Separating Evidence From Marketing
Dyslexia affects a meaningful share of any viewing audience — commonly cited estimates put it at somewhere between five and ten percent of the population, though estimates vary by definition and methodology — and captions, being entirely text, put a specific and significant demand on exactly the skill dyslexia affects. This makes caption design a genuine accessibility question, not a marginal one.
It is also an area with a considerable amount of confidently stated but weakly evidenced advice in general circulation, particularly around specific "dyslexia-friendly" fonts marketed directly to this need. It is worth being clear about what the research base actually supports before making design decisions, because some of the most commonly repeated advice is stronger as a comfort claim than as a demonstrated reading-performance improvement, and applying scrutiny here produces better outcomes than adopting whatever is most confidently marketed.
Specialized "dyslexia fonts" have not consistently demonstrated a reading-speed or comprehension advantage over well-designed conventional fonts in controlled studies, despite being widely marketed and often preferred subjectively by dyslexic readers when asked directly. Subjective preference and measured performance improvement are different findings, and conflating them is common in general advice on this topic. This does not mean typeface is irrelevant — it means the specific commercial "dyslexia font" category is not as clearly superior to ordinary well-designed sans-serif fonts as marketing for it often implies.
What the evidence more consistently supports relates to general typographic clarity principles that benefit dyslexic readers and also benefit non-dyslexic readers under difficult reading conditions — which captions, watched briefly and under time pressure, always are — rather than a single specific typeface being the deciding factor.
What Consistently Helps
Adequate letter spacing and word spacing. Cramped spacing makes letter and word boundaries harder to parse quickly, which is a real and broadly supported factor in reading ease for dyslexic readers specifically, and is more consistently supported by evidence than typeface choice itself.
Sans-serif fonts without excessive stylistic flourish, avoiding fonts with letterforms that are easily confused with each other — a lowercase l that looks identical to a capital I, for instance — which is a legibility issue affecting many readers and disproportionately affecting dyslexic readers who may already be relying more heavily on letter-shape recognition strategies that ambiguous letterforms undermine.
Sufficient size relative to viewing distance and device, since undersized captions increase the cognitive load of the reading task itself before any content processing even begins, which compounds for any reader for whom the basic decoding task already requires more effort.
High and consistent contrast between text and background, including a background treatment — a solid box or a strong drop shadow — robust to whatever is happening in the video behind the text, since inconsistent contrast forces constant re-adjustment that is a bigger tax on a reader already working harder at the decoding task.
Left-aligned text with a ragged right edge, not justified text. Justified text introduces variable word spacing to fill each line evenly, and that spacing variability specifically disrupts the consistent visual rhythm that helps readers track across a line, a factor with reasonably direct relevance to dyslexia given the specific reading mechanisms it affects.
Timing Matters More Than Typography
This is the point most general accessibility advice underweights relative to its actual importance for dyslexic viewers specifically, and it is where video captioning has genuine control that static text does not.
Reading speed limits matter more for dyslexic viewers than for the general viewing population, because decoding text takes measurably longer on average for a dyslexic reader even when comprehension, once decoded, is entirely unimpaired. A caption reading speed that is comfortable for the general population can be genuinely too fast for a meaningful share of dyslexic viewers to read in full before it disappears, even though the same text would be entirely legible and unambiguous to them given adequate time.
This is a case where standard practice and best accessible practice genuinely diverge, and it is worth being explicit about that rather than assuming meeting standard reading-speed guidelines is sufficient. Standard subtitle guidelines commonly target reading speeds designed for the general population's comfortable reading pace. A meaningfully slower rate — allowing noticeably more time per character — measurably improves completion and comprehension for dyslexic and other slower readers, at the direct cost of either showing less text per cue or requiring more cues, which has real production and editorial implications rather than being a free adjustment.
Where an accessible-specific caption track is offered as an alternative to the standard track — rather than replacing it outright — this tension can be resolved by offering both: a standard-paced track for the general audience, and a slower-paced or more heavily segmented track as an explicit alternative, similar in spirit to how audio description and standard tracks coexist rather than one replacing the other.
Consistent, predictable pacing also helps independent of raw speed. Content where caption speed varies dramatically from one segment to another - very fast during a quick exchange, very slow elsewhere - is measurably harder to adapt to than content with more even pacing throughout, since a dyslexic reader benefits particularly from a stable, predictable rhythm rather than an average speed that includes some segments far outside their comfortable range even if others compensate for it.
Line Breaking and Segmentation
Break lines at natural syntactic boundaries — the end of a clause, before a conjunction, at a natural pause point — rather than at an arbitrary character count, since forcing a reader to hold a fragment in working memory across an awkward line break adds a cognitive burden that compounds with the base difficulty of the decoding task itself for a dyslexic reader.
Shorter cues with clean boundaries generally outperform longer cues packed with more text, even at an identical overall reading speed measured in characters per second, because shorter units reduce the working-memory demand of holding a longer unbroken passage in mind while still decoding it word by word.
Avoid splitting a single grammatical unit — a name, a compound noun, a fixed phrase — across two lines or two separate cues, since this specific kind of break forces exactly the kind of cross-boundary memory work that is disproportionately taxing.
One idea per cue where content allows it, mirroring plain-language writing guidance discussed elsewhere — the same structural discipline that helps comprehension for second-language viewers and cognitively diverse viewers generally also specifically helps dyslexic readers, since much of what helps one accessibility need in caption structure tends to help several simultaneously rather than requiring entirely separate treatments.
What Users Actually Report Wanting
User research and direct feedback from dyslexic viewers consistently emphasises control — the ability to adjust font, size, background, and speed themselves — over any single fixed "correct" configuration being imposed on them, since dyslexia varies considerably in its specific manifestation and severity between individuals, and what helps one reader can be neutral or even counterproductive for another.
Platform-level and player-level caption customisation, where a viewer can adjust font, size, colour, and background independent of the content provider's default styling, is consistently reported as more valuable than any single "best" default configuration a content provider might choose on the viewer's behalf, and building or choosing a delivery pipeline that supports this customisation is a more robust investment than optimising your default caption styling for one specific accessibility need at the expense of others.
Where customisation is not available or not sufficient — burned-in captions being the clear failure case here, since they cannot be adjusted by the viewer at all — the responsibility shifts entirely to the content provider's default choices, which is a strong practical argument for soft, player-rendered captions over burned-in ones specifically where accessibility for varied reading needs is a genuine priority rather than a secondary consideration.
Practical Recommendations
Prioritise soft, toggleable, player-rendered captions that support viewer-side customisation over burned-in captions, wherever the platform and distribution channel allow it, since this single choice does more for accessibility across the full range of individual needs than any specific default styling decision could.
Where you do control default styling, favour clear sans-serif fonts with good spacing, solid contrasting backgrounds, left-aligned unjustified text, and generous rather than minimal line spacing, treating this as good general caption design rather than a specialized dyslexia-specific override, since these choices benefit the whole viewing population under real-world viewing conditions.
Treat reading speed as the highest-leverage lever specifically for dyslexic and other slower readers, and consider offering a slower-paced alternative track for content where completion and comprehension genuinely matter, rather than only adjusting typography and assuming that alone resolves the difficulty.
Break lines and segment cues at natural syntactic boundaries, applying essentially the same plain-language and structural discipline that improves comprehension broadly, since this benefit compounds across several audience needs rather than serving only one.
Do not treat a specific marketed "dyslexia font" as a solved-problem checkbox. If a genuine improvement in completion or comprehension is the goal rather than a compliance checkbox, the timing, structural, and general typographic factors above have a stronger evidence base and are worth prioritising ahead of typeface selection specifically.
A Working Checklist
- Prefer soft, player-rendered, viewer-customisable captions over burned-in captions wherever the platform allows.
- Use clear sans-serif fonts with adequate letter and word spacing as a general default rather than relying on a single marketed "dyslexia font."
- Avoid letterforms easily confused with each other in your chosen caption font.
- Set caption size adequately for typical viewing distance and device, not at a minimum.
- Use a solid or strongly contrasting background behind caption text, robust to variation in the video content behind it.
- Left-align caption text with a ragged edge; never justify subtitle text.
- Treat reading speed as the single highest-leverage factor and consider a genuinely slower alternative track for content where full comprehension matters most.
- Aim for consistent, predictable caption pacing rather than a wildly variable speed that averages out.
- Break cues at natural syntactic boundaries, not at arbitrary character counts.
- Prefer shorter cues with clean boundaries over longer, denser cues at the same overall reading speed.
- Never split a name, compound term, or fixed phrase across a line or cue boundary.
- Support and surface player-level caption customisation wherever your delivery pipeline allows it.
- Distinguish evidence-based structural and timing improvements from marketing claims when prioritising your effort.
Frequently Asked Questions
Do dyslexia-specific fonts actually improve caption readability?
The evidence is weaker than the marketing for them suggests. Controlled studies have not consistently found a measurable reading-speed or comprehension advantage for specialized "dyslexia fonts" over well-designed conventional sans-serif fonts, even though dyslexic readers often report a subjective preference for them when asked directly. Subjective preference and measured performance improvement are different findings, and general typographic clarity — spacing, contrast, unambiguous letterforms — has a more consistent evidence base than any single specific typeface.
What matters more for dyslexic viewers: font choice or caption timing?
Timing, based on the available evidence, though both matter. Decoding text takes measurably longer on average for many dyslexic readers, so a caption reading speed comfortable for the general population can be genuinely too fast to complete before the text disappears, independent of how clear the typography is. This is an area where following only standard subtitle reading-speed guidelines is not necessarily sufficient for full accessibility, and offering a slower-paced alternative track is a stronger intervention than typography adjustments alone.
Should captions ever be justified for a cleaner appearance?
No. Justified text introduces variable spacing between words to fill each line evenly, and that variability disrupts the consistent visual rhythm that helps readers, dyslexic readers in particular, track smoothly across a line. Left-aligned text with a ragged right edge is the better choice for subtitle and caption content generally, independent of any specific accessibility consideration, and this is one of the more clearly evidence-supported recommendations in this area.
Is burned-in captioning ever acceptable for accessibility purposes?
It is worse than soft, player-rendered captions specifically because it removes the viewer's ability to customise font, size, contrast, and background to their own needs — something user research consistently shows dyslexic viewers value more than any single fixed default configuration a content provider might choose. Where platform or distribution constraints require burned-in captions, the responsibility for good typographic and timing choices shifts entirely onto the content provider's fixed defaults, since the viewer has no way to adjust anything themselves.
How much slower should captions be paced for a dyslexia-accessible track?
There is no single universally agreed figure, since dyslexia's severity and specific manifestation vary considerably between individuals, but the practical approach is offering a genuinely and noticeably slower alternative track rather than a marginal adjustment to the standard pacing, and validating it with actual dyslexic viewer testing on your specific content rather than assuming a fixed percentage adjustment from standard guidelines will be adequate for the range of need in your audience.
Does improving captions for dyslexic viewers help other viewers too?
Largely yes, which is a genuinely useful property of this specific accessibility area. Clear spacing, unambiguous letterforms, consistent contrast, natural-boundary line breaks, and unjustified text all improve caption legibility for the general viewing population under real-world conditions — a small phone screen, a distracted viewing environment, a second-language viewer — even though the underlying research motivation for several of these specific recommendations comes from dyslexia-focused study.
Related reading: Subtitle Reading Speed Guide | Plain Language Video Accessibility | Video Accessibility Guide
