How does the Semantic Priming Test work?
Each trial starts with a cross for 500 ms, then a word in capitals for 200 ms: the prime. After a 50 ms gap a string of letters in lower case appears, and you decide as fast as you can whether it is a real English word. The prime needs nothing from you; just look at it.
There are 80 trials. In 40 the second string is a real word: half the time after a related prime (LEFT, then right), half the time after an unrelated one. In the other 40 it is a made-up word that looks real. Answer with J or → for a word, F or ← for not, or the two buttons; no answer within 2.5 seconds is a miss.
How is the Semantic Priming effect calculated?
Your priming effect is your average time on real words after an unrelated prime, minus your average after a related one, right answers only. A positive number means the related word had already woken up the next one, so you recognized it sooner. The made-up words aren't part of the effect; they are there so that answering “word” every time doesn't work.
Each string is timed from the animation frame that draws it to the moment your key or finger goes down, using the timestamp your browser gives the press. Which pairs are related changes every game, so the same target can be related one time and unrelated the next.
Where do the Semantic Priming word pairs come from?
From the University of South Florida free association norms (Douglas Nelson, Cathy McEvoy and Thomas Schreiber, 2004), in which thousands of people wrote the first word a word brought to mind. Each related pair here is a word and the answer 64% to 94% of people gave first: TEXT and book, LOW and high, GIRL and boy, TALL and short, among 60. Every word is used once, and an unrelated prime is never one the norms link to its target in either direction.
The made-up words are common words with one vowel changed (“weth”, “othur”), so they look and sound like English, and each was checked against a list of over 370,000 English words to make sure it isn't one. The lists were built on 31 October 2026 by a script that keeps its sources.
What is semantic priming?
It is how a word speeds up words related to it. David Meyer and Roger Schvaneveldt (1971) showed pairs of letter strings and asked whether both were words: BREAD and BUTTER were recognized faster than a pair with nothing in common. James Neely (1977) separated the two kinds of priming: with only a quarter of a second between prime and target, it happens automatically, before you can form any expectation. This test uses that short gap, 250 ms.
What can affect your Semantic Priming result?
How fast you read, and how at home you are in English: the norms come from students at the University of South Florida, so some pairs are stronger for some readers than others. Rushing raises errors, and errors leave fewer answers in each average. With 20 words in each group, a few slow answers can move the number, so a second run is worth it. For words and how they relate, try the Verbal Analogy Test; to see how fast you read whole passages, the Reading Speed Test.