Dictating Phonics Testing

 I am marking Year 8 end of unit tests. I wrote in a previous post about how we have introduced new tests. We wanted to make sure that our tests really did test pupils' knowledge of what they had been taught. And not just label pupils with good literacy skills as "good at languages". And label pupils with weaker literacy as "bad at languages". We also made some changes to focus on phonics and the knowledge that underpins the skills of Listening.

Here is Question 1 from the Year 8 Unit 2 test.

You will see that we are testing careful listening to similar sounds. It isn't exactly NCELP style minimal pairs (vin / vend) and it does have the words in context. Because this is a listening exercise that ultimately is about understanding meaning.

Interestingly, lots of pupils got question d. wrong. In fact I witnessed the gasps of delighted recognition as pupils happily and wrongly circled "je préfère". Familiarity, creation of meaning, making it all make sense, had taken over from carefully listening to what they had actually heard.

For me, this wasn't a flaw in the test. I was ultimately looking to test accurate understanding of meaning, by offering alternatives which were both possible and where the difference in sound was important precisely because it resulted in a different meaning.

The interesting thing is that following the philosophy of the Ofsted Research Review, this error shouldn't have happened. According to the Ofsted Research Review, when learners read or listen, they arrive at meaning by processing sounds and parsing known words in order to arrive at meaning. The idea that we hold meaning and form in a constant tension is alien to them.

In fact of course, it is impossible to listen to French and process it by turning sounds into words in a one way street from form to meaning. You can't tell if someone has said porc or port by processing the sound and then building the sentence. You have to keep both possibilities in mind as you make sense of the whole sentence. Processing the word as porc or port is not the first step. And it's not the last step either. It's a constant feedback circle between what the sound is and what the meaning is. That's where the Ofsted view falls down.

You can't tell if someone has said ça me dit or samedi even if you have processed the sentence around it. You have to take into account the whole interaction and maybe even ask for clarification before you turn up to a date on the wrong day!

You can't tell, in spoken English, if someone is asking the way to the Eiffel Tower or the weight of the Eiffel Tower without context.

So what is best for the new GCSE phonics dictation test? Isolated words? Known words? Unknown words? Made up words?

I think the guidance is that the dictation should contain genuine French words but which are not all on the GCSE vocabulary list. Because with known words, it wouldn't be testing their ability to transcribe the sound-spelling clusters. So probably words unknown to the pupils.

If, like Ofsted, the new GCSE panel believe that there is no back channel from meaning to processing of sound into words, then this might help pupils focus purely on the sound. But if, as actually happens, pupils have a strong conscious and unconscious pull towards making the sound into something that makes sense, then they may well distort the test words into the closest word they do know. Our brains don't seem to like non-sense, and try very hard to find meaning in what we hear.

So the pupils will be hearing unknown words and writing them down. This isn't a dictation, it's a sort of phonetic transcription. Dictation means writing down what someone says so the words make sense and the sentence is grammatically correct. So the sentence context and meaning would determine the correct spelling of the word - aller allez allé allés allée allées (allais?) / vers ver verre vert vaire verres verts. So a dictation isn't a phonics test at all.

What happens when it's a transcription of the phonics of unknown words? And the pupil can't be penalised for correctly coming up with a spelling that isn't an actual word? Because the whole point was that they didn't know the word. Is there a French utterance that can't be rendered a multiplicity of ways?

Dans, dent, dend, dand, d'ans, d'en, d'an, d'en, dant, dents... And to the ver list we would have to add verrent, vairent as potential silent endings.

Of course, this isn't going to be put into practice by Ofsted researchers or the new GCSE panel of experts. This is going to be a headache for the exam boards. Whose main concern is that the test discriminates well as an instrument for grading pupils from 1 to 9. So the dictation on Higher Tier won't be a friendly phonics check. It will be a test where a Grade 6 candidate has to get half of it wrong. 

Let's hope this is all thoroughly tested and trialed so we can see what works and what doesn't work. Let's hope that schools very quickly have some clarity so we can know how it's going to work. And so we can make sure pupils are being tested on what they have been taught.