SALT2CHAT – SALT Transcription to CHAT Conversion
Status: Current Last updated: 2026-05-12 13:36 EDT
Purpose
Converts SALT (Systematic Analysis of Language Transcripts) transcription files into CHAT format. SALT is a widely used clinical transcription system with its own conventions for speaker codes, morpheme annotations, and error marking.
Usage
chatter clan salt2chat input.slt
Options
| Option | Default | Description |
|---|---|---|
-l, --language | "eng" | ISO 639 language code for the @Languages header |
-o, --output | stdout | Output CHAT file path |
The corpus name in @ID headers is hardcoded to "salt_corpus"
(crates/clan/clan-core/src/converters/salt2chat.rs:193); there is
no CLI flag to override it. Same pattern as the other converters
in this directory.
Speaker Mapping
| SALT code | CHAT speaker | Role |
|---|---|---|
C | CHI | Target_Child |
E | EXA | Investigator |
P | PAR | (Parent) |
I | INV | (Investigator) |
SALT Annotation Stripping
SALT-specific annotations are removed during conversion:
- Morpheme codes (
word/3s–>word) - Error markers (
word*–>word) - Maze markers (
(word)–> skipped) - Comment markers (
{...},[...]–> skipped) - Bound morpheme markers (
_word–>word)
Input Format
SALT transcription files with header lines (starting with $ or +) followed by speaker-prefixed utterance lines. SALT uses single-letter speaker codes and inline annotation conventions.
Output
A well-formed CHAT file with SALT speakers mapped to standard CHAT speaker codes, SALT-specific annotations stripped, and proper CHAT headers generated. Header metadata (participant name, age, gender, context) is extracted from SALT $ lines when available.
Differences from CLAN
- Uses typed AST for CHAT generation
- Produces valid, well-formed CHAT output