Addressing a device in public is still socially marked

What it is

Four people face the elevator door. Someone says “remind me to call the dentist after work.” The sentence holds almost no secret, and still both the speaker and the other three flinch. Addressing a device is collusion with a device: people present cannot file the utterance under “phone call,” “talking to oneself,” or “talking to a companion.” The discomfort is that the act of address is socially marked, not that some content was understood. Harmless content still carries the mark.

Why it happens

Face-to-face talk has a visible recipient: gaze, torso, a phone against an ear. A dialogue system puts the recipient inside a box or a bud; bystanders see a mouth speaking into air. The participation framework does not parse. Others do not know whether to avert, to answer, or whether the public silence has been broken. Elevators and lift lobbies, which keep distance by not talking, make the clash hardest. A phone can still be read as “someone is on the other end.” A wake word names the other end as a machine and publishes the collusion. The cost can be paid before the mouth opens: people decide whether to use voice while judging “will this look like talking to the air.”

Studying it

Split social-acceptability ratings from content sensitivity. Film the same harmless command in an empty room, an elevator, and an open bench; judges score how strange it looks. Speakers do video-cued recall: did they feel watched. Independent variables: the place’s participation norm (talk allowed / silence expected), device stance (phone visibly held / speaking into air / a readable earpiece-call pose). Dependent measures: awkwardness and willingness to speak on the spot. Do not let “afraid a secret would be heard” items explain the whole variance.

Where it stops holding

An earpiece-and-murmur phone pose is already read as legitimate collusion, so markedness drops even when some content is still audible. Walkie-talkies, counters, and classroom roll-call have an institutional frame that recodes “talking to a device” as work. Alone in a car or an empty corridor, the mark loses its audience and the discomfort goes toward zero. Attributing every setting’s discomfort to privacy worry erases “looks like talking to oneself” as its own mechanism. Children present sometimes make device address more tolerable, because the frame can be read as play with a toy.

Applying it

• Make public wakes look like a call (phone raised, an obvious earpiece) rather than speech into air. A politer wake word will not carry this.
• Give even harmless short commands a silent equivalent (crown, keyboard, steering-wheel control) so “I will not speak in this elevator” has an exit, instead of asking people to overpower the mark.
• Do not ship demo films that only show confident talk in empty rooms. Acceptance footage should include an elevator or an equally silent space, and the voice-over should not still be selling a loud on-the-spot wake.
• How to check: in real silent spaces, show unaware colleagues the silent video and ask what the person is doing. If most say “talking to themselves” or “that looks odd,” the wake pose has not yet built a readable collusion frame.

Related

• Same group: M4.01.1 Speaking to a dialogue system hands the content to everyone in earshot · M4.01.3 Social cost drives people to abandon voice in public · M4.01.4 Bystanders pay a cost for being forced to overhear · M4.01.5 Spoken replies leak more than spoken commands · M4.01.6 Headphones privatize the reply, not the request
• Nearby: M1.01 When voice-first is appropriate · C7.06 Hands-free and eyes-free use · M4.11 Gender and stereotypes in voice assistants
• Search terms: collusion with a device · social acceptability · participation framework