The Representation of Translation Equivalents in Bilingual Memory

The Representation of Translation Equivalents in Bilingual Memory

Cogniiive Processing in Bilinguals - R.J. llarris (Ediior) 0 1992 Elsevier Science Publishers B.V. All righrs reserved. I57 The Representation of Tr...

1016KB Sizes 0 Downloads 52 Views

Cogniiive Processing in Bilinguals - R.J. llarris (Ediior) 0 1992 Elsevier Science Publishers B.V. All righrs reserved.

I57

The Representation of Translation Equivalents in Bilingual Memory Jeanette Altarriba University of Massachusetts at Amherst Abstract A fluent bilingual generally has two lexical representations for a single concept in memory, one in each of two languages. Researchers in bilingual memory have investigated the structure of the representation of a word and its translation in memory. Early research focused on translation equivalents in episodic memory while more recent research has focused on their representation in semantic memory. Current studies on facilitation effects in priming for translation equivalents suggest that in fluent bilinguals both words are linked at a conceptual level in memory. It appears that translation equivalents share a common semantic representation. Implications for learning a second language are discussed. Introduction The acquisition of a second language generally begins by learning the translation equivalents for words already known in one's first language. In translating from one language to another, the lexical code is varied while a particular concept remains invariant (Leont'ev, 1973). As learning translation equivalents is fundamental to the acquisition of a second language, it appears useful to investigate the structure of their representation in memory and the process by which the two lexical codes are connected. A primary debate in the bilingual literature centers on the issue of whether a bilingual has two unique memory stores, one for each language, or a single semantic store in which words are "tagged" as to language at the time of output. The first view has been referred to as the independent or dual- store view while the second has been referred to as the interdependent or single-store view (Kolers, 1966; McCormack, 1974, 1977). Another way to characterize this distinction is in terms of semantic or conceptual representation. That is, do translation equivalents have separate semantic representations or do they share a common semantic representation? The aim of this chapter is to review the experimental evidence regarding the representation of translation equivalents in both episodic and semantic memory and to present new data which suggests that for fluent bilinguals, translation equivalents are connected at a conceptual level.

J. Altarriba

I58

Translation Equivalents in Episodic Memory Experimental Evidence Free recall. Lopez and Young (1974) investigated language representation in bilinguals by examining transfer effects for translation equivalents. In their experiment, subjects were familiarized with a list of either Spanish words or English words, i.e., they were asked to read the list several times. Subjects were then placed in one of three conditions. They learned either the same list they had studied, a list containing translation equivalents of the words they had originally studied, or a new list of words. In the learning phase, the word list was presented auditorily, and the subjects then participated in free recall. The subjects were bilingual in English and Spanish and rated themselves higher in English fluency than Spanish fluency. Subjects in the experimental conditions recalled significantly more words than subjects in the control condition, and there was no difference in the amount of transfer observed for the two experimental conditions. Facilitation was equal in the Spanish to English and the English to Spanish conditions. These results appear to support an interdependent view of bilingual memory. If the familiarization of a word or its translation equivalent results in the activation of the same semantic representation, then this familiarization should result in equal amounts of transfer in learning regardless of whether or not the learning is in the same language as the familiarization. These results should also occur regardless of which language is dominant. Although Lopez and Young (1974) discussed their results in terms of the semantic representation of words in memory, their task involved the episodic processing of words and their translation equivalents. In most free recall tasks, memory is episodic and involves memory for a particular event rather than one’s long-term, semantic memory. However, a possible flaw in this study is that subjects might have translated the words during familiarization producing a transfer effect in learning. This might have been true especially if subjects had been asked to rate their fluency prior to participating in the experiment and were aware of the nature of the study. Liepmann and Saegert (1974) also used list learning to investigate the representation of translation equivalents. Arabic-English bilinguals were shown fifteen lists of words in English or mixed in English and Arabic. Items overlapped across lists. In the English lists, a subset of items was repeated across lists. In the mixed lists, the language of a subset of items alternated across lists. The words were presented on slides, and subjects participated in free recall immediately following each list. Recall for successive lists was found to decrease for the mixed lists as compared to the unilingual lists. For mixed lists, subjects had to retain information as to which list and which language an item appeared in whereas for the unilingual lists, subjects only had to remember which list an item appeared in. The results appear to support an interdependent view of language representation in episodic memory. If items had been stored separately or had separate representations in memory, then subjects’ performance would have been better with the mixed lists than with the unilingual lists

Representation of Translation Equivalents in Bilingual Memory

I59

because subjects would have been able to use language to mark items more effectively and would have had less overlap in items from successive lists.

In contrast, the results of another recall study seem to suggest that a word and its translation are represented independently. Watkins and Peynircioglu (1983) investigated the extent to which a word in a study list would facilitate the completion of a word fragment of its translation in a subsequent cued-recall test. Turkish-English bilinguals were presented lists of words in either Turkish or English. The authors hoped that this would minimize translation during study as might have occurred in Lopez and Young’s (1974) study. Each list contained 60 words, 10 of which were buffer words. The words were presented auditorily along with line drawings of each word at a rate of 1 word every 4 seconds. Immediately following the study period, subjects participated in free recall. Subjects were then shown fragment cues for 25 of the 50 words presented in the study list (the consistent condition) and for the translations of the other 25 (the inconsistent condition). Subjects were also shown 50 fragment cues for words that had not been presented in the study list, 25 in Turkish and 25 in English (the no-presentation condition). Subjects performed best in the consistent condition, and their performance was equal for the inconsistent and the no- presentation conditions. There were no facilitation effects in cued recall for the translations of studied words. These results seem to suggest that a word and its translation equivalent do not share the same conceptual representation in episodic memory. The authors concluded that word recall can be direct rather than concept-mediated. Savings. MacLeod (1976) also investigated the representation of information in episodic memory. He used a savings method to examine the information that remained in episodic memory for items that were not recalled (i.e., forgotten) after a five-week retention period. French-English bilinguals learned a list of number-word paired associates in which the words were either in English or in French (e.g., 81apple; 81-pornrne). Five weeks later, subjects were tested for their retention of the words. They were presented with the numbers and were asked to recall the corresponding words. For those words that were forgotten, new words were substituted that matched the original words in meaning or in language. Savings is found when the residual information in the forgotten word facilitates relearning of the new word as compared to an unrelated new word. Subjects showed a savings effect for the meaning of the original words but not for the language in which they appeared. It appears that the savings residual for nonrecallable words contained semantic information in a relatively language-free form. These results appear to support the notion of a single underlying concept for translation equivalents in episodic memory. The translation effect. Clifton, Sorce, Schaye, and Fiszman (1978) used a memory search task to investigate language representation. Spanish-English bilinguals saw sets of 1 to 4 Spanish or 1 to 4 English words. After each set, they were presented with a probe word that was either a word from the set or a translation of a word that had been in the set. Before beginning each session, subjects were told what type of probe

I60

J . Alturrihu

to expect. Subjects’ mean reaction times were longer when the probe word was a translation of an item in the set than when the probe was an actual item from the set. It appears that in the translation condition, subjects held both the presented and translated forms of the words in memory. Subjects then searched through both the presented forms and the translated forms before making a decision. Thus, response times were longer in this condition than in the “no translation” condition. The authors concluded that when a task encourages encoding in a language-specific manner, subjects are able to do so. Although one might be tempted to conclude that the representation of translation equivalents is language- specific, these results reflect task demands of the experiment and do not address issues of conceptual representation.

Limitations of EDisodic Memorv Tasks The tasks described above reflect the means by which translation equivalents are encoded rather than the structure of their semantic representations. The results suggest that bilingual subjects have the ability to encode information in a language-specific manner when necessary. However, subjects can also attend to the meanings of words to the apparent exclusion of language information. Thus, these tasks are informative about language processing in episodic memory and demonstrate how bilinguals can use their language skills to organize information and enhance performance. These studies do not provide sufficient evidence to distinguish the independent from the interdependent view of semantic representation described above. The following section is a review of bilingual studies involving semantic memory tasks. These studies investigated the representation of translation equivalents at a conceptual level in semantic memory. Translation Equivalents in Semantic Memory Experimental Evidence Visual masking. Another paradigm that has been used to examine the representation of translation equivalents is visual masking. O’Neill (1977) used a masking procedure to test the hypothesis that translation equivalents functioned like superordinate-subordinate associates (e.g., animal-horse). Ten French-English balanced bilinguals participated in this study. Target-mask pairs were either translation equivalents, unrelated French and English words, bilingual homophones (e.g., phonefame), unilingual superordinates and subordinates, or unrelated unilingual words. On each trial, subjects were presented with a target word for 20 milliseconds (ms) followed by a 10-ms delay interval. The mask was then presented for 120 ms. The delay interval was increased in steps of 5 ms until the subject correctly identified t h e target. The results are shown in Table 1.

Representation of Translation Equivalents in Bilingual Memory

161

Table 1 Mean Maskine Thresholds as a Function of Tarpet-Mask Relation (adaDted from O’Neill, 1977)

Target-Mask Relation unilingual superordinate-subordinate translation equivalents unilingual unrelated words bilingual unrelated words homophones

Mean Masking Threshold (ms)

77 93 125 149 165

There was no significant difference in mean masking threshold for the first two conditions above. Masking thresholds for translation equivalents differed significantly from masking thresholds for bilingual unrelated words. The mean threshold for homophones was significantly greater than the mean threshold for unilingual unrelated words. The results from O’Neill’s (1977) study suggested that the relationship between translation equivalents in memory is similar to that between same-language superordinate and subordinate words. Translation equivalents were less effective masks than different-language unrelated masks demonstrating the facilitative effect of semantic relatedness in the masking procedure. These results support the idea that translation equivalents share a common conceptual representation in bilingual memory. Priming. A large part of the evidence regarding the organization of words or concepts in bilingual memory comes from studies examining priming effects in lexical decisions for words and their translation equivalents. In a typical lexical decision trial, a subject is presented with one or two strings of letters on a computer screen and is asked to decide whether or not the letters form a real word. Lists include both words and nonwords (e.g., blit). Response time is measured to a target item following presentation of a related prime. The time interval or stimulus onset asynchrony (SOA) between the presentation of the prime and the target may vary. It is assumed that a long SOA leads to more elaborate processing of the prime.

One effect that regularly occurs in the lexical decision task is semantic priming (e.g., Meyer & Schvaneveldt, 1971). That is, a subject’s decision as to whether a letter string is a word or a nonword is facilitated when the target word (e.g., tiger) is preceded by a semantically-related word (e.g., lion).

I62

J. Altarriha

Results of this kind are predicted by the theory of spreading activation (Quillian, 1962). In this theory, concepts are viewed as nodes in a semantic network. The nodes are connected through associative pathways. Activation spreads through the pathways to related areas when the network becomes activated. This spread of activation makes the related areas more available for future processing. Spreading activation theory assumes that when an event is processed, other events are activated to the extent that they are closely related to that event. In the bilingual case, if the recognition of a word (e.g., dog) is facilitated by the previous presentation of its translation equivalent (e.g., pem) it may be assumed that a word and its translation are linked through an underlying concept in semantic memory. Although several studies have examined cross-language priming for semantically-related prime-target pairs (see, e.g., Altarriba, 1990; Frenck & Pynte, 1987; Grainger & Beauvillain, 1988, Keatley, Spinks, & de Gelder, 1990; Kirsner, Smith, Lockhart, King, & Jain, 1984; Meyer & Ruddy, 1974; Schwanenflugel & Rey, 1986), few studies have examined priming for translation equivalents. Kirsner, Brown, Abrol, Chadha, and Sharma (1980), Kirsner et al. (1984), Experiments 1 and 3, and Scarborough, Gerard, and Cortese (1984) failed to find facilitation effects in priming for translation equivalents. However, these studies included relatively long intervals of 10 minutes or more between the presentation of the primes and the targets. These studies were designed to examine the repetition effect, i.e., faster recognition of a word following its repetition as compared to its first presentation. Scarborough et al. argued that the repetition effect at long lags depends on the physical similarity of the stimuli rather than processing at a conceptual level. Therefore, failure to find a repetition effect for translation equivalents does not imply the absence of a common conceptual representation for translation equivalents. Three experiments examined translation priming effects with very short lags between the presentation of the prime and the target. One of these was a study conducted by Chen and Ng (1989). In this study, semantic facilitation and translation priming effects in Chinese-English bilingual speakers were demonstrated with a lexical decision task. Subjects were presented with two blocks of prime-target pairs, one with 30 Chinese-English pairs and another with 30 English-Chinese pairs. Each trial of the experiment began with the presentation of a star signal for one second in the center of the visual field, followed immediately by the display of a prime item for 300 ms. The prime was then replaced by the target item. The subjects’ task was to decide whether the presented target was a word or not. Subjects’ lexical decision responses were facilitated to a greater extent when primed with a translation equivalent than with an unrelated, between-language word. However, this study had several methodological problems that raise questions as to the validity of the results. The SOA (300 rns) might have been long enough to permit the translation of the prime or the use of other strategies on the part of the subjects. Also, the response times averaged over 800 ms in the translation condition. This is quite a departure from the 600-700 ms average reported in previous studies with similar procedures (see Neely & Keefe, 1989, and Neely, 1990, for complete reviews).

Representation of Translation Equivalents in Bilingual Memory

163

If subjects were not responding quickly to the targets, the results might have been influenced by strategic processes. A second study that examined translation priming effects with short prime-target intervals was a study conducted by Jin (1990). Korean-English bilinguals were shown 50 Korean-English pairs and 50 English-Korean pairs in a lexical decision task. Word-word pairs included translation equivalents, associates, and unrelated prime-target pairs. In each trial, a fixation point was shown in the center of a computer screen for 750 ms followed by the presentation of the prime word for 150 ms. The target replaced the prime word, and the subject's task was to decide whether or not the target was a legitimate word. These results support the view that translation equivalents are closely integrated through a common, conceptual representation.

Significant facilitation effects in priming were found for translation equivalents as compared to unrelated target words. These effects were significantly greater for Korean-English word pairs (150 ms) than for English-Korean word pairs (36 ms). One problem with this study, however, was the high proportion of related prime-target word pairs in the stimulus lists (.67). The relatedness proportion effect is the finding that semantic priming increases in magnitude with increases in the proportion of related prime-target word trials (de Groot, 1984; Neely, Keefe, & Ross, 1989; Tweedy & Lapinski, 1981). Tweedy and Lapinski (1981) found that a gradual increase in the number of related prime-target pairs per block of trials enhanced the priming effect in successive blocks presented to the same subjects. A decrease in the number of related prime-target pairs in successive blocks presented to a second group of subjects reduced the effect. One explanation for this effect is that as subjects become aware of the presence of related prime-target pairs, they begin to expect a target to be related to a preceding prime. Subjects may use this information to facilitate their response as they go through a stimulus list. As a result, they will be quick to respond "yes" to the target. Another aspect of the stimulus list that may influence the size of the priming effect is the nonword ratio. This ratio is the probability that a target is 2 nonword, given that it is unrelated to its prime. Researchers have noted that subjects in the lexical decision task check whether the target is related or unrelated to its preceding word prime (e.g., Balota & Lorch, 1986; de Groot, 1984; McNamara & Altarriba, 1988, Neely, 1976, 1977; Neely & Keefe, 1989). Subjects use information about prime-target relatedness to facilitate their wordhonword response. The stimulus lists used by Jin (1990) had a high nonword ratio (.54), and subjects might have used this information to bias a "nonword" response. Again, their results might have been influenced by strategic processes. In order to reduce the use of semantic checking strategies, Neely et al. (1989) suggested using a low nonword ratio. A third study conducted by de Groot and Nas (1991, Experiment 3) examined priming for cognates and translation equivalents in masked and unmasked conditions. The authors argued that when primes are clearly visible, subjects may try to use strategies such as the ones mentioned above to facilitate their responses to target

I64

J. Altarriha

words. Thus, masking the prime should minimize the use of strategies and lead to automatic processing. Dutch-English bilinguals performed lexical decisions on English-English or Dutch-English word pairs. In the masked condition, the prime was preceded by a row of 11 hash marks for 480 ms followed by a 20 ms blank interval. The prime was then presented for 40 ms, and 20 ms later the target appeared. In the unmasked condition the SOA was 240 ms. Cross-language priming for translation equivalents was found in both the masked and unmasked conditions (35 ms and 113 ms, respectively). Although it appears that the masking procedure may have minimized the use of controlled strategies by the subjects, it is unclear whether the subjects actually perceived the prime in a sufficient number of trials (this information was reported for Experiment 2 but not for Experiment 3). Also, as noted by the authors, it is difficult to disentangle the effects of masking vs. SOA as the SOA was always larger in the unmasked condition (240 ms) than in the masked condition (60 ms).

The important point in reference to the present discussion is that de Groot and Nas (1991) did not control relatedness proportion and nonword ratio within their stimulus lists. There has been only one bilingual priming study in which the proportion of related prime-target pairs within stimulus lists was controlled. Keatley et al., (1990) used a relatedness proportion of .25 in their study of cross-language priming in Chinese-English and Belgian-Dutch bilinguals and found evidence of facilitation effects in priming for semantically-related word pairs as compared to unrelated word pairs. However, they investigated priming for primary associates and did not include prime-target word pairs that were translation equivalents. The study described below investigated priming for translation equivalents under conditions that were designed to minimize subjects’ use of the strategies mentioned above. Priming for Translation Equivalents in Contextually Constrained Conditions: A Study Previous cross-language priming experiments have reported facilitation in reaction time for paired translation equivalents. However, a critical examination of this literature suggests that the observed facilitation effects may be attributed to the operation of predictive strategies and, therefore, do not necessarily imply the existence of structural links between the two languages. In some cases, subjects might have had time to translate the prime before processing the target word, and any priming effects observed could have been purely intralingual rather than interlingual. The current study provides a more appropriate test of whether or not translation equivalents are connected at a conceptual level by constraining the lexical decision task to insure that primes will be read without enabling the use of active translations or other conscious, effortful strategies. The approach taken by Neely et al. (1989) in minimizing the role of prime-generated expectancies was adopted here. Spanish-English bilinguals performed lexical decisions on within- and cross-language word pairs. The proportion of related word-primehord-target trials (RP) and the nonword ratio (NR) within the stimulus lists were held low and constant (.33), while

Representation of Translation Equivalents in Bilingual Memory

I65

the stimulus onset asynchrony (SOA) between prime and target was manipulated (200 ms and lo00 ms). Method Subjects. Sixty-four Spanish-English bilinguals were recruited from Florida International University, Miami, Florida to participate in this study. Subjects received credit or payment in the amount of $7.00 for their participation. Table 2 contains a summary of their language backgrounds. Table 2 Summary of Subjects’ Laneuaee Histories SOA Group

Mean age in years Mean number of years in the US. Mean number of years in U S . schools Mean number of years each language spoken: English Spanish Ratings on a 7-point scale (1 =strongly disagree, 7=strongly agree) I read Spanish very well I read English very well Percentage of the day each language spoken: English Spanish

200 SOA N=32

1000 SOA N=32

24 16

25 17

11

12

16 24

18 24

5.9 6.7

5.7 6.4

59% 41%

54%

46%

Desien. The design included one between-subjects factor, SOA (200 ms vs. lo00 ms), and three within-subjects factors, prime-target relation (semantically related, semantically unrelated, direct translation, and neutral), prime language (English and Spanish), and target language (English and Spanish).

166

J . Altarriba

Materials. English word primes were chosen from the association word norms of Postman and Keppel (1970). These primes were then translated into Spanish (Smith, Davies, & Hall, 1989). Thirty Spanish-English bilinguals at Florida International University were asked to provide word associations to the Spanish primes. The most frequently given response in both languages was chosen as the target in each pair for the semantically-related condition. Examples of the words used in this study can be seen in Table 3. Word-nonword pairs were also included. Nonwords were formed by changing one or two letters in unrelated words chosen from the same source as the critical stimuli. Table 3

Relation

semantically-related unrelated repetition/translation neutral

Language Same SUGAR-sweet AZUCAR-duke DIRT-sweet TIERRA-duke SWEET-sweet DULCE-duke READY-sweet LISTO-duke

Different SUGAR-dulce AZUCAR-sweet DIRT-dulce TIERRA-sweet SWEET-duke DULCE-sweet READY-dulce LISTO-sweet

Note. As the present discussion focuses on the representation of translation equivalents, only the response times for those pairs will be discussed in the results. A complete description of the results can be found in Altarriba (1990).

Table 4 shows the number of items per stimulus list used as a function of RP and NR. Critical items were counterbalanced across lists and varied from list to list. A total of sixteen stimulus lists were formed. Each subject saw one list of 256 pairs which was presented with one of the following block orders: EE SE SS ES; SE SS ES EE; S S ES EE SE; or Es EE SE SS. Each of the four blocks was preceded by a practice block of 16 trials. Four obsemations were collected for each subject in each experimental condition.

Representation of Translation Equivalents in Bilingual Memory

I67

Table 4 Number of Items Der Stimulus List as a Function of RP f.33) and NR c.33)

Word mime related word targets semantically related translations unrelated word targets nonword targets Neutral mime word targets nonword targets

96 32

16 16

TOTAL

256

96

*TP=Total pairs; CP=Critical pairs

Note.

To clarify, the 256 trials were divided equally into the four language blocks within each list with the constraint that the RP and NR were maintained within each block as well as within each list overall. Procedure. Subjects were given written instructions which informed them that they would be shown pairs of letter strings on a computer screen and that their task was to decide whether or not the second letter string of each pair was a word in a particular language. A message appeared on the screen at the beginning of each block of trials informing subjects of the language of the primes and targets in that particular block. Subjects were instructed to press the "m" key on the computer keyboard if the second letter string was a real word or the "z"key if it was not. Subjects were encouraged to respond as quickly and as accurately as possible. On each trial, the first letter string appeared in uppercase, left-justified in the center of the screen. After a 200- or 1000-ms delay, the second letter string replaced the first. The second letter string was presented in lower-case letters. Successive letter string pairs appeared as subjects responded. If subjects responded with the wrong key, the word "ERROR" appeared centered on the computer screen. Following the lexical decision task, subjects completed a Language History Questionnaire. An experimental session lasted approximately 45 minutes.

J . AIfarriha

168

Results and Discussion Mean reaction times were computed for each subject and each condition and are shown in Table 5. Response times outside the "upper outer fence" (Tukey, 1977) for each condition were classified as outliers and not included in the analyses of response times. In order to obtain a & for the evaluation of the planned contrasts, the data were submitted to separate 3 (priming condition: related, repetition/translation, or neutral) X 2 (prime language: English or Spanish) X 2 (target language: English or Spanish) ANOVAs for each SOA group. Table 5 Mean Response Times (ms) to Word Targets in the Unrelated and Translation Priminr! Conditions and Mean Priming Effects SOA CroupPrime-Target Language

Condition

SE

200 ES

1000 SE ES

Unrelated

609

749

613

697

Translation

592

679

561

621

Priming Effects:

+17

+70*

+52* +76*

Note.

Priming effects in the translation condition were computed by subtracting the reaction times in that condition from the reaction times in the unrelated condition. The least significant difference for each facilitation effect was based on the & for the Priming Condition X Prime Language X Target Language interaction for each of the two SOA groups: 200-ms SOA = 3,779 and 1000-ms SOA = 3,542. SOA = stimulus onset asynchrony. SE = Spanish-English. ES = English-Spanish. *E < .005, two-tailed. Translation priming effects were highly significant for English-Spanish word pairs at both levels of SOA. Priming effects were also significant for Spanish-English word pairs at the long SOA but not at the short SOA. As explained above, increases in SOA lead to increases in the magnitude of the priming effect as subjects have more time to elaborate and perhaps translate the prime word before responding to the target word.

Representation of Translation Equivalents in Bilingual Memory

I69

It appears that priming effects were greater for English- Spanish word pairs than for Spanish-English word pairs, although this difference was not significant at the long SOA (g > .05). These results were similar to those reported by Jin (1990). In his study, translation priming effects were greater when the primes were in the subjects’ first language than when the primes were in the subjects’ second language. In the present study, the subjects appeared to be more proficient in reading English than reading Spanish. When asked to rate how well they read in each language, the subjects rated English higher than Spanish in the 200-SOA group (E(1,31) = 7.54, & = 1.194, 2 = .Ol) and in the 1000-SOA group (E(1,31) = 6.10, MSe = 1.24, 2 = .02) (see Table 2). Based on this result and the fact that the subjects have been in United States schools for an average of 12 years, it appears that subjects might best be described as English-Spanish bilinguals for the purposes of this study. Thus, the trend in the data suggests that translation priming effects may be stronger from L1 to L2 than from L2 to L1. This result will be addressed in the final section of this chapter. Conclusions An important issue that has been studied in the area of bilingual memory is whether the two languages of a bilingual are stored independently or are mediated through a common conceptual representation. Early investigations involved episodic or short-term memory tasks in which subjects were given an initial orienting task during a study period and were then unexpectedly tested on their retention for the studied material. The language of both the study and test materials was varied. These studies addressed issues of language processing and the representation of meaning in episodic memory rather than the representation of meaning in long-term, semantic memory.

Semantic memory tasks have also been used to investigate bilingual memory. Most of the experimental evidence regarding semantic representation in bilinguals has been gathered using visual masking and semantic priming techniques. Response times to target words are measured following the presentation of a translation equivalent or a semantically-related prime word. The new evidence presented here strongly suggests that translation equivalents share an underlying conceptual representation, and that this representation mediates facilitation in priming between the two words. Although not included in this review, two other tasks that have been used to investigate bilingual memory are bilingual naming and translation (see, e.g., Chen, 1990; Chen & Leung, 1989; Kroll & Curley, 1987; Potter, So, Von Eckardt, & Feldman, 1984). In these tasks, subjects either name words or pictures in a particular language or translate words presented in one language into another language. These tasks are most informative about the nature of lexical processing in bilinguals. Their results suggest that both methods of language acquisition and proficiency in a non-native language are important determinants for the patterns of lexical processing in bilinguals. The following section discusses the implications of the current results for issues concerning second language acquisition and language development.

170

J. Altarrihn Acquiring a Second Language

Research on first- and second- language acquisition is of interest for both theoretical and practical reasons, as research findings may be utilized to teach a second language (see, e.g., Winitz, 1981). A model of language representation recently developed by Kroll and Stewart (1990) and discussed by Kroll, Altarriba, Sholl, Mazibuko, and Stewart (1991) is supported by the current results and illustrates a developmental process in second language acquisition. The model assumes that a bilingual has a large lexical store for hisher first language and a smaller store for the second language. A third store, a conceptual store, is closely linked to the bilingual’s first language. As the bilingual acquires words in the second language, those words are connected via lexical links to words in the first language store. Subsequently, as a bilingual becomes more proficient in the second language, direct conceptual links from the second language store to conceptual memory are also acquired. The model assumes that conceptual links between languages are stronger from L1 to L2 than from L2 to L1, as the former links are the first to be acquired. Therefore, priming is facilitated to the degree that the prime is a member of the larger language store. As a bilingual becomes more proficient in the second language to the point of becoming equally proficient in both languages, it is predicted that facilitation should be about equal in both directions. The above developmental model suggests that as one becomes more proficient in a second language, one is able to directly access conceptual memory from a word in the second language. This conclusion suggests that new methodologies in teaching should be aimed at enhancing conceptual mediation to facilitate second language acquisition. Methods emphasizing concepts and conceptual representation in addition to lexical, word-to-word representation are to be preferred. Acknowledgments This work was supported by NIH Grant HD07327 for training in psycholinguistics and NIMH Grant MH44246 awarded to Judith F. Kroll. The research reported here was based on a doctoral dissertation submitted by the author to Vanderbilt University. References Altarriba, J. (1990). Constraints on interlingualfacilitation effects in priming in m Spanish-English bilinguals. Unpublished doctoral dissertation, Vanderbilt University. Balota, D. A., & Lorch, R. F. (1986). Depth of semantic spreading activation: Mediated priming effects in pronunciation but not in lexical decision. Journal of Experimenral Psychology: Learning Memory, and Cognition, 2,336-345.

Representation of Translation Equivalents in Bilingual Memory

171

Chen, H.-C. (1990). Lexical processing in a non-native language: Effects of language proficiency and learning strategy. Memory and Cognition, 18,279-288. Chen, H.-C., & L u n g , Y. S. (1989). Patterns of lexical processing in a non-native language. Journal of Experimental Psychology: Learning, Memory, & Cognition, 15, 316-325. Chen, H.-C., & Ng, M.-L. (1989). Semantic facilitation and translation priming effects in Chinese-English bilinguals. Memory & Cognition, l7,454-462. Clifton, C., Sorce, P., Schaye, P., & Fiszman, A. (1978). Translation between language in memory. American Journal of Psychology, 9l,237-249. de Groot, A. M. B. (1984). Primed lexical decision: Combined effects of the proportion of related prime-target pairs and the stimulus-onset asynchrony of prime and target. Quarterly Journal of Experimental Psychology, 253-280. de Groot, A. M. B., & Nas, G. L. J. (1991). Lexical representation of cognates and noncognates in compound bilinguals. Journal of Memory and Language,30,90-123. Frenck, C., & F’ynte, J. (1987). Semantic representation and surface forms: A look at across-language priming in bilinguals. - Joumal of Psycholinguistic Research, -6 l, 383-396. Grainger, J., & Beauvillain, C. (1988). Associative priming in bilinguals: Some limits of interlingual facilitation effects. Canadian Journal of Psychology, 42, 261-273. Jin, Y.-S. (1990). Effects of concreteness on cross-language priming in lexical decisions. Perceptual and Motor Skills, 70, 1139-1154. Keatley, C., Spinks, J., & de Gelder, B.(1990). Asymmetrical semantic facilitation bemeen languages: Evidence for separate representational systems in bilingual memory. Unpublished manuscript, University of Tilberg, The Netherlands. Kirsner, K., Brown, H. L., Abrol, S., Chadha, N. K., & Sharma, N. K. (1980). Bilingualism and lexical representation. Quarrerly Journal of Experimental Psychology, 32, 585-594. Kirsner, K., Smith, M. C., Lockhart, R. L. S., King, M. L., & Jain, M. (1984). The bilingual lexicon: Language-specific units in an integrated network. Journal of Verbal Leaming and Verbal Behavior, -32 5 19-539. Kolers, P. A. (1966). Interlingual facilitation of short-term memory. Journal of Verbal Learning and Verbal Behavior, 5, 314-319.

I12

J . Altarriba

Kroll, J. F., Altarriba, J., Sholl, A, Mazibuko, T., & Stewart, E. (1991, June). Asymmetric connections in bilingual memoty. Poster presented at the annual meeting of the American Psychological Society, Washington, D.C. Kroll, J. F., & Curley, J. (1987). Lexical memory in novice bilinguals: The role of concepts in retrieving second language words. In M. M. Gruneberg, P. E. Morris, & R. N. Sykes (Eds.), Practical aspects of memory: Current research and issues Vol. 2 (pp. 389-395). London: Wiley. Kroll, J. F., & Stewart, E. (1990, November). Concept mediation in bilingual translation. Paper presented at the annual meeting of the Psychonomic Society, New Orleans. Leont’ev, A. (1973). Some psychological aspects of language acquisition: Verbal memory, translation, and the significance of errors. Soviet Psychology, lJ, 84-92. Liepmann, D., & Saegert, J. (1974). Language tagging in bilingual free recall. Journal of Experimental Psychology, l O 3 ,1137-1141. Lopez, M., & Young, R. K. (1974). The linguistic interdependence of bilinguals. Journal of Expenmental Psychology, 102, 981-983. MacLeod, C. M. (1976). Bilingual episodic memory: Acquisition and forgetting. Journal of Verbal Learning and Verbal Behavior, l5,347-364. McCormack, P. D. (1974). Bilingual linguistic memory: Independence or interdependence: Two stores or one? In S. T. Carey (ed.), Bilingualbm, biculturalism and education. Edmonton, Canada: University of Alberta. McCormack, P. D. (1977). Bilingual linguistic memory: The independenceinterdependence issue revisited. In P. A. Hornby (Ed.), Bilingualism. New York: Academic Press. McNamara, T. P., & Altarriba, J. (1988). Depth of spreading activation revisited: Semantic mediated priming occurs in lexical decisions. Journal of Memory and Language, 27, 545-559. Meyer, D. E., & Ruddy, M. G. (1974, April). Bilingual word recognition: Organization and retrieval of alternative lexical codes. Paper presented at the annual meeting of the Eastern Psychological Association, Philadelphia, PA. Meyer, D. E., & Schvaneveldt, R. W. (1971). Facilitation in recognizing pairs of words: Evidence of dependence between retrieval operations. Journal of Experimental Psychology, 3, 227-234.

Representation of Translation Equivalents in Bilingual Memory

173

Neely, J. H. (1976). Semantic priming and retrieval from lexical memory: Evidence for facilatory and inhibitory processes. Memory & Cognition, 4,648-654. Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading activation and limited capacity attention. Journal of Experimental Psychology: General, 106,226-254. Neely, J. H. (1990). Semantic priming effects in visual word recognition: A selective review of current findings and theories. In D. Besner & G. Humphreys (Eds.), Basic processes in reading: f i u a l word recognition (pp. 264-336). Hillsdale, N. J.: Erlbaum. Neely, J. H., & Keefe, D. E. (1989). Semantic context effects on visual word processing: A hybrid prospectivehetrospective processing theory. In G. H. Bower (Ed.), llze psychology of learning and motivation: Advances in research and theory Vol. 24, (pp. 207-248). New York: Academic Press. Neely, J. H., Keefe, D. E., & Ross, K. L. (1989). Semantic priming in the lexical decision task: Roles of prospective prime-generated expectancies and retrospective semantic matching. Journal of Experimental Psychology: Learning Memory, and Cognition, l5,1003-1019. O’Neill, W. (1977). Visual masking by translation equivalents in bilinguals. Perceptual and Motor Skills, 45, 1311-1314. Postman, L., & Keppel, G. (Eds.) (1970). Norms of word associafion. New York: Academic Press. Potter, M. C., So, K.-F., Von Eckardt, B., & Feldman, L. B. (1984). Lexical and conceptual representation in beginning and proficient bilinguals. Joumal of Verbal Learning and Verbal Behavior, 23, 23-38. Quillian, M. R. (1962). A revised design for an understanding machine. Mechanical Translation, 2, 17-29. Scarborough, D. L., Gerard, L., & Cortese, C. (1984). Independence of lexical access in bilingual word recognition. Journal of Verbal Learning and Verbal Behavior, 23, 84-99. Schwanenflugel, P. J., & Rey, M. (1986). Interlingual semantic facilitation: Evidence for a common representational system in the bilingual lexicon. Journal of Mernoty and Language, -52 605-618. Smith, C. C., Davies, G. A., & Hall, H. B. (1989). Langenscheidt’s compact Spanish dictionary. New York: Langenscheidt.

I74

J . Altarriba

Tukey, J. W. (1977). Exploratoory data analysk. Reading, M A Addison-Wesley. Tweedy, J. R., & Lapinski, R. H. (1981). Facilitating word recognition: Evidence for strategic and automatic factors. Quarterly Journal of Erperimental Psychology, 3 3 4 51-59. Watkins, M. J., & Peynircioglu, Z. F. (1983). On the nature of word recall: Evidence for linguistic specificity. Journal of Verbal Learning and Verbal Behavior, 2, 385-394. Winitz, H. (Ed.) (1981). Native language and foreign language acqubirion. New York: New York Academy of Sciences.