A fundamental rule of assessment is that there must be clear agreement about what is being assessed which is why “oracy” has so many problems. Even if you specify elements like “presentation skill, dialogic reasoning, confidence, listening,turn-taking, ability to disagree politely” etc., the problem is that these are all very different constructs. Vocabulary is knowledge, presentation is a performance and confidence is personal trait that is questionable as something you can even teach. Can they all be grouped together and assessed by individual teachers as a “level 7”? Or a “grade B”? These elements don’t all travel together or map onto one thing either: a confident speaker can have a very weak argument and vice versa. It’s not one thing. Do you grade each separately and average them? So a pupil scoring 9 for confidence and 3 for reasoning gets the same 6 in oracy as a pupil scoring 3 for confidence and 9 in reasoning? And “Dialogic reasoning” depends almost entirely on what the pupil knows about the topic under discussion: nobody reasons well aloud about something they understand poorly, so a reasoning score is basically a knowledge score without the knowledge specified. Another one is “Listening” which is essentially unobservable; we can only really infer it from what someone says next after the assumed listening. How do you measure that? And if a pupil responds well to what was just said, was that listening? Or was it knowing enough about the topic to have something to say, or being confident enough to say it, or being quick enough to get in before someone else? To get a speaking assessment anywhere near the reliability of a written one, you’d need many different tasks, each rated by trained assessors. Even if that was agreed, the workload would inevitably fall on overworked teachers. We can all agree dialogic talk is a very laudable end but it’s largely a philosophical disposition, but assessment as Dylan points out is something very different. It’s a question of precision, of measurement, of validity and reliability which doesn’t care about how philosophically worthy something is.