A screen is the most effective supplement

What it is

The most effective external store for “what can I do” is a screen: scannable, pauseable, glance-backable. A cash-machine voice that also lists the three things speakable now on the display does not force the customer to hold the list in the ear. A head-up display that parks “say: navigate / incoming call / volume” in peripheral vision is cheaper than reading the menu again. Voice still executes; the list goes to the eyes. That is a division of labour for discoverability, not the same sentence played twice.

Why it happens

Vision turns options into objects that are co-present. Scanning, jumping back, and ignoring cost no system turn. Legal wording can sit next to the option, so people read it aloud and coverage lines up with discoverability in one move. If speech carries the list itself, it must play serially, vanishes as it plays, and competes with the current task for the ear. A screen does not require the whole list to be encoded on the intro turn. It changes memory into “look when you need it.”

The supplement works only if the on-screen items can be referred to by voice, if gaze has a chance to land on that region, and if the content is speakable now, not a product poster. Wallpaper of twenty brand skills recreates undiscoverability, now in vision. The list has to follow state (signed in, account already chosen) so the eyes see the current grammar.

Studying it

Same goals, speech only versus speech plus a now-speakable list on a screen. Dependent measures: trials before the first legal wording, task completion, whether gaze actually hits the list (eye tracking or first-person video), and whether people speak what is on the screen. Independent variables: list length, whether it filters by state, whether gaze is locked by a primary task (driving versus standing at a kiosk).

Under a screen, Yankelovich’s question should be rewritten: did people learn what to say from the display, or are they still guessing. If eye tracking shows the list was never looked at, the screen is décor. A lab that turns the list into a giant crib sheet will overstate the supplement; field screens are smaller, more peripheral, and more often off the line of sight.

Where it stops holding

When both hands and eyes are locked by a primary task (hands on a machine, eyes required on the road), the screen is physically absent and this supplement fails; situated intros or a very short auditory cue are what remain. A lit screen whose type cannot be read (glare, distance, size) fails the same way. Using the screen to display the spoken menu verbatim duplicates channels and adds no new scan. Talking about a screen supplement on screenless hardware is empty; the design has to change, not pretend a display will be added later.

Applying it

• If there is a screen, ship a state-dependent speakable list, about three items, in words people say, not internal names. Do not have voice read those three items aloud as well.
• List items must be verbally addressable (“the first one,” “navigate”). If pointing at them by voice fails, this is not a discoverability supplement; it is a second broken visual.
• Screenless: do not impersonate a supplement with “see the manual.” Use situated intros or accept that those capabilities will not be found.
• How to check: video the share of sessions where the list is looked at and spoken from. Unlooked-at lists need a new place, a new moment, or deletion. First-legal-wording counts for speech-only versus with-screen should show a gap; otherwise the screen is not doing discoverability work.

Related

• Same group: M2.06.1 A voice interface has no scannable feature list · M2.06.2 Introduce capabilities at the right moment · M2.06.4 People ask for help after failure · M2.06.5 Capability intros need speakable examples · M2.06.6 “What else can you do” is a bad question to answer
• Nearby: M3.05 Voice and screen complementarity · M3.10 Multimodal complementarity of voice and screen · C7.16 Visible Feedback for Voice Input
• Search terms: screen as discoverability supplement · speakable overlay · multimodal command list