic blueprint, not on its physical
isolability. For example, he asserts that 'language is to be thought of on
the *analogy* of a physical organ' (REFLECTIONS ON LANG, p. 59) and that
'we may *usefully* think of the [language faculty as] *analogous* to the
heart or the visual system or the system of motor coordination and
planning' (RULES & REPS, p. 39). So Chomsky clearly is thinking of
language as something like an organ in a physiological, but not narrowly
anatomical, sense.

Steve Pinker has provided an interesting argument why the language faculty
is *not* confined to one area of the brain. He notes that hips and hearts
as 'organs that move stuff around in the physical world' have to have
cohesive shapes. But the brain, as 'an organ of computation', needs only
connectivity of neural microcircuitry to perform its specific tasks Q
there is no reason that evolution would have favored each task being
confined to one specific center.

So, to conclude, AS is an empirical hypothesis and one which, I am sure,
can be productively debated on Funknet. But it is important to focus on
those questions that bear on its adequacy and to put aside irrelevant or
tangential issues.

--Fritz Newmeyer