INDENT – Align CA Overlap Markers

Status: Current

Purpose

Aligns overlap markers in Conversation Analysis (CA) transcripts. The legacy manual describes INDENT simply as a program for realigning overlap marks in CA files, and notes that the files must use a fixed-width font such as CAFont.

clan-core aligns closing overlap markers (⌊, U+230A) by column position with their matching opening overlap markers (⌈, U+2308) on a preceding speaker tier.

Usage

chatter clan indent file.cha
chatter clan indent file.cha -o aligned.cha

Algorithm

1. Parse the file into tiers (speaker prefix + content text)
2. For each main tier (*SPK:), scan for opening overlap markers ⌈ and record their column positions and optional numeric suffixes
3. Scan up to 30 subsequent tiers from different speakers for closing overlap markers ⌊
4. Match open/close pairs by numeric suffix (or sequentially if unnumbered)
5. Insert or remove spaces before the closing marker to align columns
6. Report unmatched markers as warnings

Example

Before:

*CHI:	I want ⌈ cookies ⌉ .
*MOT:	⌊ yeah ⌋ okay .

After:

*CHI:	I want ⌈ cookies ⌉ .
*MOT:	       ⌊ yeah ⌋ okay .

Numbered overlaps (⌈1, ⌈2, etc.) are matched by their numeric suffix, allowing multiple simultaneous overlaps to be aligned independently.

Differences from CLAN

• Manual intent: INDENT is a layout command, not a semantic CHAT analysis command.
• Operates on UTF-8 text using Rust’s char-based column counting rather than C byte-level scanning.
• Uses the text-based transform pattern (no AST round-trip) to preserve original formatting outside of overlap alignment.
• Maximum 10 alignment passes (CLAN’s goto beginAgain loop has no bound, causing infinite loops on some inputs).
• Column counting treats each Unicode scalar value as width 1.