After writing my post on gap-fills and cloze tests in language lessons, I went back to a blog post that I'd read some time ago, which was part of the Pools project that had introduced me to technology such as Hot Potatoes. I commented about how I had never seen the purpose of fill-in-the-blanks as a learning tool and his response was that it was a test.

Now, it has often been observed that repeated testing aids student retention, so most teachers would assume that anything that is a test is valid as a retention aid. Is this true?

As I said in my earlier post, the cloze test and the gap-fill rely on having a sound internal model of the language under test.  Does doing a cloze or gap-fill help build that knowledge?

In the previous post, I said that doing these tests early appeals to conscious knowledge.  Many of the most successful students will look at a fill-in-the-blanks exercise and reason through it.  If you ask them how they did it, they'll say things like "that's a noun and that's an article, so the thing between them must be an adjective".  As the conscious strategy proves so successful, the learner will continue to apply it and will perhaps never develop the gestalt.  And because any attempt to use gestalt at this time will appeal to the student's first language, the student who approaches the test in the intended way will be penalised.

In effect, the student learns how to pass the test, rather than learning the specific competencies that the test was originally designed to measure.

How much damage does this really do?  In my opinion, a lot.  As students go through their academic career, they will be expected to do more complicated things, and they will be expected to do them quicker.  But there is only so fast that we can conscious churn through these rules.  Sooner or later, we can only succeed by gestalt, but how can we encourage that?  The mechanics of testing militate against going by your gut, because this leaves you with a worse mark.

I'm not saying that we shouldn't use testing as a learning mechanism early on, but that we should look again at what constitutes appropriate testing - what tests the student can carry out in a way that supports, rather than hinders, learning.

Testing, testing, 1, 2, 3...
This of course does not only hold for testing with fill-in-the-blanks.  It has been said by a number of academics, with figures to back them up, that testing aids recall.  I'd like to present an argument against that.

"Are you mad?" I hear you cry.  Yes, but that's beside the point.

My point isn't that their figures are wrong, but that the word "test" is something of a red-herring.  What is it we do when we test a student's knowledge?  We check that we can recall accurately what they have been taught.

If we discuss this as "testing", we have in our mind a goal -- scores, marks, grades.  We don't want to focus on that goal when we use testing as a learning aid -- we need to focus on the process.

The process involved in testing is accurate recall and application of learned information.  A task can be designed to require recall and application without being strictly a test.  If the figures say that testing is a successful classroom strategy, might it not be because they are comparing "tests" against other tasks that do not require any recall or genuine sentence construction skills?  I'd like to demostrate how some of the most common exercises fail to rely on these skills.

The problem with drills 
Form drills, pattern drills, substitution drills; these repetitive exercises fail to teach because the core language we expect the student to learn from the task is never subject to recall.  The teacher says it and the student repeats it, changing only a vocabulary item of very small grammatical features.  We cannot learn to recall something without doing it.

The problem with communicative tasks
Anyone who has been at either end of a classroom recently is likely to have come across communicative exercises based on an idea like the "knowledge gap".  Each student has partial information or a partial picture, and the students have to talk to each other in the target language to get the information from each other.  However, being understood by your classmates is different from being understood by a native speaker, and accuracy not only slows down the task but potentially renders you incomprehensible to a classmate of lower ability.


In general, though, regardless of the exact nature of the task, the class starts by presenting or otherwise providing the students with the language that they are going to use.  If we attempt to do this by eliciting the information from the class, we may get one or two to recall it, but most will not -- instead they will end up holding the patterns in working memory, often meaninglessly and mechanically, and parroting the phrases during the class.

Conclusion

We all want to receive or provide the best education possible.  Empirically we know that "testing" aids this, but that statement is an oversimplification of the real situation.  There is nothing magical about "a test" that makes it more effective than "an exercise".  We must examine what the core activity is in what we consider a test to be, and we must find ways to incorporate that into the day-to-day teaching process.

I contend that the distinction implied between "teaching" and "testing" is artificial -- teaching we consider to mean presenting information and going through some kind of repetitive "training" regimen, whereas testing is an unsupported check of recall.

The idea that "testing aids retention" is therefore an obstacle to good practice, because it prevents us from looking at the nature of the tests to identify what really happens.  I believe the real point is that recall practice aids retention.  If so, our task design must always be built around developing recall, or the student will never be able to spontaneously produce language.

What __ this?

I think most _____ will be familiar with this ____ of thing.  At some point in ____ education, you will have ____ across a piece __ paper that looks a ______ like this.  Do you ____ what it's called?

Like some of the gaps in the above there are two answers.  Unlike the gaps above, one of them is wrong.

Some people will call this a "cloze test", others will call it a "gap-fill".  Many teachers will say that these are the same thing.  That is a mistake, because while a cloze test and a gap-fill may look alike, they have different goals, and a learning exercise will not be particularly effective it is written without any thought to its goals.

So what are both these things?

Cloze test

The Cloze test, as I was first taught it, is a measure of reading fluency.  The theory goes that there is enough redundancy in language that you don't need to hear or see every word to make meaning from a passage.  In fact, they go so far as to suggest that your brain will usually be able to automatically pick the word to fill the gap.

This arose from the Gestalt school of psychology.  The idea of "gestalt" is quite simple, but very powerful.  The gestaltists said that the brain isn't quite as literal or as primitive as previous psychology suggested.  The idea behind gestalt is that when the eye sees part of something, the brain understands it in terms of the whole.  If you see half a face through a window, your brain knows full well that there's another half to that face.  If you see a man standing behind a low wall, you won't see his feet but you will not only know that he has feet, but your brain will actually make a good estimate of where they are and what size they are based on your knowledge of other people.

This ability to fill in what isn't there is called "closure", and that's where the name "cloze" test comes from.  A cloze test is therefore suggested as an effective way of verifying whether someone can read -- if they have the proper map of language and a proper connection between written and spoken language, it should be effortless.

The gap fill

The gap-fill exercise was, as I understand it, proposed by the cognitive psychologists as part of a structure of meaningful learning as a means of supported recall.  At its most basic, a gap-fill is a way of avoiding questions. For example, instead of asking a simple science question like "What is the boiling point of water?" You can give a gap-fill of "Water boils at __ degrees Celsius".

But gap-fills are more subtle than that.  When asking questions, you would usually ask them at random.  The special thing about the gap-fills proposed by the psychologists is that they are used in a logical progression that reinforces the structure of the topic at hand.  So the example earlier might continue: "Water boils at __ degrees Celsius. The gas formed is called water _____." This is where the magic happens.  If you simply asked the question "What is the name of the gas formed when water is boiled?" you might get back the (wrong) answer "steam", even though in the class you would have already taught the difference between "steam" and "water vapour".

The problem that the gap-fill is designed to solve is a simple one in theory, but tricky in practice: how do you teach a new fact such that it "overwrites" a previously learned erroneous fact?  A well-designed gap fill presents enough information to provoke the memory that the designer wants to reinforce (water vapour) but also uses the structure of the sentence to supress the wrong answer (the learner can't answer "steam" without producing the phrase"*water steam", which they will automatically realise is wrong).  It takes advantage of both the gestalt of the phrase and conscious thought.

That's the difference, but what difference does it make?

Language lessons will often include something that looks like a gap-fill or a cloze test, but is it either?

First of all, a cloze test is pure gestalt -- it starts with something you know, and takes bits away to make you prove that you know it.
The cloze/gap exercises in the language classroom test new knowledge, before the student has a solid internal model of it.  There is no "gestalt", so there is no process of closure.  It is not a cloze test.

On the other hand, the gap-fill uses the gestalt of the language it is in to assist in the recall of domain knowledge.  The medium should be known to support the learning of the new domain knowledge.

We're now in rather confusing territory because in the language classroom, the domain knowledge and the medium of the gap-fill are both the same thing -- the target language.  So we need the target language to be known in order to support the learning of the target language...!

If it's not a gap-fill or a cloze test, what is it?

I'm not sure how I'd define what we're left with in a short way, so I'll have to do it the long way.

What is it testing?  Well, gestalt thinking is a completely subconscious process, so if we're not using gestalt then we must be looking at conscious reasoning.  In language -- conscious reasoning means only one thing: metalinguistic knowledge.

Is this a bad thing?

Well, the whole point of the cloze/gap exercises in current language teaching is the notion of learning naturally, subconsciously, in the target language only.  The point is that we're supposed to be avoiding the conscious study of grammar, but in the end, many of these gapfills are about mechanically chosing appropriate conjugations or prepositions, the same mechanical metalinguistic tasks that current educational philosophy tries to avoid.

What's the moral of the story?

In essence, we all become the thing we hate most.  Reformers often hate what came before, and try to distance themselves from it.  Unfortunately, as in this case, this leads to a failure to critically evaluate the "old" method and work out what was genuinely wrong with it.  The differences often become superficial and we repeat the same mistakes of the last method but just dress them up in new cloze. (sorry)

When designing a learning exercise, teachers must be scrupulously honest with themselves about what the task is, how it works and how the student will approach it.  It is no use to present an active, conscious task like this if you genuinely feel that conscious study has no place in the language class, and if you accept the need for conscious study, you have to ask yourself why we do not accept the students' first language in the classroom.

After all, if you truly believe in the gap-fill, then you must believe in using a known language as a gestalt to support conscious knowledge of a new subject -- in this case, the target language.