Hedges — Interactive Module
Diagnose · Coach · Practice · Apply. Hedges soften a claim so it survives scrutiny — they mark a proposition as probable rather than certain.
Four key functions of hedges in academic writing
Hedges let a writer commit to a claim without overstating it. Overuse reads as timid; underuse reads as reckless. Each function is colour-coded and tiered by academic value.
Epistemic verbs
Lexical verbs that frame a claim as inference.
The default academic verb for a cautious inference from data.
- MedialThe findings suggest that motivation mediates the relationship between feedback and revision.
Slightly firmer than 'suggest'; still short of a causal claim.
- MedialThe regression results indicate a weak but reliable effect of class size.
Flags a pattern observed but not fully explained.
- MedialNovice writers appear to under-use concessive markers relative to experts.
Slightly more tentative than 'appear to'; common in discussion sections.
- MedialThese learners seem to rely on formulaic sequences when under time pressure.
Signals an inference the data licenses but does not state outright.
- MedialThe pattern of errors implies incomplete acquisition of the article system.
Explicitly marks the claim as beyond the evidence — useful sparingly, in a discussion section.
- MedialWe speculate that the null result reflects a ceiling effect in the pretest scores.
Choose the right marker
- indicatewhen the statistical pattern is fairly clear but the mechanism is not.
- suggestappear tofor the everyday cautious inference from your own data.
- speculateonly in a discussion section, and only once or twice.
- speculate— overuse in a results section undermines confidence in findings that are actually well supported.
Modal verbs
Grammaticalised possibility and likelihood.
The most neutral epistemic modal; safe default for a cautious claim.
- MedialThis asymmetry may reflect differences in working-memory capacity.
Slightly more tentative than 'may'; useful for a secondary or less likely explanation.
- MedialA second, less likely explanation might be sampling bias in the recruitment procedure.
Frames a claim as possible rather than probable; common when proposing alternative readings.
- MedialThe lower scores could also be attributable to test fatigue rather than ability.
Hedges a prediction or a hypothetical consequence rather than an observation.
- MedialIf the model is correct, we would expect the effect to disappear under load.
Marks an expectation derived from theory; weaker than a factual claim.
- MedialAccording to the account, proficiency should correlate with marker density.
An epistemic 'must' claims near-certainty from inference, not a modal hedge.
- MedialGiven the effect size, the difference must be due to the intervention.
Choose the right marker
- maymightfor the default epistemic hedge on a claim you believe but cannot prove.
- couldwhen proposing an alternative explanation among several.
- mustavoid as a hedge — it asserts near-certainty, the opposite of hedging.
- must— using 'must' to hedge a tentative finding overstates the evidence; prefer 'may' or 'is likely to'.
Hedging adverbs
Adverbs and adjectives that scale certainty.
A clean, low-commitment adverb for a plausible but unconfirmed reading.
- InitialPossibly, the discrepancy stems from differences in coding reliability.
Slightly more conversational than 'possibly'; common at clause-initial position.
- InitialPerhaps the strongest evidence comes from the delayed post-test.
Hedges a scalar claim by making it comparative rather than absolute.
- MedialThe sample size is relatively small for a variationist study.
Softens the degree of a claim without denying it.
- MedialThe results are somewhat consistent with the predictions of usage-based theory.
Frames a claim as defensible while flagging that it is contestable.
- InitialArguably, the most influential variable is prior exposure rather than instruction.
Conversational hedges that read as imprecise in academic prose.
- MedialThe two groups are kind of similar in their error patterns.
Choose the right marker
- possiblyperhapssentence-initial, to soften an entire claim rather than one word in it.
- relativelysomewhatto hedge a comparison or a scalar judgement.
- kind ofsort ofavoid in academic prose; too conversational.
- kind of / sort of— replace with 'somewhat' or 'to some extent' in formal writing.
Shields & approximators
Distance the claim from the writer or the number.
Attributes a claim to the literature rather than asserting it directly.
- InitialIt has been suggested that hedging density correlates with disciplinary norms.
Standard approximator for a reported quantity that is not exact.
- MedialApproximately 60% of the tokens occurred in sentence-initial position.
Scopes a claim so it does not overreach the evidence.
- FinalThe intervention improved fluency scores, to some extent.
Signals that exceptions exist without listing them.
- InitialIn general, advanced learners hedge more consistently than intermediate learners.
Shields the writer from the claim by attributing it to a named source.
- InitialAccording to Hyland (2005), hedges and boosters together constitute stance.
Distances the writer so far from the claim that it can look evasive if overused.
- InitialOne could argue that the whole methodology conflates two distinct constructs.
Choose the right marker
- it has been suggested thataccording to Xto attribute a claim rather than own it outright.
- approximatelywhenever reporting a rounded or estimated quantity.
- one could argue thatsparingly, for a genuinely contestable claim.
- one could argue that— fine once per paper for a genuinely contested point, not as a habitual opener.
Calibration & density
A hedge is a promise about how confident to be — no more, no less.
- Match the hedge to the strength of the evidence: 'indicate' for a clear pattern, 'suggest' for a weaker one.
- One hedge per claim is usually enough — stacking hedges ('may possibly suggest') reads as anxious.
- Attribute a claim explicitly ('according to X') when you want distance from a source, not from your own data.
- An epistemic 'must' or 'clearly' after a modest effect size overstates the evidence.
- Conversational hedges ('kind of', 'sort of') read as imprecise rather than cautious.
- Chronic hedging on every sentence reads as a writer who does not trust their own findings.
Place 'possibly' or 'perhaps' first when the hedge applies to the whole proposition.
Use epistemic verbs and modals inside the clause when hedging a specific predicate, not the whole sentence.