Journal o f Psycholinguistic Research, VoL 25, No. 6, 1996

Comprehension of Idiomatic Expressions in Japanese--Auditory and Visual Presentations

Ichiro Miura 1,z

Accepted March 27, 1996

This study reports on the auditory and visual comprehension o f Japanese idioms having both literal and figurative meanings. Experiment I conducted the rating o f the semantic distance between the two meanings. Experiment H investigated the difference o f com- prehension between semantically f a r and close idioms. Here the materials are presented in isolation both auditorily and visually. Experiment III conducted the same investigation as Experiment II, except that idioms were presented embedded in literally and figuratively induced contexts. Experiment I V reinvestigated the f indings obtained f rom the previous experiments. The results o f these experiments show that in isolation visual presentation precedes auditory presentation, and that both in the auditory and visual presentations semantically f a r idioms are comprehended more accurately than semantically close id- ioms.

INTRODUCTION

Every human language is endowed with ambiguity, whether spoken or writ- ten. Syntactically ambiguous sentences having dual meanings have been investigated mainly by linguists, whereas sentences having ambiguity at the lexical/semantic level have been investigated mainly by psychologists. In- cluded in the latter category are metaphors, idioms, and ironies (often called figurative language) since they have both literal and figurative meanings without manifesting syntactic ambiguity. A series of studies dealing with the comprehension of figurative language (Cacciari & Tabossi, 1988; Gibbs,

1 Department of English, Kyoto University of Education, Kyoto, Japan. 2 Address all correspondence to Ichiro Miura, Fukakusa, Fujinomoricho 1, Fushimiku,

Kyoto 612, Japan. e-maih miura@wsml.kyokyo-u.ae.jp

659

0090-6905/96/1100-0659509.50/0 �9 1996 Plenum Publishing Corporation

660 NIiura

1985; Kats, Paivio, & Marschark, 1985; Marschark & Hunt, 1985; Pollio, Fabrizi, Sills, & Smith, 1984; Schweigert & Moates, 1988) proposed three processing models--literal first, figurative first, and simultaneous. These studies have suggested that the choice of model depends on the familiarity of the figurative language and the context in which it appears. Following Schraw, Trathen, Reynolds, & Lapan (1988) and Schweigert (1986), low- familiarity idioms require longer reading times than high-familiarity ones, and appropriate paraphrases of figurative meanings are obtained more fre- quently from high-familiarity idioms than from low-familiarity ones. Kroll and Schepeler (1987) and Nippold and Martin (1989) have suggested that the comprehension of figurative language is achieved more accurately in context than in isolation, granted that the context is appropriate. Also, from the productive viewpoint, Williams-Whitney, Mio, and Whitney (1992) sug- gested that experienced writers use metaphoric expressions more frequently than inexperienced writers. Besides the above-mentioned studies which deal with normal adults, studies on adults having disorders or on children have also been conducted. Following Tompkins (1990), brain-damaged adults, whether right or left hemisphere, use procedures for comprehending figu- rative language similar to those used by normal persons, although the time required for comprehension is longer. Broderick (1991) and Pearson (1990) suggested that children as young as age 3 can comprehend figurative lan- guage, and that the accuracy of comprehension increases with child age. In children too, Cacciari and Levorato (1989) observed the effect of context: Children aged 7 and 9 comprehended idioms more accurately in context than in isolation. Also, Gibbs (1987) and Nippold, Leonard, & Kail, (1984) observed the effect of syntactic variability in comprehending figurative lan- guage, that is, that children can comprehend syntactically rigid expressions more accurately than syntactically flexible ones. They suggested that this tendency is seen especially in younger children. Further, in comparing nor- mal and mentally retarded children, Ezell and Goldstein (1991) suggested that the comprehension of the literal meaning does not differ between the two groups of children whereas the comprehension of the figurative meaning is achieved more accurately in normal children than in retarded children.

The present study investigated the comprehension of Japanese idiomatic expressions having both literal and figurative meanings. The primary con- cern was to see whether comprehension differed depending on the method of material presentation, that is, on whether the presentation is auditory or visual. This was done since very few studies have been conducted by au- ditory presentation: Auditory presentation is used exclusively for younger children having poor reading ability. The second concern was to see whether the semantic distance between the literal and figurative meanings affected comprehension. This was done since, although the notion of semantic dis-

Japanese Idioms 661

tance has already been adopted (e.g., Clevenger & Edwards, 1988), the re- sults have shown some discrepancy. When presented in isolation, semantically far idioms whose figurative meanings differ greatly from their literal meanings have required less time to comprehend than semantically close idioms whose figurative meanings show a proximity to literal meanings (Mueller & Gibbs, 1987), whereas no comprehension difference was ob- served between the two kinds of idioms (Gibbs, 1987). The third concern was to see the effect of context on comprehension, especially in relation to semantic distance. To this end, literally induced and figuratively induced contexts were prepared for semantically far and close idioms, respectively. Last, the present study concerned the Japanese language. The reason for this was not only that Japanese subjects can be employed easily since the author lives in Japan, but also that most of the previous studies have used subjects whose native language was English. It is of interest to compare the results obtained from English subjects and from Japanese subjects since figurative language varies delicately from language to language. Thus Experiment I in the present study conducted the rating of semantic distance for Japanese idiomatic expressions. Utilizing the results obtained from this experiment, Experiment II investigated how comprehension differed depending on the semantic distance. Here the materials were presented in isolation both au- ditorily and visually. Experiment III also investigated the effect of semantic distance on idiom comprehension, except that the materials were presented in context. Finally, Experiment IV was conducted in order to reinvestigate the results obtained from the previous experiments.

EXPERIMENT I

Materials

Thirty-five Japanese idiomatic expressions were singled out from a dictionary of idiomatic usage. The first criterion in choosing these idioms was that they have both literal and figurative meanings; expressions having only a literal meaning were excluded. The second criterion was that the idioms have the construction of noun + particle + verb. The reason for employing this construction was that, although idioms having other construc- tions (e.g., adjective + noun) exist, they are smaller in number and can be embedded in context less easily (a factor to be considered in Experiment III). The third criterion was that the number of syllables included in the idioms be approximately the same: The nouns employed were mostly of two syllables and the verbs were of two or three syllables. An example of the materials is mune o haru (literal meaning: throw out one's chest; figurative

662 Miura

meaning: boast). The final test materials were prepared randomizing the 35 idioms and were printed in Japanese orthography (Kana and Kanji) on A4 size sheets.

Subjects

Forty-eight Japanese university students served as subjects, all o f whom were in the same grade at the same university.

Procedures

First, the literal and figurative meanings and the semantic distance be- tween the two meanings were explained to the subjects using an example idiom. Subjects were then shown the test materials and were asked to judge the semantic distance between the two meanings on a 5-point scale, where 1 denoted semantically far and 5 denoted semantically close. They were told to leave the scale blank when they were not sure of the figurative meaning. No time limit was established.

Results

Means and standard deviations o f the ratings were calculated for each idiom. Mean values ranged from 1.67 for the idiom sune o kajiru (literal meaning: gnaw the leg; figurative meaning: hang on others) to 4.35 for the idiom atama o sageru (literal meaning: lower one's head', figurative mean- ing: apologize). For almost all the idioms, standard deviations centered around 1.0, which indicates a comparatively narrow range of ratings between the subjects. Grand mean value was calculated using the mean values ob- tained from each material, which yielded 3.0 with a standard deviation of 0.6. This indicates that the materials used here were somewhat inclined toward the close category. The percentage of responses ranged from 77 for the idiom shita o dasu (literal meaning: put out one's tongue; figurative meaning: insult) to 100 for idioms such as shiri o tataku (literal meaning: spank; figurative meaning: give encouragement). The mean value was 93 with a standard deviation of 7.2, which indicates that most of the figurative meanings were known to the subjects.

E X P E R I M E N T II

Materials

From the 35 idioms used in Experiment I, five far and five close idioms were extracted. In choosing these idioms, care was taken that no noun or

Japanese Idioms 663

verb was used repeatedly and that they showed over 90% responses. Thus these were not necessarily the closest or farthest five idioms. The means of semantic distance were 2.22 for the far idioms and 3.77 for the close idioms, and a significant difference was observed between these mean values (t = 13.320, df = 369, p < 0.01). In addition, five expressions having only literal meanings were prepared as control materials, whose syntactic construction and the number of syllables matched the far and close idioms. Here also, care was taken that no nouns or verbs used in the far and close idioms appeared in these literal controls. Appendix I shows all the materials used in this experiment with their English translations. The final test materials were prepared in the form of past-tense, active sentences beginning with the third-person pronoun. An example is Kate wa ashi o araimashita (literal meaning: He washed his feet; figurative meaning: He reformed himself). For the auditory presentation, a total of 15 materials were pronounced by a female speaker of Japanese, and her utterances were recorded. She was told to pronounce each idiom in approximately 3 s (a slightly slow rate of speech) in the same manner of utterance. The recorded utterances were checked by the author and a research assistant, and two idioms were rerecorded owing to the existence of noise. The final test tape was made randomizing the 15 materials with 3-s intervals in between. Note that the randomization was done once, that is, for the entire group of subjects. For the visual presen- tation, the 15 randomized idioms (in the same order as in the auditory pres- entation) were made to appear on the cathode-ray tube (CRT) screen: Each idiom was displayed for 3 s followed by a 3-s interval.

Subjects

Thirty-two Japanese university students were allotted to the auditory presentation and 33 Japanese university students to the visual presentation. These two groups of students were in the same grade at the same university but they belonged to different classes, None of them had hearing or visual disorders and none had participated in Experiment I.

Procedures

The subjects in the auditory presentation were asked to listen to the tape and to memorize as many sentences as possible. The presentation, which lasted approximately 11/2 min, was conducted in a quiet room using a loudspeaker situated in front of the classroom. Immediately following this, the subjects were asked to write down what they had memorized within 3 min paying no attention to the order of writing. The subjects in the visual presen- tation were asked to watch the CRT screens situated in front of each of them and to memorize as many sentences as possible. Immediately following this,

664 Miura

they were also asked to write down what they had memorized within 3 rain paying no attention to the order of writing.

Scoring

Two points were given when the noun and its corresponding verb were recalled correctly, and 1 point when either the noun or the verb was recalled correctly. When the noun and the verb recalled belonged to different cate- gories, 1 point was given to the respective category. No point was given when it was difficult to define to which category the recall belonged. Also, no point was given when the recall ended incompletely. Insertions o f un- related words such as adjectives or adverbs were ignored. The scoring was done by the author and a research assistant. Of the 628 sentences recalled, 11 sentences showed a discrepancy between these two persons and were discarded from the final analysis. Thus the interscorer agreement is assumed to have been .98.

Results

Table I shows the maximums, minimums, means, and standard devia- tions for the ratings o f the auditory and visual presentations, respectively. It seems that, both for the auditory and visual presentations, the scores for the far materials were higher than those for the close materials. Using the scores obtained from each subject, a one-way analysis o f variance (ANOVA) was conducted across the three categories of materials (far, close, and control), which showed a significant difference for the auditory presentation [F(2, 95) = 8.697, p < .05]. The post hoc test (Tukey, p < .05) showed a significant difference between the far and control materials. The ANOVA also showed a significant difference for the visual presentation [F(2, 98) = 5.548, p < .01], and the post hoc test showed significant differences between the far and close materials and between the far and control materials. These facts indicate that for the visual presentation semantically far idioms were recalled

Table I. Maximums, Minimums, Means, and Standard Deviations of Recall Scores in Experiment 1I

i

Auditory Visual

Max. Min. Mean SD Max. Min. Mean SD

Far 10.0 4.0 6.8 1.8 10.0 4.0 7.7 1.7 Close 10.0 2.0 5.8 2.2 10.0 2.0 6.2 2.5 Control 10.0 2.0 5.5 2.1 10.0 2.0 6.2 2.4

Japanese Idioms 665

significantly more accurately than semantically close idioms, whereas for the auditory presentation this tendency was observed insignificantly.

E X P E R I M E N T lII

Materials

The five far and five close idioms used in Experiment II (sentence form) were again utilized as target sentences. Each idiom was embedded in con- texts that would induce literal and figurative meanings, respectively: One sentence each was added before and after the target sentences. The preceding sentences always began with the proper name Taro (Roy) and the following sentences always begin with the conjunction soshite (then). The number of syllables included in these sentences were mostly 9 or 10, and all the sen- tences were simplex. Here, control materials were not prepared since their addition might require too much load on memory. Thus a total of 20 contexts were made, each three sentences long. These materials were divided into two lists: List A included five far-literal and five close-figurative contexts, and List B included five close-literal and five far-figurative contexts. The contexts were made by the author and were revised by a research assistant, and a final check was done by the two persons. Appendix II shows Japanese test materials and their English translations. For the auditory presentation, the same speaker employed in Experiment II was asked to pronounce each list, and her utterances were recorded. She was told to pronounce each ma- terial in approximately 10 s (a slightly slow rate of speech) in the same manner of utterance. The recorded utterances were checked by the author and the research assistant, and three materials were rerecorded owing to differences in the manner of utterance. The final tape was made randomizing the 10 materials with 5-s intervals in between. As in Experiment II, the randomization was made once for the entire group of subjects. For the visual presentation, the 10 randomized materials (in the same order as in the au- ditory presentation) were made to appear on the CRT screen: Each material was displayed for 10 s followed by a 5-s interval.

Subjects"

For the auditory presentation, 20 and 25 Japanese university students were allotted to List A and List B, respectively. For the visual presentation, 24 and 25 Japanese university students were allotted to List A and List B, respectively. These four groups of students were in the same grade at the same university but they belonged to different classes. None of them had

666 Miura

hearing or visual disorders and none had participated in the previous exper- iments.

Procedures

The subjects in the auditory presentation were asked to listen to the tape and to memorize as many sentences as possible, trying to understand the contents. The presentation, which lasted approximately 21/2 rain, was conducted in a quiet room using a loudspeaker situated in front of the classroom. Immediately following this, the subjects were asked to write down the second sentences in each material within 3 rain, paying no attention to the order of writing. For facilitating the recall, they were told that the second sentences would always begin with Kate wa. The subjects in the visual presentation were asked to watch the CRT screens situated in front of each of them and to mem- orize as many sentences as possible, trying to understand the contents. The recall session was conducted in the same manner as in the auditory presentation.

Scoring

Scoring was done in the same manner as in Experiment II by the author and a research assistant. Of the 699 sentences recalled, three showed a dis- crepancy between these two persons and were discarded from the final anal- ysis. Thus the interscorer agreement is assumed to have been .99.

Results

Table II shows the maximums, minimums, means, and standard devi- ations for the ratings of the far-literal, far-figurative, close-literal, and close- figurative contexts, respectively. It seems that, both for the auditory and visual presentations, the scores for the far materials were higher than those for the close materials. Using the score obtained from each subject, a one- way ANOVA was conducted across the four categories of contexts. A sig-

Table II. Maximums, Minimums, Means, and Standard Deviations of Recall Scores in Experiment III

Auditory Visual

Max. Min. Mean SD Max. Min. Mean SD

Far-literal 10.0 4.0 7.9 2.2 10.0 4.0 6.5 1.8 Far-figurative 9.0 0.0 7.2 1.9 10.0 2.0 7.0 1.8 Close-literal 9.0 2.0 4.8 1.8 9.0 2.0 5.1 2.1 Close-figurative 10.0 4.0 6.6 2.2 10.0 4.0 5.5 2.5

Japanese Idioms 667

nificant difference was observed for the auditory presentation [F(3, 89) = 9.917, p < .01], and the post hoc test showed significant differences between the far-literal and close-literal, the far-literal and close-figurative, the far- figurative and close-literal, and the close-literal and close-figurative contexts. For the visual presentation also, the ANOVA showed a significant difference [F(3, 103) = 5.020, p < 0.01]. The post hoe test showed significant differ- ences between the far-literal and close-literal, between the far-figurative and close-figurative, and between the far-figurative and close-literal contexts. These facts indicate that, both for the auditory and visual presentations, semantically far idioms were recalled more accurately than semantically close idioms.

E X P E R I M E N T I V

Materials

Five far and five close idioms used in Experiment II plus an additional five far and five close idioms were employed. The latter 10 idioms were extracted from the idioms used in Experiment I. The means of semantic distance were 2.22 for the 10 far idioms and 3.75 for the 10 close idioms, and a significant difference was observed between these mean values (t = 12.439, df= 781,p < .01). Note that all these 20 idioms showed over 90% responses, but some nouns and verbs were used repeatedly. These test ma- terials were presented in the original syntactic form (noun + particle + verb).

For the auditory presentation, the same speaker employed in the pre- vious experiments was asked to pronounce each material at a slow rate of speech, and her utterances were recorded. The recorded utterances were checked by the author and a research assistant but no rerecording was re- quired. The final test tape was made randomizing the 20 materials with 5-s intervals in between. Here, in order to see how comprehension develops, the memory and recall trial was repeated three times, and the randomization of the materials was made to differ across the three trials. For the visual pres- entation, the randomized materials (in the same order as in the auditory presentation throughout the three trials) were made to appear on the CRT screen. Here, the presentation time o f each material was made to match exactly the duration of the recorded utterance. For this, the recorded utter- ances were processed acoustically: They were low-pass filtered at 7 kHz and were digitized by a 12-bit analog-to-digital (A/D) converter at a sampling rate of 16.6 kHz. The duration of each material was measured to the nearest 1 ms by manipulating the computer-controlled cursor and by listening to the

668 Miura

sounds from a 12-bit D/A converter. The presentation time ranged from 728 to 1230 ms, the mean value being 978 ms.

Subjects

Thirty-four university students in one class were employed, none of whom had hearing or visual disorders and none of whom had participated in the previous experiments. They were randomly divided into two groups (17 persons each): One group was allotted to the auditory presentation and the other group to the visual presentation.

Procedures

The procedures were almost the same as in Experiment II. One change was that the memory and recall trial was repeated three times in succession, and therefore the entire experiment lasted approximately 15 min. The other change was that the subjects in the visual presentation were told they would see each material appearing on the CRT only once. For this, they were trained to conduct a one-time fixation using example phrases.

Scoring

Scoring was done in the same manner as in Experiment II by the author and a research assistant. Of the 1,516 materials recalled, 14 showed a dis- crepancy between these two persons and were discarded from the final anal- ysis. Thus the interscorer agreement is assumed to have been .99.

Results

Table III shows the maximums, minimums, means, and standard de- viations for the ratings of the auditory and visual presentations, respectively.

Table III. Maximums, Minimums, Means, and Standard Deviations of Recall Scores in Experiment IV

Auditory Visual Trial number Max. Min. Mean SD Max. Min. Mean SD

1 Far 16.0 6.0 10.7 3.0 17.0 4.0 9.8 3.3 Close 15.0 4.0 9.9 3.1 16.0 2.0 7.8 3.7

2 Far 20.0 11.0 15.9 2.7 18.0 10.0 14.6 2.4 Close 20.0 10.0 15.9 2.5 20.0 6,0 14.9 4.1

3 Far 20.0 16.0 18.5 1.3 20.0 14.0 18.1 1.8 Close 20.0 14.0 18.1 2.2 20.0 14.0 18.7 2.1

Japanese Idioms 669

This table reveals that the recall scores increase as the trial progresses, which indicates a developmental trend in comprehension. In order to verify this, a one-way ANOVA was conducted across the three trials, which showed sig- nificant differences for the far materials in the auditory presentation [F(2, 50) = 43.343, p < .001], the close materials in the auditory presentation [F(2, 50) = 42.334, p < .001], the far materials in the visual presentation IF(2, 50) = 42.313, p < .001], and the close materials in the visual pres- entation IF(2, 50) = 43.561, p < .001]. Also, the far materials were recalled more accurately than the close materials in the first trial, whereas recall scores differed little between the far and close materials in the second and third trials. A t-test was conducted between the far and close materials for each of the six pairs, but none of them reached the significant level.

G E N E R A L DISCUSSION

The results of Experiment I showed that, although syntactic and lexical restrictions were required in singling out the materials, the distribution of semantic distance ranged from low to high. Meanwhile, the percentage of responses showed high scores for most of the materials: Twelve out of 35 idioms show 100%. The mean percentage of responses obtained here (93) coincided fairly well with the 95% responses obtained from the results of English idioms (Shweigert & Moates, 1988). The percentage of responses may represent one aspect of familiarity, since the subjects were not allowed to respond when they were unaware of the figurative meanings. A correlation coefficient was calculated between the semantic distance and the percentage of responses, which yielded a low negative value (r = - .40) . Thus it can be said that semantic distance and familiarity were independent. Notice that familiarity in the current study meant the familiarity with an idiom's figu- rative meaning, while in studies of English idioms word frequency was independent of familiarity (Cronk, Lima, & Schweigert, 1993).

The results of Experiment II showed that, in both the auditory and visual presentations, the far idioms were recalled more accurately than the close idioms. This is compatible with the results related to English idioms (Mueller & Gibbs, 1987), although Mueller and Gibbs used response laten- cies as a measure of comprehension. The reason for this tendency may be that all words or idioms are not stored alike in the mental lexicon, and words or idioms with several meanings are accessed faster than those with fewer meanings (Jastrzembski, 1981). In other words, the far idioms include double the amount of information compared to the close idioms. This is confirmed by the fact that the control materials having only the literal meaning were recalled as accurately as the close materials. Following Gibbs, Nayak, and

670 Miura

Cutting (1989) and Reagan (1987), semantic distance correlated with syn- tactic variability. Recall that in the present study, the target sentences were presented in the form of past-tense, active sentences. Unlike English idioms, Japanese idioms are seen to be syntactically rigid: They are rarely used in the passive voice or in cleft constructions even when appropriate contexts are provided. However, it remains to be seen whether the results would have differed if target sentences having other syntactic constructions had been used. Comparing the auditory and visual presentations, more accurate recall is seen to have been achieved by the latter than by the former. In order to verify this, a t-test was conducted between the auditory and visual presen- tations for the three categories of materials, respectively, and a significant difference was observed only in the far materials (t = 2.047, df = 63, p < .05). Notice, however, that the subjects in the visual presentation had the possibility of seeing each material more than once.

The results of Experiment III showed that, in both the auditory and visual presentations, the materials in the far category were recalled more accurately than those in the close category. In order to verify this, a t-test was conducted between the far and close categories, with the literal and figurative categories combined. Significant differences were observed both for the auditory presentation (t = 4.362, df = 88, p < .01) and the visual presentation (t = 3.671, df = 102, p < .01). This is incompatible with the study of Gibbs (1987), where children were given questions after listening to stories. It remains to be seen whether the discrepancy was caused by the experimental method (recall or question) or the age of the subjects (adults or children). When the far and near categories were combined, a t-test showed no significant difference between the literal and figurative categories both for the auditory and visual presentations. This is compatible with the study of Shinjo and Myers (1987), which suggests that the comprehension procedures are mostly the same for the literal and figurative meanings of English metaphors. Comparing the auditory and visual presentations, more accurate recall is seen to have been achieved by the former than by the latter. This is a reverse tendency from Experiment II. In order to verify this, a t-test was conducted for the four categories of contexts, respectively, and a significant difference was observed only in the far-literal context (t = 2.337, df = 42, p < .05). Notice, however, that the four groups of the subjects came from different classes although they belonged to the same university.

The results of Experiment IV showed that the recall scores increased as the trial progressed. Recall, however, that the scores between the far and close idioms, and between the auditory and visual presentations differed little in the third trial. This may mean that the scores came close to the ceiling, that is, the maximum point. A step-by-step investigation of individual data

Japanese Idioms 671

suggests that approximately one-half of the subjects got the maximum point in the third trial. In contrast, no subject reached the maximum point in the first trial. Recall that although the far idioms were seen to be recalled more accurately than the close idioms, especially in the visual presentation, no significant difference was observed. This may be due to the fact that the number of subjects in Experiment IV was somewhat smaller than the number in Experiment II. When the far and close categories were combined, a t-test showed a marginal difference between the auditory and visual presentations in the first trial (t = 1.823, df = 66, p < .1). Also, when the auditory and visual presentations were combined, a t-test showed a marginal difference between the far and close idioms in the first trial (t = 1.748, df= 66,p < .1). Notice that the two groups of subjects are seen to have been equal since they came from the same class, and that the subjects in the visual presen- tation were assumed to conduct a one-time fixation. Thus it can be said that, in the first trial, more accurate recall was achieved by the auditory presen- tation than by the visual presentation, and the far idioms were recalled more accurately than the close idioms.

Visual presentation preceded auditory presentation in isolation (Exper- iment II), whereas auditory presentation preceded visual presentation in con- text (Experiment III). Experiment IV confirmed the precedence of visual presentation over auditory presentation in isolation. The precedence of visual presentation over auditory presentation has also been observed in the recall of Chinese sentences (Ho & Chen, 1993). Notice that Japanese orthography includes Kanji, similar to the logographic writing system of Chinese, and that this writing system activates encoding strategies which emphasize visual codes (Chen & Juola, 1982). One possible reason for the precedence of auditory presentation over visual presentation in context may lie in the re- corded utterances used for the auditory presentation. Recall that in Experi- ment III three materials were rerecorded owing to a diverse manner of utterance. In fact, the three target sentences in the figurative-induced con- texts were uttered rather idiomatically in the original recording. That is, noun + particle + verb constructions were pronounced as if they constituted groups, with a somewhat fast rate of speech and unclear articulation. This manner of utterance is considered appropriate for manifesting the figurative meaning. However, in order to make the utterance manner uniform, these three utterances were replaced by clearer ones. Thus the results would have differed if unclear original utterances had been used. Another possible reason for the precedence of visual presentation in context may lie in the appro- priateness of context. Indeed, context facilitates comprehension, as men- tioned in the Introduction, but its own comprehension requires much load on the memory. Thus context effect would have appeared more clearly if two-sentence-long contexts had been used. Another factor to be considered

672 Miura

is the recal l me thod used as a measure o f comprehens ion . This me thod m a y be appropr ia te in the case o f isolat ion, but o ther methods such as q u e s t i o n - answer or l a tency t ime measu remen t might be more appropr ia te for assess ing context comprehens ion .

Conclusion

The present s tudy has r evea led that, both for the audi tory and visual presenta t ions , semant ica l ly far id ioms show more accurate comprehens ion

than semant ica l ly c lose idioms. It has also r evea led that, in isolat ion, visual comprehens ion precedes audi tory comprehens ion . Further , it has been found that the results ob ta ined f rom other studies us ing the Engl i sh language in genera l co inc ide fa i r ly wel l wi th the present results us ing the Japanese lan-

guage. Fur ther s tudies should focus on how Japanese chi ldren acquire id io-

mat ic express ions f rom the p roduc t ive as wel l as the comprehens ive v iew- points . It should also focus on how Japanese learners o f Engl i sh comprehend Engl i sh id ioms, s ince the t ransla t ion into another l anguage includes diffi- cult ies even for b i l inguals (Nelson, 1992). Further , it should be concerned with f igurat ive language other than idioms.

A P P E N D I X I

Materials used in Experiment II are shown with their English translations of literal and figurative meanings

Idiom Literal meaning Figurative meaning

Far ashi o arau wash the foot reform onself mizu o sasu pour water make mischief kimo o tsubusu mash the liver amazed hone o oru break the bone make efforts kata o motsu grasp the shoulder take sides

Close kubi o hineru twist the neck look doubtful te o dasu reach the hand meddle fude o oku lay down the brush cease me o samasu wake from sleep undeceived yume o miru have a dream take a fancy

Control fuku o kau buy a dress e o kaku draw a picture ha o migaku brush teeth hon o yomu read a book kutsu o haku wear shoes

Japanese Idioms 673

A P P E N D I X II

Original Japanese materials and their English translations used in Ex- periment III are shown. The target sentences (to be recalled) are in the middle position.

List A

Far-Literal

1. Taro-wa hadashi-de aruki-mashita. Kare-wa ashi-o arai-mashita. Soshite heya-ni hairi-mashita. Roy walked barefooted. He washed his feet. Then he entered the room.

2. Taro-wa soba-o yugaite-imashita. Kare-wa mizu-o sashi-mashita. Soshite hi-o tome-mashita. Roy was cooking the noodles. He poured water. Then he lowered the flame.

3. Taro-wa ryori-o tsukutte-imashita. Kare-wa kimo-o tsubushi-mash- ita. Soshite nabe-ni ire-mashita. Roy was cooking. He mashed the liver. Then he put it into a pan.

4. Taro-wa jitensha-de korobi-mashita. Kare-wa hone-o ori-mashita. Soshite byoin-e iki-mashita. Roy fell off his bicycle. He broke a bone. Then he was carried to the hospital.

5. Taro-wa jyudo-o shite-imashita. Kare-wa kata-o mochi-mashita. Soshite okiku nage-mashita. Roy was playing Judo. He grasped his opponent's shoulder. Then he threw him down.

Close-Figura tive

1. Taro-wa kuruma-no kagi-o nakushi-mashita. Kare-wa kubi-o hineri- mashita. Sohite iejyu-o sagashi-mashita. Roy lost the key of his car. He looked doubtful. Then he searched all over the house.

2. Taro-wa kabu-no hanashi-o kiki-mashita. Kare-wa te-o dashi-mash- ita. Soshite binbo-ni nari-mashita. Roy heard about stocks. He meddled with it. Then he lost much money.

3. Taro-wa shigoto-o shite-imashita. Kare-wa fude-o oki-mashita. Soshite asobi-ni dekake-mashita. Roy was doing his work. He ceased work. Then he went to play.

674 Miura

4. Taro-wa kiraku-ni kurashite-imashita. Kare-wa me-o samashi-mash- ita. Soshite majime-ni hataraki-mashita. Roy was living in comfort. He was undeceived. Then he began to work hard.

5. Taro-wa tenshoku-o kanga-mashita. Kare-wa yume-o mi-mashita. Soshite jitsugen sase-mashita. Roy thought o f changing his job. He took a fancy. Then he actu- alized it.

List B

Close-Literal

1. Taro-wa ragubi-o shite-imashita. Kare-wa kubi-o hineri-mashita. Soshite shippu-o hari-mashita. Roy was playing Rugby. He twisted his neck. Then he put a wet compress on it.

2. Taro-wa hitsuji-ni dekuwashi-mashita. Kare-wa te-o dashi-mashita. Soshite ke-ni sawari-mashita. Roy encountered a sheep. He reached out his hand. Then he touched its wool.

3. Taro-wa shuji-o shite-imashita. Kare-wa fude-o oki-mashita. Soshite sumi-o suri-mashita. Roy was doing calligraphy. He laid down his brush. Then he rubbed down an ink stick.

4. Taro-wa monooto-ni kizuki-mashita. Kare-wa me-o samashi-mash- ita. Soshite akari-o tsuke-mashita. Roy noticed a strange sound. He woke from his sleep. Then he turned on the light.

5. Taro-wa hirune-o shite-imashita. Kare-wa yume-o mima-shita. Sosh- ite negaeri-o uchi-mashita. Roy was taking a nap. He had a dream. Then he shifted sleeping positions.

Far-Figurat ive

1. Taro-wa dorobo-o shite-imashita. Kare-wa ashi-o arai-mashita. Soshite shigoto-ni hagemi-mashita. Roy was committing theft. He reformed himself. Then he worked hard.

Japanese Idioms 675

2. Taro-wa tomodachi- to hanashi-te i-mashita. Kare-wa mizu-o sashi-

mashita. Soshite naka-ga wakuru nari-mashita.

Roy was talking with his friends. He made mischief. Then he parted from them.

3. Taro-wa ooki-na daibutsu-o mi-mashita. Kare-wa kimo-o tsubushi- mashita. Soshite shashin-ni tori-mashita.

Roy saw a big Buddha. He was amazed. Then he took a picture of

it.

4. Taro-wa konnan-na koto-ni deai-mashita. Kare-wa hone-o orimash- ita. Soshite umaku yaritoge-mashita.

Roy faced a difficult task. He made efforts. Then he accomplished

it.

5. Taro-wa haha-towa nakayoshi deshita. Kare-wa kata-o mochi-mash- ita. Soshite chichi-o seme-mashita.

Roy got along well with his mother. He took sides with her. Then he b lamed his father.

R E F E R E N C E S

Broderick, V. (1991). Young children's comprehension of similarities underlying meta- phor. Journal o f Psyeholinguistic Research, 20, 65-81.

Cacciafi, C., & Levorato, M. C. (1989). How children understand idioms in discourse. Journal of Child Language, 16, 387-405.

Cacciari, C., & Tabossi, P. (1988). The comprehension of idioms. Journal of Memory and Language, 27, 668 683.

Chen, H.-C., & Juola, J. F. (1982). Dimension of lexical coding in Chinese and English. Memory & Cognition, 10, 126-224.

Clevenger, T., & Edwards, R. (1988). Semantic distance as a predictor of metaphor selection, Journal o f Psycholinguistic Research, 17, 211-226.

Cronk, B. C., Lima, S. D., & Schweigert, W. A. (1993). Idioms in sentences: Effects of frequency, literalness, and familiarity, Journal of Psycholinguistic Research, 22, 59- 82.

Ezell, H. K., & Goldstein, H. (1991). Comparison of idiom comprehension of normal children and children with mental retardation. Journal o f Speech and Hearing Re- search, 34, 812-819.

Gibbs, R. W. (1985). On the process of understanding idioms. Journal o f Psycholinguistic Research, 14, 465-472.

Gibbs, R. W. (1987). Linguistic factors in children's understanding of idioms. Journal of ChiM Language, 14, 569-686.

Gibbs, R. W., Nayak, N. P., & Cutting, C. (1989). How to kick the bucket and not decompose: Analyzability and idiom processing. Journal o f Memory and Language, 28, 576 593.

Ho, C. S., &Chen, H. (1993). Effects of syntactic structure in the memory of concrete and abstract Chinese sentences, Journal o f Psycholinguistic Research, 22, 505-518.

Jastrzembski, E. (1981). Multiple meanings, number of related meanings, frequency of occurrence, and the lexicon. Cognitive Psychology, 13, 273-305.

676 Miura

Kats, A. N., Paivio, A., & Marschark, M. (1985). Poetic comparison: Psychological dimensions of metaphoric processing, Journal of Psycholinguistic Research, 14, 365-383.

Kroll, N. E. A., & Schepeler, E. M. (1987). The comprehension and recall of similes as a fimction of context and cue. Journal o f Psycholinguistic Research, 16, 101-132.

Marsehark, M., & Hunt, R. (1985). On memory for metaphor. Memory & Cognition, 13, 413-424.

Mueller, R. A. G., & Gibbs, R. W. (1987). Processing idioms with multiple meanings, Journal o f Psycholinguistic Research, 16, 63-81.

Nelson, E. M. M. (1992). Memory for metaphor by nonfluent bilinguals. Journal of Psycholinguistic Research, 21, 111-125.

Nippold, M. A., Leonard, L. B., & Karl, R. (1984). Syntactic and conceptual factors in children's understanding of metaphors. Journal o f Speech and Hearing Research, 27, 197-205.

Nippold, M. A., & Martin, S. T. (1989). Idiom interpretation in isolation versus context: A developmental study with adolescents. Journal o f Speech and Hearing Research, 32, 59-66.

Pearson, B. Z. (1990). The comprehension of metaphor by preschool children. Journal o f Child Language, 17, 185-203.

Pollio, H. R., Fabrizi, M. S., Sills, A., & Smith, M. K. (1984). Need metaphoric com- prehension take longer than literal comprehension? Journal o f Psycholinguistic Re- search, 13, 195-214.

Reagan, R. T. (1987). The syntax of English idioms: Can the dog be put on? Journal o f Psycholinguistic Research, 16, 417~41 .

Schraw, G., Trathen, W., Reynolds, R. E., & Lapan, R. T. (1988). Preferences for idioms: Restrictions due to lexicalization and familiarity. Journal o f Psycholinguistic Re- search, 17, 413-424.

Schweigert, W. A. (1986). The comprehension of familiar and less familiar idioms. Jour- nal o f Psycholinguistic Research, 15, 33-45.

Schweigert, W. A., & Moates, D. R. (1988). Familiar idioms' comprehension. Journal of Psycholinguistic Research, 17, 28 i-296.

Shinjo, M., & Myers, J. L. (1987). The role of context in metaphor comprehension. Journal o f Memory and Language, 26, 226-241.

Tompkins, C. A. (1990). Knowledge and strategies for processing lexical metaphor after right or left hemisphere brain damage. Journal o f Speech and Hearing Research, 33, 307-316.

Williams-Whitney, D., Mio, J. S., & Whitney, P. (1992). Metaphor production in creative writing. Journal o f Psycholinguistic Research, 21, 497-509.