Almost every guide to the literature review tells you to “synthesise rather than summarise” and then does not show you a synthesised paragraph. This one does. Below are three review passages written out in full, with each sentence labelled by the move it performs, so you can see the mechanism in operation instead of reading a description of it.
If you want the procedure — how to design the search, screen sources and build a synthesis matrix — that is a separate job and we have it in our step-by-step guide to writing a literature review. This article assumes you have the reading done and the blank page in front of you.
How to read the passages. They are shapes, not claims. Author names, findings and figures are marked [cite] and [finding] rather than invented, because a specimen that contained a fabricated study would be teaching you the worst possible habit. Your own referenced evidence goes in the brackets.
The difference in one picture
The test is mechanical and you can apply it to your own draft in a minute: look at what each paragraph begins with. If your paragraphs begin with author names — “Dlamini (2022) found…”, “Naidoo (2023) argued…” — you have written an annotated list. If they begin with a claim, and the authors appear inside the sentence as evidence, you have written a review.
Passage 1: the thematic paragraph
This is the workhorse of the chapter. One theme, several sources, an explicit position at the end.
[Claim opening.] Support structures, rather than the design of a curriculum itself, are consistently identified as the determinant of implementation in school settings. [Evidence, grouped.] Studies conducted in well-resourced settings converge on this point, finding that access to subject-specific mentoring and dedicated planning time predicted implementation more strongly than teachers’ reported agreement with the reform
[cite, cite]. [Evidence that complicates.] Work in lower-resourced settings has reported the same direction of effect but a markedly smaller magnitude[cite], which the authors attribute to the near-absence of the support conditions being measured. [Synthesis, the writer’s own sentence.] Taken together, these studies establish the relationship but leave open whether the mechanism operates at all below a threshold of available support — a question that cannot be answered from samples in which that threshold is always met.
What is doing the work. The paragraph has four sentences and only one of them reports a study on its own. The final sentence is the writer’s, and it is the reason the paragraph exists: it converts a body of findings into a limitation that this study can act on. A paragraph without that final move is a summary no matter how many sources it contains.
Notice too that two sources are grouped in a single citation. Grouping is the visible sign of synthesis: it tells the examiner you read the studies as a set rather than one at a time.
Passage 2: the disagreement paragraph
When the literature does not agree, resist the urge to report both sides and move on. Explain the disagreement.
[Name the disagreement.] The reported relationship between [variable A] and [variable B] is inconsistent across the South African literature. [Side one.] Studies using [method or population type] report a positive association
[cite, cite]. [Side two.] Those using [contrasting method or population] report no significant association, or a negative one[cite]. [Explain rather than report.] The split is not random: it tracks how [variable B] was operationalised, with self-report instruments producing the positive association and administrative records producing the null result. [Consequence for this study.] This matters methodologically as well as substantively, because a study measuring [variable B] by self-report is positioned on one side of an unresolved measurement question before it collects a single response.
What is doing the work. The fourth sentence. Anyone can list two contradictory findings; identifying the axis along which they divide is analysis, and it is the single most reliable way to signal that you have actually read the methods sections rather than the abstracts.
You will not always find a clean explanation. When you cannot, say so explicitly — “the inconsistency is not explained by design, sample size or setting, and remains open” — which is an honest finding rather than a gap in your reading.
Passage 3: the South African positioning paragraph
Every South African dissertation needs at least one paragraph doing this, and it is the one most often missing. Its job is to establish that the question is a question here.
[Establish the international position.] The bulk of evidence on [phenomenon] derives from [regions], where [structural condition] is broadly uniform
[cite]. [Name the local structural difference.] The South African context differs on precisely that condition: [named, referenced structural feature — a policy instrument, a sectoral structure, a demographic or linguistic reality][cite, figure]. [Report what local work exists, accurately.] Local studies have examined [adjacent aspect][cite]and [another adjacent aspect][cite], but have not addressed [your specific question]. [Do not overclaim the absence.] This is not to suggest the phenomenon is unstudied in South Africa; it is to locate a specific unexamined intersection between a well-evidenced relationship and a locally distinctive condition.
What is doing the work. The final sentence, which is a piece of defensive writing worth learning. “No research exists on this in South Africa” is a claim about your search that an examiner who knows the field can falsify in one sentence. Locating an unexamined intersection is a narrower claim that survives.
Watch synthesis done once
If the difference between listing and synthesising is still not clicking, a library walkthrough is a good complement to reading annotated prose.
The chapter’s opening and closing paragraphs
Two paragraphs carry more weight than any others, and both are short.
The opening tells the reader how the chapter is organised and why. Three sentences is enough:
This chapter reviews the literature on [topic] in three parts. The first examines [theme one], the second [theme two], and the third [theme three]. The organisation is thematic rather than chronological, because the central disagreement in this field is [conceptual or methodological], not a matter of when studies were conducted.
That last clause is optional but it is cheap insurance: it tells an examiner you chose the organisation rather than defaulting to it.
The closing is the gap paragraph, and it must connect back to your problem statement without simply repeating it:
The literature reviewed establishes [the settled ground, one clause]. It leaves unresolved [the specific gap, one clause], because [the reason the existing work cannot reach it — design, population, measurement or setting]. The present study addresses this by [one clause naming what you will do]. The following chapter sets out the methodology.
Four sentences. If your gap paragraph runs to a page, the gap is probably not yet sharp, and the place to fix that is upstream in the problem statement.
Three failure patterns, and what each looks like
- The annotated bibliography in disguise. Every paragraph opens with an author. Diagnosis takes ten seconds; the fix is to write the claim first and move the authors inside the sentence.
- The chronological march. “Early work in the 1990s… By the 2000s… More recently…”. Chronology is only the right organising principle when the field’s development is itself your argument. It rarely is.
- The uncited theme sentence. A confident claim opening a paragraph with no reference attached, followed by sources that do not quite support it. Examiners read the topic sentence against the citations that follow, and a mismatch reads as overclaiming.
How long, and how many sources
There is no universal rule and your faculty guide governs. What is worth knowing is what South African dissertations actually do: we counted the reference lists of 25 South African master’s dissertations to find out, which is a more useful anchor than a number invented for a guide.
Two practical notes on length. A review that is long because it covers more ground reads as thorough; a review that is long because each source gets its own paragraph reads as padded, and the difference is visible on a single page. And the reference list is checked mechanically at submission, so keep it consistent as you go — the repair is described in fixing a reference list that does not match the faculty guide, and the verb you choose when reporting a study carries more meaning than most writers intend, which our reporting verb bank covers.
Frequently asked questions
What does an annotated literature review example show that a how-to guide does not?
The sentence-level moves. A guide can tell you to synthesise; only written prose shows you that synthesis is usually one specific sentence at the end of a paragraph, written by you rather than reported from a source.
How many sources should each paragraph cite?
Two to four is typical for a thematic paragraph. A paragraph citing one source is usually reporting rather than synthesising; a paragraph citing twelve is usually a list.
Should the literature review be organised by theme or by variable?
By theme, in most social-science dissertations, with your variables appearing inside the themes. Organising strictly by variable tends to produce one section per variable and no synthesis across them.
Can I include South African grey literature?
Yes, where it is the best available evidence, provided you say what it is and treat it accordingly. Government reports and commissioned research are often the only source of local figures.
Do I have to critique every study I cite?
No. Critique where the weakness bears on your gap. Critiquing every study individually pushes you back towards the annotated-list structure.
Where does the theoretical framework go?
Usually a separate section or chapter, though some faculties place it inside the review. Keep the distinction visible either way: the review establishes what is known, the framework supplies the lens you will analyse through.
How recent must my sources be?
Recent enough that the field has not moved past you, with foundational work cited regardless of age. A review of a fast-moving area with nothing from the last three years invites an obvious question.
Can I reuse my proposal’s preliminary review?
As a starting point, yes, and it will need substantial rewriting. The proposal review argues that a gap exists; the chapter demonstrates it in detail and positions your findings in advance.
What if I find a study that has already done my project?
Read it carefully before panicking. Population, setting, period or measurement usually differ enough to leave a real contribution, and saying so explicitly in the review strengthens your chapter rather than weakening it.
Can I copy one of these passages?
Use the moves. The bracketed placeholders are deliberate: a specimen with real-looking citations in it would be a fabricated reference, which is the fastest way to fail a dissertation.
Get the chapter drafted
The literature review is the longest chapter in most dissertations and the one people rewrite most. Tesify helps you structure it theme by theme and keeps the references consistent while the argument stays 100% written by you.