Stephen van Vlack
Sookmyung Women`s University
Graduate School of TESOL
Human Learning and Cognition
Spring 2006
Week 9 Questions - Terry, Chapter 8 & Lamb, Chapters 14 and 15
1. What is short-term memory and how is it assessed?
The use of short-term memory involves recalling information which was acquired erewhile, and it lasts a few minutes at best under lab testing conditions. For example, to recall a counterpart's name in a conference people use their STM , and looking for new phone number will be quickly forgotten, even before reaching the phone in the absence of rehearsal. It can be said that people will have a tremendous problems in daily life if short-term memory does not work properly. It has two important characteristics. first, short-term memory can contain only a few items, "chunks" of information. Second, items remain in short-term memory only around twenty seconds.(Miller 1956). It means that STM is limited both in its duration and its capacity.
According to Brown-Peterson`s distractor task (1958), there are two reasons for rapid forgetting with the distractor technique. First, the absence of rehearsal, and the second reason seemed to occur in the absence of interference. The first reason is quite obvious for we know that rehersal is required to keep things in STM, but the second reason needs to be further explained. Interference refers to a particular theory in which forgetting was attributed to competition between similar memories. There are two types of interference, which are proactive and retroactive interference. Proactive interference is when information acquired prior to the formation of a target memory leads to forgetting of this newer memory. Retroactive interference is when information presented after the formation of a target memory causes forgetting of that previous memory. Since interference was believed to be primarily determined by similarity between competing memories, digit counting should not have caused retroactive interference with the recall of words or letters in the experiment by Brown-Peterson. However, the efficacy of STM depends on the items to be memorized and interference.
Memory span is the measure of STM capacity. The span of immediate memory is defined as the longest sequence of items that can be recalled in their correct order after a single presentation. An important determinant of memory span is the word-length effect: More items can be remembered when shorter words are the to-be-remembered items (Baddeley, Tomson, and Buchanan 1975). Memory span is unfixed so it can be increased with practice. Activities like the snowball game can be helpful for increasing memory span.
Even though there has been a large amount of research on short-term memory, the existence of STM is not perfectly clear. It might be considered as different type of long-term memory. There is also the problem of working memory and if it too exists, what its place may be in this memory triumvirate.
2. What are some characteristics of verbal STR and how does it transfer to long- term memory?
Terry, in his book, explains some characteristics of verbal short-term retention. These characteristics are those typically attributed to short-term memory, but it seems that they also apply to the phonological store in working memory. Usually in short-term memory, words, letters, or digits are remembered as they sound, as if they were being verbally rehearsed. Long-term memory, however, has been characterized as involving semantic encoding. That is, we remember the meaning or interpretation of a word in a general sense rather than the exact word (written form) or its sound. This verbal-semantic distinction explains why we can sometimes recall a just heard sentence verbatim, but in recalling it later from long-term memory we paraphrase the sentence into somewhat different words. For example, when you recall two or three words from STM, you memorize the words and their sequence acoustically. On the other hand, when you retell an interesting story you heard two or three days ago, you don`t repeat the exact words but convey the gist of the story by paraphrasing it. Of course there are exceptions here. Short-term memory is facilitated by the meaningfulness of the to-be-remembered material. On the contrary, long-term memories are sometimes encoded acoustically, e.g. poems. The difference here also shows us the relation between the to-be-remembered material and the strategies people adopt.
Also, STM has a limited capacity to hold information. The span of STM is said to be limited to about seven items or a lot less than that sometimes. Here an item is a unit already existing in long-term memory. Letters, digits, and words are each represented in permanent memory; each is an already known item. The way we can enlarge our limited span could be to increase the amount of information contained within each item. For example, the apparent capacity of memory span can be increased by parsing and encoding the material in terms of meaningful units, called chunks. When you memorize long numbers or strings of letters, you can remember better by simply dividing them into smaller meaningful units. This chunking strategy can allow us to remember better. In short, STM capacity is indeed limited and enlargements of STM are often accomplished by compiling items into more encompassing units. Short-term memory uses long-term memory. We can remember several items in STM when they are already represented somewhere in LTM.
In addition, the duration for which the short-tern memory can hold information is very short. Brown-Peterson task tells us that forgetting occurs after 15 to 30 seconds of distraction. On the contrary, information from long-term memories is fairly long. For instance, people usually remember their friends' names from their elementary school days after a long period of time. Forgetting from short-term memory was believed to be due to two reasons. One is spontaneous fading of the memory trace over time and the other is displacement of old items by new items. In forgetting, the distractors do not always cause loss of information from memory.
Then, how does STM transfer to LTM? When we decide we really need to remember (store) something in permanent memory, we seem to process information in a certain way in STM. We rehearse a name or number, try to form an image, or devise a mnemonic to aid late recall. According to several multistore theories, retention in short-term memory allows the opportunity for information to be transferred, or copied, into LTM. Among others, rehearsal allows more opportunity to encode into long-term store. STM has transfer function, and words which receive more rehearsals are generally recalled better. Verbal rehearsal may be necessary in acquiring some kinds of knowledge, such as learning new vocabulary words. New words need to be first remembered by sound, since a representation cannot be retrieved from LTM. It is suggested that residence in STM is neither necessary nor sufficient for LTM formation. It is also the case that maintenance of information in short-term memory does not guarantee entry into LTM and high level of exposure to material does not ensure retention in LTM. It should be remembered that processing in STM does aid Long-term learning. Activities such as forming mental images and actively interassociating or organizing the to-be-recalled material are controlled processes employed with short-term memory and they do benefit long-term retention.
3. What is working memory and how does it work?
According to the working memory model, STM actually consists of two different substores, which are phonological and visuospatial. The reason that we can divide STM into two different substores is from observations about what people actually do in their daily life. People seem to have no problem to complete two different tasks when the tasks are from the different substores of working memory. Since the two different tasks have their own substores in working memory, the interference between the two may be reduced.
One of the important features of the working memory approach is that dual-task performances can be assessed by this approach. As mentioned above, if two tasks use different STM stores, the interference between the two tasks can be decreased.
Furthermore the working memory works like an executive controller which is somewhat similar to convergence zones. It handles, directs, or helps the two substores to organize, plan, or perform information and actions. By doing this, it lets us know not only what to do but also what to avoid. Regarding language learning, we can teach learners what they need to know how to do in the TL, and at the same time also teach them what they need to avoid such as translation in the FL or SL situation, to become proficient language users.
The idea behind a working memory model is that there are basically two types of storage systems or modals and that these two are controlled by some central executive function. The two modalities are phonological and visuospacial. This idea is simply a further abstraction based on what has already been observed in Short-term Memory (STM). The advantage of the working memory idea over that of traditional STM views is that it allows people to engage in two modalities at once. In this model it is claimed that people can listen and see at the same time and still process both types of information without one interference from one or the other. The central executive function works to make sure that precessing is optimally fast and that attention is focussed in the right place for the right amount of time. Thus, precessing time and efficiency becomes an important factor in how the working memory functions.
4.How do we apply the STM in relation to language teaching?
As regards language learning, short-term memory plays such an important role in the processes of reception. When we read or listen to something, it is necessary to remember what the previous sentences are in order to be able to follow the meaning as we listen. Therefore it is critical for teachers to know what short-term memory is, how to use it, and also how to improve or expand students' short-term memory.
One way of aspects of STM is to use chunking strategies, since the STM can be improved and expanded through chunking items into meaningful units. In case of reading, the ability to read thought groups, which are structurally and semantically complete groups group often similar to phrases, is a very important strategy for efficient reading. By helping students to get to know how to form or find the thought groups, we can lead them to be more efficient readers and this is related to chunking.
Getting the students to focus their attention is also a good way of improving the effects of STM. Like all other things in the world, paying more attention generates better remembering. However focusing attention is expensive because it takes lots of energy. Students will not pay any attention if there is nothing to get from it. Therefore materials or lessons must be meaningful, and interesting to students. To do this, teachers sometimes need to act out or entertain. And also it is required to make the students be involved directly, and also interact with the lessons or materials.
Moreover, the long-term memory can help to build better STM. Using what students already know which is in LTM, such as schema or preassumption, helps them to remember things better. Using familiar topics or environments in language teaching can nurture students` language learning.
Chunking
Attention-getting
Limitations on new content
Using what is in the LTM
5. Why is sequencing such an important element in language and why does a neurological model
have to deal with this? (L14)
The importance of sequencing has not failed to escape us in this course. It should be clear at this point that all movement, physical, cognitive or linguistic, is reliant on sequencing. In essence this is the biggest question in the area of cognitive/neurological linguistics and the fact that we keep on coming back to this in Lamb (1999) without finding an adequate or simple answer is a clear indicator that we really do not know enough about how the brain works yet. It should be clear that sequencing is important or a problem even in this model because the PDP model which most cognitive linguists employ claims that many operation are occurring at the same time. Earlier in this course and in other classes we have spent some time dealing with the difference between serial and parallel processing models and came to the conclusion that the brain is able to do both or at least what looks like both. In essence the brain is able to sequence not just output but also input. The way the brain does this is not as simple as what one finds in a serial model nor is it complex in theory. In theory sequencing is based on the strengths of connections between different neural networks. The stronger the connections the faster they will be. This simple racing metaphor is only half the story. There are two other aspects of this. As a quicker impulse reaches a neuron and gets it to fire it triggers the neuron not only to release neuroenhancers so impulses can be sent to other networks the but also to release inhibitors in synapses unrelated to the present task. The other thing is that as each neuron fires an impulse is also sent back to get feedback. After all, the impulse needs to make sure it hasn`t missed any contextual details or has not incorporated erroneous information into the neural equation. What is also true, and Lamb does not mention this and I am not sure why, is that in almost all language use endeavors people make use of both top-down and bottom-up networks. This helps keep the information flowing to the right places.
6. What are some of the possibilities Lamb (1999) offers to explain sequencing? (L14)
Lamb (1999) discusses four possibilities through which we might be able to explain sequencing. The first two of these revolve around explanations in producing sounds while the others are little more global.
Clock timing
The basic idea behind this is that phonological elements of language are very much time sensitive. All languages have a strict timing aspect to the production of phonological units (either the syllable as in Korean or the stress group as in English). Lamb (1999) proposes a rather simple way for the brain for be able to deal with this. Clock timing is a simple and simple way of accounting for production timing. The idea is that there is a timing mechanism in the brain, like of like a beat or pulse which moves at a fixed interval which regulates how quickly certain neural networks (those related to linguistic production) are able to be used.
Feedback timing
Feedback timing has a few more applications in that it is clear that there is a feedback loop for all our productive and receptive activities and here Lamb (1999) is claiming that this feedback loop is has an effect on the timing of our production and reception. In this way we can account for slower and faster speech above the phonological level. It clearly takes longer to put together more formal discourse or to decode more formal or challenging language. This may very well be because we are being much more careful of what we are doing (we are closely monitoring both input and output) and this `being careful` generally involves the use of more and more complex feedback mechanisms. So the amount and the way we use feedback will alter the time it takes to perform a linguistic operation (or any other operation for that matter).
Self-contained delay element
A self-contained delay element might be a part of each neuron`s physical structure. Rather than firing randomly in all directions there might be a delay in the firing which gives more control to the firing pattern. This means that the neuron will be able to make use of other information and other firing patterns that are commencing at the same time. In effect the process of firing is slowed down to make it more efficient and so that fewer mistakes happen. This explains one of the most amazing things about the brain, which is it high rate of accuracy. Despite the complex patterns of neuronal firing (millions upon millions each second) there are very few mistakes or misfirings. There has to be a system which supports this high rate of accuracy.
Patterning - Structure
Throughout this course we have talked repeatedly about linguistic patterns and how these patterns form the building blocks of language. Well, it is important to mention here that pauses (in language: silence) are also a part of patterning. There are set patterns to how we should fire neurons not only in the components and the ordering of those components, but also in the timing of the components, that is the interval required between the different components. These kinds of things as well are stored in the brain and thus helps us with the timing of not only physical movements but of cognitive movements as well.
7. What are some of the linguistic illusions from previous, non-neurological studies of language? (L15)
These are becoming old hat to us now as most of them have already been beaten to death by Lamb at some point earlier in the book. The five of these generally revolve around distinctions between analytical (generative) and cognitive linguistic approaches and models.
The semiotic fallacy
This fallacy revolves around the idea that each concept has a single representation and that this representation is directly linked to the world. Cognitive linguistics has rejected this idea of a clear representation between mind and world for a quite some time based on linguistic observations of language usage patterns alone. As soon as we begin to look at neurology and how the brain actually precesses information coming in from the outside world the false nature of this highly oversimplified idea becomes even clearer. The concepts we have in our head although they are originally brought into the mind from the world outside are not the same as those in the real world., They are altered. They have to be altered in part because the brain has very limited mechanisms for storing information, but also because the brain is able to do things that are not possible in the real world like connect things that cannot connect in the real world. And it is through these limitations and also strengths that the world inside our heads (made up of concepts) winds up being quite different than the world outside (made up of physical objects and concrete actions). This is an important fallacy for us because the way the brain reshapes the world for internal representation and use is the same way that it works with language input. Language input is manipulated in the same way buy the brain. It is brought in and changed.
The monosemy fallacy
The idea here is that words and indeed word units do not have a single meaning, in fact they have no fixed meanings. When looking at how the brain stores and accesses words we can see how this idea of monosemy (Lit. single meaning) simply cannot be true. The phonological form of a word is composed of links and this neural net which is the collected composition of those links is connected to a multitude of other links related to all kinds of different information housed as well in neural networks. A `word` is composed of millions of bits of information. In using or word or understanding a word not all these bits of information are necessarily accessed. Only the bits that are triggered from and relate to the situation inn which the word is bring used will be fired. Every single time a specific word is activated it is done so on the basis of a different neuronal firing patterns. Since the situation is never exactly the same neither will the firing patterns be. In this way words are defined by the entire conglomeration of the connections, both primary (highly active) and secondary (less active or even passive) not by the link to a single neural network. This is simply not how the brain works.
Linearity in phonology
The idea of linearity in phonology (that a phonological string can be broken down into concrete units which are sequenced in a linear order) is what has been used to support the idea that there is such a thing as a concrete phoneme. Again, as we saw in the fallacy above, the brain is not a representational system where representational units are stored as single complete entities, but rather a system of connections. No where is this truer or easier to see than in phonology. All we have to do is look at any phonological string to be able to discern a large amount of blending in the units. Interestingly, as we discussed briefly above, timing is an important aspect of this. The shorter the time used, the more integrated the sounds will be. Based on this we can make a similar proposal for phonology and phonemes that we made for words above, namely that a phoneme is not a complete, intact representation of a set of phonemic features which are all necessarily present in the same form each time the sounds is due to be made, but is the conglomeration of a rather wider set/range of articulatory gestures (or really the networks that drive them) which are chosen based on the context which for phonology is the preceding and following sounds as well as the timing and, relatedly, genre of the speech event.
The word and morpheme unit fallacy
Generative grammar makes a big deal about the level of the representation and has traditionally made an important distinction between morphemes and words. This distinction was made on the basis of grammatical operations (morphemes are supposed to behave in rule-governed ways while words aren`t) and not meaning. Since cognitive linguistics emphasises meaning as the organizer of language and not syntax the word/morpheme distinction simply disappears. Eliminating this distinction does make the overall system quite redundant as morphemes (particularly inflectional ones) will be represented in many places and many times in the brain, but the brain is undoubtably redundant in its storage.
The rule fallacy
This last fallacy is the one that often inspires the most resistance not from generative linguists quite so much as language teachers who are still caught up in traditional grammar. The fact remains though that grammar rules are simply not part of our functioning linguistic system. In fact, there is no way for the brain to process any kind of rule except in the system of episodic memory. Rules are stories that humans make up to control things from the real world both in deed and understanding, but the fact remains that they exist neither in the physical world nor in language itself. There is simply no way for the brain to store rules except as episodic memory. The day to day, moment by moment functioning of the brain comes as the result of patterns of associations. That is all. Yes, these patterns are more numerous and therefore complex than we can comprehend, but in this simply the way the brain works and as we have seen in this course, the brain has many way to regulate and use this multitude of connections efficiently and virtually without mistakes.
8. How does being aware of these illusions make us better teachers of English potentially? (L15)
The main problem with these illusions, when taken collectively, is that they actually lead to the TL (in this case English) being stored in the brain in a fundamentally differently way than the L1 of the learners. From our brief review of fallacies above was can see how teaching English with these fallacies will relegate much of what the students know about English to episodic memory. Even illusions like phonemes will be stored as units of episodic memory (probably along with their IPA representations). This imposes huge restrictions on linguistic operations as the information is stored in ways which are not part of the functioning linguistic system. All this stuff has difficulty being used. The time spent learning all these erroneous facts (which are in, fact, wrong) not only take precious time away from the presentation of linguistic patterns but also crate confusion because the rules we teach are indeed wrong at least some of the time.