Stephen van Vlack
Sookmyung Women`s University
Graduate School of TESOL
Developing Bilingualism
Fall 2007
Week 14 - Answers
Kroll, J. and Tokowicz, N. (2001) The development of conceptual representation for words in a second language. In J. Nicol (ed.), pp. 49-71.
1. How do conceptual representations for words develop in subsequent language learners? What are some of the different models?
According to this article dedicated representations for words in a subsequent language learning situation develop first through the first language forms and then with use or exposure are linked in a stronger and more useful way directly to a concept. This is the simple view and it is very similar to what we already talked about in Hamers and Blanc (2000) but under the terms consecutive and coordinate and in this class in general based on Kecskes and Papp (2000), under the term Common Underlying Conceptual Base (CUCB). In short we can basically say that this hierarchical view suggests that people move from a consecutive to a coordinate type of bilingualism. This also supports data from traditional word association experiments where researchers have positive evidence that novice learners of language first form phonologically (Levelt's form) based or syntagmatic links between translation partners or synonyms and later on paradigmatic or semantic based (Levelt's lemma) links become more prevalent. Lexicalists argue that such connections form the basis of the lexicon at different levels of development. Thus, novice learners are going to have a second language mental lexicon that is first based on phonological links. Thus words from the second, new language will be fused onto or connected with similar sounding words in the first language. In these early stages of development or learning there will not be any semantic links. Semantic links only come with more practice and exposure in context. This is the standard theory anyway and the hierarchical model was developed based on this evidence.
This is the first model that Kroll and Tokowicz (2001) give us and they spend a tremendous amount of time not only explaining the model but also explaining where the evidence for the model comes from. In fact, if we look at all the publications the authors cite we find that in one way or another Kroll has been involved in the most all these different experiments. This may very well be why so much time and space is devoted to introducing this model which is obviously flawed due to its simplicity and overall lack of important considerations. In the authors' defense, they do a good job of explaining where the model comes from and then subverting the idea by adding a lot of new considerations. The only problem I have with how the article is that it spends too much time on the original, insufficient model and not enough is spent on the newer model and even a lot of the considerations which they mention are not adequately explored. Faulty also is the way in which the data was collected. I mentioned word association tests above briefly and the effect of these on how lexicalists came to view the lexicon and its development, but nothing is mentioned about this in the actual paper. Almost all the results are obtained through different types of translation texts. Remember, this is not translation in production this is simple translation, finding translation partners and rating the appropriateness of words in different languages as translation partners. As the authors themselves say this is not a very efficient or useful way of trying to figure out how language actually works in a general sense. Such an experimental design calls into question both their findings and their interpretation thereof.
So, taking a look at this original hierarchical model we find that it is overly simplistic for several reasons. The first of these is that the model never really mentions anything about L2 concepts. They mention that concepts may overlap but they never mention that there might actually be different concepts which might be more basic and, therefore, need to form prior to others or which need to be developed exclusively for the L2. Also, as we pointed out in class this actual type of acquisition, which is described in both the hierarchical model and the revised hierarchical model would be based on an extreme model of decontextualized classroom teaching. In a real situation or even in a classroom where the teacher tries to create an environment as authentic as possible we would assume that concepts might actually be learned or activated before the forms are actually learned or even perceived. (This is actually the normal order of operations in naturalistic language acquisition.) This observation is supported by some of the problematic data which they report, but again they don't really mention anything about that. Another possible problem is that they do not mention any other types of lexical linking. They do mention that words have multiple meanings (that is, all words are ambiguous to varying degrees) and that meanings are dispersed across the lexicon in neural networks, but the article is misleading in that they cannot explain what they actually mean by concepts. (How do concepts relate to the multiple meanings all words are known to have?) Certainly concepts are necessary for categorization, and categorization is necessary in order to analyze translation partners, but the meanings of words go beyond just concepts and of course this depends on the type of word in question. Abstract words obviously rely more on concepts than concrete words which have concrete imagens which more specifically help in composing the meaning. What, however, is never mentioned is the role of long-term memory in figuring out the meanings and real uses of words. Kroll and Tokowicz (2001) do mention context as an important factor in activating concepts but again they don't explain it fully or really much all. As a result we windup more or less disbelieving what they argue due to a simple lack of well presented support for a wide variety of peripheral issues.
2. How are L2 learners able to access concepts in the L1 and L2?
How well L2 learners are able to access concepts for their second-language is a very important question and a very interesting one particularly here in Korea where we see some pretty interesting things happening. Obviously, based on the hierarchical model, L2 learners would only be able to access concepts through the L1 in the initial stages, and this would include activating the form which may then interfere with the activation of the necessary concepts to achieve meaning. This basic scenario would seem to make sense because obviously L2 learners are going to need to rely on L1 concepts to a large extent until they can build their own L2 concepts, and if the L2 is never actually experienced meaningfully and in context, then L2 concepts may never be created. In this case what we might end up with is people who have a fairly large amount of L2 forms (as we do find in South Korea) but really have very few L2-specific concepts upon which to base these forms. What we wind up having is a group of students who speak, to a certain degree, the L2 but with the concepts and even the semantic structures of the L1 translation partners. In essence what these people are doing is translating as a chief means of basically effecting use in the L2. Some people are quite good at it and can actually achieve a rather high degree of fluency in certain speech repertoires, but their use of particular forms will be skewed based on their reliance on the underlying concepts and semantic structures of the L1.
Remember! We said in the beginning of this course that language is the mechanism we use to make sense of the world. Language is a kind intermediary system between our systems of perception and the outside world. If you want your students to really use English well they need to form some sort of English world view in addition to a Korean world view. This is obviously achieved by creating new concepts present only in an English system as well as adding on or altering Korean concepts to include English variants to their already present Korean conceptual counterparts. Obviously, developing these concepts will only come from language experience and use. These are not things that we can effectively teach overtly in the classroom, at least not all the time and probably not actually very much. We have to create situations which will allow our students to (1) be exposed to real (meaningful) language use and (2) be able to practice such language in an authentic way. In this way our students will develop their own L2 concepts and extend L1 concepts into a more expanse world view which incorporates elements of L2 conceptualization.
Getting back to the model one of the things that seems to be the most damning for a strictly hierarchical model is the belief that at the early stages of development learners do not have access to concepts in the L1 or the L2, but rather are reliant on just forms or form links. It seems to me from my own experiments with word association tests that learners even at the earliest stages do have access to concepts (of course these are the L1 concepts for the most part) and even L1 episodic experiences with translation partners of the L2 stimulus words. Of course the authors mentioned this briefly in the paper, but again it is a point not fully explained or made use of.
3. Why and how is control such an important mechanism in the use and access of words in bilinguals/multilinguals?
It should be clear to us as we develop a finer understanding of how both the brain and neural networks work in organizing material that activation and inhibition, control mechanisms, form the basis of how we actually are able to use items stored in the brain. We mentioned briefly last week, in relation to Grosjean`s (2001) concept of a base language, about the idea of threshold levels within each node of a neural network (synapse of a neuron). Nodes with lower threshold levels will require less activation to fire. Threshold levels are lowered through frequent use. Not so much an alternative, but rather a more complementary model is that of his strengths and weaknesses of connections themselves. Stronger connections will allow information to move faster and will be used more readily because they're more efficient to use. They become stronger with use, so we can see that both models, slightly different though they may be, support and clearly demonstrate the idea of how competition works to form the shape of our brain as well as the things that we use in our brain. Obviously language is no exception to this overall system in the brain. Knowing this explains why L1 lexical information tends to be used for the L2 whether it is intended or not. Because the access routes and connections in the L1 are so strong, due to many years and years of constant and variant use, it is going to be very hard to form the new links in the L2 which are not somehow overshadowed, dominated and even controlled by the richer, more various, and stronger L1 links. It is for this reason that immersion and submersion programs are seen as being not only greatly beneficial but possibly also necessary in a foreign language learning environment. Such an intensive and, hopefully, meaningful exposure/practice is believed to, in a way, force the learners to develop L2 connections fast and to the exclusion of the L1, which is supposed to remain inhibited over extensive periods.
In the standard Korean classroom for English there is no way that the learners are going to form their own L2 connections simply because they don't get enough practice or the right type of practice as was already mentioned above. It should not be surprising, therefore, that Korean learners of English translate a lot. The way we teach them forces them to use their L1 connections in much the same way as the revised hierarchical model might actually posit. They have few L2 concepts, but many L2 forms. These L2 forms are linked both to L1 forms, but also have strong links to L1 concepts which proficient Korean learners of English are able to use in order to speak English with some degree of fluency and maybe even proficiency. But, again this type of configuration is going to impose a strong glass ceiling on their attained level in English. What we're really talking about is a rampant use of what is become known as Konglish.
Konglish works well in South Korea because other Koreans seem to understand it with little difficulty. English learners are focusing so much on making sure their forms are okay that they don't think it all about different concepts and the meaning they are creating. Nor do they need to when there is little to no focus on meaning or functions for communicating with non-Koreans. Korean students are taught almost exclusively to direct their attention to forms because that is what can be tested. So, we as teachers not only need to enable to our students to build L2 concepts, but we also need to get them to use the L2 concepts with as little interference is possible from L1 concepts. While this is quite hard the simplest way of doing this is to create new and different contexts for the L2 and for L2 use. This will work well at least initially. Certainly at some point the two linguistic systems are functionally going to overlap, we should want this to happen, but to enhance learning in the initial stages and diminish L1 interferences new contexts for the L2 should be developed. Probably the best place for this to occur is within the educational system in schools through immersion type programs.
Herdina and Jessner (2003), Chapter 7 - Answers
1. What are three types of multilingualism according to DMM? Briefly describe them. (pp. 117-125)
Balanced bilingualism
- Ambilingualism : assumed native-like competence in both languages.
- Ambilingual balanced bilingualism is the simplest form of stable multilingualism.
Note that it is assumed that both language systems are fully developed to an ideal native speaker proficiency level and it is presented in non-simultaneous form to assist the graphic representation. (see Figure 25, p.119)
- Non-ambilingual balanced bilingualism must be considered the more likely form of bilingualism where both systems are not considered developed to ideal speaker proficiency level. In non-ambilingual balanced bilingualism two language systems are equally developed but below a native speaker proficiency level. (see Figure 26, p.120)
Unbalanced or asymmetrical bilingualism
We can expect to encounter dominant bilingualism which itself can be divided into transitional bilingualism and stable dominant bilingualism, far more frequently than balanced forms of bilingualism.
- Transitional bilingualism can be considered the most obvious proof of the validity of the maintenance effort hypothesis and the natural monolingualism hypothesis as in this instance LME can be seen as insufficient to maintain more than one full language system. (see Figure 27, p.121)
One language system is gradually replaced by another to result in a long-term reversion to monoligualism. Transitional bilingualism is a gradual transition from LS1 to LS2 as the dominant language system. This development is to be found in immigrant individuals rather than communities, that is individuals who, due to outward circumstances, lose their native communities and have to adapt to new communicative needs as a specified later. 1) In the first phase of transitional bilingualism we will observe transfer phenomena occurring in the developing L2, whilst interference will be low as we observe in normal immersion settings. 2) in the second phase this process will be reversed as the LS2 is by sufficiently stable and LS1 is also under going a rapid process of erosion and has lost its originally dominant position. (See Figure 27, 2004).
Stable dominant bilingualism: The stabilising effect of LME is greater in a utility system than in learner pidgin. Both utility systems and learner language are interlanguage, (Figure 28, p.122) that is fossilisiation in a learner language makes it effectively indistinguishable from a utilitiy or pidgin system. Whilst the utility system is no longer oriented towards a target language system, learner pidgin is still perceived as (communicatively) deficient by the speaker and her/his linguistic enviroment and is therefore more likely to fluctuate, as the multilingual speaker attempts to increase her/his general language effort to overcome these perceived deficiencies. (see Figure 28, p.122).
Multilingualism (see Figure 29a, 29b, p.124)
The types of learner multilingualism are obviously dependent on the acquisition of the first
two languages. Possible means of distinguishing types of bilingualism
1) simultaneous or successive bilingualism
2) primary or secondary bilingualism
Primary learner multilingualism, that is a third language is learnt after primary bilingualism, can be described as the less frequent albeit increasing form of learner multilingualism.
Secondary learner multilingualism refers to the acquisition of the two languages simultaneously after LSp has been acquired.
2. What makes multilingual learners distinguish themselves from monolinguals? (pp. 129 - 131)
It is assumed that well-developed multilingual language systems will lead to a number of factors distinguishing them from monolingual systems, that is, the multilingual learner develops new skills, such as language learning skills, language management skills and language maintenance skills which are linked to a change in quality to be expected in the language learning process. All these skills can be seen as contributing to metalinguistic awareness.
3. What personal and psychological factors affect multilingual proficiency? (pp. 137- 140)
The factors determining language acquisition progress or more precisely, the rate of change in positive or negative language growth, can be subdivided into motivational factors, perceptional factors and anxiety. The relationship among the exemplary factors determines the rate and direction of the development of the new language system. The factors are (multi)language aptitude/metalinguistic abilities (MLA), language acquisition progress (LAP), motivation (MOT), perceived language competence (PC), self-esteem (EST), and anxiety (ANX).
1) (multi)language aptitude/ metalinguistic abilities
It forms part of the M-factor as defined in the DMM and influences the development of the language system.
2) language acquisition process(LAP)
This factor can be positive or negative, primarily an indicator of the rate of change of a language system.
3) motivation(MOT)
Motivational factors are seen as the prime factors in determining general language effort. It is easy to assume that individual motivation will show its effects on the amount of effort put into the acquisition and maintenance of a specific language system and therefore on positive or negative growth.
According to Schumann(1997), motivation is a neurobiological foundation, that will be determined by the speaker?s PCN and the speaker?s interpretation of in how far her/his language competence is able to meet the perceived communicative needs. The psycholinguistic model DMM distinguishes between effective communicative needs and perceived communicative needs, effective competence and perceived competence, is the degree of motivation available can be interpreted as resulting from the difference between the speaker?s perceived competence(how good s/he think s/he is)and the desired goal(how good s/he thinks s/he ought to be).
4) perceived language competence(PC)
The higher the perceived language competence in the language system to be developed, that is, the more the speaker is perceived to meet her/his communicative needs the smaller the effort she/he is going to put into language acquisition. This factor is not necessarily identical to language competence.
5) self-esteem(EST)
Self-esteem or self-confidence is a significant factor in determining the willingness of the individual to engage in communicative activity. It is dependent on the PC or high-self ratings of the individual language proficiency as communicative failure within a language community is to be avoided.
6) anxiety (ANX)
Anxiety is not merely a performance-related phenomenon but acquisition-related. It will also reduce the toleration of communicative blunders, without affecting self-esteem.
It is important that in the outlined prototypical developments of multilingual systems, these factors are considered to be constant, that is, not to affect the development as such.
In addition, these factors not only show their influence on language but also depend on each other in more than merely unidirectional relationships. The feedback loops contained are supposed to show that the system has the ability to determine its own conditions of growth as part of its autopoeitic behavior. The growth and size of the input will determine the rate of growth.