Chung wrote:"Unseen" as a noun is almost certainly a Britishism (not to mention jargon). I've never encountered it used that way on this side of the pond either in any of the language classes that I've taken.
Not A Britishism.
If you look up the Abebooks list, books with "Unseens" in the title have been published on both sides of the pond. It's just an old, deprecated term.
The reason Mista has seen the term is presumably because classical language teaching is much slower to change than modern language teaching.
In fact, I wouldn't be surprised if "unseen" as a noun was originally a translation from Latin, where using adjectives as nouns is perfectly acceptable. A lot of older high register English was translations and borrowings from Latin even when the Latin conventions didn't really match natural English.
The more that I think about its use as a noun in language proficiency exams, it actually seems tautological. A test by definition must (should?) have exercises that uses passages which the test-taker has not seen before - to take it to its logical conclusion, a test is supposed to comprise only of such passages and questions.
That depends how many "seens" you have encountered in the course. Very few people would be able to memorise a whole book, so if you're told that one of the tasks will be to translate a page from a particular book, you are not going to be working from memory.
Working with previously seen passages is easier, but this is not a bad thing in an exam -- it means it's quicker, so you can test more in a given length of time.
Unseen passages also make marks less reliable, because A) you often can't guarantee that no-one has actually read the text before (particularly in the classics) B) Even if none of the examinees have seen it, it's very difficult to avoid introducing idioms that haven't been covered during the course, and then it's random luck which students have come across those particular idioms in their outside study.
That last point might seem OK on the grounds that a widely-read student will do better, and that's a common way of thinking in current language testing. However, I don't think that I have ever come across a paper for the Cambridge Advanced or Proficiency exams in which I could get 100%, because there are always idioms that I, as a native speaker, don't recall ever having encountered.
After all, what kind of test would it be if I were confronted with questions/assignments that matche problem sets or a practice test given previously, right down to the variables, given constraints, and/or preambles?
If the student can't predict that that particular question will come up, it's not as big a problem as you'd think. Performance on previously-seen problems does correspond fairly reliably with performance on unseen problems. But, that said, no-one would do this is a science/maths-type exam.
"Rehearsed question" as Cainntear puts it could be covered by what I view as material that would be presented similarly rather than identically although I think that a "seen" is the antonym of an "unseen". "Rehearsed question" doesn't that give quite that edge of being an antonym to "unseen [question]".
Formal logic is pretty useless in natural language. "seen" vs "unseen" is a binary system, a dichotomy, but there are useful distinctions that go beyond the dichotomy.
The point about "rehearsed questions" is that the students will practice them in class -- they are not simply allowed to; they are actively encouraged to.
If we used a binary antonym, we'd end up talking about "unrehearsed questions", which is ambiguous. Am I allowed to show my students that question, just not rehearse? In that situation you've got a slippery slope, and slowly teachers will start to find excuses for letting more and more rehearsal of so-called "unrehearsed" questions.
Or you could go with "seen" vs "unseen". Now you lose clarity on whether there is supposed to be any in-class practice or not. Some teachers will rehearse the questions, others won't.
The specific problem here is that some teachers will always bend the rules to help their students get better marks, whereas others believe that cheating is never justified, and undermines the whole system. The experience for students therefore varies widely between teachers, and the validity of the exam is compromised, as different people are essentially being tested on different skills (memory vs real-time processing).
So if we set a test with "rehearsed questions", we have questions that are seen
and practiced. It's specific, and every teacher will show and practice the questions.
We have "unseen questions", and we know these have not been rehearsed, because the students have never been shown them.
In between the two is a possible category of "seen but not rehearsed", but this is no use, because we have no way of ensuring that students haven't rehearsed these questions with classmates or private tutors (therefore turning them into "rehearsed questions"
The biggest mistake we can make in testing is to look for binary categories based on simple antonyms -- humans don't work that way.