This study aimed at examining the effects of styles of presenting multiple-choice options, that is, the style in which the options are presented in written format and the style in which they are presented in recorded format, on the results of a Japanese listening comprehension test. In two Shimada cases (2004 and 2006a), the test consisting of items with written options was less difficult than the test with recorded options. In this study, based on the Shimada (2006a) data, the effect of the ability of each individual examinee on the test results was examined. An analysis of four groups of examinees classified according to the test scores showed that, as for Step 1, or the lowest score level, in the Level-3 item group for the Japanese Language Proficiency Test, the average score was higher for the style in which the options were presented in recorded format. As for Step 2, in most item groups the average score was higher for the style in which the options were presented in written format, and as for Step 3, in the Level-2 item group for the Japanese Language Proficiency Test, the average score was higher for the style in which the options were presented in written format. As for Step 4, or the highest score level,no significant difference was observed between the two styles at any item groups. These results show that the written options helped medium-level learners in answering the questions, whereas reading the options constituted a burden for beginner-level learners.

抄録全体を表示

This study explores an analysis unit suitable for listening recall. It follows Iimura (2007) and focuses on idea unit (IU). It examines three types of IUs-Carrell version, Sakai version, and Spencer version. The recall protocols of 41 Japanese university students are scored against these types. Their scores on each type are compared to an external criterion measure, the listening section of the computerized assessment system for English communication (CASEC). Although the results indicate that all the versions are significantly correlated with the external measure, by grouping recall protocols, this study reveals the characteristics of each type and identifies Sakai version to be the most suitable analysis unit for the listening recall of the three IU types.

抄録全体を表示

Mochizuki's (1998) Vocabulary Size Tests (VSTs) have three forms, which are made up of seven word-frequency levels (1,000, 2,000, 3,000, 4,000, 5,000, 6,000, and 7,000 word-frequency levels) to measure how many of the most frequently-utilized 7,000 words an examinee knows. In this study, the author will examine whether the rationale of estimating receptive vocabulary size of Japanese learners of English based on each word-frequency level, which adopts the stratified sampling method, is appropriate (in other words, it will partially examine the validity of Mochizuki's VSTs) from the psychometric viewpoint as well as reviewing the previous studies utilizing or analyzing Mochizuki's VSTs. University students majoring in English took six sections (2,000-7,000 word-frequency levels) in each of the three forms of Mochizuki's VST. Since six variables (six subtotals of correct answers in each section) were obtained in each form, Pearson's product-moment correlation coefficients among the six variables were calculated in each form, and exploratory factor analyses were conducted. The results showed that six sections (2,000-7,000 word-frequency levels) had high correlations with one another (especially with the closer word-frequency-level sections) and that only one factor was extracted in each of the three forms. This may indicate that no section (no word-frequency level) is heterogeneous and that each word-frequency-level section is consecutively connected. Theoretically no section (no word-frequency level) should be heterogeneous and each word-frequency level section should be connected successively. In this study, these theoretical corollaries may have been partially confirmed empirically from the psychometric results obtained through the interaction between the examinees and the VSTs.

抄録全体を表示

This is the third report in a study series that examined the relationships between English learning experiences in elementary school days and English proficiency and attitude toward learning it in high school years. Participants were 661 students learning at either of the two courses at a SELHi, one regular and the other English-focused. They took a survey on pre-junior-high English learning experiences and current drive to learn the language. They also sat for an ACE battery test with listening, reading, grammar, and vocabulary components. T-tests indicated significantly higher means for those with the early learning experience, in motivation and listening, but not in reading, grammar, or vocabulary. The effects on motivation and listening were significant even after the influences of sex, school course and year were partialed out by multiple regression. There was also some evidence that early English learning had a greater impact on motivation and proficiency if it had continued for three or more, than for two or fewer, years.

抄録全体を表示

The purpose of this study was to conduct DIF analysis of Japanese Can-do-statements, focusing on the learner's mother tongue, either Chinese or Korean, and investigate (1) concrete verbal behavioral situations that cause DIF and (2) the extent to which differences in methods for detecting DIF affects the results. Results indicated that (1) DIF was advantageous to a speaker whose mother tongue was Chinese, in the direction of 'Can-do' for 5 question items on reading and writing skills. On the other hand, DIF was advantageous to the speaker whose mother tongue was Korean, in the direction of 'Can-do' for 5 question items regarding speaking and listening skills. (2) Similar results, from the perspective detecting DIF, were obtained using SIBTEST, BILOG-MG DIF command, Likelihood ratio test, the Mantel-Haenszel statistical method, and Logistic regression method.

抄録全体を表示

This research addressed the comparison of concurrent calibration between a polytomous IRT model and a dichotomous IRT model using English achievement test data. Two forms of English achievement tests for senior high school students were composed of testlets (groups of items) to eliminate the effect of the dependence among within-testlet items. The two forms were equated with common testlets through a polytomous IRT model. The testlet parameter estimates and the category characteristic curves were analyzed on a common scale. The result showed that one form was more difficult than the other, as test designers had intended. The mean of the ability parameter estimates of the more difficult form was higher than that of the easier form. These findings yielded useful feedback for test designers. Item parameter estimates of independent dichotomous items, ability parameter estimates and the amount of test information derived by concurrent calibration under the graded response model (polytomous IRT model) and the two-parameter logistic model (dichotomous IRT model) were compared. The results showed similar parameter estimates for the two IRT models. The standard errors of ability parameter estimates for both models also were highly correlated. The two-parameter logistic model provided a greater amount of test information than the graded response model.

抄録全体を表示