{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp05_complete_proposal_portfolio20_20260803","cell_id":"negative_space_design__linguistics_semiotics","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_id":"nsd_ls_001_silence_lane_transcript","proposal_index":1,"version":0,"title":"Silence-Lane Transcript for Adjacency-Pair Analysis","problem":"In a conversation-analysis workflow, analysts inspect recorded exchanges to classify how a response relates to an initiating move. The working transcript places turns, timestamps, overlap marks, glosses, speaker attributes, and analytic annotations into a nearly continuous block. A gap after an initiating move is therefore represented as another small token among many, even though its duration and location may be part of the interactional event. The candidate problem is not missing transcription data; it is that the display gives measured non-speech too little protected perceptual form.","actors":["Conversation analysts who interpret turn sequences","Transcribers who align speech and non-speech intervals to recordings","Corpus or project lead responsible for the transcript-view specification","Participants whose recorded interaction is represented in the corpus"],"observable_state":"For a bounded set of recorded adjacency-pair excerpts, the source alignment contains a measurable interval between the end of a first move and the next vocal contribution, but the standard analyst view collapses that interval into compact punctuation or an inline duration label surrounded by other annotations. Analysts must inspect the waveform, hover metadata, or reread the line to distinguish the gap from adjacent marks; the same surface can also leave recording loss, unmeasured silence, and measured silence insufficiently differentiated.","consequence":"Analysts may overlook the placement of silence, confuse deliberate or interactionally consequential non-response with display compression or missing data, or allow surrounding annotations to dominate judgments about response timing. This can weaken the traceability and consistency of sequence classifications without any defect in the underlying recording or alignment.","affected_objective":"Make the timing and status of response-relevant silence perceptually available while preserving the linguistic context and source evidence required to interpret it.","intervention":"Add an optional silence-lane view to the transcript renderer. When a verified non-speech interval occurs after a configured first-pair-part boundary, render a protected vertical band between the adjacent turns. Give the band a bounded height derived from duration bins rather than unlimited proportional scaling. Leave its center visually quiet and place only a small status cue at its edge distinguishing measured silence, unmeasured gap, recording discontinuity, and analyst-excluded material. Move nonessential glosses, speaker attributes, and provisional analytic codes from the band’s immediate edges into an on-demand side panel, without deleting them. Preserve the initiating turn, responding turn, overlap information, uncertainty markers, and a direct link to the synchronized audio. A visible control returns all hidden annotations, and the ordinary transcript remains available unchanged.","structural_mapping":[{"archetype_element":"Attention Competition Map","domain_realization":"Inventory which inline marks compete with perception of the interval between an initiating move and its response, separating necessary timing and uncertainty evidence from deferrable glosses and analytic codes."},{"archetype_element":"Omission Candidate","domain_realization":"Treat nearby nonessential metadata and provisional interpretation labels as candidates for temporary removal from the focal view, not as candidates for deletion from the corpus."},{"archetype_element":"Protected Empty Space","domain_realization":"Reserve a bounded vertical silence band that transcript text and optional annotations cannot occupy."},{"archetype_element":"Positive Form Relationship","domain_realization":"Tie the empty band specifically to the preceding initiating move and following contribution, so the absence frames their sequential relationship."},{"archetype_element":"Absence Boundary","domain_realization":"Use clear upper and lower turn boundaries, duration bins, and a narrow status edge to distinguish the designed interval from ordinary page whitespace."},{"archetype_element":"Meaning-of-Absence Check","domain_realization":"Encode whether the region represents measured non-speech, unknown timing, recording loss, or excluded material rather than using one undifferentiated blank."},{"archetype_element":"Context Preservation Frame","domain_realization":"Keep adjacent turns, uncertainty and overlap marks, speaker identity, timestamps, and synchronized audio recoverable around the silence band."},{"archetype_element":"Reintroduction Trigger","domain_realization":"Restore deferred annotations immediately when the analyst selects the band, opens the side panel, or disables silence-lane mode."},{"archetype_element":"Clarity or Effect Test","domain_realization":"Test whether analysts correctly locate and classify interval status and sequence structure, rather than accepting visual sparseness as sufficient evidence."}],"mechanism_mapping":[{"mechanism_slug":"whitespace","role":"The gap itself signals separation and makes the temporal relation between the two turns perceptually distinct without adding a competing diagram.","counterfactual_removal":"If the protected gap were replaced only by another inline symbol, absence would cease to organize perception and the intervention would revert to denser annotation."},{"mechanism_slug":"editorial_cut","role":"Temporarily removes redundant or premature metadata from the focal region while retaining every cut item in a recoverable side panel.","counterfactual_removal":"Without the recoverable cut, surrounding glosses and codes could continue to crowd the interval; permanent deletion would instead damage context and auditability."},{"mechanism_slug":"empty_state_design","role":"Diagnoses and labels different reasons for apparent absence so measured silence is not confused with unavailable audio, unknown timing, or excluded content.","counterfactual_removal":"Without typed absence states, a blank band could imply several incompatible conditions and would be unsafe to interpret."},{"mechanism_slug":"margin_and_gutter_system","role":"Defines reusable spacing and duration-bin rules that protect the silence lane consistently across excerpts and display sizes.","counterfactual_removal":"Without a codified boundary, annotations could gradually encroach on the band and its meaning would vary by transcript or renderer."},{"mechanism_slug":"focus_mode_or_control_hiding","role":"Hides only secondary display controls and annotations during sequence inspection and provides an obvious one-action restoration path.","counterfactual_removal":"Without reversible hiding, the view would either remain crowded or strand analysts without supporting metadata."}],"causal_chain":["Inline annotations and compressed gap notation compete with the perceptual evidence carried by response timing.","The renderer identifies a verified interval at a sequence-relevant boundary and assigns it a typed absence status.","Deferrable annotations are moved out of the immediate region while load-bearing context and audio access remain present.","A bounded, protected silence band gives the interval visible extent and isolates the turns it relates.","The edge cue prevents the quiet region from being mistaken for missing data or a rendering fault.","Analysts can notice the interval before optional analytic labels and then recover all deferred evidence when needed.","If the representation works, interval-status identification and sequence judgments should become more traceable without reducing access to source context."],"baseline":"Use the project’s ordinary transcript view: consecutive turn lines with compact inline pause notation, timestamps, glosses, speaker metadata, overlap marks, and analytic codes shown together. Analysts may consult synchronized audio or waveform views as they normally would. The underlying corpus, excerpts, instructions, and available evidence remain identical across baseline and intervention conditions.","nearest_rivals":["An inline duration label that makes the pause value more prominent but does not protect an empty region around it","A synchronized waveform view that exposes acoustic timing but separates it from the turn-sequence reading task","Color highlighting of pause symbols or response delays, which adds salience through a positive mark","A filter that hides selected annotation classes without giving the resulting absence a bounded meaning or relation to adjacent turns"],"remaining_contrastive_claim":"The candidate’s distinguishing testable feature is not merely fewer annotations, stronger pause notation, or access to timing data. It is a typed and protected empty interval whose boundaries are structurally attached to the initiating and responding turns, with displaced context recoverable on demand. Any advantage must disappear if that protected absence is replaced by an equally informative inline mark or ordinary filtering.","authority_safety":{"decision_authority":"The corpus or project lead may authorize a reversible prototype view and a bounded analyst study; individual analysts retain authority over their classifications, and corpus stewards retain authority over source transcripts and recordings.","authorized_first_step":"Create a read-only prototype for a small, access-controlled excerpt set and compare it with the unchanged baseline view; do not alter source alignments or publish participant material.","excluded_actions":["Deleting or rewriting source transcript annotations","Inferring participant intent from silence alone","Treating recording loss, redaction, or unmeasured time as observed silence","Hiding uncertainty, overlap, consent restrictions, or provenance","Making the silence-lane view the project default before the bounded comparison","Releasing recordings or transcript excerpts beyond their existing access permissions"],"halt_rollback":"Halt the test if the view causes analysts to confuse absence types, conceals load-bearing context, exposes restricted material, or produces display failures that resemble silence. Roll back by disabling the optional renderer and retaining the unchanged transcript, alignment, annotations, and study logs."},"negative_tests":{"strongest_counterevidence":"In a blinded comparison using identical excerpts, analysts perform as well or better with a prominent inline duration label or waveform-supported baseline, while reporting that the silence bands fragment turn sequences or overstate timing as evidence.","problem_falsifier":"The candidate problem is falsified for the target workflow if analysts already notice and correctly distinguish sequence-relevant measured silence, unknown gaps, and recording discontinuities in the ordinary view without extra navigation or recurrent confusion.","intervention_falsifier":"The intervention is falsified if the silence-lane view does not improve correct interval-status identification or traceable sequence judgments relative to the baseline, or if any improvement is matched by an equally informative inline-label rival without protected empty space.","risks":["A visually large band may imply interactional importance that duration alone does not justify.","Duration bins may distort fine timing distinctions or appear metrically precise when alignment is uncertain.","Removing provisional codes from immediate view may slow analysts who legitimately need dense comparison.","The band may be read as participant refusal, discomfort, or intent without supporting contextual evidence.","Screen readers or compact displays may not receive an equivalent ordering and typed-status cue.","Analysts may learn condition-specific display conventions during a crossover comparison."]},"next_evidence_step":"Select 12 access-controlled excerpts containing a balanced set of measured silence, unmeasured gaps, recording discontinuities, and near-zero transitions, with statuses established from existing alignment metadata rather than new interpretation. Build static or read-only interactive baseline, silence-lane, and prominent-inline-label views. In counterbalanced order, ask 6 eligible analysts to identify interval status, locate the relevant turn boundary, state a sequence classification with cited evidence, and rate whether any necessary context was inaccessible. Record accuracy against the established interval status, navigation actions, classification agreement, context-recovery failures, and qualitative misreadings. Use the result only to decide whether a larger evaluation or redesign is warranted.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"Not assessed in this sealed single-proposal generation.","revision_record":{"parent_version":null,"progress_targets_addressed":[],"conceptual_changes":[],"operational_changes":[],"evidence_changes":[],"claim_changes":[]}}