Paraphrase accurately without changing the source's claim

An accurate paraphrase expresses a source’s meaning in wording that fits your argument. Its test is not how many words changed: it is whether the reader receives the same claim, with the same important limits, and can identify the source.

Replacing words one by one is therefore a poor method. It can retain the source’s sentence structure while quietly changing its meaning. A reliable process starts with understanding the claim, writing from notes and then comparing the result with the original.

If you are still deciding what a source should contribute to your reasoning, first see how to build a paragraph around a claim and evidence.

Record the claim before rewriting it

Read enough context to know what the passage is doing. Then make a short claim record:

Main claim: What does the author actually conclude?
Evidence: What observation, data or reasoning supports it?
Scope: Which people, places, period or conditions does it cover?
Strength: Does the source prove, find, suggest, estimate, question or merely describe?
Limits: Which qualifications stop the claim becoming broader or more certain?
Attribution: Whose idea or evidence is this?

University of Wollongong’s paraphrasing guidance defines a good paraphrase by accurate meaning, substantial rewording and appropriate referencing—not by whether it is shorter. The University of Technology Sydney similarly advises readers to identify essential meaning, draft in a new structure, compare against the source and cite it.

If you cannot fill in the claim record, do not paraphrase yet. Re-read the surrounding section, look up unfamiliar terminology and check what any pronoun, comparison or statistical value refers to.

Use a clearly fictional source

The following passage, authors, institution, pilot and numbers are invented teaching material. They are not a real study and must not be cited in an assignment.

During a six-week pilot at the fictional Westmere College Library, 68 students voluntarily registered for evening planning sessions and 42 attended at least four. The participating group’s average self-rated confidence in planning assignments increased from 2.8 to 3.4 on a five-point scale. Because registration was voluntary, there was no comparison group, and survey responses were anonymous rather than linked to individual attendance, the evaluation cannot determine whether the sessions caused the change or whether the result applies to students beyond the pilot. The findings suggest that a longer, controlled evaluation is warranted (Dale & Okoro, 2026, p. 14).

The source’s claim record is:

Claim: the pilot warrants a longer, controlled evaluation;
Evidence: group-level self-rated confidence rose from 2.8 to 3.4;
Scope: voluntary participants in one fictional six-week library pilot;
Strength: “suggest,” not “prove”;
Limits: self-selection, no comparison group, anonymous responses and no individual attendance–response link; and
Attribution: fictional authors Dale and Okoro.

The attendance figure is context, not proof that the sessions produced the confidence change. Forty-two divided by 68 is approximately 61.8%, but calculating that percentage would not resolve the study-design limits.

Compare four paraphrase attempts

Accurate paraphrase

In a fictional six-week library pilot, participants reported higher average confidence in assignment planning, but voluntary registration, the absence of a comparison group and unlinked anonymous responses prevent a causal or wider claim. Dale and Okoro (2026) therefore argue for a longer controlled evaluation.

This version changes the wording and organisation while retaining the direction of the result, the main limitations, the cautious recommendation and the attribution. It leaves out the exact scores because this sentence uses the study’s conclusion rather than the size of the change. If the size matters to the surrounding argument, include both 2.8 and 3.4.

Overstated paraphrase

Library planning sessions improved students’ confidence and should be expanded across the college (Dale & Okoro, 2026).

This is inaccurate. “Improved” turns an observed group-level change into a causal effect, “students” broadens voluntary participants to a wider population, and immediate expansion replaces the source’s recommendation for controlled evaluation.

Qualifier-losing paraphrase

Most registered students attended regularly, and confidence rose during the six-week programme (Dale & Okoro, 2026).

The attendance statement is arithmetically defensible only if “regularly” is defined as the source’s threshold of at least four sessions, but the wording shifts attention away from the actual conclusion and omits every design limitation. A true fragment can still create a misleading account of the source.

Patchwritten paraphrase

Over a six-week trial at Westmere College Library, 68 learners voluntarily signed up for evening planning workshops and 42 participated in at least four.

This swaps several words but keeps the original information order and sentence frame. It also omits attribution. Even if a citation were added, the paraphrase would remain too dependent on the source’s construction and would not show what the evidence contributes to the writer’s point.

Audit meaning with a comparison table

On a narrow screen, scroll the table sideways to read every column.

How each paraphrase treats the fictional source's claim
Feature	Source	Accurate version	Misleading risk
Population	Voluntary participants in one pilot	Keeps the group bounded to pilot participants	“Students” may imply the whole college or a general population
Result	Average self-rated confidence rose from 2.8 to 3.4	Reports higher average self-rated confidence	“Improved confidence” can imply an established causal effect
Certainty	Findings suggest further evaluation	Retains the recommendation for controlled evaluation	“Proves” or “shows conclusively” strengthens the source
Limitations	Self-selection, no comparison group and anonymous unlinked responses	Keeps the limits material to causal and wider claims	Deleting limits makes the same numbers appear more decisive
Attribution	Dale and Okoro, fictional 2026 report	Names and cites the fictional source	A reference-list entry alone may not show where borrowed ideas begin and end

The University of New England’s paraphrasing resource emphasises retaining meaning, preserving technical terms where necessary, restructuring sentence patterns, using an accurate reporting verb and referencing the source. Reporting verbs matter: “concludes,” “suggests,” “estimates” and “speculates” do not communicate the same level of confidence.

Draft from notes, then reopen the source

A practical sequence is:

Read the relevant passage and enough surrounding context to understand it.
Write the claim record in short notes without copying the source’s syntax.
Close or cover the source.
Draft a sentence that serves your paragraph’s purpose.
Reopen the source and compare claim, scope, strength, limitations and technical terms.
Add the in-text citation and confirm the full reference.
Mark any distinctive exact wording as a quotation or rewrite it more independently.

Australian National University’s using-sources guidance recommends paraphrasing from notes and changing both structure and wording. UMGC likewise states that paraphrasing uses the writer’s syntax while crediting the source and rejects simple reordering or synonym substitution.

Writing from notes does not mean trusting memory blindly. The final comparison is essential because memory may drop a negative, reverse a comparison, merge two groups or turn “may” into “will.”

Preserve technical terms and relationships

Some words should not be replaced. A discipline’s recognised term, named construct, chemical, statute, diagnostic category or statistical measure may lose precision if forced through a thesaurus. Keep the term, cite the source and rebuild the explanation around it.

Pay special attention to:

negatives: “did not differ” is not “differed slightly”;
direction: an increase cannot become a decrease;
comparison groups: “higher than the control” cannot become “high”;
population: “survey respondents” cannot become “people”;
time: a short follow-up cannot support a permanent effect;
association and causation: “was associated with” cannot become “caused”;
modality: “may,” “could” and “is likely” cannot become “will”; and
quantity: “some,” “most” and “all” are not interchangeable.

This is where paraphrasing becomes analytical work rather than vocabulary replacement.

Keep attribution visible

A paraphrase still uses someone else’s idea or evidence. Cite it in the style required by the assignment and place the citation so a reader can tell which material it supports.

UNSW’s writing guidance distinguishes paraphrase, summary and quotation but requires attribution for each. It also recommends reporting phrases that separate the source’s position from the student’s own.

When a paragraph moves among several authors, a citation at the final full stop may be ambiguous. Use names and reporting verbs where they help:

If two or more sources need to contribute to one point, see our guide to bringing two sources into one essay paragraph.

Dale and Okoro (2026) suggest that the pilot warrants further evaluation. By contrast, Chen’s fictional review questions whether self-rated confidence is an adequate outcome.

Both sources in that sentence are invented examples. In real work, every cited source must exist and support the attributed claim.

Know when to quote instead

Use a direct quotation when the exact words are themselves important—for example, a definition you will analyse, a distinctive formulation or primary-source language whose wording matters. The guide to integrating a quotation and explaining why it matters shows how to select, contextualise and analyse those words. Use quotation marks or block formatting and the location information required by your referencing style.

Paraphrase when you need to integrate a source’s idea, evidence or reasoning into your own sentence structure. Summarise when only the broader point matters. Follow the brief and disciplinary conventions; neither quotation nor paraphrase is automatically superior.

The goal is faithful, attributed use of evidence—not manipulating a similarity score or disguising copied wording. A detector result cannot tell you whether you preserved the source’s claim accurately.

Run a source-to-paraphrase check

Before submitting, ask:

Can I state the source’s main claim without looking at my version?
Did I preserve who or what the claim covers?
Did I retain important negatives, conditions, comparisons and uncertainty?
Did I accidentally upgrade association to causation?
Are technical terms accurate rather than replaced for novelty?
Is my syntax genuinely independent from the original?
Does the reporting verb match the author’s stance?
Can the reader see where the borrowed idea begins and ends?
Does every in-text citation lead to a real, retrievable reference?
Would quotation be more honest if the exact phrasing matters?

If you are still collecting evidence, keep the source wording, locator and your own reaction separate with the guide to traceable source notes. If the paraphrase is accurate but awkward, revise the sentence for flow after the meaning audit. Our guide to academic editing and proofreading explains when that becomes a language-level edit rather than a source-use problem.

If you want guided feedback on source integration in a draft you are developing, academic writing development & revision can help you compare your paraphrases with the original sources while you retain authorship of the assessed submission.