Direct answer
You remove plagiarism from a research paper by fixing attribution and rewriting language, not by deleting content: put quotation marks around verbatim text and cite it, restate borrowed ideas in your own sentence structure with a citation, replace patchwritten sentences with genuine synthesis, and disclose any reuse of your own earlier work. Then re-check the draft and read every remaining match individually, because a similarity score is a percentage of matched text, not a plagiarism verdict [1]. The goal is a paper where every borrowed idea is traceable to a source, not a paper with a low number.
What Are the Legitimate Ways to Remove Plagiarism From a Research Paper?
The first thing to separate is unattributed reuse from matched text. Turnitin's own guidance on plagiarism types treats direct quotation as entirely legitimate when it is enclosed in quotation marks and accompanied by a full citation — the problem was never the borrowed words, it was the missing attribution [2]. So the highest-priority fix in almost every draft is not rewriting but labelling: find every verbatim string, wrap it in quotation marks, and attach the author, year, and page. Deleting a well-chosen quotation usually weakens your argument without improving your integrity.
The second fix is genuine paraphrase. Paraphrasing means restating the idea in your own sentence structure and vocabulary while still citing the source — the citation does not become optional because the words changed [2]. A practical test is to close the source, write the idea from memory in a sentence you would actually say aloud, then reopen the source and check that you have not reproduced its clause order. If your sentence can be laid over the original and still line up, you have not paraphrased.
The third fix is the one most students skip: patchwriting. Swapping synonyms while keeping the original sentence architecture is still treated as plagiarism under most institutional policies, and it is exactly the pattern that survives a naive "rewrite" pass [2]. The remedy is structural — merge two sources into one claim, add your own transition, or subordinate the borrowed point to your own analysis so the sentence exists because of your argument rather than because of the source.
Finally, treat reuse of your own earlier work as a real category. Submitting text you have already submitted elsewhere, or lifting a paragraph from your own previous essay, is self-plagiarism unless it is disclosed and cited [2]. If a method section genuinely needs to repeat earlier wording, say so in a footnote or cite the prior work — disclosure converts a concealment problem into a normal scholarly practice.
Does Rewording and Adding Citations Actually Lower a Similarity Score?
Yes — but only in specific ways, and the number is a weaker signal than most students assume. The similarity score is recalculated from the current version of the document against the comparison database, so changing the matched strings genuinely changes the match set [3]. Quoting and citing a passage does not delete the match, but it changes what the match means, which is the distinction that matters in an integrity review.
The complication is that instructors can configure the report to exclude quotations, the bibliography, and small matches [3]. Two students with identical text can therefore see very different percentages depending on the settings their department uses. This is why chasing a specific number is a poor strategy: you may be optimising against a filter configuration you cannot see, while the underlying attribution problem is untouched.
It also cuts the other way. Excluding a source or applying a filter changes the displayed number without changing a single word of your text [3]. A score that drops because of settings rather than because of rewriting is not evidence that anything was fixed — and if your institution reviews the unfiltered report, the discrepancy becomes its own problem.
What reliably moves a score in the right direction is structural rewriting plus correct citation: shorter, better-integrated quotations, paraphrases with new clause order, and synthesis paragraphs where multiple sources are compressed into your own claim. What reliably does not work is synonym substitution, sentence shuffling, or inserting invisible characters — those leave the match intact or create a new integrity issue, and they are trivially visible to anyone reading the document rather than the score.
How Can I Check My Rewritten Draft Before I Submit It?
The safest verification is the one your institution already provides. Where draft submissions are enabled, Turnitin lets students submit a draft and review the similarity report before the final deadline, so attribution problems surface while they are still fixable rather than after a misconduct referral [4]. Ask your module lead whether draft submission is switched on for your assignment — it is a per-assignment setting, not a universal feature, so the answer varies by course.
When you do get a report, read it the way an instructor reads it. Open each match, look at how much text is highlighted, and ask a single question: is this attributed? A 24% score built from properly quoted and cited material is a healthy report. A 9% score containing one unattributed paragraph lifted from a source is a serious problem [1]. Working match by match, in descending order of matched length, is far more productive than watching the headline percentage.
Where draft submission is not available, you can still verify the draft privately against the same report format your instructors see, so that you are reviewing real match highlights and a real similarity summary rather than guessing. Whichever route you take, keep the corrected file and note what you changed — if a match is later queried, a short record of "quotation added, citation added, paragraph rewritten" resolves the conversation in one email.
The habit worth keeping beyond this paper is to verify before the deadline rather than after. Students who check a draft routinely describe the same experience: the report arrives fast, the flagged passages are specific, and the fix is a thirty-minute editing session instead of a panic [1][4].
Once you have rewritten the flagged passages, the remaining question is whether the fixes actually landed — and that is a question about the report, not about your confidence. turnitin0 produces the same Turnitin AI detection and similarity reports your professors see in their LMS, so you can open each match, confirm the quotation marks and citations are in place, and re-check the draft as many times as your deadline allows. English documents only, 300–30,000 words, results typically back in under 15 minutes, and your file is never added to Turnitin's student paper database.
※ Turnitin0.com - Actual [Turnitin AI Report](https://www.turnitin0.com/guides/us/purchase-humanizer) Cover, Score, Flag And Similarity Summary
Get [Real Turnitin](https://www.turnitin0.com/) AI & Similarity Report
FAQ
Is it enough to paraphrase without citing?
No. Paraphrasing changes the wording, not the ownership of the idea, so the citation is still required [2]. Removing the citation to make a match disappear converts a formatting issue into a plagiarism issue.
What similarity percentage is safe to submit?
There is no universal safe number, and Turnitin advises against treating a threshold as a misconduct finding on its own [1]. Read the matches: attributed quotations and bibliography inflate a score harmlessly, while a single unattributed paragraph is a problem at any percentage.
Does Turnitin remove quoted and cited text from the score?
It can, if the instructor enables the exclude-quotations and exclude-bibliography settings [3]. Because those settings vary by department, do not assume your visible score matches the one a marker will see.
Can I check my draft before the final submission?
Often yes, but only if draft submission is enabled for that specific assignment [4]. If it is not, verify the draft privately in the same report format your instructors use, then re-check after you edit.
Will rewriting a flagged paragraph always clear the match?
Not always — a rewritten sentence can still match if it keeps the original clause order or distinctive phrasing. Rewrite structurally, cite the source, and re-check the passage rather than assuming the edit worked [3].