On 15 October, the Education Secretary stood up at the CST conference and said: "I will give oracy equal status to reading and writing, as a new strand of the English curriculum."
That's a big promise. It also raises an awkward problem for every department. Writing stays on the page. You can re-read it, second-mark it and moderate it in June. Talk is gone the moment it's said.
So before anyone tells you how oracy will be assessed, it's worth agreeing what good talk actually sounds like in your department.
What Powell actually announced
Lucy Powell's delivered speech went further than many of us expected:
"Both in content and in assessment, speaking, listening and communicating will be a core part of every child's education, from primary to GCSE and beyond."
And on how the change will be made:
"Working with you, working with Ofqual, and as part of our curriculum review, oracy will light up every classroom."
Compare that with November 2025. In its response to the Curriculum and Assessment Review, the government promised an oracy framework, guidance for teachers on how to formatively assess oracy, and said it would "consider the role of non-statutory assessment". That was cautious, and it was formative.
Assessing speaking all the way up to GCSE is a different order of ambition.
What it doesn't settle yet
What we don't have is just as important as what we do:
- No specification.
- No timeline.
- No detail on which GCSEs, or how talk would count.
- No Ofqual consultation or statement yet.
The baseline today is the spoken language endorsement in GCSE English Language. As Tes noted in its coverage, it's reported separately (Pass, Merit or Distinction) and doesn't count towards the grade. We won't guess what replaces it. Nobody outside the DfE and Ofqual knows yet, and anyone who tells you otherwise is selling something.
Why talk is the hardest thing to mark consistently
If you've ever moderated extended writing, you know how much disagreement there is between good teachers looking at the same script. Now take the script away.
With talk:
- It's gone once it's said. There's nothing to re-read and no second marker after the event.
- It's judged live. You're listening, managing the room and deciding on a mark at the same time.
- It's open to bias. Confident, fluent pupils sound better than they sometimes are. Quieter pupils, pupils with EAL and pupils who think before they speak can be underrated.
Powell said it herself: "for too long it's been a marker of privilege." If assessment rewards the pupils who already sound most at ease, it will widen that gap rather than close it. Consistency isn't an admin issue here. It's an equity issue.
Four things that hold whatever Ofqual decides
You don't need a specification to do any of these. And none of them will be wasted, whatever the final rules look like.
1. Shared success criteria, by subject
Good talk in science isn't the same as good talk in history. In science it might mean using precise vocabulary and explaining cause and effect step by step. In history it might mean weighing evidence and responding to someone else's argument. Write down three or four criteria per subject, in pupil language, and agree them as a department.
2. A bank of recorded samples
You already use exemplar scripts. Do the same with talk. Record a handful of short clips at different levels (with the right consent in place) and keep them as anchors. When a new teacher asks "what does a strong answer sound like?", you can play them one.
3. Standardise it the way you moderate writing
Take one clip into a department meeting. Everyone marks it blind against your criteria, then compare. The gaps you find are the conversation you need. Do it once a half term and your judgements will start to converge, the same way they do with extended writing.
4. Keep KS3 low-stakes and formative
Until the rules exist, don't build a high-stakes oracy grade into Year 7–9 reporting. Use talk to give feedback and to practise, not to rank. You can tighten it once you know what's being asked.
None of this is new pedagogy. It's the same discipline good departments already apply to writing, pointed at talk. If you're writing a wider policy on AI and assessment at the same time, our guide to AI policy for schools covers the "policy before tools" questions.
Where we're going
Excelas Learn marks written scripts today. We're building audio marking.
The idea is simple: upload a recording of pupils talking, have it scored against your own rubric, and a teacher reviews and owns the mark. The point isn't to take the judgement away from you. It's to give you a consistent first pass and a record you can go back to, which is exactly what talk has always lacked.
Our principle doesn't change for audio: AI supports the teacher's marking, it never replaces it, and a teacher reviews and owns every final mark. (We've written more on why that matters in keeping humans in the loop.)
If you'd like early access to upload a recording, have it scored against your rubric and own the final mark, book a demo and tell us you're interested in audio.