Idiom comprehension in aphasic patients

Abstract

Idiom comprehension was assessed in 10 aphasic patients with semantic deficits by means of a string-to-picture matching task. Patients were also submitted to an oral explanation of the same idioms, and to a word comprehension task. The stimuli of this last task were the words following the verb in the idioms. Idiom comprehension was severely impaired, with a bias toward the literal interpretation. Very few errors were produced with words, making impossible to establish a correlation between comprehension of idioms and of individual words. The difficulties in idiom comprehension seemed to be due to the fact that patients rely on a literal-first strategy, accessing a figurative interpretation only when the linguistic analysis fails to yield acceptable results.

Introduction

In our speech we make use of expressions, which are not necessarily interpreted on a literal ground: idioms are among the most common forms of figurative language (Gibbs, 1999). These expressions are characterised by a semantic eccentricity: their meaning is not a direct function of the meanings of their component words. For example, the meanings of the words kick, the, and bucket, composed according to the syntactic relations among them, do not produce the meaning die suddenly. This is conventionally associated to the string and a person must know the convention in order to correctly understand it (Glucksberg, 2001). Moreover, idioms typically occur only in limited syntactic constructions, even though there is a large variability among them, and while some are almost completely unconstrained (e.g., spill the beans, let the cat out of the bag), others allow very few operations, if any (e.g., by and large) (Fraser, 1970).

Although idioms do not appear to serve a logical function in language, they are very common and reflect our tendency to say things in conventionalised ways (Jackendoff, 1995; Johnson-Laird, 1993).

These characteristics have rendered idiomatic expressions an object of great interest in current psycholinguistics. To date, there is a general consensus that the comprehension and production of idiomatic expressions call for processes that often differ from the processing of other forms of figurative language, most notably metaphors (Cacciari & Tabossi, 1993). Furthermore, most researchers agree that the traditional view, according to which idioms are mentally represented and recognised as long, morphologically complex words is not correct (Swinney & Cutler, 1979). In fact, the processes leading to the identification of verbal idioms during spoken language comprehension differ from spoken word recognition (Tabossi & Zardon, 1993; Tabossi & Zardon, 1995). Also, idioms may exhibit a certain degree of lexical flexibility (e.g., hit the hay → hit the sack) and productivity (e.g., break the ice → shatter the ice) (Gibbs, Nayak, & Cutting, 1989; McGlone, Glucksberg, & Cacciari, 1994) and undergo syntactic processing even after their figurative meaning has been retrieved (Peterson, Burgess, Dell, & Eberhard, 2001). These different types of evidence converge on suggesting that idioms are not a unitary class of expressions. Rather, they range from expressions that are almost like long words (e.g., by and large) to expressions that are like metaphors (e.g., skate on thin ice). In between, there are the majority of idioms, in particular verbal idioms that are syntactically and semantically processed, and accordingly can undergo syntactic and semantic variations.

At present, a number of dimensions are considered important in determining the processes underlying the identification and comprehension of the different types of idioms and the syntactic and semantic operations they can accept. Two interesting dimensions are compositionality and transparency (Glucksberg, 2001). The former refers to the degree to which the phrasal meaning, once known, can be analysed in terms of the contribution of the idiom parts. In spill the beans, for example, there is a clear correspondence between spill and beans and the relevant parts of its figurative meaning ‘divulge information’ (Nunberg, Sag, & Wasow, 1994). In this idiom, the correspondence is motivationally unclear to most people, i.e., along the dimension of transparency spill the beans is opaque. Transparency refers to the extent to which the idiom meaning can be inferred from the figuration (metaphor, hyperboles, etc.) it involves (Nunberg et al., 1994). Although there may be some relation between the two notions, they do not overlap. While some idioms, like spill the beans, may be decomposable and opaque, others may be non-decomposable and transparent. Saw logs, for example, has no identifiable parts mapping onto the meaning; yet, the acoustic communalities between a saw cutting woods and some sleepers is not hard to appreciate. The exact role of these characteristics is still an open question, but there may be little doubt that all the processes involved in the comprehension of literal strings, from lexical retrieval to syntactic parsing, are at work during idiom comprehension.

These lines of research clearly suggest that injuries to the left hemisphere, typically resulting in aphasic impairments, ought to damage, along with other linguistic skills, patients’ ability to comprehend idiomatic expressions. However, at odds with this tenet, a widely accepted view in current neuropsychological research assumes that damage to the left hemisphere may have no major consequences, and it is the non-dominant right hemisphere that is important for the processing of idiomatic expressions (Kempler, Van Lancker, & Bates, 1999; Van Lancker & Kempler, 1987).

However, recent studies have challenged this view. Papagno, Oliveri, and Romero (2002) and Oliveri, Romero, and Papagno (2003) used repetitive transcranial magnetic stimulation (rTMS) to test the ability of 15 healthy young right-handed Italian speakers in the comprehension of idioms. The scalp positions for stimulation were a postero-superior temporal site corresponding to T5–T6 positions, and a frontal site corresponding to a position between F5(6) and F7(8), both on the left and right hemisphere. A sentence-to-picture matching task was performed, both with a two-choice syntactic test and a two-choice idiom comprehension test, where the two alternatives in this last case were the picture corresponding to the figurative meaning and that corresponding to the literal meaning. The Authors found that only left temporal rTMS significantly interfered with participants’ accuracy and reaction times.

Similarly, Tompkins, Boada, and McGarry (1992) reported evidence against a right hemisphere involvement. They suggested that some of the deficits previously reported for right brain-damaged (RBD) adults might be partly an artefact of the differential plausibility of depicted alternative meanings (Huber, 1990), alone or in combination with patients’ visuospatial or visuoperceptual impairments.

The main goal of this study was to assess whether the comprehension of idiomatic expressions in left brain-damaged, aphasic patients is actually good, as claimed by the right hemisphere hypothesis, or it is impaired as follows from current psycholinguistic models of the mental representation and processing of idioms.

To this end, we selected 34 Italian verbal idioms. These expressions were all highly familiar and unambiguously idiomatic: none of them would normally be open to a literal interpretation, either because they are ill-formed (fare fiasco/make flask=to fail utterly), or because they do not have a literal meaning (e.g., montarsi la testa/mount one’s head=to pride oneself) or because they have a literal meaning, but that meaning would be pragmatically inappropriate in normal circumstances (e.g., essere sulla cattiva strada/be on the bad road=to misbehave). They were also opaque. Given these characteristics, the meaning of these expressions must once have been learned and their comprehension requires, among other things, their identification and the retrieval of their meaning from memory (Papagno, 2001).

We tested the ability of 10 aphasic patients to understand these expressions by means of a string-to-picture matching task, in which patients’ performance was compared with the performance of a matched control group.

We predicted that, in agreement with current psycholinguistic models of idiom processing, aphasic patients with lexical-semantic deficits would be impaired in the comprehension of these expressions and hence perform reliably worse than matched controls.