Stephen van Vlack

Sookmyung Women`s University

Graduate School of TESOL

Second Language Learning Theories

Spring 2007

Week 2 - Answers

M&M, Chapter 1: Second language learning: key concepts and issues

1. What components are necessary for a good theory of SLA?

As we mentioned last week Second language learning theories are based on general theories of language in addition to generalized theories of learning and possibly more specific theories of language learning. In any case an SLA theory needs a well investigated and explained theory to support it, but often teachers employing specific SLA theories are not aware of or do not share the view of language used to design the theory. Even more tricky is the issue of learning. SLA theories tend to be strong in the area of generalized learning (such as behaviorist models) or in the area of language (Chomsky) and often it is hard to mix the two (Lydia White, for example, is one researcher desperately trying to develop language learning theories that correlate to generative linguistic views). The important point to remember is that theories of SLA are reliant on generalized theories regarding the nature of both language and learning in the same way that teaching methods and classroom practices are reliant on theories of SLA.

Thus, we can take the issue of second language teaching and lay it out on a continuum with models of language and learning on one side and teaching methods and practices on the other and in the middle are theories of SLA. It is essential for us as teachers and researchers as well to have a clear view of this chain of relationships. As should be evident to all of you by now as a result of both your lives and careers, nothing come from nowhere. All ideas and practices have an origin and there is no such things as a completely new or original idea. Our job as reflective teachers and/or researchers is to accept this sometimes meandering chain of development and embrace it, for it is only with this knowledge that we can make additions to our own knowledge and develop new and wonderful ideas to help our students and ourselves.

2. How has language been viewed over the last fifty years or so and how have these affected SLA theories?

The basic view of the last fifty years which has had a great effect on SLA has been the generative model of Noam Chomsky. This view while radical in some ways is also very traditional in others and specifically in the units of language it sees as being basic as well as the different areas of language study and the extent of their relationship to each other. This view (and it is archetype traditional dating back several thousand years to early Greek grammarians) can be described as componential (highly modular) in that it sees language as consisting of variant components which do not share features and as such behave differently. Thus, from an investigative point of view these different components also need to be treated differently. In this respect, generative linguistics under Chomsky simply continued to redefine the componential approach continuously adhered to in grammar and lauded in structural linguistics 100 years ago although the number of components has been reduced. Thus, generativists have taken the traditional components of language (phonemes, morphemes, words, phrases, clauses, sentences, discourse) and their respective linguistic systems and not coincidentally areas of study (phonetics-phonology, morphology, syntax, discourse analysis, semantics, pragmatics) and reduced them to two systems (a computational system consisting basically of syntax (principles and parameters), LF (semantics and maybe some pragmatics) and PF (phonetics, phonology and morphology)) and a lexicon.

While it is not necessarily of great importance to be aware of the most recent developments in the generative model (most language teachers do just great without knowing all this), it is however, of great importance to be aware of the general effects of a componential view of language as regards learning and specifically language learning. Because each of these components (regardless of the number) functions independently they each need their own system and rules. If we look at this strictly and take it seriously, then it is very hard to effectively attach general learning principles which can account for language acquisition in the first place and even less for second language acquisition, where a functioning linguistic system is already there due to the presence of such a conglomeration of variant systems. Additionally, if a specific component functions according to its own rules and systems then it is further seen as operating independent of outside effects which includes any kind of generalized learning process. Thus, based on the componential view of language, each system needs to be learned independently and, therefore, can be and maybe should be dealt with independently and general learning principles as encountered in the environment will have little or no effect. This is a strict interpretation of such a model. It is, more or less, where we are today in mainstream SLA theories.

3. What are some of the aspects of the learning process which affect SLA?

Mitchell and Myles (2004) mention some important aspects which are claimed to affect the language learning process. Interestingly, many of the ones they mention are related to the generative approach, thus underscoring the predominance of such approaches in SLA theories.

Innateness

Innateness is an interesting theory and again is closely related to all componential approaches to language and the generative model in particular. In componential approaches because each component operates according to its own self-contained and idiosyncratic system they each need to have special rules which cannot be learned through generalized learning procedures and as a result of environmental exposures. Generalized learning theories do not./cannot work, yet people do learn. An explanation had to be found for language learning that was outside of accepted learning theories, and the theory of innateness was developed as a kind of faith theory of learning. In a strictly componential model of language, the only possible explanation for learning is to posit innate structures which aid (and to various degrees in different models) the acquisition process. Innateness is a necessity in componential models.

According to Chomsky (1995), there are two sides to language; the core and the periphery. The core contains all the universal properties of languages. The periphery is built up on the core and contains all the language specific properties and units (words) of the language the speaker has. It is the core which is hard-wired into the brain (is innate). While the core makes up a relatively small part of the total language mass, it is the core which allows the periphery to develop and work. About 90 percent of language is the periphery, but the periphery cannot exist without the core, in that it is the core which ensures the proper use of the periphery in this model. Thus, even language specific elements in the periphery are made possible by the innate structures of the core.

Now the big problem is that for SLA it seems that proposed innate mechanisms may not be work the same in all learners. Since the process of SLA seems to bring different outcomes for different people even under similar conditions, researchers have often wondered what happened to such innate structures. If the innate structures were still working properly, as they are claimed to do invariably in first language acquisition (FirLA, not only would SLA occur in the same fashion as FirLA, but we would not expect to find so many differences between learners. In a model, generally the more powerful the innate structures the less important environmental effects will be. This is something to consider.

Modularity

Modularity is a simple idea. It states that the brain is made up of different components or sections which are all specialized for a certain function. This means that variant kinds of information is housed and processed in each of the different `modules` of the brain. To use the terms of Fodor (1984) the brain is composed of modules which are all `informationally encapsulated.` While different modules are all connected in some way, they operate separately. Thus, for example, certain cases have been claimed to show that language skill and other types of intelligence are indeed separate in that a person can suffer from a general cognitive impairment, but show strong language skills as in those with William`s Syndrome. Likewise, language can be impaired without the rest of the brain being so as in the famous case of the epileptic Swiss monk, brother Gregory. The modularity issue is very important in linguistics in that the whole idea of UG and language specific mechanisms in the brain (componential models) is only possible if there is a large degree of modularity, at least that has been the mainstream idea for a long time. For us modularity supports the idea that we learn and need to teach language separately and differently from other elements/modules in the brain.

Systematicity

The idea here is that language is a system unlike other systems (a separate module) and that even within language there are different systems/components (more separate modules) which need to be learned. In learning language one does not kneed to learn not only the language but also the system which runs the language; i.e., the rules and often for some reason it is the system which takes precedence over the language itself in many models of teaching, particularly here in Korea.. In this view, knowing the units of a language (the words, for argument`s sake) is useless without knowing the system which dictates how such units are used in the language. Modular systems are also serial systems both in use and, by extension, in learning. Thus, language learning is primarily seen as the development of systematic knowledge needed to make the system work. So, the system must be learned before the language can be. Using this logic, language teachers need to teach the system previously to the language, the units or components of the system.

Of course comparing English language teachers in English speaking countries and our situation as English teachers in a non-English-speaking country, we need to look carefully at the units of language. In the UK, for example, the forms of the target language (TL) are everywhere and are accessed and used every day constantly. Teachers and researchers in such environments, often without ever really thinking about it, do not recognize the importance of these units because their students/subjects already have them. They have gotten them outside te classroom without teaching and teachers and researchers can be fooled into thinking that such things are simply not important. For them the main focus is fine-tuning the knowledge their students have so that they can more effectively use language. Units are delayed until conscious knowledge of some systems (written ones) are built. Once the systems are built then we can focus on the units (hard ones). We in the EFL universe, however, cannot wait to develop units, but that is exactly what we do. Back in 1984 Paul Meara wrote a now seminal paper lamenting the virtually non-existent role of vocabulary in language research and teaching. While much research on vocabulary is now taking place it still holds a minor place compared to structural systems in language teaching. The serial model prevails in which we need o get our students up to level on one component before focussing on another. Sadly, this whole systems-model approach has been complicated by the issue of pidgins and creoles.

Parallels have been drawn between the process of acquiring the system of a second language and the development of a pidgin into creole. A pidgin is a system of communication created by people of diverse native languages, but no common language, who have to work together for simple exchanges of information. It is a highly simplified system of basic communication with little grammar. When two pidgin speakers have children these children often only have the pidgin as their input, but they create a creole (a complete language system) out of the pidgin. This process is often claimed as being not unlike L2 development in that they both are the result of greatly diminished input and they both seems to go through stages of development in which a fully functional grammatical system is built where there was none.

Usage and creativity

The issue here is that learners need to be able to prove what they have in their head (their competence) through their actual use of the TL. Different SLA models attach different levels of importance to usage and creativity. The usage based-models, as the name would certainly suggest, attach a central role to this and claim that the usage issue works the same for both FirLA and SLA while other models see usage primarily as just a means of getting a limited view of the system. For Chomsky, for example, use is uninteresting and unimportant because it is unsystematic and incapable of being studied effectively. His suggestion is, therefore, to ignore use issues. As foreign language teachers we might be justifiably baffled with such a claim.

Failure

The idea that L2 learning is doomed to failure (and it bears repeating that this is predominant belief in SLA. In fact it is one highly acceptable way of defining SLA in relation to FirLA.) is based on the fact that there are too many factors (sticky use-related problems) getting in the way of acquisition. It has been pointed out time and time again by almost everyone involved in the SLA process and tradition that few people ever acquire their L2 to the level of their L1. If the goal of second language learning is to create a complete system exactly like an L1 then L2 learning is, in fact, doomed to failure from the very outset because this is not how language (or the brain as the basis of language) works. As language researchers we need to figure out why this happens, but also appropriately reset our goals as language teachers and learners.

Transfer

Transfer is an important idea in second language acquisition because it claims that learners can make use of or reapply the knowledge they already have in one area to another. This is a generalized aspect of learning. In fact all learning is transfer of one kind or another, to the extent that without transfer there is NO learning of any kind at all. Researchers in language often focus predominantly or exclusively on the negative side of transfer possibly because of the prevalent SLA doom scenario, but it should be acknowledged that positive transfer is a potentially very powerful force in SLA. It is the key to all learning, language learning included, the predominant componential model notwithstanding.

Cross-linguistic Influence

There has been a tremendous amount of debate on this issue and we can put the accepted views into a hierarchy ranging from the strong belief that L2 is totally dominated by the L1 to the belief that there is no connection what-so-ever between the different langayegs person might have. Most modern theories of L2 acquisition have downplayed the former prominent role of the L1 in the acquisition process. We, as people dealing primarily with the acquisition of a subsequent language (English) generally felt to be very different from the L1 (Korean), might have a different take on the issue. The distinction can be made in the core and the periphery, to use some terms from Chomsky. In the core, all languages are equidistant: that is, there are no languages more or less similar from each other in the core, for the core comes from an innate UG system. In the periphery, however, there are languages that do or don`t share more elements. We need to somehow figure out if L2 acquisition is core or periphery driven, or maybe neither. The might not even be a core a periphery. The distinction might simply be false.

Also, there is the issue of perceived language distance. Some researchers would argue that it is not the actual distance between languages, but the perceived distance between the languages that is important in the subsequent language learning process. Following this idea, Korean and English are no more or less similar than Korean and Japanese, but Koreans feel that Japanese is more similar and English less so and thus it is harder to learn English because the mind set of the Koreans makes it so.

4. What are some if the main learner variables that effect the process of SLA?

The important thing to remember about a lot of these different learner variables is that all of them are affected by and ultimately affect the working of the brain. This is important because any learning is going to have to occur under the confines of neuronal architecture. Here, we are also going to take the modern view that everybody's neuronal architecture is different based on experience. Thus, we see these variables as not only of rising from experience but also affecting experience and creating particular limitations in experience, all of which have a huge effect on learning and obviously output.

Affective variables

These are variables related to the emotional situation of the learner. They basically relate to their feelings about the target language and institutions related to that language. Traditionally affective variables were linked to the idea of motivation, but recently in current brain research motivation, itself has undergone a kind of transformation into a cognitive system all it own. For many researchers consciousness and motivation are seen as being part of the same macro-function (system) and it is this system of consciousness/motivation which regulates and drives all other cognitive systems. Thus it is not possible to underestimate the importance of such affective variables.

Cognitive variables

Cognitive variables relate to the learner`s brain as is determined by their past experience both with the target language and with the world in general. Thus variables related to cognitive styles and strategies are derived form the learner`s experiences of success and/or failure in all the different endeavors they have undertaken. All this experience is encoded in the brain`s structure and determines what the brain likes to do (does well) and how it perceives things. It must be understood that in this view, proposed above, the brain is interpreted as a fluid (not fixed) organism. This means that the brain, or really the connections within the brain (which define the brain), are in a constant state of flux. They're constantly changing, from second to second, based on information which is coming in both from the outside world and also from the systems already present inside the brain. It is in this way, I propose, that cognitive variables arise and come to dominate the entire process of learning, in a somewhat systematic way.

Social variables

These are variables determined by the social environments in which the learner lives or feels a part of. These exist at different levels such as the national, regional local, family, social class, etc. Each of them plays an important part of determining both the affective as a well as the cognitive variables of a learner.

Johnson, Chapter 1: Three Major Scientific Research Traditions

5. Of the three major approaches to SLA which one is dominant and why?

The cognitive/computational tradition in SLA has been the dominant one for the last 50 years (basically the entire history of SLA as a separate field of study) for a few different reasons. The first of these clearly shows that cognitive models rose in response to the failure of behaviorism. The demise of behaviorism came about certainly partially as a result of the new cognitive theories, but behaviorism died more due its own stunning failure to explain complex human behavior, and particularly language. This is particularly interesting when we think that both the behaviorist and the cognitive approaches prescribe to the same Cartesian philosophy which claims that there is a fundamental brain/body split. The difference between the two, however, is in their specific point of view. In the cognitive/computational approaches the brain is dominant over the body. For behaviorists, it is the experience of the body which affects the working of the brain and all the brain does is store and organize this information. Bodily reactions are based on previous experience and not on thought or cognitive processes.

There is one further point for why the cognitive/computational models might have become dominant and one which Johnson (2004) fails to point out. In the cognitive models the human brain is seen as being wholly unique. It is fundamentally different from the brains of other animals. This is an appealing theory for the creature proposing their own inherent dominance. In the behaviorist view there was no theoretical difference between human behavior and animal behavior. All sentient beings were seen as possessing the same basic learning mechanisms (humans could process more obviously) and the same systems of cognition. In essence humans were bodily-driven animals. Not surprisingly, these views upset many researchers with strongly Christian values. Many of the basic theories of the cognitive model correlate exactly to Christian philosophy. In both views, the body is seen as being a problematic, often evil, entity which must be controlled and wi do this through will power directed by the brain. The brain, in this view, is special. The brain is what controls the body. It needs to because the body is bad. And the human brain is fundamentally different from all other brains. It needs to be because it needs to control our bodies, something animal brains don`t do well. The implication is that human brain was specially designed just for humans, and presumably to be able to control some of the problematic urges of the body. Again, these are the views that Christian philosophers have been expanding since the dark ages and provide a further reason for the predominance of the cognitive view.

6. What is it that behaviorist and cognitivist views of language research share in common?

Both of them rely on a Cartesian view to support their ideas but they examine this from opposite points of view. I think we went over this well enough above, So I will leave it at that.

7. How is the dialogical view different why does Johnson think it is better?

The dialogical view is based on Vygotsky`s belief that the social environment is extremely important in and necessary for cognitive development. The beauty of this view is that in rejecting the Cartesian view of the mind it embraces the idea that there is a fundamental connection between bodily experience and the brain/cognition in forming an overall comprehensive system. The two are connected in that while bodily experience is necessary for setting up an initial cognitive systems, it is the systems (with language being one of the most important) which at some point begin to control and regulate bodily experience. For the dialogical view bodily experience is different than just embodiment, it is achieved through interaction first of a generalized type like embodiment, and then through an increasingly sophisticated kind of interaction mediated by cognitive systems and controlled predominantly through language (language is thought in this model - a view flatly rejected by innatist, modularity-based philosophers).

An interesting aspect of the dialogical views revolves around their age. Vygotsky developed the basic format for the dialogical view about 80 years ago when there was practically nothing known about the actual functioning and architecture of the brain. Modern neurological studies, however, are providing evidence in direct support of this dialogical model. Modern neurologists (Damasio 1994, Ratey 2001, LeDoux 2002) soundly reject the Cartesian model and are finding that the body mined connection is an essential part of our cognition.

8. Which tradition do you personally think has the most potential and why?

In reality, I think we need to try to take bits and pieces from all of these three main approaches and fuse them together into one comprehensive type of theory. One possible problem which still exists in relation to the dialogical models is the fact that they are still highly philosophical. This is that they have not yet achieved a larger volume of actual physical evidence to support them. While such evidence would have been impossible in Vygotsky's day, it is certainly possible today. By using these three models together we can hopefully, with a neurological/cognitive support mechanism to provide evidence for the dialogical models. Researchers like Tomasello are working in a parallel tradition to the dialogical models called usage-based models which show tremendous promise in linking all three traditions and fusing them into one holistic theory both of language and generalized learning.

Singleton, D. (1995) Introduction: A critical look at the critical period hypothesis in second language acquisition research. In D. Singleton and Z. Lengyel (eds.) The age factor in second language acquisition. Clevedon: Multilingual Matters, pp. 1-29.

9. What can be derived about the very nature of subsequent language learning by taking a critical look at the critical period hypothesis (CPH)?

Well, what we basically see from all the studies mentioned by Singleton is that the second language acquisition process seems to be much more variant then one could possibly have expected. We get a fair amount of contradictory evidence and studies which disapprove each other. This means that it is very hard to come up with some sort of systematic view of the second language learning process. Interestingly all this seems to coincide with the general dialogical view, wherein there does not seem to be any one particular system which all learners necessarily follow. Such things bring into question the entire idea of Interlanguage is a specific type of linguistic system. In general this is a calling card telling us that there are no simple or easy ways for us as teachers and researchers to deal with the issue of second-language acquisition. It seems to be a rather personal process as opposed to one imposed on learners by particular innate aspects of our commonly-shared human brain structure. This is a very important point.

10. After reading all this how do you think English teachers in Korea can best think about the CPH?

As hard as it might seem, as annoying as it will certainly be, there's nothing we can do except to accept the reality that every learner is different. When something as seemingly simple as the Critical Period Hypothesis becomes a convoluted mess in the face of various studies we really begin to question what it is we have been doing. In research the most important thing is to try to come up with the right questions. The answers are only useful if the right questions are being asked. If a particular question does not produce satisfactory answers, then maybe we are asking the wrong types of questions. As should be clear from our entire discussion this week, any kind of questions people will pose come from their background ideas. Obviously the critical period hypothesis was developed in response to a Cartesian model of a brain and accompanying acquisition. Researchers desperately needed to try to find a way to explain why second-language learners fared differently both from each other and presumably from first language learners. Maybe it was just the wrong question from the wrong point of view and the most we can get out of this is we need to ask different questions, from a different point of view.