Writing a Literature Review Without Drowning in Sources

A second-year PhD student I know has, at last count, 214 PDFs in a folder on her laptop called lit review. Another folder, nested inside it, is called lit review v2. Inside that is a folder called lit review REAL. She has not started writing. When I asked her what the problem was, she said, without hesitation, that she didn’t have enough sources yet.

She has plenty of sources. The thing she does not have is an argument.

This is the single most common failure mode of the literature review, and it is almost always misdiagnosed by the person experiencing it. The anxiety presents itself as a gap in coverage. If I could just read three more things, I could start writing. But reading three more things does not produce an argument. It produces three more PDFs. The lit review that never gets written is rarely a lit review of insufficient sources. It is a lit review whose author has not yet decided what they think.

This is worth saying plainly, because the confusion is baked into how we teach the genre. Students are told that a literature review surveys the existing scholarship on a topic. That description is technically accurate and practically useless. It suggests that the writer’s job is comprehensive coverage, which it is not, and that the writer’s voice should be neutral, which it should not be. A literature review is a piece of writing that makes an argument about the state of a field. The argument might be that two camps have been talking past each other. It might be that a well-known finding rests on a narrower empirical base than people realize. It might be that a question everyone treats as settled is actually wide open. Without an argument of this kind, a lit review is a bibliography with transitions.

The scholars who write well about academic writing are surprisingly consistent on this point. Pat Thomson, who has spent years blogging to doctoral students about the craft, returns again and again to the idea that a lit review is a conversation you are joining, not a wall you are building. Wendy Belcher, whose Writing Your Journal Article in Twelve Weeks has become something close to a standard reference in the humanities and social sciences, frames the lit review as an act of positioning the writer’s own contribution. Howard Becker, in Writing for Social Scientists, was bluntly impatient with the idea that scholarship is about covering everything. He thought it was about saying something that was worth saying, to people who might disagree, in language they could follow.

So what does a sane sequence look like, if you are the student with the 214 PDFs?

First, skim widely and badly. Not carefully. The first pass is not for comprehension, it is for the shape of the conversation. Read abstracts, introductions, and conclusions. Notice who cites whom. Notice which names appear in every bibliography and which appear only in one subset of papers. This tells you where the clusters are. A week of fast skimming will teach you more about the structure of a field than a month of careful reading of ten papers in the wrong order.

Second, pick a question. Not a topic. A question. What happens to working memory under cognitive load is a topic. Whether the dual-task paradigm used in most working-memory studies is actually measuring what it claims to measure is a question. A question has a possible answer, which means it has possible disagreement, which means it has the thing you need: a conversation in which people have taken sides. If you cannot yet find a question you care about, read more introductions and fewer methods sections.

Third, map the conversation. This is where the work actually gets done, and it is the step most students skip. Open a document or a set of linked notes and start making claims of the form X argues that Y, but Z has shown that W. Do this for ten or fifteen of the central papers. You will find, almost immediately, that the field divides into positions, and that certain papers cluster together because they are making the same move. This mapping is not your lit review. It is the scaffolding your lit review will hang on.

A Zettelkasten-style linking system is genuinely useful here, because the move you are making, connecting claims across papers and letting a structure emerge from the connections, is the exact thing that note-linking tools were designed for. A good lit review is, more or less literally, a well-written traversal of such a network. If you are looking for help turning dense papers into usable summaries as a first pass, AI writing assistants can be real help as source summarizers, though I would not trust them to tell you what matters. That part is yours.

Fourth, manage your references with a tool and stop thinking about them. Zotero is free and works. Mendeley works. EndNote works if your university pays for it. The tool does not matter. What matters is that you stop losing citations, stop retyping author names, and stop spending emotional energy on a problem that was solved twenty years ago. Set it up in an afternoon and forget about it.

Fifth, write thematically, not chronologically. The temptation to organize a lit review by date, marching from the 1970s to the present, is almost always a sign that the writer has not yet found their argument. Chronology is a default. Themes are a decision. If your review is structured around three or four positions in the debate you identified, with each section explaining what that position claims, who holds it, and what its weaknesses are, you will have written something that makes an argument. If it is structured around a decade-by-decade tour, you will have written an annotated timeline.

There is one more shift worth naming, which is about what a lit review is for in the larger project. The lit review is not a hoop. It is where you earn your seat at the table. By the time a reader finishes it, they should know what conversation you are in, what is at stake in that conversation, and why the thing you are about to do next is the right move to make. If your lit review does not do those three jobs, the rest of your thesis will struggle, no matter how good the data.

The student with 214 PDFs does not need more PDFs. She needs to close the folder, open a blank document, and write a paragraph that begins The central disagreement in this field is. Whatever comes after that sentence is the beginning of her review.

Photo via Unsplash.