Working memory training and its effects on oral proficiency in English: implications to automatically graded speaking tests

控制组和实验组的约40名学生在第一学期和第二学期的13次工作记忆训练前后按计划被记录下来。从复杂性、准确性和流利性三个方面对话语进行转录和分析。第一学期的结果显示了任务效应，即引出学习者言语的问题意外地影响了成绩。因此，工作记忆训练的效果只能推断，因为预期效果与任务效果是分不开的。结合第二学期的语料，对学生的言语变化进行了较为详细的分析。复杂性(句法复杂性和句子长度)的改善与流利性之间没有显著的相关性，这意味着句子变长了而不牺牲流利性。另一方面，复杂程度的变化与错误数量之间存在显著的相关关系，表明复杂程度和准确性之间存在权衡效应；而流利度方面，前测结果与前后流利程度的差值呈显著负相关(r=-.59)，表明学生最初的流利程度越高，这些练习的效果就越差。

Utterances of about forty students in the control and experimental groups were recorded before and after the 13 working memory training sessions in the first and second semesters as planned. The utterances are transcribedand analyzed in terms of complexity, accuracy and fluency. The results of the first semester showed task effects, namely, the questions that elicited learner speech unexpectedly influenced the performance. Therefore, the effects of the working memory training could be only inferred since the expected effects could not be separated from the task effects. With the data recorded in the second semester, the changes in students’speech were analyzed more in detail. There were no significant correlations between the improvement in complexity (syntactic complexity and sentence length) and fluency, meaning that sentences became longer without sacrificing fluency. On the other hand,significant correlation was found between the changes in complexity and in number of errors, implying a tradeoff effect between complexity and accuracy.As for fluency, there was significant negative correlation between the results of the pre-test and the difference in fluency scores between pre-and post-tests (r=-.59), indicating that the more fluent the students were initially, the less effective these practices were.