Cain, Towse & Knight (2009)

Cain, Towse & Knight (2009) investigated how semantic and contextual processing skills would contribute to differences in idiom comprehension in the field of first language acquisition. These two processing skills have been highlighted in previous literature regarding figurative language comprehension (Gibb 1987, Nippold & Martin 1989, Levorato & Cacciarri 1992, Nippold & Taylor 1995, Nippold 1998). Previous studies have claimed that both semantic analysis and contextual inference facilitate idiom comprehension. However, the conflicting findings from the different designs in these related studies motivated further research in Cain, Towse & Knight (2009). As stated in Cain, Towse & Knight (2009), more work should be done for clarifying the role of semantic analysis and contextual inference in children’s idiom comprehension.

As defined in Cain, Towse & Knight (2009:281), “semantic analysis of an idiomatic

phrase involves retrieving alternate meaning of key words and computing a nonliteral meaning of the phrase.” Iidioms can be categorized as semantically transparent idioms and opaque

idioms. For semantically transparent idioms, there are cues for children to successfully derive the meaning through semantic analysis. On the other hand, semantic analysis cannot serve as a trigger for semantically transparent idioms. For example, it is not possible to analyze the

components in a semantically opaque English idiom like ‘kick the bucket’ and derive the correct meaning ‘to die.’

Contextual processing skill was the second target to be investigated in Cain, Towse &

Knight (2009). When interpreting a semantically opaque or unfamiliar idiom, the inference from context proved to be useful for young children. As stated in the GEM (Levorato &

Cacciari 1995), children of age eleven and older could make use of context to comprehend figurative expressions. Especially when it was a semantically opaque idiom, children could actually depend on their coherence of the context to successfully comprehend the meaning of the opaque idiom.

The aim of Cain, Towse & Knight was to reconsider these two important processing skills in different age groups. Subjects ranging from seven to eleven years old were chosen because of the previous findings related to the developmental trend of the GEM reported in Levorato &

Cacciari (1999). In addition, related studies discuss the exact age at which semantic analysis

can be used as a processing skill by children, although disparate findings have been reported (Gibb 1987, 1991; Nippold & Taylor 1995; Levorto & Cacciari 1999).

In Cain, Towse & Knight (2009), novel idioms were adopted to test the subjects’ idiom comprehension without tapping into their memory of previously learned idioms. In previous related literature, researchers tended to use idioms which the subjects might already be familiar

with. Through the use of novel idioms, Cain, Towse & Knight aimed to exclude the factor of

‘prior knowledge regarding the idioms,’ as this knowledge highly correlates with the age factor.

Two experiments were conducted to address four research questions extending from the previous related works. The first question was to examine the contextual effect within the same group of the subjects. In contrast to a between-group design (Levorato & Cacciari 1999), Cain, Towse & Knight aimed to observe how the influence of context regarding a single idiom would affect the same children; thus, the design of Cain et al. (2009) examined a within-group performance. The second question was to compare the performance of the two age groups in comprehending transparent novel idioms when contextual inference involved. Next, the third question was to assess the subjects’ semantic analysis skills in comprehending figurative language. The last question was to determine whether comprehension differed between the adolescent and adult groups in a development period.

Experiment 1 recruited 20 seven-year-old (Mean age=7;10) children and 20 nine-year-old (Mean age=9;11) children in two village schools in northwest England to investigate whether

these two groups’ comprehension varied with semantic and contextual processing skills. In addition, 19 undergraduate school students (Mean age=19;4) from Lancaster University also participated in Experiment 1. All the subjects spoke British English as their first language.

The materials were twelve familiar British English idioms and twelve translations of European idioms which were considered as the novel idioms. The type of familiar and opaque idioms were further divided into two categories, transparent and opaque, and this division was set for the investigation of semantic analyzing skills. Regarding the inference of context, all three groups first completed the tasks without context. After a minimum of two weeks, all three groups were given the tasks with context. In total, three major variables were manipulated in Experiment 1 including the transparency of the tested idioms, the novelty of the idioms and the presence of context. The use of a multiple-choice task design was to assess the children’s idiom

comprehension, and the answers fell into four categories, idiomatic, contextually plausible, contextually implausible and literal were designed to observe the children’s interpretation of

idioms.

Based on the results in Experiment 1, the effect of context was indeed observed since the main effect was found in both the younger age group and the older age group. That is, the two younger groups showed a tendency to depend on the aid of context and selected the target interpretation. As for the novel transparent type of idioms, when context was not provided, the adult group and the older group preferred the idiomatic interpretation. That is to say, both

groups were able to use semantic analyzing skills to have the correct interpretation. The significant age factor proved to be found in the analysis of novel idioms and Cain, Towse &

Knight (2009:290) stated that this result ‘demonstrated that the language processing skills that aid idiom comprehension and acquisition are still not fully developed in 9- and 10-year-olds. ’ Experiment 2 was designed to address two issues which had not yet been solved in the previous literature. First, three more groups were recruited for the purpose of having the clearer picture of the development phase since Experiment 1 only included 7- and 9 year-olds. Thus, subjects of ages 8, 11 and 12 were recruited. Second, the work in Levorato & Cacciari (1995) suggested that the age period from 7 to 11 was a critical period for children in which they developed skills of figurative language comprehension. Nonetheless, previous works had not considered novel idioms among the tested items.

The subjects in Experiment 2 were twenty children in Grade 3 (7- or 8-year olds), twenty children in Grade 5 (9- or 10-year olds) and twenty-two children in Grade 7 (11- or 12-year olds) and all subjects studied in the same schools in northeast England. The same procedure used in Experiment 1 was conducted in Experiment 2; additionally, the task content was the same as in Experiment 1. However, an adjustment in the multiple-choice options was made to exclude literal meanings in the task. In this way, Experiment 2 eliminated the possibility that the youngest group depended on literal choice when selecting their answers.

The results in Experiment 2 were reported as follows. The t tests showed that all the groups

relied on cues of context when the subjects encountered familiar and novel idioms. Also, when context was absent, the subjects in all groups showed a tendency to derive idiomatic meanings

from semantic analysis. Based on this result, Cain, Towse & Knight (2009:295) suggested that

‘the performance of 7- and 8-year olds may have been underestimated in Experiment 1.’ It was

reported that developmental differences in the 11- and 12-year-old group were observed so knowledge of idioms were proved to continue developing beyond the primary school years.

These two groups performed well on familiar opaque idioms when the context was not provided.