The topics people chose to talk about, as measured in CANDOR transcripts by a simple keyword dictionary, reflect the ebb and flow of societal issues in an unusually tumultuous year. COVID-19 (red) surged from unknown to the talk of the nation by mid-2020, matching or even exceeding family-related discussion (blue), a reliable staple of conversation. CANDOR frequencies for the presidential election (purple) and policing (green) highlight the trajectories of the nationally debated issues.
Candor Dataset
I co-lead the development of Candor (Conversation: A Naturalistic Dataset of Online Recordings), which is among the largest, publicly available, multimodal datasets of naturalistic conversation.
Dataset is available
In the manuscript that accompanied the public release of the data, published in Science Advances, we explore the corpus by embracing three principles:
- Conversation in constructed around a highly cooperative system of turn-taking.
- Understanding the full complexity of conversation requires insights from a variety of disciplines that, although they examine the same phenomenon, often remain siloed in their research questions.
- Examining conversation computationally and at scale is an enduring challenge, but new technologies, particularly advances in machine learning, promise to unlock many aspects of conversation that were previously inaccessible.
Key Features
- The full data download includes the raw audio and video files, along with hundreds of measures that capture everything from people’s overall enjoyment and their impressions of their conversation partner to millisecond-by-millisecond measures of turn taking, emotion expression, and vocal prosody.
- We qualitatively reviewed every conversation to ensured data quality, and in the process, identified conversations that were particularly high in rapport – often among conversation partners who needed to bridge significant demographic and sociocultural divides.
- The corpus consists of Americans, aged 19-66, who talked during the year 2020, thus offering a unique lens into how the national discourse shifted in one of the most tumultuous years in recent memory, including the onset of a global pandemic and hotly contested presidential election.
Analysis Highlights
1. Topic flow within the CANDOR corpus
2. A framework for studying conversation
The results are organized according to an analytic framework that distinguishes between three related levels of conversation. Low-level features can be observed directly, vary over short time periods, and often relate to conversational structure (e.g., a pause at the end of a speaker’s turn). Mid-level features are generally inferred indirectly by human perceivers or algorithms that approximate human perception, vary on a medium-frequency or turn-by-turn basis, and capture linguistic or paralinguistic conversational content (e.g., a happy facial expression or vocal emotional intensity). High-level features relate to people’s subjective judgments of a conversation (e.g., post-conversation-reported enjoyment or people’s evaluations of their partner). Work exploring the interplay between levels will represent an increasingly common and important type of research.
3. Positive affect is significantly greater after than before a conversation
Each row of density plots corresponds to an age group. Respondents were asked to report their mood immediately before (red) and after (blue) their conversation. Conversation’s effect on people’s mood was positive, significant, and of considerable magnitude
4. Behavior patterns of good and bad conversationalists. (AtoF)
The behavioral patterns of good conversationalists (top 25% of partner-rated conversationalist score; depicted in blue) and bad conversationalists (bottom 25%; depicted in red) are depicted. Horizontal axes denote turn-level feature deciles. The y-axis indicates the mean proportion of turns in a category for a good or bad conversationalist. Error bars represent 95% confidence intervals. Top, middle, and bottom rows correspond to text, audio, and visual modalities, respectively; left and right columns include features that can be observed directly and those that require an additional layer of machine learning to estimate.
5. The frequency of backchannel words across the corpus
Backchannel words are a foundational element of conversation that occurs at an approximate rate of 1000 per hour of speech; listeners deployed them in nearly two-thirds of speaker turns that were five words or longer.“Generic” continues, such as“uh huh,” may function to signal to speakers that they should keep talking. In contrast,“specific” backchannel words, such as“wow,” may convey context-specific responses such as mirroring a speaker’s emotion while a story is told. This distribution of backchannels reflects English spoken in the United States in 2020.
6. Distribution of gaps and overlaps across speaker transitions.
Negative intervals are classified as “overlaps,” indicating the presence of simultaneous speech on the part of two conversation partners. Positive values are“gaps,” indicating a stretch of silence between turns. The results indicated that the median between-speaker turn interval was +80 ms and was distributed approximately normally. These results are similar to those previously observed in conversations across many cultures and communication modalities. Short gaps are consistently the most common type of turn transition in naturally occurring conversation
