以写作样本的文本指标评价读后续写题

Evaluating the Story Continuation Task through Text Indicators Generated from Writing Samples

为了更全面验证高考英语读后续写题的效度,本研究考察该题型所激发的语言样本是否实现了命题者的意图。为此,研究者收集了247名考生在一次模拟考中产出的应用文和续写样本,经语料分析获得155项文本指标,通过相关、主成分分析和回归分析,从中筛选出符合理论构念并能可靠预测作文成绩的4项指标,据此对应用文和续写文本进行比较。结果表明,续写任务能够促使考生产出更丰富的内容,使用更多低频单词和更复杂词语,符合命题机构和倡导者的预期。

In an attempt to validate the NMET story continuation task,the current study evaluates the extent to which the language samples elicited by the task live up to the expectation of the test developer.The writing samples of 247 test takers on a letter writing task and a continuation task in a NMET mock test were analyzed with corpus tools to generate 155 text indicators for each sample on each task.Correlational,principal component,and regression analyses were conducted to categorize the indicators and select a group of four indicators that matched the dimensions of the construct and consistently predicted the writing scores on both tasks.The two tasks were then compared in their mean values on the four indicators.The results showed that the continuation tasks had elicited richer content,more low-frequency words,and more complex word forms,to the expectation of the test developer and advocators.