AbilityBenchWord retrieval under a clock
Verbal fluency test scored for clusters and switches
Type as many animals as you can in sixty seconds, or as many words beginning with F, A and S in a minute each — free, no account, and the whole analysis on screen the moment the clock runs out. Alongside the count the run groups your words into the runs of related items they came out in and counts the jumps between those runs, which is the pair of numbers the task was built to produce and the pair almost every browser version drops. Every word you typed is listed back with a verdict beside it, so you can see which ones the rules removed and put any of them back.
- 100% free
- No signup
- 60 seconds a trial
- Clusters and switches scored
- Typed, no microphone
Sixty seconds a trial, one word per line, Enter after each. The count is the easy half; the run also groups your words into the pockets of memory they came out of and counts the jumps between pockets, which is the analysis this task was designed around and the one browser versions leave out.
What the clock does
The minute is measured with the same frame clock the rest of this site times reactions with, and it stops when this tab goes behind another window — so the minute you get is a minute of trying rather than a minute of wall time. Each stop is counted and printed with your result, because a paused minute with a search engine open in it is not a measurement of retrieval.
There is no untimed practice, and that is not an oversight: the clock is the measurement here, so a version without it would be a different task. The M trial is the practice — half a minute on a letter none of the scored trials uses, so nothing you warm up on is spent.
How to take the verbal fluency test
One minute, one word per line, and an analysis you can argue with afterwards.
Choose the animal minute, the three letter minutes, or all four
The animal trial asks for any animal at all and scores clustering against a tagged list. The letter trials ask for words beginning with F, then A, then S, one minute each with a break you control between them, and the three counts add up to the total the published FAS figure refers to. Take all four if you want both halves: letter fluency tracks education more closely and animal naming tracks age more closely, which is exactly why a full administration runs both instead of picking the quicker one.
Type, press Enter, keep going
Enter or a comma commits a word; a space does not, so two-word animals like polar bear go in whole. There is an Add button beside the field for a phone whose keyboard hides the return key. Do not stop to check spelling — the rules read a normalized form, so capital letters, stray punctuation and a doubled space change nothing. If you catch yourself typing a proper name or a number, let it go in and strike it out afterwards; every word is clickable on the results screen.
Read the runs, not just the total
The results screen leads with how many words were counted and what each exclusion rule removed, then draws every cluster it found as a chain: dog, cat, hamster is one run of pet-tagged animals, and the jump to eagle is a switch. Mean cluster size and switch count sit beside the total, and the four fifteen-second bins show whether the minute ran out of words or you ran out of them. Nothing is compared to a percentile, because neither published figure supports one.
Technical specifications
| Trials | Animals for 60 seconds; F, A and S for 60 seconds each, run back to back with an untimed break between them; plus an optional 30-second trial on the letter M that is scored the same way and counted nowhere |
|---|---|
| Input | Typed. Enter or a comma commits a word, a space does not, and an Add button covers a soft keyboard with no return key. A word added by button is timestamped at the last frame rather than at the tap, which costs it up to one frame of placement in the fifteen-second bins and nothing else |
| What counts, and who decides | Removed automatically: an exact repeat; a word that an earlier word from the same trial plus one of 10 listed endings spells, so walk then walking counts once; and, on a letter trial, anything not starting with that letter. Left to you: proper names and numbers, which the published procedure excludes and no program can spot, since France and Fred are ordinary strings beginning with F. Every word on the results screen is a button that overrules its verdict in either direction — the ending rule is spelling-based, so moth followed by mother is flagged and has to be put back by hand — and the total, the clusters and the switches all recompute from what is left |
| Animal list | 431 animals, each tagged with one or more of 25 subcategories — habitat, farm or pet use, taxonomic family. An animal you type that is not on the list is shown in its own group rather than deleted, because the list is ours and being absent from it is not evidence of anything |
| Clustering, animal trial | Two adjacent words join the same run when they share at least one subcategory tag. Cluster size is the number of words in a run minus one, so a word standing alone scores zero and still counts toward the mean; switches are the transitions between runs, one fewer than the number of runs |
| Clustering, letter trials | Three of the four published phonemic rules: the same first two letters, the same final three letters, or the same consonants with different vowels. The fourth rule joins homonyms and is not applied, because it needs a pronunciation dictionary this page does not ship — pare followed by pear reads as a switch here and would be one cluster on paper |
| Reference figures | ≈ 20 words for the animal minute and ≈ 40 words across the three letter minutes — Tombaugh, Kozak & Rees (1999), Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming, Archives of Clinical Neuropsychology. Both are printed as orientation with their caveats attached and neither is converted into a rank |
| What leaves the page | Nothing. The word list, the timestamps and the cluster analysis exist only in this tab; the copy button writes a plain-text summary to your own clipboard and that summary does not contain your words |
Frequently asked questions
Why is my count so much lower than the figures I have seen quoted?
Because you are typing and the published figures were spoken. A minute of speech carries far more words than a minute of typing for almost everybody, so the ceiling here is your keyboard rather than your vocabulary, and the gap is largest for the people who type slowest. There is no microphone anywhere on this site and that is deliberate rather than an oversight: this domain asks the browser for no permission at all, so the choice was a typed version with the limitation stated or no version. Compare a typed run with your own typed run a week later and the comparison holds; compare it with a clinic transcript and it does not.
Do proper names, numbers and repeats with a different ending count?
No, none of the three, and this page enforces two of them for you. An exact repeat is caught, and so is a word that an earlier word plus a listed ending spells — give walk and then walking on a letter trial and only the first is counted, because the published rule treats them as one word. Proper names and numbers are the ones no program can see: France and Fred are ordinary strings starting with F, and there is no dictionary in the page to tell them apart from fern. So every word on the results screen is a button that overrules the verdict on it, and it works in both directions: a word the rules counted drops out, and a word a rule rejected goes back in. That second direction matters because the ending rule is spelling-based and will occasionally be wrong — give both moth and mother and the second reads as the first plus an ending, which is exactly what it looks like on paper.
What do clustering and switching tell me that the total does not?
They separate two ways of arriving at the same number. Memory for this kind of material is organized in pockets — farm animals, sea creatures, words that start with the same two sounds — and producing many words needs both a deep sweep inside a pocket and a willingness to leave it when it empties. Two people who both say eighteen animals can have done something entirely different: one gave six pets, then six birds, then six fish, and the other gave eighteen unrelated animals one at a time. Mean cluster size describes the sweeping; the switch count describes the leaving. The runs are drawn on screen so you can check where the analysis put each boundary.
Why F, A and S rather than any other three letters?
Because that is the set the reference figure was collected on, and letters are not interchangeable. English has far more common words beginning with S than with F, and far fewer beginning with Q or X than either, so a total from a different set of three letters is a different quantity wearing the same label. If a tool offers you a letter of your choosing and then compares the result with an FAS figure, the comparison is broken before it starts. The optional M trial here exists for the opposite reason: it is a letter none of the scored trials uses, so warming up on it spends nothing.
My animal was not on your list. Did I lose a word?
Not from the run — it is shown in its own group on the results screen, labeled as unlisted rather than deleted, and you can see exactly which words fell there. The list holds a few hundred animals and every list of animals is arbitrary at the edges: breeds, extinct species, dinosaurs and the names of individual pets all sit somewhere in the argument. Rather than pretend to a boundary it cannot defend, this page separates what it recognized from what it did not and leaves the second group visible. What the list cannot do is put an unrecognized animal into a cluster, since clustering runs on the subcategory tags.
Why does the timer stop when I switch to another window?
Because a browser stops handing out animation frames to a tab nobody is looking at, and a clock that keeps counting through that is measuring the window manager. The minute here is a minute of the tab being in front, so switching away pauses it and coming back resumes it where it stopped. The number of stops is printed with your result, and the reason it is printed is blunter than the reason it is measured: a paused clock is also long enough to look up a list of animals, and the honest place to record that is next to the score rather than nowhere.
Can I show this number to a neurologist or a speech therapist?
As a description of what you did on a website, yes; as data, no. The published tables are split into cells by age and by years of education, and the same raw count sits in a different place in each cell — which is why this page does not ask for either and does not place you. On top of that the administration differs in every way that matters: spoken instead of typed, an examiner applying the exclusion rules live instead of a program applying three of them, and a quiet room instead of whatever is in front of you. Bring the observation that prompted you to search for this, not the number.
What clustering and switching measure, and why the count on its own is thin
The task is old and the interesting part of it is not. Asking someone to name as many animals as they can in a minute goes back to the middle of the last century and sits in most clinical batteries because it is quick and drops early in several conditions. What it produced for decades was one number, and one number turned out to be ambiguous: fluency depends both on the size of the pocket you are searching and on how readily you abandon a pocket that has run dry, and a total blurs the two together. Troyer, Moscovitch and Freedman separated them by scoring runs of words from the same subcategory as clusters and the moves between them as switches, and the separation has held up because the two components behave differently — they load on different abilities and they come apart in different conditions. That is the analysis this page runs, and it is the reason the page shows you the runs rather than only the numbers derived from them.
Three details decide whether a browser version of this is worth anything. The first is what counts: the published procedure excludes proper names, numbers, and a second form of a word already given, and a tool that counts every line you typed is inflating the score by whatever your habits happen to be. The second is the cluster convention, which is where implementations quietly disagree — a run of one word has size zero here and is included in the mean, and a tool that instead averages only over runs of two or more reports a visibly larger mean cluster size from identical input. Neither is wrong; what is wrong is printing the figure without saying which was used. The third is the letter set, because F, A and S are not a generic stand-in for “three letters” and a total built from other letters does not belong next to the FAS reference figure. If you want the vocabulary this task is drawing on measured directly rather than under a clock, the vocabulary test estimates size with a guessing correction and the synonym test probes the sense relations rather than the count.
No percentile appears anywhere on this page, and the reason is specific enough to be worth stating rather than waving at. The FAS and animal-naming norms are spoken and scored live by an examiner. A typed browser version is bounded by typing speed, which is a different ability from word retrieval, and the stratified cell values could not be verified here to the digit. So the two figures beside your result are what they are: ≈ 20 words for the animal minute and ≈ 40 words across the three letter minutes, from Tombaugh, Kozak & Rees (1999), Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming, Archives of Clinical Neuropsychology, printed as orientation and never as a rank. Retrieval is only one face of the language system, and the neighbours measure the others: recognition speed at the level of a single word is what the lexical decision task times to the millisecond, and the language processing test is the short battery that measures three levels at once and says which page to open for each. If the thing you actually wanted to know is how fast you get through text, the reading speed test pairs a rate with a comprehension check. Two neighbours sit either side of the minute itself: the sentence completion test also asks you to produce a word, but with the search narrowed by grammar and register rather than by a category or a letter, so the two together separate a retrieval problem from a knowing-which-word-fits problem. And if English is not your first language, run the english level test before this one: it places you on the CEFR scale, and a minute of retrieval in a second language measures the language long before it measures the retrieval.
One note about the clock, since every timed page on this site owes its visitor an account of its own error and this one’s account is unusual. Nothing here is a reaction time: the page reports a sixty-second window and the second each word landed in, not the interval between a stimulus and a key, so the frame-level floor that limits the millisecond figures elsewhere on this site does not apply and is not quoted. What limits this window instead is that it is measured in delivered animation frames, which makes it accurate to well under a tenth of a second across a minute, and that it stops entirely whenever the tab is not the one on screen. The count of those stops is printed with your result.
This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish dementia, aphasia, ADHD or any frontal-lobe problem. Only a qualified professional, working with more than a browser, can make that judgment.
Where the words you type go
Every number on this page is worked out by JavaScript running in the tab you are reading it in. Your answers, your reaction times and your score are never uploaded, logged or kept — which is also why the test carries on working after you disconnect from the network, and why nothing here can be held back behind an email address.
The word list lives in a JavaScript array inside this tab for as long as the tab is open, and there is no autocomplete, no spell-check request and no suggestion service behind the field — the input has spelling, autocorrect and capitalization all switched off, so the browser is not consulting anything either. Reloading the page loses a finished run, which is the trade for keeping nothing.