Examples
Worked examples
- Is an instance
A results section reproduces two sentences from a previously published paper word-for-word, with a citation in the reference list but no quotation marks around the borrowed sentences — this is verbatim textual plagiarism even though a citation exists, because the citation doesn't disclose that the wording itself was copied.
- Is an instance
An introduction paraphrases a source passage by swapping several words for synonyms and reordering a clause, while keeping the original's sentence structure and argument sequence intact, with no citation at all — this is patchwriting (a form of mosaic/textual plagiarism), and it remains plagiarism even where a citation is added, per COPE and university academic-integrity guidance, unless the borrowed wording is also quoted.
Counter-examples
Looks similar, but isn't
- Not an instance
An author reads a source's argument, fully internalizes it, and writes an original passage in their own words and sentence structure that reaches a similar conclusion — but omits a citation to the source that originated the idea. No wording matches, so a similarity checker would not flag it, but this is still a citation failure; ORI and journal-ethics bodies would classify it as plagiarism of ideas, not textual plagiarism, because the violation is about crediting the source of the contribution rather than about copied wording.
Editorial commentary
Textual plagiarism is the unattributed copying of another person’s written words. The U.S. Office of Research Integrity (ORI) draws this distinction explicitly in its own guidance: plagiarism of text is copying a portion of text from a source without giving appropriate credit, while plagiarism of ideas is borrowing someone else’s argument, framework, or findings and expressing it in fully original wording without crediting where the contribution came from. The two categories often overlap in a real case, but the distinction matters operationally, because they are detected and remediated differently.
What makes something textual plagiarism
ORI’s guideline is specific: any verbatim text taken from another source must be enclosed in quotation marks (or set off as a block quote) and accompanied by a citation. A citation by itself is not sufficient — it tells a reader the idea or fact traces to a source, but it doesn’t disclose that the exact wording was copied. Textual plagiarism therefore covers a range of severity, from copying a full passage outright to copying a sentence or two embedded in otherwise original text:
- Verbatim (direct) copying — reproducing another author’s sentences or paragraphs exactly, with no quotation marks, whether or not a citation is present.
- Patchwriting — copying a passage while swapping some words for synonyms, deleting or inserting a few words, or lightly reordering clauses, but keeping the source’s underlying sentence structure and phrasing intact. Patchwriting is a well-documented, common failure mode in academic writing — it often occurs when a writer intends to paraphrase but stays too close to the source’s language — and it is treated as plagiarism regardless of intent once a citation and quotation marks are both missing.
- Mosaic plagiarism — interweaving a source’s own ideas and phrasing with the writer’s original sentences throughout a passage, so that borrowed and original material become difficult for a reader to distinguish without comparing against the source directly.
See the Self-plagiarism entry for the related but distinct case of an author reusing their own previously published text without disclosure — ORI’s federal misconduct definition (42 CFR Part 93.103) does not classify self-plagiarism as research misconduct, because that regulation is scoped to appropriating another person’s words or ideas, but most journals and COPE guidance still treat undisclosed text recycling as a publication-ethics violation.
Why the text/ideas distinction matters for detection
Textual plagiarism is defined at the level of matching wording, which is exactly what similarity-checking software is built to find. Tools such as Turnitin Similarity, iThenticate, and Crossref’s Similarity Check (which runs on the iThenticate engine) compare a submitted manuscript against a reference corpus of web content, previously submitted papers, and licensed publications, and report a similarity percentage showing how much text string-matches something already in that corpus. See Anti-Plagiarism Software: How Turnitin, iThenticate, and Crossref Similarity Check Work for how these tools are actually used editorially, and Originality Report for how to read the output.
Plagiarism of ideas produces no textual match at all when the wording has been thoroughly reworded, so a similarity report cannot surface it — catching it depends on a knowledgeable reviewer or editor recognizing an uncredited argument, framework, or finding. This is also why Turnitin’s own guidance is explicit that a similarity score does not itself determine whether plagiarism occurred; a human reviewer still has to read what’s underlined and judge whether it constitutes textual plagiarism, permissible paraphrase, properly quoted material, or a false positive (shared boilerplate language, standard methods phrasing, block-quoted material that is already properly quoted and cited).
Where the numeric threshold question comes up
Because similarity tools return a percentage, authors and even some editors often ask what percentage of textual match is “acceptable.” No publisher body, funder, or the ORI’s federal misconduct definition sets a universal numeric threshold — see Acceptable Similarity Percentage for why any specific figure in circulation is a local institutional policy rather than a field-wide standard, and why a low overall score can still mask a serious instance of textual plagiarism concentrated in a few sentences.
Editorial and institutional handling
COPE’s flowcharts give editors a graduated response depending on the extent and apparent intentionality of textual plagiarism found in a submitted or published manuscript — from asking an author to properly quote and cite a short borrowed passage, through requiring a correction, to retraction for extensive, undisclosed copying. See the general Plagiarism entry for how textual plagiarism fits within the broader plagiarism spectrum COPE and ORI address, including misappropriation of unpublished ideas encountered during peer or grant review.
References
- U.S. Office of Research Integrity, “Plagiarism of Text” (ori.hhs.gov/plagiarism-text)
- 42 CFR Part 93 (ORI’s federal research-misconduct definitions)
- COPE Core Practices (2017)
- Turnitin Guides, “Understanding the similarity score”
- Crossref, “Similarity Check” documentation (crossref.org/documentation/similarity-check/)
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="Textual Plagiarism"
vocab-term-identifier="https://casrai.org/dictionary/term/textual-plagiarism" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/textual-plagiarism",
"name": "Textual Plagiarism",
"identifier": "https://casrai.org/dictionary/term/textual-plagiarism",
"description": "Textual plagiarism is unattributed copying of another person's written words — as distinct from plagiarism of ideas, where an author paraphrases someone else's argument or findings thoroughly enough that no wording matches but the underlying contribution still goes uncredited. The U.S. Office of Research Integrity (ORI) uses this exact two-way split — plagiarism of text vs. plagiarism of ideas — in its own research-integrity guidance. A passage counts as an instance of textual plagiarism when matching or near-matching wording appears without both a citation to the source AND quotation marks (or block-quote formatting) marking exactly which words are borrowed; a citation alone, without quotation marks around verbatim language, is not sufficient. Because it is defined by the words on the page rather than by intent, textual plagiarism is the category similarity-checking software (Turnitin, iThenticate, Crossref Similarity Check) is built to surface — these tools compare submitted text against a reference corpus and report string-level matches. They cannot detect plagiarism of ideas, since a thoroughly reworded passage produces no textual match at all.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/research-integrity#set",
"url": "https://casrai.org/dictionary/term/textual-plagiarism",
"sameAs": [],
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"dateModified": "2026-07-23T12:15:59",
"inLanguage": "en"
}
