The lexical decision method
Asking people whether they know a word produces inflated answers, because half-recognition feels like knowledge. Language researchers get around this by mixing real words with invented ones and scoring both halves — a design known as a lexical decision task, and the basis of the widely used LexTALE test. Your result here combines how many real words you accepted with how many invented ones you correctly rejected.
This is why the test cannot be gamed. Saying yes to everything gives you every real word and every fake, which scores exactly fifty percent — chance level, the same as answering at random. Saying no to everything does the same from the other direction. Only genuine discrimination between the two moves the number.
What makes an invented word convincing
The fakes are not random letters. They follow English spelling and sound patterns closely enough to be pronounceable, and they are usually built from real fragments — a genuine prefix, a familiar root shape, a common ending. Something like proquilent has nothing wrong with it structurally, which is exactly why it is a useful item.
English is unusually vulnerable to this because it borrowed so heavily from Latin, French and Greek. A large part of its formal vocabulary is assembled from parts, so a plausible new assembly is hard to tell from a real one you have simply not met — and the more academic reading you have done, the more the fakes will tempt you.
How to answer well
- Answer fast. Lexical decision is designed around first recognition, and a word you have to reason about is usually one you do not actually know.
- Ask whether you could use it, not whether you have seen it. Having seen a shape is not the same as knowing a word.
- Do not credit yourself for a familiar-looking prefix. That is precisely the reflex the invented items are testing.
- Say no when genuinely unsure. A rejected real word costs you a little; an accepted fake costs you more.
- Sixty judgments take about two minutes. Rushing the last ten because you are bored is the most common way to distort your own result.
Frequently asked questions
How is this different from the self-report vocabulary test?
The self-report test asks you to check words you know — fast, but it relies on honesty. This check makes you judge every word as real or fake, so over-claiming is punished automatically: accepting fake words lowers both your accuracy score and your verified word count. Take both and compare!
What is a good accuracy score?
50% means pure guessing. Around 60–70% is typical for intermediate learners, 80%+ suggests advanced proficiency, and 90%+ is native-like — those are the ranges reported for LexTALE-style tests in research settings.
Can I get on the leaderboard?
Yes — unlike the self-report test, this score is objective and cheat-resistant, so it feeds a global leaderboard. Submit your accuracy score with a nickname after the check.
Why do the fake English words look so convincing?
They're pseudowords — invented strings that obey normal English spelling and sound patterns, the same technique used in psycholinguistics labs. Because they feel pronounceable and vaguely familiar, gut feeling alone can't filter them out; only genuine word knowledge can. That's exactly what makes the resulting score so hard to inflate.
How much vocabulary do I need to read English comfortably?
Reading researchers estimate you need to know about 95–98% of the words on a page to read comfortably without a dictionary. That typically means several thousand of the most frequent word families for everyday texts, and more for novels and newspapers. A high accuracy score here suggests you're at or near that comfortable-reading zone.
How often should I retake the check to track progress?
Every few weeks is a good rhythm — vocabulary grows slowly, so daily retakes mostly show sampling noise rather than learning. Each attempt draws a fresh mix of real and fake words, making it a fair re-measurement. Log your accuracy over months and the upward trend becomes an honest record of progress.