مدارک و شواهد برای برانگیختن تکرار طولانی مدت بین زبانی در انجام وظایف حافظه ضمنی مفهومی

Previous studies have failed to find evidence for long-term cross-language repetition priming (e.g., presentation of the English word frog does not facilitate responding to its Dutch translation equivalent kikker on a later presentation). The present study tested the hypothesis that failure to find cross-language repetition priming in previous studies was due to the use of tasks that rely primarily on lexical or orthographic processing of the stimuli instead of conceptual processing. Consistent with this hypothesis we obtained reliable cross-language repetition priming when conceptual implicit memory tasks were used. The present results support theories of bilingual memory that assume shared conceptual representations for translation equivalents. In particular, our results support the concept mediation model as they indicate that bilinguals access the conceptional representation directly from the L2 lexical representation (i.e., without first accessing the L1 lexical representation).

It is estimated that the majority of the people in the world are bilingual. It is not surprising therefore that many researchers have become interested in how language is represented and processed in people who master more than one language. A question that has received considerable attention in the literature is whether the two languages of a bilingual person have separate or shared (also called integrated) representations. Initially researchers assumed separate representations for translation equivalents (e.g., Kolers, 1963). Many current theories of bilingual processing and memory representation, however, assume that representations of translation equivalents are integrated to some degree (e.g., Kroll & Stewart, 1994; Potter, So, Von Eckhart, & Feldman, 1984). A number of findings have been taken to support this view. For example, semantic priming effects are obtained even when the prime and target words are presented in different languages (e.g., de Groot & Nas, 1991; Kirsner, Smith, Lockhart, King, & Jain, 1984). Additionally, Stroop interference is obtained when ink colors must be named in a language different from the one in which the words are printed (see MacLeod, 1991, for a review). Also, in a category-verification task, reaction times are as fast when the category and its member are presented in the same language as when they are presented in different languages (e.g., Caramazza & Brones, 1980). These and other findings are consistent with the notion of shared bilingual memory representations (see Francis, 1999, for a review). Results obtained with the long-term repetition priming paradigm, however, have provided little evidence for shared representations. 1 Repetition priming refers to the common finding, well documented in the monolingual memory and language literature, that participants respond faster and more accurately to recently studied words than to words that have not been studied recently (e.g., Jacoby & Dallas, 1981; Scarborough, Cortese, & Scarborough, 1977; Zeelenberg, Wagenmakers, & Raaijmakers, 2002). The question is whether repetition priming is also obtained when the first and second presentation of a word in the experiment are in a different language (e.g., does the previous presentation of the English word frog facilitate responding to its Dutch translation equivalent kikker?). A number of experiments using the lexical decision task indicate that no such effect is obtained ( Gerard & Scarborough, 1989; Kirsner, Brown, Abrol, Chadna, & Sharma, 1980; Kirsner et al., 1984; Scarborough, Gerard, & Cortese, 1984) when the translation equivalents are noncognates (i.e., when they are orthographically and phonologically dissimilar). The absence of cross-language repetition priming seems to be inconsistent with the idea of shared representations. Several researchers (for recent reviews, see de Groot, 2002; Gollan & Kroll, 2001) have argued, however, that cross-language repetition priming is not obtained in lexical decision because performance in lexical decision depends primarily on orthographic or lexical processes. In accordance with many current theories of monolingual and bilingual language processing, these researchers distinguish between a lexical and a conceptual level of representation. It is often assumed that the lexical level represents the orthography of a word whereas the semantic or conceptual level represents the meaning of a word. Most current theories of bilingual memory representation assume separate or language specific representations for translation equivalents at the lexical level of representation and shared or integrated representations at the conceptual level of representation. Because these theories assume separate lexical representations for translation equivalents they predict no cross-language repetition priming for tasks that tap lexical processes. A closely related explanation for the absence of cross-language repetition priming in lexical decision is provided by the transfer appropriate processing (TAP) framework (Blaxton, 1989; Morris, Bransford, & Franks, 1977; Roediger, 1990; Roediger & Blaxton, 1987). The TAP framework accounts for a number of findings in the implicit memory literature (see Roediger & McDermott, 1993, for a review). According to the TAP framework, memory performance and hence priming, depends on the extent to which the processes at retrieval recapitulate the processes at encoding. Within the TAP framework a distinction is often made between perceptual and conceptual memory tasks. Performance in perceptual tasks relies primarily on the processing of the physical attributes of the presented stimuli whereas performance in conceptual tasks relies primarily on the processing of the semantic attributes of the presented stimuli. Cross-language repetition priming is not found in tasks that emphasize perceptual processes, because there is little or no overlap in the perceptual processes at study and test. The interpretation that priming in lexical decision depends primarily on perceptual, orthographic or lexical processes and not conceptual processes is consistent with the ideas of many researchers in the field of monolingual repetition priming (e.g., Becker, Moscovitch, Behrmann, & Joordens, 1997; Bowers, 2000; Zeelenberg & Pecher, 2002). It should be noted, however, that the evidence supporting the hypothesis that cross-language repetition priming is not obtained in lexical decision because priming depends on lower level form-based processing is circumstantial. There is no direct evidence supporting this hypothesis. The aim of the present study was to provide a more direct test of this hypothesis. A prediction that follows from the reasoning outlined above is that cross-language repetition priming should be obtained in implicit memory tasks that rely on conceptual processing. Some support for this idea was obtained by Durgunoglu and Roediger (1987). They used the TAP perspective to account for the fact that in some memory tasks cross-language transfer is observed while in other tasks no such transfer is observed. They predicted that evidence for cross-language transfer would be found in tasks that emphasize conceptual processing, but not in tasks that emphasize perceptual processing. In agreement with this prediction, their results showed that in free recall, a conceptually driven task, language at study played little role. However, in word fragment completion, a data-driven task, language at study played a major role and performance was no better for words studied in a different language than for nonstudied words (i.e., there was no evidence for cross-language repetition priming). The results of Durgunoglu and Roediger (1987) provide some support for the idea that cross-language repetition priming might be obtained in tasks that rely on conceptual processing. It should be noted though that the Durgunoglu and Roediger study provides no definitive evidence because word fragment completion and free recall not only differ in the extent to which they rely on perceptual versus conceptual processes. These tasks also differ in the sense that word fragment completion is an implicit memory task and free recall is an explicit memory task. Numerous studies have shown dissociations between implicit and explicit tasks even when only conceptual tasks are considered (e.g., Goshen-Gottstein & Kempinsky, 2001; Humphreys, Tehan, O’Shea, & Boland, 2000; Shimamura & Squire, 1984; Weldon & Coyote, 1996). Thus, an alternative interpretation of the Durgunoglu and Roediger results is that cross-language transfer is obtained in explicit tasks but not in implicit tasks. A more specific reason why cross-language transfer might be obtained in explicit tasks but not in implicit tasks is that in explicit tasks memory performance might be influenced by translation strategies at test (see Francis, 1999, for a more elaborate discussion of translation strategies). Such strategies are less likely to occur in implicit memory tasks, because subjects are not tested for their memory, but are simply asked to perform a certain task (e.g., decide whether or not a string of letters forms an existing word or decide whether a word refers to something ‘animate’ or ‘inanimate’). Thus, it remains to be shown that cross-language repetition priming effects occur in conceptual implicit memory tasks. The large majority of studies investigating repetition priming have used tasks that rely primarily on perceptual or orthographic processes.2 Recently, however, priming in conceptual implicit memory tasks has gained more attention (e.g., Becker et al., 1997; Mulligan & Stone, 1999; Vriezen, Moscovitch, & Bellos, 1995). Some examples of conceptual implicit memory tasks that have been used in previous studies are animacy decision, man-made decision, free association and category-exemplar production. These conceptual implicit memory tasks, however, have been used almost uniquely in monolingual repetition priming studies. To our knowledge, the only experiment that investigated cross-language repetition priming in a conceptual implicit memory task (animacy decision) was performed by Kirsner et al. (1984, Experiment 3). Somewhat surprisingly, Kirsner et al. failed to find evidence for cross-language repetition priming. Thus, even in conceptual implicit memory tasks there is no evidence for cross-language repetition priming. The absence of cross-language repetition priming in the Kirsner et al. (1984) study seems to suggest that the explanation that cross-language repetition priming in lexical decision is not obtained because priming in this task depends primarily on perceptual or orthographic processes and not conceptual processes might be incorrect. Given, however, that this is the only published experiment investigating cross-language repetition priming in a conceptual implicit memory task we wanted to take a further look at cross-language repetition priming in these tasks. More specifically, the present study investigated cross-language repetition priming in animacy decision and man-made decision, two semantic classification tasks. It is important to ensure that priming in semantic classification tasks is indeed due to conceptual processing, otherwise we might still fail to obtain evidence for cross-language repetition priming. A number of findings from the monolingual repetition priming literature indicate, however, that priming in semantic classification tasks indeed depends on conceptual processing and therefore these tasks are well suited to study cross-language repetition priming. These monolingual studies have shown that in semantic classification tasks: (a) repetition of the lexical processing of an item produces little or no priming (Franks, Bilbrey, Lien, & McNamara, 2000; Vriezen et al., 1995), (b) priming is not affected by changes in modality of presentation (auditory vs. visual) from study to test (Thompson-Schill & Gabrieli, 1999), and (c) priming is affected by the extent to which there is an overlap in the type of conceptual processing at study and test (Huntjens et al., 2002; Thompson-Schill & Gabrieli, 1999; Vriezen et al., 1995). The latter point is of particular importance because it suggests that we may fail to obtain repetition priming if the study and test tasks emphasize different aspects of the conceptual knowledge of a word stored in memory.3 Because research indicates that priming in semantic classification tasks depends largely on specific conceptual knowledge, words in Experiment 1 were presented in an animacy decision task (i.e., participants judged words to be living or nonliving) during both study and test. In the same-language condition, the target words were presented in Dutch (the native language of our participants) during both study and test. In the cross-language condition, target words were presented in English during study and in Dutch during test. The question was whether cross-language repetition priming would be observed under these circumstances, as is predicted by theories that assume shared representations for translation equivalents at the conceptual level of representation.