2 cents on Nerbonne and Kiss

[Louise McNally]

Hello all,

I have to admit that wasn't paying much attention
to this discussion until I read John Nerbonne's
message, so I'm putting in my 2 cents at the risk
of redundancy and/or putting my foot in my mouth.
Please hit the delete key immediately if you find
me redundant and accept my apologies for posting
without having done the homework of reading all
the previous messages. But I couldn't
agree more with what Nerbonne said, and so I wanted 1)
to say *that* 2) to offer a bit of perspective from
outside the syntax world, and 3) to comment on
the following statement:

>To the best of my knowledge (and I have some knowledge
>in infinite V syntax) there is no *standard controlled
>VP analysis*, i.e. nobody claimed in a GB/MP analysis
>that the pertinent constituents in infinite V syntax
>under *raising or control* are VPs....HPSG analyses
>have devoted some time and effort to explicitly argue
>against a CP analysis and *in favour* of a VP analysis.
>In which way does a non-existent analysis differ from
>vague conjectures and tacit assumptions?

If I'm not mistaken, John Moore's 1991 UCSC thesis
(subsequently published by Garland Press)
argued for a VP-complement analysis of at least a
subset of control constructions in Spanish (e.g.
with verbs like "querer", 'want'), and of at
least a subset of raising verbs (e.g. "empezar", 'begin'),
and against a biclausal analysis within GB.
I'm not sure exactly what the adjective "standard"
means in Tibor Kiss's message, but if it means "what a
naive linguist would think a VP complement analysis
means", then Moore's analysis is indeed standard.
I supposed it's slipped past people because the title of
his thesis and at least those papers of his on the
topic that occur to me use the term "reduced" constructions
in the title rather than "control" or "raising".

That said, a perspective from outside syntax:
I learned syntax at a place and time (UC Santa
Cruz, the late 80s) where a big
issue was made about learning how to deal with
various frameworks, how to compare analyses within
frameworks, etc. etc. Nonetheless, the majority of
people who studied with me and worked in syntax/semantics
used GB or its successors more than in any other
framework, myself included, even though some of us did
or (in my case even still do) some things in HPSG or
other frameworks. I would not hesitate to affirm
that in the vast majority of cases, this was done
for one of two reasons:  in the case of
syntacticians, as a strategy to increase readership
and to improve chances of getting a job; in the case
of semanticists, because in the end,
a tree offered enough syntax to get the job done
(to put it crudely) and required less formatting
and explaining to the reader. Interestingly, the only
thesis I remember written using HPSG was written by the only
individual who (if I remember correctly)
had it clear from day 1 that he wanted an industry
job and wasn't interested in an academic one.

The fact that the decision is strategic or practical
does not mean that the choice between syntactic
frameworks doesn't matter or isn't important, and
obviously it's unfortunate that the reality is what
it is, that some people make strategic decisions
with the result that they work in frameworks they
don't particularly like, and don't work in frameworks
that they might like better (though Moore, at least,
has moved on). But until it's clear to
everyone that this practical or strategic motivation
is a BIG reality, the focus on substantive comparisons
between frameworks probably won't be very effective
in increasing the number of people who use syntactic
frameworks other than the latest MIT product.

Another important factor is the profile of the syntactician.
A lot of syntacticians I know are in language departments and
come from what used to be called the philologies.
My extensive experience with the philologies in a
country where the discipline really still exists
has left me with the impression that
for many of these people, the precision of one's
specific analysis of phenomenon X is just not as
high a priority as the ability of a framework to give you
a handle on (even perhaps coarse) similarities and
differences between languages, or to be able to
help one understand how different morphological and
syntactic phenomena are correlated with each other
within given languages. "Principles and parameters"
is very appealing for this kind of audience, I think, and its
lack of precision is not a defect (another caveat:
I'm not implying that one can't be as precise as
one wants with GB/P&P/MP, simply that that doesn't
appear to be a top priority in many works in the framework
that I have read). Even though I get very frustrated
at the imprecision of much work I read in
this framework, I'm sympathetic to
the attitude that I think contributes to it appeal.
(As an aside, if that last sentence makes me susceptible
to accusations of unscientificness, if
what's at issue is scientificness, I would say I
find at least as troubling, if not more so, the
way data are collected and treated in linguistics than
the tools that are used to analyze that data.)

So if the impression
is created (and I do honestly think that this impression
has been created, even if undeservedly) that HPSG's main
selling point is its precision as a framework, that is
not going to be enough of a compensation to the sort
of audience I've just described, especially if they have
to deal with the hassle of
formatting AVMs to boot (which -- why not say it? -- has
always been a real obstacle for me personally).
I understood John Nerbonne as saying that it was
essential to emphasize the OTHER selling points of
HPSG, if possible in a precise but nontechnical
way, and with that, as I said at the beginning of
my message, I wholeheartly agree.

Louise

**************
Louise McNally
Dept. Traducció i Filologia, UPF
La Rambla, 30-32
E-08002 Barcelona
(+34)935422248 /FAX: (+34)935421617
louise.mcnally at upf.edu