Marking the Conversation at GCSE (AQA) -- Not funny

 There are two meanings to the word "a joke". One is something deliberately risible, to make those in the know laugh. The other meaning is that something is an object of derision. Sometimes both meanings coincide. As in "this definition of Good Development" is a joke.

This is from the new specification for GCSE in French, German and Spanish from AQA. And anyone who knows anything about teaching languages will spot the joke. I like to eat carrots because it is interesting is the kind of desperate answer we don't accept from pupils. Our teaching consists of trying to move them on from this kind of answer. There are teachers who ban it is interesting because it's seen as the last resort of pupils who can think of nothing to say and have turned up to an exam utterly unprepared.

Yet here we have "I don't like social media because it's boring" as the very definition of "Good Development."

What is it trying to signal? Two clauses is what is considered a well developed answer. And three clauses counts as an extended response. It is signalling that long pre-learned answers are not required in order to perform well. Its message is all about what is NOT wanted, rather than thinking through what might be required.

In exactly the same way, it is signalling that deliberately fancy expressions thrown in to wow the examiner, are not wanted. In terms of amount of information, it is boring is no worse than a pre-learned autant que je sache. Unfortunately it also means it is boring is just as good as a thoughtful because it takes up too much of my time.

We are looking at very complex arguments being played out, about the nature of the level of difficulty in languages. There are lots more ironic jokes at play here. Like the fact that the autant que je sache was a favourite of teachers most strongly associated with the "Knowledge Curriculum". Supposedly to show how well their pupils could do if taught "properly". When all they were doing was showing that the supposed hierarchy of difficulty is bogus. Just as using je vais used to be a whole National Curriculum level higher than je dois because je vais is the future.

Revenons à nos moutons:

This really should have had no place in the Specification. It too clearly smacks of in-jokes and point scoring in the spat between exam boards and the GCSE panel in the creation of the new exam. It is too focused on what is NOT wanted (rote answers and fancy expressions), rather than thinking through what IS wanted. As such, it could have been guidance on the conduct of the exam rather than the marking criteria. In fact, it is strongly emphasised everywhere that having a narrow list of questions that all pupils are going to be asked, is malpractice.

Given the parameters of the exam, what might be wanted? Firstly, it was advertised as a conversation of between 4 and a half and 5 and a half minutes. On just one theme. So twice as long as the previous GCSE, which had a similar length conversation, but on two themes. Clearly, I don't like social media because it is boring is not going to see you through 5 minutes. Five minutes of such short answers would require about 30 questions in a relentless back and forth. I don't think I have 30 questions on most of the topics in these themes. And I don't think pupils have 30 variations on I don't like... because it is... So clearly, while the exemplars in the specification served their tangential purpose of sending a strong message as to what was NOT wanted, we had to figure out for ourselves what was wanted.

And it seemed reasonable to think that if we teach pupils to develop their answers spontaneously, and to respond to prompts from the teacher which would interrupt any pre-planned answer, then this would be rewarded.

The idea that a pupil who extended their answers spontaneously would be penalised is ridiculous. Or that a teacher who interjects Why? For example? And so? And then? would be penalising their pupils, is also ridiculous.

That is what happened with this week's guidance, now apparently hastily withdrawn. Although I have yet to find anything official from AQA either presenting the guidance or withdrawing it.

A pupil who extended their answers spontaneously, would not necessarily reach the 17 questions total. A teacher who interrupts to prompt or redirect a pupil, pushing them to spontaneously develop an answer, would fragment the 3 clauses into a series of "minimal" answers.

Thank goodness AQA did publish the guidance. Imagine if it was being marked this way. And pupils who spoke and interacted spontaneously were marked down for answering fewer questions or for having fragmented clauses responding to the examiner's interjections. And we wouldn't know why it was happening.

This is the key thing that AQA missed. They think they have to quantify "amount of information." And they think it's only fair to publish it. What they fail to realise is that this then determines what answers we have to train pupils to give. Instead of evaluating what pupils say, the exam board are determining what they have to say. We have to train them to answer 17 questions with 3 clauses (some may fall short of 3 clauses). And because everyone will be doing this, it throws the emphasis onto the other criteria: accuracy and variety.

17 answers carefully box ticked, carefully accurate, deliberately including variety. This is recreating the exact conditions for fancy rote-learned pre-prepared answers. The very thing they were trying to get away from.

NOT Funny.

Part 3 of A Spanner in the Works. AQA Guidance for Marking the Conversation.

I don't know where to find this information officially from AQA, but I am hearing there is going to be a re-think. I don't know what will replace it. Hopefully a 5 minute conversation as specified in the spec. Note also that the "I don't like social media because it is boring" exemplification of "good development" is in the spec. I think the problem lies in the way the guidance set hoops to jump through that were going to determine/distort/limit pupil performance instead of assessing their performance. Same goes for the photocard guidance.

 Now we know that pupils will NOT have to speak for 5 minutes on one theme in the Conversation, what should their answers look like?

We have to interrogate the exemplars from the specification. They are likely to creak under this exercise, as they were originally intended to be examples. But now they are being forced into the role of definitions of "extended" and "good".

Here they are:

You can see that they have been chosen to exemplify that pupils will NOT need to have extended answers or fancy language. The choices are deliberately, even knowingly at the level which previously we have aimed to move pupils away from. Now "I don't like social media because it is boring"  is the definition of Good development of an answer. The GCSE panel wanted to remove the Conversation because it led to rote delivery of long answers containing fancy language. The exam boards put the Conversation back in, and are signalling that it will not reward long memorised answers and deliberately inserted fancy language.

Of course, this also avoids rewarding pupils who can spontaneously develop answers and naturally use sophisticated language as part of a complex narration.

Looking closer, we can see that "amount of information" is being interpreted in a weird grammatical way. The exemplar for "extended response" includes three clauses.

So I think we are to assume that an answer delivering more information, but all in one clause, would not count as extended.

I love to go to the cinema in Norwich with my friends or family but not on my own to see an action film or another good film most weekends in a cinema with a big screen and a great sound system.

This example only has one conjugated verb I love... And although it contains a greater "amount of information" than the exemplar, we would have to count it as "minimal development". "Minimal development" of the "amount of information".

So verbs are crucial. Not the "amount of information".

What about the fact that the exemplar for "extended response" contains three different verbs. This is all we have to go on. So are we to assume this is also a requirement? What if I repeat the same verb?

I love to go to the cinema and I love to go to Norwich. I love to go with my friends or my family, but not on my own. I love to see an action film but I also love other sorts of film and I love to go most weekends to a cinema with a big screen and I love a cinema with a great sound system.

Is that now an "extended response"? Or is it disqualified because it is the same as the earlier "minimal response" with the verb repeated?

This makes a difference. One of these would be "good development" and the other one would be "an extended response"? Or not?

I like to play tennis because it is fun and exciting.

I like to play tennis because it is fun and it is exciting.

And the overriding question remains. Is "I go to the cinema and I watch films. I love films" really what is required for a grade 9? If so, we have got an awful lot of thinking to do about what we are teaching.

Of course, this exemplification was there all along, and isn't changed by the new 17 question guidance.

What is changed, is the dropping of the requirement to talk for between 4 and a half and five and a half minutes on just one theme. This has been replaced by the requirement to give short simple accurate answers with 3 verbs for 17 questions (some of them can fall short of 3 clauses). 

What also has changed, is that everyone will make sure that pupils can tick this box, so the Conversation is now the equivalent of Controlled Assessment. Planned and prepared against a tickbox that everyone meets, so effectively irrelevant in its effect on the grade. And remember, AQA have already done the same thing to the Photo Card. We are right back in the bad old days of 2016 and the Baukham report, with the wrong answer to the wrong problem.

This is exactly what this new exam was meant to avoid. And exactly what I feared it would do right from the start. An exam explicitly designed to change the way we teach. Ends up ruining language teaching again.

Part 2 of A Spanner in the Works. The AQA guidance on marking the Conversation.

I don't know where to find this information officially from AQA, but I am hearing there is going to be a re-think. I don't know what will replace it. Hopefully a 5 minute conversation as specified in the spec. Note also that the "I don't like social media because it is boring" exemplification of "good development" is in the spec. I think the problem lies in the way the guidance set hoops to jump through that were going to determine/distort/limit pupil performance instead of assessing their performance. Same goes for the photocard guidance.

 This is going to make a lot more sense if you have read Part 1 of how AQA have thrown a Spanner in the Works for how my department teach the Conversation part of the Speaking Exam.

A huge amount of thinking, collaboration and planning has gone into teaching this new GCSE, and in particular, the new Speaking Exam. Our KS3 is designed to teach pupils how to use a growing repertoire of language across topics, with an emphasis on not just learning more language, but on learning how to use it. Pupils work on thinking up what to say, how to make it personal, coherent, interesting and developed.

We start Year 10 with Module 0, showing them how their KS3 French already enables them to tackle the role play, unexpected questions and some conversation questions. In Year 10, we build up language, carefully transferring it across topics, and making sure pupils see how they can deploy it in the exam. I feel we are doing our best to put in place best practice, in dialogue with the ideas behind the new GCSE.

Last year we had the opportunity for Year 10 to do a Speaking Exam. Rather than an exam, it was more of a run-through, to familiarise them and us with the elements and demands of the exam. They had the Role Play, Read Aloud and Photo Cards in advance, so they could turn up and do the Exam in 10 minutes without the need for invigilators or prep time.

What did we discover? Not to be afraid of the exam. The Role Play - short answers containing a verb. The Read Aloud - stunning. The Unexpected Questions - a bit of explaining that you have to guess what you think the question is, say something related, then say a couple more random things that might be related. The Photo Card - say there is or is 8 times for each picture, without risking trying to say anything else. (Post here on the negative effect of the AQA marking guidance on the Photo Card.)

That left the Conversation. We had NOT prepared answers to a list of questions. But pupils knew they would get questions that they could answer using their repertoire of opinions, reasons and tenses. They knew that we would prompt them for more using and, so, for example, why...? We didn't stick to one theme, but used it as an opportunity for them to show off their French across all different topics.

So what we discovered was that their French was up to the task. But the demands (at the end of all the other tasks) and the cognitive load of thinking up what to say and how to say it in French, was too much. After a while their answers ran out of ideas and became repetitive, or we had to switch topics to keep them going. Or we said "Well done" and stopped before the full five and a half minutes.

This then, was our focus for going into Year 11. Tweaks to the Scheme of Work. The Department Plan. Inset in September and department meetings. Individual teachers' Performance Management Targets. All in a coherent focus on managing the balance between having ideas prepared, but not memorising answers. Managing the balance between preparation of ideas, and spontaneous improvisation of the French. So that the pupils could talk for 5 minutes on just one theme (double the time compared to the previous GCSE) without having memorised answers. What is the best way to teach pupils to talk for 5 minutes? Prepared answers is not the best way. Because the more prepared the answer, the quicker it is to deliver and you end up having to learn ridiculous amounts to fill the time. Better to have a genuine conversation, with some ideas prepared, but making up the answer in response to the examiner's questions. A balance of prepared ideas, but spontaneous French.

So we do have a booklet of possible starter questions for the conversation. And pupils are challenged to answer the questions spontaneously in speech and to plan their ideas in writing. They do not memorise their answers, but they do have their ideas ready. We have been careful to mix the questions up across topics so the pupils are deploying the same repertoire irrespective of theme, and there is definitely no set list or order of questions. When we come back to practising questions, we don't let them look at their planned answers - they have to improvise a new answer based on the ideas they had come up with, just as we did with the previous GCSE. We work on creating answers in layers. So they can give an immediate response. Then back it up with reasons or if sentences or examples in past or future. They know that the teacher will prompt for this kind of extended detail with follow-up questions such as et... ?  alors...? par exemple...? Pourquoi ? The conjunctions dice game in the second half of this post has featured heavily in getting them to extend answers and respond to being pushed in different directions by the throw of the dice. We have worked hard on the different directions a story can go in, with one idea leading to another, so you don't get stuck thinking up what to say next, as in the mouse and the cookie. We have even turned the order of Year 9 units around, to start with developing ideas into stories. 

A huge collaborative and joined-up effort of the entire department, based on taking stock from the Y10 Speaking, and gearing up for the mock speaking next month.

Then AQA put out their guidance and it's hard not to feel as if the rug has been pulled out from under our feet.

It's not the 17 questions. They were always going to have to define "amount of information." Although defining it means everyone will make sure they meet it. With a planned and monitored set of questions. And because everyone meets it, the emphasis that swings the grade will fall on the other criteria: accuracy. And it's not the fact that AQA "extended answers" mean very short basic answers. Although a requirement for a set number of 3 clause accurate answers is perhaps best met through planned, prepared, rote learned answers. And it's not even the fact that redirecting prompts like and... so... for example... why?  would now invalidate the pupils' responses, by breaking up the 3 clauses.

Well, yes. It is all that. But the main thing is the removal of the timings. If you no longer have to fill 5 minutes (remember in the old GCSE, there were minimum times on each theme), then you no longer need to have a repertoire you can riff on confidently and indefinitely.

It's not hopeless. We just have to adapt. First we have to audit our questions for each theme. Are there 17 questions? If we include short prompts as further questions, can we do this without disqualifying the pupils' answer from reaching 3 clauses? How many more questions do we need so as to avoid repeating the same questions? How do we allow pupils to show what they can do in terms of inventive longer answers, but still get through 17 questions? Is the AQA exemplar answer "I go to the cinema and we watch action films. I love action films" really going to get a grade 9? If what our pupils can do is superfluous to requirements, what elements of the exam should we have been focusing on? And if the thing that is going to swing the exam is now the accuracy marks, does this mean our pupils now should learn scripted answers off by heart?

Part 1 of: A Spanner in the Works? The new AQA guidance for marking the Conversation.

I don't know where to find this information officially from AQA, but I am hearing there is going to be a re-think. I don't know what will replace it. Hopefully a 5 minute conversation as specified in the spec. Note also that the "I don't like social media because it is boring" exemplification of "good development" is in the spec. I think the problem lies in the way the guidance set hoops to jump through that were going to determine/distort/limit pupil performance instead of assessing their performance. Same goes for the photocard guidance.

I don't know where to start with this post. Or where it is going. But I think I know what the key problem is. And it's NOT the 17 questions. Do I know what to do about it? I'm working it out. But it may take more than one post...

This is going to be big. It wasn't supposed to be. It was supposed to quietly define "amount of information" in the Conversation part of the new AQA GCSE Speaking Exam. 

It was even supposed to disincentivise rote learning of scripted answers. I'm not sure how. Because even my immediate reaction to this is to check how many questions I have for each theme, and how many of them pupils would have extended answers for. And I live and breathe spontaneous answers in my teaching.

It's worth mentioning straight away that an AQA "extended answer" is not what we understand by an extended answer. For our Year 9s, working on extended answers means things like the examples below, moving from a random stream of French, to coherent answers, to past tense stories with cheats, to telling stories. These are written examples, but we spend much more time working on speaking and spontaneity, with strategies like Being Ben or telling stories round the class to develop pupils' ability to think what to say next.

No. For AQA, an extended answer looks like this:

Three clauses of particularly uninspiring language, containing an opinion and a conjugated verb. The example given for "Good Development" because it is boring seems a deliberately knowing and sarcastic inclusion. Because this is clearly a response to the initial attempt to do away with fancy pre-learned answers by the GCSE panel when they originally proposed getting rid of the Conversation completely.

Both the GCSE panel and the exam board in their different ways are trying to get rid of pre-learned scripted rote answers.

But I can't see how this is not going to mean a return to rote learned answers. The ticking off of a specified number of answers means teachers having to carefully plan and keep track. Everyone will be making sure they hit the magic number. This then means that what differentiates one pupil's performance from another will be the criteria for Accuracy. And the need to deliver a set number of highly accurate answers will lead to... rote learned answers.

Is the number of questions so prohibitively high that no-one would dream of learning that amount? 17 three clause answers for each of three themes. With lots of cross-over where a question could be used in more than one of the themes. This is prime "learn by rote" territory.

I actually don't think the 17 answers is the problem. They were always going to have to define "amount of information". And I already suspected that the reduction in topic content was going to shift the balance back to pre-learned answers.

The actual problem is the ditching of the times. Nominally, the Conversation is supposed to last between four and a half and five and a half minutes. A long time to talk on just one theme. The old GCSE Conversation was this long over two themes. So even though I knew that 3 clauses was all that was required for an "extended" answer, I had spotted that filling 5 minutes was going to need pupils to have more to say. Our pupils work on developing answers spontaneously, responding to teacher prompts such as et alors... ? par exemple... ? Pourquoi ? I will look at exactly where we are up to in terms of being able to riff on these prompts to fill 5 minutes in a later post. But all that may now have to go. Perhaps we were fooling ourselves all along that it was what was wanted.

And the five minutes also could have been a disincentive to learn and deliver pre-learned answers. A pre-learned answer, delivered fluently takes up less time than an improvised answer. Like Achilles chasing the tortoise, the more you fall back on pre-learned answers, the more you find you have to say.

So we were pleased to convince ourselves that improvised answers with the teacher intervening to prompt for more detail, was the best way to fill five minutes.

That's what's gone. It's not 5 minutes anymore. We're left with the requirement to give 17 short but accurate answers. How does this not tip the balance back towards having prepared answers?

Incidentally, this is exactly the same mistake that AQA have made with their interpretation of the specification markscheme for the photo card. With similarly negative consequences for teaching and learning, as I found in this post.

What about the idea of the examiner prompting the pupil for more detail, to push the pupils to extend and develop? Things like et alors... ? pourquoi ? par exemple...? all count as questions, so would make it easy to get to the 17 number. But what they also do is fragment the "extended answer" into single clauses. So instead of demonstrating the pupil's ability to extend spontaneously, they now disqualify the answers from counting as "extended" as each response may now fail to meet the 3 clause threshold.

I have plenty more to say about exactly where we are and what to do next. But that's enough for now. It's NOT the 17 questions. It's the ditching of the 5 minutes. That changes everything.

I know the exam board had to define "amount of information" and don't want to see rote learned answers. They will have tried out how the marking works out on sample recordings of conversation. Have they done the opposite? Have they tried out what sort of conversation you get when you specify 17 short accurate answers? I hope they are right that this means we are still better off teaching pupils to extend their answers spontaneously. I'll explore that in another post...

Recent Posts on the Way Forward for Languages.

Here's selection of recent posts on the two systemic problems facing MFL, why they are so bad, and what we could do about it.

 The impact of unfair grading:

https://whoteacheslanguages.blogspot.com/2025/08/unfair-grading-and-its-impact-in.html

Could the success of the languages for all pilot offer hope for being able to offer mainstream language learning post 16? https://whoteacheslanguages.blogspot.com/2025/07/hope-is-in-air.html

The two things that need sorting to allow MFL to flourish. https://whoteacheslanguages.blogspot.com/2025/05/can-we-sort-out-languages-in-english.html

A Level Spanish 2025. How they make the exam too hard for even the tiny minority who take it. https://whoteacheslanguages.blogspot.com/2025/06/really-cool-translation-challenge.html

How bad is the reformed A Level? https://whoteacheslanguages.blogspot.com/2025/05/the-problem-with-this-level.html

And worth getting it straight from the horse's mouth. Look at the contempt for language learning in this submission from the people who "reformed" A Level: https://alevelcontent.wordpress.com/wp-content/uploads/2014/10/alcab-rationale-for-english-essay.pdf

The ugly reason why things are so bad for MFL post 16. https://whoteacheslanguages.blogspot.com/2023/10/colonial-curriculum.html

One easy thing to put right: https://whoteacheslanguages.blogspot.com/2025/01/one-thing-that-costs-nothing-which.html

Am I about to come unstuck? - How much can you rely on a metaphor for learning?

 How much can you rely on a metaphor for learning to guide your practice? All models of learning are metaphors. Starting with the popular "storage and retrieval" model. This seems a particularly circular metaphor, based on comparing the brain to computer memory, which in turn is a metaphor based on human memory. Metaphors for the brain often go hand in hand with current technology. This post looks at how previous models included cogs, hydraulics, cables... And of course in languages we have been presented with the metaphor of pillars which I examined in this post, showing how the metaphor revealed more than I expected: carefully constructed classically impressive pillars of free-standing stand-alone grammar, vocabulary and phonics, was deliberately an act of "folly".

You will know that my favourite metaphor for language learning is the snowball.

A few years ago, the day after a light snowfall, I was walking round the school with a pupil who had been sent out of his French class, to calm things down. He was telling me he didn't mind French lessons, but he just didn't know any French he could use. We stopped and I asked him where all the snow from yesterday had gone. He said, "It all melted, Sir." I asked him, "And where's your French?"

He was quick on the uptake (he is now a vicar, after a time in the police force), and said, "Oooh. Nice metaphor, Sir." He had been there in lessons while all the French was happening, but he hadn't managed to grab hold of any, roll it into a snowball, and stop it from melting.

This is the first use of the metaphor. To warn pupils that their French will melt. That it's their responsibility to grab hold of some and make it theirs. To roll it into a ball and stop it melting. And that more and more French will stick to it.

Then there's the message to teachers. We need to spend time making sure that pupils have a core of sticky French. That they are making it theirs and not letting it melt. It's important that our curriculum is designed so that we develop this core of language, using the same language over and over. And it's important not to design a syllabus where everything is ticked off once. The metaphor tells us that an even coverage of language will melt. What we want is a snowball of language that rolls on from topic to topic, getting bigger and bigger around a sticky core.

This post examines how to design a syllabus where new language adds a layer of accretion to the snowball. It starts from how to add new language to the pupils' existing language. Not by chopping up the language into bitesize chunks of omelette and hoping the pupils can make their own omelette out of it. Mixed metaphor alert. But cooking an omelette out of raw egg is the equivalent of the snowball approach. Chopping up the cold dead omelette is the equivalent of the even coverage approach.

So far so good. But how far does the snowball approach get me with the new GCSE?

For the speaking and writing, it's fundamental to our vision. We use it explicitly in our resources to show pupils how to tackle the demands of the speaking exam, whatever the topic. 

But this new GCSE has a huge gulf between the language needed for the Speaking and Writing exams and the vocabulary list for the Listening and Reading exams. The vocabulary list is not designed to be based on the language needed for the topics or for the tasks of the exam. For example when you get to the Jobs unit, there are fewer than 10 jobs words in the list. The topic of Jobs is just another arena to meet the non topic vocabulary. It's the even coverage idea reimagined. This time it's meant to be such intense snowfall that layer upon layer of French has fallen, before the previous layer has had time to melt. It risks leaving my pupils with their pathetic snowballs they were so proud of, lost in a snowy wastes landscape that stretches off to the horizon. Or at least that's what it's starting to feel like.

But is anyone achieving this deep layer upon layer of snow? Does it mean having to do listening and reading activities from the textbook totally by the book, missing out nothing because without the intensity of repeated snowfall, melting will happen? To achieve this, we would have to abandon the lessons focused on getting pupils better at using their snowball of French, all the lessons on practising speaking, thinking what to say next, getting really good at using their snowball. Do the textbooks actually deliver the meticulous coverage and re-coverage required for this permanent even coverage not to melt?

Anyway, for me and my Year 11s, who only started Spanish in Year 10, it's too late. The snowball approach is doing its job for the speaking and writing exams. Are we scared of the vocabulary for the listening and reading exams? What I am hoping is that having such a huge snowball will take care of it. Now their snowball is so big with all the things they can say or write, surely the little rocks, bits of grass, sticks, abandoned carrots and coal from other people's melted snowpersons... it can all stick to their snowball as they roll it round and round...? Can it?

This is the idea. That by making sure the pupils have their own snowball which has got bigger and bigger, more and more Spanish will stick to it. Including words that aren't nicely adding a natural layer, but which are odd words that don't seem to stick, but get swept up along with the snow.

Can I rely on this metaphor to get me and my pupils through the exam? Or will the whole thing come unstuck?