Newsgroups: sci.lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!pipex!uknet!festival!edcogsci!steve
From: steve@cogsci.ed.ac.uk (Steve Finch)
Subject: Re: IS LINGUISTICS A SCIENCE?
Message-ID: <Czrz64.J2M@cogsci.ed.ac.uk>
Organization: Centre for Cognitive Science, Edinburgh, UK
References: <3aj6ek$pig@ruccs.rutgers.edu> <3aqodb$39n@newsbf01.news.aol.com> <3atepr$hgl@ruccs.rutgers.edu> <Czq1rt.EsI@cogsci.ed.ac.uk> <3avtvi$nu2@ruccs.rutgers.edu>
Date: Thu, 24 Nov 1994 14:06:00 GMT
Lines: 71

jirifkin@ruccs.rutgers.edu (Jay Rifkin) writes:

>steve@cogsci.ed.ac.uk (Steve Finch) writes:

>>whether linguistic intuitions count as scientific data.  To so count,
>>they have to be:

>>1. Repeatable(ish).
>>2. Objectively measurable.

>>In particular, there has been no scientific evidence I have been aware
>>of to show that anything which linguists would like to call
>>"competence" has repeatable objectively measureable data associated
>>with it.  

>Actually, there is a large body of fairly uncontroversial data which
>proves repeatable and objectively measurable, where by objectively
>measurable I mean there are statistically significant levels of 
>agreement.  When you say linguistic "competence," I take it you mean
>that part of the cognitive faculty which deals with purely linguistic
>(syntatic, phonological, morphological, (scope)) matters, rather than
>with the conceptual content of utterances (or written written sentences).
>Let me provide an example which came to mind when you mentioned
>center-embedded sentences.

> [description of experiment which shows processing differences between
> centre-emedded and non-centre-embedded sentences]

Yes, I can beieve that, but how does this bear on what linguists do
all the time --- explain competence?  The study of performance data
_is_ science, and none of this has proved that "competence" is
anything more than a convenient fiction which allows folk to apply
formal language theory to natural language.

In particular, is there a performance difference (neural measurements
or otherwise) which will distinguish between two linguistic accounts
of a phenomenom as there is in (some) other theoretical science.  I
hold that although scientific methodologies can be used in linguistics
(and have been used for many years in psycholinguistics), none of this
can be used to "prove" or "disprove" linguistic accounts.

Example: 

One linguistic theory (1) proposes that the following two forms of
sentence both contain lots of nested dependencies, while another (2)
proposes that only B: has nested dependencies.

A: The can in which Mary put the box on which John put the match is red.

B: The can which Mary put the box which John put the match on in is red.

Barry & Pickering ran an experiment which showed a significant
processing difference between them indicating (they claim) that A: was
not being processed as if it had a nested dependency (centre embedded)
construction.

Did this prove anything atall about the validity of the linguistic
accounts?  Not a bit of it!  All it proves to theory 1 (GB, I believe)
is that some kinds of nested constructions are easier to process than
others.  Both theories are attempting to explain competence data, not
performance data, so evidence such as this from performance data is
irrelevant.

But the point I was trying to make is that the data which linguists
primarily interested in competence theories use doesn't count as
scientific.  They don't use or take notice of this data.  period.

Cheers,

Steve.

