How can NZ English teachers mark a class set of essays faster?

Where does the time actually go when you mark a class set of essays?

Ask an English department where the marking hours disappear and the answer is rarely the reading. A senior essay takes a few minutes to read and understand. What takes the time is writing a comment specific to that script, checking the judgement back against the standard, and then doing both again twenty-seven more times without drifting.

There is a second cost that is easy to miss. Marking a class set is a run of comparative decisions, and comparisons get harder as you tire. The script you read at nine on a Sunday night is being judged against a memory of the scripts you read on Thursday. Most of the inconsistency in a set comes from that, not from any real disagreement about the standard.

In New Zealand there is a third layer. Internally assessed NCEA standards are externally moderated by NZQA, so a judgement has to be defensible weeks after you made it. The reasoning behind a Merit rather than an Achieved needs to exist somewhere other than in your head.

• Reading and understanding each response.
• Writing feedback that is specific to that script.
• Applying the standard evenly from the first script to the last.
• Holding a consistent tone late in the pile.
• Recording marks and keeping evidence for moderation.

How do you turn an achievement standard into criteria you can mark against?

Standards are written to be precise, not to be marked from quickly. The first useful step is to rewrite the standard into four to six statements you can actually point at in a script. For an essay that usually means the clarity and control of the argument, the selection and integration of textual evidence, the analysis of the writer’s choices, the shape of the response, and the control of written expression.

Then write the difference between the grades in your own words, for this task and this text. In NCEA English the step from Achieved to Merit to Excellence generally turns on the depth and perception of the analysis rather than the amount of material. One sharp line per boundary, in the vocabulary of the actual assessment, is worth more than a page of generic descriptors, because it is what you will apply at speed on a Sunday night.

Hand the same sheet to students before they write. Criteria that arrive with the mark are a report. Criteria that arrive with the task are teaching, and they shorten your marking, because a student who has seen them produces fewer of the problems you would otherwise have to explain in the margin.

• Phrase each criterion as something visible in the script, not as a quality of the student.
• Keep the list short enough to hold in your head while reading.
• Write one line for each grade boundary as it applies to this task.
• Use the same wording in the task sheet, the marking grid and the feedback.

How do you keep the first script and the last one consistent?

Calibrate before you mark anything. Pull five or six scripts from across the range and read them without a pen. Place them provisionally, then say out loud what separates the top one from the one below it. That is your working boundary for this cohort. Ten minutes there saves the re-marking that happens when you realise on script nineteen that you have been generous all evening.

Mark in blocks and protect the ends of the set. Work in runs of eight to ten, take a real break, and re-read your first three scripts after you finish the last one. Keep a small pile for the scripts you cannot place and decide those together rather than one at a time, because boundary calls are comparative by nature and you make them better side by side.

Record the reason, not just the grade. One line per script naming the criterion that decided it is enough. It costs a few seconds while the script is still in front of you, and it is the thing that lets you answer a moderator, a parent or a student six weeks later without re-reading the essay.

What goes in a comment bank, and how do you build one?

A comment bank is not something software invented. It is the set of phrases you already write over and over, written down once. Mark six essays the way you normally would, then highlight anything you wrote more than twice. That is your first bank, and it is usually longer than teachers expect it to be.

Structure each entry so it does work. A comment that earns its place names what you noticed, ties it to a criterion, and gives one next step. “Your evidence is quoted but not analysed. After each quotation, add a sentence on the effect of the word choice” is reusable across a whole class and still specific enough to act on. “Good analysis” is not.

Teachers sometimes worry that banked comments make feedback impersonal. In practice the opposite happens. What distinguishes expert essay marking is not novelty of wording, it is accuracy and consistency of judgement, and a bank frees your attention for the scripts that need an original response. Tag each entry by criterion so you can find it, and cut anything students visibly ignore.

What kind of feedback do students actually act on?

Cap it. Two or three points per essay, one of them a clear next step, does more than a page of annotation. A script covered in comments tends to get read for the grade and then filed, and the effort you spent on the twelfth marginal note is effort you did not have for the next twelve students.

Separate the class problem from the individual problem. If a third of the set made the same mistake, that is a teaching point for Monday, not twenty-eight separate comments. Whole-class feedback built around two anonymised paragraphs on the board handles it faster and better than repeating yourself in the margins.

Give the feedback somewhere to go. A comment only matters if the student has a task that uses it: redrafting one paragraph, applying the next step to a different text, or working through study guides and exemplars for the standard before the next assessment. Feedback with no follow-up task is a mark with more words attached.

Where does an AI marking tool fit into this workflow?

Everything above works with a pen and a photocopied grid. A marking tool becomes worth considering once the criteria and the comment bank already exist, because those are what it needs in order to be useful. Feed any tool a vague rubric and you get vague feedback faster, which helps nobody.

What a tool changes is the order of the work. Instead of composing a comment for each script from a blank page, you get a first draft of feedback and a suggested grade for every student in the class, produced against the same criteria, and your job becomes reviewing and correcting. Reviewing is faster than writing, and review is where a teacher’s judgement does its best work anyway.

That is the shape of JeddAI, the marking engine inside Jeddle. It drafts feedback and a suggested mark against the rubric, success criteria and comment banks you supply, so if you built those from the achievement standard, the drafts reference the same qualities a moderator would look at. Nothing is finalised automatically. Every mark and every comment is a draft you review, edit or throw out before a student sees it.

Two cautions are worth holding on to. Drafted feedback is strongest on structure, evidence use and expression, and weakest on the things that need context: an unusual but valid reading of a text, a student whose progress matters more than their score, an argument that breaks the rubric in a good way. Those are the scripts to slow down on. It is also worth checking your school’s position on AI tools before you use one on assessed work.

If you want to test the idea, take one class and one task you genuinely have to mark. Enter the criteria you already use, add ten or fifteen banked comments, run the set, and compare the drafts against how you would have marked three scripts yourself. If it tracks your judgement on those three, it will save you time on the rest. If it does not, the criteria usually need sharpening before the tool does.