I was discussing this topic with a couple of students at the UHI Post-graduate research students’ conference this week. It sounds like a trick question, but in the same way that a long walk, or a marathon, begins with the first step, so does the literature review begin with the first academic paper that the student reads. Normally, when starting out, the supervisor will direct the research student to three or four good papers which are especially relevant to the research topic under consideration. The student will read these, take notes, and then turn to the reference list at the end. As a guiding rule, the researcher should follow-up any and all of the references cited in the text of the paper which are in any way interesting, challenging, or crucial to the argument being made. From these first few papers, the research student will possibly discover half-a-dozen or more references in each paper which need to be read. At this level of study there is an expectation that if the researcher is going to quote anything, or even make a reference to previous academic work in context, they really need to have read the original article. It is not enough to say, “As was noted (Bloggs in Somebodyelse, 2015) it is clear that…” because it may be that the way that Somebodyelse used the idea(s) of Bloggs was wrong, or not how Bloggs intended that data to be used. There is always the danger of misinterpretation. The student needs to reach their own understanding.
So, over a period of time, the student will read a lot of academic articles, track down the evidence which is cited in these articles, then read these follow-up articles too. And so on. From an initial 3 or 4 papers, the lines of follow-up investigation spread out like the roots of a tree. One of the signs that will tell the student when they are coming to the end of their required reading, is when they start to see the same papers starting to crop up again and again. There will undoubtedly be some blind alleys, when the reading slips off in one particular direction or another which is not really useful to the current research project. It might be because the papers referred to are old and the knowledge has been surpassed in later years (this will vary between subject areas, such as fast-moving subjects in science, computing, or e-learning). In some subjects it will be necessary to reference much older publications, either because they set a marker in the development of the subject, or because you want to contrast them with contemporary methods and disciplinary thinking).
Another good idea is to visit the college library collection of previous PhD dissertations for a similar or related subject area. My preference is to search the online index of theses to which my university subscribes. This gives me access to every PhD abstract that has been produced in the UK, and an opportunity to order a copy if I find a particularly interesting match. The object of seeking out a similar PhD is not simply to read about the subject area (which you will extend and surpass anyway) but also to get a feel for the structure of the PhD dissertation, and to get a fast-track on the references that have been used to provide the evidence for this thesis. In the old days, people used to talk about students “reading for a degree in X” rather than studying it, and it is certainly true that the more effort that is initially spent on reading the background and the latest information on the subject, the better placed the student will be to make informed decisions when they begin to gather research data.