You'll know if you've been following this blog for a while that I'm opposed to the view that learning by translating is simply learning to translate. But why do so many people hold this view? Well in many cases I imagine it's due to their own experience, and awareness of their own blockages when learning.
Most of us who have studied languages will have felt at some point this sensation of juddering along from word to word, one by one. I certainly have, but it's not something I put down to translation; and if I'm right, avoiding translation not only fails to solve the problem, but actually risks making the problem worse.
What I came to realise was that juddering wasn't caused by looking for the English word, just simply by looking for the “word”. That is to say that I had one form of the word in front of me, but in order to understand it, I needed to recall a “reference form”. Their are two candidates for the reference form: the dictionary headword in the target language and the dictionary headword in your native language. In order to translate a word back to your native language, you need to first disinflect* the form you see back to the headword form in your target language, a task which is going to be a lot more difficult than recalling the link between the target and native headword forms (which should be instantaneous because you get far too much practice of this anyway).
(* No, disinflect isn't a real word, but it's just so beautifully apt here that I couldn't resist. It just seems so sterile, so clinical, so... “unlanguage”....)
Don't believe me? Well consider you're learning a verb from a language with a reasonable number of conjugations (think Spanish or German). If we take a totally regular verb, we may have 6 forms in each tense, and we've got multiple tenses – let's say 6. If I quiz a student on all 36 conjugations, and they have to convert back to the headword form, they perform the conversion of each form to the headword once only. If they take it one step further and translate to the headword form in their own language, they've practised that conversion 36 times. At a ratio of 1:36 you can see that the translating step will very quickly be learned to the point of automaticity.
So it's the disinflection – the removal of conjugation or declension – that's takes the time, and it's that that we need to eliminate.
And yet that's precisely what a lot of target-language-only courses encourage, but explicitly asking the students to conjugate time and time again, reinforcing this unintuitive, dictionary-led model where the meaning exists only in the headword form. Consider this sort of task:
Put the following verbs into the appropriate tense:
- I ____(tell)_____ you if I knew. (answer: would tell)
- I gave him a biscuit after he ____(stop cry)____ (answer: stopped/had stopped crying)
As we go along, then, we're reinforcing the headword form in every task, and the individual inflected forms a fraction of the time. Little wonder, then, that the students have no problems with the headwords and massive problems with everything else, and we end up descending into Tarzan-speak: I … go … shop … yesterday.
(Which leads back to my ponderings the other day about whether creole languages are really created by the speakers, or if teaching has a role to play in their creation. In TPR you could suggest that the imperative supplants the dictionary headword as the “reference form” of choice.)
The consequences of this headword fixation can be quite embarrassing, and can even cause offense, when you consider that headwords in the Romance languages, for example, are usually male. You risk ruining the effect of a chat-up line if you accidentally call a woman beautiful in the masculine form....
But this type of error is difficult to eliminate after the fact, because no matter how many remedial exercises you give the student, if they solve the problem by reference to the headword form, then they're reinforcing the behaviour that causes the problem.
Summarising the problem and the misunderstanding
The anti-translation camp thinks it is a problem if we label the equivalent of “to go” in our target language with the English “to go”. If it's Spanish, we should label it with “ir”, if it's German it should be “gehen”.
But it is the very notion of having a label in the first place that is wrong. We need to think of it as a fuzzy concept, a collection of go, went, gone, goes and going with none of them being more important than the others.
Headwords are for dictionaries, not for our heads.
So what do we do?
The headword form is a form with little communicative function (ironic then that the communicative approach places so much emphasis on it!) and while it may be useful at the very beginning, we have to eliminate it as soon as possible.
And that means no vocabulary tests: if you want to test that a student knows a word, you need to give them the opportunity to use it in all it's forms, individually and separately, but you also need to give them a reason to use it, and a context in which it actually means something.
That means varied practice, and in the context of controlled practice, it means that target-language-only is unlikely to cut the mustard, as it is very difficult to avoid a reference for except in very specific circumstances, such as converting between direct and indirect speech in English:
“I'm tired,” he said. ↔ He said he was tired.
Which leads back to a conclusion that you may (rightly) accuse me of taking as a premise anyway: translation is a Good Thing if you do it right.