Frontiers in Psychology (May 2015)
Effects of familiarity, context, and abstract representations on idiom processing in aphasia
Abstract
People with aphasia (PWA) often have impaired idiom processing. This could result from difficulty suppressing the literal meaning (Cacciari et al., 2006; Papagno et al., 2004). Supporting this, in an idiom-probe matching task in which PWA had to choose which of four probe words went best with familiar idioms, Cacciari et al. (2006) found that most errors were associated with the idiom’s literal meaning. However a post hoc analysis showed that PWA also chose more abstract foils than concrete foils. Cacciari et al. interpreted the choice of an abstract foil as indicating that the participant did not know the figurative meaning of the idiom, but knew that the literal meaning was not correct. This account predicts that the less accessible the figurative meaning of an idiom is, the more abstract errors PWA will make. Familiarity and context both can influence the accessibility of figurative meanings for idioms. Highly-familiar idioms have more familiar figurative meanings than literal meanings (Nordmann et al., 2014). Likewise, a figuratively-biased context can speed access of an idiom’s figurative meaning (Ortony et al., 1978), and aid healthy young adults’ comprehension of unfamiliar idioms (Qualls et al., 2003). The current study investigates the effects of the accessibility of figurative meaning on idiom comprehension in PWA by examining error patterns after reading high- and low-familiarity idioms presented with and without figuratively-biased contexts. PWA (n=18) and healthy age-matched controls (n=32) read sentence pairs in a 3x2 design, crossing idiom type (highly-familiar vs. less-familiar vs. a literal paraphrase of the idiom’s figurative meaning) with the position of a figuratively-biased context sentence (before vs. after the idiom sentence; 1a-f, Figure 1a). Participants then chose which of four probe words (figurative target, literal foil, unrelated concrete foil, unrelated abstract foil) went best with the pair. For controls, the odds of choosing the target increased for high vs. low familiarity idioms (β=1.44; p<.01), but there were no effects of sentence order or literality, and no interaction (Figure 1b). In PWA, getting context after the idiom decreased the odds of choosing the target (β=-0.5495; p<.01), as did reading an idiomatic sentence compared to a literal sentence regardless of familiarity (β=-1.42; p<.01). Similar to controls, the odds of choosing the target increased for high vs. low familiarity idioms (β=0.514; p<.05). PWA made literal and abstract errors at the same rate, regardless of sentence order, sentence literality, or idiom familiarity. However, the odds of making an abstract error were higher for PWA than controls (β=1.67; p<.01). PWA’s accuracy was lower than controls’, but it improved with higher familiarity and supportive context, indicating that these manipulations boosted their access of figurative meanings. Interestingly, the rate of PWA’s literal and abstract errors did not differ by condition. This is unexpected under Cacciari et al.’s (2006) account that abstract errors result from difficulty accessing idiom meaning. Instead, abstract errors could result from PWA’s weaker access of both abstract-word meaning (e.g. Sandberg & Kiran, 2014) and idiom meaning if weaker meanings are more acceptable as potential matches than stronger meanings (i.e. for concrete words).
Keywords
