The fourth research question explores whether the acceptability ratings of frequency adverbs modifying events, especially ET2-1, will be raised with the support of contextual clues. Section 4.4.1 reports the subjects’ overall acceptability ratings for frequency adverbs in the context-free and context-provided tasks. Section 4.4.2 discusses the effects caused by the contextual clues.
87
4.4.1 Overall Findings
Table 4-13 demonstrates the mean scores of events in which frequency adverbs occurred in the context-free and context-provided tasks. As indicated in Table 4-13, the subjects accepted frequency adverbs when they modified events in the context-free task (M = 1.18) significantly better than in the context-provided task (M = 0.96) (F (1,1998)
= 31.65, p < .001).
Table 4-13 Subjects’ Overall Acceptability of Frequency Adverbs Modifying Events in the Context-free and Context-provided Tasks
Context M SD F p-value
Context-free 1.18 0.89 31.65 2.11e-08 ***
Context-provided 0.96 0.90
Note: *** indicates that the p-value is smaller than .001.
Figure 4-5 presents each group’s acceptability ratings for events where frequency adverbs occurred in the context-free and context-provided tasks. Like what we have found in Sections 4.2 and 4.3, only the kindergartners responded differently from the other groups’.
88
Figure 4-5 Each Group’s Acceptability Ratings for Frequency Adverbs in the Context-free and Context-provided Tasks
Likewise, one-way ANOVA indicated that there was a significant difference among the five groups in the two tasks (Context-free: F (4,995) = 2.82, p < .05; Context-provided:
F (4,995) = 7.11, p < .001). When we look at each group’s acceptability ratings, only the kindergartners gave higher ratings to the events in which frequency adverbs
occurred in the context-provided task (M = 1.16) than in the context-free task (M = 1.06). The other groups accepted frequency adverbs when they modified events
in the context-free task better than those in the context-provided task (Grade 2: Mcf = 1.23 > Mcp = 1.12; Grade 4: Mcf = 1.12 > Mcp = 0.87; Grade 6: Mcf = 1.18 > Mcp = 0.80;
Control: Mcf = 1.33 > Mcp = 0.85). As for each age group’s acceptability of frequency adverbs in the two tasks, the within-group differences are provided in Table 4-14.
Kindergartner Grade 2 Grade 4 Grade 6 Control
Context-free 1.06 1.23 1.12 1.18 1.33
C-f SD 0.75 0.84 0.9 0.93 0.88
Context-provided 1.16 1.12 0.87 0.8 0.85
C-p SD 0.88 0.86 0.88 0.95 0.89
89
Table 4-14 The p-values for the within-group Differences between the Context-free and the Context-provided Tasks
Kindergartner Grade 2 Grade 4 Grade 6 Control
F 1.40 1.82 8.18 16.44 29.97
p .24 .18 .004** .000*** .000***
Note: ** indicates that the p-value is smaller than .01 and *** is smaller than .001.
A significant difference was found in the older children and the control group (Grade 4:
F (1,398) = 8.18, p < .01; Grade 6: F (1,398) = 16.44, p < .001; Control: F (1,398) = 29.97, p < .001), indicating that they accepted frequency adverbs when they modified events in the provided task in a significantly lower degree than in the context-free task. That is, the contextual clues reduced their acceptability ratings. On the other hand, the younger children did not respond significantly differently to the two tasks (Kindergartner: F (1,398) = 1.40, p > .05; Grade 2: F (1,398) = 1.82, p > .05), indicating that they did not benefit from contextual clues.
Regarding the between-group differences, Table 4-15 shows the p-values for the differences in the acceptability of frequency adverb in the two tasks.
90
Table 4-15 The p-values for the between-group Differences in the Acceptability of Frequency Adverbs in the Context-free and Context-provided Tasks
Grade 2 Grade 4 Grade 6 Control
C-F C-P C-F C-P C-F C-P C-F C-P
Kindergartner .35 1.00 1.00 ** .96 *** * **
Grade 2 - - .98 * 1.00 ** 1.00 *
Grade 4 - - - - 1.00 1.00 .17 1.00
Grade 6 - - - .64 1.00
Note: * indicates that the p-value is smaller than .05, ** is smaller than .01 and *** is smaller than .001.
The post hoc comparison showed that in the context-free task, only the adult controls responded significantly differently from the kindergartners (p < .01). The other comparisons of between-group differences did not reach a significant difference (p
> .05). As for the context-provided task, the adult controls, Grade 6 and Grade 4 showed significantly different responses from Grade 2 (p < .05, p < .01, p < .05, respectively) and the kindergartners (p < .01, p < .001, p < .01, respectively). However, Grade 2 did not respond significantly differently from the kindergartners (p > .05).
In addition to the general acceptability ratings for frequency adverbs modifying events, Zou (2007) argues that some atelic events with a verb like xiang ‘think’ become telic once they receive an endpoint from the outside. Therefore, ET2-1 is examined to see whether contextual clues, which include an endpoint, will turn atelic events into being able to be modified by frequency adverbs. Table 4-16 presents the overall
91
acceptability ratings for frequency adverbs when they modified ET2-1 in the two tasks.
Table 4-16 Overall Acceptability Ratings for Frequency Adverbs Modifying ET2-1 in the Context-free and Context-provided Tasks
Event Type Context M SD F p-value
ET2-1 Context-free 1.10 0.87 25.74 .000***
Context-provided 0.66 0.84
Note: *** indicates that the p-value is smaller than .001.
As shown in the table above, the acceptability ratings for frequency adverbs in the context-free task (M = 1.10) was higher than those in the context-provided task (M = 0.66), and the difference reached a significant level (p < .001). In other words, the acceptability rating for frequency adverbs in ET2-1 was significantly lower with the support of contextual clues.
4.4.2 General Discussion
The role of contextual clues has been widely discussed in a variety of acquisition studies (e.g., Beyth-Marom 1982, Schatz and Baldwin 1986, Stanovich 1986, Cacciari and Levorato 1989, Qian 1996, Walqui 2000, Friedrich and Friederici 2004, Belke et al.
2005, Hsieh 2008, Hoff 2010, among others). Most of them concluded that the subjects had better comprehension with the assistance of contextual clues. However, there are a few studies arguing against the positive effect of contextual clues. For instance, Schatz
92
and Baldwin (1986) conducted experiments to examine how contextual clues helped students when they inferred the meanings of unknown words and they concluded that the effect of contextual clues had been vastly overestimated and that contextual clues did not provide enough clues for the subjects to comprehend low-frequency words.
They even claimed that contextual clues were unreliable predictors of word meanings.
The findings of the present study indicated that contextual clues did not generally help our subjects accept frequency adverbs when they modify events. For the younger children (the kindergartners and Grade 2), providing contextual clues neither raised nor lowered their acceptability ratings. As for the older children (Grades 4 and 6) and the adult controls, the acceptability of frequency adverbs in the context-free task became higher than those in the context-provided task with a significant difference. The results demonstrated that contextual clues actually exerted bad influence on the older children and the controls. There are two possible reasons for this phenomenon. First, some people may consider the event nong po dongxi ‘to broke something into pieces’
happening twice in a week as ouer ‘occasionally’ but other people may consider it as changchang ‘usually.’ In other words, the subjects may take the frequency into
consideration based on their subjectivity, and individual differences in frequency can be a factor. Second, lexical meanings of verbs may be a factor as well. For instance,
93
although the two events nong po dongxi ‘to broke something into pieces’ and xizao ‘to take a shower’ happened to someone twice a week, Taiwanese people would not use the same frequency adverb to describe the frequencies, especially in hot summer days.
Regarding the acceptability of frequency adverbs in ET2-1, the results indicated that the acceptability ratings of frequency adverbs in the context-provided task (i.e., with the added endpoints) were actually significantly lower than those in the context-free task (i.e., without the endpoints). Therefore, the argument of Zou (2007) is problematic in that other elements are necessary to make some atelic events acceptable to be modified by frequency adverbs.