European Schools

Marking European Baccalaureate work

EB marking has two features that most marking tools handle badly: a 0–10 scale with decimals rather than a bank of raw marks, and a room full of students writing serious analysis in their second or third language. Both change what a fair mark looks like.

The 0–10 scale

Work in the European Schools system is marked on a 0–10 scale, reported to a decimal place, against harmonised descriptors of what performance at each point looks like. That is a different instrument from a GCSE scheme that adds up to 25 raw marks, and it changes the conversation with a student: the question is not which points they missed, it is which description of performance their work actually matches.

It also compresses everything. A tenth of a mark is a real distinction, and the difference between a 6.5 and a 7.5 carries the weight that four or five raw marks would elsewhere. Small inconsistencies across a class set are proportionally larger, which makes drift between the first script and the twenty-fifth a bigger problem than it is at GCSE, not a smaller one.

In practice most departments mark S6 and S7 written work against their own criteria — a set of headings with marks attached, or indicative content agreed between the teachers of a subject — and then place the result on the scale. The scheme you actually use is the one your department wrote, which is exactly the scheme a marking tool needs to be reading.

Second-language writing

History, geography and economics are frequently taught and examined in L2. A student is constructing a historical argument in a language they are still acquiring, and the surface of the prose will show it: article errors, calques from L1, an occasional sentence assembled the long way round.

None of that is what is being assessed. If the meaning is clear, non-standard grammar and dialect should not cost marks in a subject essay, and the student should be judged on the thinking rather than the accent of the writing. This is easy to say and hard to do consistently at eleven at night, and it is also where automated marking most often quietly misbehaves — fluency reads as quality, and the most native-sounding script in the pile drifts upward.

The reverse error matters just as much. A student who writes beautifully in their first language should not be rewarded for the fluency itself when the analysis underneath it is thin.

What consistency is worth here

The strongest argument for marking a class set mechanically first is not speed, it is that every script meets the same standard. Experienced examiners agree with one another on a given script only about half the time, and any individual marker drifts over the course of an evening. On a compressed 0–10 scale, that drift is the difference between a 7 and an 8 for two students who wrote equivalently well.

Applying one scheme identically to all of them is the part a person genuinely cannot do by hand. Deciding whether the result is right is the part only you can do.

How Paddle handles it

Paddle marks against the scheme you paste, at the level you state, and is explicitly instructed not to penalise non-standard grammar or dialect where the meaning is clear — so a student writing in L2 is marked on their argument. It credits quotation only for selection and use, never for the quality of the source's prose, and where you supply the source material separately it is excluded from any judgement of the student's writing.

Every criterion quotes the phrase from the student's own work the mark rests on, and every mark carries a confidence score. Paddle was built in Brussels and its first real testing was on European Baccalaureate History scripts, which is not a coincidence.