A regression is a backward eye movement to text you have already passed. Everybody makes them: in skilled adult readers roughly 10 to 15 percent of eye movements go backwards rather than forwards. So if you catch yourself re-reading, you are not doing something unusual. The useful question is which of your re-reads are repairing your understanding and which are happening because your attention was somewhere else.
What is a regression in reading?
When you read, your eyes do not glide. They stop on a word for about a quarter of a second (a fixation), jump seven to nine letters forward (a saccade), and stop again. A regression is a saccade in the opposite direction. The mechanics are covered in more detail in how your eyes move when you read.
Regressions come in two sizes, and they mean different things:
- Short regressions go back a few letters or one word. Most regressions are of this kind. They are quick corrections: the eyes overshot, or a word needed a second look.
- Long regressions go back several words, to an earlier line or to the start of the sentence. They become more frequent when the text is difficult or when comprehension has broken down.
One thing that looks like a regression but is not: the long sweep from the end of one line to the start of the next. That return sweep is part of normal reading. It matters here because when it lands on the wrong line you read the same line twice, which is a layout problem rather than a comprehension problem.
How common are regressions?
Keith Rayner's 2009 review of eye-movement research puts regressions at about 10 to 15 percent of saccades in skilled readers of English, and the 2016 review by Rayner and colleagues repeats the figure. The rate is not fixed. It rises as text gets harder, and beginning readers, less skilled readers and readers with dyslexia make more regressions than skilled adult readers do.
That gives you a realistic target. Zero regressions is not what good readers do. A good reader looks back less often on easy material and more often on hard material, and that flexibility is part of the skill.
Do regressions help comprehension?
Yes, and there is a direct experiment on it. Schotter, Tran and Rayner (2014) had 40 university students read sentences on a screen while an eye tracker followed their gaze. In one condition each word was replaced by a row of x's as soon as the eyes moved past it, so looking back showed nothing. Comprehension was worse when re-reading was blocked, and this held for straightforward sentences as well as for ambiguous ones.
The reason is easy to feel in a sentence such as "The old man the boats." Your first reading takes "old man" as a noun phrase, the sentence collapses, and your eyes go back to rebuild it with "man" as the verb. Without that backward look you would be left with a misreading.
Regressions are a repair tool. Speed-reading advice that tells you to eliminate them asks you to give up comprehension you would otherwise have had.
Why do I keep re-reading the same line?
Repeated re-reading of the same line or paragraph usually has one of four causes. Work out which one applies before you try to fix anything, because the fixes are different.
| Cause | How it feels | What helps |
|---|---|---|
| The text is hard | You read the words but the sentence does not add up, so you go back to a specific phrase. | Let it happen. Slow down, look up the unfamiliar term, and re-read on purpose. |
| Attention drifted | You reach the end of a paragraph and realise you were thinking about something else. | Shorter sessions, a set purpose before you start, a pacer, and a break when it keeps happening. |
| Habitual checking | You understood the sentence but go back anyway, to be sure. | Read easy material with a pacer and test yourself with a one-sentence recall instead of a second pass. |
| Lost your place | After a line break you land on the line you just finished, or skip one. | Narrower columns, larger line spacing, or a card or finger under the line. |
The first cause is reading working as intended. The other three are the ones worth training, and none of them is fixed by forcing your eyes forward. They are fixed by keeping your attention on the text, which is why a steady pace helps more than a fast one.
A note on symptoms. If you regularly lose your place together with headaches, double vision, blurred text or tired, sore eyes, see an optometrist. Reading drills and eye exercises do not treat vision problems. If reading has always been slow and effortful for you, an assessment for dyslexia is more useful than any technique here.
Does a pacer reduce regressions?
A pacer is anything that moves along the line at a steady rate for your eyes to follow: a finger, a pen, a card, or an underline that advances on a screen. It gives your eyes a place to be, which makes drifting off and aimless back-tracking less likely, and it makes any look-back a decision rather than a reflex.
Be clear about the evidence, though. We could not find controlled studies showing that finger or software pacing raises reading speed while keeping comprehension. Treat a pacer as an attention aid. If you notice it dragging you past sentences you did not understand, it is set too fast.
What does RSVP do to regressions?
RSVP shows one word or short group of words at a time in a fixed spot, so there is no earlier text on screen to look back at. Regressions are not reduced; they are removed. That is the main reason researchers are cautious about it. Rayner and colleagues report that RSVP comprehension is comparable to ordinary reading at around 200 words per minute and falls as the rate goes up, and the eye movements that RSVP saves account for only about a tenth of reading time in the first place.
In practice this means three things when you use an RSVP reader:
- Keep it for easier material, or for a first pass where the gist is enough.
- Use pause and step-back as your regressions. Going back a chunk is not cheating.
- For anything dense, prefer a mode that keeps the full text visible, so your eyes can still go back on their own.
How to practise: a two-week plan for needless re-reading
The aim is fewer regressions caused by drift and habit, with the useful ones left alone. Ten minutes a day is enough. This is a practical routine, not a research-tested dose.
- Day 1: take a baseline. Read an ordinary article for five minutes with a pencil in hand. Each time you go back more than a few words, make a mark: M if you had lost the meaning, D if you had drifted or were only checking. Count both.
- Days 2 to 7: paced reading on easy text.Pick material you find easy. Preview the headings for 30 seconds, then read for five minutes with a pacer at about your normal speed; for most adults that is somewhere near 240 words per minute for non-fiction. In Tachi's Pacer mode, start at 225 or 250 WPM.
- Apply one rule.You may go back only when you can say what you missed ("I don't know who 'they' refers to"). If you cannot name it, keep going. On a screen, pause first, then step back.
- Recall instead of re-reading. At the end of each section, look away and say one sentence about what it said. If you can, the urge to check was habit. If you cannot, re-read; that regression was needed.
- Days 8 to 14: raise the difficulty, not just the speed. Move to your real reading material. Raise the pace by 25 WPM only after two sessions in a row where your recall sentences came easily, and drop back as soon as they do not.
- Day 14: repeat the baseline.Same kind of text, same five minutes. Expect the D marks to fall. The M marks should stay roughly where the text's difficulty puts them.
If attention is the main problem, the wider habits in how to read faster matter more than any eye technique: a clear purpose, the right level of difficulty, and no phone within reach. And if you are wondering how far this can go, does speed reading work sets out what the research supports. Definitions of the terms used here are in the glossary.
Frequently asked questions
Is re-reading a bad habit?
Not in itself. Looking back is how readers repair a misread word or a sentence that did not make sense, and blocking it in experiments makes comprehension worse. It becomes a problem only when you go back without a reason, because your mind wandered or because you do not trust that you understood. Those are the cases worth working on.
How many regressions are normal when reading?
In skilled adult readers of English, about 10 to 15 percent of eye movements go backwards. Most are small, to the previous word. The share rises with difficult text and is higher in beginning readers, less skilled readers and readers with dyslexia. A rate of zero is not normal and is not a sensible target.
Why do I read the same line twice?
Landing on the line you just finished is usually a return-sweep error, not a comprehension problem. It is more likely with long lines, tight line spacing or when you are tired. A narrower column, more line spacing or a card under the line usually helps. If it comes with headaches, blur or double vision, see an optometrist.
Do speed-reading apps stop regressions?
Apps that show one word at a time remove regressions completely, because there is no earlier text on screen. That is a cost as well as a saving: comprehension tends to fall at higher speeds when you cannot look back. Choose an app that lets you pause and step back, or one that keeps the full text visible.
Can I train myself to stop regressing?
You can reduce the needless ones. Read at a steady pace with a pacer, set a purpose before you start, and check your understanding with a one-sentence recall instead of a second pass. Keep going back whenever you can name what you missed. Expect a modest change in focus and flow, not a dramatic change in speed.
Sources
- Rayner (2009), Quarterly Journal of Experimental Psychology, 62(8), 1457–1506. Eye movements and attention in reading, scene perception, and visual search
- Schotter, Tran & Rayner (2014), Psychological Science, 25(6), 1218–1226. Don't believe what you read (only once): Comprehension is supported by regressions during reading
- UC San Diego (2014), news release on Schotter, Tran & Rayner. Don't Believe What You Read (Only Once): Speed-reading apps may impair reading comprehension
- Rayner, Schotter, Masson, Potter & Treiman (2016), Psychological Science in the Public Interest, 17(1), 4–34. So much to read, so little time: How do we read, and can speed reading help?
- Brysbaert (2019), Journal of Memory and Language, 109, 104047. How many words do we read per minute? A review and meta-analysis of reading rate