Everything Counts: The Managed Omnirelevance of Speech in Human-Voice Agent Interaction
Pith reviewed 2026-05-18 04:23 UTC · model grok-4.3
The pith
Humans adapt their speech around the constant risk that voice agents might respond to any utterance.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The omnirelevance of human speech, meaning the possibility that a conversational agent may erroneously respond to any speech it detects, emerged as a constitutive feature of human-agent encounters. Humans produced their conduct in reference to this ever-present risk, and the analysis documents the practices through which they managed the artificial agents' turn-taking conduct.
What carries the argument
The omnirelevance of human speech as the ever-present risk of erroneous agent responses to any detected speech, which acts as an interactional constraint for users.
Load-bearing premise
The detailed analysis of naturalistic data from specific agents and users reveals practices generalizable to other voice agent systems and populations.
What would settle it
A study finding no evidence of speech management practices in response to erroneous agent responses across multiple voice agent interactions would falsify the centrality of this omnirelevance.
read the original abstract
To this day, turn-taking models determining voice agents' conduct have been examined primarily from a technical point of view, while the ways in which they emerge as interactional constraints or resources for human conversationalists in situ remain underexplored. Drawing on a detailed analysis of corpora of naturalistic data, we document how humans' conduct was produced in reference to the ever-present risk that, each time they spoke, their talk might trigger a new uncalled-for contribution from the artificial agent. We examine this phenomenon in interactions involving rule-based robots from a 'pre-LLM era' as well as the most recent voice agents. This 'omnirelevance of human speech' (i.e., the possibility that a conversational agent may erroneously respond to any speech it detects) emerged as a constitutive feature of these human-agent encounters. We describe some of the practices through which humans managed these artificial agents' turn-taking conduct. Given recent improvements in voice capture technology, we ask whether this 'omnirelevance of human speech' weighs even more heavily on human practices today than in the past.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper claims that the 'omnirelevance of human speech'—the ever-present risk that any human utterance may trigger an erroneous response from a voice agent—functions as a constitutive feature of human-agent encounters. Drawing on conversation-analytic examination of naturalistic corpora spanning pre-LLM rule-based robots and contemporary voice agents, the authors document interactional practices through which humans manage agents' turn-taking conduct and suggest that recent improvements in voice capture may intensify this constraint.
Significance. If the empirical grounding holds, the work offers a valuable interactional perspective on voice-agent design that complements technical turn-taking models. By treating omnirelevance as an endogenous constraint rather than a technical flaw, the analysis could inform more robust agent architectures and user-interface guidelines. The diachronic comparison across eras is a strength, as is the focus on observable participant orientations rather than post-hoc self-report.
major comments (2)
- [Data and Methods] Data and Methods: The manuscript states that the claim rests on 'detailed analysis of corpora of naturalistic data' yet supplies no information on corpus size, number of recordings or participants, selection criteria, platform diversity, or analytic procedures (e.g., transcription conventions, validation steps, or inter-analyst reliability). Without these details the constitutive status of omnirelevance cannot be evaluated against the possibility that the observed practices are tied to particular technical setups or user groups.
- [§5] §5 (Practices of Management): The description of human practices for managing omnirelevance is presented at a high level of generality. Specific excerpts illustrating how participants orient to the risk in real time, together with the sequential environments in which these orientations occur, are required to substantiate that the phenomenon is constitutive rather than occasional.
minor comments (2)
- [Introduction] The abstract and introduction use 'omnirelevance' without an explicit definition or contrast to related CA concepts such as 'relevance' or 'conditional relevance'; a brief definitional paragraph would aid readers unfamiliar with the subfield.
- [Data and Methods] Figure or table summarizing the two corpora (pre-LLM vs. recent agents) would help readers assess the scope of the comparison.
Simulated Author's Rebuttal
We thank the referee for their constructive and positive assessment of the manuscript's potential contribution, particularly the value of the interactional perspective and the diachronic comparison. We address each major comment below and will revise the manuscript to incorporate the requested details and evidence.
read point-by-point responses
-
Referee: [Data and Methods] The manuscript states that the claim rests on 'detailed analysis of corpora of naturalistic data' yet supplies no information on corpus size, number of recordings or participants, selection criteria, platform diversity, or analytic procedures (e.g., transcription conventions, validation steps, or inter-analyst reliability). Without these details the constitutive status of omnirelevance cannot be evaluated against the possibility that the observed practices are tied to particular technical setups or user groups.
Authors: We agree that the manuscript would be strengthened by greater transparency regarding the data and analytic procedures. Although the study employs conversation-analytic methods that emphasize sequential detail over statistical sampling, we will add a dedicated Data and Methods section (or expand the existing description) to specify corpus sizes, number of recordings and participants, selection criteria, platform diversity, transcription conventions, and analytic procedures. This will enable readers to assess the grounding of the claims more fully and address concerns about potential ties to specific technical setups. revision: yes
-
Referee: [§5] §5 (Practices of Management): The description of human practices for managing omnirelevance is presented at a high level of generality. Specific excerpts illustrating how participants orient to the risk in real time, together with the sequential environments in which these orientations occur, are required to substantiate that the phenomenon is constitutive rather than occasional.
Authors: We concur that concrete excerpts are essential to demonstrate the real-time, sequential orientations that support the constitutive status of omnirelevance. In the revised manuscript we will include additional transcribed excerpts from the corpora in §5, each accompanied by detailed sequential analysis showing participants' orientations to the risk of erroneous agent responses. These will illustrate the specific interactional environments in which the management practices occur and thereby strengthen the argument that omnirelevance is an endogenous feature rather than an occasional occurrence. revision: yes
Circularity Check
No circularity: observational claims rest on external naturalistic corpora
full rationale
The paper is a qualitative conversation-analytic study drawing on detailed examination of naturalistic interaction data from human-voice agent encounters. No equations, fitted parameters, self-referential derivations, or load-bearing self-citations appear in the derivation chain. The central claim about omnirelevance of speech as a constitutive feature is presented as emerging from analysis of external corpora spanning pre-LLM and recent agents; this is independent of the paper's own structure and does not reduce to its inputs by construction. Generalizability concerns are separate from circularity.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Naturalistic data corpora accurately reflect real-world interactional practices in human-agent encounters.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.