A Second Look at the Word-in-Isolation and Word-in-Context Tasks

Considering the inconsistency in the within-group difference listed in Table 4-9, we found there was a within-group significant difference in G1, G2, G3 and G5, and yet no significant difference in G4 was observed. Thus, a further examination of the data is required. As discussed in the previous issue concerning the type of idiom, transparency of chou-related idioms was found to have a significant impact on the ² results. After we removed the frequency counts of semantically transparent idioms, we could have a revised look at the data, as drafted in Table 4-10 below:

Literal Metaphorical Literal Metaphorical

G1 54 4 49 36 25.600 1 .000^**

In accordance with the results of the ² test, all experimental groups and the adult control

in-Context task than in the Word-in-Isolation task (G1: n in WIC= 36 > n in WII= 4; G2: n in WIC= 63 > n in WII =11; G3: n in WIC= 77 > n in WII= 44; G4: n in WIC= 93 > n in WII=

69; G5: n in WIC= 104 > n in WII= 79; Adult: n in WIC= 115 > n in WII= 97). There was a significant within-group difference in G1, G2 and G3, indicating there was a marked improvement in their metaphorical frequency when aided by the contextual effect.

4.4.3 General Discussion

According to the results of the ² test, there was a significant interaction between the four

interpretational categories and the two different approaches (p=.000). As a result, the metaphorical frequency counts in the Word-in-Context task were significantly more than those obtained in the Word-in-Isolation task. The frequency counts for the other category (the participants responding with ‘I don't know’) in the Word-in-Isolation task also decreased (n in WII= 200 > n in WIC= 74). From Table 4-10, we can find that all six groups had more metaphorical interpretations in the Word-in-Context task than the Word-in-Isolation task, supporting the fact that context plays an essential role in idiom comprehension.

The contextual effect has been extensively discussed in previous literature in order to investigate how children, adolescents and adults comprehend idiomatic expressions (Ackerman 1982, Gibbs 1987, Nippold & Martin 1989, Levorato & Cacciari 1992, 1995, 1999, 2002, Laval 2003, Hsieh 2008, Cain et al. 2009, Hsieh & Hsu 2010, Hsu 2013). Nevertheless, various approaches have been employed to measure participants’ idiom comprehension.

For example, in Levorato & Cacciari (1995), both a multiple choice task design (as a form of context) and a paraphrase task design were adopted to observe the complexity of two different approaches. As reported in the previous literature (Ackerman 1982, Prinz 1983), a multiple choice task design could facilitate correct interpretations from participants since this design helped them to recognize idiomatic interpretations with greater accuracy. However, Levorato & Cacciari found that these facilitating factors, contrary to prior research, did not result in a significant effect on participants' comprehension; in other words, both task designs were of equal difficulty for participants, and neither led to a significantly greater amount of correct interpretations. As long as the participants were provided with substantial informational context, in the paraphrase task, they were capable of producing correct interpretations.

Multiple choice task design was implemented in the pilot study, but the options were found to mislead our participants’ interpretations. In addition, the present study included

quadra-syllabic idioms as a part of the task design. Most quadra-quadra-syllabic idioms in Chinese are semantically opaque idiomatic expressions and it is relatively difficult to design a literal meaning into the options since the storyline is embedded in the Word-in-Context task. Thus, the present study attempts to adopt interpretation tasks in the design of both experimental tasks.

In this way, possible facilitating factors in multiple choice tasks can be avoided and it becomes easier to directly assess each participant’s idiom comprehension.

In light of the above, the present study only employs interpretation tasks and not multiple

choice tasks as the major measure. Although previous literature has emphasized the fact that younger children lack the necessary metalinguistic awareness to properly interpret metaphorical meanings (Hsu 2013), according to Hsieh & Hsu (2010) however, children whose mean age was six years old were able to adequately complete an interpretation task. For the preschoolers (who were of age six) in the present study, if we further look at this group’s frequency in the other category within both tasks, their performance improved and the frequency counts for this category in the interpretation task sharply diminished after a contextual factor was added into the task by means of the Word-in-Context design (G1: n in others in Word-in-Isolation = 68 > n in others in Word-in-Context = 25). In addition to this, based on the collected data, G1 was capable of explaining the idiomatic meaning of transparent idioms once context was a factor. Take for example the transparent quadra-syllabic

chou-related idiom chou-qi-chong-tian ‘a scent stinking to high heaven.’ Contrary to frequent responses to the other category seen under the Isolation design, in the

Word-in-Context task, most children in G1 were able to produce the correct metaphorical interpretation

“the smell in the bathroom is too disgusting and nobody in the restaurant can stand such a foul odor.”

Similar findings were found in Gibbs (1987). In Gibbs’ study, the kindergarteners and first graders had the ability to map the meanings of transparent idioms to their metaphorical domain.

Gibbs further stated that “children more easily understood metaphorically transparent idioms

because they metaphorically extend their literal meanings to other contexts to create their idiomatic interpretations” (Gibbs 1987:581). G1 in the present study was able to metaphorically

extend the literal meaning of chou to express one’s inner feeling towards unpleasant smells.

This result therefore demonstrates the relation between semantic transparency and context.

Leverato (1999:63) also found that “context plays an important role, particularly in the early phases of idiom acquisition, with an effect which adds to that of semantic transparency.”