Return to the evidence before correcting it
Debate retrieval · Original evidence only
Open the unedited Lesson 5 audio, burden maps, examiner notes and live-evidence log. Locate the saved successful inversion and two effortful seams.
Recover what happened in the interaction before proposing a repair.
- For each time stamp, transcribe the preceding challenge, the exact response and the next listener or opponent move. Preserve false starts and self-repairs.
- State the intended job: restrict, correct, add, continue, test a condition or move to action. Do not diagnose from the isolated sentence.
- Mark what the listener actually recovered: source, condition, claim scale, payoff and responsible actor. Leave the wording unchanged for now.
Choose the action that preserves valid evidence for repair.
1. The speaker’s notes contain a polished inversion, but the audio shows a different, incomplete response. Which text belongs in the repair record?
2. A listener answers the wrong condition after a grammatically correct inversion. What should be logged first?
Diagnose the first layer that changed the argument
Repair order · Meaning before surface form
| Layer | Question | Typical evidence |
|---|---|---|
| 1 · Source and claim scale | Did the turn preserve who said what, about whom and with what limit? | An opt-in 68% becomes “residents agree”; a concern becomes a prohibition |
| 2 · Responsiveness and register | Did the response answer the live claim in language suited to this formal debate? | A rehearsed flourish replaces the answer; five marked openings obscure hierarchy |
| 3 · Clause target and conditional time | Is the correct clause inverted, and does the condition keep its intended reality and time? | Only if will… can…; an unreal past premise presented as present proof |
| 4 · Auxiliary and completion | Is one auxiliary correctly placed, and is every promised relation completed? | A doubled auxiliary; not only without but also |
| 5 · Prosody and distribution | Can listeners hear restriction, consequence and release without theatrical overload? | Question-like closure; equal heavy stress; marked syntax in every turn |
| 6 · Plain control and retest | Does a simpler version preserve the proposition and improve recovery? | A plain clause wins when inversion adds no useful argumentative job |
Assign each debate seam to the first repair layer that explains its listener effect. A later grammatical fault may still need correction.
S Source or claim scale
R Responsiveness or register
C Clause target or condition time
A Auxiliary or completion
P Prosody or distribution
1. “Never have Porto Claro residents accepted secret systems,” although the record contains only an opt-in consultation.
2. Asked who would fund human review, the speaker delivers a prepared line about transparency and leaves the cost question unanswered.
3. “Only if will the council define deadlines can residents test the appeal route.”
4. “Not only could a pilot generate evidence.” The turn ends there.
5. A correct restrictive inversion closes with a rise, so the chair hears it as a question rather than a finding.
6. Four consecutive answers begin with marked inversion, and listeners cannot identify which correction carries the argument.
Make the smallest repair that restores the move
Repair only as far as the evidence requires. First restore the proposition: source, population, time, condition and degree. Then decide whether marked syntax still performs a useful job. With only if and not until, keep ordinary order in the subordinate clause and invert the main clause. With had, were and should, keep the result compatible with the condition’s time and reality. With not only, hardly or no sooner, complete the promised relation.
Over-formality is a fit problem. A formal council debate licenses inversion, but a grammatical form can still sound staged when it ignores the live claim, repeats without hierarchy or receives all-heavy stress. Retain it when restriction, correction, contingency or addition deserves foregrounding. Otherwise use ordinary order and let evidence carry the authority.
Match each repaired turn to the principle that licenses it.
Choose the version that repairs the named seam without changing the case.
1. The consultation was opt-in. Which response repairs “Never have residents supported opaque systems”?
2. Which continuation keeps a should conditional prospective rather than certain?
3. When is plain order the stronger repair in this formal debate?
4. A grammatical inversion is delivered with equal stress on every word. What is the smallest next test?
Correct the layer, then test whether inversion survives
Correct each turn at the named layer. Preserve the Porto Claro evidence and use marked syntax only where the repair still earns it.
1. Only if will the council publish every supplier field can residents understand the system. — clause target and overclaim
2. Had the consultation represented the city, it proves that 68% of residents demand immediate adoption. — condition time and claim scale
3. Not only does a pilot create evidence. — completion and certainty
4. No sooner would publication occur when accountability would follow. — pair word and causal guarantee
5. Under no circumstances may a security exception be permitted, because suppliers mentioned risk. — source status and excessive force
6. Never has transparency been more essential, nor has accountability been more vital, nor should secrecy ever prevail. — unsupported absolutes and over-formality
Rewrite each turn in the direction shown. Preserve proposition, source boundary and commitment; change the rhetorical distribution only.
Rehearsed flourish → responsive plain answer
Never has the need for democratic accountability been greater. [Question: Who funds human review?]
Doubled inversion → one earned restriction
Only if will the council define deadlines can residents test the appeal route.
Unsupported absolute → bounded plain claim
Never have residents accepted automated decisions without explanation.
Flat sequence → one marked correction plus release
The inventory does not prove that every system is hidden. It shows that nine of fourteen lack a public description. The audit team should close the inventory gap.
Accumulated formality → clear distribution
Rarely has visibility mattered more; only then can trust emerge; nor should any exception remain; never must remedy be delayed.
Run a repair conference on the live debate
Pushed output · Diagnose, repair, blind-retest
Use the saved successful turn and two consequential seams from Lesson 5. A partner acts as listener and challenge reader; working alone, record the same protocol and compare delayed reconstructions.
Produce two locally repaired debate turns and prove whether marked or plain syntax gives the listener the more accurate argument map.
- Minutes 1–3 · reconstruct: for each seam, align preceding claim, exact response, intended job and next move. Rank the loss: evidence, condition, payoff, action or stance.
- Minutes 4–6 · diagnose: identify the first consequential layer and any downstream fault. Cite exact words and audio evidence; do not label a sentence “too formal” without a register, distribution or responsiveness effect.
- Minutes 7–9 · build two versions: make the smallest defensible repair and a plain control. Preserve source, population, time, modality and policy burden. Mark thought-group boundaries and one contrastive stress.
- Minutes 10–12 · blind retest: hear both versions after the original challenge in a concealed order. The listener records claim, evidence limit, condition or payoff and next actor, then chooses the easier accurate recovery.
- Minutes 13–14 · pressure replay: deliver the winning version without pausing, answer one follow-up and release any inversion into ordinary reasoning. Keep the original position unless the evidence requires a concession.
- Minute 15 · archive: save original, diagnosis, both repairs, listener maps, winning rationale and replay. Carry the repaired moves into Lesson 7 planning; do not draft the essay yet.
Repair ledger · Evidence required
| Field | Record |
|---|---|
| Context | Time stamp, preceding challenge, exact turn and following response |
| Intended move | Restriction, correction, addition, continuation, condition or action |
| First layer | Source/scale, responsiveness/register, clause/time, auxiliary/completion or prosody/distribution |
| Minimal repair | Only the changes needed to restore the proposition and intended move |
| Plain control | The same defensible proposition in ordinary order |
| Listener test | Recovered claim, limit, condition/payoff and next actor for both versions |
| Decision | Winning version and a reason based on accurate recovery rather than sophistication |
Keep the version that repairs listener recovery
Choose the conclusion supported by the retest evidence.
1. The marked and plain versions are grammatical, but only the plain one leads listeners to identify the finance office as the next actor. Which wins here?
2. A corrected only if sentence is recovered accurately until the speaker uses a final rise and the listener records a question. What remains to repair?
3. One learner repeatedly inverts the subordinate only if clause across baseline, practice and replay. What may the teacher record?
Carry repaired control into the essay plan
Choose what should enter Lesson 7.
1. Which item belongs in the 440-word essay planning record?
Standing informal-input bridge
Finish the selected full Intelligence Squared debate and continue the high-quality op-ed. Revisit the saved challenge–response sequence and paragraph turn: create one marked and one plain version, then record which better preserves source, condition, payoff and response. Bring exact source locations, the Lesson 5 artifacts and today’s repair ledger to Lesson 7. Do not draft the 440-word essay before its brief and evidence architecture are set.