Verbal memory test: method

What it measures

Verbal recognition memory: telling words already seen in this run from new ones, as the list and the interference grow.

Procedure

Each run draws 400 words at random from a list of everyday concrete nouns in the page language (English list: 854 nouns written for this test). One word at a time, no time limit. The player answers Seen or New (buttons, or two letter keys, S and N in English). Once 3 words have been shown, each next word is a repeat with probability 0.5, chosen at random from all words shown so far (never the word just shown); otherwise a new word. Right or wrong is shown briefly. Three mistakes end the run (cap: 1,500 words). No practice. The run is not timed, so leaving the tab does not end it.

Parameters

Scoring

For each word, the truth is 'seen' if it appeared earlier in the run, otherwise 'new'. Score = number of correct answers before the third mistake; higher is better. Also: words shown, different words, mistakes. Hit and false-alarm rates or d′ are not computed. No player comparison is shown.

Valid-run rules

The truth is recomputed from the order of the words, so there is no answer key to trust. Any word not in the language's list rejects the run. Answers after the third mistake are ignored. The server rebuilds every run from the raw log (and the run's seed where there is one) and computes the score itself; anything the page says about right or wrong is ignored. A log that breaks the game's rules is rejected, and nothing is stored. Stored runs can carry quality flags that are never shown and never block a result: 10 or more runs from one IP address, a missing visitor id, tab switches, and automation signals (automated browser, impossible speed, no input events, machine-regular timing).

Differences from the classic task

Shepard and Teghtsoonian (1961) used three-digit numbers in a long fixed series and analysed recognition by the lag since the first showing. This is an open-ended game with words, random lags, a 50% repeat rate and a stop after 3 mistakes, so the score mixes memory with caution and luck in the lags. No lag-based analysis.

Limitations

Runs vary: early repeats with short lags make a run easier, an early slip makes it shorter. Word lists differ between languages, so scores are not comparable across languages. Playing often makes the list familiar, so new words can feel seen. Unsupervised setting, no norms, not a clinical test.

Languages and review status

Native lists of everyday concrete nouns for English and 57 other languages (58 in total), written for each language rather than translated from the English list. Pages in other languages use the English list.