Most adults read about 238 words a minute without thinking about it. Turn the page upside down and, in the classic studies, the same lines took about three times as long. Here is what researchers have found about reading flipped and mirrored text, and some text to look at. There is no timer and nothing to score.
Have a look first
One sentence, five ways. Read each one at your own pace.
Mara left her red umbrella on the morning train, and it rode to the coast without her.
Mara left her red umbrella on the morning train, and it rode to the coast without her.
Mara left her red umbrella on the morning train, and it rode to the coast without her.
Mraa lfet her red urebllma on the minorng trian, and it rdoe to the casot wthiout her.
araM tfel reh der allerbmu no eht gninrom niart, dna ti edor ot eht tsaoc tuohtiw reh.
Most people find the jumbled one easiest after normal, and the backwards one hardest. The studies below roughly agree.
How fast we normally read
A review of 190 studies and 18,573 readers put silent reading of non-fiction at about 238 words a minute, and fiction at about 260. Most adults fall between 175 and 300. The often quoted 300 is above average, not normal.
Upside down: about three times as long on day one
In Paul Kolers’ classic studies from the 1960s, people read 25 lines aloud. Normal text took about a minute. Upside down text took about three minutes on the first day, and about two and a half after eight days of practice, as a 2017 review sums up. These were small groups reading aloud, so treat the numbers as a rough guide.
Single letters are not the problem. In one set of experiments with Hebrew letter strings, a lone upside down letter was recognised almost as fast as an upright one. A whole upside down word took more than twice as long: about 2 seconds instead of under 1 (Koriat and Norman, 1989). The cost grew with every extra letter. The trouble seems to be the order of the letters, which now arrive back to front.
Here is the odd part. Flip text top to bottom only and it is close to unreadable. Flip it left to right as well, which turns it fully upside down, and it gets easier again. Nobody has fully explained why (Erlikhman and others, 2017).
Mirrors: slow, but still real reading
When letters are mirrored, even skilled readers slow right down, but eye tracking shows they are still reading, not solving a puzzle word by word.
Mirror reading is also a skill you keep. In a well known 1980 study, people with severe amnesia learned to read mirrored words as fast as everyone else and still had the skill three months later, even though they did not remember practising (Cohen and Squire).
Practice helps, but only for what you practise
In a small modern study of 14 people, three practice sessions cut the time to read a mirrored paragraph from about 245 seconds to about 157 (Jimura and others, 2014). Kolers found students still read pages they had practised upside down faster more than a year later.
The gain is narrow, though. Practice with one kind of changed text carried over mostly to the same kind (Masson, 1986). Getting good at mirror text will not make you good at upside down text.
Jumbled and backwards words
Jumbled middle letters slowed readers by about 11%, and moving the first letters slowed them by about 36% (Rayner and others, 2006). The famous email saying a Cambridge study proved letter order does not matter was not a Cambridge study at all. We wrote about that here.
Words spelled backwards are the hardest change in one eye tracking study. Readers almost never skipped a word and paused longer on each one (Chandra and others, 2020). Nobody has published a clean words a minute figure for it yet.
Fast is not the same as understood
A major review found that reading much faster than normal almost always means understanding less. Jumping from about 250 to 500 or more words a minute with the same understanding is unlikely (Rayner and others, 2016). That is why every nnLabs passage ends with three questions about what it said.
What nobody knows yet
Most of these numbers come from small lab groups, often reading aloud, sometimes in other languages. Why a full turn beats a single flip is still open. And there is no good speed figure for words spelled backwards.
What our readers show
Not enough runs yet. Once each script has at least 20 verified runs from at least 10 different readers, we will add anonymous totals here: typical speed in each script, compared with normal text. Our readers choose to come and choose their script, so these will be nnLabs readers so far, not a scientific sample.
Want to see your own speed?
Try one passage in the script you like. No sign up, and nothing about you is saved. Try one passage
There is also a free Monthly Challenge for adults 18+, with US$100 for the live round winner.
Sources
- Brysbaert, M. (2019). How many words do we read per minute? Journal of Memory and Language 109. doi.org/10.1016/j.jml.2019.104047
- Erlikhman, G. and others (2017). On the legibility of mirror reflected and rotated text. Symmetry 9(3), 28. doi.org/10.3390/sym9030028
- Koriat, A. and Norman, J. (1989). Journal of Experimental Psychology: Human Perception and Performance 15(1). doi.org/10.1037/0096-1523.15.1.153
- Pittrich, K. and Schroeder, S. (2023). Reading vertically and horizontally mirrored text. Quarterly Journal of Experimental Psychology 76(2). doi.org/10.1177/17470218221085943
- Cohen, N.J. and Squire, L.R. (1980). Science 210. doi.org/10.1126/science.7414331
- Jimura, K. and others (2014). Frontiers in Human Neuroscience 8, 339. doi.org/10.3389/fnhum.2014.00339
- Kolers, P.A. (1976). Pattern analyzing memory. Science 191. doi.org/10.1126/science.1257750
- Masson, M.E.J. (1986). Journal of Experimental Psychology: Learning, Memory, and Cognition 12(4). doi.org/10.1037/0278-7393.12.4.479
- Rayner, K., White, S.J., Johnson, R.L. and Liversedge, S.P. (2006). Raeding wrods with jubmled lettres. Psychological Science 17(3). doi.org/10.1111/j.1467-9280.2006.01684.x
- Chandra, J., Krügel, A. and Engbert, R. (2020). Scientific Reports 10, 4210. doi.org/10.1038/s41598-020-60833-6
- Rayner, K. and others (2016). So much to read, so little time. Psychological Science in the Public Interest 17(1). doi.org/10.1177/1529100615623267