Power to the pupils; software in language learning

This paper attempts to state and justify some methodological principles behind CALL, to describe and classify available software for language learners and to talk about trends, in other words to indulge in prophecy. It is not about machinery. Machines themselves are of no importance, or should not be; they are the paper on which the software is written and, just as you only think about paper when it becomes so torn and dirty that you can't read it, so you should only need to think about machines when they are so awkward and unreliable that you have difficulty using them. Unfortunately that happens more often than it should, but they are getting better all the time, better in the sense of being unobtrusive and unimportant.

Success story

In 1975 I conducted an oral examination in English for a group of pupils in Bangkok. Most of them had reached the standard you would expect; they could use English, but their vocabulary was limited, they made a lot of grammar errors, and their pronunciation was heavily influenced by Thai. There was one student, however, who was different. He understood everything I said; his accent was almost perfect, and he hardly made any grammar errors. What made him different? I looked at his record form, but could find nothing to explain it. He had not lived overseas and did not seem to have attended any particularly well-known schools. “Who were your teachers?” I asked him, thinking he might have had a native speaker in his school class. “My teachers?” he said, “my teachers were Richard Burton, Laurence Olivier, Vivien Leigh and Gregory Peck.” He was a cinema buff, and watched movies with the kind of empathy that led him to imitate the behaviour of the screen heroes and heroines, including their language behaviour. "Pretentious little git," you may be thinking, but there was not much wrong with his English.

A naïve reaction to a success story like this might be to say: school language classes are useless; let us instead spend all the hours for English on the timetable at the cinema. We could run a controlled experiment, perhaps, with one group getting ordinary textbook work and the other group seeing classic films. Then we could publish learned articles showing that the boys in the cinema group performed 2.1% better than the control group and that this is significant at the p=.99 level, while the girls showed only 1.7% gains and this is significant at the p=.9 level.

A more sensible reaction would be to notice that what distinguished this pupil was his enthusiasm, his love not so much of English as of some users of English. English for him had become a hobby rather than just a subject. Once you have harnessed that kind of enthusiasm and alertness to a subject, then there is no further problem in getting the subject learned. The quality of the teacher and the materials hardly matters; old-fashioned materials serve just as well as the most modern ones.

There are many teachers who would be uncomfortable to have that student in their class. It is usually unpleasant to have learners who do not succeed in learning what you teach them, but it may be even more disturbing to have learners who learn what you do not teach them. Yet there is bound to be a mismatch between teaching and learning (using learning for the time being in a broad sense which includes acquisition). If there were not, then education would be a continuous process of attrition, since you cannot pass on the totality of your knowledge. What we know about language as competent (not necessarily native) speakers could not possibly be stated explicitly in a finite time.

Latent grammar

A simple experiment can illustrate this. Imagine you are doing a grammar exercise, a straightforward sentence transformation task. Write down the first response that occurs to you without agonizing over whether it is right. Here is the task: make this sentence singular:

They write books.

You may have written She is writing a book (or He is writing a book) doing exactly what I asked you to. You may have written She writes a book but then worried about it, perhaps crossed it out. The point is that I did not tell you to change the verb aspect, and yet that is what you would automatically do if you were thinking about meaning rather than form. If we consider the following four sentences:

all four of them are grammatical, but the first two are natural, while the third and fourth require rather elaborate contextual justification. I could dream up contexts for them, for instance for number three:

She keeps getting divorced, yet she still manages to look cheerful. How does she do it?
—Oh, every time she ditches a husband, she writes a book.

Sally used to go to the bridge club regularly, but she has given that up. She's writing books instead.

If you want a linguistic explanation, you could say that there is a semantic feature plus or minus specific which applies to both verb phrase and noun phrase. A noun phrase with a can be either, but it defaults to plus specific, i.e. we expect the speaker to have a particular book in mind unless the context makes that unlikely. On the other hand the plural books defaults to minus specific. Similarly the continuous verb aspect is writing defaults to plus specific; we expect the speaker to have a particular act of writing in mind, whereas the simple present writes defaults to minus specific. Now it doesn't really matter whether you made that analysis or even understood it. What matters is that you instinctively would have chosen the right form in any meaningful context. How did you do it? Nobody ever taught it to you; it is not in the textbooks as a grammar rule. If textbooks came down to this kind of detail, they would weigh tons and cost fortunes. Textbooks are necessarily meagre, a characteristic that is often overlooked. The quantity of teaching they contain is insufficient to do the job unaided. As managers of learning we have to make sure that we open windows on a wider world of language outside the textbook.

Passive learning

The outside world, and even some people within the language teaching profession, often seem to have a Barber's Shop image of education. You want your hair cut? OK, go and sit down on that chair, turn your head the way the barber tells you, and wait until he or she has finished cutting. You want to learn English? OK, sit on that bench. Open your mouth when the teacher tells you to, and wait until he or she has finished teaching. I am exaggerating, of course. Nobody seriously imagines that learning is quite as passive a process as getting a haircut. However, this kind of thinking accounts for the spectacular failure of certain kinds of CALL. One claim that used to be made often was that computers would let students make up work that they had missed through illness. (The same claims had been made for programmed learning and language laboratories.) Another claim was that one could learn at home. The strongest claim was that made by Tom Stonier that in future home would be the place one went to learn, and school would be only a place for socialising (1983:172). Yet any child who has tried to make up more than a few hours of missed work, or any parent who has tried to buy their offspring a real head start by getting a computer and some disks, will know that it doesn’t work. It is easy to blame the machines and the software. But do you think that a compulsory series of visits to the cinema to see old Gregory Peck movies would work any better? One cannot simply give a child a disk and say: “Away you go and learn your irregular verbs.” Learning involves involvement. Parents who succeed in improving their children’s education are most often those who learn alongside them.

The best software is software which involves teachers, learners, and parents in the same enterprise. We should distrust the kind of software that tries to do the whole job, that tries to foresee every possible student error and supply an appropriate feedback message for each. Software is only one element of a learning environment that includes many other components, such as the dormant knowledge of the individual, what you can work out or remember if you are not under threat of public humiliation, such as the pooled knowledge of a group who collaborate on a task, such as the use of a teacher as a consultant—“Come and tell us why this did not work”, and plain old-fashioned reference books. Learners often succeed in making software change its nature by using it in new ways; drills become games, toys become tools, and tools become toys. Long may this continue. In particular drill activities which are timed are sometimes turned spontaneously into races, and anything which involves scoring becomes a competition, either a competition against others or one against oneself, a constant effort to improve on one's best scores. Classroom teachers can also harness the competitive instinct among learners, but the language laboratory, with its pre-recorded fixed response, could never do this.

Types of software

The four-way classification scheme offered below is rough-and-ready, and I doubt whether every program could be made to fit one of the four types comfortably.

DO WHAT I TELL YOU: this is the first and most familiar category. Here the machine controls to a great extent the nature and order of events. This includes drills, exercises, quizzes and tests, and at the most organised end of this spectrum programmed learning. The machine gives you a task, say a sentence to complete or a question to answer. You can answer by typing out a word or phrase, or by selecting an answer in a multiple choice format, say by pressing a number key. The machine tells you if you are right or wrong and then invites you to try again if you were wrong. There may be some feedback about why the answer was wrong. As soon as you have found the right answer, you are moved on to the next task. These programs, just like books, can be good or bad, error-free or error-ridden, relevant or irrelevant. They are often welcomed by learners simply because the procedures are familiar; the rules of this learning game are well understood.
However one interesting point has emerged from recent experience, which is embodied into some of the best software, particularly Wida Software’s TESTMASTER. The conventional assumption used to be that you commented on wrong answers but, as soon as a student has found a right answer you say “Well done” and move on to the next item. What Chris Jones, the author of TESTMASTER, has pointed out is that students have not necessarily lost interest in an item just because they got it right. Perhaps they would like to know if there are other possible right answers. Perhaps they want to know what is wrong with a different answer which they were thinking of giving. This is a point which class teachers should take note of, not just programmers.

GUESS WHAT WAS THERE, the second major category, is perhaps the most significant innovation in language teaching that CALL has supplied in its short life. Tim Johns invented the form with his TEXTBAG program, in which all the words of a text were masked out and you have to point to single words and buy them, buying the minimum number of words that you needed in order to answer a comprehension question displayed at the end of the text. I followed this up with STORYBOARD, and its latest manifestation ECLIPSE, in which the words of a masked out text had to be guessed, and each occurrence of a successfully guessed word was supplied wherever it occurred, in technical terms a type-replacement activity rather than a token replacement activity. STORYBOARD probably has the distinction of being the most widely imitated CALL program ever; there must be more than a dozen versions current. It has to some extent changed our perception of teacher/learner roles.

CAN I HELP YOU?, the third category, describes uses of the computer as tool. If we as writers or administrators find them useful, why should we cut our students off from the benefits? Let them explore the language with their aid. Let them have the satisfaction of producing attractively presented work that they can display with pride, edited and corrected (by the teacher and by each other) with none of the tedium of making a fair copy by hand. The computer's natural role is that of a slave, obeying orders and carrying out jobs for its master on demand, and the obvious language job is word-processing, in other words displaying fair copy of the master's second thoughts. For all that we may value hard work, it is generally the case that when we make work easier, we get more of it done. Whatever the pocket calculator may have done to our mental arithmetic skills, it has certainly led to more calculations being performed.
All word-processors have a search-and-replace function, which can be used to create practice material from any piece of text which has been typed in. You can replace all the articles with **** and then print out the text so that the learners can write them back in. For a more difficult task one can use a piece of first-person narrative and replace all the first-person pronouns with ****. Now ask learners to re-write the text as third-person narrative, putting appropriate pronouns or names into the spaces and making any other necessary changes. Most word-processors include outliners, spelling-checkers, thesauruses, and grammar- or style-checkers. These are of very little use as writing aids, but they are fascinating as ways of exploring the language, as programs to play with. Feed in a nonsense word and watch what alternatives the spelling-checker offers you. Look at the set of synonyms offered by the thesaurus and then look for synonyms of the synonyms. Discuss which words will fit the context, and what is wrong with the others. Type in some Shakespeare and watch the style-checker offer its naïve comments.
The most powerful aid of all for letting learners play with language is a concordancer, ie a program which finds particular words in a text and prints out the line on which the word occurs. A Google search functions as a concordance. True concordance programs produce Key-Word-in-Context (KWIC) displays, where you see the key word in the centre of the screen with half a line of context before and after, making it much easier to notice the way words come together, eg which preposition occurs after believe or the fact that the word same is almost always preceded by the. There are also many public-domain utilities which let you turn a text into a list of all the words used in it, sorted either alphabetically or by their frequency.

HOW DO I GET OUT OF THIS? the fourth category, consists of activities such as simulations, games and puzzles, many of which were not created for language learners at all. Even those that were, such as my INVENNTION program, often lead to the comment, “But that's a maths program, not an English program”. (INVENNTION uses Venn diagrams as the basis of logic puzzles, hence the spelling of the name.) The response I give to that, of course, is that we live in a world full of mathematics, the shape of a building, the speed of a car, the cost of a glass of beer. We talk about these concepts in language, and we are not obliged to limit such talk to the maths lesson. We should always be trying to open a window on a world in which language can be acquired, a world where language is used in a concentrated way to receive and transmit meanings.

The future

So finally to predictions. Where are we going from here? When I first gave this paper in 1988 I suggested that with interactive video and CD-ROMs we would soon have a world where the whole of the Library of Congress can be consulted from one’s bedroom by pushing a few keys, along with tele-conferencing and the global village of instant communication. Twenty-five years later these things have more or less come in the form of Wikipedia and Google Books, and are largely taken for granted. Meanwhile what of the learner and the teacher? Learners can use these things just as easily as anybody else. Machines supply power. However, we will need to learn how to use it. Although we will be able to get answers to an infinity of questions, we need to train ourselves to ask more questions. When we have the training we will have power to learn from a much wider part of our environment than a single teacher can provide.

The learning instinct is strong in all normal circumstances; we are constantly engaged in an effort to make sense of our environment, and in making predictions. There are two cases in which this learning instinct will be suppressed. One is when the environment is too complex to be explained; that is what happens to me when I hear conversation in Chinese. The other is where the environment is so impoverished and predictable that it contains nothing which requires explanation, no problems to be solved and no uncertainties to be predicted. This might occur if I had to carry out an automatic drill on sentences whose meaning doesn't matter. Both cases, too much challenge and too little, lead to the same outcome, boredom or the switching off of the learning instinct.

What computers are good at, because of the rapid and slavish way they respond, is supplying optimal challenge, something a teacher with forty students can hardly ever do. Optimal challenge is individually determined and can vary from minute to minute; at times I will want to be stretched, at others to receive all the help that is available. With the computer as a slave, learners and small groups of learners can readily find optimal challenges. This more than any other reason accounts for the so-called addictiveness of computers, the fact that even children with short attention spans can spend long periods with them.

The outcome is not likely to be that everyone will uniformly learn more and improve their performance. Since computers supply opportunities, not fixed paths to follow, improvement will be sporadic and unpredictable; we are likely to encounter more individuals like my star pupil in Thailand without necessarily seeing an overall jump in standards. The pioneers of CALL saw themselves as handicappers in a horse race, bringing all the horses to the finishing line abreast. What I predict instead is that computers will act like adrenalin, sending some of the horses racing far ahead of the field. In the process teachers are going to find their authority, in particular their authority as “knowers” of their subject, will be undermined, since some of the learners will seize opportunities to discover things the teacher does not know. Teachers will have to find new roles, as advisers, as managers, even as fellow learners discovering new insights into language by using the same facilities as their students. For some teachers it will be an uncomfortable experience. Good luck!