Sorry, folks, my message earlier today
came out messy and hard to read (it didn't
look like that on my screen when I wrote
it). I hope this attempt is less rough
on the eye. Best, Robyn
----------------------
Dear all
The following thoughts were prompted (or reactivated) by reading the recent
interesting paper by Ira Noveck, Maryse Bianco and Alain Castry describing
some experimental work on metaphor processing (the paper was publicised on
this list). I was going to address the authors specifically but then
thought the issue might be of wider interest to people involved in
relevance theory. However, be warned that this is not a short message, but
more like a short essay. I would very much appreciate corrections/comments
if anyone makes it through to the end.
Ira Noveck et al.'s basic position is that metaphorical utterances take
more processing effort and have more cognitive effects than 'non-figurative
equivalents'. Building on some work by Ray Gibbs (1990), they ran some
experiments to test this hypothesis and found that the results supported
it. Their claim is that this is precisely what the relevance-theoretic
account of metaphor predicts.
Gibbs (1994) also takes the 'extra processing effort' prediction to follow
from RT and finds it to be at fault on these very grounds: '... the
psychological research clearly shows that listeners do not ordinarily
devote extra processing resources to understanding metaphors compared with
more literal utterances. The metaphor-as-loose-talk view, ... may not see
metaphors as violations of communication norms but still incorrectly
assumes that metaphors, and other tropes ... obligatorily demand
additional cognitive effort to be understood' (p.232). It is this
psychological evidence that Noveck et al. are calling into question.
However, the opposite view of what RT predicts about metaphor processing is
also to be found, espoused by another staunch relevance theorist. In his
excellent book on relevance and poetic effects, Pilkington (2000) discusses
psycholinguistic work by Gibbs and by Gerrig (1989) which seems to show
that 'utterances used metaphorically do not necessarily take longer to
process than the same utterances used literally' (p.89). According to
Pilkington 'the relevance theory account of metaphor interpretation is
quite consistent with [these] psycholinguistic results' (p.90).
So what's going on here? On the face of it, it looks as if someone has to
be wrong about what the RT prediction is. Before I check back on what the
founding father and mother have to say, note that talk of extra or more
processing effort raises the question of 'more than what?'. What is the
metaphorical utterance being compared with? I think there are two
different comparisons at work and that this may be the source of the
apparently contradictory positions.
Sperber & Wilson (1985/86), discussing such highly conventional metaphors
as 'Jeremy is a lion', which communicates the strong implicature that
Jeremy is brave, say that such metaphors 'must suggest some further line of
thought if their relative indirectness and its extra processing cost is to
be justified ...' (p.167) and, of course, this point holds all the more for
more creative metaphors. The processing effort comparison here is between
uttering 'Jeremy is a lion' and uttering 'Jeremy is brave', or between 'You
are a piglet' and 'You are a grubby kid'. It is this sort of comparison
that Noveck et al. have in mind.
However, consider now an example in Sperber (2000, p.132): 'John is a
soldier'. In a particular context (A), a hearer might access an
interpretation along the lines that John is very dutiful, loyal, obedient,
a team-player, self-sacrificing, but not including that John is a member of
the military, has weapons, etc. This would be a case of a loose or
metaphorical use of 'soldier'. In a different context (B), the 'literal'
properties of being a member of the military, being trained to kill, etc.
might be accessed instead. In both cases, as ever, the hearer follows a
path of least effort in accessing the interpretation and stops when his
expectation of relevance is satisfied. The issue of different degrees of
processing effort is not Sperber's concern here, so is not discussed, but
it seems clear that there is no prediction that there will be any
significant processing effort difference between the loose/metaphorical use
in context A and the literal use in context B. It also seems pretty clear
that in certain contexts the loose/metaphorical reading is going to be
easier to access than the literal one. I think these are the sorts of
comparison that Gerrig and Pilkington are making, and they are quite
different from that discussed in the previous paragraph.
So is that it? - problem solved? Probably not entirely. I'm not deeply
enough steeped in the full weighty mass of psycholinguistic experiments in
this area to be sure of my ground here, but I assume that not only cases
such as the 'soldier' one, but also cases like the 'piglet/dirty kid' one
have been tested and have given a result of no significant difference in
processing effort/time (a word of confirmation from Ray Gibbs and/or Sam
Glucksberg would be most welcome at this point). If so, this would still
seem to be at odds with the RT account, unless, of course, the measures
(reading times, etc) used in the experiments simply cannot tap the sort of
effort differences at issue.
Furthermore, there seems to be a striking difference in the psycho results
that I'm aware of between metaphorical predications and metaphorical
referring expressions. While the predications (which all the examples so
far discussed are) apparently take no more effort than literal paraphrases,
metaphorical reference (e.g. 'The princess' used to refer to a cat, 'That
junkyard' used to refer to someone's office) does take significantly longer
than literal reference. This was pointed out by Gibbs himself (1990) and
reinforced by Onishi & Murphy (1993), who tried everything they could to
ease the way for metaphorical references, but had to conclude that there
just is a heavier processing load for metaphorical referring expressions
than for corresponding literal ones. Of course, there are plenty of
possible explanations for this reference/predication difference, some of
which are discussed by Gibbs and O&M. Now, Noveck et al.'s experiments in
which a metaphorical/literal processing difference was found were all cases
of referring expressions. So their results further reinforce those of
Gibbs and O&M, but they could therefore be subject to an explanation for
the extra effort in terms of special properties of reference that make
metaphor difficult when used in this way. It might be of greater interest
for the RT prediction about 'piglet/dirty kid', etc. if there were
experiments which revealed effort differences when these are used in
predicating properties of an already determined referent. I wonder if
these are planned ...
Best wishes, Robyn
References:
Gerrig, R. 1989. Empirical constraints on computational theories of
metaphor. Cognitive Science 13, 235-241.
Gibbs, R. 1990. Comprehending figurative referential descriptions.
Journal of Experimental Psychology: Learning, Memory and Cognition 16, 56-66.
Gibbs, R. 1994. The Poetics of Mind. Cambridge: Cambridge University Press.
Noveck, I. et al. 2001. The costs and benefits of metaphor. Metaphor
and Symbol 16, 109-121.
Onishi, K. & Murphy, G. 1993. Metaphoric reference: When metaphors are
not understood as easily as literal expressions. Memory & Cognition 21,
763-772.
Pilkington, A. 2000. Poetic Effects: A Relevance Theory Perspective.
John Benjamins.
Sperber, D. 2000. Metarepresentations in an evolutionary perspective. In
D. Sperber (ed.) Metarepresentations: A Multidisciplinary Perspective,
117-137. Oxford: Oxford University Press.
Sperber, D. & Wilson, D. 1985/86. Loose talk. Proceedings of the
Aristotelian Society 86, 153-171.
-------------------------------------------------
Robyn Carston
Department of Phonetics & Linguistics, UCL
Gower Street, London WC1E 6BT, UK
Tel: + 44 020 7679 3174
Fax: + 44 020 7383 4108
URL http://www.phon.ucl.ac.uk/home/robyn/home.htm
-------------------------------------------------
This archive was generated by hypermail 2b29 : Tue Jun 26 2001 - 18:46:46 GMT