Fidelity: say what the paper says, not what you wish it said
The most common failure in summarizing is not omission but drift. You read a paper because it is relevant to your argument, so your attention is already shaped by what you need it to say, and the summary comes out tilted toward that need. The tilt is rarely deliberate and almost always visible to anyone who reads the original.
Watch two specific moves. The first is upgrading a verb: a paper that reports an association arrives in your notes as demonstrating a mechanism. The second is dropping a boundary condition: a finding that held in one age group, one country, or one experimental setup becomes a general result. Original: “Among first-year undergraduates at a single institution, structured feedback was associated with higher revision rates.” Drifted: “Structured feedback increases how much students revise.” Two conditions gone, and a correlation promoted to a cause.
A faithful version keeps the shape: “In a single-institution study of first-year undergraduates, students who received structured feedback revised more often, though the design does not establish that the feedback caused the change.” Longer, yes — and now it is a sentence you can defend in a viva.
Proportion: let the paper set the shape
Proportion is the quiet half of accuracy. A summary can contain nothing but true statements and still misrepresent a paper if it devotes half its length to a robustness check the authors mention once. Readers infer importance from space, so where you spend words is itself a claim about the paper.
Before drafting, sketch the paper’s own weighting: roughly how much of it is method development, how much is the primary result, how much is secondary analysis. Then allocate your summary the same way, rounded to whole sentences. If your budget is four sentences and the paper is three-quarters methods, one sentence on findings is the honest split.
The exception is a purpose-built summary — an entry in a literature review that exists to serve your argument. There, it is legitimate to foreground the part relevant to you, provided you say so. “Relevant here is their secondary finding that …” signals to the reader that you are selecting, which is honest, rather than implying the paper was about your topic.
Summary, abstract, and annotation do different jobs
An abstract is written by the authors to make you read the paper; it advertises. A summary is written by you to make the paper usable without rereading it; it records. An annotation in an annotated bibliography adds a third layer — an assessment of quality and relevance in your own project. Confusing them produces summaries that sound promotional and annotations that forget to evaluate.
Practically, this means you should never build a summary by lightly rewording the abstract. Beyond the plagiarism problem, an abstract omits precisely what you will need later: sample size, exclusions, what the control condition actually was, which comparison the headline number came from.
For an annotation, add two sentences the paper cannot supply: how strong you judge the evidence to be, and what role the source plays in your own argument. Six months later those two sentences are worth more than the rest of the note.
- Abstract — the authors’ pitch, written before you read.
- Summary — your compressed record, written so you never have to reopen the PDF.
- Annotation — summary plus your judgment of quality and relevance.
Making the summary usable a year later
Write summaries for the version of you who has forgotten the paper. That means naming the study rather than using pronouns, spelling out abbreviations once, and recording the design in plain terms — “a randomized trial with a waitlist control” rather than “an RCT.” Notes that assume context are notes you will rewrite.
Include one line on how the paper connects to the others you have read: it replicates, contradicts, or extends something specific. A stack of summaries with no connective tissue is a reading log; summaries carrying those links are already the skeleton of a literature review.
Keep the full reference attached to the summary at the moment you write it. Reconstructing bibliographies from half-remembered author names is a distinctive kind of misery, and it is entirely avoidable.
