How do you mark HSC English essays faster without dropping the standard?

What actually slows down HSC English marking?

The reading is not the bottleneck. A Paper 2 essay takes six or seven minutes to read properly. What takes the rest of the evening is re-deciding, on every single script, what a Band 5 response actually looks like, and then writing a version of the same comment for the tenth time. Fix those two things and the pile moves.

The second cost is drift. Standards move over a long session. The essay marked at nine at night rarely gets the same reading as the one marked at ten in the morning, and three strong responses in a row will quietly raise the bar for the fourth. Drift is not a failure of skill. It is what happens when the standard lives only in your head.

So the useful question is not how to read faster. It is how to decide the standard once, write each comment once, and hold both steady to the end of the pile. Everything that follows is a way of doing that, and none of it depends on any particular software.

How do you turn band descriptors into criteria you can mark against?

Start with the official material. NESA publishes past HSC exam papers for Stage 6 courses with marking guidelines and feedback, and those are the closest thing a faculty has to a shared standard. But the language in them has to cover every text and every module, so it is written broadly on purpose. Sophisticated is a judgement. It is not something you can underline in a script.

Before you mark, rewrite the criteria for the task in front of you. Do it at the level of the question, not the course. A Texts and Human Experiences response and a Module B critical study reward different moves, so the criteria should differ too. Three or four criteria is enough. Twelve is a checklist nobody applies consistently, including the person who wrote it.

The test for each criterion is simple. Could a colleague who has never seen the task point to the place in a script where it is met or missed? If the answer is no, keep rewriting. Four criteria that survive that test will do more for consistency than any amount of rubric detail.

• Argument: does the response answer this question, or a nearby question the student preferred?
• Evidence: are quotations short, embedded in the sentence, and used to prove a claim rather than decorate one?
• Analysis: does the student say what a technique does, not only that it is there?
• Expression: can each sentence be read once and understood?

What marking routine keeps a standard steady across forty scripts?

Benchmark before you mark. Read six or seven scripts without writing a word or awarding a mark, then sort them into rough piles: strong, middle, weak. Pick one script from each pile and keep it on the desk. From then on the question is no longer whether this is a Band 5, which is hard, but whether it is stronger or weaker than that one, which is fast and far more reliable.

Mark in blocks of eight to ten and then stop. At the end of the pile, go back and re-read your first five scripts against the anchor from the middle. That is where drift shows up, and it is a ten minute check that stops one class being marked to two different standards.

Flag boundary scripts instead of agonising over them in the moment. Put a dot in the corner, keep moving, and decide them together at the end when the full range is in front of you. Borderline calls made in isolation at eleven at night are the ones that come back at moderation.

Some markers work vertically, taking one criterion across the whole pile before starting the next. It holds a single standard in working memory at a time and it is quick. The cost is that you lose the sense of the response as a whole, so it suits an analytic rubric better than a holistic one.

How do you write essay feedback a student can act on?

Tie every comment to a criterion and to a spot on the page. Needs more analysis tells a student nothing they can do on Monday. Your paragraph on the storm names the metaphor and stops, so add one sentence on what it changes about how we read the narrator, tells them exactly what to do next. Feedback earns its time when it is tied to explicit criteria and points at an action.

Cap the volume. Two things that worked and one thing to change is a realistic load for a student. Fourteen margin notes produce no change at all, and they cost the most to write. The habits behind expert essay marking are mostly habits of reuse: the comment you wrote about embedded quotations in March is the comment you will need again in August.

So write each one once. Keep a comment bank, in a document or in whatever marking tool you use, and put short codes in the margin with a key on the front page. Students learn the codes within a term. A comment that also sends a student back to their own notes or to the study guides they already use for that module is worth more than a longer comment read once and forgotten.

Say the mark and the reason in the same breath. A band with no criterion attached invites an argument about the number. A band attached to two criteria the student can see in their own script invites a conversation about the work, which is the one worth having.

How does a faculty keep marking consistent between teachers?

Within one teacher’s pile, anchors and a routine are usually enough. Across four teachers marking a common task, they are not. The cheapest fix is cross-marking a sample. Everyone marks the same three scripts blind before the real marking starts, and the faculty meets for twenty minutes to argue about the two they disagreed on.

Run that meeting on scripts, not on numbers. Comparing spreadsheets tells you there is a gap. Comparing two responses that one teacher placed at the top of Band 4 and another at the bottom of Band 5 tells you why, and it is the only conversation that actually moves anyone’s marking.

Then keep the artefacts. The rewritten criteria, the anchor scripts and the comment bank belong with the task, not in one teacher’s drive. Next year the setup cost is already paid, new staff inherit the faculty’s standard instead of guessing at it, and the same assessment gets marked the same way twice.

Where does an AI marking assistant fit into this?

One section on software, because the question comes up and the honest answer is narrow. An AI marking tool does not know the NSW syllabus. Jeddle’s engine, JeddAI, drafts against the rubric, criteria and comment bank you load for the task, which means the alignment you get is the alignment you wrote. Load vague criteria and the drafts come back vague.

What a tool like this can take on is the repetitive part described above. It applies the same criteria to the fortieth script as to the first, and it produces a first draft of the comments so you are editing rather than composing. You still read the essay, you still set the mark, and you still rewrite anything that misses. Nothing about a drafted comment makes it true.

What it should not touch is the judgement. Borderline band placements, an original argument that does not fit the rubric, questions of authenticity, and feedback shaped by what you know about a particular student are all yours. Used that way, JeddAI removes the typing rather than the marking, and the standard on the page is still the one the faculty agreed on.

• Worth delegating: first-draft comments, applying the same criteria across a full class set, flagging where a criterion is not evidenced.
• Not worth delegating: borderline band decisions, unconventional arguments, authenticity concerns, and feedback that depends on knowing the student.