Reconstruct the failure before editing it
Retrieval · Lesson 5 evidence
Place the unedited meeting recording, role notes, final statement, listener trace, secure response and time-stamped repair ticket side by side.
Identify the first point where intended meaning and listener reconstruction diverged.
- Play only the prior claim, target turn and following response around the marked time stamp. Do not explain the intended answer first.
- Write four lines: exact wording, intended proposition, listener reconstruction and decision consequence.
- Name one candidate layer—complement, source stance, scope, relationship, condition, responsibility, sequence, responsiveness, pause or prominence—without editing yet.
- Keep the secure response available as a comparison. Lesson 6 repairs local evidence; it does not replace the full meeting with a polished script.
Choose the repair ticket that supports a testable diagnosis.
1. Which ticket gives enough evidence for a local repair?
One listener symptom can have different causes
Symptom evidence from an Aster meeting
| Listener symptom | Possible first cause | Evidence needed before repair |
|---|---|---|
| Projection heard as verified return | Source or stance disappeared; projection verb was weakened by delivery | Exact source phrase, reporting verb, audible grouping and reconstructed owner |
| Survey result heard as an all-workforce finding | Respondent scope was absent, remote or deaccented | Location and prominence of among 62 respondents and 24 |
| Observed difference heard as causal proof | Relationship overclaim or causal boundary placed too late | Verb choice, lower-volume fact, design limit and point of listener divergence |
| Recommendation heard as weak or evasive | Advice itself was over-hedged, or the answer was delayed behind caveats | First finite clause, action verb, condition and response to the actual question |
| Recommendation heard as unconditional | Gate or owner was missing, vague or prosodically detached | Action scope, condition, responsible role, evidence threshold and thought-group boundary |
| Exact fact heard as uncertain | Probability or approximation language attached to an established layer | Shared record, lexical/modal stack and whether the audience questioned that fact |
Match each transcript-and-listener trace to the first layer that needs inspection. Do not infer a global weakness from one seam.
F Reporting form or complement
S Source or stance
P Scope
R Relationship strength
C Condition or responsibility
D Delivery or sequence
1. Speaker: “The vendor suggested us to roll out.” The listener understands the advice but flags the structure as unstable.
2. Speaker: “It is expected to create 1.1 million credits.” The listener cannot identify who produced or reviewed the estimate.
3. Speaker: “Employees reported disruption.” The listener attributes the finding to all 84 pilot employees.
4. Speaker: “Aster reduced the median by 2.2 hours.” The listener records a proven platform effect.
5. Speaker: “We can extend, subject to further review.” The listener cannot name who reviews what or when.
6. The speaker gives the compliance gate after a long pause and after the chair has moved on; the listener records only “extend”.
Minimal-repair rule
Repair the first layer that can explain the observed divergence. If changing The vendor projects restores ownership, do not add three modal hedges as well. If the wording is recovered correctly when read but not in the meeting audio, test grouping or prominence before rewriting the sentence. Preserve every successful part so the retest can identify what caused the gain.
Use a seven-step calibration diagnosis
From proposition to listener recovery
| Step | Diagnostic question | Minimal repair |
|---|---|---|
| 1 · Record accuracy | Do number, source, population, time, relationship and status match the Aster record? | Restore the evidence floor before adjusting style |
| 2 · Reporting form | Does the verb take the chosen that-clause, -ing form, to-infinitive or object plus to-infinitive? | Repair only the complement or choose a verb whose stance and grammar fit |
| 3 · Source and stance | Can the listener identify who observed, reported, projected, interpreted or recommended the proposition? | Name the source and use a verb that does not silently upgrade or downgrade it |
| 4 · Qualification scope | Is uncertainty attached to fact, frequency, magnitude, scope, relationship, time or condition? | Move, replace or delete the hedge so it reaches only the open layer |
| 5 · Action architecture | Are recommendation, condition, responsible role, evidence threshold and next decision point recoverable? | Own the advice and make the gate operational |
| 6 · Sequence and responsiveness | Does the answer address the live question before adding support, and has accepted information moved into common ground? | Reorder, compress or release repeated caveats into the next action |
| 7 · Prosody | Do thought groups keep source and condition attached while prominence marks the consequential contrast? | Retest delivery with wording held constant |
Match each local fault to the revision that repairs it without changing the record.
Choose what to test first in each diagnosis.
1. The sentence is grammatical, but “among respondents” is absent and the listener generalizes to all staff. What comes first?
2. The written sentence contains the correct source and condition, but both disappear in the audio reconstruction. What is the cleanest next test?
3. A listener understands a sentence despite “recommend to extend”. What must still be repaired?
Brazilian-Portuguese transfer remains a testable local hypothesis. Compare a repeated choice in your own English with how you would handle the same proposition and pressure in Brazilian Portuguese only when contexts are comparable. Test a predicted listener outcome—such as delayed answer, lost source, distributed prominence or excessive/insufficient qualification. Do not use nationality as the diagnosis, and mark a single or contradictory pattern unverified.
Repair form, scope and action separately
Correct the named fault while preserving every defensible part of the proposition. Use the shared Aster record.
1. The vendor suggested Northbank to purchase the annual licence. — reporting complement
2. The dashboard argues that Aster cut resolution time by 2.2 hours. — source stance and causation
3. Perhaps 84 employees may have participated for roughly 12 weeks. — established facts over-hedged
4. Staff reported that routing alerts disrupted concentrated work. — subset generalized
5. The speed improvement clearly proves that Aster will work across all six teams. — relationship and transfer overclaim
6. It has been recommended that there could possibly be an extension, subject to further checks. — responsibility and stacked caution
7. Management admitted that accessibility testing is open. — unjustified stance
8. The incident might have happened, and the controls should probably prevent it again. — certainty on wrong layers
Rewrite each meeting line by changing only the layer named in the direction. Do not silently strengthen the evidence.
Source hidden → explicit ownership
It is expected to create 1.1 million credits in value.
Answer delayed → recommendation first
Given the lower volume, volunteer teams, survey response and unresolved testing, there are several matters, so perhaps we could extend.
Condition vague → operational gate
Extend subject to further review.
Established layer weakened → direct fact
The response rate may possibly have fallen to around 21%.
Relationship too strong → bounded interpretation
The reopened-case figures confirm a quality improvement.
Repeated caveat → accepted premise plus next action
The median was lower, although volume was lower and causation is unclear. As I said, the median was lower, although volume was lower and causation is unclear.
Prosody-only and sequence-only trials
Select one accurate Lesson 5 sentence whose source, scope or condition the listener missed, plus one accurate answer that arrived too late.
Test delivery and sequence before altering successful wording.
- For the prosody trial, record A with original grouping and B with source or condition kept in the same thought unit and consequential contrast selectively prominent.
- For the sequence trial, preserve every proposition but move the answer before its rationale. Keep lexical choices constant.
- Use blind reconstructions. Retain a change only if the targeted meaning becomes more recoverable without a new distortion.
Run a controlled A/B repair
Local repair · Original meeting preserved
Use the first time-stamped Lesson 5 seam where the intended and recovered source, evidence strength, action scope or condition diverged. Keep the preceding challenge and following response in both versions.
Test one minimal revision of a 30–60-second exchange and convert the result into a conditional drafting principle.
- Minutes 1–2 · align: mark exact wording, intended proposition, listener reconstruction and professional consequence at the first divergence.
- Minutes 3–4 · diagnose: apply the seven steps in order and select one primary cause. Record a secondary cause only when separate evidence requires it.
- Minutes 5–6 · revise minimally: change one variable—reporting complement, source verb, hedge scope, action gate, owner, sentence order, thought group or prominence. Preserve the Aster record and successful surrounding language.
- Minutes 7–9 · blind retest: present A and B in unlabelled order. The listener reconstructs source, evidence status, action scope, condition and owner, then identifies the clearer version and cites the language or delivery cue. Working alone, record both and reconstruct after a delay.
- Minute 10 · decide: accept B only if it improves the targeted recovery without changing another proposition. Otherwise retain A or mark a different layer for retest.
- Minute 11 · save: keep both exchanges, both traces, diagnosis and decision. Write one “When X, use Y because Z improved” principle for the 440-word analysis in Lesson 7.
A/B evidence record
| Field | Required entry |
|---|---|
| Context | Exact time stamp, preceding challenge and target professional decision |
| Version A | Original wording and delivery; never overwrite the baseline |
| Diagnosis | First failed layer and evidence connecting it to the listener symptom |
| Version B | One controlled change with all other relevant features held stable |
| Blind result | Recovered source, status, scope, relationship, action, condition and owner |
| Decision | Accept, reject or retest, with a conditional principle for Lesson 7 |
Accept only what the retest demonstrates
Choose the conclusion supported by each retest result.
1. Version B names the vendor and uses projects; the listener now identifies both source and forecast status while all numbers remain unchanged. What follows?
2. A prosody-only Version B still produces an unconditional-rollout reconstruction. What follows?
3. One English repair resembles a preferred Brazilian-Portuguese pattern, but no repeated comparable examples exist. What can be logged?
Error log · Evidence and limit
Complete one row using the original meeting, controlled revision and blind traces.
Carry one tested repair and one unresolved seam into drafting rather than a vague confidence judgement.
- Record context, Version A, listener symptom, first failed layer, Version B, controlled variable and blind outcome.
- Mark repaired only when recovery improves and the evidence status remains intact; otherwise mark retest.
- Keep the secure Lesson 5 response as a positive model of conditions under which the original language already worked.
Carry a conditional principle into the analysis
Choose the principle that can responsibly guide Lesson 7.
1. Which note is appropriately evidence-bound?
Standing informal-input bridge
Return to the saved challenge-response moment in the same full Conversations with Tyler episode and Harvard Business Review feature. Identify one local choice that preserves source, evidence strength or an action condition, then compare it with the accepted Aster repair without copying extended wording. Bring exact source locations, both conditional principles, the shared decision record and the midpoint brief to Lesson 7 for the 440-word professional analysis.