【作者】 刘倩；

【导师】 程伟；

【作者基本信息】 暨南大学 ， 外国语言学及应用语言学， 2023， 硕士

【摘要】 电商直播已然成为促进销售的强劲动力。然而,对其中微观语言单位的关注相对稀缺。本研究旨在研究中国电子商务直播(ECLS)中多模态话语的互动性,探究电商直播话语中口语和视觉元话语资源的分布模式,以及其中的模态关系,并且考察直播类型(场景型与非场景型)对口头和视觉元话语资源使用的影响。本研究自建了一个电商直播小型语料库。基于淘宝直播平台上六个高质量直播间的60段直播视频,创建了一个235分钟、55,448字的多模态语料库,涵盖了多种类型的产品促销。根据一个三层三级的编码方案对该语料库进行标注,并采用定量和定性相结合的方法对数据进行分析。研究结果表明,口头和视觉元话语资源在电商直播中使用广泛。在口头元话语方面,立场特征比介入特征应用得更频繁。自我中心提及和听众中心提及分别是立场和介入特征中最常使用的元话语资源。其次,视觉元话语从主播和非主播两方面进行考察。主播方面的视觉数据显示,在电商主播的凝视时长中,占比最大的是看向摄像机。非主播方面的视觉数据显示,电商直播界面在分块和外部构架方面是一致的,不同的分块承担着不同的功能。最后,就模态关系而言,强化关系在电商直播中占主导地位,其中最主要的子类型是主次关系。研究还发现,直播类型影响了电商直播中口头和视觉元话语资源的使用。就口头元话语而言,非场景型中的主播使用更多的立场特征,尤其是态度标记语。就视觉元话语而言,主播方面的数据显示,非场景型直播中的主播凝视摄像机和电脑的时间更长,并且使用更多的节奏型手势。而场景型中的主播倾向于使用远镜头,更多地将自己定位在产品后面和旁边。非主播视觉元话语中,场景型和非场景型直播之间最显著的区别在于它们的标题和背景设计。在模态关系方面,场景型直播中突出和协调关系出现的频次明显更多,而扩展和冗余关系在非场景型直播中更多。综上,电商直播中的口头和视觉元话语资源可以实现与自我披露、身份趋同、权力转移和社会在场感相关的意义,有助于构建直播间和观众之间的互动关系。本文不仅为电商直播中互动意义建构的探索提供启示,还通过对实时商业话语体裁的考察丰富了元话语和多模态话语分析的研究,同时也可以为直播从业者提供职业指导,并帮助消费者甄别话语意图。更多还原

【Abstract】 E-commerce live streaming has become a strong driving force in boosting sales.However,far too little attention has been paid to the micro linguistic units in it.The present study purports to examine the interactivity of multimodal discourse in Chinese e-commerce live streaming(ECLS).Particularly,it investigates the distributional patterns of spoken and visual metadiscourse resources in ECLS discourse,as well as their mode relations.It also considers the role of live streaming type(scenario-based vs.non-scenario-based)in the use of spoken and visual metadiscourse resources.This study self-built a mini-corpus of ECLS.Based on 60 live streaming videos from six high-quality live streaming studios on Taobao Live platform,a multimodal corpus of 235 minutes in length and a transcribed word count of 55,448 was created,covering product promotions of various types.The corpus was then annotated based on a threefold and three-tiered coding scheme,and the data were analyzed by using quantitative and qualitative methods.Findings from the study have shown that spoken and visual metadiscourse resources are widely used in ECLS.In terms of spoken metadiscourse,stance features are applied more often than engagement features.Self-oriented mentions and listeneroriented mentions are the most frequently used metadiscourse resources in stance and engagement features,respectively.Second,visual metadiscourse is examined from both the streamer and nonstreamer side.The streamer side of visual data showed that the largest percentage of streamers’ gaze time in ECLS is directed to camera.The nonstreamer side of visual data revealed that ECLS interfaces are consistent in terms of chunking and external skeleton,and distinct chunks in ECLS bear different functions.Finally,as for mode relations,reinforcements predominate in ECLS,with the most dominating subtype being the relationship of primary and secondary.It is also found that live streaming type exerts an influence on the use of spoken and visual metadiscourse resources in ECLS.As far as spoken metadiscourse is concerned,streamers in non-scenario-based live streaming use more stance features,especially attitude markers,than streamers in scenario-based live streaming.As for visual metadiscourse,while streamers in non-scenario-based live streaming gaze significantly more at the camera and computer and use significantly more beats,streamers in scenario-based live streaming use long-shots and locate themselves behind and beside the product significantly more often.On the nonstreamer side,the most significant difference lies in heft in terms of their titles and background settings,which lead to different first impressions.In terms of mode relations,there are significantly more highlighting and coordination relations in scenario-based live streaming,while extension and redundancy relations are more in non-scenario-based live streaming.Taken together,the findings suggest that the use of spoken and visual metadiscourse resources in ECLS can achieve meaning related to self-disclosure,identity convergence,authority transfer and social presence,which help construct interactional relations between the studio and the audience.This study not only sheds light on the exploration of interactional meaning construction in e-commerce live streaming but also enriches the research on metadiscourse and multimodal discourse analysis by examining data in real-time business verbal genres.It also provides vocational guidance for live streaming practitioners and helps consumers to screen discourse intentions.更多还原