Hedging in academic writing means stating a claim exactly as firmly as its evidence allows. "May", "suggests" and "in this sample" are hedges; "clearly" and "shows" are boosters, and they have their place too.
On 26 September 2026 we read seven universities' guides: Manchester's Academic Phrasebank, Oxford and Warwick in the UK; George Mason and San José State in the US; Pretoria in South Africa and La Trobe in Australia. We also read two abstracts by the linguist Ken Hyland and five studies whose figures run through our examples.
How firmly to state a claim depends on how much evidence exists, and there EdCitation helps. Find sources searches by the claim itself and filters for systematic reviews and meta-analyses; Cite a source turns a DOI into a reference taken from the publisher's own record, and Verify references confirms that each entry on your list is real. Choosing the hedges stays with you: EdCitation does not write, rephrase or soften any of your sentences.
What is hedging in academic writing, and why do writers hedge?
Hedging is cautious language that makes a claim fit its evidence. Manchester's Academic Phrasebank describes the academic habit of trying to "avoid expressing absolute certainty, where there may be a small degree of uncertainty" (Morley, n.d.).
The research treats hedging as a skill. Hyland (1996), analysing 26 research articles in molecular biology, argues that hedges help a claim gain acceptance by stating it with precision, caution and deference to colleagues who may disagree. In a later study of 56 papers from eight disciplines, with interviews with expert informants, Hyland (1998) found hedges and boosters both important, and their use varying widely between fields. We read both studies at their abstracts, not in full.
University of Pretoria Library (2026) gives the practical reason: a single study seldom proves anything beyond doubt. San José State University Writing Center (2021) adds that total certainty implies no further research is needed.
What are the forms of hedging?
Hedges take seven main forms, and the guides list broadly the same ones.
| Form | Examples | Listed by |
|---|---|---|
| Modal verbs | may, might, could, would | George Mason, San José State, Pretoria, Oxford |
| Lexical verbs | suggest, indicate, appear, seem, tend | George Mason, San José State, Pretoria, Oxford |
| "Appears to", "seems to" | the effect appears to be small | Pretoria, San José State |
| Adverbs of probability | possibly, probably, perhaps | George Mason, San José State, Oxford |
| Adverbs of frequency | often, usually, generally | George Mason, Pretoria, Manchester |
| Quantifiers | some, many, most (instead of all or none) | Pretoria, San José State, George Mason |
| Limiting phrases | based on the limited data, according to this preliminary study | George Mason, San José State |
Approximators and shields
University of Oxford (n.d.) sorts hedges into two kinds. Approximators signal imprecision ("roughly", "typically", "estimate"); shields signal that the writer is not fully committed ("may", "appear", "suggest"). Use an approximator when a figure is fuzzy and a shield when the conclusion is uncertain; a shield where an approximator was meant makes a sound finding sound doubtful.
Limiting phrases: the hedge that carries information
In our view the best hedge is often a phrase that names the limit: "in adults over 50", "over 16 weeks", "in this sample". George Mason University Writing Center (2020) lists phrases of this kind. "May" says only that you are unsure; "in one 16-week trial with adults over 50" says why.
When should a claim be stated firmly?
When the evidence is strong, consistent and covers what the claim says. University of Pretoria Library (2026) keeps boosters such as "clearly" for well-established facts backed by overwhelming evidence; George Mason notes that boosters are less common than hedges; San José State says there is a time and place to sound confident. A booster is right, we suggest, when a reader who checked your sources would agree without hesitation.
A firm claim, bounded by its evidence
Taylor et al. (2014) pooled five cohort studies covering 1,256,407 children and five case-control studies covering 9,920, and found no relationship between vaccination and autism (odds ratio 0.99, 95% CI 0.92 to 1.06, in the cohort data). Their title states it flatly; their abstract says the findings "suggest" it. A student can write firmly that the meta-analysis found no association. What it does not license is a wider boost, such as "vaccines are completely safe": it was about autism.
Over-hedging: when caution becomes vagueness
University of Oxford (n.d.) warns that hedging indiscriminately makes writing sound uncertain and speculative. University of Pretoria Library (2026) names the double hedge, as in "might possibly", which sounds indecisive rather than careful. A sentence with "may", "perhaps" and "some" on one point has three.
How do you match the verb to the evidence?
Choose the verb by the design behind the claim. The table is our summary of the guides and studies in this article.
| Evidence behind the claim | Verbs that fit | Verbs that overstate |
|---|---|---|
| One small or early study | found, reported, suggests | shows that, proves |
| A meta-analysis of many trials | found, shows, indicates | proves, settles |
| An observational study | is associated with, is linked to | reduces, causes, prevents |
| A randomised controlled trial | reduced, improved (in that sample) | cures, works for everyone |
| A theory or argument | argues, proposes | shows, found |
A single study against a meta-analysis
One trial supports a narrow claim; a meta-analysis a broader one, with its own grade of confidence. Blumenthal et al. (1999) randomly assigned 156 adults aged 50 or over with major depression to aerobic exercise, sertraline, or both; after 16 weeks the groups did not differ statistically (P = .67), and the abstract lists no placebo group.
Noetel et al. (2024) combined 218 trials with 14,170 participants and found moderate reductions in depression for walking or jogging (Hedges' g = −0.62), yoga (−0.55) and strength training (−0.49), and concluded firmly that exercise is an effective treatment. Yet only one study met the Cochrane criteria for low risk of bias, and the authors rated their confidence low for walking or jogging and very low for the rest. Report the conclusion and the grade together.
Correlation against causation
A correlation supports "is associated with"; only a design that rules out other explanations supports "causes" (causality). Freedman et al. (2012) followed 229,119 men and 173,141 women aged 50 to 71. Adjusted only for age, coffee drinkers were more likely to die; they were also more likely to smoke. After adjustment for smoking and other confounding factors the association reversed: a hazard ratio of 0.88 for men drinking four or five cups a day, 0.84 for women. The authors' conclusion is a model hedge: their data cannot say whether the finding is causal.
Sumner et al. (2014) compared 462 press releases from 20 leading UK universities with the papers behind them: 33% made causal claims beyond the correlational research, and when one did, 81% of the news stories that followed did too, against 18% when it did not.
Which reporting verbs show how strong a source's claim is?
Reporting verbs tell the reader how strongly the source made its claim, and the lists disagree on where common verbs belong.
| Verb | Warwick (UK) | La Trobe (Australia) | San José State (US) | Oxford's hedging list (UK) |
|---|---|---|---|---|
| argue | strong | strong | neutral | not listed |
| claim | strong | not listed | neutral | a hedge (shield) |
| show | neutral | not listed | writer agrees | not listed |
| demonstrate | neutral | strong | writer agrees | not listed |
| find | not listed | neutral | neutral | not listed |
| suggest | tentative | tentative | not listed | a hedge (shield) |
University of Warwick (2020) calls "argues" rather than "suggests" a major difference of meaning. Coelho (2014), for San José State, sorts verbs by the writer's attitude to the source, which is why "show" sits under agreement there; La Trobe University Library (n.d.) counts "demonstrate" as strong.
Our advice: pick the verb for what the source did. "Found" and "reported" suit results, "argues" suits a position, "suggests" a finding its authors hedged. Avoid "claims" for a finding you accept, since one list reads it as strong and another as a hedge. "Prove" is in none of the four lists.
What do over-stated, over-hedged and right claims look like?
All nine sentences are ours; the studies and figures are real, read at PubMed on 26 September 2026.
Exercise and depression.
Over-stated: Exercise is as effective as antidepressants (Blumenthal et al., 1999).
Over-hedged: It could perhaps be possible that exercise might help with some symptoms of depression in some people.
Right: In a 16-week trial with 156 adults aged 50 or over, depression scores after exercise did not differ statistically from those after sertraline (Blumenthal et al., 1999); a network meta-analysis of 218 trials has since found moderate reductions for walking or jogging, yoga and strength training, with evidence its authors rated low or very low in confidence (Noetel et al., 2024).
Coffee and mortality.
Over-stated: Drinking coffee helps people live longer (Freedman et al., 2012).
Over-hedged: Coffee may perhaps be somewhat linked to a potentially lower risk of death for certain people.
Right: Among more than 400,000 adults aged 50 to 71, coffee drinkers were less likely to die during follow-up once smoking and other factors were adjusted for, although the authors state that their data cannot show whether the link is causal (Freedman et al., 2012).
Vaccines and autism.
Over-stated: Vaccines have been proven completely safe (Taylor et al., 2014).
Over-hedged: It is possible that vaccines might not necessarily be linked to autism in some studies.
Right: Vaccines are not associated with autism: a meta-analysis of ten studies covering more than 1.2 million children found no relationship (Taylor et al., 2014).
What changed:
- Over-stated: one trial generalised, an association turned into a cause, a finding about autism widened to all safety.
- Over-hedged: hedges stacked, figures dropped.
- Right: the design, the sample and the effect, then the verb that design supports. The third needs no hedge. Deciding which claims deserve judgement is in the guide to critical writing; "I" and the passive are in our guide to tone.
How do you check the strength of the claims in your draft?
Go through the draft claim by claim once the argument is settled.
- Mark each claim and the evidence behind it: one study, several, a review, or your own reasoning.
- Search the claim in Find sources and switch on the "Systematic reviews and meta-analyses" filter. A review changes how firmly you can write.
- Read the source's own conclusion, and never state it more firmly than its authors did.
- Match the verb to the design, using the table above.
- Search for double hedges ("may possibly", "might perhaps") and cut one of each pair.
- Search for boosters ("clearly", "proves", "always") and keep only those your sources carry.
- Turn each source into an entry with Cite a source, then paste the whole list into Verify references.
- Read the brief for what the argument must do; Check your paper turns it into a checklist, free.
How claims connect from sentence to sentence is in signposting and linking words.
How EdCitation helps you judge how firmly to state a claim
For the evidence side of this job EdCitation is, we think, the best tool: it shows what has been published on a claim, and the references it supplies are copied from records it located at the publisher, never composed the way a chatbot composes one. Your wording stays yours. Our runs, on 26 September 2026:
Find sources: how much evidence exists
Find sources covers about 300 million published works and needs no account or payment. "Vaccines autism" returned 23,267 results; with the "Kind of study" filter set to systematic reviews and meta-analyses, 798, led by Taylor et al. (2014). "Exercise reduces depression" went from 548,617 to 29,625, with Noetel et al. (2024) second; "coffee consumption mortality" from 51,231 to 2,234. Because the filter goes by how a paper describes itself in its title or abstract, open each result before relying on it, and search in the studies' own words: "coffee drinkers live longer" gave 63 filtered results, the top ones off the topic.
Building and checking the entries
From the Noetel DOI, Cite a source produced this APA 7 entry, quoted exactly; it reported no retraction and linked the 2024 correction:
Noetel, M., Sanders, T., Gallardo-Gómez, D., Taylor, P., del Pozo Cruz, B., van den Hoek, D., Smith, J. J., Mahoney, J., Spathis, J., Moresi, M., Pagano, R., Pagano, L., Vasconcellos, R., Arnott, H., Varley, B., Parker, P., Biddle, S., & Lonsdale, C. (2024). Effect of exercise for depression: Systematic review and network meta-analysis of randomised controlled trials. BMJ, 384, Article e075847. https://doi.org/10.1136/bmj-2023-075847
Cited the same way, Freedman et al. came back without the 2012 erratum PubMed lists, so check a key paper's PubMed page too. We then pasted five entries into Verify references. Four were verified; the fifth, where we spelt Freedman "Freeman" on purpose, came back doubtful ("check this"): "The DOI and title are right, but the first author differs. Check the names." An entry it cannot reach is marked "could not check", never "not found".
With Pro or Max: the whole paper
Once the draft is done, upload it to References from a file, which tests the whole list against the published record and flags a citation with no entry, or an entry never cited. A free account gets three runs a month; Pro ($8 a month) removes the limit and includes Mechanics QA, a rule-based pass over formatting, citations, figures and language. On Max ($24 a month), Theoretics QA weighs the finished paper against the instructions for your assignment, quotes your own sentences as evidence, and leaves the text as you wrote it (pricing). No tool here can promise a mark.
Quick questions
Is "may" enough to hedge a claim?
Usually, one hedge per claim is enough. A phrase naming the limit, such as "in this sample of 156 adults", often works better.
What is the difference between a hedge and a booster?
A hedge ("suggests", "may") weakens a claim to fit uncertain evidence; a booster ("clearly", "shows") strengthens one where the evidence is strong. Hyland found both in published papers, varying by discipline.
Can I write "proves" in an essay?
Rarely. Outside mathematics and logic a single study seldom proves anything, and none of the four verb lists we read includes "prove". Use "found" or "shows".
Is "claim" a neutral reporting verb?
Not reliably. Warwick lists it as strong, San José State as neutral and Oxford among its hedges, so a reader may hear doubt. Use "found", "reported" or "argues".
Can EdCitation hedge my sentences for me?
No. Every word of the essay, hedges included, stays yours; EdCitation does not write or edit essays. Find sources shows how much evidence exists for a claim, reviews included, so you can decide how firmly to state it.
References
- Blumenthal, J. A., Babyak, M. A., Moore, K. A., Craighead, W. E., Herman, S., Khatri, P., Waugh, R., Napolitano, M. A., Forman, L. M., Appelbaum, M., Doraiswamy, P. M., & Krishnan, K. R. (1999). Effects of exercise training on older patients with major depression. Archives of Internal Medicine, 159(19), 2349–2356. https://doi.org/10.1001/archinte.159.19.2349
- Coelho, F. (2014). Using reporting verbs [Handout]. San José State University Writing Center. https://www.sjsu.edu/writingcenter/docs/handouts/Reporting%20Verbs.pdf
- Freedman, N. D., Park, Y., Abnet, C. C., Hollenbeck, A. R., & Sinha, R. (2012). Association of coffee drinking with total and cause-specific mortality. New England Journal of Medicine, 366(20), 1891–1904. https://doi.org/10.1056/NEJMoa1112010
- George Mason University Writing Center. (2020, October 17). Hedges: Softening claims in academic writing. https://writingcenter.gmu.edu/writing-resources/research-based-writing/hedges-softening-claims-in-academic-writing
- Hyland, K. (1996). Writing without conviction? Hedging in science research articles. Applied Linguistics, 17(4), 433–454. https://doi.org/10.1093/applin/17.4.433
- Hyland, K. (1998). Boosting, hedging and the negotiation of academic knowledge. Text, 18(3), 349–382. https://doi.org/10.1515/text.1.1998.18.3.349
- La Trobe University Library. (n.d.). Reporting verbs. Writing. https://latrobe.libguides.com/writing/reporting-verbs
- Morley, J. (n.d.). Being cautious. Academic Phrasebank. The University of Manchester. https://www.phrasebank.manchester.ac.uk/using-cautious-language/
- Noetel, M., Sanders, T., Gallardo-Gómez, D., Taylor, P., del Pozo Cruz, B., van den Hoek, D., Smith, J. J., Mahoney, J., Spathis, J., Moresi, M., Pagano, R., Pagano, L., Vasconcellos, R., Arnott, H., Varley, B., Parker, P., Biddle, S., & Lonsdale, C. (2024). Effect of exercise for depression: Systematic review and network meta-analysis of randomised controlled trials. BMJ, 384, Article e075847. https://doi.org/10.1136/bmj-2023-075847
- San José State University Writing Center. (2021). Hedging, softening, and writer's distance [Handout]. https://www.sjsu.edu/writingcenter/docs/handouts/Hedging-Softening-Distance.pdf
- Sumner, P., Vivian-Griffiths, S., Boivin, J., Williams, A., Venetis, C. A., Davies, A., Ogden, J., Whelan, L., Hughes, B., Dalton, B., Boy, F., & Chambers, C. D. (2014). The association between exaggeration in health related science news and academic press releases: Retrospective observational study. BMJ, 349, Article g7015. https://doi.org/10.1136/bmj.g7015
- Taylor, L. E., Swerdfeger, A. L., & Eslick, G. D. (2014). Vaccines are not associated with autism: An evidence-based meta-analysis of case-control and cohort studies. Vaccine, 32(29), 3623–3629. https://doi.org/10.1016/j.vaccine.2014.04.085
- University of Oxford. (n.d.). Hedging. Oxford Lifelong Learning. https://lifelong-learning.ox.ac.uk/hedging/
- University of Pretoria Library. (2026, April 25). Using hedging language. Academic Writing. https://library.up.ac.za/c.php?g=1505780&p=11264954
- University of Warwick. (2020, July 15). Reporting verbs. Open House, Global PAD. https://warwick.ac.uk/fac/soc/al/globalpad-rip/openhouse/academicenglishskills/grammar/reportingverbs