Synthesis and summary do different jobs
A summary reports what one study found. A synthesis reports what several studies, taken together, mean. The difference is visible in the subject of your sentences: summary sentences start with authors, synthesis sentences start with findings. If most of your paragraphs open with a name and a year, you have written a reading list in prose form.
Compare two versions of the same material. Weak: “Study A found that feedback improved accuracy. Study B also found improvements. Study C reported a smaller effect.” Stronger: “Feedback improves accuracy consistently, but the size of the effect depends on timing — it is largest when feedback arrives in the same session (Studies A and B) and much smaller after a week’s delay (Study C).” The second version states something none of the three papers said on its own.
The test is simple. Ask whether a sentence could have been written after reading a single paper. If it could, it is summary. Aim for paragraphs in which at least one sentence required all of the cited sources to write.
Organize by theme, not by author
Most reviews default to one paragraph per source because that is the order in which the reading happened. Readers do not care about that order. They want the field sorted into the questions it is arguing about, so the organizing unit should be a claim, a construct, a method, or a disagreement — never a citation.
Theme names are a good diagnostic. “Studies on motivation” is a container, not a theme: nothing follows from it. “Motivation is measured differently in classroom and laboratory settings, and the two literatures disagree about the size of the effect” is a theme, because it tells you what the paragraph has to establish and what would count as evidence against it.
A handful of organizing schemes cover almost every review. Pick one as the primary structure and allow a second inside it — for example, thematic at the top level with a chronological sub-order inside a theme where the field genuinely moved. Mixing schemes at the same level is what produces the review that seems to restart every few pages.
- Thematic — by construct, question, or claim. The default, and usually the right one.
- Methodological — by how the evidence was produced, when method drives the disagreement.
- Theoretical — by the framework each group of studies works inside.
- Chronological — only when the argument is that thinking changed over time.
The sentences that carry a synthesis
Synthesis lives in a small set of sentence moves, and they are worth practicing directly. Convergence: “Across settings, X is reported consistently.” Divergence: “Findings split along the line of Y.” Qualification: “The effect holds for A, though not when B is present.” Each of these needs more than one citation to be true, which is exactly the point.
Reporting verbs carry your judgment, so choose them deliberately. “Demonstrates” and “establishes” commit you to the finding; “suggests”, “reports”, and “claims” hold it at arm’s length. If you use the same neutral verb for every source, the reader cannot tell which evidence you find persuasive, and the review flattens into a list again.
Citation placement matters as much as verb choice. A parenthetical citation at the end of your own sentence keeps you as the speaker; a narrative citation at the start hands the sentence to the author. Use narrative citations sparingly, for the few studies whose identity actually matters to your argument.
Deciding what to leave out
Coverage anxiety produces bloated reviews. You are not obliged to mention every paper on the topic, and a review that tries usually loses its argument in the process. The obligation is to cover everything that would change a reader’s judgment of your claims — including work that cuts against you.
Two things earn a source a place: it is load-bearing for a theme, or it is the strongest counter-evidence to one. Foundational work that everyone cites can be handled in a single sentence with a couple of references; recent work that directly touches your question deserves detail. Anything that only repeats a point already made should go, no matter how much time you spent reading it.
Keep a discard file rather than deleting notes. Sources that do not fit the review often turn out to be exactly what the discussion needs when your results land somewhere unexpected. Ten minutes of filing while you read saves an afternoon of searching six months later.
Landing on the gap
The last paragraph is the one your reader will remember, because it is where the review stops describing the field and starts justifying your study. A gap statement has two parts: what is missing, and what would change if it were not. The second part is the one people skip.
Gaps come in recognizable kinds. Something has been studied in one population but not another; two lines of evidence contradict each other and nobody has tested why; a construct has been measured only in the laboratory; a method has become available that makes an old question answerable. Naming the kind of gap keeps you from asserting novelty that a reviewer can disprove with one search.
Weak: “No study has examined this in a Portuguese sample.” Stronger: “The effect has been demonstrated only in settings where feedback is anonymous, so it is unclear whether it depends on anonymity or on the feedback itself — a distinction that determines whether the practice transfers to classroom use.” The second names a gap that has consequences.
