本文以1993-2014年期间SSCI期刊《应用语言学》发表的学术论文摘要为语料,对英文摘要语步结构标注后,利用语料库和自然语言处理方法,提取有效语步预测特征,通过语步结构类别训练学习分类器(条件随机场),获得摘要语步结构自动识别模型。本研究突破了长久以来依赖于人工的语步分析法,对语类研究以及英语学术写作具有借鉴意义和实用价值。 Based on the abstracts of scholarly essays published in Applied Linguistics of SSCI in 1993-2014 as the corpus and the structure of English abstracts, we extracted the effective step predictive features by corpus and natural language processing, Category training learning classifier (conditional random field), access to digest step structure automatic identification model. This study has broken through the long-time dependence on the step-by-step analysis of human language, and has reference and practical value for language studies and English academic writing.