L2 Orthographic Activation In Reading L3 Sentences

Bilingual phenomenon nowadays is quite common around the world. Second language acquisition (SLA) has drawn great attention among numerous researchers because of its complicated mechanism. Bilingual representation, bilingual lexical access and cross-linguistic influence (CLI) are among the hottest issues in SLA studies. Researchers have developed a number of explanatory theories to address these issues, and new evidence is continuously explored to justify these theories. In recent years, a new language phenomenon, trilingual acquisition has begun to establish itself as an independent discipline, providing new perspectives to bilingual researches.Previous studies suggest that an earlier acquired language will be automatically activated in the processing of a later acquired language, leading to either facilitation or inhibition. It is of theoretical and pedagogical significance to find out why and how the earlier acquired language(s) become activated in multilinguals. So far, the studies on automatic activation in trilingual process have mainly focused on the automatic activation of first language (L1). Of the few studies which highlighted the second language (L2) activation, participants are generally from western countries whose languages are alphabetic. This study therefore will take Chinese trilinguals(English-German) as participants to explore the activation mechanism of L2 orthography in third language (L3) processing with the help of eye-tracking technology.Two experiments were carried out to answer three general research questions: (1)Will L2 orthography be activated in L3 processing? (2) Will semantic constraint interfere with the activation of L2 orthography in L3 processing? (3) Will L3 proficiency affect the activation of L2 orthography in L3 processing? Both experiments were set up with tasks by reading natural L3 sentences. Experiment 1 included 36 English-German cognates (18 identical; 18 non-identical) and 18 controls,and 18 sets in total. For each set of materials, a sentence was constructed that could comprise both the cognate (identical and non-identical) and its control as one of the middle words. Altogether there were 54 sentences divided into 18 groups accordingly,which constituted a single factor experimental design with 3 conditions. Experiment 2 consisted of 19 English-German cognates and 19 matched controls. For each target word, a low- and high- constraint sentence context were constructed respectively,resulting in 76 sentences in total. Thus, a 2 (word type: cognates vs. matched control)×2 (sentence type: high-constraint vs. low-constraint) experimental design was conducted. 46 participants volunteered to participate in the experiments. All the participants were instructed to comprehend a L3 sentence presented on the screen.Eyelink 2000 eye-tracking system provided by SR Research Ltd. was used as both the experiment builder and data recorder. All data collected were imported to Statistical Package for the Social Science Software (SPSS 15.0), by which descriptive analysis and repeated measures ANOVA were conducted.This study provided answers to the above mentioned research questions: (1) L2 orthographic features were automatically activated during L3 processing, and this activation was observed for both early and late measurement. It was evidenced that L2 activation started from the shallow level with orthography and moved on to a deeper level, which is consistent with the predication of BIA+ Model and TIA Model. (2)Semantic constraint exhibited relatively limited interference to the activation of L2 orthography, and the semantic constraint had an early influence on L2 orthographic activation, that is, the semantic constraint effect occurred in an early stage, and at a shallow level of processing. (3) The influence of L3 proficiency was not as significant as that of the semantic constraint. Specifically, the influence of L3 proficiency was observed at neither early stage nor late stage, while it was obvious in the processing depth of L2 orthographic activation. High-L3 participants were inclined to take the processing to a deep level, indicated by greater degree of cognitive loadings in recognizing words of different cognate conditions or words from different semantic contexts; whereas the low-L3 participants were to process at a shallow level. These findings of current study are consistent with the prediction of RHM, and are also in accordance with the prediction of TIA Model usually applied in a trilingual context.The significance of this study lies in the following aspects: (1) theoretically, this study presented a theoretical model which could be used to justify the mechanism of the L2 automatic activation in L3 processing; In addition, this study involved both semantic factors and orthographic factors in language processing, which would surely provide new perspectives to the research on CLI; (2) methodologically, this study provided more evidence for the investigation of trilingual representation and access by taking Chinese trilinguals (English-German) into consideration and by employing eye-tracking techniques which would raise the accuracy of the data collected. In addition, a new set of materials was compiled, which might provide reference for the researches in the future; (3) to language policy and planning in our country, the findings obtained in this study could cast light on the design and improvement of the programs for foreign language (FL) teaching and learning, especially those programs involving more than one FL, and which are in urgent need of a scientific way to cope with the negative influence from L3.