Stephen van Vlack
Sookmyung Women`s University
Graduate School of TESOL
Developing Bilingualism
Fall 2007
Week 15 - Answers
Forster, K and N. Jiang. (2001) The nature of the bilingual lexicon: Experiments with the masking paradigm. In J. Nicol (ed.), pp. 72-83.
1. How are priming effects used to investigate the nature of the bilingual lexicon and what are the results?
At this point it should be clear that priming is one of the most important and one of the most basic phenomenon which have been observed in the brain. The strength of priming as a research tool is that it seems to occur in all different types of brain structures and functions, which is quite telling in and of itself. In fact it seems to be a basic universal, corresponding to the basic design of the brain and how neurons and neural networks function. For this reason, priming has become one of the main techniques used in studies of brain-related functions and structures, such as language.
Priming, as you should all by now, is a phenomenon which affects connections between elements or nodes in the brain. In psycholinguistic studies, priming is often used as a specific (and psychologically real) variable to try to get a clear determination of what types of connections there might be between different information structures in the subject`s brain. If a prime is made (observed) on a particular item within a particular structure or network shows an effect on a target item in another network then it can be hypothesized that there must be a connection between these two items. From such observations, researchers can determine that this type of phenomenon would play a role in organizing the lexicon or any such array of networks in the brain. Thinking, now, about bilinguals are multilinguals it can then be seen how priming would be of the utmost importance in trying to determine not only the types of connections but also the strength, density or amount of connections between different nodes or even between linguistic codes (systems?). Researchers like Forster and Jiang try to use priming as a way of doing just that.
Forster and Jiang want to basically see if words from one language can prime words from another language for bilinguals. This is a very old and well-documented type pf priming experiment. If priming effects are clearly observed then obviously the languages are going to be more directly and strongly connected. If, on the other hand, priming is not observed then it means that the languages, or at least those words, are either not connected or not connected through that particular variable. One major concern which Forster specifically mentions is that of trying to reduce the overt control over the process of, in the case of his studies, lexical decision tasks. This means he doesn't want people to really be able to plan when they are involved in the experiment. He wants to try to get natural reactions to primes and not top down control in determining the language of the target item. Therefore, he developed what he calls a` masked` priming technique. While we are no position to critique the masked priming technique specifically, we can say is that clearly what Forster is doing in the laboratory is not at all similar to what people do in the real world. In his own studies and in his citations from other studies Forster finds that there does not seem to be priming from an L2 to and L1. This would indicate a seemingly high level of asymmetry in the lexicons of bilinguals. It would be interesting to see what the results might be on a study incorporating three or four languages. Forster does make some predictions of what he thinks might happen with more balanced bilinguals or multilinguals for that matter. Disappointingly, Forster and Jiang don't seem to make any kind of statement about what the actual target group is in their study. All we know is that they are Chinese speakers who have been living in the U.S. for a rather wide variety of years. There is no mention of a proficiency test or grouping or are even other language behavior while the subjects are in the U.S. thus we're really left wondering who these people are. This is a major flaw either in their experimental technique or in the way they report the experiment (It is possible and really almost unthinkable that they did not take extra-special care in choosing their subjects as this is a basic in experimental techniques). It should also be mentioned that priming between two languages seems to be bidirectional in different types of lexical items.
2. What role does episodic memory play in the retrieval and activation of lexical entries?
This is the big question. It is clear from both their studies (Forster and Jiang) and my own as well that episodic memory plays an extremely important role in lexical organization and acquisition, but I think Forster and Jiang miss the boat in their description. The point that Forster and Jiang are missing, I believe, is the episodic memory is indeed connected to semantic memory and lexical memory. They are sticking too closely to Tulving`s (1985) now classical distinctions between memory types. Remember that lexical memory is often thought of as a specific type of semantic memory. In fact, the results of the study are interesting because they support something which I personally believe which is basically that all memories must somehow start off as episodic memories.
Episodic memory is primary. This certainly makes sense particularly when we think about how people deal with the world when they are young, in a pre-literate state. Everything a child learns occurs through direct contact with the world. That is, every learning experience comes as a result of a particular episode in someone`s life. Obviously, all memories are in effect initially episodic memories. Even the episodes we experience indirectly through language (people telling us stories of their own or other`s experiences) are probably first coded as episodic memory. Certainly procedural memory is also initially episodic. But memories change over time. As the same or similar elements occur packaged in different events the episodic element to the memory becomes less salient than some of the component parts, especially when they are found recurring in other episodic memories. The more experience a child has with the world the more they are able to prune unimportant elements away (and recurrence is an important aspect of this) revealing a kind of prototypical inner core to the components of the memory, which not surprisingly, becomes semantic memory. Now, if these memories happen to deal with language then they also go through a further stage of moving from basic semantic memory to lexical memory. Each system, episodic----> semantic----------> lexical, is a little more specific and specialized than the other, not because of some weird modular structure in the brain but by virtue of the actual elements or components which make up the system. A lexical system is lexical because it houses only lexical information in its core. Again it is extremely important to remember that the lexical system also has its base in an episodic system and connections still attach both in a profound way across millions of connections.
The lack of L2-->L1 priming effects which Forster and Jiang report may not be due to the episodic, lexical differences alone but may also have a lot to do with the actual task that they had students perform in their experiment. We already mentioned in the answer to the above question that lexical decision tasks are not something that people in the real world would ever really have to do, especially when reading. Taking all these things into account (and thinking about the activation model proposed by Grosjean (2001)) we can not only critique what Forster and Jiang have done but can also find some very useful information in the paper which helps us further understand the process of lexicalization, which we can use to help better teach language to our students.
Muysken, Pieter. (1995) Code-switching and grammatical theory. In L. Milroy and P. Muysken (eds.), pp. 177-198.
3. What are some of the structural limitations on code switching?
There is an old adage, `seek and ye shall find` which correlates nicely to Muyskens article. The issue of code-switching and code-mixing has risen like a tidal wave. It started off very slowly as a weird or especially a negative phenomenon of bilingual or language learner behavior which was thought to be mostly lexical and based on lexical deficits. As researchers found out more about this phenomenon it became clear that code-switching and code-mixing is not merely a result of lexical deficits (although such deficit-based phenomenon is a real part of learner language use). It was also noted that this phenomenon posed a real challenge to the well-developed and widely accepted model of Chomskian generative linguistics. Chomskian generative linguistics (to make the matter as simple as possible) sees language use as occurring within a single system. The nature of the system is determined by lexical properties which spell out the structures that a word can engage in. In addition, there are parameters which are language specific concerns like simple head ordering which determine the basic structural formations a language system will have. Code-switching and code-mixing is a major problem for this kind of model because there are bound to be conflicts between the lexical restrictions and spell outs coming along with the words from a different system that are used within the overall language matrix structure set up by the parameter settings of the base language. In a strict adaptation of the Chomskian model it would seems that code-switching and, particularly, code-mixing would be impossible. For this reason, code-switching and code-mixing phenomenon have become a veritable battleground issue for researchers with differing theoretical affiliations. Muysken is in search of a way to explain such phenomenon within a Chomskian framework. He is seeking - but does he find? That is the real question.
To try to find solution to this problem Muysken investigates carefully where code-switching and code-mixing occurs. In the paper he looks exhaustively at different aspects and constraints of both code-switching/mixing and grammar. In the end, it seems that code-switching is really a phenomenon dependent on both grammatical and lexical aspects of the units being switched and the place they are being switched.
4. Does evidence from code-switching give us evidense to support the primacy of grammatical or lexical phenomena in language?
Neither side is definitively supported here. Although I would tend to interpret the data as supporting a lexical position, it is clear that a grammatical position could be equally well supported based on the data presented. This is the nature of language and language study. As we keep on saying, one of the most important things you can learn as a graduate student is realize that there is no one clear or definitive answer to questions, especially questions about language. This is where critical thinking really comes in. We develop our skills as language researchers, even teachers of language, by being confronted with unanswerable questions. The basic realization is that there are very few absolute truths in the world. All there really is are beliefs which can be momentarily substantiated to varying degrees as new research comes in. As a potential researcher or teacher, you need to find your position on the questions (After having found the questions which is often a feat in and of itself) and then conduct research to find support for your particular belief. This is how it is done, but make sure that you don`t somehow believe that you are proving anything beyond a shadow of a doubt in doing your research. Researchers compile evidence, but do not establish truths. As new evidence comes out the `truth` will often shift.
I like this question in particular because it really does show us how this type of research works. Muysken goes into this fully expecting to find mounting evidence for more structural bases to code-switching, but in fairness to him he is fair and that he freely acknowledges lexical affects even though he was not initially looking for them. As is often the case, Muysken`s original hypothesis is not supported by what he actually finds in his investigation. While he is not sure what to make exactly of all this he does handle the situation by at least mentioning it neutrally.
Herdina and Jessner (2002), Chapters 8 and 9.
5. What are the main repercussions of a modular view of language?
One of the main repercussions of modularity is that studies of second language acquisition and certainly bilinguality have been kept out of the field of linguistics. The Chomsky model of language, what has come to be known as Generative Linguistics is very much reliant, or even centered on, on a modular view of language. Chomsky uses modularity as a way of justifying the distinction between competence and performance which is of the utmost importance in his model. Based on this fairly strong view of the modularity hypothesis, language is a totally separate entity in its basic nature from other cognitive concerns, such as cognition, knowledge, or culture. What this means for us as budding bilingual researchers is that according to this modular view, both language and languages themselves should also be separate inside a person`s brain. So a bilingual person has two separate linguistic systems which do not affect each other. They come from different `computational systems`. This is perhaps the biggest repercussion for us as it relates directly to this course on bilingualism.
What is interesting about this whole thing is that we can really see how one could get away with not looking at the successes of bilingualism over the last hundred years. Because of a strict adherence to a rather strong version of the modularity hypothesis, bilingual studies have never really become an integral part of the field of linguistics and not even, shockingly, of SLA. Second language acquisition research have always taken the point of view that a second language is built up virtually anew, rather than through transfer as a foundation for new language learning. In this modular view, for every language that we learn we need to start from scratch over again with all new materials and all the tools and the two entities are not going to affect each other.
Let`s take a familiar example, we can think of the Sookdae campus as a hypothetical model of the systems within the brain. Each building on the campus represents a language or a specific cognitive system. As new buildings (or languages/systems) are constructed they are (may be) done so with very little consideration of the other buildings already present. From one point of view, it may seem that each building is a separate self-contained organism. Now, having said that, from a different point of view, the people inside each of the buildings (those who actually use the buildings) are profoundly affected by subsequent buildings that go up. Certainly, we in the TESOL MA were greatly affected by the destruction of the old graduate building and the construction of the new Injaegwan, as were were by our departure from it, but our concerns/happiness/feelings had nothing to do with either the destruction of the old graduate building or the new construction of the Injaegwan. These buildings were built, and even designed, separately from the people that would be in them. This is really how modularity would work in the brain as seen by the Chomsky model. It is the framework, the actual building, which is the most important thing. It is hardwired and each one is built irrespective of any other.
Now, in this class I hope that you have somehow formed a judgment which tells you intuitively that this modular model does not work very well. It certainly does not work for bilingual and multilingual people (the vast majority to people in the world) and as a result probably does not really work for monolinguals either. The different systems of knowledge a person possesses do affect each other. And certainly different languages affect each other, and to varying degrees depending on the different learning as well as usage situations. Modularity is an oversimplification which simply cannot capture the essence of the interconnectivity of the brain. The repercussion of all this is that a huge amount of research simply goes unnoticed or is never undertaken because the wrong questions are asked. What a shame.
6. How does the holistic view of language better deal with multilingual phenomena?
The holistic view of not only language but the brain in general deals better with both multilingual and general cognitive phenomena simply because it is more flexible. One thing we should all be aware of at this point is that different people learn differently, even in the same situations. Because of this simple truism we need more flexible models. Any model of language, as well as cognition, needs to be able to account for all different types of language learning and use phenomena. Modularity just doesn`t cut it. At the same time, however, one must realize that some systems are much more integrated than others. The beauty of the holistic view is that there is no strict definition of where and when and how different cognitive systems (linguistic or otherwise) are going to interface. The main repercussion of this is that researchers are going to need to develop new experimental techniques for testing hypotheses related to integrated approaches. As much as generative linguists and SLA researchers belittle the behaviorists it is interesting to note that they still generally follow behaviorist techniques in the way that they investigate language and other cognitive phenomena. Experimental techniques have not varied dramatically for a hundred years. If this holistic view of language is ever to become a mainstream view than experimental techniques will have to change quite drastically. We are going to have to move from much more quantitative too much more qualitative types of analyses and research paradigms.