Be rigorous at the start and the marriage will survive

The relationship between a PhD student and their supervisor can be a tricky one. Susan Bassnett has a recipe for success.

Sooner or later, most academics find themselves examining a doctoral thesis. When a thesis is good, the viva can be a rewarding experience for all concerned. It is always a moving moment when the candidate learns that he or she has satisfied the examiners and the doctorate is in the bag.

But what about those times when it isn't possible to pass a thesis and when the examiner confronts the unsuccessful candidate's misery? I have found myself faced with the dilemma of referring a thesis or, worse still, recommending a lower degree. On another occasion, once the candidate had fled from the room in tears, I was placed behind a desk to give the verdict because the internal examiner felt the candidate might attack me.

Each failed thesis is a distressing event. Not least because there is often a doubt as to where the responsibility for failure lies. Sometimes the cause of the problem is clear: the candidate was not up to writing a doctorate, the research was not sufficiently rigorous or was unoriginal, or the topic was wrongly conceived. But there may also be other problems - poor structuring of an argument, poor presentation, an inadequate bibliography - and here we enter the grey area of shared responsibility between candidate and supervisor. At a time when there is a consultation document doing the rounds seeking to improve the quality of doctoral work generally, it is fair to reflect on the role of the supervisor and to ask what exactly is expected of them.

We have all heard tales of supervisors who saw their students once a year for ten minutes in the pub. These days there is talk of extensive training for supervisors, of mentoring and guidelines and I even heard it suggested that supervisors should sign a declaration accepting liability should something go wrong in the viva. Clearly, the pendulum is swinging away from the idea of a supervisor as a source of expert advice to someone who is expected to take ever greater responsibility for students' work. This doesn't seem reasonable to me, particularly when the number of doctoral students is rising and many of those students do not have English as their first language. What should a supervisor do?

For a start, there needs to be a clear division of labour, ideally a contract clarifying who does what and what supervisor and candidate can expect in terms of contact hours and reading of drafts. Some institutions have such guidelines, some don't. If there isn't a system, you need to set up your own. I take the view that creating a bibliography is part of the process of acquiring a doctorate; the supervisor can make suggestions, but bibliographical research is not their job.

I also take a tough line on drafts: some candidates rewrite so many times that you would go mad if you read every version. One student once gave me a 400-page "final" draft then changed it completely without telling me. I refused to read the new version.

It is also important to ensure that a draft is as close to perfect as possible. There is no point reading something that is scrappy, lacking footnotes and has not been proofread. My tactic in such cases is to read one chapter in that condition, noting how many hours it takes me to do so and correcting every single error. This is generally so shocking to the candidate that the next chapter arrives in much better shape.

Should a supervisor proofread a thesis? If you don't help students whose English is poor or who can't write or spell, they will be disadvantaged, but the final responsibility must be the candidate's. Consequently, supervision needs to vary at different points in the doctoral lifecycle: too much talk doesn't help anyone, rigorous correcting does, and there should always be detailed written feedback because information given orally will be misheard or forgotten.

A good relationship between supervisor and student is one of the most powerful and lasting relationships you can have. Get it right and the bond will be sealed for life. Get it wrong and the candidate will blame you forever. Not unlike a marriage.

Susan Bassnett is professor at the Centre for Translation and Comparative Cultural Studies, University of Warwick.