UPDATE: Fixed a broken link to lit review examples. Now it points to a really good article on what a lit review should be like. Thanks to Alysson Webb!
Several days ago, I wrote my very first serious literature review. It was tough. I encountered many issues along the way. But, in the spirit of “what doesn’t kill you make you stronger”, below are the things I learned.
Spare a lot of time for reading
I grossly overestimated my ability to read an academic paper. When I read a paper, I read only the intro and the conclusion thinking that’s enough so I procrastinated reading the whole thing. It may be so for most purposes, but not this. I was wrong. That resulted in having only a few days before the deadline to read and understand 10+ papers which was torturous. Reading paper is hard work. The intro and motivation sections sometimes are sure easy, but the actual content is most likely not. Start reading as early as possible, especially if you’re not used to reading academic literature. Not reading early is probably my biggest mistake.
Do not read the whole paper at once
As mentioned before, reading paper is hard work. Make sure that the paper you’re going to read is actually relevant to your topic or you’ll waste your precious time otherwise. I have found that reading the abstract, the first sentence of each paragraph in the intro, and the conclusion is enough to get the gist and judge its relevance. Sometimes even just the abstract is enough. When the time comes to read the whole paper, do it slowly, bits by bits. When stumbling upon something you don’t understand, at first try harder to understand it, but after 2-3 times don’t hesitate to leave it and revisit later. Sometimes you need to consult other sources which may take days and you know what, it’s fine. Of course this is only ideal if you have spared a lot of time to read.
Do NOT just highlight, annotate!
I have a habit of highlighting sentences I deem important or interesting when I read a paper. This is good to help me find critical details but oftentimes I forget why I highlight them in the first place. A remedy to this issue is annotation. Leave a note of why you think this part is essential, or questionable, or even ridiculous. This can also serve as a sort of indexing if your annotation tool can display all annotations in a paper. Annotate each important part with things like “motivation” or “datasets” or “evaluation metric”, and to find them later, you only need to scroll over a short list of annotations. This is, of course, provided that you annotate sparingly. And you should. Those annotations lose their importance once you annotate frequently.
Take a break
I cannot stress this enough. Thoroughly reading several papers a day is exhausting. Taking breaks is an absolute necessity to keep you sane and motivated so that you can go on. I personally use Pomodoro technique, ensuring myself a 5 minutes break after about half an hour working. It is going great for me and I recommend you try it too!
Write a summary for each paper read
After reading a paper, I find it helpful to write a summary of what I just read. This helps to recall and solidify my understanding, as well as progress to the actual writing. What I mean by the latter is that when writing the literature review, I can copy-paste my summary as the content. Thus, it greatly reduces the actual work when I write. In order to do so, however, the summary must be written such that it contains all the important info from the paper that I want to convey and, at the same time, succinct enough that I can fit it in a paragraph, for instance. I have tried doing this and I must say it’s not an easy task, but done properly this could help tremendously.
Write an outline
After all the papers have been read, do not jump into writing immediately! Write an outline first. Outline aids our writing; it provides structure and forces us to think what each of the section in our review should communicate. I personally would go as far as a paragraph-level outline. I can’t overstate how writing each of the paragraph topic sentence made it easy for me to avoid writer’s block during my writing. I also put the reference in my outline. I know exactly which reference should go where in my review, reducing the amount of work I need to do when searching for summary I want to refer in my review.
Let the thesis statement be the guide
Like any other argumentative writing, a literature review should have a thesis statement, sometimes called hypothesis, i.e. a one-liner stating your goal in writing the review. All your topic sentences should be in line with this goal. When writing, keep asking yourself whether this is aligned with the thesis statement. This aids your writing to stay focused, and it helps the reader to understand your writing too. Should the thesis statement be explicitly written? From my understanding, usually thesis statement is placed as the last sentence in introductory paragraph and the first sentence in concluding paragraph. But I think it depends on what kind of writing it is. This article explains in more detail what a literature review should look like.
Use the best tool for the job
Use a reference management software (Mendeley, Zotero) to help you organize your references. Use LaTeX, especially if your topic is a science or engineering one, and BibTeX while you’re at it. Use version control (Git, Mercurial) so you’ll never lose your work. Tooling seems superficial but I assure you, great tooling matters. No tool is perfect, but some are better than others for some tasks, so keep a lookout on a new tool and don’t hesitate to try or even migrate to the new one if it means more productivity.