Verbal Fluency Task
Version: v1 (current)
Show a target letter or category, then capture spoken word production for a fixed amount of time. The task family includes letter (phonemic) fluency tests such as FAS/COWAT and semantic (category) fluency tests such as naming as many animals as possible.
Overview
The participant sees one target: a letter (for example "F") or a category (for example "Animals"), and is asked to say out loud as many qualifying words as they can, for a fixed amount of time (60 seconds by default). The window always runs its full duration: there is no way for the participant to end it early, because the standard paradigm requires every participant to get exactly the same amount of time so that word counts are comparable. Their spoken production is captured and the researcher counts and scores it afterward.
One task instance runs one production window for one letter or category. A battery that tests several letters or categories (for example the classic FAS three-letter set: F, A, S) is built by adding several Verbal Fluency tasks to the study, each configured with its own target. See "Building a multi-letter or multi-category battery" below.
Scientific background
- Letter (phonemic) fluency, commonly FAS or COWAT, asks the participant to name as many words as possible starting with a given letter, excluding proper nouns (names of people or places) and excluding repeating the same word with a different ending (for example "run" and "running" count as one word).
- Category (semantic) fluency asks the participant to name as many members of a category as possible (for example animals, fruits, or occupations).
- The standard administration window is 60 seconds per letter or category. The primary measure is the count of valid, non-repeated words; intrusions (words that break the rule) and perseverations (repeated words) are typically scored separately. Because judging word validity requires linguistic judgment, scoring is always done by a person, never automatically by the task.
How production is captured
Word production is captured by in-task browser speech recognition (the browser's own speech recognition, running in open dictation for the whole production window). This works on Chrome and Edge with the microphone permission granted. If the browser does not support it or the participant denies the permission, the task continues and records a marker noting that this source was unavailable.
The recognized words are never shown to the participant during the task. Only the target letter or category and a countdown clock are on screen; there is no live transcript. This keeps the participant's attention on producing words rather than reading (and possibly second-guessing) an imperfect live transcript.
The session recording is always on, independent of this in-task capture. If you add a Speech-to-Text analysis to the study, the recording is transcribed on the server as well, which is generally more accurate; the production window is marked on the timeline so you can find it in the transcription review screen. When both a browser transcript and a server transcript exist, treat them as two distinct, unverified inference sources: compare them and fall back to the raw audio for anything unclear.
Fluency mode
- Letter (phonemic) (default): set a Target letter (for example "F"). Use this for FAS/COWAT-style protocols.
- Category (semantic): set a Target category (for example "Animals"). Use this for semantic fluency protocols.
Only the fields for the selected mode are shown in the configuration panel.
Timing: a fixed window, no early exit
Unlike some other production tasks on the platform, verbal fluency has no "I'm done" button during production: the window always runs for the full configured Production duration (60 seconds by default), matching the standard clinical/research administration. Once the window closes:
- Complete automatically when the timer ends (default, on): the task ends immediately and the study moves on to the next task. This is the common case and needs no participant action.
- Turned off: a Next button appears instead, and the participant (or the moderator, in a strictly moderated session) must click it to continue. An optional Delay after timer ends before Next appears setting adds a short pause before that button becomes clickable.
In a strictly moderated session, the moderator can always end the production window immediately and move on, regardless of the auto-advance setting, which is useful for a stuck or unresponsive session.
Building a multi-letter or multi-category battery
Classic protocols test several letters or categories in sequence. Build this by chaining instances:
- Add a Verbal Fluency task for the first letter (for example "Letter F") with Target letter set to F.
- Copy it for each additional letter (for example "Letter A", "Letter S").
- For a semantic battery, use Category mode instead, with one task per category (for example "Category: Animals", "Category: Fruits").
Give each instance a distinct name so the timeline markers and task index make each production window easy to find.
Reading production from the transcription review
When a Speech-to-Text analysis is configured, open the transcription review for the participation and locate the production window using the production markers on the timeline (each Verbal Fluency instance has its own start and end). Read the words produced, apply the standard scoring rules (excluding proper nouns and repeated word roots, and noting intrusions), and record the count. When only the in-task browser transcript exists, read it from the response data and cross-check against the raw audio.
Parameters
| Parameter | Default | Notes |
|---|---|---|
| Fluency mode | Letter (phonemic) | Or Category (semantic) |
| Target letter | F | Shown in letter mode |
| Target category | Animals | Shown in category mode |
| Production duration | 60000 ms | The fixed window length; the standard letter/category fluency administration is 60 seconds |
| Show microphone level meter | On | A visible cue that the microphone is capturing; not the capture channel itself |
| Complete automatically when the timer ends | On | Off reveals a Next button instead of auto-completing |
| Delay after timer ends before Next appears | 0 ms | Only applies when auto-complete is off |
Exported data
The task produces exactly one row per task run (a single production record), plus a summary file. The row carries the fluency mode, the target letter or category, the in-task browser transcript and segment count, and the production window's start and end. There is no per-word or per-trial breakdown from the task itself and no automatic word count or correctness judgment: word counting and scoring are entirely up to the researcher, using the recorded audio and transcript(s).
Validity caveats
- In-task speech recognition is effectively Chrome/Edge only and streams audio to the browser vendor; disclose this in your consent text (the platform already uses it for the Stroop and Verbal Learning tasks).
- The in-task transcript is an automated inference and can misrender unusual or rapid speech. Score from the audio recording (and the server transcript, when configured); treat the in-task transcript as a convenience cross-check, not the scientific record.
- Because there is no early-exit control, make sure participants understand from the instructions that they should keep producing words until the timer ends, even through a pause: stopping does not end the task early.