You swapped a few words for synonyms, moved a clause to the front, stuck a citation on the end. It felt like paraphrasing. Then the checker highlighted it anyway, and a marker reading closely can still see the original sitting underneath, like a stencil.

I see this more than any other source-use problem. Dishonesty is almost never what is behind it. What is behind it is a mechanical habit: people paraphrase with the source sitting open in front of them, and that alone is enough to guarantee the original sentence shape survives into their version. So the thing to change is the procedure. Your vocabulary was never the problem.

Why synonym-swapping fails

A paraphrase is supposed to restate an idea in your own words and your own structure. Swapping words changes only the first half. The clause order, the sentence rhythm, and the logical scaffolding all still belong to the author you read, which is why matching software still recognizes the passage and why an experienced marker still hears the other writer's voice.

There is a name for this halfway state, and researchers treat it as its own category rather than lumping it in with copying. It sits in the gap between quoting and genuinely rewriting, and it is where most accidental trouble lives.

The one change that fixes most of it

Close the source.

Purdue OWL's six steps for paraphrasing put this at the center: reread the passage until you understand it fully, then set the original aside and write your version without looking. Check it against the source afterward for accuracy, and put quotation marks around any distinctive phrase you could not avoid reusing.

That instruction to look away is doing the real work. You cannot reproduce someone's syntax from memory unless the sentence is extremely short. Forced to reconstruct the idea, you naturally rebuild it with your own scaffolding, because that is the only scaffolding you have.

Before and after

Here is a source sentence I wrote so we can pull it apart freely:

Municipal recycling programs collapsed across much of the United States after 2018, when China stopped accepting most imported plastic waste and domestic processors could not absorb the surplus.

Patchwriting, the version that gets flagged:

City recycling schemes fell apart across much of America after 2018, when China ceased accepting the majority of imported plastic refuse and local processors were unable to absorb the excess (Author 12).

Every content word has a synonym. The citation is there. But the clause order is identical, the "when X and Y" structure is identical, and the sentence is the same length with the same rhythm. This is the version people are shocked to see highlighted.

An actual paraphrase:

China's 2018 decision to stop importing most plastic waste left American cities with more recyclable material than their own facilities could handle, and many local programs shut down as a result (Author 12).

Same facts, same citation, different architecture. The cause now opens the sentence instead of arriving in a subordinate clause, "collapsed" has become a consequence stated at the end, and the sentence reflects a decision about what matters most rather than a walk through the original in order.

A test that takes ten seconds

Put your version beside the original and ask: if I deleted every noun and verb, would the two skeletons still match?

If the clauses come in the same order and the connective words line up, you have patchwritten regardless of how many synonyms you found. If a reader could not reconstruct the original's sentence shape from yours, you have paraphrased.

A second check, useful for longer passages: could you have written your version from a one-line note? If your paraphrase contains detail you could only have produced with the source open in front of you, that detail probably arrived with its original phrasing attached.

Paraphrasing more than one sentence at a time

Everything above assumes you are rewriting a single sentence, which is the easy case and also the case that creates the most patchwriting. The moment you are working from a paragraph or a page, the sentence-by-sentence approach is the problem itself.

Marching through a source sentence by sentence guarantees your version inherits the author's order, and order is an argument. The author decided what to establish first and what to hold back. Reproducing that sequence in different words is still borrowing their thinking, even when no individual sentence looks copied.

Work from the whole passage instead. I read it once, put it down, and ask what two or three claims in there actually matter for the point I am making. Then I write those, in whatever order my own argument wants them, which is almost never the order the source used. The rest gets cut. That cutting is what makes the result mine.

Where the citation goes

A question I get constantly, because getting it wrong looks like hiding something even when nothing is hidden.

One paraphrased sentence takes a citation at the end of that sentence. A paragraph paraphrasing one source needs the attribution to cover the whole stretch: name the source at the start, in the text, then cite at the end. Something like "Okonkwo argues that…" followed by your three sentences and a closing citation makes the boundaries obvious to a reader.

Trouble comes from a paragraph of paraphrase with one citation stranded at the end and no signal at the start. Your marker cannot tell whether that citation covers the whole paragraph or just the last sentence, and the ambiguity always reads badly. Worse: paraphrase two sources in one paragraph without naming either, and nobody can reconstruct which idea came from where. At that point the citation is decorative.

When to quote instead

Paraphrasing is not always the right move, and the instinct to paraphrase everything is part of what creates patchwriting. Quote when the exact wording is the point: a definition the field argues about, a phrase the author coined, a legal or policy sentence where changing words changes meaning, or a line whose style you plan to analyze.

The rest of the time, paraphrase or summarize. And prefer summary more often than you probably do. Compressing three pages into two sentences of your own is harder to do badly, because you cannot summarize something you have not understood.

Keeping notes that do not sabotage you

Most patchwriting is created at the note-taking stage and only discovered at the drafting stage, weeks later, when nobody remembers which lines were copied.

Never paste source text into your draft file "temporarily." It will survive.
Keep quotes in a separate document, always inside quotation marks, always with the page number.
Write notes in your own words at the moment you read, while you still understand the argument. A note written as a paraphrase is a paraphrase you never have to make later.
Mark anything uncertain with a symbol you can search for before you submit.

Check the passages you rewrote hardest

Scan a finished draft and the highlights worth your attention are the passages you worked hardest on, not the ones you dashed off. Sentences you struggled to reword are exactly where the source's structure tends to survive. A checker that shows the matching source lets you hold the two side by side instead of guessing, so run the draft through our plagiarism checker and read the yellow highlights rather than the percentage. Yellow is reworded text still tracking a source too closely.

Then rewrite those passages the closed-book way. It is faster than negotiating with a synonym dictionary, and it produces sentences that sound like you, which is the actual goal. The percentage at the top of the report was never the thing to fix. Understanding the difference between what a similarity score measures and what it proves is what tells you which highlights to act on.