conventional interpretations are not functions of the meanings of their individ-
ual parts (Chafe 1970; Chomsky 1965, 1980; Fraser 1970; Katz 1973; Weinreich
1969). For instance, the conventional, nonliteral interpretations ofblow your
stackorto get pissed off(i.e., ‘to get very angry’) cannot be determined through
an analysis of their individual word meanings. Many linguists have also noted
that the noncompositional nature of idioms explains why idioms tend to be
limited in their syntactic and lexical productivity. For example, one cannot
syntactically transform the phraseJohn kicked the bucketinto a passive con-
struction (i.e., *The bucket was kicked by John) without disrupting its nonliteral
meaning. Similarly, the noncompositionality of idioms also explains why idi-
oms are lexically frozen (i.e., why one cannot changekick the bucketintokick the
pailwithout disrupting its figurative meaning of ‘to die’). Finally, speakers
learn the meanings of idioms by forming arbitrary links between idioms and
their nonliteral meanings (e.g., forming links betweenspill the beansand ‘to
reveal a secret,’button your lipsand ‘keep a secret,’lose your marblesand ‘go
crazy’). Thus, children and second language learners presumably learn idioms
in a rote manner or simply infer the meanings of idioms from context.
How do we comprehend what idioms mean? The noncompositional view
of idioms suggests that idioms are understood through the retrieval of their
stipulated meanings from the lexicon once their literal meanings have been
rejected as inappropriate, or in parallel to the processing of their literal mean-
ings, or directly without any analysis of their overall literal meanings as phrases.
A variety of experimental studies in psycholinguistics indicate that figurative
uses of idioms are easier to process than literal uses (cf. Gibbs 1980, 1985, 1986).
The common explanation of this finding is that people do not perform normal
analyses on the individual lexical items when understanding idiom phrases, an
assumption that makes sense given the traditional view of idioms as having
noncompositional meaning.
My aim in this chapter is to challenge many of these traditional assumptions
about idiomaticity. I argue that most linguists and psychologists are simply
wrong to assert that idioms are noncompositional and have meanings that are
derived from dead metaphors. A great deal of evidence in linguistics and psy-
chology shows that many idioms are, at least to some extent, compositional or
analyzable. People do not simply assume that the meanings of idioms are arbi-
trary or fixed by convention. Instead, people make sense of idiomatic expres-
sions precisely because of their ordinary metaphorical and, to a lesser extent,
metonymic knowledge that provides part of the link between these phrases
and their figurative interpretations. Research in cognitive linguistics and ex-
perimental psychology supports the idea that idioms retain much of their
metaphoricity. Many idioms are partly motivated by pervasive, preexisting
metaphorical concepts that can account for significant aspects of the linguis-
tic behavior of idioms as well as for the acquisition and comprehension of idi-
oms. The metaphorical mappings underlying many idiomatic phrases give rise
to multiple entailments, one reason why people understand idioms as having
complex interpretations, contrary to the traditional view of idiomaticity.
My most general claim in this chapter is that the empirical study of idioma-
ticity reveals important aspects of how people think and reason about the con-
cepts to which idioms refer. In this way, the study of idiomaticity in language
734 Raymond W. Gibbs Jr.