Writing summaries is an important skill not only in higher education but also in real-world contexts. In second language (L2) classrooms, summarizing texts in L2 is widely used to enhance language skills (e.g., reading and writing) and cognitive skills, including critical thinking (CT). However, little is known about the relationship between word limits in summary tasks and the quality of L2 learners’ summaries. Thus, this study aims to investigate the impact of word limits on summary writing performance. A total of 52 Japanese university students completed six summary tasks, with two tasks at each of three target lengths (40, 60, and 80 words). Three trained raters evaluated the summaries using a rubric focusing on CT skills. The results of ordinal logistic regression analyses indicated that total scores differed significantly between the 40- and 80-word summaries, as well as between the 60- and 80-word summaries. Further analyses showed that students found the 80-word summaries the most challenging in terms of identifying and including key points, maintaining coherence, and paraphrasing effectively. Therefore, when incorporating summary tasks into instruction, teachers should recognize that the required reading, writing, and CT skills may vary depending on task length.

This study examines the functioning of the functional adequacy (FA) rating scale and rater effects in assessing English writing of Japanese high school students. Ninety-seven high school students completed three writing tasks, and three raters evaluated the responses using an analytic FA scale with four criteria: Content, Task requirements, Comprehensibility, and Coherence and cohesion, each scored on a six-category scale. Many-facet Rasch measurement revealed problems in the scale’s functioning. Some score categories did not function as expected. Raters used the scale consistently, although differences in severity were observed. No substantial bias interaction was found between raters and criteria. To address the malfunctioning of the scale, a four-category scale was tested by combining adjacent categories. The reanalysis showed improved functioning and reduced severity differences, while maintaining rating consistency. No bias interaction was found between raters and criteria either. These findings suggest that the four-category FA scale may be more appropriate for evaluating English writing of Japanese high school students, though this suggestion remains subject to certain limitations.

This study investigates the effect of Early English Exposure/Education (EEE) post-adolescence on 252 Japanese students’ listening and reading proficiency, measured four times longitudinally based on the TOEIC Bridge IP tests, over three years (2004–2006) of senior high school (SHS) in Japan. A 5-item questionnaire determined if students had learned English in pre-kindergarten, kindergarten, and/or grades 1–2/3–4/5–6 of elementary school. Five exposed groups (EGs) with EEE experience were formed based on the EEE period. The longitudinal data of the five EGs and the unexposed group (UG) were analyzed using marginal means and hierarchical models. The results suggest that at SHS entrance (intercepts), the EGs with two or more years of EEE have higher estimated values than the UG in both listening and reading scores. However, there are negligible differences in growth (slopes) over three years of SHS in both listening and reading. Thus, the difference between the EGs and the UG at SHS entrance persists until graduation. If the UG could match the EGs in the total time spent on English exposure/education in their lives after graduating from SHS, they could catch up with the EGs.

This study investigated how two different speaking test formats—interview tests (i.e., student–teacher interaction) and paired tests (i.e., student–student interaction)—influence the type of linguistic competence elicited. Both the formats were administered to 40 second-year Japanese junior high school students, and their performance was assessed in terms of accuracy, fluency, and complexity. Significant differences were found in elicited fluency (words per minute) and lexical complexity, with the interview test producing more extensive speech and a wider vocabulary range. By contrast, no significant differences were observed in terms of accuracy, disfluency (repetition and self-correction), or syntactic complexity. These results suggest that the interview format is more effective when the aim is to elicit more fluent and lexically varied speech. However, the interview test format limited students’ opportunities to formulate questions—an essential aspect of interactive communication. Therefore, paired tests should also be administered. Furthermore, a close examination of students’ utterances revealed that their accuracy, fluency, and complexity in the paired test were influenced by their partners, particularly their degree of familiarity. Because the paired format may elicit varying linguistic performance depending on the interlocutor, multiple assessments with different partners are recommended to ensure more reliable and valid evaluations.

This study examines the construct validity of the Critical Thinking Test for English Education in Japan (CTTEEJ). This critical thinking (CT) test was developed in a multiple-choice format consisting of 22 items for the context of Japanese English education. A total of 143 university and senior high school Japanese learners completed the CTTEEJ in approximately 50 minutes. To validate the test construct, exploratory factor analysis (EFA) was conducted after examining item properties and reliability coefficients. Additionally, given that CT has been theoretically and empirically suggested to be an elusive concept, cluster analysis was also employed to explore the characteristics of CT skills measured by the CTTEEJ. The item analysis results indicated sufficient reliability as a measure of CT skills. Moreover, both EFA and cluster analysis confirmed that the internal structure of the CTTEEJ was generally consistent with the hypothesized subscales proposed in this study. However, several items were classified into subscales different from expected ones, suggesting the complexity of CT skills. These findings provide mostly support for the construct validity of the CTTEEJ and reflect the elusive nature of CT. While the analysis of this study could collect validity evidence of the CTTEEJ, further research is needed to strengthen the degree of test validity.