Formal Test

After the pilot study, several revisions of the questionnaire were made. Seeing that the performances of the mid and high groups were quite similar in the pilot study, we then included only two experimental groups--mid²⁶ and high groups-- in the present study. Furthermore, in light of the avoidance of the RVC found in the pilot study, the event description was dropped out of the formal test. Consequently, the final version of the questionnaire used for the formal test was composed of two tasks, the GJ and SI tasks. In the formal test, before the subjects started to complete the questionnaire, they were told to sign the consent form and fill out their background information. Then, some instructions were given in spoken Chinese or if they preferred, in English. Finally, the participants were reminded not to go back and change their answers that had already been written down in the tasks. The data coding and the type of statistics employed in our data analysis are reported below.

With respect to the scoring, in the GJ task, for each test item, those who provided an ‘X’ to mark the ungrammaticality or the combinatory infelicity of an RVC and a durative linguistic expression and then underlined the part of the sentence that was problematic were given one point. Notice also that test items that were marked ‘X’

with no indication of the part of the sentence that was problematic were not given the point. In the SI task, the learners’ answers were grouped into the following categories:

26 The low group in the pilot study was labeled as the mid group in the present study. Because the learners in this group had an average of 1 year Mandarin learning experience and were named Beginner-High, Intermediate-Low and Intermediate learners at the MTC, it is reasonable to label these learners as mid-level subjects in this study.

(1) an action reading and (2) a result reading.²⁷ Results obtained from the experimental group were categorized as frequencies and then compared to those from the control group in respect to the frequencies of occurrence for the action and result responses to each test item. It should also be noted that in the data we collected, all our subjects chose either the action interpretation or the result interpretation in the test questions without the adverb chayidianr ‘almost’ while in test items with the adverb chayidianr, some of our subjects chose the option irrelevant to the interpretation to the

target structure. To find out if questions with the adverb chayidianr would receive more result interpretation than those with the adverb due to the possible L1 positive transfer, as we predict in Chapter Two, we compared and contrasted only the frequency of result interpretation response in our data analysis.

After the scoring, a mixed designed ANOVA was designed and used to analyze the data collected in the GJ task. In the case where we obtained a significant F, we performed subsequent analyses to find out where in the data the significance occurred.

In the SI task, we used the Chi square analysis to examine whether there were statistically significant differences between subjects in the experimental and control groups with respect to their action and result responses to test questions. In addition,

27 The main goal of the SI task was to falsify Tai’s (2003) claim that the English-speaking people would attend more to the action part of the event than to the result part, as compared to Chinese-speaking people. We analyzed only the first (action) and the second (result) readings in the SI task. The third reading, an irrelevant reading, was not what we concerned, so test items marked with this answer by the learners were not analyzed.

the data analysis technique of Chi square was used to explore if the difference observed between the learners was statistically meaningful.

3.4 Summary

In this chapter, we introduce the experimental design of this study. In terms of the subject, there were 40 English learners of Chinese participated in the research.

They were divided into two distinct proficiency levels, each comprising 20 subjects.

The grammaticality judgment and sentence interpretation tasks were designed to disclose the learners’ knowledge of the basic properties of RVCs. The GJ task was designed to examine whether the English learners are aware of the feature that the Act-R, Sem-R or Sta-R RVC is not allowed to go with durative grammatical structures such as the imperfective marker zhengzai, and the two verbs, kaishi ‘begin’

and tingzhi ‘stop’. The SI task was used to test if the subjects know that the result interpretation expresses the main idea of the RVC sentence. In the following chapter, the results of the experiment will be reported. And, the four research purposes will be addressed with the findings interpreted.

Chapter Four Results and Discussion

In this chapter, we present the results of this study and discuss the findings with respect to the four research purposes stated in Chapter One. Section 4.1 examines the acquisition difference among the three types of RVCs. Section 4.2 explores the L1 influence on the L2 acquisition of RVCs. Section 4.3 addresses the issue on the interplay between language and though. Section 4.4 summarizes the main points of this chapter.