Bilingual Memory - Essay Example

? Research in bilingualism has been very extensive since Uriel Weinreich, Ervin and Osgood paved the way through their seminal work. These scholars made suggestions compound/coordinate distinction that have since helped in the evaluation of the question on language storage aspect. That is, how do bilinguals store and organize language; is it as a single entity or in two small stores? Prior to this, there were three experimental approaches as used in bilingualism. They include word association and naming, language transfer and interference and recognition and recall. Experiments conducted (based on the three approaches) however, showed very contradicting results with regard to knowledge organization. For purposes of reconciling the three methods, scholars developed tag models. This involved the ‘tagging’ of each word in the store with the language that it belonged to. To date, this aspect of language-tagging has reminded very fundamental in bilingualism. In its modified form, it is evident in various bilingualism models. However, language tagging still remains a very debatable issue in bilingualism. There are various factors that have paved way for the research in bilingual memory in the present time. Firstly, experimental psychology has been influenced to a great extent by information processing (IP). Secondly, there has been increased understanding on the importance of experimenting data from a task-specific analysis approach. Thirdly and perhaps most important is the realization that there is an urgent need in bilingualism to separate semantics from their lexical instantiation. The information processing approach resulted in experimentation using automatic processing tasks. These are tasks that are driven by spreading activation. They are then measured by reaction times. The tasks in question included priming and interference tasks among others. In addition, during the 70s, it was observed that in order to understand the contradictory results as evidenced in prior research, there was need to interpret data gathered from bilingual memory in a task-specific approach. This led to an agreement though from a general point of view on the need to separate conceptual and lexical levels. At this point, conceptual level was shared by the two languages while the lexical level was specific to each. There was support for a single conceptual store as it was evident from various results showing cross-language semantic priming. On the other hand, support for separate lexical stores was evidenced by the lack of cross-language repetition priming. ‘Hierarchical models’ are three and largely based on this conceptual and lexical separation. The models care concept- mediation, word-association, mixed and revised hierarchical models. All of them share a common structure whereby each consists of two separate lexical stores (for the two languages) and one conceptual store. Location and weighting of between the first language (L1) and second language (L2) lexical and conceptual nodes are the common d determinants of a specific hierarchical model. Further, knowledge on bilingual memory organization has been added from studies on data from bilingual patients who have brain lesions. For instance, translation disorders are evident in some bilingual aphasics. This supports to a large extent, the revised hierarchical model. Some disorders are therefore described as breakdowns of specific links in the above model. Critics of hierarchical models base their argument on memory structure. They argue that the memory structure of a bilingual showed variations on numerous factors. Some of these factors include the abstractness of a given word, the part of speech it belongs to and whether or not the translation of the word was cognate. In the hierarchical models in question, the two lexicons may interact to a certain extent, however, they are separate. This separation as proposed was as a result of the lack of long-term repetition priming. For instance, the presence of the ‘chien’ in a list containing French words may not in any way make word recognition faster compared to the orthographically dissimilar translation that is equivalent to the word ‘dog’ in a list containing English words. Such an argument on separate-lexicon structure with regard to bilingual memory still poses a ‘level-of-observation’ problem. When one considers this argument from a perceptual perspective, it is observed that repetition priming consists of similar perceptual components. If no repetition priming is observed between orthographically dissimilar synonyms, one cannot conclude that each word belongs to a different lexicon. The question therefore is, why should it be concluded so when equivalent priming effects are observed between translation-equivalents which are orthographically dissimilar. There are two more arguments in bilingual memory organization for separate lexical stores. The first one is that when language between two lists that are supposed to memorized are changed, what is observed is a release from proactive interference. The second argument is based on language recall whereby bilinguals perforce language specific recall effectively of words presented previously. Again, the above arguments are still open to ‘level-of-observation’ problem. A critical look at the release from proactive interference reveals this. While it has been a useful investigation monolingual studies using two lists of semantic category, there has been no conclusion that there are two ‘lexicons’ one for each category. In the same manner, why should it be concluded with regard to a bilingual study that there are two separate ‘lexicons’ (in this case language lexicons)? In addition, recall performance in specific the role that context plays is a well-established concept. Studies conducted by Marian and Neisser reveal that with regard to memory retrieval, language is the context cue. As such, good language-recall performance can also be attributed as the product and not necessarily as separate language storage. Most importantly it might be a product of the contextual effect. This is further discussed below. Various studies on language conducted at the neuro-anatomical level are seen to further weaken the separate lexical stores and language processing argument. The studies show that the phenomenon cannot be understood without regard to various factors that range from age of acquisition of language, level of processing to language proficiency. They further emphasize on the variability extent even at the bilingual level. Recent research shows that in semantic processes, there is a possibility of a common neural substrate. One specific issue of Bilingualism: Language and Cognition discussed broadely on the concept of functional magnetic resonance imaging (fMRI) as well as evoked response potential (ERP) approaches used in the study of different aspects of bilingual memory. These aspects include inter-lingual homograph recognition, language switching and bilingual aphasia. The description of bilingual memory has taken centre stage in the traditional approach of the phenomena. However, not much time has been taken to study the emergence and evolution of that organization. It is out of this concern that current models of neural network in the study of bilingual memory indicate that some types of organization may have been as a result of learning (language acquisition) mechanisms that were encompassed into the model. Specifically, these are unsupervised learning mechanisms. There are three connectionist models have been developed. Firstly, the bilingual simple recurrent network (BSRN) was developed by the French. It revealed that after an individual has been exposed to both languages sufficiently, the internal word representations form a cluster with regard to language, regardless of the fact that the activation patterns of the said representations still overlap. The second model is the self organization of bilingual memory (SOMBIP) whereby the production of language separation relies on unsupervised learning. The third distributed connectionist model was developed by Thomas (Thomas, 1997). The organizational structure of both BSRN and SOMBIP is produced as a product of their inputs (respectively). The third model produces its organizational structure from ‘static’ tags. In this case, problems are encountered with regard to the origin of the high level languages tags that are explicit. The question herein does not lie on whether tags are essential in the production of effective language separation. Rather, it is whether they are necessary in the production of that separation. The SOMBIP model relies on a series of phonological cues while the BSRN model relies on word associations statistics. In both, language separation is as a result of input to the specific model. As earlier discussed, there is a rather general agreement on overlapping organization with regard to bilingual memory. As such, this discussion will now delve on lexical access and lexicon selection. The term lexical access is used to refer to the process whereby one is able to activate the ‘right’ word in a specific context. This concept is interdisciplinary as it is also used outside the concepts of bilinguals. Various scholars have contributed on the topic of whether lexicon specific access exists. This refers to the access to a specific language at a time (J.G. Van Hell, 1998). However, many studies have revealed that non-selective access exists. This refers to access of words in both languages at the same time. The latter is the generally accepted notion. One of the scholars who counters this issue is De Groot interprets Gerard and Scarboroughin’s selective access approach as a non-selective access approach. The question however, is, if non-selective access exists in one intergrated bilingual lexicon, why do bilinguals experience very little trouble when it is necessary to remain with one language? Again, there lies the problem of lexicon selection. One of the attempts that have been utilized successfully in explaining lexical selection is the bilingual interactive activation (BIA) model. It is based on the McClelland and Rumelhart’s proto-connectionist interactive activation model whose basic assumption is an integrated lexicon. During its development, its major concern lied in exhibiting how the recognition of a word (that it belongs to one language or the other) occurs. The model has proved very successful in the extension to bilinguals the single-language effects. Again, this model does not escape the ‘level-of-observation problems as seen in the BIA. This is because of its assumption of non-selective access and single integrated bilingual lexicon. Notably, regardless of the fact that BIA can be ‘integrated at the lexical items level - it uses language nodes which are equivalent to tags in order to inhibit (selectively) all words in one language thereby producing the lexical selection – above this lexical level, it loses its integrated nature. There are various problems that still remain at the lexical organization level. They include evidence of the need to double represent inter-lingual homographs. Fortunately, the extended version of the BIA and SOPHIA (BIA’s computer implementation) have done overcome these problems to a large extent through explicit tagging. The architecture of the BIA involves the incorporation of phonetic information. It then makes clear distinction between new component that are designed to handle task demands and the identification system. This echo’s the non-selective access model referred to as Green’s inhibitory control (IC) model. Both the BIA and IC concur with Green’s proposition that there should be separation between a bilingual lexico-semantic system and those procedures set to act on the system. Green’s model is actually not the only one that specifies on the use of inhibitory mechanisms in the production of lexicon selection. However, it is unique in that it insists on two mechanisms, one that operates on bilingual lexicon-semantic system inputs and the other, on the outputs of the system. The two distinct mechanisms work through a process whereby they allow for lexical selection which is highly determined by the task schema. They also limit any influence that may be caused by the not-in-use language. Additionally, they involve two processes that are discussed below. The first is an automatic/bottom up process which occurs within the lexicon-semantic system. It is driven by stimulus input and involves the modification of activation level in the system (lexicon-semantic). The second is an intentional/top-down process that changes the response of an individual to signals from the bilingual lexicon-semantic system. However, it does not modify the levels of activation in the system. The description above is the process of the bilingual lexicon-semantic system regulation that works to explain Grosjean’s proposed language modes. In his proposition, bilinguals remain in either one or the other language without any difficulty. This means that the IC model has inhibitory mechanisms that are not language-specific. Contrary, there are other scholars who propose that language specificity in lexical selection can be achieved when one considers lexical nodes in the activation level only of the language in question. The BIA, IC and the BIAt models have challenges with regard to lexical organization. However, they are unique due to their bottom-up and top-down processing of lexical selection aspect. Additionally, they consider what the problem-solving strategies and goals of the participants are. This is mainly through an ‘experimental task’ approach. This indeed paves way to allow for the incorporation in these models of other aspects. For instance, there is need to determine whether perceptual processing alone is adequate for task demands or semantics have any role. In addition, when studies are conducted in a bilingual setting, provides better understanding on bilingual memory mechanisms as wells as other general mechanisms of memory. Research conducted recently on the interactions present between cognitive function and linguistic experience indicates that bilingualism has a positive influence on both non-linguistic cognition and some linguistic processing notions. With regard to the non-linguistic aspect, it has been concluded that bilingualism facilitates in children (Bialystok & Martin, 2004) selective attention and inhibitory control. The same has been observed in adults as evidenced by Bialystok, Craik & Ryan (2006). In the linguistic aspect, bilingualism has been said to be responsible for the performance of meta-phonological tasks in children. Numerous authors have successfully demonstrated the effect of mechanisms that are involved in task demands. Not only have these authors discussed various that are involved in the modification of performances among bilinguals such as age of acquisition and proficiency, they have also exhibited the dissociations that are evident between tasks. Some of them have delved deeper to analyze a variety of mechanisms involved in the processing when a task is performed. They have taken into account, both within language and cross-language aspects. Two of these authors however, stand out in this argument. Zeelenberg and Pecher argue that there exists identical process when one considers monolingual and bilingual aspects. An example that demonstrates this can be analyzed in Marian and Neisser’s experiment in explaining the role of the task. These two authors validate the language dependency of autobiographical memory. This is later interpreted to imply that indeed, there exists separate stores. However, they are keen to point out that these results can also be attributed to the process of encoding and retrieval of the test. They also revealed that autobiographical recall is highly influenced by the linguistic context. Although the above two papers differ with regard to experimental tasks and procedure, they are based on similar logic which is largely an extension of classical memory theory to the issue in question, bilingual memory. They convey a message that bilingual memory is memory with the implication that its study may be guided by the general principles of memory. There are some scholars who delved on the relationship between emotion and bilingual memory. One of these research looked at the cross cultural recognition of basic emotions (five). They are joy, anger, sadness, disgust and fear. It was revealed by the study that recognition was better in the native language compared to any other. One questionable aspect however, is to what extent did linguistic ability influence performance. Altarriba (2002) posits that there are more extensive studies concerning the memory and perception of monolinguals compared to bilinguals. A study conducted by Javier (1993) involved individuals recalling an emotional memory and later describing it in the opposite language. They were bilingual Spanish-English speakers. It was found out that greater detail of the story as well as better organizational skills were present when they performed the task in the native language compared to the other language. To add on to that, there are some scholars who have a general concurring statement that people’s first language is the best when it comes to working memory. In the education realm, information is mainly conveyed verbally usually involving groups of students. As such, listening abilities as well as language comprehension become fundamental tools in learning. They require that one has good skills in auditory perception as well as auditory working memory (WM). Working memory is defined as a system that involves maintain and manipulating information on a temporal basis, that is necessary for complex cognitive activities which include reasoning, comprehension and learning (Baddeley, 1992). The working memory consists of a verbal working memory (vWM) often referred to as the ‘phonological loop’, a critical factor in language acquisition not only in the development stage, but also in daily life’s language processing. In contrast, it has been noted that WM may be limited when students undertake learning in an environment where the instruction language is not their native language (McDonald, 2006). This is explained by the lack of WM resources in L2. In an experiment, participants were asked to complete four tasks consisting of different conditions. The first involved a L1 memory task with disturbance while the second, L1 without disturbance. The third task involved L2 memory task without disturbance and the fourth, L2 with disturbance. Ten English or Chinese words were shown to the participants. The result was that there was better performance when participants engaged in their first language. Additionally, in situations where there was no disturbance, performances were better compared to those with distraction. Read More