| |
Abstract:
Questions. Two different paradigms are discussed in the
literature with respect to online idiom comprehension during spoken
sentence processing: the modular and the configurational account
(e.g., Hillert & Swinney, 2000a; b; Cacciari & Tabossi,
1988; Titone & Connine, 1994). However, the data are not as
contradictory as it has been claimed. For instance, in Cacciari and
Tabossi's CMP (cross-modal priming) studies they probed at the
offset of the idiomatic phrase. It was said that only idiomatic but
not literal priming was found. However, the data show that subjects
responded to literal probes clearly faster than to control probes,
and statistics measured a tendency to significance. It appears that
not only the idiomatic meaning but also the literal meaning was
primed. A Type II statistical error seems likely. More recently,
Hillert and Swinney reported evidence that the perceiver seems to
access the idiomatic as well as the literal meaning at the head of
an idiomatic (German) compound - independent of whether the
compound was literally interpretable (e.g., Wasserratte; gloss:
water-rat; fig. someone who likes to be in the water; lit. 'cf.
gloss') or not (e.g., Lackaffe; gloss: lac-monkey; fig. dandy; lit.
N/A). The online evidence clearly favors a modular access account
of idiom processing. The question arises at which computational
stage the perceiver accesses the non-literal meaning.
Answers. Two experiments with the CMP-paradigm for lexical
decision were conducted. Subjects were asked to perform a lexical
decision task at the offset of the first element of the compound as
well as at the offset of the compound, which is identical with the
offset of the head - the compound's second element. The
experimental words were idiomatic terms with and without a literal
interpretation (e.g., Indian summer, talk turkey). The idiomatic
terms were embedded either in an idiomatic or in a literal biasing
context. First, the priming patterns replicated the patterns found
for German compounds (Hillert & Swinney, 2000b): At the offset
of the head literal and idiomatic priming was found for ambiguous
terms as well as for solely figurative terms. This priming effect
seems to be independent of certain lexical-syntactic structures,
because the English terms are syntactically more relaxed compared
to the strict N-N structure of German compounds. Moreover, at the
offset of the first element no priming was found for the literal or
for the idiomatic meaning. This finding is not predicted by a
number of lexical access models. For example, according to the
cohort model a perceiver accesses a word when encountering the
onset of a word (Marslen-Wilson, 1976). Thus, the model cannot
explain why the perceiver did not immediately activate the
idiomatic meaning, in addition to the literal meaning of the word.
In morphological complex words such as idiomatic compounds or
phrasal idioms the onset may consist of more than one word or of
several syllables before the lexical entry of the idiom can be
accessed as a whole. The experimental evidence supports the view
that idiomatic meanings are accessed in a modular fashion.
Cacciari, C. & Tabossi, P. (1988). The Comprehension of
Idioms. Journal of Language and Memory, 27, 668-683.
Hillert, D. & Swinney, D. (2000a). The Processing of Fixed
Expressions during Sentence Comprehension. In A. Cienki (ed.)
Conceptual Structure, Discourse, and Language. Stanford: CSLI. (in
press)
Hillert, D. & Swinney, D. (2000b). Accessing Idioms in Spoken
Sentence Perception. Evidence for Linguistic Modularity.
(submitted)
Marslen-Wilson, W.D., 1976. Linguistic Descriptions and the
Psychological Assumption is the Study of Sentence Perception. In
R.J. Wales and E.C.T. Walker (eds.), New Approaches to Language
Mechanisms. Amsterdam: North-Holland.
|
