Table 4-3 A Test Sample Used in the PT
The correct answer for the example is the indefinite article a, which bears the feature [+SR, -HK] and gives an indefinite reading. The NP a student in Table 4-3 belongs to the [+SR, -HK] settings since the speaker has a specific referent in mind although the referring referent is unknown to the hearer. Therefore, from the hearer’s points of view, the NP renders an indefinite reading.
4.3 Procedures
This section consists of three parts. Section 4.3.1 recapitulates the pilot study and its inadequacies. Section 4.3.2 illustrates the procedures of the formal study. The scoring and the statistical analysis of the study are reported in Section 4.3.3.
4.3.1 Pilot Study
A pilot study was conducted based on Ionin et al. (2004), attempting to verify the Article Choice Parameter Hypothesis,1 and to investigate whether Mandarin Chinese
1 Ionin et al. 2004 aims to investigate the parameter settings of (in-) definiteness. They speculate that if article acquisition is constrained by Universal Grammar (UG), article choices must be derived from parameter settings. As a result, they propose a semantic parameter - Article Choice Parameter. Based on the cross-linguistic distinctions on the article choice, Ionin et al. (2004) suggest two options for the Article Choice Parameter: (i) the definiteness setting and (ii) the specificity setting because there are languages such as English, in which article choice is constrained by the definiteness setting.
Tom calls his teacher, Ms. Lee, to talk about his term paper.
Ms. Lee: Hi, this is Susan Lee.
Tom: Hello, Miss. Lee. This is Tom. May I speak to you for a second? It’s about my term paper.
Ms. Lee: Sorry. Can you call me back in an hour? I’m meeting with _________
student from my English class. I have to give her a makeup test.
Answer: a
44
speakers preferred to choose the definite English article the in a specific environment and the indefinite English article a in a non-specific environment.
The pilot sample consisted of fifty-six students from NTNU, and they were divided into an experimental group and a control group. For the experimental group, the participants were all non-English majoring freshman from three English proficiency levels: advanced, intermediate and basic, based on their test scores on the university designed English placement test. Each proficiency level comprised fifteen subjects;
thus, the total number of the participants in the experimental group was forty-five. As for the control group, eleven native English speakers were recruited from the Mandarin Chinese Training Center of NTNU.
There were two written tasks employed in the pilot study, an acceptability judgment task and a forced choice production task. The former consisted of thirty questions in the form of single sentences, and the later was eighteen questions manifested in a dialogue. In the acceptability judgment task, the participants were asked to grade the test items from 0 to 3 (3: Very Acceptable; 2: Acceptable; 1:
Somewhat Acceptable; 0: Not Acceptable). In the forced choice production task, the test materials were systematically controlled with the priming of either [+Specific] or [–Specific] features. The participants needed to read through the dialogues and choose the appropriate English article according to the contexts.
The results of the pilot study (see Appendix C) are as follows. With regard to the production task, the accuracy rate directly corresponded to the subjects’ proficiency levels (Advanced 24%; Intermediate 33%; Beginner 41%). When comparing the accuracy rate between the [+Specific] context and the [-Specific] context, unexpected results indicated that the subjects showed higher accuracy rates in choosing the indefinite article a in the [–Definite, +Specific] context rather than the [–Definite, –Specific] context. However, the Scheffe post hoc indicated that the significances
45
were found mainly between the advanced group and the beginner group (p = .004). As for the comprehension task, the experimental group inclined to rate target sentences with the indefinite article a 2 (Acceptable) or 3 (Very Acceptable) when the context was [-Definite, -Specific], yet they did not show preference for rating the target sentences 0 (Not Acceptable) or 1 (Somewhat Acceptable) when the context was [-Definite, +Specific]. However, an increase of tendency rating the target sentences 2 or 3 was saliently formed when the context was changed from [-Definite, -Specific] to [-Definite, +Specific]. Finally, the Scheffe post hoc showed out that the significances were found between the advanced group and the intermediate group (p = .02), and the intermediate group and the native control (p = .038).
A few limitations were observed in the pilot study. To begin with, the subject pool was too small to render significant results. In addition, the test items were not properly controlled. For example, a few test materials were ambiguous. Last, the materials used in the production task included only the indefinite article use. Therefore, the pilot study did not provide a complete picture of the L2 acquisition of English articles by Chinese learners.
4.3.2 Formal Study
In light of the inadequacies in the pilot study and the goals of exploring additional variables that might affect the acquisition of English articles by Chinese learners, test formats and test materials were revised in the formal study. English articles were re-categorized to meet the analytic purpose, and the materials were systematically organized. The number of participants also increased to reach a more legitimate result.
Prior to commencing the study, ethical clearance was made to the recruited participants by asking them to sign a consent form (see Appendix D) that stated the purpose of the study and informed them that their responses would only be used for
46
academic purposes. Next, to ensure the participants’ understanding of the formats of the tasks, clear instructions and examples of the two tasks were provided. The participants were asked to begin with the comprehension task, and then proceed to the production task. Both the comprehension and the production tasks were completed together in one shot and in one controlled setting (i.e., a classroom), but the testing date for each proficiency group was different. It took each participant approximately 20 to 30 minutes to complete the two tasks.
4.3.3 Scoring and Statistical Analysis
Once the data were collected, they were typed, scored, and organized into sets based on the proposed article classification in Table 4-1. Regarding the grading criteria for both tasks, each correct answer received one point; for wrong answers, 0 point was given. Data management and analysis for both tasks were performed using SPSS 22.
One-way analysis of variance (ANOVA) and the post hoc Scheffe with the significance level set at .05 were executed to examine the extent of differences across the subject groups and the difficulty order of the article types and features. Besides statistical analyses, an error analysis concerning the error patterns made by the subjects was also employed.