CHILDES Kazakh Balausa1 Corpus


Ilya Razorenov
EMCL++
University of Groningen

Michael Riessler
Philosophical Faculty
University of Eastern Finland

Irina Sekerina
Psychology
College of Staten Island, CUNY

Participants: 8
Type of Study: cross-sectional
Location: Kazakhstan
Media type: audio
DOI: doi:10.21415/vk5j-kw20

Browsable transcripts

Download transcripts

Link to media folder

Citation information

Publications based on these data should cite one of the above articles.

Project Description

The Balausa1 corpus comprises a collection of spontaneous caregiver–child interaction samples in Kazakh. The corpus includes 13 recordings from eight typically developing Kazakh-speaking children between the ages of 1;09 and 5;01 interacting with parents and other family members in naturalistic home environments across Kazakhstan. Most children contributed a single recording, while several children were recorded on multiple occasions, providing limited developmental coverage across time.

Recordings were collected by parents during everyday activities, including play, family conversations, storytelling, mealtime interactions, and routine household activities. The current release contains approximately five hours of time-aligned audio recordings and transcripts. The corpus includes 3,621 child utterances (approximately 7,005 tokens) and 4,177 adult utterances (approximately 10,780 tokens). Morphological annotation (%mor tier) was generated automatically using Batchalign2 and Universal Dependencies resources and subsequently manually verified for nouns, proper nouns, verbs, and auxiliary verbs with the help of trained native Kazakh speakers.

Balausa1 represents the first release of the broader Balausa initiative, which aims to expand the documentation of Kazakh child language in future versions of the corpus. In addition to the CHILDES release, supplementary materials, corpus reports, and project resources are available on GitHub (https://github.com/langdoc/kaza1248-aquisition).

Acknowledgment