A Common Scenario
A graduate student spent three weeks on a literature review. She changed every sentence from her sources — different words, same ideas. The paper came back flagged for plagiarism. She was devastated. She had changed the words. What she had not changed was the sentence structure, the argument order, or the ideas themselves — and she had cited nothing. She had been paraphrasing the way most students are taught to paraphrase, and most students are taught wrong.
This scenario is common enough to have a name in academic integrity literature: patchwriting. According to a study published in the Journal of Writing Research, patchwriting accounts for approximately 67% of student plagiarism cases. The fix is not harder synonym substitution. It is a different understanding of what paraphrasing actually requires.
Key Takeaways
- ▸Paraphrasing is not synonym substitution. Patchwriting — keeping original structure and swapping words — is classified as plagiarism by most institutions. Proper paraphrasing requires restructuring at the clause and sentence level, not just the word level.
- ▸You still must cite after paraphrasing. The International Center for Academic Integrity's 2025 survey found 58% of students incorrectly believe paraphrasing eliminates the citation requirement. It does not. Attribution is required regardless of how thoroughly wording is changed.
- ▸The “cover and write” method is the gold standard. Read the source, set it aside, write from memory, then check accuracy against the original. This process almost always produces genuinely original phrasing — not because the words change, but because you are synthesizing rather than translating.
- ▸Paraphrasing tools have a specific failure mode. Most tools, including QuillBot in standard mode, produce patchwriting — they change words but not structure. Faculty familiar with paraphrasing tools identify the output reliably. Manual paraphrasing is significantly more defensible for academic work.
- ▸Similarity percentages are not the right metric. A paper with 5% similarity and no citations can be plagiarism; a paper with 30% similarity and proper attribution is academically valid. The citations matter more than the number.
What Paraphrasing Actually Means — and Doesn't Mean
Let's establish what proper paraphrasing is, because the definition is more demanding than most students expect.
Paraphrasing is the process of restating another author's ideas in your own words and sentence structures, with attribution to the original source, for the purpose of integrating that idea into your own argument. Three elements are all required: your own words and your own structure and a citation. Remove any one element, and the result is either plagiarism or incomplete scholarship.
Patchwriting — the most common error — gets the first element partly right (some words change) but fails on structure and often on citation. The APA Style Guide (7th Edition) describes patchwriting as “a form of text that uses the original author's words or sentence structure to such a degree that the passage cannot be considered original writing, even when a citation is present.” That last clause matters: even with a citation, patchwriting can be academic misconduct because it misrepresents the author's ability to process and communicate information independently.
This distinction is increasingly relevant in 2026, where AI-generated paraphrasing tools and AI writing assistants have made synonym substitution faster than ever — while making patchwriting detection by faculty simultaneously easier, because the pattern of tool-generated paraphrasing is recognizable. According to Turnitin's 2025 Academic Integrity Insights report, patchwriting flagged in student submissions increased 31% between 2023 and 2025, correlated with the adoption of paraphrasing tools.
The Five Core Paraphrasing Techniques
Effective paraphrasing draws on multiple techniques, typically combined. No single technique is sufficient on its own. Here are the five core approaches, with examples of each.
Technique 1: Sentence Restructuring
Change the grammatical architecture of the sentence — not just the words. This includes:
- Active to passive voice (or vice versa)
- Reversing the order of main and subordinate clauses
- Converting a statement to a question or conditional
- Moving the main point from the end of the sentence to the beginning
Example: Sentence Restructuring
Original
“Researchers have found that students who receive immediate feedback on their writing show significantly greater improvement over the course of a semester than those who receive delayed feedback.”
Patchwriting (Wrong)
“Studies have discovered that learners who get instant feedback on their essays demonstrate considerably higher progress across a semester compared to those who receive late feedback.”
Same structure, only vocabulary changed — this is patchwriting.
Proper Paraphrase (Correct)
“Feedback timing has measurable effects on writing development: semester-long improvement is significantly larger when feedback is provided immediately after submission rather than delayed (Smith & Johnson, 2024).”
The clause order is reversed, the focus shifts to “feedback timing,” and the citation is present.
Technique 2: Vocabulary Substitution (Used Carefully)
Vocabulary substitution is appropriate — but as one of several techniques, not the only one. The key distinction: choose synonyms that are appropriate to the field and that carry the same technical meaning, not just any thesaurus replacement. In academic writing, “significant” has a specific statistical meaning; replacing it with “important” changes the implied claim. “Demonstrates” implies evidence-backed showing; replacing it with “suggests” weakens the claim.
Vocabulary substitution that serves the paraphrase versus vocabulary substitution that undermines it:
| Original Term | Good Substitution | Problematic Substitution | Why It Matters |
|---|---|---|---|
| Statistically significant | Reached statistical significance | Very important | Changes the technical claim |
| Demonstrates | Shows / Establishes | Suggests / Implies | Weakens or strengthens the evidential claim |
| Participants | Subjects / Respondents | People | Loses technical precision |
| Longitudinal study | Extended cohort study | Long-term research | Changes the methodological description |
Technique 3: Expansion
Break one source sentence into two or more, adding explanatory context that demonstrates your understanding of the material. This is particularly effective for technical or data-dense sentences where the original is compressed and assumes background knowledge that your readers may not have.
Expansion Example
Original (dense, compressed):
“Neural language models trained on internet-scale corpora exhibit emergent capabilities not present in smaller models (Wei et al., 2022).”
Expanded paraphrase:
“As neural language models scale up in size and training data, they begin to exhibit capabilities that were not present in smaller versions of the same architecture. Wei et al. (2022) documented this “emergence” phenomenon — the sudden appearance of abilities like multi-step reasoning and language translation at specific scale thresholds that could not have been predicted by interpolating smaller model performance.”
Technique 4: Condensation
The inverse of expansion: synthesize several related sentences from a source into one cohesive statement. Useful when integrating findings from a source whose methodology details you do not need to reproduce. Condensation requires genuine comprehension — you must understand the full argument before you can synthesize it accurately.
Condensation Example
Source (three paragraphs describing a study):
“We recruited 342 undergraduate students... We tested four conditions... Analysis of variance revealed... Post-hoc Tukey tests confirmed... These findings suggest that self-regulation strategies significantly improve academic performance across all tested conditions.”
Condensed paraphrase:
“In a controlled study of 342 undergraduates across four experimental conditions, self-regulation strategies consistently improved academic performance — a finding that held across all conditions tested (Chen et al., 2025).”
Technique 5: Perspective Shift
Reframe the original statement from the author's first-person perspective to a statement about what “the field” or “the literature” shows, or into an analytical observation about what the finding implies. This technique is powerful in literature reviews because it naturally integrates the source into your argument structure rather than just reporting what someone else found.
Perspective Shift Example
Original (author perspective):
“We concluded that climate anxiety among adolescents is significantly correlated with social media exposure to climate-related content.”
Perspective shift paraphrase:
“Adolescent climate anxiety appears to be partly mediated by digital information environments — a pattern suggesting that social media content moderation policy may have direct mental health implications (Garcia et al., 2024).”
The paraphrase integrates the finding into a broader argument about policy implication, shifting from reporting what the authors concluded to interpreting what the finding means.
The Gold Standard: Cover-and-Write Method
The most reliable method for producing genuinely original paraphrases — recommended by the Johns Hopkins University Writing Program, the Purdue Online Writing Lab, and essentially every academic writing center — is the cover-and-write method. It sounds simple because it is. Its effectiveness comes from what it forces cognitively.
Step 1: Read the source passage carefully
Read it until you can explain the main idea without looking at it. This is the comprehension step — and it is where most paraphrasing fails. If you cannot explain it without looking, you do not yet understand it well enough to paraphrase it accurately.
Step 2: Cover or close the source
Physically set aside the original — close the tab, turn the book over, move the paper. This is the constraint that makes the method work. Without the source visible, you cannot patchwrite it.
Step 3: Write the idea in your own words from memory
Write how you would explain this to a colleague — in your language, your sentence structure, your level of formality. The result almost always produces genuinely original phrasing without any deliberate effort to “change words.”
Step 4: Check accuracy against the original
Compare your paraphrase to the source to verify you have not misrepresented the idea. Accuracy is non-negotiable — a paraphrase that changes the meaning of the original to fit your argument is worse than quotation.
Step 5: Add the citation
In the format required by your style guide (APA, MLA, Chicago, or other). For paraphrases, APA 7th Edition requires at minimum Author, Year. Adding a page number is encouraged but optional for paraphrase — required for direct quotation.
Cambridge University's academic skills library describes the cover-and-write method as “the most effective single tool for producing original paraphrases” and notes that students who use it consistently show significantly lower rates of inadvertent plagiarism than those who paraphrase with the source visible. The reason is cognitive: when the source is visible, the brain optimizes for matching rather than generating. Cover it, and synthesis replaces translation.
Citations: What the Style Guides Actually Require
Paraphrasing without citation is plagiarism regardless of how thoroughly the wording has been changed. This point is not sufficiently emphasized in many writing guides. The International Center for Academic Integrity's 2025 survey of 8,000 students at 140 institutions found that 58% of students incorrectly believed that paraphrasing eliminated the citation requirement. It does not. Here is what each major style guide requires:
| Style Guide | Paraphrase In-Text Citation Format | Page Number Required? | Reference List Required? |
|---|---|---|---|
| APA 7th Edition | (Author, Year) — e.g., (Smith, 2024) | Encouraged, not required | Yes — References page |
| MLA 9th Edition | (Author Page) — e.g., (Smith 142) | Yes, when available | Yes — Works Cited page |
| Chicago Author-Date | (Author Year, Page) — e.g., (Smith 2024, 142) | Yes, when available | Yes — Bibliography |
| Harvard | (Author Year: page) — e.g., (Smith 2024: 142) | Recommended | Yes — References |
| IEEE | [Number] — e.g., [1] | No (numeric system) | Yes — References |
An important exception: common knowledge does not require citation. Facts widely known and not attributable to any specific source — that the Earth orbits the Sun, that World War II ended in 1945, that DNA carries genetic information — can be stated without attribution. The test is whether a reasonable person in your field would expect a citation. Specific data, original arguments, interpretations, and findings always require it.
Paraphrasing Tools: What They Do and What They Miss
Paraphrasing tools have become ubiquitous — QuillBot alone reports over 30 million active users as of 2025. Understanding their specific failure modes is important for using them without falling into the patchwriting trap.
Most paraphrasing tools, including QuillBot in standard mode, operate primarily through synonym substitution and limited syntactic restructuring. This addresses the surface vocabulary of patchwriting but typically preserves the clause structure and logical progression of the original sentence. Independent testing in 2026 found that QuillBot's standard mode produced output that experienced faculty identified as patchwriting in approximately 41% of academic text samples — not because the vocabulary matched the original, but because the syntactic pattern was preserved and the result “felt like the original, arranged differently.”
QuillBot's Improve mode and several dedicated academic writing tools do perform more structural changes and produce genuinely more original output — but this comes at the cost of occasionally changing the meaning. Our analysis of the best paraphrasing tools found that no automated tool reliably produces the level of structural transformation that manual cover-and-write paraphrasing achieves.
The defensible use case for paraphrasing tools is as a starting point, not a finished product. Run the source through a tool to get an initial rewrite, then apply the five techniques above to transform the output further. This hybrid approach produces more defensible paraphrases than either tool-only or scratch paraphrasing for most intermediate writers. For high-stakes academic submissions, the cover-and-write method without tool assistance is still the most defensible approach.
Common Paraphrasing Mistakes — and How to Fix Them
Beyond patchwriting, several other paraphrasing errors compromise academic integrity or writing quality:
Mistake 1: Changing Meaning Through Paraphrase
Paraphrasing that distorts the original author's meaning — even unintentionally — is a form of misrepresentation. This happens most frequently when a nuanced finding is paraphrased as an absolute claim, or when a qualified conclusion (“our findings suggest”) is rewritten as an established fact (“research proves”). Always check that your paraphrase accurately represents the author's intended claim, including their qualifications and limitations.
Mistake 2: Over-Citation (Citing Everything)
Some students, anxious about plagiarism, cite every sentence — including common knowledge and their own analytical conclusions. This undermines your credibility as a writer because it implies that every idea in your paper belongs to someone else. The goal is to cite specific claims, data, and findings that originate with specific authors. Your analysis, your synthesis, your conclusions — these should be presented as yours.
Mistake 3: Using Multiple Sources Without Distinguishing Them
A paragraph that draws on three sources but cites only one parenthetical at the end implies that all the information came from that one source. When combining multiple sources in a single paragraph, attribute each specific claim to its specific source in-text. “Smith (2024) found X, while Jones (2025) reported Y, suggesting that Z” is cleaner and more accurate than mixing the ideas and attaching one citation to the paragraph.
Mistake 4: Paraphrasing Without Understanding
This is the root cause of most paraphrasing failures. If you do not understand the source well enough to explain it without looking at it, you will either reproduce it too closely (patchwriting) or misrepresent it (meaning distortion). The solution is not a better paraphrasing technique — it is spending more time comprehending the source before attempting to paraphrase. Understanding must precede paraphrase.
Paraphrasing vs. Quoting: When to Use Each
Paraphrase is not always superior to direct quotation. Knowing when to quote directly is part of academic writing competency. The general principle: quote when the specific wording matters; paraphrase when the idea matters but the phrasing does not.
| Use Direct Quotation When: | Use Paraphrase When: |
|---|---|
| The author's exact language is being analyzed or critiqued | The idea matters but the specific phrasing does not |
| The phrasing is so precise that any change would lose meaning | The source is long and you need to integrate its main point economically |
| The original wording is iconic or widely recognized | You are synthesizing findings from multiple sources |
| You are analyzing literary or legal texts where wording is the subject | The source uses jargon or technical language your readers may not have |
| The original is already concise and clear | You want to demonstrate your comprehension of the material, not just report it |
One practical rule from the Purdue OWL guide on paraphrasing: if more than 10% of your paper consists of direct quotation, you are over-quoting. Quotation should be the exception, not the default. Heavy quotation-dependence suggests that you are reporting sources rather than engaging with them analytically.
Similarity Percentages: What Turnitin Scores Actually Mean
Because Turnitin similarity scores are the most commonly cited metric in academic integrity discussions, it is worth being precise about what they measure and what they do not.
Turnitin's similarity percentage reflects the proportion of your text that matches indexed sources in Turnitin's database — which includes previously submitted student work, internet content, and published academic papers. A high similarity score does not automatically indicate plagiarism. A paper with 35% similarity that properly quotes and cites all matching passages is academically clean. A paper with 5% similarity that paraphrases without any citation is plagiarism. The score is a starting point for investigation, not a verdict.
Most universities set informal similarity thresholds in guidance, though these vary significantly:
- Undergraduate essays: typically flagged for review above 15–20%
- Postgraduate dissertations: often stricter, with review above 10–15%
- Literature reviews: may allow 20–25% due to extensive cited quotation
- Scientific papers (IMRaD format): Methods sections typically show high similarity to standardized language — most institutions exclude boilerplate methodological phrasing from the count
For practical purposes: keep your own writing below 15% similarity, properly cite all quoted material, and understand that the number is a signal for human review, not an automatic penalty trigger. Use a plagiarism checker before submission to review your similarity report with the same lens a professor would apply — look at what is highlighted and whether it is properly attributed, not just at the percentage.
Frequently Asked Questions
Does paraphrasing avoid plagiarism?
Proper paraphrasing avoids plagiarism only when combined with a citation. Paraphrasing restates an idea in your own words — it does not make the idea yours. The International Center for Academic Integrity's 2025 survey found 58% of students incorrectly believe paraphrasing eliminates the citation requirement. It does not. Proper paraphrasing changes both wording and sentence structure, then cites the source: attribution plus transformation, not transformation alone.
What are the 5 main paraphrasing techniques?
The five core techniques are: (1) Sentence restructuring — changing grammatical architecture, not just words; (2) Vocabulary substitution — replacing terms with field-appropriate synonyms; (3) Expansion — breaking one source sentence into multiple sentences with added context; (4) Condensation — synthesizing multiple source sentences into one; and (5) Perspective shift — reframing the original claim into your argument structure. Effective paraphrasing typically combines at least three of these simultaneously.
What is the most common paraphrasing mistake?
Patchwriting — keeping original sentence structure and swapping individual words for synonyms — is the most common error and is classified as plagiarism by most institutions. A Journal of Writing Research study found patchwriting accounts for approximately 67% of student plagiarism cases. The second most common mistake is failing to cite after paraphrasing, based on the mistaken belief that changed wording eliminates the attribution requirement.
Do you need to cite a paraphrase?
Yes — always. Paraphrasing restates someone else's idea. The idea still belongs to the original author and requires attribution regardless of how thoroughly the wording has been changed. APA, MLA, and Chicago all require citation for paraphrased material. The only exception is common knowledge — facts so widely known that no single source can claim ownership. Specific data, arguments, and interpretations always require citation.
How is paraphrasing different from summarizing?
Paraphrasing restates a specific passage in different words at similar length — it conveys the same level of detail. Summarizing condenses a longer text to key points, discarding supporting detail and reducing length significantly. Both require citation. Paraphrase when the specific argument or detail matters; summarize when you need the main conclusion without the supporting structure. Both are attribution-plus-transformation, not alternatives to citation.
What percentage similarity is acceptable in academic writing?
Most universities flag for review above 15–20% for standard submissions, though thresholds vary by assignment type and institution. Dissertations often require under 10%; literature reviews may allow 20–25% when properly attributed. Similarity percentage alone is not a plagiarism indicator — properly cited quotations that raise the similarity score are academically valid. What matters is whether highlighted passages are attributed, not just the number itself.
Can I paraphrase from Wikipedia?
Technically yes — Wikipedia is CC-licensed. In practice, most academic institutions discourage or prohibit Wikipedia as a primary source because its content is crowd-edited, not peer-reviewed. Follow the citations in Wikipedia articles to the original academic or journalistic source, and cite that primary source instead. Wikipedia is a useful starting point for research orientation; it should rarely be the cited source in academic or professional writing.
How do paraphrasing tools compare to manual paraphrasing?
Most paraphrasing tools perform synonym substitution without structural change — which is patchwriting. 2026 testing found QuillBot standard mode output was still flagged as patchwriting by experienced faculty in approximately 41% of academic text samples. Manual cover-and-write paraphrasing consistently produces more original output because it forces synthesis rather than translation. For academic submissions, manual paraphrasing is significantly more defensible than tool-only output.
Check Your Paraphrased Text for Plagiarism
Before submitting, run a plagiarism check to see exactly which passages are flagged and whether your citations cover them. EyeSift's free tools also detect AI-generated content — so you get both checks in one place.