TEXT
COMMERCIAL
LABELED
Contributors declare the two languages they mix, then write as they would message a friend who speaks both, switching inside the sentence, not between paragraphs. Monolingual text, sentence-boundary switching and anything translated are refused, and the brief tells people not to force it. Recruitment is confined to 11 Southeast Asian countries, where Taglish, Indonesian-English and Malay-English mixing is the ordinary way to text; three pieces per author at most, each tagged with its language pair. A large share of the world texts in two languages at once and almost none of it reaches training data, because it rarely appears in published text.
Volume
50,000+
Pieces
Contributors
25,000+
Regions
Southeast Asia (11 countries)
Updated
What’s in the collection and how it’s delivered.
Data modality
Text
Annotated
Labeled (declared language pair)
Exclusivity
Non-exclusive rights
Capture tool
Typed in the Kled app
Use cases
Code-switched language modeling, token-level language identification, informal chat-style generation
Format
Plain text
Resolution
Plain text, UTF-8
Total volume
50,000+ pieces (approx. 250 MB)
Regions
Southeast Asia (11 countries)
Purity level
99%+ passed QA review
Where this data comes from and how it’s licensed.
Every file comes straight from the contributor who provided it and is rights-cleared before it’s listed, so your license traces back to the source.
Contributor
Provides the data and uploads it straight to Kled.
Kled
Clears the rights before anything is listed.
Your team
Licenses it under the terms below.
License terms
Non-exclusive commercial training license.
Consent
Each contributor opted in to this use when they uploaded.
Specs reflect the current collection and may change as new data is added. Final terms are set in your license agreement.



