Word or Not: a lexical decision test online

A string of letters appears, such as boat or trolk. Decide as quickly as you can whether it is a real English word or a made-up one. 96 strings, half of each kind.

Start Word or Not

96 strings · about 4 min · tap, click or F and J

After the last string you see how many you sorted correctly and how quickly, and a free account keeps the result.

trolk Word Not a word F J
trolk: word or not?Made up, so Not a word · up to 3 s each

How it works

6 practice strings with feedback, then 96 scored strings with none. Each string waits at most 3 seconds for you, so the whole run takes about 4 minutes.

  1. 1 · First

    6 practice strings

    After each practice answer the screen says whether the string was a word. If fewer than half are right, the practice runs once more with a one-line reminder. None of it is scored.

  2. 2 · Each string

    A cross, then letters

    A small cross holds the centre for 400 ms. The letters replace it and stay until you answer or 3 seconds pass. A 300 ms blank comes before the next cross.

  3. 3 · Your answer

    Word or Not a word

    Two buttons sit at the bottom of the screen. Tap or click Word for a real word and Not a word for a made-up one. On a keyboard, F or the Left arrow means word, J or the Right arrow means not.

  4. 4 · The mix

    48 words, 48 made up

    The real words come in 3 bands of 16, from common to rare, and run 4 to 7 letters. The made-up strings obey English spelling, so none can be dismissed at a glance.

Once the scored part starts, nothing tells you whether you were right. A string left unanswered for 3 seconds counts as a miss and the next one appears. The order is fixed: Form A or Form B, each with its own 96 strings. On the same device a retake switches form, A then B then A, so you never meet one list twice in a row.

David Meyer and Roger Schvaneveldt introduced this kind of task in 1971, timing people as they judged whether strings of letters were words. It has since become a standard way to study how quickly a reader finds a written word in memory. Researchers often build the made-up strings with generators such as Wuggy (Keuleers and Brysbaert, 2010); the strings here were written for this check.

What it measures

Sorting words from made-up strings at speed shows how directly you recognise written words, without spelling them out letter by letter.

Real words

  • Recognising a familiar written word as a whole, which practised readers do in well under a second.
  • Word frequency: common words such as boat are found faster than rare ones, and the low minus high frequency gap shows how big that effect is for you.
  • Vocabulary and reading experience, since a word you have never read is hard to accept as real.

Made-up strings

  • Checking a string like trolk or spanet against the words you know and finding no match.
  • Rejecting with confidence, which usually takes longer than accepting: the made-up minus words gap measures it.
  • Not being fooled by strings that look and sound plausible, because they follow the rules of English spelling.

In a study of students entering higher education, Callens, Tops and Brysbaert (2012) found the largest differences between students with and without a dyslexia diagnosis on word reading, spelling and phonological tasks. In Spanish, which is spelled far more regularly than English, Suárez-Coalla and Cuetos (2015) found that adults with dyslexia still read more slowly than other adults. Speed is often where the difference shows, so the result screen weighs speed and accuracy together.

This task says nothing about reading comprehension. Someone can follow long, demanding texts and still be slow to accept or reject single strings, and the reverse happens too.

A test, not a diagnosis.

96 quick decisions describe one sitting. They cannot diagnose dyslexia or rule it out.

What you get

Your numbers appear as soon as the 96th string is over.

On the result screen

  • How many of the 96 strings you sorted correctly, and the same as a percentage.
  • Your median time on correct answers, and a speed and accuracy figure built from it, where lower is better.
  • A small table of real and made-up words, each with its accuracy and median time.
  • A plain note that there is no comparison group yet, so the screen shows your own numbers only.
  • Full data for clinicians: the frequency gap, the made-up minus words gap, how many strings got no answer, the form and the device type, plus a CSV file of every answer.

When you keep it

  • A free account saves the numbers so you can open them on any device.
  • The same task is one of the 3 in the dyslexia check, which reads it next to the other two tasks and a few questions about reading.
  • Continue takes you back to the tests hub.

How this version is scored

Each answer is stored with the string, its type, its frequency band, your choice and your time in milliseconds.

Correct
Word for a real word, Not a word for a made-up one. A string with no answer within 3 seconds is a miss.
Response time
Counted from the first frame that shows the letters to your tap or key press. Medians use correct answers between 250 and 3,000 ms, which leaves out answers given too fast to follow any reading.
Speed and accuracy
inverse efficiency = median correct time ÷ proportion correct, in milliseconds. Lower is better. Racing through with many errors cannot score well, and neither can a careful but very slow attempt.
Made-up minus words
Your median correct time on made-up strings minus your median correct time on real words. Most readers need longer to reject than to accept.
Frequency gap
Your median correct time on the 16 rarest words minus the one on the 16 most common.
Clear reading
The screen says the attempt did not give a clear reading when overall accuracy is under 65%, when accuracy on words or on made-up strings is under 50%, or when more than 20% of the strings get no answer.

Phones and computers. A touch screen adds a small delay to every tap, and screens differ in how fast they draw. So times are only compared within the same kind of device, never a phone attempt against a laptop one.

Why there is no percentile yet. No published table fits these 96 strings, so the screen shows raw numbers. Percentiles will come once 300 qualifying people per language and device type have taken this version. Until then, provisional cut-offs exist only inside the dyslexia check.

Before you start

Speed counts here, so the setting matters more than it does in an untimed task.

For a cleaner attempt

  • Set aside about 4 quiet minutes, with your glasses on if you wear them.
  • On a computer, rest two fingers on F and J before the first string.
  • Answer as soon as you know; waiting earns nothing, and after 3 seconds the string is gone.
  • Use the device you would use next time, so a later attempt can be compared.

Interpret cautiously when

  • English is not the language you learned to read in, or you read it less often than another one.
  • You were tired, rushed or interrupted, or you tapped without reading.
  • Low vision or a very small screen made the letters hard to see.
  • You switched between a phone and a computer from one attempt to the next.

Questions

What is a lexical decision task?

A timed task in which you decide whether a string of letters is a real word. Researchers have used it since the early 1970s to study how quickly people recognise written words. The usual measures are accuracy and time on correct answers, and this version also combines them into one speed and accuracy figure.

What does a slow result mean?

On its own, not much. Speed on this task also depends on your device, your vocabulary, how much you read and whether English is your first reading language. A slow result is a reason to look at the other two reading tasks and at how reading goes for you day to day, not a finding by itself.

Does Word or Not diagnose dyslexia?

No. It is one short reading task. Dyslexia is diagnosed by a qualified professional after a full assessment with a history, standard tests and more time. The dyslexia check on this site combines this task with two others and a few questions, and it cannot diagnose dyslexia either.

Can I take it on my phone?

Yes. The 2 buttons sit at the bottom of the screen within reach of a thumb. Every tap on a phone adds a little delay, so compare phone attempts only with other phone attempts.

Can I take it again?

Yes. A retake on the same device switches to the other form, so the second attempt uses a different set of 96 strings. The third attempt returns to the first form.

Sources

By Cristian Dominguez Rein-Loring, Creative Technology Lead · Updated October 20, 2026