Speaker Filtering

Status: Current Last updated: 2026-05-11 18:58 EDT

Speaker filters restrict which speakers’ utterances are analyzed. This is one of the most frequently used filters — most CLAN analyses target a specific participant (e.g., the child in a child language study).

Include speakers

Analyze only specific speakers:

chatter clan freq --speaker CHI file.cha
chatter clan mlu --speaker CHI --speaker MOT file.cha

CLAN equivalent: +t*CHI, +t*CHI +t*MOT

Multiple --speaker flags use OR logic: utterances from any listed speaker are included.

Exclude speakers

Remove specific speakers from analysis:

chatter clan freq --exclude-speaker INV file.cha

CLAN equivalent: -t*INV

@ID filtering

Filter speakers by metadata fields in the @ID header. The filter is pipe-delimited, in CHAT @ID column order:

lang|corpus|speaker|age|sex|group|ses|role|education|custom

Each field is either * (wildcard), empty (also wildcard), or a literal exact match. Trailing fields may be omitted entirely.

# All children (role = Target_Child, at position 8)
chatter clan freq --id-filter "*|*|*|*|*|*|*|Target_Child" file.cha

# English-language speakers only
chatter clan freq --id-filter "eng|" file.cha

# A specific speaker code (position 3)
chatter clan freq --id-filter "*|*|CHI" file.cha

CLAN equivalent: +t@ID="*|*|*|*|*|*|*|Target_Child".

Authoritative parser: crates/clan/clan-core/src/framework/id_filter.rs. The language field is a comma-separated list in the underlying @ID; the filter matches if any language in the @ID matches the pattern.

Interaction with other filters

• Speaker filtering is applied first, before range or word filters
• Include and exclude can be combined; excludes are applied after includes
• Speaker codes are case-sensitive and must match the @Participants header exactly (e.g., CHI, not chi)