Should Researchers Paraphrase Online Posts to Reduce Searchability?

If exact quotations can identify people, should researchers rewrite them?

A researcher wants to protect the identity of people whose online posts appear in a study. Usernames are removed, but several proposed quotations can be copied into a search engine and traced back to the original accounts.

One solution seems obvious: paraphrase the posts.

Changing the wording can disrupt exact-string searches and reduce one important route to identification. Yet researchers are also expected to represent their evidence accurately. If the study analyzes rhetoric, discourse, grammar, hashtags, humor, slang, or other features of language, rewriting a post may destroy the very data being analyzed.

Paraphrasing is therefore a useful protection in some online research, but it should be a methodological decision rather than an automatic anonymization ritual.

Paraphrase when it meaningfully reduces risk without undermining the analysis

Paraphrasing trades textual fidelity for greater protection from traceability

Why researchers paraphrase online material

Verbatim online text can be unusually identifying because the same wording may remain indexed on a website, forum, or social media platform.

ESRC work on social media research has specifically warned that direct quotations can be traced to their original authors through search engines, creating risks for anonymity. Current UKRI guidance likewise identifies online identifiability as an important ethical concern and recommends protecting identities wherever possible unless disclosure is necessary or permission has been obtained.

Paraphrasing changes the exact sequence of words, making straightforward quotation searches less likely to locate the original source.

Paraphrasing protects against one identification route, not every route

Researchers should avoid describing paraphrasing as anonymization by itself.

A rewritten post can remain identifiable through its content. An unusual event, exact date, rare occupation, small community, location, employer, hashtag, or combination of personal details may reveal the source even when every sentence has been rewritten.

Paraphrasing can contribute to the second goal, but primarily by addressing the first.

The analytical role of language should determine whether wording can be changed

Before paraphrasing, ask what counts as evidence in the study.

If the exact language is the data, rewriting it can turn protection into distortion.

A good paraphrase should preserve meaning without reproducing a searchable fingerprint

Paraphrasing for privacy is not the same as casually summarizing what the researcher thinks a participant meant.

A defensible paraphrase should preserve the substantive content relevant to the analysis while changing enough of the linguistic surface to reduce traceability. Researchers should be particularly careful not to intensify emotional language, remove uncertainty, introduce motives, or replace participants' concepts with the researcher's preferred terminology.

For example, a post expressing uncertainty should not become a paraphrase expressing certainty merely because the latter is shorter. The privacy protection is useful only if the evidence remains faithful.

Do not place rewritten text inside quotation marks as though it were verbatim

Readers need to know whether words attributed to online users are exact or reconstructed.

If a passage has been paraphrased, it should not be presented typographically as an exact quotation unless the reporting convention clearly indicates that the wording has been altered. The methods section should explain the approach used to protect online sources.

Transparency matters because readers may otherwise reasonably assume that language inside quotation marks reproduces the source.

Paraphrasing can conceal analytically important language

Suppose a study investigates how people use euphemisms when discussing layoffs. Replacing those euphemisms with straightforward researcher language would erase the phenomenon.

Similarly, rewriting a post can remove irony, uncertainty, dialect, identity markers, rhetorical positioning, or emotional tone. These features may matter even in studies not formally labeled linguistic analysis.

Researchers should therefore decide what must remain invariant before altering text.

Exact quotation may sometimes be worth the identification risk

There are situations in which verbatim quotation is scientifically necessary or ethically appropriate.

A study may analyze public political rhetoric, institutional communication, misinformation narratives, public advocacy, or discourse in which exact wording and authorship are integral to the research question. An author may also have provided permission for identifiable quotation.

In such circumstances, researchers should not paraphrase merely to create the appearance of anonymity. They should instead address identification openly and ensure that the reporting strategy is consistent with the conditions for quoting social media material.

The more sensitive the material, the stronger the case for reducing traceability

Paraphrasing becomes especially relevant when researchers analyze pseudonymous discussions of experiences that could create significant consequences if linked to their authors.

UKRI's ethics guidance highlights risks involving privacy, reputation, employment, family relationships, sensitive behavior, and stigmatization. It also identifies potentially identifiable social media quotations, particularly around sensitive topics or closed discussions, as circumstances that may justify fuller ethics scrutiny.

The question is therefore not simply whether an online quotation can identify its author, but whether that identification would matter.

Paraphrasing should be part of a broader disclosure-control strategy

Changing wording while retaining an exact username, location, employer, rare diagnosis, event date, and community name accomplishes rather little.

Researchers should examine the entire representation of the person or post. Other safeguards may include removing unnecessary contextual details, broadening dates or locations where methodologically defensible, avoiding unnecessary platform identification, separating examples, limiting metadata, or obtaining permission for identifiable reporting.

The appropriate combination depends on the research and commitments made to participants or data subjects.

Do not fabricate a cleaner or more convenient participant voice

There is another risk at the opposite end of the spectrum. Once researchers permit themselves to alter wording for privacy, it becomes easy to “improve” a quotation.

That should be resisted.

Where substantial transformation is required to protect identity, researchers should consider whether a summary, composite illustration, or omission would be more transparent, provided the chosen approach is methodologically appropriate and clearly disclosed.

Researchers should document the transformation process

If paraphrasing is used systematically, the research team should establish a consistent approach rather than rewriting examples ad hoc during manuscript preparation.

Documentation can specify what information must be preserved, what identifying features may be altered or omitted, how transformed passages are checked against originals, and how the publication will disclose that quotations have been paraphrased.

This makes privacy protection part of the method rather than a last-minute editorial repair.

When exact wording adds identification risk but little analytical value

Here, paraphrasing removes little analytical value while reducing an obvious identification pathway. In a study analyzing the contributor's precise rhetorical choices, the same decision would be much harder to justify.

Paraphrasing is useful, but it is not an automatic solution

Decide what must remain exact before changing the words

If your online data are searchable, treat quotation strategy as part of research design rather than merely manuscript formatting.

The aim is not to make the quotation unrecognizable at any cost. It is to remove unnecessary traceability while preserving whatever features of the evidence the research genuinely needs.

Before paraphrasing online posts, protect both privacy and analytical fidelity

Questions about paraphrasing online research data

Paraphrase to remove unnecessary traceability, not necessary evidence

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.