【作者】 张薇；

【导师】 陈晓湘；

【作者基本信息】 湖南大学 ， 外国语言学及应用语言学， 2020， 博士

【摘要】 修正性反馈指教师或会话者对学习者的错误做出的回应,是近年来语言习得领域备受关注的话题之一。现有的研究主要集中于考察修正性反馈对词汇、语法、语用等语言习得不同方面的作用,而鲜有研究将修正性反馈引入语音领域进行研究。首次将修正性反馈引介至二语语音习得领域的是Saito&Lyster(2012a),之后便陆续出现了一批对修正性反馈在二语语音习得领域的应用潜力、适用语境和互动效果等各个方面进行了尝试性的探索研究(Bryfonski&Ma,2019;Gooch et al.,2016;Lee&Lyster,2016a,2016b,2017;Parlak&Ziegler,2017;Saito&Lyster,2012b;Saito,2013a,2013b;Saito&Wu,2014;Saito&Saito,2017),但目前该领域的研究仍处于萌芽阶段,且学界对修正性反馈在二语语音感知与产出中的影响一直都存有争议。鉴于此,本文首先以12项国外实证研究为数据来源,运用元分析方法综合分析了修正性反馈在二语语音感知与产出中的总体效用,并着重考察了反馈类型、研究环境、习得背景、目标语结构、训练活动、干预时长、传递方式、后测时间、打分方式及学习者语言水平几大调节变量的有效性,研究结果显示:1)修正性反馈具有显著的积极效应,2)反馈类型、目标语结构、训练活动、干预时长、后测时间及学习者语言水平六大变量有效调节了修正性反馈的效用,3)亚组分析表明:提升+重铸,重铸两类反馈的效应值显著优于提升、明晰纠错两类反馈的效应值;针对超音段进行反馈的效应值显著高于针对音段进行反馈的效应值;基于感知训练的修正性反馈效应值显著高于基于产出训练的修正性反馈效应值;干预时间长的效应值显著高于干预时间短的效应值;即时后测的效用值显著高于延时后测的效应值;中等二语水平学习者的效应值显著高于初学者的效应值。基于元分析结果,本文将上述调节变量对修正性反馈的具体影响纳入本实验的综合考察范畴,进而形成本文实验设计的主旨:修正性反馈在中国外语学习者感知和产出英语松紧元音对中的影响机制。本文选取重铸(直供输入类)、引导(诱导输出类)、引导+重铸(诱导输出类+直供输入类)三类修正性反馈,以英语松紧元音对/i/-/(?)/为目标结构,试图从更微观层面厘清三类最具代表性的修正型反馈之于中国外语学习者感知和产出英语松紧元音对的影响。实验共持续五周,分别进行了一次前测,六轮任务型交互活动,一次即时后测,一次延时后测,及一次问卷调查,实验过程全程录音。在前后测中包括感知测试、控制型产出测试和自发型产出测试。其中,感知测试以强制辨认方式进行,要求受试根据听感选择其认为正确的选项。语料共包含44对单词,其中包括目标单词22对(已训14对,未训8对)和干扰项22对,且每对目标单词中均属包含了/i/-/(?)/的最小对立体(如sheep-ship)。控制型产出测试中,测试员向受试发放14张活页,每张活页有2个单词,且要求受试在承载句“I said…”中以正常语速、相同韵律朗读14对单词,其中包括目标单词7对(已训4对,未训3对)和干扰项7对。自发型产出测试共包含8对关键词,其中包括目标单词4对(已训2对,未训2对)和干扰项4对,描述性图片16张。测试员首先要求受试在20秒内记住2对关键词,其中,目标单词和干扰项各1对。20秒后,待受试确定记住上述2对关键词后,测试员向受试分发1张活页,活页上有4张图片,在无时间准备的前提下,测试员要求受试运用先前记住的2对关键词逐一对图片内容进行描述,且每张图片描述需包括1个关键词。每轮任务型交互活动基于受试对目标元音对的感知能力进行训练。任务型交互活动以意识唤醒的方式进行,分为英语Karuta任务,宾戈任务和填空任务三个步骤。本文以100名中国非英语专业大一新生为研究对象,所有受试随机分为四组,分别接受不同类型的修正性反馈。其中:重铸组25人,当受试出现目标语音感知错误时,测试员用降调提供目标单词的正确发音,且不进一步解释;引导组25人,当受试出现目标语音感知错误时,测试员以问句形式重复该目标单词的不正确发音,也不进一步解释;引导+重铸组25人,当受试出现目标语音感知错误时,测试员先以问句形式重复该目标单词的不正确发音,再用降调提供目标单词的正确发音,亦不进一步解释;控制组25人,当受试出现目标语音感知错误时,测试员不提供任何形式的反馈。延时后测结束后,所有受试接受了问卷调查,汇报他们在任务型交互活动中对测试员反馈的注意程度以及对整个实验活动的看法。本文的具体研究问题如下:(1)不同类型的修正性反馈对英语松紧元音对/i/-/(?)/感知的影响力如何?(2)不同类型的修正性反馈对英语松紧元音对/i/-/(?)/产出的影响力如何?(3)经修正性反馈后的感知进步是否是产出进步的重要标志?本文采用计量统计方法对实验数据进行定量分析,主要运用多元测量方差分析、两因素重复测量方差分析和回归分析。通过对统计数据的分析,本文得出以下结论:(1)感知方面:在已训和未训词汇条件下,重铸、引导、引导+重铸在英语松紧元音对/i/-/(?)/的感知中均发挥了显著效应,且三类修正性反馈的有效性无显著差异;(2)产出方面:在已训和未训词汇条件下,重铸、引导+重铸两类修正性反馈在英语松紧元音对/i/-/(?)/的产出中的效用显著高于引导类修正性反馈;(3)在感知与产出的相关性方面:在已训和未训词汇条件下,经修正性反馈后的感知进步是产出进步的重要标志。基于以上研究发现,本文在对中国外语课堂语音教学提出了相应的建议之后,亦指出了未来可能的研究方向。更多还原

【Abstract】 CF refers to the responses to learner utterances containing an error,which has been one of the hotly debated topics in the field of language acquisition in recent years.In general,previous research mainly concentrated on the effectiveness of CF in lexical,morphosyntactic,pragmatics aspects,little research considered the effectiveness of CF in L2 speech perception and production.Since Sait o & Lyster(2012a)published their influential work,a number of studies have been conducted to explore the availability and application of CF in L2 speech perception and production(Bryfonski & Ma,2019;Gooch et al.,2016;Lee & Lyster,2016 a,2016b,2017;Parlak & Ziegler,2017;Saito & Lyster,2012b;Saito,2013 a,2013b;Saito & Wu,2014;Saito & Saito,2017).However,the research on CF in L2 speech perception and production is in its infancy,and its role remains unclear.To investigate the pedagogical effectiveness of CF in L2 speech,the present study initially conducts a meta-analysis to synthesize extant empirical research,including 12 overseas empirical studies.The analysis is designed to explore whether CF is effective in L2 speech and,if so,whether its effectiveness varies according to CF types,research setting,research context,target structure,training activities,treatment duration,mode of delivery,time of post-test,scoring method and proficiency.Results show that CF has medium-to-large effects on L2 speech,and its effectiveness is mediated by a host of variables,including CF types,target structure,training activities,treatment duration,time of post-test and proficiency.Sub-group analysis further shows that the effects are larger for recasts+prompts and recasts than prompts and explicit correction,for suprasegments than segments,for perception-based training than production-based training,for long treatment than short treatment,for immediate post-test than delayed post-test and for intermediate learners than beginners.Based on the results of meta-analysis,the above mediating variables of CF effectiveness provide a structured framework in the following experimental design and form the propose of the present study: the influential mechanism of CF in perception and production of English tense-lax vowel contrast /i/-/(?)/ by Chinese EFL learners.Next,the present study examines the effectiveness of three types of CF-recasts(input-providing),elicitations(output-providing),elicitations+recasts(output-providing+input-providing)on the perception and production of English tense-lax vowel contrast /i/-/(?)/.This empirical study lasts about 5 weeks,and employs a pre-test,6 task-based interactional treatment sessions,an immediat e post-test,a delayed post-test and an exit questionnaire,and all the instances of CF and the participants’ immediate uptake are video-recorded.The effectiveness of CF is examined using 3 versions of perception test,controlled production test and spontaneous production test.The perception test is designed with forced-choice identification task.There are 44 sets of minimally paired words in total,22 sets are minimally paired words targeting /i/ and /(?)/(i.e.,sheep-ship)with 22 sets of distractors.1-14 sets are used as trained items and 15-22 sets are used as untrained items.As for the controlled production test,there are 7 sets of target paired words with 7 sets of distractors,1-4 sets are used as trained items and 5-7 sets are used as untrained items.The participants are shown 14 worksheets one by one,each of which has 2 words,they are required to produce utterances using a carrier sentence “I said [X].” As for the spontaneous production test,there are 4 sets of target paired words with 4 sets of distractors,1-2 sets are used as trained items and 3-4 sets are used as untrained items.Participants are first given 20 seconds to memorize a written list of 4 key words,including 2 target paired words and 2 distractors.Immediately after the list is taken away,they are given a piece of paper containing four pictures and asked to describe the pictures one after another by using all of the key words they just memorized without any preparation,1 key word for each picture.Moreover,3 awareness tasks including English Karuta task,bingo task and fill-in-the-blank task are employed during the treatment sessions.100 non-English major freshmen at university are randomly assigned into 4 groups: recasts group(n=25): the instructor responds to participants’ nontarget forms by enhancing the salience of those forms through recasts,and pause for a short time to wait for participants’ repair.If the participants cannot provide a repair,the instructor will continue the conversation without pushing them to produce a repair.In order to control the variables,the recasts in the present study are all partial,didactic,ended in falling tone;elicitations group(n=25): the instructor responds to participants’ nontarget forms by means of an open-ended elicitation with rising tone but there is no target form provided;elicitations+recasts group(n=25): both elicitations and recasts are provided to participants for their nontarget forms.The first step for the instructor is to respond to participants’ nontarget forms by means of an open-ended elicitation with nontarget form,and then immediately followed a recast with target form;control group(n=25): no feedback is provided.Questionnaires are also used for all groups to gather information on their noticing of CF and their perceptions towards the activities included in the experiment.Three principle research questions are formulated as follows:(1)How do three CF types differentially impact Chinese EFL lea rners’ perception of English tense-lax vowel contrast /i/-/(?)/?(2)How do three CF types differentially impact Chinese EFL lea rners’ production of English tense-lax vowel contrast /i/-/(?)/?(3)Does improvement in perception accuracy resulting from CF predict improvement in production accuracy?The quantitative analysis is conducted as econometric and statistical method.The econometric and statistical techniques employed for analysis are MANOVA,Two-way Repeated Measures ANOVA and Regression Analysis.Through the statistical analysis,three conclusions are gene rated:(1)As for both the trained and untrained items,recasts,elicitations and elicitations+recasts all have beneficial effects on Chinese EFL learners’ perception of English target vowel contrast /i/-/(?)/,and there is no significant difference among CF.(2)As for both the trained and untrained items,recasts and elicitations+recasts demonstrate significantly superior effects than elicitations group on Chinese EFL learners’ production of English target vowel contrast /i/-/(?)/.(3)As for both the trained and untrained items,the regression analysis demonstrates that improvement in perception accuracy resulting from CF is a significant indication of production accuracy.Based on the above conclusions,the pedagogical implications of the present study for speech instruction in Chinese EFL classroom are recommended and several topics that future studies should examine are pointed out.更多还原