Marking AQA GCSE English Language Paper 2 with AI

If Paper 1 eats a weekend, Paper 2 quietly eats the one after it. AQA GCSE English Language Paper 2 pairs two unseen non-fiction texts — often from different centuries — with a reading section that runs from a short true/false question up to a demanding comparison of writers' attitudes, and then a Section B writing task that asks students to argue a viewpoint. Marking AQA GCSE English Language Paper 2 means holding two source texts, a student's response, and a mark scheme spanning AO1, AO2, AO3, AO4, AO5, and AO6 in your head at once, script after script. It is slow, exacting work, and a full class set is a day gone.

This post walks through how GradeOrbit handles Paper 2 specifically: getting handwritten scripts in without retyping, loading the AQA mark scheme and both source texts, reviewing the AI-suggested marks against the assessment objectives, and turning the output into feedback your students can actually act on.

Why Paper 2 Is So Slow to Mark

Paper 2's difficulty is the comparison. Question 2 asks students to summarise differences between the two sources; Question 4 asks them to compare how the writers convey their attitudes, weighing AO3 across both texts at once. To mark that fairly you have to keep both sources live in your mind and judge whether the student has genuinely compared, rather than described each text in turn. That is real cognitive load, and it does not ease as you work down the pile — if anything it wears you down faster than Paper 1's more mechanical opening questions.

Then comes Section B: a piece of viewpoint writing — a letter, article, or speech — marked on AO5 for content and organisation and AO6 for technical accuracy. After an hour of close comparative reading, switching to assess a persuasive argument on its own terms is another gear change. The paper keeps asking you to think differently, and it is that constant switching, more than the volume, that makes it exhausting to mark well.

Scanning Handwritten Responses With the QR Phone Upload

Step one is getting the scripts into GradeOrbit without retyping anything. Open the marking workflow on your laptop, scan the QR code with your phone, and the phone becomes a connected scanner over a direct WebRTC connection. Lay each script flat, photograph each page, and the images stream straight into the marking session. No app to install, no email-to-self loop, no queue at the staffroom scanner.

Before you submit, redact any names, candidate numbers, or identifying details by drawing black boxes directly on the images. The redactions are burnt into the image client-side using the Canvas API, so the AI never sees the original. GradeOrbit does not store student work at any point, and the redaction step adds a second layer of privacy you control yourself. For more on how the scanning workflow handles physical paper across other subjects, see how to mark physical exam papers faster with AI.

Loading the AQA Mark Scheme and Both Source Texts

Set the qualification to GCSE, the exam board to AQA, and the subject to English Language. Then upload the full AQA Paper 2 mark scheme as the grading criteria — or paste the level descriptors and assessment objective weightings directly. GradeOrbit reads the band descriptors and holds them as the static prompt context for every script in the batch.

Crucially for Paper 2, give the system both source texts. The AI cannot meaningfully assess the summary at Question 2 or the comparison at Question 4 without the material the student is working from. Upload both extracts once at the start of the batch and they stay in context for the whole class, so every comparison is judged against the same sources you gave the students.

Reviewing AI-Suggested Marks Against AO1-AO6

GradeOrbit returns a suggested mark per question, broken down by assessment objective where the mark scheme demands it. For Question 4 you will see how the model weighed the comparison across both texts, with the specific phrases it judged to be genuinely comparative rather than sequential description. For Section B you get a marks-based judgement split across AO5 and AO6. This is where you, the teacher, do the real work — agreeing, adjusting, or overriding each mark.

The point is not to accept the marks blindly. It is that you are no longer deciding from cold whether a Question 4 response sits in Level 3 or Level 4 with both sources open on your desk — you are reviewing a draft with the relevant student quotes already pulled out, and confirming the band or moving it. Most teachers find the time per script drops from fifteen to twenty minutes down to four to six once they trust the workflow. The marking gets faster; the marks themselves stay yours. Our guide on how to mark GCSE English Language papers faster covers the wider approach.

Turning Marks Into Student-Facing Feedback

Once you have agreed the marks, GradeOrbit generates categorised feedback per student: what they did well, what to focus on next, and concrete examples drawn from their own writing. You can edit any of it before it goes anywhere — nothing is published or shared automatically.

For Paper 2, the feedback students actually use is comparison-specific. "Your Question 4 answer analysed each source well but rarely linked them — you described the first writer's anger, then the second writer's calm, without comparing how each conveys their attitude" is far more useful than a generic level comment. GradeOrbit defaults to that specificity because it has both sources and the mark scheme in context. For the principles behind feedback that lands, see writing effective feedback for students.

Try GradeOrbit on Your Next Paper 2 Pile

The next time a class set of AQA Paper 2 mocks lands on your desk, try GradeOrbit on the first ten before you mark the rest by hand. Compare the suggested marks against your own, see how close they sit across the comparison questions and Section B, and decide whether the time saved is worth it for you. New accounts get free starter credits — enough to run a class through and form a real opinion.

Head to the GradeOrbit homepage to create an account and try it on a real script today.