English CHILDES
Overview
| type | only spoken |
| available since | 2.16 |
| link | https://github.com/UniversalDependencies/UD_English-CHILDES |
| genre | spoken |
| contact | xy236@georgetown.edu |
| sentences | 48183 |
| tokens | 289817 |
Issue draft: UD_English-CHILDES
Modality identification
Is spoken part clearly identifiable? N/A
Metadata review
doc (and paragraphs) metadata
(none found) - corpus_name (6 distinct values: Brown, Braunwald, Providence, Kuczaj, Weist, Thomas) is currently sentence-level and its sentences are interleaved throughout the file rather than grouped.
| Field | Advice |
|---|---|
corpus_name |
recompose: sort sentences by original_sent_id within each corpus_name, then set corpus_name once per document as # newdoc id |
Caveat to confirm with maintainer: corpus_name is the CHILDES study name, not a single recording - e.g. Brown alone contains three different children (Adam, Eve, Sarah) recorded across different ages/sessions, and Providence contains three more (Lily, Naima, Violet). If a “document” should mean one recording session, newdoc id may need to key on (corpus_name, child_name, child_age) instead of corpus_name alone
speaker metadata
| Field | Advice |
|---|---|
child_name |
rename to speaker_id |
child_age |
rename to speaker_age |
child_gender |
rename to speaker_gender |
chi l d |
data bug, not a real field: a single malformed line (# chi l d = 37.29..., 1 occurrence) - almost certainly a corrupted child_age entry; flag to maintainer |