描述
开 本: 16开纸 张: 胶版纸包 装: 平装-胶订是否套装: 否国际标准书号ISBN: 9787302512684
Preface Ⅰ
List of Abbreviations Ⅺ
Chapter One Introduction 1
Chapter Two Theoretical and Empirical Background 9
2.1 Definition of Key Terms 9
2.1.1 Bilinguals 9
2.1.2 Mental Lexicon 11
2.1.3 Lexical Representation 11
2.1.4 Lexical Processing 12
2.1.5 Lexical Automatic Activation 13
2.1.6 Priming 13
2.2 Theories on Mental Lexicon 14
2.3 Theoretical Models of Bilingual Lexical Processing 17
2.3.1 Bilingual Interactive Activation Model 18
2.3.2 Semantic, Orthographic and Phonological Interactive Activation Model 21
2.3.3 BIA Model 22
2.3.4 Weinrich’s Model 23
2.3.5 Concept Mediation Model and Word Association Model 24
2.3.6 Revised Hierarchical Model 26
2.3.7 Distributed Concept Feature Model 28
2.3.8 Shared Asymmetrical Model 29
2.3.9 Modified Hierarchical Model 30 2.3.10 Three-Stage Model 32
2.3.11 Sense Model 33
2.3.12 Language Mode Hypothesis 34
2.3.13 Inhibitory Control Model 34
2.3.14 Summary of the Theoretical Models 35
2.4 The Empirical Studies on Bilingual Activation 37
2.4.1 Lexical Decision Task 39
2.4.2 Semantic Relatedness Judgment 55
2.4.3 Translation 62
2.4.4 Picture-Word Interference Task 68
2.4.5 Other Tasks 76
2.4.6 Limitations of the Previous Empirical Studies 79
2.4.6.1 Studies at the Sub-Lexical Level 79
2.4.6.2 Types of Bilinguals Investigated 80
2.4.6.3 Research Technique 81
2.4.6.4 Stimuli 82
2.4.6.5 Stimulus Onset Asynchrony (SOA) 82
2.5 Summary 83
Chapter Three Methodology
3.1 Research Questions 85
3.2 Overview of the Experiment Rationale and Design 87
3.2.1 Rationale 87
3.2.2 Design 90
3.3 Methods in Experiment 1 91
3.3.1 Participants 91
3.3.2 Experiment Design and Materials 92
3.3.3 Experiment Procedures 95
3.3.4 Data Recording and Analyses 97
3.3.4.1 Behavioral Data 97
3.3.4.2 Electrophysiological Data 98
3.4 Methods in Experiment 2 99
3.4.1 Participants 99
3.4.2 Experiment Design and Materials 100
3.4.3 Experiment Procedures 101
3.4.4 Data Recording and Analyses 103
3.4.4.1 Behavioral Data 103
3.4.4.2 Electrophysiological Data 104
3.5 Summary 104
Chapter Four Results and Discussion
4.1 Results 108
4.1.1 Behavioral Data 108
4.1.2 Electrophysiological Data 117
4.1.2.1 Visual Inspection 117
4.1.2.2 Midline Statistical Analyses 126
4.1.2.3 Lateral Statistical Analyses 135
4.2 Discussion 146
4.2.1 The Locus of L1 Chinese Automatic Activation During L2 English Processing 146
4.2.1.1 The Locus of L1 Chinese Automatic Activation in Deep Processing 147
4.2.1.2 The Locus of L1 Chinese Automatic Activation in Shallow Processing 155
4.2.1.3 Summary of the Discussion on Locus 157
4.2.2 The Time Course of L1 Chinese Automatic Activation During L2 English Processing 157
4.2.2.1 The Time Course of L1 Chinese Automatic Activation in Deep Processing 159
4.2.2.2 The Time Course of L1 Chinese Automatic Activation in Shallow Processing 161
4.2.2.3 Summary of the Discussion on Time Course 171
4.3 Summary 171
Chapter Five General Discussion 173
5.1 Predictions of Two Theoretical Models 173
5.1.1 Predictions of the BIA Model 173
5.1.2 Predictions of the RHM 177
5.2 Extended Hierarchical Model 179
5.3 The Cognitive Mechanism of L1 Chinese Automatic Activation During L2 English Processing as Is Predicted in the EHM 180
5.4 Summary 183
Chapter Six Conclusion 185
6.1 Major Findings 185
6.2 Implications 188
6.3 Limitations and Recommendations 191
Bibliography 195 Appendices 211
Appendix A: Language Learning Background Questionnaire 211 Appendix B: Experimental Stimuli Materials 212
Vocabulary knowledge is a core component of language com-petence, and vocabulary acquisition plays a vital role in second language acquisition research. In recent years, with the integration with other disciplines such as psycholinguistics, neurolinguistics and cognitive sciences, vocabulary acquisition of second language has massively expanded its research field and has nurtured a series of new hot issues. One of them is the bilingual lexical automatic activation. Many studies have found out that L1 may be automatically activated during the processing of L2 words. By far, the research of L1 automatic activation in L2 word processing has made great progress, yet there are still some limitations. In terms of research participants, most studies take second language learners with alphabetic L1 as participants, but little attention has been paid to English learners in China with non-alphabetic L1. In terms of research content, there is still a lack of comprehensive and systematic exploration on the locus (conceptual level/lexical level/sub-lexical level) and time course of L1 automatic activation. Methodologically, most studies have only used the basic reaction time (RT) technique rather than the event-related potentials (ERP) technique, which has a high temporal resolution and is able to make a real time record of the cognitive processing.
Besides, the cross-language tasks that have widely been used provide a bilingual context themselves and bias toward a dual-language activation pattern, which may lead to participants’ conscious rather than automatic activation of L1 and hence result in interference to the observation of L1 automatic activation. Theoretically, most of the theoretical models have been proposed based only on the studies on alphabetic languages. It needs verification whether or not these models can be applied to the ideographic languages such as Chinese, but little research has worked on it.
Regarding these facts, the present study intends to investigate the L1 automatic activation during L2 word processing of the highly proficient and unbalanced Chinese-English bilinguals (Chinese learners of English), by combining RT technique and the ERP technique, and applying the semantic relatedness judgment task and masked lexical decision task. The research questions involve:
(1) Where is the locus of L1 Chinese automatic activation during L2 English processing? That is, at which level or levels (conceptual level/ lexical level/sub-lexical level) does the activation take place? What is the difference in activation locus between the deep processing and shallow processing? (2) What characterizes the time course of L1 Chinese automatic activation during L2 English processing? That is, what are the features of the activation at the early and the late stages of L2 English processing respectively? What is the difference in activation time course between the deep processing and shallow processing?
To answer these questions, two experiments have been carried out. Experiment 1 was a semantic relatedness judgment task. Twenty-one participants judged whether the prime words and the target words were semantically related. Experiment 2 was a masked lexical decision task. Fourteen participants judged whether the target words were real words. The core materials in the two experiments were the same 210 English word pairs. Half of them were semantically related while the other half were not. Although these English word pairs did not share any morphological repetition, their Chinese translation equivalents eventually concealed a hidden repetition of the first character, repetition of the second character or no repetition. Thus, the manipulation of the two variables came to a within-subject 2 (semantic relatedness: related vs. unrelated) × 3 (hidden character repetition: first character repetition vs. second character repetition vs. no repetition) factorial design. Based on this design, the semantic relatedness was manipulated to measure the L1 automatic activation at the conceptual level and the position of hidden character repetition was manipulated to measure the L1 automatic activation at the lexical and sub-lexical levels. The ERP components N400 and LPC were extracted to measure the L1 automatic activation at the early stages and the late stages respectively. The SOAs of the first and the second experiment were set 700ms and 67ms respectively, in order to investigate the L1 automatic activation in deep processing and shallow processing.
Based on the analyses of the RT and ERP data, the following findings have been revealed:
1) In shallow processing, no semantic priming effect or hidden repetition priming effect was observed. However, in deep processing, there was a significant effect of semantic priming and hidden repetition priming when Chinese learners of English were processing L2 words. It indicated that L1 Chinese automatic activation took place at the conceptual level, lexical level and sub-lexical level. That is to say, L2 word processing was accompanied with multi-level L1 automatic activation. The results supported the predictions of hierarchical structure of bilingual mental lexicon in the BIA Model. Besides, the asymmetry of hidden character repetition effect at the sub-lexical level has been found out for the first time. That is to say, the effect of the hidden character priming of the second character was more salient than that of the first character, which indicated that the automatic activation of the second character was stronger than the first character.
2) In shallow processing, no ERP effect was observed. However, in deep processing, there was a significant N400 effect at the early stages and an LPC effect at the late stages during L2 word processing of Chinese learners of English. It indicated that the L1 Chinese got automatically activated quickly at the early stages during L2 English processing and lasted for a relatively long time to the late stages. Take the above mentioned findings into consideration, it could be inferred that in L2 word processing, the visually input L2 word forms activated the concepts first, which subsequently activated the corresponding L1 word forms, and further activated the second character and the first character at the sub-lexical level. The L1 was not intended to be consciously activated but was the “by-product” of the activation of L2. The results supported the predictions of pathway of activation in the RHM.
In consideration that the above findings only partially fit the theoretical predictions of the mainstream models such as the BIA Model and RHM, the present study tentatively analyzed the features and limitations of the BIA and RHM that were widely applied to alphabetic languages, and proposed the Extended Hierarchical Model (EHM), which was compatible with the features of Chinese. It turned out that this model could better account for the phenomenon of L1 automatic activation during L2 word processing of English learners in China.
The contributions of the present study are mainly manifested in
the following four aspects. (1) In terms of research participants, the investigation into the high proficiency unbalanced Chinese-English bilinguals enriched the types of research participants and provided more evidence for the universality of bilingual activation. (2) In terms of research content, this study made a comprehensive exploration of the three loci (conceptual level, lexical level and sub-lexical level), two time periods (the early stages and the late stages) and two processing depths (the deep processing and the shallow processing) of L1 automatic activation, which expanded the research dimensions of bilingual activation. (3) Methodologically, the measurement of this study involved the combination of two techniques (RT and ERP), which expanded the dimension of data collection. Furthermore, this study modified the hidden repetition priming experimental paradigm used in previous studies and designed a module of stimulus materials for the measurement of bilingual activation at the sub-lexical level, which provided methodological reference for the future research. (4) Theoretically, this study tentatively constructed the Extended Hierarchical Model of L1 automatic activation during L2 word processing. This model could be applied to account for the features and mechanism of Chinese-English bilinguals’ L1 automatic activation.
This book is organized in the following way. Chapter One is the introduction of the study, including genesis and significance. Chapter Two defines several key terms in the present study and reviews theoretical models and empirical studies in bilingual activation. Chapter Three proposes the research questions, discusses the rationale, and reports the methods used in the two experiments. Chapter Four reports and discusses the findings on the locus of L1 automatic activation during L2 processing, as well as the time course of L1 automatic activation during L2 processing. Chapter Five continues to make some general discussions based on several theoretical models, and proposes the Extended Hierarchical Model to account for the findings of the present study. In the end, Chapter Six concludes the major findings, summarizes the theoretical, empirical, methodological and pedagogical implications, points out the limitations of the present study, and comes up with recommendations for future studies.
Xiao Wei 2017.11
Figure 2-7 Revised Hierarchical Model (Kroll & Stewart, 1994)
2.3.7 Distributed Concept Feature Model
Distributed Concept Feature Model (DCFM) (De Groot, 1992) satisfactorily overcomes the problem of the disagreement in form-concept mapping. As the name indicates, DCFM is a distributed model. This model assumes that a concept of a word can be further divided into several tiny, basic units. There are links from a word to its corresponding conceptual features. In a bilingual lexicon, the representations of some words are largely shared across languages, while the representations of some words share fewer semantic features and are less overlapped. As De Groot (1992, 1993) claimed, the translation equivalents for concrete words share more overlap in semantic features than those for abstract words, due to the fact that concrete concepts tend to be more steady across languages than abstract concepts, which are more influenced by particular cultures and societies.
Despite of DCFM’s superiority, its weaknesses are as obvious. First, compared to RHM, it lacks a focus on the dynamic changes of L2. In the RHM, the status of connections in mental lexicon changes over time, while in the DCFM everything seems to be set without change. This is obviously contradictory to the facts in L2 development. Second, the assumption of distributed features resembles the classic theory of semantic features, which is a simplified version of meaning description. Cognitive linguistics has already found out the effect of prototype. Some semantic features in a category are more basic and salient, while some may be peripheral. The DCFM assumes equal importance of the features, which is contradictory to the prototype theory. Third, it assumes an equal amount of semantic feature overlap across languages, which implies a symmetrical linking. This assumption ignores the asymmetry of L1 and L2 lexical-conceptual linking, hence it has difficulty in explaining the commonly found asymmetry effects in bilingual studies.
Vader Father Idee Idea
Lexical memory
Conceptual memory
Concrete words Abstract words
Figure 2-8 Distributed Concept Feature Model (De Groot, 1992)
2.3.8 Shared Asymmetrical Model
Combining the advantages of RHM and DCFM, Dong et al. (2005) proposed Shared Asymmetrical Model (SAM). In this model, L1 and L2 forms are linked with each other. They are also linked to the common conceptual elements, L1 specific elements and L2 specific elements. Therefore, the differences between L1 and L2 can be clearly seen in the model. There are L1-specific, L2-specific and common elements in the conceptual level, which hypothesize that L1 and L2 have their unique elements as well as shared parts. Hence, the problem of cross-language nonequivalence as is seen in the DCFM can be solved. Different from DCFM, in the SAM, it is also possible that the forms in one language can be linked to the concepts in another, rather than be restrained in the language-specific elements as is depicted in the DCFM. Besides, the dynamic changes in the second language acquisition are reflected in the SAM. The form-concept connections are usually stronger within the same language than cross language. The L1 forms have stronger connections to the common elements than the L2 forms do, mainly due to the fact that the L1 form–shared concept link may have already be set up before the involvement of L2.
Figure 2-9 Shared Asymmetrical Model (Dong et al., 2005)
2.3.9 Modified Hierarchical Model
Modified Hierarchical Model (MHM) retains the developmental progression from lexical to conceptual mediation in L2 development, as is assumed in the RHM, and the notion of full, partial and no share in bilingual concepts, as is assumed in the SAM or DCFM. Therefore, it has several new advantages compared to the earlier models.
Compared to the RHM, which assumes an integrated conceptual storage, the MHM assumes that the concepts may be fully shared, partially overlapped or fully language-specific. This is superior to the RHM in that there is always conceptual nonequivalence across languages, and MHM faithfully assumes this feature. This theoretical hypothesis leads to a very interesting inference. If a bilingual speaks in one language and would like to express a concept that is specific in another language, he or she has to resort to the lexical form of the other language by code switching, which results in cost of cognitive resources. From this point of view, the MHM comes to a similar conclusion on language switch with the Inhibition-control model proposed by Green (1998).
Compared to the DCFM or SAM, the MHM is able to better account for the dynamic changes with L2 learning. It regards L2 learning as processes of conceptual reconstruction. The main goal of L2 learning is to reconstruct the concepts and to develop the target-like linguistic categories (Pavlenko, 2009). The processes of restructuring go along with the L2 acquisition. The L2-specific categories have been neglected before L2 learning. These categories are not differentiated in L1, thus difficult to be acquired. For example, Chinese does not have the subjunctive aspect while English has such an aspect. A Chinese learner of L2 English usually finds the subjunctive aspect difficult to learn because it is not necessary to make such a distinction in the L1 Chinese. Only after the category of subjunctive aspect is set up can a leaner develops the relevant L2 abilities.
Lexical.links
评论
还没有评论。