北京大学中文系学士、理论语言学硕士，荷兰Leiden大学心理语言学博士；心理学、语言学、信息科学及技术教授；宾夕法尼亚州立大学神经科学中心主任，美国心理计算学会主席等；Bilingualism: Language and Cognition主编，Frontiers in Language Science副主编，Journal of Cognitive Science编委等 。研究兴趣包括认知神经科学、心理语言学、双语加工、语言习得的计算及神经机制等。
Neurocognitive and Computational Mechanisms of
Lexical Acquisition and Representation
Department of Psychology and Center for Language Science
Pennsylvania State University
How does a child rapidly acquire a vast number of words and develop a structured mental organization for them within the first years of life? How does a bilingual individual deal with the even more complicated task of learning and representing two lexicons? In this talk, I outline an approach based on our research that takes a dynamical perspective toward the lexicon, and discuss how this approach can be applied to account for lexical organization, structural representation, and dynamic competition within and across languages. In particular, I provide computational evidence based on the DevLex model, a self-organizing neural network model, and neuroimaging evidence based on functional magnetic resonance imaging (fMRI) studies, to illustrate how children and adults learn and represent the lexicon in their first and second languages. Our computational and neuroimaging studies suggest that language-specific input, linguistic experience, and learner characteristics jointly contribute to the time course of learning and the neurocognitive representation of single and multiple languages. Our research sheds light on the cascading effects of early learning on later development and the child-adult differences in language acquisition.
Motivator-based contingency model in the resolution
of Syntactic ambiguity
莫雷 (Lei Mo)
South China Normal University
Two competing and largely incompatible classes of model dominate current sentence processing research. One is called Linear view, in which the processor makes initial decisions on the basis of strategies defined in terms of syntactic information alone and uses thematic information (such as determined by minimal attachment and late closure) (e.g., Ferreira & Clifton, 1986).According to this class, the processor computes syntactic analyses serially, in two stages. In the first stage, it draws on a restricted range of information to construct an initial analysis. During the second stage, it accesses other sources of information, which may sometimes cause it to abandon its initial analysis and compute another.
The second class of model assumes that the processor can activate multiple analyses in parallel. It employs both syntactic and nonsyntactic information in a single stage to foreground one analysis, but other analyses remain activated (e.g., MacDonald, 1994; Trueswell, Tanenhaus, & Garnsey, 1994), which is called Parallel view or competing model. The best known account of this class is the constraint-based model.
In the framework of the contingency model, which mechanism the subjects will explore to deal with the syntactic ambiguity corpus depends on the contingent properties of the ambiguity motivator. The so-called "ambiguity motivator" refers to the information point that promotes and forces people to make a choice between the alternative analysis. According to the contingent properties of the ambiguity motivator, there are three conditions in the resolution of syntactic ambiguity.
In the first case, the ambiguity motivator does not appear in any position of the corpus. In this case, the individual will generate an analysis based on their syntactic knowledge, context and their preference and so on. They won’t attempt to do any deep processing to resolute the ambiguity. In the second case, the ambiguity motivator appears in the ambiguity region of the corpus. In this case, people will use the competition mechanism to parse the ambiguity corpus. In the third case, the ambiguity motivator appears in the region of the disambiguation region of the corpus. In this case, people will select the reanalysis mechanism to parse the corpus of the disambiguation and the subsequent regions.
In current paper, according to the motivator-based contingency model，the series of experiments were designed. Both the English syntactic ambiguity corpus and the Chinese syntactic ambiguity corpus were used to investigate the mechanism that people explored to process the ambiguity corpus in three conditions which composed of the absence of the ambiguity motivator, the ambiguity motivator in the disambiguation region and ambiguity motivator in the ambiguity region. The results of the experiments showed that when the ambiguity motivator didn’t appear in any region of the corpus, people would apply the mechanism that they used to parse the unambiguity corpus to process syntactic ambiguity coupus. When the ambiguity motivator appeared in the disambiguation region, people would adopt the reanalysis mechanism to parse the corpus of the disambiguiation and the subsequent region. When the ambiguity motivator appeared in the ambiguity region, people would select the competition mechanism to parse the ambiguity corpus.
Keywords: syntactic ambiguity; reanalysis model; competing model; ambiguity motivator；contingency model
Typical and dyslexic development in learning to read Chinese
State Key Laboratory of Cognitive Neuroscience and Learning
Beijing Normal University
Compared with research on alphabetic languages, research on reading acquisition and impairment in Chinese has a relatively short history. In the past of thirty years, lots of research has explored reading acquisition of Chinese children and the mechanism of developmental dyslexia in Chinese. However, many research questions related to the universal and specific properties of Chinese normal and dyslexic children are still unanswered.
In my talk, I will firstly describe some important features of Chinese language, and how these features influence reading acquisition of normal Chinese children. Then, I will summarize a series of studies on dyslexic development in learning to read Chinese, in which the critical cognitive deficits of Chinese dyslexic children and the underlying dysfunction were identified. Finally, a longitudinal study will be reviewed, in which the early predictors and developmental trajectories of reading acquisition and impairment in Chinese children were explored. Research has revealed that Chinese children with dyslexia are mainly suffered at the accuracy and speed of character or word recognition. Phonological awareness and naming-speed are the two deficits shared by both dyslexic children in Chinese and in alphabetic languages. Morphological and orthographic awareness are particularly important to consider in understanding Chinese reading development and dyslexia. The longitudinal studies reveal that it is possible to identify the school-age poor readers from early stage.
葉彩燕教授于美國德州大學奧斯丁分校獲得語言學學士學位，並于南加州大學獲得博士學位。研究興趣包括: 雙語獲得、第二語言獲得、粵語、潮州話及中國語言比較語法、心理語言學及認知科學。著作包括: Interlanguage and Learnability: from Chinese to English (由Benjamins出版社出版)；合著了一系列關於粵語語法的書籍，如Cantonese: A Comprehensive Grammar (已有日語譯本) 、Basic Cantonese、 Intermediate Cantonese (由Routledge出版社出版)。她的專題著作，The Bilingual Child: Early Development and Language Contact（與Stephen Matthews合著，由劍橋大學出版社出版）榮獲了美國語言學學會頒發的2009年Leonard Bloomfield 圖書獎。
葉彩燕教授帶領其團隊創建了香港雙語兒童語料庫（the Hong Kong Bilingual Child Language Corpus），該語料庫是首個記錄粵英雙語兒童縱向發展的雙語語料庫，更是位於卡內基梅隆大學（Carnegie Mellon University）兒童語言研究資源交換系統(CHILDES) 中最大的雙語多媒體語料庫。葉彩燕教授是香港中文大學兒童雙語研究中心主任，同時擔任Bilingualism: Language and Cognition 及 Second Language Research的編輯委員會委員。
One Child, Two Languages: Chinese Windows
on Bilingual Development
Childhood Bilingualism Research Centre
Department of Linguistics and Modern Languages
Chinese University of Hong Kong
In the simultaneous acquisition of two languages, the relationship between the target languages plays an important role in determining the nature and direction of crosslinguistic influence. So far the majority of studies in BFLA involve Indo-European language pairs which represent a limited sample of the possible language combinations in terms of global linguistic diversity. The diversity of language pairs in bilingual acquisition offers new windows into what the possibilities are in development. BFLA will be better understood when investigated against a rich background of linguistic diversity, extending the database to typologically unrelated languages with very different structures. For example, properties of Chinese such as null arguments, word order in relative clauses and wh-questions raise new possibilities for interaction between a bilingual child’s developing linguistic systems.
A number of influential studies in BFLA have concluded that the child’s two languages do not interact significantly, developing much like their monolingual counterparts. These studies have largely involved European language pairs such as English with Dutch or French with German. Recent studies of the bilingual development of Chinese and English call this picture into question. In this paper, we review findings in Cantonese-English grammatical development showing that core properties of Chinese grammar can be transferred to English, including wh-in situ, null object and prenominal relative clauses. Certain properties of Cantonese are also susceptible to influence from English. In some domains, such as that of verb particle constructions and directional verb complements, bidirectional influence is seen. We discuss possible reasons why interactive development is found in some studies but not in others. Studies of bilingual development involving Chinese languages provide windows for viewing developmental processes and pathways and enriching our theoretical and empirical investigation. These studies also provide unequivocal evidence for crosslinguistic influence in bilingual development.
Annette de Groot is Professor of Experimental Psycholinguistics at the University of Amsterdam. Her work has concentrated on word recognition and the structure of the mental lexicon, the psychology of reading, and bilingualism and multilingualism. Her present research focuses on bilingualism, with a special interest in bilingual word processing and foreign-language vocabulary acquisition. She is the author of Language and Cognition in Bilinguals and Multilinguals: An Introduction, which was recently published by Psychology Press.
Parallel language activation during bilingual word processing
Annette M. B. de Groot
University of Amsterdam
The results of a substantial body of studies have led some authors to conclude that bilingual word processing is fundamentally language-nonselective in nature. This means that bilinguals, while processing words, cannot switch of the contextually inappropriate language, even not when task performance takes place under unilingual circumstances. In other words, the two languages of bilinguals are always both activated while they perform some word-processing task and the co-activated knowledge structures of the non-target language influence processing.
In the vast majority of the studies that gave rise to the above conclusion that bilingual word processing is language nonselective the words to be responded to (the “targets”) were presented to the participants visually or aurally (instead of having to be generated from internal thought processes, as in word production tasks) and each target was presented in isolation, that is, without being embedded in a context such as a sentence or larger linguistic fragment. Studies of this type have shown, in various ways, that lexical information in the non-target language is generally co-activated with lexical information in the target language.
Still, this type of evidence does not yet warrant the conclusion that bilingual word processing is fundamentally language nonselective. After all, during normal language processing words are never presented in isolation but as part of larger and meaningful linguistic structures. The larger linguistic context of a word may modulate the recognition process, for instance by constraining activation to the contextually relevant language. To legitimately draw the conclusion that bilingual word recognition is fundamentally language-nonselective, evidence of co-activation of lexical information in the contextually inappropriate language should therefore be obtained from experiments that present the critical words in natural linguistic units such as sentences.
In addition, evidence of language-nonselective word recognition as obtained in both in-context and out-of-context experiments would still let in the possibility that only word recognition, not word production, is fundamentally language-nonselective. In word production the speaker is not driven by bottom-up information (as the reader and listener are) but intentionally chooses the target language. It is therefore possible that the speaker can exert some control over the memory representations that are activated during speech production (e.g., Costa & Santesteban, 2004).
In fact, the present question regarding the language-(non)-selective nature of bilingual word processing has also been addressed in a couple of experiments that examined bilingual word production out-of-context and in a few that examined bilingual word recognition in sentence context. Furthermore, in our own laboratory we recently ran what we believe to be the first set of experiments that examined the present question by looking at bilingual word production in sentence context (Starreveld, De Groot, Rossmark, & Van Hell, submitted). In my talk I will review the evidence from these four types of studies on bilingual word processing – those on word recognition in context and out of context, and those on word production, again in and out of context – and discuss a few variables that appear to constrain the language-nonselective nature of bilingual word processing. In addition I will briefly present some views on how bilinguals, despite their relatively “noisy” language system, manage to keep their languages separate.