Thursday, 12 December 2013

Efficiency and Redundancy in the L2 Classroom

1. Information Theory
This is the Shannon-Weaver Model of Communication –

It was originally developed (and is still incredibly influential) for mechanical and electronic communication: telegraphy, telephony, the internet and the like. We need to be very careful about taking it too completely as a metaphor for spoken communication. That said, it can still be a useful model. We can re-label it for L2 or lingua franca exchanges like this –

Don’t get too concerned with the diagrams. The important point is that there is always noise. ‘Perfect’ transmission from source to destination is impossible, and so communicators have to adopt strategies to deal with this. These strategies, consciously or not, revolve around the two key concepts of Efficiency and Redundancy.

2. Efficiency and Redundancy
Shannon thought of ‘information’ as something unexpected. Something which contains no new ideas – something that tells you what you expected to hear – contains very little information. For example, this sequence –


If the next letter in the sequence is B, that’s very predictable and so contains little information. If, however, the next letter is C, that, like turkeys on Christmas Eve, is an unexpected event that you wouldn’t have predicted from the preceding data and so has a very high informational value (Shannon called this value ‘entropy’, but don’t worry about that now). Informational value, then, is essentially probabilistic.

Communication methods, including ‘natural’ human languages, exploit this probability for more efficient communication. Information that can be successfully derived from context is often omitted (like the Japanese omission of subjects and sometimes even objects). An ‘efficient’ way of expressing the above sequence would be something like –

ABAB... and so on.

Much shorter. However, this obviously presents a problem if we ever get a C. In fact it presents two –

              1. External Noise. Remember, perfect transmission is impossible. The receiver could just mishear C as B, or the transmitter may mispronounce it.
              2. Internal Noise. This is the big one. Because the receiver is actually expecting a B, they are more likely to hear that. Their expectations create internal interference which prevents correct reception. Run, turkeys, run!

This Internal Noise is entirely natural. In many respects it’s a mark of quite sophisticated language use: in order for something to fail to meet your expectations you must have been able to establish those expectations in the first place. And because it’s natural communicators adopt strategies to compensate for it, the primary one of which is redundancy.

At the simplest level, this is just repeating yourself –


However, doing that so bluntly is fantastically inefficient, as well as lacking a certain finesse (and risks insulting your audience by implying that they’re dumb) so a number of other, less blatant strategies are invoked, some of which are embedded in the language at a fairly fundamental level. In English, the past and future tenses and pronoun agreement are often good examples of this –

A - “My sister went to the shops with her friend last Saturday.”

In this example, ‘last weekend’ explicitly places the timeframe of the question in the past, so using the past tense of ‘go’ adds no new information. Likewise, the gender of my sister is already given in the word ‘sister’, so the informational content of ‘her’ is redundant, compare –

B – “My sister went to the shops with his friend last Saturday.”

‘He’ is unexpected, at least in most situations (and that opens up a whole other can of worms about sociocultural power relationships and marked and unmarked language that we won’t go into here). It is new information, in a way that ‘her’ isn’t. However, ‘her’ still serves to indicate a possessive (similar to in Japanese), but this function too is already served by ‘with’. ‘My’ is understandable from context, and loads of languages function perfectly well without articles. The destination is known (‘shops’) as is the fact of movement (‘go’), and so the directionality of ‘to’ is also unnecessary. After all that, we get this maximally efficient version of example A, which it is impossible to simplify without losing any information:

C – “Sister go shops with friend last Saturday.”

Now this looks a lot like something that some of my lower level learners might produce. That’s not a coincidence.

3. Learner Perceptions
Language learners, and lower level learners especially, are incredibly efficient communicators. This has the side effect of leaving very little room for error: the meaning of example C is the same as A, but take out (or replace) just one more word and the meaning changes. That’s not the point here though.

The point here is that, because learners produce very efficient sentences, that’s what they’re expecting to receive as well. Because every word they say is important, they expect that every word they hear (or indeed read) is going to be similarly important. Redundancy (such as the way that last sentence paraphrased the one before it) is NOT an aid to communication in the same way as it is for native speakers.

This is important, and the thing that native speakers often overlook when talking to L2 learners. That redundancy is hard-coded into the grammar of a language, so to a certain extent it’s unavoidable if you want to speak with anything approaching fluency. More significantly, at a level beyond grammar redundancy is one of the key strategies language users adopt to increase comprehensibility. Metaphor, paraphrase, idiom; all rely on repetition but, because just straight-up repeating yourself lacks elegance, they introduce slight variations as well. These variations increase informational content, but information, you remember, is the unexpected. Information by definition is unexpected. Learners don’t expect this (again, by definition).

The greater the informational content of an utterance, the more complex it is and the more work listeners need to do to understand it. But the whole point of communication is to convey information. The key in the L2 classroom is to convey only the necessary information, with as much efficiency as possible.

I realise that, not without a certain irony, I’ve just used a huge amount of language to make a very basic point (“SPEAK SIMPLY”), but that’s part of what I’m saying: the strategies native speakers automatically adopt to explain things don’t work with language learners, in fact they’re actively counter-productive. Therefore, as a native speaker in the L2 classroom, you have to give much more thought to what you’re saying, not just at the level of vocabulary, but also higher-level discourse strategies. What seems inelegant and insulting for fluent speakers isn’t for learners, and repetition is fine, but only repetition of what really matters. If native speakers don’t understand something you usually try to give them more information, but if students aren’t understanding then you have to give them less – less but more important. I know this sounds obvious, but all the available evidence suggests it’s pretty hard to do, because you’re working against decades of learning that more information is better.

No. Better information is better. And that requires thought.

1 comment:

  1. That was actually quite interesting, though as you rightly state at the end, the whole point is quite simple really: less is more. Yep. See also, Twitter, wabi-sabi, TEDex and all that. Or as Twain said or probably said, something along the lines that if he has to write 10,000 words, he can have it on your desk by tomorrow. Bu if it’s 1,000, he’ll need a week. Something like that.