be defined in some context-free manner. Idiom researchers face the challenge of
understanding the exact contribution of lexical semantics in people’s interpre-
tation of idiomatic phrases. My suspicion is that idioms differ quite a bit in
terms of how their parts contribute to their meaning as a whole (see Nunberg
1978). Thus, even two analyzable expressions, such aspop the questionandspill
the beans, differ in that the wordquestionhas a fairly conventional meaning in
the context of the idiom (i.e., having to do with the question about marriage),
whereas the wordbeanshas little to do with its conventional meaning as a food
item. Various empirical studies have begun to examine how the meanings of
individual words, whether these are characterized as literal, metaphorical,
conventional, or whatever, are accessed during the processing of familiar and
unfamiliar idioms (Cacciari 1993; Flores d’Arcais 1993). Some of these studies
have revealed that many idioms have key points or uniqueness points, places at
which idioms become uniquely identifiable (Cacciari & Tabossi 1988; Tabossi &
Zardon 1993). These different studies provide evidence that is consistent with
the data on the analyzability of idioms. But much work needs to be done in
both linguistics and psychology on the different types of word meanings that
play a role in the linguistic behavior (e.g., syntactic productivity) and in the
learning, use, and understanding of idioms.
There are a number of interesting linguistic and behavioral consequences of
the idea that idioms differ in their degree of analyzability. One series of studies
showed that the semantic analyzability of an idiom affects people’s intuitions
about its syntactic productivity (Gibbs & Nayak 1989). For instance, people find
semantically analyzable or decomposable idioms more syntactically flexible
than unanalyzable idioms. Thus, an analyzable phrase such asJohn laid down
the lawcan be syntactically altered intoThe law was laid down by Johnwithout
disrupting its figurative meaning. However, semantically unanalyzable idioms
tend to be much more syntactically frozen (e.g., one cannot changeJohn kicked
the bucketintoThe bucket was kicked by Johnwithout disrupting its figurative
meaning).
Another series of studies indicated that semantic analyzability influences
people’s intuitions about the lexical flexibility of idioms (Gibbs, Nayak, Bolton,
& Keppel 1989). Thus, analyzable idioms can be lexically altered without sig-
nificant disruption of their nonliteral meanings (e.g.,button your lipstofasten
your lips), but semantically unanalyzable phrases cannot (e.g.,kick the bucketto
punt the bucket). More dramatically, the individual words in many idioms can
be changed to create new idiomatic meanings that are based on both the origi-
nal idiom’s meaning and the new words. For example, the idiombreak the ice
can be altered to formshatter the ice, which now has the meaning of something
like ‘break down an uncomfortable and stiff social situation flamboyantly in
one fell swoop!’ (McGlone, Glucksberg, & Cacciari 1994). McGlone et al.
argued thatshatter the iceis an example not of lexical flexibility but of semantic
productivity. People can understand semantically productive idiom variants
(e.g.,Sam didn’t spill a single bean) quite readily; and the more familiar the orig-
inal idiom, the more comprehensible the variant. Variant idioms can also be
understood as quickly as their literal paraphrases (e.g.,Sam didn’t spill a single
beanversusSam didn’t say a single word) (McGlone, Glucksberg, & Cacciari
1994).
Idiomaticity and Human Cognition 737