Stephen van Vlack

Sookmyung Women`s University

Graduate School of TESOL

Developing Bilingualism

Fall 2007

Week 13 - Answers

Grosjean (2001) The bilingual`s language modes.

1. What is language mode in Grosjean`s view and how does this different from other views we have read?

Grosjean's view of language mode basically revolves around the idea of one specific linguistic code forming the base system from which language will be produced. This particular code functions as a base simply because it receives and has received higher activation. It is more highly activated and harder to inhibit so it is, in effect, always available for use. This is a very different view than the switch hypothesis which we read about earlier in Hamers and Blanc (2000) because it allows two or more language codes to operate at the same time within a single base system. The model also links well with the idea of the CUCB.

       This theory of language mode was developed based on Grosjean's observations of code-switching and code-mixing in both bilingual and multilingual individuals. In such situations, for the most part, it seems that bilingual speakers function within the confines of certain base code and that the switches into different codes occur on top of, or in respect to, a specific base. This is particularly true in the case of code mixing (when certain smaller elements from a different code are inserted in small amounts into a longer piece of discourse). Now it is important to remember that just because there is a specific base at any one time doesn't mean that the base can't or in fact doesn`t change. In fact, what Hamers and Blanc (2000) defined as `code switching` Grosjean would probably be referred to as a `base change` by Grosjean.

       He also identifies three basic different states that bilinguals or multilinguals find themselves in when they speak. The first of these is the monolingual mode. In this mode the speaker sticks very closely to the norms of the base system and does not deviate very much from that system. The second mode is an intermediate stage in which a small but easily recognizable series of code mixes occurs on top of or in respect to the base system. These are more tentative, often pragmatic types of code mixes. The final mode would be one of fully bilingual or multilingual mode where there are much more and much larger switches either on top of a still presiding base or there can even be base system switches. In this way we concede that Grosjean's model, while seemingly simple, is quite a departure from some of the other models we have studied to date. Whether he understands it or not Grosjean is inadvertently providing support for a dynamic or holistic model of language because his model assumes that systems can be accessed at the same time, which means they might need to be part of one larger, interconnected system.

       It should be clear in looking at this model that it is reliant on the idea of an integrated system for different languages. If different languages were totally separate then there would be little way for them to code mix so quickly and effectively. To code mix quickly we need a common store, so it would seem, at least at the conceptual level. Obviously, words which share the same concepts will have different lexical properties in relation to their different codes. Likewise, this model provides counter-evidence against a Chomskyan model of language which sees linguistic competence as being equated with grammatical knowledge alone. Evidence from code switching and language modes shows language to be primarily lexical. Here it should be obvious that grammatical systems play only a secondary role.

2. What are some of the variables that affect language mode?

Grosjean claims in this paper that language mode itself is more a product of subconscious choice. This doesn't mean that bilinguals don't sometimes make conscious decisions, but most of the time since language production occurs so quickly speakers have to do things without actually consciously being aware of what they're doing (so dies a strict version of the monitor hypothesis). In order for this to happen there must be some sort of trigger which causes us to behave in a certain way. As with everything else in language, and as we now know about all natural systems in the world, triggers can come from both external and internal sources (causes). In the case of language mode the external trigger is going to be the overriding factor in making the decision about which linguistic state a speaker will be in, but this does not mean that internal triggers are not also very powerful and in reality an external trigger usually works to trip an internal one - the work in cahoots. The external trigger is simply the audience or other elements of external context, like location. If a speaker (Wendy, to give her a name) of language A and B is speaking to somebody (Peter) who only speaks language B then obviously Wendy is going to be limited to the monolingual speaker mode, or at least she was going to try to stick to that mode as best she can. In such a situation it makes no sense to code switch (unless she dislikes or is angry at him and wants to disaccomodate) because Peter is simply not going to understand her. Now, if Wendy is now talking to Maurice, who also speaks both languages A and B, she has the possibility to code mix. And she certainly may mix a bit. How much she ends up code mixing or indeed code switching depends on the location, situation, and how formal it may be as well as how well she knows Maurice. Remember that code switching is often a sign of familiarity or closeness. It is akin to speech accommodation. It is a way of bringing someone closer by accentuating similarities between the speakers.

       People who just met each other even if they are somehow aware that they know the same languages will probably limit their code switching (intermediate mode) until they know each other little better. So this most simple variable is the one which first allows someone to code switch or not code mix at all. That is, are they going to be in monolingual mode or some other mode. Once that decision has been made then depending on how well they know the person, or the topic of conversation, or the situation they are in they will make a decision about the amount and the type code switching they will engage in. That is, they will decide which mode is possible, an intermediate or a bilingual mode. All these decisions are based on external triggers which help people make important decisions about the language and the language forms they are going to use. What is also interesting is that these external triggers seem to be more or less predictable. Nothing to tricky here in relation to external triggers. This is not necessarily the case for the internal triggers.

Internal triggers are tricky because they vary widely from person to person based on their internal cognitive makeup. The internal triggers come from the types of connections between different types of knowledge networks and the strength of these connections within the entire linguistic system of the speaker (memory plays a large role in this), but one major aspect also relates to the overall activation levels of the specific code within the language system as a whole. Simply put, the amount of use (internal or external) a specific code gets will more or less determine with its resting activation levels. The more a language is used the higher its resting activation levels will be. This means that it takes much less electricity (energy) to use a code with high activation levels. A specific linguistic code which has low resting activation levels will generally take much more energy to use and may, therefore, be less used.

       The specific code which has the highest resting activation levels is simply the dominant language among all the other languages that a speaker may have. The fact that one language is easier to use provides a very simple explanation for language interference. If we return to our hypothetical speaker, Wendy, who we said speaks languages A and B, and she is speaking to Maurice, who also speaks A and B, but in this case Wendy's dominant language is A while Maurice's dominant language is B then we could have some trouble. Let's assume that Wendy is a nice person and she likes Maurice A LOT, then she is going to try to accommodate to Maurice A LOT. She will then try to use B as the base code for her communication. In such situation, depending on how strongly dominant A is in Wendy's system she will probably not be able to escape from code switching, and if she's talking to Maurice it's OK. If, however, she's talking to Peter, who speaks only B and does not speak A, she may not seem like a very good speaker to him. This is in part due to internal triggers resulting from the high activation levels of connection points within A which she has difficulty inhibiting. Therefore, elements of language A are going to seep into your her production of the B base and through this we can see how internal triggers can explain language interference. Bear in mind that this is just one small aspect of the possible range of internal triggers. Exactly how this works will be reviewed in the answer to the next question.

3. Does Grosjean`s idea of language mode have psychological reality and if it does how would this affect our view of language.

Grosjean's concept of language mode does seem to have psychological reality simply because it is observable in the actual behavior of all bilingual multilingual users of language. Everybody who has access to more than one code does in fact code switch and code mix, often without realizing it. In fact, as Grosjean mentions in the paper, it is very easy to extend this idea from bilinguals and multilinguals to monolinguals. The idea is that even within a monolingual system there are different linguistic strands, in a way codes, which the speaker uses with different people and in different situations. This is really no different (in theory) than code switching and code mixing. Speakers of all languages always need to alter each of their languages in order to fit the situation, which is exactly what Grosjean is trying to define here as part of bilingual and multilingual language behavior. In the same vein it is extremely important and useful that we come up with models of language and linguistic behavior which accurately describe all possible language use situations. Therefore, we need theories which accurately describe monolingualism, bilingualism, and multilingualism in actual use.

       Right now, the Chomskyan model, based on the superiority of the monolingual speaker simply does not do this. In order to describe second-language acquisition from a Chomskyan perspective, we need to come up with all different models for how people learn. In short, the current model of UG does not explain second-language acquisition or use at all adequately. Nor can it explain the idea of language mode. Again, this is because the UG model is reliant on the idea of modularity, or separateness of cognitive systems. This is the idea that all different elements of language are separate from each other. For Grosjean's concept of language mode to be psychologically real, as we think it is, then the different linguistic systems a speaker has need to be interconnected in a profound way, and would need to be part of one co-dependent system. Let us now take a moment to try to model, in a very simple way, of what the system might be like.

       The first thing we need to think of is a system where everything is somehow interconnected and where connections, of all sorts, take place. Because lexical (phonological, morphological, syntactic as well as semantic) systems tend to be different from language to language and blending is often not tolerated, we need to look for a part of the linguistic system which would be underlyingly similar for all different codes and the obvious answer to that is the conceptual system. The conceptual system is the system of underlying meanings which while not entirely identical from language code to language code, as we discussed in relation to Hamers and Blanc (2000), there are very important points of intersection. Based on this we, therefore, claimed that systems integrate through the lexicon (albeit at the underlying conceptual level) which is the storehouse of all verbal information, which in many newer theories of language would necessarily include both phonological and syntactic information in addition to the traditional view of semantic information. It should also be noted that we would have two basic different types of strains, one related to language production, which is based on semantic-functional links, and one related to language reception, which is based on phonological and or orthographic links.

       Remember that the lexicon is not some simple little list. The basic function of the lexicon is as an organizer. Therefore lexical entries each form their own neural network of interconnected information, which is based on the semantic/functional nature of the item and this would include cross linguistic correspondences. Thus, the semantic network for a concept like `late` is going to include a multitude of different information across all different codes that the speaker has. Remember as well that neural networks are not self-contained units. They are deeply connected with and blend into other neural networks (including ones encompassing different linguistic codes) forming one gigantic mass of interconnected information, which is constantly changing. Thus, as we go to use language we're going to choose the code we want to use and selectively activate the connections within the elements of that specific code while suppressing connections to other connected codes. In order to do this we not only need to activate certain elements within a specific code, but we also need to suppress a large amount of unimportant or erroneous information which is both in the context of the contextual situation but is also in the connected neural network systems. If we are using (or trying to use) a language code as a base which has lower general activation levels than some of our other codes then this is going to be a very taxing process indeed as the speaker will need not only to activate the somewhat dormant code but also will be required to suppress the more dominant codes. This fits in with the model of DMM, where language maintenance can be seen as a mechanism for strengthening links between elements in a system and raising passive activation levels, which actually go in tandem. Based on this we can begin to get a fix on how Grosjean's concept of language mode actually fits in with the DMM system.

Herdina and Jessner, Chapter 6 - Answers

1. What are some of the characteristics of a dynamic multilingual system?

The main idea of DMM can be affected all the micro and macro level of influence and according to the different amount of influence DMM continues to work on adapting new organization in time. There are some of the characteristics of a dynamic multilingual system.

First of all, its non linearity. Language development shows the similar pattern of growth as a sine curve which means at the beginning of the curve starts the acceleration of its rate and growing fast and rounding and then slowing down to keep equilibrium.
Second, as system gets the energy to keep equilibrium as maintenance, the system could show retarding curve as gradual language loss because of not enough time and refreshing the knowledge of an L2 or L3.

Third, linking to the second, DMM is in need to keep stability as maintenance which need continuous enough time and effort. Therefore, if it is out of intention, it can be instability status gradually. In addition, there are some variables such as the number of language learner dealing with, age for language learning, relative stability established, the level of proficiency at which this take place and the time span over which the language system is maintained.

Fourth, interdependence is DMM`s most important characteristics for even SLA, FLA because it could be the evidence of psycholinguistic model language systems not as dependant system. Language system can be referred to the behavior of previous and subsequent systems that can`t be analysis as isolated component or system. As the evidence of it, there is linguistic mixing such as, code mixing, code switching, borrowing, and loan blending from lexical level to over sentence level.

Fifth, DMM can`t be simple because every single transfer or interference is able to lead to divergent results in different systems. It suggests that both transfer or interference is working negatively as decaying development process.

The final one is that DMM has a change in quality as qualitative changes in the speaker`s psycholinguistic system adapt to meet a new psychological and social requirement that also changes its nature.

2. What are two main key functions of DMM?

The two key factors of DMM is gradual language loss and language maintenance. In muiltilingualism, we often see the phenomenon of language loss but the complete loss of language system in a short time is empirically unlikely. Rather the standard form of loss must be ‘language attrition,’ which in a systems model is consistently called gradual loss. Language attrition is a gradual and much less spectacular process than abrupt complete language loss. If noticeable, the speaker will generally try to counteract the effect of language attrition by increased compensatory strategies. Language attrition is not observable because, at least at an early stage, it expresses itself in the form of an increased scatter of performance. According to DMM, easily perceived forms of language loss will follow a phase of declining language use which appears to have no effect on a language system. We can therefore expect the system to attempt to absorb the effect of decreasing language use by internal adjustments which are not perceived by the outward observer. Also, language loss will affect different linguistic subsystems to an unequal degree. DMM also assumes that the speaker can compensate for a shrinking language system by increased monitoring effort in the course of speech production and interpretation. It is also important to note that measures of language loss are necessarily unreliable as it is extremely difficult to measure language use – that is, exposure is as difficult to trace as to avoid. If we assume the existence of crosslinguistic interaction, lack of exposure to first language does not necessarily mean that components of this language system are not activated. Thus we may also assume that one reason why absolute language loss is generally not observed in native speakers is because the tendency towards a state of linguistic entropy is counteracted by the process of language maintenance.

       The other factor, language maintenance is meaningful only in relation to language loss. Use of language counteracts language loss or decay. The model of language maintenance effort, which can be seen as an essential part of the systems model, is composed of language use factor, that is, activation of parts of the linguistic system for communicative purposes resulting in a renewal of parts of the subsystem, and linguistic hypothesis verification or corroboration factor, that is the renewal of parts of the speaker`s explicit knowledge of a linguistic subsystem by means of a verification of hypotheses concerning the language system. Language learning thus requires an increase in language work and it is self-evident that the amount of language maintenance required to guarantee homeostasis within a linguistic system will increase with the accumulation of linguistic knowledge. The learning consequences must be seen as a result of interaction between the individual system and the environment. An idealized abstract language learning process taking the form of exponential growth, however, in reality, a certain degree of underachievement is to be expected, influenced by personal limiting factors. Furthermore, it is important to realize that the process of language learning and gradual language loss do not take place in an abstract psycholinguistic vacuum but in a concrete sociolinguistic and economic environment, which forces the individual to comply with the principle of economy of effort or least effort. The frequently observed and much researched phenomenon of switching codes thus contains a new interpretation, as it can be taken as a possible indication of the erosion of both or one of the competing language systems.

       In conclusion, neither language acquisition nor language attrition can adequately be understood as language learning processes in isolation as is still commonly done in the literature. Also, loss should be regarded as an integral part of language acquisition studies as a whole. As will be analyzed in more detail in the next chapter, they have to be seen as an integrated part of an evolving dynamic system where language attrition is defined as a function of language acquisition, with language maintenance providing the necessary link between the two processes and thus making clear that the two processes depend to a much higher degree on each other than has been suggested so far.