Analysis for Each Research Question

11 Since there were only three cases of RR+IC, they were included in IC+RR pattern. There was no reverse order found in the other patterns.

of Chinese?

The analysis for this question was twofold. The first level of analysis concerned the quantitative and qualitative assessments of all the tokens produced by the Americans and the Chinese across the eight scenarios in the pre-instructional production questionnaires. For the quantitative analysis, each meaning unit was coded and all the tokens produced by these two groups were calculated respectively. A frequency table and a polygon were depicted to show general tendencies. Then a chi-square test was run to see if the American performances differed significantly from the Chinese performances in terms of strategy distributions.¹² If there was a statistically significant difference, further tests were conducted to examine which strategy contributed to the significance. Furthermore, the six strategies were classified as opting out/making complaints to see if the Americans and the Chinese reacted similarly when an offense occurred. Also, the strategies were classified as either on-record or off-record. According to Brown & Levinson (1987), the major difference between on-record and off-record strategies is determined by whether the communicative intention the speaker conveys is clear to the hearer. On-record strategies refer to those which express “just one unambiguous attributable intention”, while off-record strategies refer to those which may lead the hearer to have more than one interpretation. Based on this criterion, on-record complaints refer to the strategies with which the speaker not only mentions the complainable, but also holds the hearer responsible for the offensive act, while off-record complaints refer to the strategies with which the speaker may or may not mention the complainable, and does not relate the complainable to the hearer in an explicit way. Therefore, in the present study, on-record strategies include indirect accusation, direct complaint and threat because these strategies contain not only the complainable, but also the complainee.

12 The significance level was p<.05

Off-record strategy is indirect complaint because it does not contain the complainee.

Request for repair can be on-record when the speaker directly asks the hearer to do an act (e.g. “Can you…”) or off-record (e.g. “Can I…”) when the speaker only mentions his/her desires. Chi-square tests were run to see if the Americans and the Chinese differed significantly in opting out/making complaints as well as on-record/off-record taxonomies. The qualitative analysis compared the linguistic forms of each strategy used by the Americans and the Chinese. The investigation was conducted because languages often differ from each other in form-function mapping.

The second level of analysis examined how the Americans and the Chinese used two-strategy patterns. For the quantitative analysis, the number of the patterns was calculated respectively for each group of productions. A frequency table and a polygon were diagrammed to show general tendencies. A chi-square test was run to see if there was a statistically significant difference between the American and the Chinese performances in the distribution. If there was a statistically significant difference, further tests were run to see which pattern contributed to the significance.

In addition, the six patterns were classified in terms of on-record/off-record category and requestive/non-requestive category. On-record patterns refer to those whose first strategies were on-record. They included patterns such as DC+RR, IA+IC, RR+TH, and IA+IA. Off-record patterns refer to those whose first strategies were off-record, which included patterns such as IC+RR, IC+IC.¹³ Requestive complaints refer to those which contain request for repair. Patterns such as IC+RR, DC+RR and RR+TH belong to this category. On the other hand, non-requestive complaints refer to those which contain no request for repair. Patterns such as IC+IC, IA+IC and IA+IA belong

13 Trosborg (1995) points out that native speakers of English tend to prepare for and justify their complaints by using less direct strategies first to pave the way for more direct ones. Therefore, in the successive expressions of complaints, more direct strategies followed by less direct strategies sound more indirect and polite than less direct strategies followed by more direct strategies. The different degree of politeness investment is categorized through on-record and off-record dichotomy in the present study.

to this category. Chi-square tests were conducted to see if the Americans and Chinese differed significantly in these two categories. The qualitative analysis compared the content of the same strategy pattern produced by the Americans and the Chinese because it generally reflects the sociocultural information between societies (Bardovi-Harlig & Griffin, 2005).

2. To what extent does pragmatic transfer affect the learners’ complaint behaviors in American English?

Since research has shown that learners rely heavily on L1 knowledge to produce L2 targets (Olshtain & Cohen, 1989), the learners in the present study were predicted to conduct positive transfer if there were L1-L2 similarities. On the other hand, they were predicted to conduct negative transfer if there were L1-L2 differences.

Like research question 1, there were two levels of analyses. The first level of analysis compared the total number of tokens produced by the Americans, the high-proficiency learners and the low-proficiency learners. Transfer predictions were made based on the results of research question 1. For the quantitative analysis, a chi-square test was conducted to see if the learner productions matched predictions. If not, further tests were conducted to see which group or strategy contributed to the deviations.

Furthermore, all the tokens were classified as either on-record or off-record. Transfer predictions were also tested by the chi-square assessment. The qualitative analysis investigated the learner deviations from the American performances in terms of linguistic forms, and provided explanations for the deviations.

The second level of analysis dealt with two-strategy patterns. The learners’

performances were compared to the American performances. For the quantitative analysis, a chi-square test was conducted to see if transfer occurred as expected. These patterns were also classified in terms of on-record/off-record and requestive/non-requestive complaints. Chi-square tests were also conducted to see if

the learner performances matched the American performances. In addition, the learners were expected to transfer L1 sociocultural information to their productions, so a qualitative analysis was conducted to see if the content produced by the learners differed from that produced by the Americans.

In addition to the production questionnaires, the learners’ role play performances were also compared to the American performances. Like the DCT, predictions were made in terms of strategy distribution, linguistic forms and content.

However, since role play tasks were performed by only eight learners, simple descriptive statistics was employed as opposed to inferential statistics, and the qualitative analysis played a more important role.

3. What are the similarities and differences between the learners’ productions and perceptions of the DCT and role plays?

To answer this question, the analysis was composed of two parts. The first part concerned the learners’ DCT and role play productions in terms of strategy distribution, linguistic forms and content, which were also the major foci of research questions 1 and 2. The second part addressed the learners’ perceptions of the two elicitation methods. The learners’ interview protocols were transcribed verbatim and content analysis was conducted.

4. To what extent does explicit teaching affect the learners in the acquisition of complaint behaviors?

For this question, the learners’ post-instructional production questionnaires and role play performances were compared to those produced by the Americans. The quantitative analysis examined if after instruction, the learners produced native-like performances in terms of strategy distribution (including the overall tokens and the two strategy patterns), on-record/off-record strategies or patterns, and requestive/non-requestive complaints. A chi-square test was run to see if there was no

statistically significant difference between the learner and American performances. If the comparison across the three groups did not reach significance, it means that the learners approximated closely to the Americans after treatments. However, if the comparisons reached statistical significance, further tests were conducted to see which learner group contributed to the difference. Their performances before and after instruction were then investigated because progress could still be observable even though they did not produce near native performances. In addition, the qualitative analysis focused on whether the learners achieved native-like usage in terms of linguistic forms and content.

5. What are the learners’ perceptions of explicit teaching of complaint behaviors?

To answer this question, the learners’ self-reports completed after instruction were investigated. Analyses were made in terms of general impression, teaching materials and perceived difficulties.

In conclusion, details are provided in this section as to how the data collected from Phases I and III were analyzed. The coding scheme is presented first for the analysis of pre- and post-instructional DCTs and role plays. Then the steps for analyzing the data for each research question are described in detail.