KIDEVAL – Child Language Evaluation
Status: Current
Purpose
Produces a comprehensive child language evaluation report by combining multiple analysis methods into a single per-speaker summary. KIDEVAL is designed for evaluating children’s language development and aggregates results from several individual CLAN commands.
See the CLAN manual for the original KIDEVAL command specification.
Usage
chatter clan kideval file.cha
chatter clan kideval --speaker CHI file.cha
chatter clan kideval --format json file.cha
chatter clan kideval --dss-rules english.scr file.cha
Options
| Option | Description |
|---|---|
--speaker <CODE> | Include speaker |
--exclude-speaker <CODE> | Exclude speaker |
--dss-rules <PATH> | Custom DSS rules file (.scr) |
--ipsyn-rules <PATH> | Custom IPSYN rules file |
--format <FMT> | Output format: text, json, csv, clan |
KIDEVAL does not expose per-component utterance caps as CLI flags
(despite the standalone DSS / IPSYN commands having --max-utterances
of their own); the combined report uses each component’s built-in
default (DSS 50, IPSYN 100). To override, run those components
separately and aggregate.
Combined Metrics
KIDEVAL produces a single report combining:
| Metric | Source | Details |
|---|---|---|
| MLU (words and morphemes) | Main tier + %mor | See MLU |
| NDW / TTR | Main tier word types/tokens | See FREQ |
| DSS score | %mor tier | See DSS |
| VOCD (D statistic) | Main tier words | See VOCD |
| IPSyn score | %mor tier | See IPSYN |
| POS category counts | %mor tier | Nouns, verbs, auxiliaries, etc. |
| Error counts | [*] markers | Word-level errors |
This is the primary tool for clinical assessment of child language samples, providing a comprehensive profile in a single command invocation.
Differences from CLAN
VOCD simplification
KIDEVAL uses a simplified TTR-based D estimate rather than the full bootstrap sampling approach used by the standalone VOCD command. This trades precision for speed when computing the combined report.
IPSYN rules
Uses the built-in simplified rule subset unless a custom rules file is provided via --ipsyn-rules. For full 56-rule coverage, supply the official IPSYN rules file.
DSS rules
Uses the built-in simplified rule subset unless a custom rules file is provided via --dss-rules. For full clinical scoring, supply a complete .scr rules file.
AST-based analysis
All component analyses share the same AST-based infrastructure, ensuring consistent word identification and morpheme counting across all metrics. In CLAN, each component command has its own independent word-filtering logic, which can lead to subtle inconsistencies.
Golden test parity
Verified against CLAN C binary output.