目标语词块的频率和搭配强度影响写作成绩的结构方程模型研究

打开文本图片集

中图分类号：H319 文献标识码：A 文章编号：1000-0100（2026）01-0087 -9

DOI编码：10.16263/j.cnki.23-1071/h.2026.01.013

Drawing on 6OO argumentative essaysand 6O request leterscomposed by Chinese high school English learners，the curent studyusedCollram，anaturallnguageprocessingtool，toanalethenativelikebigramsandtrigramsinthsecompositioad to measure their mean frequencyand meanasociation（Mutual Information，or MI）intheCorpus ofContemporaryAmerican English（COCA）.Wethenused Structural Equation Modeling to investigate the path relation through which the mean frequency andmeanMIassociatedwiththese nativelikechunksafectedtheratedscoresofthetwotypesofcompositions，andto explore whethertheir predictiveefectsonscoresare moderatedbygenresandchunk lengths.Theresultsshowthat thedirect fectof mean frequencyof thesenativelikelexicalchunkson writing scores wassmalerthantheir indirecteffects mediated bymean MI. Bothdirect and indirectefects of meanfrequencyof these nativelikelexicalchunksonscores didnotvarybygenre becausehigh scholEnglishlearnersusedmanysamenativelikechunksinthetwogenes.Aditionall，theeffectofmeanMIofnativeliketrigramsontheratedscoresofbothtypesofcompositionsoutperformednativelikebigrams，underscoringthemoderatingeffectof chunk lengths.Theseresults indicatethatChinesehighscholEnglishlearners excesivelydependonalimitedrepertoireof nativelike chunks and underscore their underdeveloped ability of using genre-specific lexical chunks.

Key words：nativelike chunks； frequency；association； structural equation modeling；English writing

1引言

词块指若干个毗邻的单词在一起复现所形成的连续性多词组合，它是介于词汇和句法之间的固定或半固定的结构。（剩余13557字）