Reply to Fritz Newmeyer

[Frederick Newmeyer]

I find myself in the awkward and undesirable position of telling Matthew
Dryer that he doesn't appreciate the implications of his own
ground-breaking work.

The sensitivity of speakers to the abstract (formal, structural) notion
'Phrase structure branching-direction', a notion that doesn't (as Dryer
shows) correlate perfectly with semantic notions such as 'head-dependent'
supports the idea that speakers mentally represent abstract
phrase-structure relations INDEPENDENTLY of their semantic and functional
'implementations'. That is, it supports the autonomy of syntax.

Matthew writes:

> As for Fritz' claim that my evidence that the word order
> correlations involve branching direction rather than head position
> provides an argument for the autonomy of syntax, I would argue the
> opposite.  Those who assume that the correlations reflect head
> position generally treat consistent head position as an explanation
> in itself.  For example, a common position among formal linguists
> most closely aligned with Chomsky is that there is some sort of
> head-position parameter that is part of innate knowledge.

Yeah, but those people are wrong. As Dryer has shown, it's branching
direction, not head position, that is relevant. My view is that branching
direction is a MORE hard core 'formal' notion than head position, the
determination of which always seemed to involve semantic considerations.

And note that I never said anything about 'innateness' in this context.
The question of the autonomy of syntax and that of innate syntactic
principles have to be kept separate. The 'autonomy of syntax' is a fact
about mature grammars. What the language learner draws upon to construct
that grammar is a separate issue. So while innate syntactic principles
entail (I think) autonomous syntax, autonomous syntax does not ential
innate syntactic principles.

Matthew goes on to write:

 > Conversely, I have suggested that the tendency towards consistent
 > branching direction reflects (in addition to grammaticization
 > factors) parsing problems associated with mixed branching, i.e.
 > "performance"problems extracting the intended meaning.  If this
 > view is correct, then the explanation lies in the nature of human
 > working memory, and thus is inconsistent with notions (1) and (2) of
 > autonomous syntax.

I agree that there is a functional explanation (parsing-based) for why
speakers prefer consistent branching direction. That doesn't challenge the
autonomy of syntax, since AS is a claim about what speakers mentally
represent, not what they 'prefer'. In other words, functional explanation
is perfectly compatible with AS. Let me give a chess analogy. Nobody could
deny that the rules of chess (pieces, possible moves, etc.) form an
autonomous system. But functional factors could have (indeed, surely did)
enter into the design of the system. A ruling from the International Chess
Authority could change the rules (resulting in a different, but still
autonomous, system). Furthermore, when playing a game we have a choice as
which pieces to play, which moves to make.

Syntax, then, is autonomous in very much the same way that chess is
autonomous. We mentally represent an autonomous system. Why that system
has the properties that it has is another question. One answer is, as
Matthew points out, is pressure from the parser. Another is probably
pressure for iconic representations (see my paper in LANGUAGE of a few
years ago.) Another may be innate syntactic principles.

By the way, the most compelling, in my view, parsing explanation for
Dryer's generalzation is Jack Hawkins' principle of 'Early Immediate
Constituents' (see his book A PERFORMANCE THEORY OF ORDER AND
CONSTITUENCY). Hawkins is absolutely explicit that parsing explanations
are compatible with autonomy; ideed, he sees the former as a partial
explanation for the latter. Hawkins goes on to write:

"More generally, the very autonomy of syntax may owe its existence to the
need to make semantics processable for both the speaker and the hearer,
and it remains to be seen whether any precision can be given to the
formula: semantics + processing = syntax." (Hawkins 1994: 439)

While his formula (as he would I am sure agree) is too simple, I basically
agree with him.

--fritz newmeyer