Reading Speed in a Second Language: What Actually Works
Em resumo
- Brysbaert pooled 190 studies and 18,573 participants: adults reading silently in their first language average 238 words per minute for non-fiction and 260 for fiction. Second-language readers are often at half that.
- A 2016 review in Psychological Science in the Public Interest by Rayner and colleagues concluded that speed reading trades comprehension for speed, and that the way to get through text faster while keeping comprehension is to practise reading and become a more skilled language user.
- Reading speed is a by-product of automatic word recognition. When recognising a word costs no conscious effort, speed follows.
- Repeated reading — reading the same easy passage three or four times and tracking the rate — is one of the cheapest and best-supported ways to raise reading rate.
- Reading aloud caps you at speaking rate. Brysbaert puts the average oral reading rate at 183 words per minute across 77 studies, so it is pronunciation practice rather than speed practice.
Marc Brysbaert pooled 190 studies covering 18,573 participants and found that adults reading silently in their first language average 238 words per minute for non-fiction and 260 for fiction. In a second language the same person is often at half that on comparable text, and the gap is what makes timed exams brutal and long articles exhausting.
The good news is that the mechanism is well understood. The bad news is that it does not respond to the techniques people usually try first.
Why speed reading does not transfer
Speed reading methods promise 700, 1,000, 2,000 words per minute through wider eye spans, suppressed subvocalisation and elimination of regressions.
In 2016, Keith Rayner and colleagues published a long review of the evidence in Psychological Science in the Public Interest. Their conclusion was blunt: there is a real trade-off between speed and accuracy, and it cannot be trained away. The large gains claimed by these methods come from skipping text, and comprehension drops accordingly. Their recommendation for anyone who genuinely wants to read faster was, in their own words, to practise reading and become a more skilled language user — through increased vocabulary in particular.
That is unglamorous advice, and for second-language readers it is exactly the right advice, because the constraint is different from a first-language reader's.
What actually limits you
Reading speed is downstream of automatic word recognition.
There is a large difference between "I can work out what this word means" and "I read it without noticing I read it". The first costs conscious effort and, crucially, occupies working memory. The second is free.
When enough words in a sentence are in the first category, two things happen at once. You go slowly, and by the time you reach the end of the sentence you have lost the beginning, because the effort of decoding consumed the capacity you needed to hold the structure together. This is why re-reading a sentence three times is such a common experience in a second language and so rare in a first one.
So slow reading is not a bad habit. It is a symptom, and it has two causes worth separating:
The text is too hard. If you are meeting more than about two unknown words per hundred, you are decoding rather than reading, and no technique will fix that. Check with the readability checker or just count the unknown words on a page — the method is in how to choose reading material at your level.
The words are known but not automatic. This is the interesting case, and the one that responds to training.
Measuring your rate honestly
Before training anything, get a number.
- Take a passage of known length — 500 words is convenient.
- Read it once, at a speed where you genuinely understand it.
- Time yourself in seconds.
- Words per minute = word count divided by minutes. For 500 words in 3 minutes 20 seconds, that is 500 divided by 3.33, which is 150 words per minute.
One rule: always record the difficulty alongside the number. A rate measured on a graded reader and a rate measured on a research paper are not the same measurement, and mixing them will make your progress look random.
| Rate on easy material | Where you are |
|---|---|
| Under 100 wpm | Decoding. The text is probably too hard for the measurement to mean anything. |
| 100-150 wpm | Common starting point for intermediate learners on ordinary text. |
| 150-200 wpm | Functional. Enough for most timed exams if sustained. |
| 200-250 wpm | Comfortable. Around the 238 wpm first-language average for non-fiction. |
| Above 250 wpm | Fluent for the material in question. Move up a difficulty level. |
The 238 wpm reference point is Brysbaert's pooled figure for first-language adults reading non-fiction silently; his fiction figure is 260. Those are averages across a very wide spread, not a standard.
The three things that work
1. Repeated reading
Take a passage you have already read and understood. Read it again, timed. Then again. Then again.
Because comprehension is already settled, every pass is spent purely on recognising words faster — which is precisely the thing you are trying to train. Rates rise between the first and third pass on the same text, and the interesting question is how much carries over to new text. Be sceptical of anyone quoting a precise transfer figure: the honest position is that repeated reading reliably speeds up the practised text and that transfer to new text is smaller and more variable.
It costs nothing, needs no new material, and almost nobody does it, because re-reading feels like not making progress.
2. Timed reading at an easy level
This is the structure behind Paul Nation-style speed reading courses: a set of passages of equal length, all controlled to a vocabulary level well below the learner's, read against the clock with the rate charted each time. Sonia Millett's free speed reading courses, published by Victoria University of Wellington alongside Paul Nation's other resources, are built exactly this way: every text is 500 standard words long so the timing is directly comparable pass to pass.
The controlled vocabulary matters. The whole point is to remove decoding from the equation so the only variable is recognition speed. Doing timed readings on hard text trains anxiety, not fluency.
A practical version: pick twenty passages of about 500 words at a level where you know essentially every word. Read one a day, timed, and chart it. The curve is usually flat for a week and then moves.
3. Volume
Everything above is targeted practice. Underneath it, the thing that moves the baseline is the number of words you read per year.
Automaticity comes from repeated exposure. There is no shortcut, and the arithmetic is unforgiving: twenty minutes a day at 150 words per minute is about a million words a year, which is roughly ten novels. That is the level at which reading rate changes noticeably. Fifteen minutes twice a week is not.
The mechanics of that — how easy the material should be, and how to balance it against close study — are covered in extensive vs intensive reading.
Things that do not work, or barely
Suppressing subvocalisation. The inner voice is not the bottleneck it is claimed to be, and skilled first-language readers subvocalise too. Rayner and colleagues address this directly; trying to stop it mostly produces distraction.
Widening your eye span. The perceptual span in reading is narrower than speed reading courses claim — this is one of the central findings the Rayner review is built on — and it is not trainable in the way they suggest.
Skimming as a substitute for reading. Skimming is a genuine, useful skill — for finding information, previewing structure, or deciding whether an article is worth your time. It is not fast reading. Do not confuse the two, and do not practise skimming when you meant to practise reading.
Reading aloud. Excellent for pronunciation and for internalising sentence rhythm. Useless for speed, because it caps you at speaking rate — 183 words per minute on average across the 77 oral-reading studies Brysbaert pooled, against 238 for silent non-fiction.
If you are doing this for an exam
Exam reading is a rate problem wearing a comprehension problem's clothes. IELTS Academic Reading is 2,150 to 2,750 words plus 40 questions in 60 minutes, transfer time included. TOEFL iBT reading, since ETS rebuilt it in January 2026, is around 50 items in about 30 minutes of base time. Neither is generous.
But the fix six weeks out is not a speed reading course. It is timed practice for strategy — IELTS reading time management and TOEFL reading strategy cover that — sitting on top of months of volume for the underlying rate. Candidates who do only the first plateau, reliably, because strategy redistributes an hour and does not make anyone faster.
The IELTS reading timer will tell you the words-per-minute pace your target actually requires, which is a more useful starting point than a generic goal.
For the volume half, the practical obstacle is usually supply: finding enough text at exactly the right difficulty, day after day. LingoRead generates passages at a CEFR level you choose with questions attached, which keeps the difficulty constant while you train the rate — the same principle as a controlled-vocabulary speed reading course, without having to assemble the set yourself.
Sources
- Brysbaert, M. (2019). How many words do we read per minute? A review and meta-analysis of reading rate. Journal of Memory and Language, 109, 104047. Source for 238 wpm silent non-fiction, 260 wpm fiction and 183 wpm oral reading.
- Rayner, K., Schotter, E. R., Masson, M. E. J., Potter, M. C. and Treiman, R. (2016). So much to read, so little time: how do we read, and can speed reading help? Psychological Science in the Public Interest, 17(1), 4-34. Source for the speed-accuracy trade-off, the perceptual span, and the recommendation to become a more skilled language user.
- Millett, S. Speed reading and listening fluency, Te Herenga Waka - Victoria University of Wellington. Free controlled-vocabulary speed reading courses, 500 standard words per text.
- IELTS. IELTS Academic: Reading test format. Source for the 2,150-2,750 word total and the 60-minute limit.
- ETS. TOEFL iBT test content and structure. Source for the post-January-2026 reading section timing.