Most undergraduate literature reviews fail in the same way. They are accurate, well-referenced, and organised one study per paragraph — and the marker writes “descriptive” in the margin. The difference between a descriptive review and a critical one is not effort or reading volume. It is a handful of sentence-level moves that most students have never been shown.
This guide shows you an annotated extract from a literature review chapter, with every critical move labelled, so you can see exactly where synthesis happens. Then it gives you the sentence patterns and the chapter structure to reproduce it. The example is fictional but written as a real chapter would be.
What the chapter has to do
A literature review chapter in a UK undergraduate dissertation does four jobs, in this order: it maps what is known, it organises that knowledge into positions or themes, it evaluates the evidence behind those positions, and it identifies what remains unresolved. That last job is the one the whole chapter exists for, because it is what makes your research question necessary rather than arbitrary.
Note that “critical” does not mean negative. Evaluating a study means saying what its evidence can and cannot support, given how it was done. A well-designed study that answers a narrow question is not weak; it is narrow, and saying so precisely is criticality.
The annotated extract
Below is a section from a literature review on hybrid working and employee retention. Each critical move is labelled in bold brackets. Read it once for the argument, then again for the labels.
[Thematic signpost] A second strand of research moves beyond satisfaction to ask whether flexible arrangements change the social structure of the workplace itself.
[Synthesis across studies] Three studies converge on the finding that reduced co-presence weakens informal knowledge transfer, though they measure it in different ways — through self-reported mentoring contact, through observed collaboration patterns, and through time-to-competence for new hires. [Evaluative move] That convergence is more persuasive than any one of the studies alone, precisely because the three operationalisations are independent: an artefact of one measure would be unlikely to appear in all three.
[Counter-position] Against this, a smaller body of work reports no such effect, and in one case a positive association between remote working and reported learning. [Reconciling move] The discrepancy appears to be a function of seniority rather than a genuine contradiction: the studies reporting no effect sampled experienced professionals, for whom the relevant knowledge is already held, whereas those reporting a decline sampled early-career employees. [Implication] If that reading is correct, the question is not whether hybrid working harms learning but for whom.
[Methodological evaluation] A limitation common to nearly all of this work is that it relies on cross-sectional self-report collected during a period when hybrid arrangements were still novel. Novelty effects and self-selection are both plausible, and neither can be ruled out by the designs used. [Scope of the claim] The literature therefore supports an association between reduced co-presence and weaker informal learning among junior staff, but does not establish that the arrangement causes it.
[Context gap] A further constraint is contextual. Almost all of this evidence comes from large organisations with formal HR functions, in which hybrid working is a written policy applied consistently. [Why it matters] In smaller firms, arrangements are typically set informally by an owner-manager and vary between individuals, which means the independent variable itself is different in kind. [The gap, stated] Whether the association observed in large organisations holds where the arrangement is informal and individually negotiated has not, to this author’s knowledge, been examined.
Count the studies actually named in that passage: none, by name, and yet every sentence is doing evidential work. That is the tell of a synthesised review. The studies are grouped by what they found and how they found it, not introduced one at a time as characters.
The sentence patterns that produce it
You can build almost any critical paragraph from these six patterns. Keep them somewhere visible while drafting.
| Move | Pattern |
|---|---|
| Synthesis (agreement) | “Studies of X converge on [finding], although they differ in [method/population]…” |
| Synthesis (disagreement) | “Findings on X diverge. Where [group A] report…, [group B] report…” |
| Reconciling | “The discrepancy appears to reflect [difference in sample/measure] rather than a genuine contradiction…” |
| Methodological evaluation | “Because these studies rely on [design], they can establish [X] but not [Y]…” |
| Scope of the claim | “The evidence therefore supports [narrow claim], but does not establish [broader claim]…” |
| Gap | “Whether [relationship] holds in [context] has not been examined…” |
The pattern students are missing is almost always the fourth. Saying what a design can and cannot show is the cheapest criticality available, and it is available for every study you cite.
Structuring the chapter
A workable undergraduate literature review chapter has this shape:
- Introduction to the chapter (150–250 words). What the chapter covers, in what order, and how the material was selected.
- Conceptual definitions (200–400 words). The two or three terms your dissertation depends on, defined from the literature and then fixed for the rest of the document.
- Theme 1, Theme 2, Theme 3 (the bulk). Each theme is a position or a debate, not a study and not a subtopic.
- Conceptual framework (200–400 words). What the literature implies for your study, expressed as the variables, relationships or themes you will investigate.
- The gap and the research questions (200–300 words). Stated explicitly, so the methodology chapter follows without argument.
The single most useful structural test: read only your subheadings. If they are author names, dates or source types, restructure. If they are claims or debates — “Flexibility and informal learning”, “The seniority moderator” — the chapter is organised correctly.
How to get there from a pile of reading
The practical bridge between reading and writing is a synthesis matrix. Build a table with one row per source and columns for: the research question, the population, the method, the key finding, and the limitation. Fill it as you read. When it is full, read it down the columns rather than across the rows — the themes will be visible as clusters of similar findings, and the gaps as empty regions of the table.
Where to search matters as much as how you read, and the databases differ sharply by subject — our comparison of where to search for your dissertation literature covers them. If your whole dissertation is a review rather than an empirical study, the requirements are different again: see literature review or systematic review for which one your department wants, and our guide to the search strategy section of a nursing dissertation for how to document a formal search.
Five things markers flag
- One paragraph per study. The clearest sign of summary rather than synthesis. Group by finding instead.
- Critical language with no evidence behind it. “This study is flawed” earns nothing. “This study’s cross-sectional design cannot separate cause from selection” earns the mark.
- No conceptual framework. The chapter ends and the reader still cannot say what you will measure or look for.
- A gap that does not follow. If the gap in your final paragraph was not visible in the preceding pages, it reads as bolted on.
- Sources that were never opened. A run of secondary citations — classic works cited through other people’s summaries — is obvious to a marker and raises an integrity question as well as a quality one.
How many sources is enough is the question everyone asks; our guide to how many references a dissertation should have explains why there is no number and what markers assess instead. If you are using AI anywhere in this process, the line your university draws matters more than the tool — our guide to writing a literature review with AI, honestly sets out what is reasonably delegated and what never is.
Where this chapter sits in the wider document, and how long it should be relative to the others, depends on your project’s shape — see our guide to undergraduate dissertation structure by subject.
If your reading is done and the chapter will not assemble, that is a structuring problem rather than a writing one. Tesify organises your sources into themes and drafts the chapter around them, with every claim tied to a source you actually added. Everything stays 100% written by you, and it is free to start.
Frequently asked questions
How long should an undergraduate literature review chapter be?
Commonly 25–30% of the total word count, so roughly 2,500–3,500 words in a 10,000–12,000-word dissertation. Your handbook governs; see our guide to word counts and what they include.
What does “critical” actually mean in a literature review?
Saying what each body of evidence can and cannot support, given how it was produced, and reconciling findings that appear to conflict. It does not mean criticising authors.
Should I organise by theme or chronologically?
By theme, in nearly all cases. Chronological organisation is appropriate only when the development of thinking over time is itself your subject.
How do I write the gap without overclaiming?
Describe it as an absence you observed, hedged honestly: “has not, to this author’s knowledge, been examined”. Claiming that nobody has ever studied something is rarely true and easy to disprove.
Can I cite a source I found in another paper’s reference list?
Only if you read it. If you cite it through the intermediary, say so as a secondary reference — the format is in our Harvard referencing guide.
Where does the conceptual framework go?
At the end of the literature review in most subjects, so that it converts the chapter into the variables or themes your methodology then operationalises. Some business departments ask for it as a separate short chapter.
Do I have to include studies that contradict my expectations?
Yes, and they are an opportunity rather than a problem. A review that engages with disconfirming evidence and explains it reads as far stronger than one that quietly omits it.
Is a literature review the same as the background section?
No. Background gives the reader context — prevalence, policy, setting. The literature review evaluates research evidence and ends in a gap. Some structures contain both.