复杂度
预测能力
计算机科学
任务(项目管理)
可预测性
语言学
质量(理念)
自然语言处理
写作评估
人工智能
心理学
数学教育
统计
数学
经济
管理
社会学
哲学
认识论
社会科学
作者
Yujie Zhang,Jinghui Ouyang
标识
DOI:10.1016/j.asw.2023.100727
摘要
This study investigated the predictive power of syntactic, lexical, and phraseological complexity indices over human-assessed scores of independent non-prompt and integrated reading-to-write tasks of different genres. Their similarities and differences were then compared and discussed. To address the research questions, corpora of writing samples collected from intermediate EFL learners were built, rated, and analysed accordingly. The regression models indicated that 1) the predictability of syntactic complexity remained stable across task types and genres within the 20–30% range; 2) fine-grained syntactic indices, especially phrasal complexity and syntactic sophistication indices, played a stronger role in predicting scores; 3) the predictive power of lexical indices topped over the other two variables and was stronger in independent writing tasks; 4) the predictive power of phraseological indices was noticeably stronger in integrated writing tasks. Implications based on the findings were discussed for language teaching, learning and assessment.
科研通智能强力驱动
Strongly Powered by AbleSci AI