Paraphrasing, quoting and summarising: which to use when
Short answer
Quote when the exact wording is the evidence. Paraphrase when you need one specific point in your own words at roughly the original length. Summarise when you need the gist of something much longer, compressed. All three require a citation; only quoting requires quotation marks.
These three get taught as a list of definitions and then used interchangeably, which is where the trouble starts. They are not three styles of the same operation. They are three different tools, and using the wrong one is what produces both weak writing and accidental plagiarism.
The question that selects between them is not "how much of this do I want" but "what is this source doing in my argument".
Quoting: when the words themselves are the evidence
Quote when the specific wording matters — a statutory phrase, a definition under dispute, a line whose ambiguity is your point, or a claim so surprising that a reader will suspect you of paraphrasing it unfairly.
Outside those cases, quoting is usually a sign that you have not finished thinking. A paragraph built from stitched-together quotations shows you can locate relevant passages but not that you understood them.
Keep it short
Quote the clause that carries the meaning, not the surrounding sentence. Long block quotations are routinely skipped by readers, which makes them an expensive way to say nothing.
Introduce and interpret
A quotation dropped in without a frame does no work. Say whose it is and what you take it to mean, before and after.
Reproduce it exactly
Including errors, marked with [sic]. Use an ellipsis for omissions and square brackets for anything you add. Tidying a quotation silently misrepresents the source.
Paraphrasing: one passage, your words, same detail
A paraphrase restates a specific passage in your own words at roughly the same length. You use it when the content matters but the phrasing does not, which is most of the time.
The test of a real paraphrase is that it could not be turned back into the original by reversing a thesaurus. The structure should be yours, not just the vocabulary.
The reliable method is the same one that makes summarising work: read the passage, close the source, write what it said, then reopen and check for both accuracy and accidental similarity. Writing with the source visible is how patchwriting happens.
Summarising: a lot of source, a little of your text
A summary compresses something substantially longer into its main points. It is what you use when a reader needs to know what a source argued but does not need its detail.
The skill it depends on is deciding what counts as a main point, which is the same judgement as finding the main idea at paragraph scale. Get that wrong and you produce a shorter text that misrepresents the original.
Summaries carry a specific risk: compression strips qualifiers. "Associated with, in one cohort, under these conditions" becomes "causes" with alarming ease. Check your summary against the original specifically for strength of claim.
The patchwriting trap
Most plagiarism findings against students who were genuinely trying to do the right thing come from patchwriting: keeping the source sentence and swapping words for synonyms.
It happens because the source is open while you write. Your brain takes the path of least resistance and edits rather than composes, and the result retains the original architecture even though no phrase is identical.
The structural fix is to never write with the source visible. Take notes, close it, write from the notes. If you cannot write the point without looking, you do not yet understand it well enough to use it — which is useful information in itself.
Choosing quickly
Ask what breaks if you change the words. If the argument breaks, quote. If nothing breaks and you need the whole point, paraphrase. If nothing breaks and you only need the gist of something long, summarise.
Highlighting helps before any of this, by marking the passages worth using at all. Deciding which twenty sentences in a paper are candidates is a separate job from deciding how to deploy them, and doing it first makes the second decision much faster.