Epistemic Modesty and the Art of Hedging
This session focuses on calibrating how certain your claims sound. You will learn to deploy modal verbs, tentative verbs, probability adverbs, and author-distancing frames, and to avoid the twin pitfalls of under-hedging and over-hedging.
By the end of this session, you will be able to:
- Explain the concept of epistemic modesty and why scholarly writers avoid expressing absolute certainty.
- Differentiate between unhedged categorical assertions and proportionate, evidence-grounded claims.
- Identify and deploy four key hedging devices: modal verbs, tentative verbs, probability adverbs and quantifiers, and author-distancing frames.
- Apply discipline-appropriate hedging, recognising when to hedge and when to state facts directly.
Understanding Epistemic Modesty and Hedging
Skip controls become available when the audio information has loaded. A full transcript is available below.
Show transcript
In the first two sessions, you worked on sounding formal and sounding objective. This session adds another essential skill, sounding appropriately certain. That may sound strange at first. Surely academic writing is supposed to sound confident? It is. But confidence in academic writing does not mean claiming more than your evidence can support. It means being exact about what you know, what you can reasonably infer, and what still remains uncertain. This is the purpose of hedging. A hedge is not an apology, and it is not a way of making your writing weak or vague. It is a signal to the reader that you understand the limits of your own evidence.
When you write that results suggest a relationship, or that a factor may contribute to an outcome, you are not backing away from your argument. You are calibrating it. Compare these two claims. Remote work increases productivity. And the findings suggest that remote work may increase productivity in some contexts. The second version is stronger academically, even though it sounds less absolute. Why? Because it leaves room for different industries, different conditions, and future evidence that might complicate the picture.
Defining epistemic modesty
In academic discourse, writers take care not to appear certain where doubt exists, and they systematically avoid broad over-generalisations where exceptions might occur. This caution is not a weakness — it is a discipline.
What is hedging?
Hedging devices are linguistic mechanisms that mitigate or lessen the epistemological strength — the strength of knowledge — of a statement or claim. They allow a writer to signal exactly how confident the evidence justifies being, rather than defaulting to either full certainty or vague uncertainty.
Why hedge at all?
Softening the tone of assertions enables writers to make claims that are strictly proportionate to the scope and quality of the available evidence. Far from signalling weakness, epistemic hedging demonstrates academic rigour, respects the limits of the data, and leaves room for future inquiry or counter-evidence. A claim that is hedged appropriately is more defensible, not less persuasive.
| Overconfident claim | Epistemically modest claim |
|---|---|
| “This proves the theory is correct.” | “This finding supports the theory.” |
| “All participants improved after training.” | “Most participants showed improvement after training.” |
| “The policy failed because of poor design.” | “The policy’s limited success may be attributable to design factors.” |
Linguistic Mechanics of Hedging
There are four primary categories of hedging device. Strong academic writers move between all four rather than relying on just one.
1. Tentative Verbs
Replace absolute verbs with verbs that denote possibility or probability: suggests, appears to, seems, tends to, indicates.
Unhedged: “Drinking alcohol causes breast cancer in women.”
Hedged: “Some studies suggest that drinking alcohol increases the risk of breast cancer.”
2. Modal verbs
Use may, might, or could to bound possibilities rather than state them as fact.
“This inconsistency may be due to sample limitations.”
“Extreme weather could be a result of climate change.”
3. Probability adverbs and quantifiers
Soften generalisations using adverbs (generally, likely, possibly, conceivably) and bounded quantifiers (most, several, a number of).
“It is likely that these conditions are less common in urban populations.”
4. Author-distancing frames
Use passive or impersonal constructions that distance the author from an unverified proposition: It is widely believed that…, Recent research has suggested that…, There is some evidence to indicate that….
| Hedging device | Examples |
|---|---|
| Tentative verbs | suggests, appears to, seems, tends to, indicates |
| Modal verbs | may, might, could |
| Probability adverbs / quantifiers | generally, likely, possibly, conceivably, most, several |
| Author-distancing frames | it is widely believed that, recent research has suggested that |
“Sample size limitations explain the inconsistent results.”
“The intervention improves long-term patient outcomes.”
“Students prefer digital feedback to written comments.”
“I think that early intervention improves language outcomes.”
More device-spotting checks
Extra hedging practice
“The decline in attendance is due to the new timetable.”
“Climate change is the reason for the longer droughts.”
“The data prove that the training works.”
“Social media causes anxiety in teenagers.”
“Employees prefer flexible working hours.”
“Online courses are cheaper than classroom courses.”
“Bilingual children are better at multitasking.”
“The new curriculum raises exam scores in every school.”
“It could perhaps possibly be suggested that the drug might potentially reduce pain.”
“Remote work makes people more productive. It also improves their health. Companies must offer it to every employee.”
Discipline-Specific Contexts and Placement
Where hedging concentrates
Discussion and conclusion sections of research articles are particularly rich in hedging devices, especially when authors offer explanations for findings or suggest broader implications. Methods and results sections, by contrast, tend to state what was measured directly.
Disciplinary variations
| Field | Typical hedging pattern |
|---|---|
| Social sciences and humanities | Hedging used frequently, reflecting complex, multi-causal phenomena and open-ended interpretation. |
| Natural sciences | Hedging used more sparingly, reserved mainly for interpreting unexpected findings or unproven hypotheses; verified empirical data are stated directly. |
Hedging across the sections of a research article
Hedging is not spread evenly across an article. Writers use more of it where they interpret, explain or speculate, and less where they report what was done or measured.
| Section | Amount of hedging | Typical purpose |
|---|---|---|
| Introduction | Low to moderate | Reviewing earlier research; noting gaps without overstating them |
| Methods | Very low | Stating exactly what was done |
| Results | Low | Reporting what was measured or observed |
| Discussion | High | Interpreting findings, explaining causes, acknowledging limitations |
| Conclusion | Moderate to high | Suggesting implications and future research |
Hedge the interpretation, not the evidence
Results (direct): “Mean test scores rose from 52 to 61 after the intervention.”
Discussion (hedged): “This increase may reflect the additional practice time that students received.”
The measured change is stated as fact. The explanation for the change is hedged, because the study cannot fully rule out other causes.
What to hedge and what to state directly
| Usually hedged | Usually stated directly |
|---|---|
| Explanations and causes | Measured results and verified data |
| Generalisations beyond the sample | Methods and procedures |
| Predictions and implications | Established facts and definitions |
| Interpretations of unexpected findings | Descriptions of the study’s own design |
| Claims about the limits of the study | Numbers, dates and quotations |
The same finding in two disciplines
| Field | Example |
|---|---|
| Natural sciences | “The yield was lower than predicted. Impurities in the reagents may account for this difference.” |
| Humanities | “The poem can be read as an expression of the poet’s grief, although other readings are possible.” |
In both examples the observation is stated plainly. Only the explanation or interpretation is hedged.
More placement checks
Placement practice
Results: “Test scores increased by 8% after the intervention.”
“Participants in the control group reported lower stress because the control task was more relaxing.”
“Schools must adopt this programme.”
“The poem expresses the poet’s grief.”
“The reaction produced less product than predicted. Impurities caused this.”
“The study measured blood pressure in 60 adults. Blood pressure fell by 5 mmHg. The drug is responsible for this fall and will help all patients.”
Common Pitfalls: Under-Hedging vs Over-Hedging
Under-hedging
Making sweeping, categorical statements — for example, “The new policy definitely fixes urban housing issues” — leaves the author vulnerable to immediate refutation. A single counterexample can undermine the entire claim.
Over-hedging
Compounding too many hedging devices in a single sentence — for example, “It might possibly seem that the data could perhaps suggest…” — dilutes the argument until it loses all communicative value. One well-chosen hedge is stronger than four stacked together.
| Problem | Example | Better version |
|---|---|---|
| Under-hedged | “The new policy definitely fixes urban housing issues.” | “The new policy appears to improve some urban housing outcomes.” |
| Over-hedged | “It might possibly seem that the data could perhaps suggest a link.” | “The data suggest a possible link.” |
More under-hedging and over-hedging checks
Claim Calibration Workshop
Calibrate each overconfident or absolute claim using appropriate hedging devices. Aim for one well-chosen hedge per sentence rather than stacking several.
The Hedging Audit and Calibration Exercise
Rewrite the discussion-section draft below by introducing modal verbs, tentative verbs, and probability adverbs to align every claim with proper epistemic modesty. Aim for balance — avoid leaving claims unhedged, but also avoid stacking multiple hedges in a single sentence.
Session 3 Summary Sheet
Download a PDF summarising the hedging device categories, article-section placement, what to hedge and what to state directly, and the under-hedging versus over-hedging pitfalls from this session.
Session 3: Epistemic Modesty and the Art of Hedging
Hedging device categories · discipline variation · under vs over-hedging · calibration examples
The same summary is available below as HTML. The PDF has not been verified as a tagged, accessible document.
Read the summary on this page
1. Overconfident vs Epistemically Modest Claims
| Overconfident | Epistemically modest |
|---|---|
| “This proves the theory is correct.” | “This finding supports the theory.” |
| “All participants improved after training.” | “Most participants showed improvement after training.” |
| “The policy failed because of poor design.” | “The policy’s limited success may be attributable to design factors.” |
2. Four Hedging Device Categories
| Device | Examples |
|---|---|
| Tentative verbs | suggests, appears to, seems, tends to, indicates |
| Modal verbs | may, might, could |
| Probability adverbs / quantifiers | generally, likely, possibly, most, several |
| Author-distancing frames | it is widely believed that, recent research has suggested that, there is some evidence to indicate that |
3. Hedging by Article Section
| Section | Amount of hedging |
|---|---|
| Introduction | Low to moderate |
| Methods | Very low: state exactly what was done |
| Results | Low: report what was measured |
| Discussion | High: interpret, explain, acknowledge limitations |
| Conclusion | Moderate to high: implications and future research |
Social sciences and humanities hedge frequently (multi-causal phenomena); natural sciences hedge sparingly, mainly for unexpected findings or unproven hypotheses.
4. What to Hedge and What to State Directly
| Usually hedged | Usually stated directly |
|---|---|
| Explanations and causes | Measured results and verified data |
| Generalisations beyond the sample | Methods and procedures |
| Predictions and implications | Established facts and definitions |
| Interpretations of unexpected findings | The study’s own design |
| Limits of the study | Numbers, dates and quotations |
Results (direct): “Mean test scores rose from 52 to 61.”
Discussion (hedged): “This increase may reflect the additional practice time that students received.”
5. Under-Hedging vs Over-Hedging
Under-hedged: “The new policy definitely fixes urban housing issues.”
Over-hedged: “It might possibly seem that the data could perhaps suggest…”
Well-calibrated: “The results may indicate a trend.” / “The data suggest a possible link.”
Aim for one well-chosen hedge per sentence rather than stacking several.
Example Transformation
“The survey proves that students hate online lectures.”
“Survey findings suggest that a significant number of students may find online lectures less engaging than in-person sessions.”
Self-Check Before You Submit
[ ] Claims are proportionate to the evidence (avoid proves, always, all and definitely without support).
[ ] Interpretations and causes are hedged; measured data and methods are stated directly.
[ ] No more than one or two hedges per sentence.
[ ] Hedging is concentrated in the Discussion and Conclusion.
[ ] Hedging style fits the discipline.
Session 3 complete
You have covered epistemic modesty, the four hedging device categories, discipline-specific placement, and the pitfalls of under-hedging and over-hedging. Download your takeaway summary any time from the previous step.