Stephen van Vlack
Sookmyung Women`s University
Graduate School of TESOL
Human Learning and Cognition
Spring 2006
Week 7 Questions - Terry, Chapter 6 & Lamb, Chapters 10 and 11
1. What is verbal learning and what was Ebbinhaus trying to prove with his experiments?
Verbal learning has to do with memorizing or learning lists of words. Basically this is done as a method of studying how the memory really works. We need to be careful how we assess this because a verbal learning task is often very different then how people actually learn language in the real world. Unfortunately, verbal learning is very similar, in many ways, to how people learn language in the classroom. Think about how many times you giving your students lists of vocabulary items which they didn`t really know very much about or, what`s more, didn't care about. Ebbinghaus was obviously aware that his verbal learning was not really about language learning (this is evidenced in his use of meaningless syllables as well as his use of more meaningful syllables). That was not his chief concern, or really his concern at all. What he was really concerned about was simply trying to identify some elements of memory as it relates to learning. Chief among his concerns was how information is housed in the brain so that it can be recalled with efficiency. This really is the spirit in which we want to evaluate these early types of experiments. They are highly controlled and not very authentic which means that we will need to try to work out slightly different ways in order to test similar things. If we are to do our own experiments following this type of model than we are going to have to try to inject a little more meaning into what we are doing. At the same time, however, we need to nature that we do control our experiments as much as possible.
2. What is serial learning and what are some of its variables?
Serial learning is basically learning a list of items according to their sequence. Again, this is basically what Ebbinhaus was doing with his experiments. He was primarily concerned with the idea of listing. This might seem like something that's not actually very important in relation to the real world. After all, how often are we really exposed to lists? Upon further consideration, however, we find that serial learning is very much a part of our daily lives and that lists are extremely prevalent in our lives. In fact they might be so prevalent that we don`t even notice them most of the time. Almost everything we do occurs in a particular sequence. This extends from how we move through physical space to how we think about things and go about solving problems. In Ratey (2001) we already read about how movement systems and how movement relates to cognition, or really that cognition is a metaphorical extension of movement. Well, movement is really only possible when built on a basis of serial learning. We not only need to sequence movements and actions, but we also need to understand the consequences of each particular movement in relation to each other and to the whole. This is pretty complex stuff and it makes serial learning, which might seem extremely boring at first, actually very interesting and relevant when seen from the point of view.
There are several things about serial learning which we need to focus our attention on, and which do actually relate to a large way to language learning, for language itself is prey to sequencing. The whole idea of syntax relates to nothing but how linguistic units (lexemes, maybe) are sequenced. One thing we see is that the position of an item within the list has a very profound effect on how well it is going to be learned and recalled. Not surprisingly, it is items at the beginning and the end of a list which are more easily recalled. I indicated that this was not surprising because we find the same kind of effects in studies of lexical recall and presumed storage, wherein lexical items are accessed primarily based on beginning sequences and final sequences while the middle sequences are actually harder to access, and actually are often not required. Researchers have drawn up an inverted U-shaped curve to show these kind of effects. Aitchison (1987), in writing about accessing units in the mental lexicon, refers to this as `the bathtub effect`. If we think about sentence structure now we will see how this relates to the actual meaning in sentences. It may not at all be a coincidence that subjects, or agents have a strong tendency to come first in most languages. Objects, which are arguably less important, appear in the middle of the sentence in most languages. Also, following data presented in Nichols (1992) verb last languages are the most common type of language in the world, thus supporting this idea as having some sort of effective in language.
Another interesting effect observed in serial learning is the idea of anchors. Anything that is distinctive or doesn't seem to fit in with the rest can often be seen as an anchor. An anchor can also be an overall type of context, such as a fixed space or time. Items in the list are associated with this type of anchor. Because, as has been observed above, middle items in a list do not seem to be linked to the context itself but to each other, it is actually quite efficient to try to create an anchor right in the middle of the list which will allow these middle items to associate with some sort of contextual clue. Anchors, while often facilitative in the learning process, can also show inhibiting or interference factors. If the anchor is too strong it might actually inhibit some of the items in the list from being remembered. For basic anchors researchers describe two types of interference proactive and retroactive. Both of these types of interference prey on middle items in the list, thus explaining why people possibly have more difficulty remembering middle items.
The last aspect we need to go over in relation to serial learning is the idea of associations. These are the associations between the connective points. There are basically two different opinions here. The initial assumption is that associations progress on an item to item basis. Following this the use of one item would trigger the memory of another item, which would trigger the next item, and the next, and the next until the entire list was complete. Critics of this have argued that it simply is not fast enough and that it also ignores some observations have been made in the real world. Alternative theories are the remote association theory and also the group association theory. In these two, somewhat related theories, the idea of associations is taken to be a little more complex. In the former, associations can occur across distances. What this theory really involves in the idea of different weights of associations, with more closely related words physically having stronger associations, but other items being associated in a weaker way but still associated. The latter theory expands on the idea that items are associated as groups or, to use terminology from the field of lexical study, chunks. In reality things are probably associated using all these different kinds of links.
Following all these kinds of observations we can see that serial learning experiments can tell us not only about how the memory works, but more importantly the results seem to tie in with what we know about language structure as well.
3. What is paired-associate learning and how can it be analyzed?
The basic idea behind paired-associate learning is basely that which we saw in classical conditioning. Simply put, two items which happen to co-occur are associated. This is the simple idea behind paired associate learning. This type of learning is usually done in relation to lists of pairs. If we manage to extend our view of this out of the highly controlled situation of laboratories we can actually find out some specific variables in the process of paired associate learning. It is easy to see how this, again as we discussed earlier, is really part of our everyday life and quite specifically the very basis of language learning, both L1 and L2 (+) learning. The first idea to be tackled or controlled in paired associate learning is for the learner to be able to distinguish the two items that are being associated from other items in the world. This might seem like a simple thing, but even when we just think about language we find that it is very easy to get items confused if they are not properly differentiated. Such confusion is a common phenomenon in subsequent language learning and listening comprehension in large part because second language learners may not be not able to distinguish sounds in the target language and therefore are not able to distinguish lexical items. Thus, as teachers the first thing we need to do is to be able to make sure that students can actually distinguish the units that we're trying to get them to associate. We do this by accentuating certain distinctive features of the items. In the case of lexical items we would overemphasize certain phonemic code items. Following this, it would seem that features are very important in this process. Once we've learned to distinguish different items only then can we begin to actually learn them through association.
Different factors seem to affect how efficient stimulus response learning is going to be. Obviously, responses which can be linked immediately to many other things which have already been learned will be easier to learn. This is a simple idea of meaningful learning which no doubt all of you have heard about. Connections seem to matter. In a way, if your brain is ready to take in a particular item then it will do so much more easily. Teachers then need to make sure that students can see how new items can be fitted within their overall cognitive structure and framework. This of course relates in a very direct way to what we talked about in clutter theory. There has to be space were something can fit otherwise it won't, at least not very easily. Basically what were talking about here is what was already mentioned under the name of feature hopping. The more features we can hop from one new item to another or to pre-existing items in the brain the easier it will be to learn that new item. This is the same idea we find in cognitive elaboration. We forced the students to elaborate on many different aspects of the new item so that they can create new associations. In effect, we draw the students` attention to associations which already exist or features which can hop from one item to another.
What is also interesting here is order. The order which we set up the association seems to be important. This is something which is going to play out actually very differently in first language acquisition and subsequent language acquisition depending of course on how the subsequent language is learned, or more precisely how the input is handled. Terry (2006) points out that we seem to have one-way types of associations. This is definitely true when the talking about first language. This is basically what we call collocation. So, for example, baby will associated with boy, but boy will not associate for baby. This reflects what's actually encoded in the brain. It needs to be this way and it's perfectly fine when we're dealing with first language. Ultimately this same basic relationship also has to hold true for the second or subsequent languages that are learned because all languages are based on collocational links and stored chunks in the mind which are one-way associations. If, however, you as a teacher are practising translation from one code (language) to another then you need to think carefully about the direction of the translations you have the students engage in. Normally we have students translate from the target language into the first language, thus creating a very strong link in that direction and that direction only. It seems to be important to reverse the order in order to enhance learning, since students will not be thinking in English so much as thinking in Korean. We need to create more links the move from Korean to English and not from English to Korean. So start off by translating from Korean into English to try to form associations between two items that would seem to also help to try to have the students do the opposite and try to get them to also sometimes move from English into Korean is a way of practice in order to solidify this type of association. The only danger here, as mentioned above, is that such two-way links don't regularly exist in most language, or single language codes.
4. What are some of the variables that affect free recall?
There are basically three variables that seem to affect free recall types of experiments. At this point none of them should come as a huge surprise for you are we have already mentioned them in the other types of verbal learning experiments. They are:
Serial position
The issues underlying serial position should be fairly simple. Again we know that words of the beginning of lists and at the end of lists seem to be easier to remember. There to possible explanations for this one has to do with priming (beginning) and the other has to do with recency (end). The other explanation is one related to differences in short-term memory and long-term memory. As a teacher it is important to remember this when you are looking at reading texts as well as other types of input. Authentic input should probably follow the natural cognitive makeup of the writer because they are writing in a more natural way. Contrived input may very well violate these types of normal ordering principles, simply because the writer is engaged in so many other concerns which make the writing less natural. A contrived text may have very important information not in the beginning or the end but in the middle and maybe even without any kind of anchor to draw attention to it. Teachers should be aware of this.
Rehearsal
The idea behind rehearsal is simple. The more rehearsal the better and this seems to correlate with education level. Those who have been successful students have some how learned how to rehearse and to rehearse effectively. This is shown in the fact that distributed rehearsal works better and that more experienced students seem to know this. Rehearsal can also be correlated with memory and this correlation can be used to explain why first words are mostly effected by rehearsal. The most important thing we want to remember about rehearsal is that we want to over hers in different ways. Certainly repetition is an aspect of reversal, but simple repetition in the same form in the same context is not going to be extremely effective. Rehearsal using different context different modes is going to be more effective. Make it happen.
Organization
Organization seems to also have a strong effect on recall. Interestingly, there seems to be threshold effects on organization in the that the more words someone already has the easier it will be to organize and the therefore the easier it will be to learn new items using strategies like organization. Thus, for us as teachers we need to try to organize the new linguistic information we are presenting our students with. Maybe the simplest way of doing this is not be presenting them with meaningless unorganized lists but by giving them authentic texts which should have their own type of inner organization and meaningfulness. Teachers don`t organize either because they don`t know how to organize or they are simply lazy. Alphabetic order is not a meaningful way to organize things for the brain. We can organize things for logically, semantically, schematically, etc. The bottom line is that you need to organize because organization is your secret weapon as a teacher and it really is a large part of our job as teachers. If there is one thing that language teachers do it is at least to a talent to organize the language that the students are supposed to learn. Do it.
5. How do available and accessible memories differ and why do they do so?
Available memories simply include those things which have made it to the long-term memory store but you cannot talk about them or even really clearly remember them. This group includes all the things that you have ever learned, in theory anyway. Studies have shown that even though material may be available it may not be recalled or recognized. A relearning test may be required to see what is actually stored. These kind of relearning effects clearly show that available memory is not only just in the brain, but it has an effect on but we actually do. We make decisions based on this huge pool of available memories which really are the basis of our cognitive makeup, and we don`t even know about it.
Accessible memories are those things which we have learned (are stored in the long-term memory) and which may be called upon for use when necessary. Accessible memories are ones we generally have no trouble calling up and using. It is important to recognize it and teaching we pay absolutely no attention to available memories even though available memories are really the base of the iceberg whereas accessible memories are merely the tiny little tip of the same giant iceberg. We spend all our efforts (and a huge amount of our students` energy and patience as well) trying to create accessible memories which we can test specifically. In reality decibel memories are probably less important than available memory simply because they are such a small number in relation. The main difference between these two, as defined by what we think we know about how the brain works would be related to their activation levels.
All memories, or really parts of memories in the brain (OK, Let`s call them bits of information.): All bit of information in the brain are activated through a change in the chemical composition of the neurons holding the information. This is achieved through an electro-chemical process called firing. Neurons vary greatly in their degree of stability. Generally, the more stable or inert the set of neurons are, the harder it is to activate them: more energy is required for firing. Accessible memories in the brain represent bits of information with lower activation levels. Activation levels are set based on use, either direct or indirect. It is important to remember that memories, according to certain theories are not housed as solid blocks in the brain and they are housed on a large number of connected neurons, which are in turn connected to other neurons. If a memory (i) or parts of a memory (i) are connected to other memories (j-x) that are highly salient and therefore much used then all memories linked to (j-x) will be fired residually and will have lower activation levels as a result. Following this, we can see just how important connections really are.
6. How and why do mnemonics work?
What all the hoopla about mnemonics comes down to is connections. It may actually be a special type of learning which simply takes advantage of the way the brain is structured and functions. I call it learning because mnemonics is just about forming associations: that is learning associations between two unrelated concepts or a series of concepts. The big trick is the concept upon which the new information is anchored. For mnemonics to work the new information must be associated with old and highly used or meaningful information (very active neural networks). This means that the activation levels for the new, and invariably less useful information, will immediately start out as being low and will probably stay that way. Mnemonics make new and basically meaningless information immediately accessible by forming a series of connections with readily accessible information. The only big problem with mnemonics is that they are hard to come up with. This makes them somewhat less practical in the real world. It is not only difficult to find mnemonics that work well but they can also be harmful. Especially when we are dealing with language, which is so intertwined with culture and specific experiences and people, trying to use mnemonic devices such as keywords can end up having detrimental long-term effects. Linking the TL to the L1 as a single type of learning strategy is not really a very sound singular long-term strategy because it actually will weaken the possibilities for forming good connections within the L2 system.
7. What is lexicalization and how does it work? (L10)
Lexicalization is the process through which a lexeme comes to be stored in the brain. This is what learning is all about. It is the simple process by which new nections (neurons) are recruited to store and link information. When this happens related to lexemes then this is called a lexicalization process, or the process of coding new lexical items. It should be understood that lexicalization is not simply a process of sticking a word into a box. I mention it in this way because this is pretty much how people in the past thought about word learning and the mental lexicon. The mental lexicon was basically something that we can compare to a child`s toy box. It was a specific part of the brain which housed lexical information, most importantly the form of the word, and meanings. The process of lexicalization was linking the form to the meaning and then it got shoved into this box. We are going to take a much more modern, refined, and cognitively explicit view of lexicalization. At this point we should clearly understand that lexicalization is a process of laying down new associations. These associations move in two different directions, bottom-up and top-down. We associate from the top down meaning from the concept which we already have to the entire lexeme unit. We associate from the bottom of moving from articulatory gestures (maybe) to specific phonemes, to the lexeme unit. The lexeme then, in this process of lexicalization, is really at the center of all the information. This is where all the links lead to and extend out from.
Lamb (1999) spends an awful lot of time trying to convince us of lexicalization as a process which relates to all different sized type lexical units. Once we accept this simple model for a single word, such as dog (the animal), then it is not hard to use exactly the same system to extend this out to include multi-word units like a three dog night (meaning `a really cold night`). In order to make this jump (If we want to call it that. It`s really not a jump at all, but rather a normal part of the process of learning language) all we need to do is create more associations not just to the word dog from above and below but also within the level of lexis from dog to a to three and night, thus forming the idiomatic chunk a three dog night.
How the process of lexicalization works is really simple and no different than any other way learning occurs in the brain. Remember lexicalization is a type of learning, more specifically lexeme learning. Lexicalization is based on frequency and saliency. The simple truth (of course we will find out later that this is not really quite so simple) is that we learn things because we encounter them over and over again. If we decide to focus attention on them, because they are salient then they will be encoded in the brain. This is the basic truth behind learning, it requires repetition in the real world. In the language classroom, in lieu of repetition, we can use rehearsal as a way of mimicking the repetition process of the real world. In order to build chunks like a three dog night we have to hear the chunk and probably many, many times. It is through this simple repetition that information moves from being active to being accessible. We can see from this that the chef from active to accessible is not just a matter of a strengthening the connections but also laying down different connections. Redundancy is born.
8. How is it that learning can be described as Darwinian? (L10)
Learning can be described as Darwinian in that only the strongest connections which survive. Although we are born with a multitude of latent connections, established in part through the process of neuronal migration, many of these connections will die out or will be appropriated and reappropriated again and again for use in different neural networks which are more strongly activated. For a year on and its system of connections (synapses) to survive it must be used and to be used it must somehow be selected among millions of other neurons and billions of synapses. How does this happen? How are neurons selected for firing? At some point the system itself (remember these are all self-organizing systems) makes choices, not cognitive choices like we make (there is no cute little homunculus), but based on preexisting strengths of connections and contextual salience about which synapses will be used and through this use be further strengthened. As our experiences in the world changed so to do the connections in our brain. Our brain is Darwinian and evolutionary and that it is constantly adapting to new demands, which come from differing situations. Learning is then written by this situation which requires the learning. As teachers of language we need to seriously think about this. Students will learn what they think they need to learn, what the situation demands them to learn. If students feel that all they need to learn is material for a test, then that is all they are going to learn and trying to get them to learn anything else will be an extremely difficult endeavor. If we as teachers really want to try to control our students learning or, in better words, enhance our students learning, then you will need to come up with better reasons for them to learn. We need to embrace and use these Darwinian aspects of our cognitive system to our students advantage and not disadvantage.
9. How is it that speaking, comprehending and thinking are all one in the same operation? (L11)
In the cognitive linguistic model, in strong contrast to the analytic or generative linguistic model, language and thought are believed to be linked. The basis of this link comes from the simple observation that both language and thought are all connected to concepts. In fact they are both based on concepts. It is concepts which derive all linguistic communication as well is thought. For the cognitive linguist it seems ridiculous to try to posit different concepts for thought and language. Obviously both of them are derived from, created from the same concepts. Looking into language we can also split language into speaking (writing) and comprehending, that is production and reception. Again, it seems obvious that both of these are reliant upon concepts. The main difference between speaking and writing will occur below the lexical level, in neural connections related to either phonological or graphemic representations and the ensuing either articulatory gestures or hand gestures. Everything about these basic low-level neural networks related to actual movement will be the same. Similar observations can be made when we want to distinguish production from reception. Really, the main difference between these two is simply the direction in which the activation is going to spread. Lamb (1999) called for two different systems, one for section and one for production. I rejected this based on the idea that the two do affect each other. Looking now at directionality we can posit with some confidence that really there are no two different systems, but production and reception differ in the different directions or neurnonal pathways which they use. This is important because we now know that different directions will be either stronger or weaker. It is through these different strengths in the actual pathways that we it can account rather simply for differences between production and reception. Perceptive pathways will be stronger.
Having done all this, we can now clearly argue that speaking, comprehending, and thinking for all really the same thing. They differ only in the extent of the pathways which they use and/or don`t use. Thought will use only higher-level connections while speaking in comprehending will use the same higher-level connections in addition to some different lower-level connections. In this way it is easy to understand how they are all interconnected and really different aspects of the same thing.
10. How can linguistic learning be relegated to vocabulary learning? (L11)
As we saw when looking at the process of lexicalization, the formation of new lexical items, in less precise terms vocabulary learning, is really the basis of all language learning. Lexis is the highest system in the family of linguistic systems and all connections to other linguistic networks run to and from lexis. Once more it is not only connections to other linguistic systems but also extralinguistic systems which run through lexis. If we are thinking about comprehension, speaking, and thought it is really lexis which ties all these together. Language certainly doesn`t have thought and since it does affect thought it does this the level of lexis or words. It is through the process of lexicalization that concepts become linked to words, thus making them potentially much more accessible. We can think of concepts as existing as active bits of knowledge in our brain, and it is really through the process of lexicalization that they become accessible. Even though these concepts are there and at some level we have acknowledged them, it is not until they are lexicalized that they actually become accessible. It is through the process of lexicalization that we can begin to talk about concepts which causes them to change. Through lexicalization we begin to think through lexical items or words. Lexis really is that all important link.