The Role Of Lexical Form And Semantic Senses In Bilingual Masked Translation Priming

Research in terms of bilingual lexicalprocessing is the focus in thefield of bilingual memory study. Masked translation priming paradigm isa very important one used in study on models of bilingual lexicalrepresentation and processing. The previous experimental results usingsuch paradigm have revealed the masked translation priming asymmetry inlexical decision but priming symmetry in semantic categorization.Both of Distributed Conceptual Feature Model proposed by de Groot etal. and the Revised Hierarchical Model proposed by Kroll et al. are unableto simultaneously well explain why translation priming asymmetry existsin lexical decision but priming symmetry in semantic categorization. In1992, Distributed Conceptual Feature Model (DCFM) was proposed claimingthat lexical nodes are associated with a distributed set of conceptualfeatures and translation priming is caused by the activation of theconceptual features common to both L1 and L2. It is unable to account forthe masked translation priming asymmetry in lexical decision. In 1994,the Revised Hierarchical Model was proposed claiming the weak L2form-meaning connection strength is the limiting factor of priming inL2-L1 direction. It claims that the connection-strength between L2lexical form and semantic representation is weaker than that between L1lexical form and semantic representation. Masked L2 prime is incapableof automatically activating the shared conceptual nodes by weak L2lexical-conceptual connection strength so that it can not pre-activateL1 translation equivalent target resulting no priming effect in thisdirection. It is able to explain masked translation priming asymmetry inlexical decision but is unable to explain the task difference, namely, why symmetric masked translation priming exists in semanticcategorization.The Sense Model was proposed by Finkbeiner et al. in 2004 claimingthat the translation priming is caused by the ratio of the primed semanticsenses to the complete semantic representation associated with target.The model is able to simultaneously explain the masked translation primingasymmetry and task difference. However, it only attributes translationpriming to semantic factor. Only considering semantic factor withoutconsidering other possibilities such as lexical form is quitequestionable.The present study aims at probing into the role of lexical form andsemantic factors in translation priming in lexical decision by usingmasked translation priming paradigm. A series of five experiments weredesigned: In Experiment 1, the bilingual masked translation primingasymmetry in lexical decision, namely, robust priming in L1-L2 directionbut no in L2-L1 direction was confirmed. In Experiment 2, we confirmedthe within-language priming asymmetry in lexical decision inFinkbeiner' s study (2004), finding robust masked many-few priming butnot in few-many direction. In Experiment 3, we probed into bilingualmasked translation priming in special condition using L2many-andLlfew-sense word pairs, finding significant priming which furtherdeveloped Finkbeiner' s study and conclusion on the role of semanticfactor in masked translation priming. We begin with the assumption thatnot only the semantic factor affects masked translation priming but alsothe lexical form does. Such assumption was tested in Experiment 4-5. InExperiment 4, we studied cross-language priming using one-sense-wordpairs and the experimental result showed robust priming in Llone-L2onedirection but no priming vice versa, indicating the possibility thatlexical form may be an important factor which can directly influences cross-language priming. Experiment 5 further explored masked priming inLlfew-L2many direction and robust priming was obtained. It can be inferredthat lexical form indeed plays a role in priming and the influence thelexical form has on priming exhibits in the output access at the same time.Considering all the experimental findings, we argued that cross-languagepriming is affected not only by the semantic factor of the language butalso by its lexical form, (e.g., the "form-meaning" input and"meaning-form" output accesses), and also proposed the "Access-SenseModel" in the present study as a possible account of these findings basedon the hypothesis mentioned above.