Tag Archives: minimal pairs

To read better, improve your pronunciation?

To read better, improve your pronunciation….

This statement jumped out at me recently, from an excellent post by Robin Walker on Pronunciation for YLs. He was reporting from a talk by Catherine Walter at IATEFL 2008, and reading this (perhaps surprising) statement sent me off to find her original research.

In her article Phonology in second language reading: not an optional extra, Walter questions the idea that the skill of reading is something which needs to be taught to second language learners. She argues that successful L1 readers already possess the cognitive skills needed to build a mental structure or representation, which is, in fact, what we mean by ‘comprehension.’ According to Walter, comprehension isn’t actually a linguistic skill, so it’s fruitless to talk about transferring it from L1 to L2.

So why are some learners competent at reading in L1, but struggle with reading in L2? Obviously, lack of linguistic knowledge plays a part. Fascinatingly, though, Walter cites a study (Robertson et al 2000)* which used MRI scanning to show that sentence comprehension and comprehension of a text as a whole take place on different sides of the brain.

Walter doesn’t mention this, but, of course, there has been a lot of speculation about left brain and right brain thinking. Interestingly, sentence level comprehension, or decoding, used the left frontal lobe (associated with analytical, logical processing) and text comprehension uses the right frontal lobe (associated with intuition).

So a learner may have effective (non-linguistic) comprehension skills, and even be able to decode L2 sentences, but still struggle. Why?

According to Walter, it may be to do with how we use our working memories. Part of the working memory is something called the phonological loop:

‘a short-term memory mechanism that stores information in phonological form and automatically rehearses that information by unconscious sub- vocalisation.’ (Walter 2008)

In other words, as we listen, we automatically ‘record’ the last two seconds of what we hear in the memory, like a little Dictaphone. It’s why we can repeat back what we have just heard, even if we weren’t really listening properly to the speaker.

There is also evidence that we do exactly the same thing as we read- that we also sub-vocalise and record the sound of what we are reading. We don’t see it, we hear it.  At least, this is true for those of us with L1s that are alphabetic (there is some evidence that learners with non-alphabetic L1s, may use more visual representations).

However, if our phonological representations of what we have just read are unreliable, we may find it difficult to associate these sounds with meanings, and thus find it difficult to keep meaning in our short term memory. This, in turn, will make it harder for us to carry out meaning building processes on the text as a whole.

Walter’s research in this paper concludes that unless learners are also poor at comprehension in L1, we would be better off teaching them to improve how they ‘mentally represent spoken language’ than teaching comprehension skills.

Some suggestions:

As much exposure as possible to the spoken language- so lots of listening and watching videos in English.

Listening while reading - hearing the spoken version while reading a text. This could mean using sub-titles, or following a transcript while listening, or listening to an audio version of a written text while reading. I would suggest that these don’t have to be instead of our usual listening or reading activities (I’m not quite ready to throw out more traditional reading and listening procedures), but as a follow up.

And finally, explicit focus on features of pronunciation, such as minimal pairs work , and focus on word stress and on how words change in the stream of speech . This kind of work will help learners to develop a more reliable repertoire of L2 sounds, which, Walter suggests, could also help them to hold what they are reading more efficiently in their short term memory, which in turn will greatly help with building up the meaning of the text.

And even if this isn’t the case, it will certainly help with developing fluency in speaking and confidence in listening, so what have we got to lose?


 * Robertson, D. A., Gernsbacher, M. A., Guidotti, S. J., Robertson, R. R. W., Irwin, W.,Mock, B. J., et al. (2000). Functional neuroanatomy of the cognitive process of mapping during discourse comprehension.

Psychological Science, 11, 255–260

 

 

8 Comments

Filed under Different ways to use a coursebook, Pronunciation, Teaching methodology

How we can use minimal pairs to help L2 listeners to decode what they hear.

At the weekend I was lucky enough to catch Sam Shephard’s lively session on pronunciation at the NATECLA conference in Liverpool. His session focused mostly on productive pronunciation, but as I was presenting on the same day on decoding skills for listening, I found myself thinking more about the role of pronunciation work in decoding- and specifically about minimal pairs.

When I first saw this advert for Berlitz language schools on youtube, I was struck with how clever it is.

But, apart from in this rather specific context, how important is it really that learners can understand or pronounce the difference between //θ/  / and /s /?

Minimal pairs, minimal importance?

It seems that misunderstandings in natural speech are rarely caused by the mispronunciation of one sound. Usually context gives us enough of a clue to understand what the speaker is trying to say. Adam Brown gives a good example in his 1995 article, Minimal pairs, minimal importance?:

‘Singapore is one of the busiest ports in the world. However, it is a tiny island (the size of the Isle of Man) with a population of three million. Consequently, land is at a premium, and there are no animal farms. The nearest most Singaporeans come to sheep is mutton curry. In short, if Singaporeans don’t pronounce the distinction between ship and sheep clearly, the chances of misunderstanding are minimal: they are almost certain to mean ship.’

Similarly, Jenkins (2000) found that /θ/ rarely caused misunderstandings between NNSs, and she also points out that many native speaker varieties don’t use it anyway, often using /t/ or /f/.

So should we chuck out the minimal pairs work?

Can minimal pairs help L2 listeners decode more effectively?

Well, according to John Field (2008) there is evidence that L2 listeners process in words, but that ‘many of the matches they make are rough approximations that do not correspond exactly to the sounds that the listener heard.’ In other words an inability to recognise certain phonemes is leading to learners making inaccurate guesses about words, which in turn could lead them quite seriously off track as they apply top down skills to their guesses. For example, the listener who hears ‘screams’ instead of ‘screens’ is likely to go quite a way off track.

It is certainly true that context could help here- but that is making the assumption that learners are able to use their top down skills effectively when, Field and others argue, learners who are unable to decode effectively, usually can’t hold onto enough meaning to start stringing ideas together.

So, therefore, there is certainly an argument for using some minimal pair work, especially at lower levels- though we probably do need to be quite selective about which phonemes we choose to focus on.

Functional load

Sounds that carry a high functional load are used to distinguish between a significant number of words. The opposite is sounds which carry a low functional load. For example, Brown (1995 above) says that the only minimal pairs in English for /ʃ/ and /ʒ/ are:

Confusion/Confucian, azure/Asher, mesher/measure , allusion/Aleutian, glacier/glazier

I think we can probably all agree that the likelihood of students even using most of these words is pretty low, never mind getting them confused!

Minimal pairs with a high functional load would include:

/e/ and /ae/ , /p/ and /b/, /e/ and /I/

A full list is available in Brown 1988

Individual needs

With a monolingual group, it should be fairly straightforward to find out which pairs are causing the most problems. A book like Learner English can be helpful, or simple observation. Obviously with a mixed nationality group, tricky minimal pairs are likely to vary, but there are some which seem to be difficult for speakers of many different languages, and have a high functional load, such as /e/ and /ae/ and /ae/ and /ʌ/

Activity ideas

The first point to make is that ideally, learners should be able to see the link between the minimal pair work and what they are listening to. For example, if a number of learners have heard ‘scream’ instead of ‘screen’, that would be a perfect opportunity to do some minimal pair work on /m/ and /n/.

The second point is not to overload the learners. I wouldn’t suggest working on more than one pair of sounds at a time.

The third point (made by Field) is that ideally words used should be relatively frequent and of roughly equal frequency. So bin and pin would be OK, but perhaps not blade and played.

There are lots of ideas for working on minimal pairs (some of which came up in Sam’s session, mentioned above)

Some different ways for learners to show they can differentiate the two sounds:

  • Put the two words in each pair on different sides of the board and learners  put up their left hand/right hand according to which they think they hear.
  • Alternatively, learners can physically move to the right or left side of the classroom.
  • Put the words on cards and learners grab the right card, either in small groups with little cards, or with big (sturdy) cards, you can haver learners line up so one from each team is in front of the board and they race to grab the right word from there.
  • For a more sedate activity, learners write down what they think they hear.
  • Learners say if the words you say are the same or different.

Obviously all the activities above can be done with a learner providing a model, but then it becomes oral work, rather than listening, and they will need help to know how to make the sounds etc.

If learners have literacy issues, the above activities could potentially be done with pictures rather than words:

 

 

And if one of the words in the pair you want to use is not very frequent (e.g. played/blade), you could still do the activity but just write the frequent word on the board and ask ‘Same or Different?’

A more contextualised task, which would make the relationship to listening clearer, might be to select a phrase or short section from something they have listened to which contains a lot of the two sounds (not necessarily in minimal pairs) and ask them to mark the two phonemes.

E.g. ‘Looking after rabbits is really easy’ might work well for /r/ and /l/.

Individualised work

Clearly working on minimal pairs is much trickier with a multi-lingual class. As mentioned earlier, there are some vowel sounds which a lot of people find tricky. Alternatively,  learners could be given different sounds to work on, according to needs. There are now quite a few websites  (for example http://www.shiporsheep.com) where learners can listen to minimal pairs, so this kind of differentiated activity could be set as homework.

11 Comments

Filed under Differentiation, Exploiting authentic recordings and videos, Pronunciation