01Topic headings and claim headings
A topic heading names a subject. It tells a reader where they are standing and nothing else, which is why a chapter made of them reads as a series of stops rather than a route. A claim heading states what the literature shows on that subject, so a reader arrives already knowing what the section is for and spends their attention checking it rather than guessing at it.
The difference is grammatical and it is easy to convert. A heading that reads as a noun phrase is a topic. A heading with a verb in it, making an assertion somebody could disagree with, is a claim. Take each of your headings and ask what the section beneath it concluded, then put that conclusion in the heading and delete the noun phrase.
- Noun phrase: the name of a concept, a population or a method, with nothing asserted about it
- Claim: a sentence with a verb, stating what the body of work collectively found
- Test one, does the heading say something a reasonable reader could argue with
- Test two, could the heading sit above a different set of studies without changing meaning
- If either test fails, the heading is a label and the section under it will read as a summary
02The sequence four or five headings have to make
Read as a set, the headings should move through four positions: what is broadly established, how it is measured and why that measurement matters, where the findings part company with each other, and what nobody has examined in the setting you care about. The last of those is where your question lives, and if it is not the last heading the chapter arrives somewhere other than your study.
That shape also solves the coverage argument for you. A reader wondering whether you have read enough is really asking whether you know where the disagreements are. A chapter whose third heading names a genuine dispute in the literature has answered that question before it is asked, and one that presents every study as broadly agreeing has not.
03The first sentence under a heading carries the synthesis
Under each heading, the opening sentence states what the group of studies collectively shows. Everything after it supports that statement with particular studies, their designs, their populations and the ways they differ. Written in that order, the individual studies become evidence. Written the other way round, they become a list with a sentence stuck on the end.
There is a fast diagnostic. Read only the first sentence under each heading, in order, ignoring the rest of the chapter. Those sentences alone should make the argument the headings promised. If any of them opens with an author's surname, that subsection is summarising rather than synthesising, and the fix is to write the collective sentence first and demote the surname into the support.
04The rewrite that repairs most chapters
Strip the document to its headings on one page. Fix the headings until the sequence works, without touching a paragraph. Then go back and move each existing paragraph to whichever heading now owns it, which is mechanical and faster than it sounds. Only then rewrite prose, and only in the sections that the new order has actually changed.
Two things surface during the move. Some paragraphs belong under no heading at all, and those are almost always studies you included out of loyalty to the reading you did rather than to the argument you are making. And some headings end up with one paragraph beneath them, which usually means the claim is real but under-evidenced, so the search needs another pass rather than the heading needing deletion.
For a chapter that has been past a committee once and is still coming back on structure, the repair belongs to a specialist inside capstone and dissertation work rather than to line editing.
FAQQuestions to control
How long should a literature review be?
Long enough to establish the dispute your study enters and no longer, which for a chapter is set by your programme and for an article by the journal. Length is the wrong variable to manage. A chapter that reaches the right question in fewer pages reads as controlled, and one that reaches it late reads as unfinished regardless of how many sources it contains.
Should the review be organised by theme or chronologically?
By theme in almost every case, because chronology only carries an argument when the argument is that the field changed over time. If it did change, one subsection can say so. Organising the whole chapter by date turns it into a record of your reading order, and readers notice, because the sections stop relating to each other except by year.
What counts as a gap I can claim?
Something that follows from the studies you have just described, rather than something asserted at the end. If your third section shows two bodies of work measuring the same thing differently and never being compared, the gap is already visible and you only have to name it. A gap nobody can see coming from the preceding pages reads as a convenience.
Does everything cited need to be recent?
No, but the recency should be uneven in a defensible way. Programmes often state a window, so follow yours for the bulk of it. Foundational work belongs wherever the field genuinely rests on it. What a reader checks is the disputed section, which should draw on the newest evidence available, because a live disagreement argued from old studies reads as a search that stopped early.
