Request a demo
Request a demo

The CANDOR Corpus

BetterUp Labs advances the science of human connection with CANDOR, the world’s most comprehensive collection of natural conversation, for use by researchers and universities worldwide.

The CANDOR Conversation Corpus was published in Science Advances, an open access journal from the American Association for the Advancement of Science (AAAS) that publishes innovative original research across all disciplines of science, made freely available to a global audience.

You can read the full paper here: Science Advances | The CANDOR Corpus: Insights from a large multimodal dataset of naturalistic conversation.

At BetterUp we know that impactful coaching hinges on powerful conversations. Conversations have the capacity to enrich people’s lives and help them reach their potential – whether they’re speaking to an expert coach, a manager, a colleague, a loved one, and even a stranger. 

Large, high-quality data sets for studying conversations have historically not been readily accessible. And without them, it’s been difficult for the field of conversation research to make significant progress. That’s why BetterUp Labs collected, and is now proud to release, the  CANDOR Conversation Corpus – the world’s largest multimodal data set of two-person conversations that includes vocal, facial, and semantic expression, along with a detailed post-conversation survey of speakers’ reflections. 

This data set captures 1,600+ conversations that took place Jan. - Nov. 2020 between U.S.-based research participants aged 18-65 who opted into having their conversations recorded and were compensated for their participation. To analyze and publish the results, BetterUp Labs partnered with leading researchers from Wharton and to explore CANDOR corpus, examining the intricate mechanics of how conversation partners take turns, the relationship between conversation and happiness, how the national discourse shifted during the onset of the pandemic, and much more.

By making this data set of 850+ hours of conversations available to the scientific community, BetterUp Labs aims to support research institutions around the world in advancing the interdisciplinary body of knowledge in a way that’s never been possible before. 250+ universities and researchers around the world including Columbia, Harvard, Dartmouth, Cornell and Carnegie Mellon are already using this data to further our understanding of linguistics, psychology, communications, sociology, semantics, speech recognition, mental health, and more. BetterUp Labs will continue to partner with research teams around the globe in this work and publish our science for the benefit of others.

BetterUp Labs’ full study protocol for the CANDOR corpus has been reviewed and approved by a third-party, federally-certified research ethics board. All participants have consented and agreed to be video and audio recorded during the study, and were notified that the video/audio recording will be shared with other researchers and/or made publicly available. No BetterUp customers, coaches or members were included in this independent research study.