I hope this question is appropriate for this site. Maybe community wiki might be a good option.

I was the TA in a course for undergraduates last term. I student approached me after the course ended and wanted to know how he could dive deeper into the subject of the course. I suggested him that we could launch a reading course, in which we would work through a research paper. I know the topic of the paper a bit, but want to use this reading course to learn about that specific subject myself.

Question: Do you have experience with such a format and can provide best practice?

What interests me in particular is: When I study a new topic on my own, that happens somehow in a non-linear manner. Together with another student, I guess it's necessary to give the whole thing quite a bit more structure and assign e.g. weekly reading assignments and specific tasks?

  • 1
    Is this an appropriate thing for a TA to do? Don't you need more, like a professor or department buy-in? I think it is a good idea, if you can manage it. (I removed the europe tag as irrelevant.
    – Buffy
    Mar 28, 2021 at 15:10
  • 3
    In principle, it’s our private pleasure. I think in Europe the relationship between TAs and students is sometimes a bit different than for example in the US, this is why I included the “Europe” tag in the first place. Mar 28, 2021 at 15:13
  • Seemingly. You can put it back of course. And I'd be interested to learn about the differences. But a reading course is actually independent of who runs it, so maybe it is still orthogonal.
    – Buffy
    Mar 28, 2021 at 15:14

2 Answers 2


My experience was primarily as a (math) student many years ago but the course didn't focus on a single paper, but on a topic of which the paper was only an example. Maybe this is something you could use.

The basic meta concept is that reading isn't enough to learn and that students need to be more active if what they read/see/hear is to actually be learned.

So, if you organize a group around a topic and select, not one, but several papers that refer to that topic. Each student (or perhaps each pair of students) is responsible for creation of a presentation on that to the others, giving key concepts and any insights they manage to develop. But, all students are responsible for reading the paper.

A session, with everyone present, has two parts. First is the presentation of one of the follows by an open discussion at which questions can be asked and further ideas developed.

If the group meets once per week, then you "cover" one paper per week.

After the presentation and the discussion, the student could be asked to then develop a write-up.

This idea can be adapted to chapters of a book, also.

Note, however, that undergraduates can be expected to be awkward at this, so you need to be pretty understanding if it is a graded/marked activity. It is a useful skill to have, but one in which they have little or no experience. And not every suggestion here need be incorporated. There can be many variations.

Edited to add:

Suppose, however that you have a large and long single paper. Such is difficult to divide up, since understanding early parts may be essential for reading later parts.

In such a case, the paper can be read in "sections" by everyone. One student (or a pair) is designated to lead the discussion of the "next" section and develop a set of initial questions. When the group meets that student leads the discussion. Alternatively, each student must turn in a written question or two on the section of the week. Or perhaps an insight along with a question. But, again, but the emphasis on student activities beyond reading and just making up questions on the fly. Insist on some actual preparation.

Over the horizon. Bonus points.

The following is probably too much to expect from undergraduates but you might offer bonus points (in any grading scheme) for any student or group of students who can come up with a research question that might arise from the paper. You could suggest what might lead there such as "changing some initial condition" or "generalizing" or "specializing" or whatever seems appropriate for the readings at hand. It might even happen, even if unlikely.

  • 1
    I agree with most of what you say, but think the rate at which you should cover papers will strongly depend on their content and one should be aware of this before fixing a schedule. There are many papers that can be quickly summarized in a single session, but there are some that are long and complicated enough to take two, or even have a natural split somewhere in the middle. And at least in math there are some 100+ page monstrosities which can be worth spending a whole semester going through them section by section.
    – mlk
    Mar 28, 2021 at 16:20
  • @mlk, yes, of course. I meant it only as an outline, but with an emphasis on the student activities beyond reading. There are a lot of variations. But a big paper is hard to "divide up" among inexperienced students unless the study is really drawn out since understanding the early part is essential to working on later parts. Let me add an addendum.
    – Buffy
    Mar 28, 2021 at 16:39
  • Thanks for your perspective! I think for a general reading course with students (as part of their regular studies) this is very suitable. However, my specific situation seems to be a bit different: My “group” could equally consist of some PhD students who want to explore a “new” subject using a longer paper. I mentioned that it is together with undergrads because this moves the burden of structuring the material and giving guidance to me. Mar 28, 2021 at 20:00

I did it several times with individual students rather than a group. What we did was just to meet once a week and go one logical unit a time (could be anything from a section to a single lemma, depending on the length and difficulty). I asked if the student had any questions first, then I let him to present the piece on the board (closed book style, but I gave hints if he got stuck anywhere), and then we decided on what the next logical unit would be for the next week. The student was also welcome to ask questions between the sessions.

We have managed this way 2/3 of J.-P. Kahane's "Random series of functions", about 1/2 of L. Hormander's "Functions of several complex variables", Carleson theorem on the a.e. convergence of the Fourier series (exposition by O.G. Jorsboe and L. Mejlbro) and a couple of other things (though I usually tended to choose books over individual papers), so I view it as a positive experience in general despite varying student's abilities and degrees of motivation.

IMHO, the main points are

  1. to let the student do most of the work,

  2. to thoroughly check understanding (the superficial knowledge of the first section of a long paper will invariably get you completely stuck by the fourth one, if not earlier),

  3. to go reasonably slowly (so plan plenty of time for everything), and

  4. to keep momentum (so for the hardest sections, the teacher has to participate in the discussion quite actively too; I usually tried to explain the general idea before I made a formal assignment for the next week in such cases).

I'm a bit wary of trying a group discussion (mainly because of point (2) I mentioned) but I'll not be surprised if some people managed to pull it through with more than just a short "one-session" paper.

The last but not least thing to mention is that you'd better read and understand the full paper yourself before assigning it. The only time I deviated from this rule resulted in a near disaster.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .