Skip to content

WF-043 · Writing · Evidence and source dates listed below

How can you revise dialogue in a separate pass?

Use a focused dialogue checklist on one scene.

Watch the briefing

Watch the full video on YouTube

Full briefing transcript

A dialogue pass can ask three questions without trying to solve them all at once: what does each speaker want, how does each phrase a tactic, and can readers track who speaks? Start with one short argument. Inspect intention first, voice second and clarity third. Then reread the whole scene, because a focused improvement can still damage a relationship elsewhere in the prose.

Here is the five-line argument. You should let me go, Mara said. I don't want to, the inspector said. I really think you should, Mara said. No, he said. Fine, she said. The speakers disagree, but their reasons and tactics remain broad. Tags make the alternation clear, yet similar phrasing gives little sense of what each person attends to. We have several possible problems, not one instruction to make every line clever.

Here is a revised argument. Give me one crossing, Mara said. Only if the repair report is complete, the inspector said. The missing page won't change the tide. It may change who's responsible. Mara lowered the form. Then tell me what that page must establish. Mara seeks a limited permission; the inspector sets a documentation condition. Her final request changes the exchange from repeated demand to clarifying what would satisfy him.

Pass one is intention. Mark Mara's permission request and the inspector's condition, then follow how each answer redirects the conversation. Pass two is voice. Mara points toward immediate circumstances, while the inspector emphasizes responsibility. Those are choices in attention and phrasing, not caricature accents. Pass three is clarity. The first two tags establish the pair; the final action and name identify Mara when the tactic changes.

The untagged middle lines rely on established two-person alternation. If a third speaker enters or an interruption breaks the pattern, add identification where ambiguity appears. Do not keep every tag merely to meet a rule or delete every tag for elegance. Similarly, the formal condition suits this inspector's role here; it is not a template for every opponent. Preserve each speaker's intention without turning them into slogans.

Try three narrow readings of one dialogue scene. First label the intended action behind each line. Next underline distinctive wording or attention and ask whether it fits the person. Finally check speaker changes and paragraph boundaries without relying on your private memory of the conversation. Make a small revision after each reading. Then read aloud and examine the scene's turn to ensure the separate fixes still work together.

Use a focused dialogue checklist on one scene. The checklist should expose a problem, not require decoration on every line. Your practical result is three margin notes, one targeted change in each pass where needed, and a final whole-scene comparison. Keep a plain line when it does the job clearly. If a revision improves voice but obscures who speaks or what changes, return to that relationship before calling the dialogue finished.

Practical takeaway

Use a focused dialogue checklist on one scene.

Disclosures

AI-assisted production and synthetic narration. Original teaching examples and diagrams; linked third-party sources retain their respective rights.

All ferry-town passages, names, comparisons and exercises are original synthetic teaching material; no source literary excerpts, private manuscripts, Leonard biography or copied imagery. Local synthesis/render completed; no external action.

Original sources and limits

  1. OpenLearn — Editing summary — checked 2026-09-18. General revision, selection and clarity foundation, not a universal priority hierarchy or checklist formula.
  2. OpenLearn — Giving feedback — checked 2026-09-18. Specific analytical feedback foundation; our intention/cue/interpretation trace and uncoached fresh-reader tests are original applications.
  3. OpenLearn — Receiving feedback — checked 2026-09-18. Reflecting on comments while retaining author choice; no claim one reader response establishes universal engagement.

Checked dates preserved from source records; no later recheck claimed.