简体中文 English
主办单位
协办单位
四川省认知语言学研究会


承办单位


大会报告(以纸质版会议手册为准)

主旨发言

20191026日 星期六

时间

主讲人

题目

主持人

地点

09:00-09:50

Julie A. Fiez

Cross-linguistic perspectives on the neural organization of the brain for reading

王穗苹

 

图书馆求实厅

 

10:20-11:10

Guillaume Thierry

Language is not magic: Towards an integrated account of linguistic relativity and embodiment theory

蔡振光

11:10-12:00

Chris Sinha

Interdisciplinarity, multiple methods and reflexivity in the cultural linguistic laboratory

周榕

20191027日 星期天

时间

主讲人

题目

主持人

地点

15:00-15:50

李兴珊

中文阅读中特异性的认知机理研究

王瑞明

图书馆求实厅

15:50-16:40

屈鸿

面向机器学习和人工智能的自然语言处理研究

梁君英

 


主旨发言1 



Abstract:

This talk will consider factors that influence how readers code orthographic knowledge and map it onto corresponding phonological and semantic knowledge. The first part of the talk will examine second language (L2) reading by individuals with a Chinese or Korean native literacy (L1) background. Data from imaging and behavioral studies will be presented that suggest Chinese-English (CE) bilinguals transfer a bias towards lexical reading procedures and holistic orthographic coding to L2 English reading, whereas Korean-English (KE) bilinguals transfer a bias towards sublexical reading procedures and analytic orthographic coding. The second part of the talk will present evidence that similar individual differences can be observed amongst L1 readers of English. Specifically, it will be shown that a marker of holistic orthographic coding is predictive of lateralization differences in the neural response to visual words, and of behavioral differences in reading style. Finally, the third part of the talk will present results from studies involving the use of artificial orthographies. In these studies, adult native English speakers were trained to read either an artificial alphabet or syllabary for English. The results indicate that learning an orthography with an alphabetic mapping principle induces robust left-lateralized neural changes in orthographic representation, whereas learning a non-alphabetic orthography induces bilateral neural changes in orthographic representation. These neural changes are accompanied by behavioral signatures of analytic versus holistic orthographic coding. Taken together, these results reveal different patterns of behavioral and neural assimilation versus accommodation to the linguistic features of an orthography that should be considered when developing theoretical models of skilled reading and its emergence through reading practice.

 

Professor Julie is a cognitive neuroscientist with expertise in functional neuroimaging research and neuropsychological studies of acquired brain damage. Her research is characterized by a strong commitment to identifying the neural mechanisms that underlie human behavior. She has a long-standing interest in (a) the neural substrates of reading and language and (b) learning and memory mechanisms. She has published more than 100 papers in international journals, including Nature Neuroscience, Journal of Neuroscience, Cortex, Human Brain Mapping, Brain and Language, Cerebral Cortex, Journal of Experimental Psychology: General etc. The total citation is now over 18,000 times and H index 55. 

 



主旨发言2



Abstract:

In this talk, I will attempt to demystify our perception of language as a mysterious epiphenomenon of human brain operation, seemingly disconnected from other types of cognitive processes such as perception, emotion, attention, or decision making. To do this, I will show how cognitive operations, which could be construed as language-independent, are in fact implicitly modulated by language context and linguistic knowledge. I will focus on results from four experiments (Gao et al., Journal of Neuroscience, 2015; Athanasopoulos et al., Psychological Science, 2015; Wu et al., Cognition, 2016; Li et al., Neuroimage, 2019) in which bilinguals individuals either (i) adjust their betting style based on the language in which they receive feedback; (ii) conceptualise motion events as more or less completed depending on the language context in which they find themselves; (iii) look differentially more at second language words that conceal a relationship to a visual, non-verbal target through translation in the native language; and (iv) access metaphors that only exist in their native language leading them to suffer spatial-temporal processing interference when neither type of information –metaphor or spatiotemporal stimulus configuration– are task-relevant. In all cases, tasks requirements were deceptively simple: (i) decision to take a bet in a game of chance, (ii) indicate whether an ambiguous motion scene looked as ongoing or nearing completion, (iii) detect circle or square shapes; and (iv) indicate whether the time gap between a spoken stimulus and present time was one or two days/years. But in all cases, the bilingual brain manifested nonverbal cognitive modulations dependent upon contextual language information, either incidentally presented or idiosyncratic to one of their languages. All in all, these findings are consistent with the idea that language representations are embodied in the human brain and they provide support for the provocative idea that language and other facets of cognition are deeply, ineluctably, and , unconsciously inter-twinned in the human mind.

 

Professor Guillaume Thierry has been working on the key issues in the field of language with his profound knowledge of cognitive neuroscience, by using interdisplinary research methods of experimental psychology and neuroscience. His research focuses on many hot-debating topics, for example, the processing of language, the neural mechanism of bilingual processing, the interaction between different languages, the influence of language on other cognition, and the neural mechanism of language disorders. He has published more than 100 papers in international journals, including Nature Neuroscience, Trends in Cognitive Science, Proceedings of the National Academy of Science USA (PNAS), Journal of Neuroscience, Cerebral Cortex, Psychological Science. The total citation is now over 5000 times and H index 37.




主旨发言3


Abstract:

Cultural and linguistic variation can be seen as providing “natural laboratory” environments for investigating both variation and constraints on variation in the human mind, human development and human natural languages. The methodological ideal is often to employ converging methods to attack the same problem. Field investigators thus borrow ethnographic methods from anthropology, and experimental methods from psychology. However, it also has to be recognized that the use of converging methods, although powerful and often highly productive, can also lead to problems of “uneasy fit”. Not only do some proponents of quantitative methods denigrate qualitative methods (and, rather less frequently, vice-versa); but also converging methods may actually produce diverging (and puzzling) results. Part of the problem is that the very notion of replicability in experimental method is antithetical to cultural comparative field research. The “same” experimentally controlled situation (materials, instructions, procedures) will have different meanings in different cultural contexts. The recognition of this by cultural and developmental psychologists in the 1970s prompted the methodological call by Michael Cole and others for ecological validity, and the recognition that experiments are social encounters, not scientific “neutral ground”. In this lecture, I will explore these vexed questions of methodology, validity and generalizability with reference to my own and my colleagues’ work, and to the methodological reflections of cultural psychologists. I will argue that “taking experiments to the field” requires a reflexive stance on the part of the researcher in just the same way as does the use of qualitative methodologies.

 

Chris Sinha is Distinguished Professor of Cognitive Science in the School of Foreign Languages, Hunan University; and Honorary Professor in the School of Politics, Philosophy, Language and Communication Studies, University of East Anglia (UK). He is Past President of the UK Cognitive Linguistics Association and of the International Cognitive Linguistics Association; Founding General Editor (2009-2016) of the journal Language and Cognition; and a current member of four international journal editorial boards and four book series editorial boards. He has authored 3 monographs and 130+ articles and edited 4 volumes. He is currently editing the Oxford Handbook of Human Symbolic Evolution. His central research interest is in the relations between language, cognition and culture, and a main aim of his research is to integrate cognitive with socio-cultural approaches to language, communication and human development. He is experienced in field experimental and observational methods.




主旨发言4



报告概要:

与西方文字系统相比,中文具有很多鲜明的特点。其中一个重要特点就是词和词之间没有空格分隔。我们在阅读中文时,也需要一些特有的认知机制来加工这种文字系统。本报告将介绍了我们近期在中文词切分等中文阅读中特异性的认知机理方面取得的最新研究进展。还将介绍利用计算建模的方法模拟语言加工认知过程的方法,以及利用计算建模技术指导认知科学实验的研究思路。

 

李兴珊,中国科学院心理研究所研究员,博士生导师,中科院“百人计划”入选者。目前是中国科学院行为科学重点实验室主任,是Quarterly Journal of Experimental Psychology副主编,《心理科学进展》编委。他长期从事中文阅读的认知机理研究,利用眼动跟踪、计算建模等研究手段,在中文阅读的词切分及眼动控制领域做了一系列研究,在Cognitive Psychology, JEP: General, JEP: HPP, JEP: LMC等期刊发表学术论文近六十篇。他的一篇研究论文获得美国实验心理学会2011年度最佳论文奖。




主旨发言5



报告概要:

机器翻译是借机器之力自动地将一种自然语言文本翻译成另一种自然语言文本。传统机器翻译模型基于规则,通过语言学专家设计的源语言-目标语言词对相关语法规则进行翻译统计机器翻译统计模型应用于机器翻译,对大量源语言-目标语言文本进行统计学分析,并根据分析结果生成可能性最大的结果。2014年起专门应用于文本生成的新型深度学习编码-解码框架的出现,基于深度学习的神经机器翻译的研究在数年内取得了极大的进展。本报告将介绍基于深度学习的机器翻译主要模型以及本团队近年来在本领域的重点研究工作及最新进展。

 

屈鸿教授是电子科技大学计算机科学与工程学院博士生导师,圭尔夫大学博士后、洪堡大学访问学者。IEEE协会会员,ACM协会会员,中国计算机学会会员CCF中国人工智能学会会员CAAI,四川省计算机学会大数据专委会副主任委员。主要研究领域包括人工智能、神经网络机器学习等。 IEEE Transactions on Neural Networks, Neurocomputing, Neural Processing LettersJCR\二区及信息领域国内外知名期刊和国际会议上发表论文50余篇,申请中国技术发明专利40余项。

 


© Copyright 2019 电子科技大学外国语学院

Admin