Chinese Queer Corpus
A research corpus of Mandarin speech, sex, gender, and sexuality
Chinese Queer Corpus (CQC) is a speech corpus of speakers from sexual and gender minority communities. The current release contains about 10.5 hours of Mandarin speech collected from public videos on Bilibili and YouTube, transcribed and time-aligned word by word and sound by sound. Researchers can apply to use the annotated data.
At a glance
sex, gender, sexuality, and speech
CQC documents variation in the speech of queer speakers. Our initial phonetic analyses show variation within every group and large overlap between some groups. A voice is not evidence of anyone’s gender identity or sexual orientation, and using the corpus to infer anyone’s identity from their voice is prohibited.
Why CQC
Research on language, sex, gender, sexuality, and the voice has produced a substantial body of work, but most of it concerns English and a small number of other languages. Many large Mandarin speech corpora were not designed for these questions: they provide limited information about sex, gender, and sexuality, and generally do not include detailed self-described gender and sexuality metadata. CQC provides a resource for studying these relationships in Mandarin Chinese.
What you can study
CQC is designed for questions such as:
- How does phonetic variation in Mandarin relate to sex, gender, and sexuality, and how much does it vary within groups?
- How do speakers use linguistic resources to construct identity?
- How do locally meaningful Chinese queer identity labels relate to linguistic practice?
- How well do speech technologies, such as speech recognition, work across different speakers and recording conditions?
Researchers in fields such as linguistics, sociology, anthropology, gender and sexuality studies, psychology, and communication can apply. See For Researchers for what each kind of project needs.
Protecting speakers
- Annotated data and audio go only to approved researchers who sign the Data Use Agreement.
- The agreement prohibits inferring identity, identifying or contacting speakers, and voice cloning.
- Speakers can ask for their material to be removed at any time. See how removal works.
Questions
What is a speech corpus?
A speech corpus is a collection of recorded speech, organized and annotated for research. In CQC, speech is transcribed, and each word and sound is time-stamped so that researchers can explore pronunciation, pitch, and timing.
Why is audio not publicly available?
Public availability does not by itself make redistribution appropriate or unrestricted. Collecting recordings into a labeled research corpus can make speakers easier to locate and analyze systematically.
Who is included?
Speakers who described themselves as members of sexual or gender minority communities in their own public videos on Bilibili or YouTube. We record only what speakers said about themselves, and never infer it from voice or appearance.
I think my video is in CQC. What can I do?
You can ask us to remove it at any time, without giving a reason. See how to request removal or the 中文说明.
I am not a linguist. Can I use CQC?
Yes. Researchers in any research field can apply. See For Researchers.