During the last semester my supervisor Prof. X held a seminar for bachelor and master students (mostly computer scientists). According to the study regulations the students have to
- give scientific presentations (based on journal or conference papers) and
- hand in seminar papers on the topic of their presentation (i.e. their papers should cover the results of the original paper, lay out details of proofs, add additional explanations or examples, etc.).
The final grade consists of 2/3*presentation + 1/3*paper.
Prologue. At the beginning of the semester Prof. X's research assistants compiled a list of interesting research papers. At the first seminar meeting the students were able to choose the paper they would like to present to the rest of the group at the end of the semester. Furthermore Prof. X made it very clear that understanding the topic is just a portion of presenting and writing down scientific ideas. Every student was assigned to a seminar supervisor to whom they could talk when they ran into problems. Half-way through the semester there was an obligatory meeting with the supervisor and a deadline for handing in a draft of the presentation slides. At the end of the semester we organized a little "conference" and the students presented "their" papers. Afterwards, they received written feedback on their presentations from all participants (students, research assistents, and Prof. X). Six weeks later they had to hand in their seminar papers.
Problem. Unfortunately the quality of many seminar papers is relatively poor; even the papers handed in by students who had understood their topic "quite well" and had given good presentations are surprisingly different from what we expected.
Even though they were allowed to use their native language, many papers had bad spelling (obviously no spell checker was used) and bad grammar.
In some cases it was impossible to understand the basic ideas if one had not already been familiar with the subject.
Imprecise language and almost no sources cited, e.g. "Algorithm Y is rather efficient in comparison with other algorithms."
Some students only cited a single source (= "their" research paper).
Questions. How should we address these issues? Of course, we are going to give some feedback on their seminar papers. I am worried about the next seminar. Reading all these papers was (mentally) exhausting. Should we require that seminar papers must be handed in first? How could we install an iterative feedback process?
I am looking forward to your ideas and experiences.