Dear All,
The puzzle, Frankie's comments on it, and the phrase 'just build a
context..' in the procedure outlined by Guy Politzer for solving it, all
seem to spotlight once again the theoretical vacuum in RT around the
mechanism of relevant context retrieval or construction.
Most of us would probably agree that attempts to model mental organisation
of 'knowledge' like semantic networks, ontologies, and binary feature trees
predict much slower retreval of relevant contexts than is observed, and I
think both Matsui 1998 (critique of Sanford and Garrod 's 1981account of
bridging references and Carston 2002 refer to work confirming that such
retrieval is much faster than all these frameworks predict*. RT seems to
be wary of cognitive semantics and mental models theories which propose
such models (eg prototype theory) because they also carry claims about
reasoning processes incompatible with RT, only the relatively innocuous
idea of schemas and scenarios organising and facilitating retrieval is
accepted.
What still seems vague is the relationship between constraint and
facilitation. For instance, all three of Frankie's plausible hypotheses to
explain our failure on the 'surgeon' puzzle, especially these extracts (my
emphasis) highlight the constraining, suppressing role of schemas evoked by
linguistic cues (with humbling implications for our pretensions at
lucidity, suggesting that we can only ever see anything at a cost of great
effort to escape them).
'.....We are given no reason to doubt this guess until we come to the final
statement, at which point our deduction system _isn't strong enough to
shake off_ our previously-held assertion.'
and '... use of the male gender _requires_ us to access concepts
of maleness in our minds which consequently spill over into our mental image
of the surgeon.'
I suppose this might be consistent with a fast, modular Fodorian operation
of relevance, like a filter.
But this seems to be the opposite of the constructive process implied in
Guy Politzer's 'way out of the impasse', viz. 'Just build a context...etc.'
, unless this expression is just being used loosely.
I hopr this makes some sense and that I am not just ignorant of this part
of relevance theory. I don't think any of the above justifies suggests
retreating from, for example, the adhoc concepts theory in favour of , for
example, natural-kind or feature-based models of how concepts and thus
utterance interpretations are constructed, but does anyone have thoughts,
information, or references about what a RT theory of context
retrieval/order of accessibility would look like?
Thanks in advance,
Robin Setton
*I'd be grateful for a reminder of the reference, by the way.
Prof. Robin Setton
Professeur à l'Unité d'interprétation
Ecole de traduction et d'interprétation (ETI)
Université de Genève
UNIMAIL
40 Bd. du Pont d'Arve
CH-1211 Genève 4
Robin.Setton@eti.unige.ch
Tel: (41-22) 705-8753
Fax (Unité) 705-8759
This archive was generated by hypermail 2b29 : Wed Jan 21 2004 - 14:44:06 GMT