FluencyBank English Davis Corpus

	Eleonora Beier Department of Psychology University of California, Davis ejbeier@ucdavis.edu		Nene (Suphasiree) Chantavarin Department of Psychology University of California, Davis schantavarin@ucdavis.edu
	Fernanda Ferreira Department of Psychology University of California, Davis fferreira@ucdavis.edu

Beier, E., Chantavarin, S., & Ferreira, F. (submitted). Age doesn’t matter, but speech rate does: A longitudinal corpus study of disfluencies.

In accordance with TalkBank rules, any use of data from this corpus must be accompanied by at least one of the above references.

The disfluency_data.csv file includes these variables:

Interviewee: person being interviewed
Interview_age: age of the interviewee
Interview_year: year the interview took place
Occupation: occupation category for the interviewee
N_segments: number of interview segments selected from longer interviews in order to reach 2 minutes of the interviewee’s speech
N_interviewer_segments: number of short interjections by someone other than the interviewee (e.g., the interviewer) within the selected interview segments
Total_time: duration of all interview segments in seconds
Interviewer_time: duration of all interviewer interjections in seconds
Speech_only_time: duration of interviewee-only speech in seconds (Total_time – Interviewer_time)
N_words: number of words uttered by the interviewee
UhUhm_Count: number of filled pauses uttered by the interviewee
Repeats_Count: number of repeats uttered by the interviewee
Repairs_Count: number of repairs uttered by the interviewee
N_syllables: number of syllables uttered by the interviewee, calculated using the nsyllable function in the R package quanteda (version 2.0.1, Benoit et al., 2018)

Additional files include:

CohMetrix_data.csv: data used to perform exploratory analyses. Interview transcriptions were analyzed using Coh-Metrix 3.0 (Graesser et al., 2004; McNamara et al., 2014). For a description of each variable, see http://cohmetrix.com/.
Transcriptions_full: folder containing text files for all interview transcriptions in their “raw” form. Each transcription contains all interview segments, and includes short interjections by the interviewer or others. {Curly brackets} denote the beginning and end of each interview segment. [Square brackets] denote the beginning and end of speech by anyone other than the interviewee (e.g., the interviewer). Unintelligible words are denoted as (?). These transcriptions also include non-words (e.g., uhm, hmm, laughs) which were later discarded; a separate count of uhs and ums was used in the analyses.
Transcriptions_clean: folder containing text files for all interview transcriptions as they were used in our analyses (e.g., to compute word and syllable rate). Speech by anyone other than the interviewee was removed. All punctuation and all non-words were also removed.
statistical_analyses.R: mixed effects models used for all analyses, using words per second as a measure of speech rate
statistical_analyses_syllablerate.R: mixed effects models used for all analyses, using syllables per second as a measure of speech rate