【作者】 罗娟；

【导师】 肖云南；

【作者基本信息】 湖南大学 ， 外国语言文学， 2019， 博士

【摘要】 近年来,为了科学诊断学生的认知结构,教育学与心理测量学领域兴起了认知诊断研究,为现代教育测试注入了新活力。认知诊断研究强调将心理测量学与认知心理学相结合,深入整个受试作答的认知加工过程中,以此将受试的认知加工特点揭示出来。语言领域的认知诊断研究在国内外已逐步开展起来,迄今为止,基础研究胜于实践研究,尚停留在理论概述和小规模应用领域,在社会大规模考试中应用不多。目前语言测试领域的阅读认知诊断研究尚有不足:首先,缺少认知诊断测试编制理论指导测试开发,以翻新已有测试为主;其二,语言学界对阅读认知属性定义不一,存在明显争议;其三,以学生个体为诊断对象,缺乏群体诊断;其四,认知诊断报告匮乏,与教学实践脱节。针对目前研究局限,本文尝试基于大学英语分级测试,编制以认知诊断为目的的阅读能力测试,构建英语阅读能力认知模型,针对非专业大学生的阅读能力,进行学生个体与群体的双重认知诊断研究,并面向师生设计诊断报告。本文的研究问题具体如下:(1)如何构建英语阅读能力认知模型?(2)大学非英语专业学生与不同水平学生群体的英语阅读能力有哪些特征?(3)针对不同需求,如何设计师/生认知诊断报告?根据研究问题,本文从阅读认知模型构建、认知诊断分析与诊断报告设计三个步骤展开。首先,本文明确了语言认知诊断测试编制步骤,依据该步骤测试开发组编制了英语阅读能力诊断测试(CEDT);以湖南大学新生入学英语分级机试(CEPT)为基础,测试专家组确认了阅读的认知属性,构建了阅读能力认知模型;其后,在实验阶段,笔者综合8名学生的有声思维报告及5位学科专家的经验判断,从双重角度确定了Q矩阵。在2016年以1032名考生为试测对象,采用G-DINA模型检验了阅读认知模型、Q矩阵的数据-模型拟合度,并对试题进行了项目分析及信度效度初步检验。于2017年以2682名考生为诊断对象进行了诊断测试,对学生个体与不同水平群体进行认知诊断分析,并从多个途径进行信度效度验证;最后,针对师生不同需求,设计了诊断报告框架。研究发现如下:(1)基于CEPT分级测试的语言框架CLB,结合教育部《课程要求》,并参考前人文献研究,CEDT测试开发专家组确定了阅读诊断测试所需的5种认知属性:A1(词语及词组)、A2(复杂句子及结构)、A3(显性细节信息)、A4(衔接与概括)与A5(推理)。本文从语言学及认知语言学角度出发,将Bachman(1990)构建的CLA模型与Kintsch(1988)的CIM模型相结合,构建了阅读能力认知模型,将5种属性整合于模型之中。该模型既评估学生的语言能力,同时探索学生在语言学习过程中认知能力的发展特征。(2)学生总体对阅读理解的5种属性的掌握程度存在明显差异,验证了阅读能力认知模型中各属性难度的层级性。从认知心理学来看,学生对阅读认知属性的掌握概率由理解词语及词组到推理逐渐降低,与属性A1(词语与词组)、A2(复杂句子及结构)与A3(显性细节信息)相比,属性A4(衔接与概括)与A5(推理)比其他三种属性的难度更大。该发现验证了在阅读能力认知模型中,它们需要对文本中的信息进行更高层次的处理,属于更高认知水平的属性。从语言学角度分析,无论是学生总体,还是初级、中级水平群体,学生对于语言知识层面所涉及属性的掌握程度优于理解策略层面涉及的属性,主要问题集中在理解策略层面。而高级群体对两个层面涉及的认知属性的掌握比较均衡。(3)本文以CEPT语言能力等级量表为标准,将分级测试结果与认知诊断信息相结合,设计了针对教师及学生的诊断报告框架。该框架采用结构化的方法构建,将诊断报告分为基于任务、基于技能和基于个性化指导三大部分,实现报告手段上量化分析与质性描述相结合、诊断对象上学生个体与群体兼顾、诊断技能上微观属性与宏观能力相平衡的原则,通过数据、文字、图表与可视化等多种方式,将各种信息清晰地展现出来。本文从语言学与认知语言学双重角度,深入剖析阅读能力的本质,解构英语语言阅读理解过程中所涉及的知识、技能与策略,同时深入分析学习者的心理加工机制,构建了科学、全面的阅读能力认知模型。本文尝试将认知诊断理论引入英语能力分级测试,结合两种测试的目的、功能与特征。在宏观上既评估学生的语言能力等级,同时在微观上探索学生语言学习过程中认知能力的发展特征,对学生总体、不同水平群体及个体进行了诊断。基于诊断结果建立了新型的分数报道机制,为英语学习者及教学者、管理者提供多元化的诊断报告,为改进教与学指明方向。通过对研究问题的探析,本文最终实现科学评估、精确诊断、有效报告一体的分级测试模式,为分级测试研究开拓了新思路,有助于我国分级测试的科学化与普及化。研究成果用于指导我国大学英语教学实践,有利于实现以考促教、以考促学、因材施教的宗旨,促进我国外语教育水平和学生外语能力。更多还原

【Abstract】 Interest in cognitive diagnostic assessment(CDA)has been growing rapidly over the past several years,which combines theories of psychometrics with cognitive psychology,intended to measure test takers’ cognitive structure and processing skills,so as to make inferences about their cognitive strengths and weaknesses.The last decade has witnessed the development of CDA in the field of language testing at home and abroad.So far,the fundamental theoretical research on EFL reading far outweighs the empirical study.Until now,few systematic empirical studies have been conducted in the field of EFL reading tests,which is confined to small-scale application domain,and some obvious limitations exist.Firstly,existing traditional proficiency tests were retrofitted for diagnostic purposes,as currently available literature offers very little guidance on how diagnostic tests should be developed;Secondly,disputes continue unabated over what attributes are involved in EFL reading process;Thirdly,prior research has focused mostly on the diagnosis of individual learners,but not the student groups at different proficiency levels;Forth,few studies have examined the effect of providing diagnostic feedback to stakeholders.Considering the above limitations,there is a dire need for more CDA research in EFL reading.At this point,the present study focuses on the English reading ability of Chinese college non-English majors,attempts to develop a cognitive diagnostic test based on EFL placement test,used for diagnosing the reading ability of individuals and student groups of different language proficiency levels.Then,test data were analyzed and developed into diagnostic reports for teachers and students.Specifically,the research questions come as follows:(1)How to construct a cognitive model of EFL reading ability?(2)What are the characteristics of English reading ability of individual students and student groups at different language proficiency levels?(3)How to design cognitive diagnostic reports specifically for teachers and students?The methodology falls into three phases: construction of reading ability cognitive model,cognitive diagnosis analysis and design of diagnostic report.Phase 1 includes reading attributes identification and cognitive model construction,and then development of the CEDT(College English Proficiency Diagnostic Test)by the experts group;Phases 2 is the experimental stage.It begins with Q matrix construction,which is confirmed by combining 8 students’ retrospective think-aloud protocol with 5 expert judgments.In 2016,1032 examinees participated in the pilot test.Based on G-DINA model,data analysis has been carried out to check the Q matrix and data-model fitting.Moreover,preliminary analysis of item parameter,test reliability and validity were carried out based on IRT analysis.In 2017,cognitive diagnostic test was conducted to 2,682 examinees,then strengths and weaknesses were identified for the overall group,three reading proficiency groups and individual learners.To verify the reliability and validity of the test,multiple ways has been applied in this study;in phase 3,framework for diagnostic reports specific to teachers and students are also proposed.The findings come as follows:(1)Based on CLB(Canadian Language Benchmarks),and College English Curriculum Requirements,five attributes are identified by the domain experts,namely,A1(Vocabulary),A2(Syntax),A3(Extracting explicit information),A4(Connecting and summarizing),A5(Making inferences).From the twofold perspectives of linguistics and cognitive psychology,English reading ability is defined within CLA model(Bachman 1990)and CIM model(construction-integration model)(Kintsch 1988).Thus,accordingly,the reading ability cognitive model consisting of the five attributes is constructed.This model is novel in the sense that it not only evaluates students’ language competence,but also explores the development characteristics of students’ cognitive competence in the process of language learning.(2)Cognitive diagnostic analysis shows that there is significant difference in overall students’ mastery probability of the five attributes,which verifies that some attributes are more difficult to master.From the perspective of cognitive psychology,among the five attributes,the mastery probability decreases gradually from understanding vocabulary to making inferences.Compared with attributes A1,A2 and A3,attribute A4 and A5 are more difficult than the other three attributes.This finding indicates that in the cognitive model of reading ability,attribute A4 and A5 required higher-level processing of information from the text and belong to higher cognitive level,thus more difficult to master.From the perspective of linguistics,the overall students,as well as student groups of primary and intermediate proficiency levels,demonstrate poor performance on the reading test with low attribute mastery probability across strategy-related attributes.That is,their mastery of knowledge-related attributes is better than strategy-related attributes,whereas the advanced students show a balanced mastery of the five attributes.(3)Considering students’ demands for feedback report,this study provides a structured approach for developing score reports,which is divided into task level,strategy level and tailored instruction level,specific to the needs of teachers and students.Guidelines for reporting and presenting diagnostic scores are based on the integration of cognitive model of reading ability and CEPT(Computerized English Placement Test)language proficiency scale.This report framework has promising features,such as reporting information in alternate forms(i.e.,narrative,and numeric),presenting different reports for various stakeholders(i.e.,the individuals and student groups of different proficiency levels),and integrating test functions of proficiency placement and attribute diagnosis.Moreover,the multiple displays of information in this framework include numeric,graphic,and narrative forms presented in a visualized method.To summarize,the study is significant in that it reveals the EFL reading processes,presents how to develop EFL reading diagnostic tests and devises ways for reporting diagnostic results and offering pedagogical guidance and mediation.Firstly,this paper incorporated models of linguistics and cognitive psychology when designing attributes and constructing reading ability cognitive model,which gives a deep insight into the components and cognitive process of reading ability simultaneously;Secondly,it makes a bold attempt to introduce CDA into EFL placement test.By integrating the purposes,functions and characteristics of the two tests,the CEDT test is developed,which not only evaluates students’ overall reading proficiency,but also explores their micro characteristics of cognitive competence,thus,introducing new ideas to the development of language testing.In addition,pedagogical implications could be drawn from this study regarding the diagnostic feedback provided to various stakeholders.This research provides teachers and learners with a clearer and more effective diagnosis of their own teaching and learning status,and is helpful to improve the efficiency of teaching and learning.更多还原