Speech output occupies the auditory channel and cannot be browsed in parallel
Aliases: auditory bottleneck · serial output · channel conflict
What it is
Speech monopolizes the auditory channel. A screen can show dozens of items side by side for the user to pick from; speech says one sentence at a time, and a second must wait for the first to finish. That seriality means two utterances cannot play at once, and a user cannot listen to one while browsing another.
Why it happens
Understanding speech requires segmenting a continuous stream into words and clauses, which depends on sustained attention and resists parallel routes. When two streams overlap, listeners typically follow one and hear the other as noise. This is why a voice assistant receiving several pending messages must queue them. Vision, by contrast, samples multiple regions at once at the cost of distributing attention; hearing has no equivalent parallel read.
Studying it
Dual-task and competing-speech paradigms are standard: have participants follow one of two simultaneous streams and measure recall accuracy, or listen while performing a visual search and watch for contention. Dependent measures include recall accuracy, completion time, and perceived workload. Control the similarity and signal-to-noise ratio of the two streams, or the source of interference cannot be attributed.
Where it stops holding
When the second stream is designed as background—low information, safely ignorable—parallel comprehension is not required and the interface need only keep the primary stream clear. Skilled listeners or highly predictable content lower the load but do not make it available. The seriality constraint bites hardest when both streams must be understood.
Applying it
- Play only one utterance that must be understood at a time; queue the rest and expose queue state.
- Make interrupt-and-resume behavior predictable so two utterances never overlap.
- Provide a visual digest of spoken content so users can bypass the auditory queue.
- Verification: construct a scenario where several messages arrive at once, check for overlapping playback, and ask users to restate each message; an omission means the channel conflict is unhandled.
Related
- Within the group: D2.06.1 Speech conveys meaning well but structure poorly · D2.06.3 Speech output must support interruption and replay
- Adjacent: D2.07 The linear, unscannable nature of hearing · D2.10 Priority arbitration among multiple alarms
- Search terms:
auditory channel·serial processing·dual-task interference