The multidetermined nature of idiom processing

Models of idiom comprehension differ in their predictions concerning compositionality: Some claim that idiomatic meaning is the result of compositional analysis initiated at the earliest stages of comprehension, whereas others claim that compositional analysis occurs only at late stages, subsequent to direct retrievalespecially for idioms that are highly familiar. We evaluated these alternatives in four experiments by using a variety of online and offline comprehension measures. In Experiment 1, we analyzed the normative characteristics of 219 idioms with respect to these predictions. Dimensions of interest included several measures of decomposability, familiarity, and word frequency of the idioms' verbs and nouns. In Experiments 2 through 4, we determined how these dimensions relate to several online measures of idiom comprehension. High familiarity was associated with good comprehension across all experiments; however, facilitative effects of decomposability were found only for tasks that required an overt semantic judgment. Word frequency, but not semantic decomposability of the idiom-initial verb, was associated with comprehension for some measures. These data support a model of idiom comprehension, according to which figurative meaning arises from the time-dependent availability of multiple linguistic constraints, and in which decomposability plays a limited role in the earliest stages of idiom comprehension. Normative data for 210 of the idiomatic phrases may be downloaded from the Psychonomic Society Web archive at www.psychonomic.org/archive/. - Idioms are classically defined as phrases whose figurative meanings are distinct from their component words, for example, kick the bucket, spill the beans, be on cloud nine, and others (Abel, 2003; Cacciari & Glucksberg, 1991; Gibbs, Nayak, & Cutting, 1989; Nunberg, 1978; Titone & Connine, 1999). In the psycholinguistic literature, three classes of theories have been proposed to explain how idioms are represented and processed during comprehension. One class of theory holds that idioms are represented as long words (Bobrow & Bell, 1973; Swinney & Cutler, 1979), thus adhering to the classic noncompositional view. A second class of theory holds that an idioms internal semantic structure is necessary for understanding idiomatic meaning, thus taking a compositional approach (Abel, 2003; Gibbs, Nayak, & Cutting, 1989; Hamblin & Gibbs, 1999; Nunberg, 1978). A third class of theory adopts a hybrid approach that incorporates features of both noncompositional and compositional theories (Cacciari & Tabossi, 1988; Titone & Connine, 1999). Accordingly, idioms have unitary representations that may be directly retrieved when idioms are familiar or predictable, but they may also be compositionally analyzed during comprehension, especially in the case of unfamiliar or unpredictable idioms. The present study tests predictions of these different views of idiom processing using several offline and online comprehension measures. Compositional theories of idiom processing have gained much ground in recent years (e.g., Gibbs & Nayak, 1989; Gibbs, Nayak, Bolton, & Keppel, 1989; Gibbs, Nayak, & Cutting, 1989; Hamblin & Gibbs, 1999). The notion that idioms are compositionally analyzed was first introduced by Nunberg (1978), who proposed a method of idiom classification that emphasized the interaction between an idioms literal and figurative meanings (see also Cacciari & Glucksberg, 1991; Geeraerts, 1995; Gibbs, Nayak, & Cutting, 1989; Nunberg, Sag, & Wasow, 1994). This classification scheme asserts that idioms vary in their semantic decomposability, and that there are different types of decomposability relations. For example, the words of normally decomposable idioms bear a direct relation to the figurative meaning (e.g., question in pop the question refers to a marriage proposal), whereas the words of abnormally decomposable idioms bear a metaphorical relation to the figurative meaning (e.g., maker in meet your maker metaphorically refers to a deity). Thus, decomposability is defined as the extent to which the words independently contribute to the figurative interpretation. With respect to comprehension, the main prediction of compositional models is that an idiom will be easier to comprehend if its words are related, in any way, to its figurative meaning. Furthermore, this advantage for decomposable idioms over nondecomposable idioms should occur at the earliest stages of comprehension (i.e., as soon as the figurative meaning accrues activation). Accordingly, the comprehension of decomposable idioms such as pop the question or meet your maker would be faster than that of nondecomposable idioms such as kick the bucket. This would occur because the idioms words are semantically consistent with a figurative interpretation for decomposable idioms but inconsistent with a figurative interpretation for nondecomposable idioms. However, it is also possible that decomposable idioms may not differ from nondecomposable idioms with respect to comprehension, or that nondecomposable idioms might actually be easier to comprehend. These two options would be inconsistent with compositional models. Indeed, to the extent that nondecomposable idioms such as kick the bucket are easier to understand than decomposable idioms such as pop the question, one may infer that direct retrieval of a complete idiomatic meaning is faster than a full compositional analysis of the string that generates the idiomatic meaning. Other models assert that idioms undergo both simultaneous compositional analysis and direct retrieval when their component words constitute a familiar and recognizable configuration (Cacciari & Tabossi, 1988; Titone & Connine, 1999), although the configuration model (Cacciari & Tabossi, 1988) does not make specific claims about the effects of decomposability. Rather, Cacciari and Tabossi (1988; see also Tabossi, Fanari, & Wolf, 2005, 2008) emphasize the dimension of predictability (i.e., the extent to which an idiom is completed idiomatically when its final word is omitted), which gates access to the idiomatic configuration, independently of other variables such as familiarity. In contrast, others (Titone & Connine, 1994b) view predictability effects as being partially determined by other variables such as subjective familiarity. For example, Titone and Connine (1994b) found a strong positive relationship between predictability and familiarity ratings for a relatively large set of English idioms. Another difference between the two approaches is that according to the hybrid view, facilitative effects of semantic decomposability should be modulated by familiarity. Thus, when comprehenders encounter highly familiar idiomatic sequences, idiomatic meanings will be directly retrieved and integrated into the ongoing discourse representation (see also Giora, 1997), irrespective of whether the component words bear any relation to the figurative meaning. In c (...truncated)