Department of Speech and Hearing Sciences
Washington State University
|Type of Study:||naturalistic|
Download CHAT transcripts, ITS files, and metadata
VanDam, M., Jessup, C., & Tully, T. (2016, November). Fathers’ and mothers’ differential talk to sons and daughters with hearing loss. Poster presented at the 172nd Meeting of the Acoustical Society of America and the 5th Joint meeting of the Acoustical Society of America and the Acoustical Society of Japan, Honolulu, HI.
VanDam, M., & Tully, T. (2016, May). Quantity of mothers’ and fathers’ speech to sons and daughters. Talk presented at the 171st Meeting of the Acoustical Society of America, Salt Lake City, UT. [PDF (83KB)]
Heid, H. N. (2016, March). Semi-automated linguistic transcription of daylong audio files. Poster presented at the Showcase for Undergraduate Research and Creative Activities, Pullman, WA.
VanDam, M., & Silbert, N. (2015, November). Fidelity of automatic speech recognition labeling. Talk presented at the American Speech-Language Hearing Association Convention, Denver, CO.
VanDam, M., & Carns, K. (2015, November). Acquisition of syntactic question types in children who are hard of hearing. Poster presented at the American Speech-Language Hearing Association Convention, Denver, CO.
VanDam, M., De Palma, P., Strong, W. E. (2015, May). Fundamental frequency of speech directed to children who have hearing loss. Poster presented at the 169th Meeting of the Acoustical Society of America, Pittsburgh, PA. abstract PDF (44KB)
Matalone, H., Stoddard, C., Bookhout, T., & Carns, K. (2015, April). Transcription reliability of the speech of preschoolers with hearing loss. Poster presented at the National Conference for Undergraduate Research (NCUR), Cheney, WA.
Carns, K. 2015. Question Exposure and Production in Preschoolers Who Are Hard-of-Hearing. Unpublished Washington State University master’s thesis. Available: http://vandammark.com/WSU/Carns_2015.pdf
In accordance with TalkBank rules, any use of data from this corpus must be accompanied by at least one of the above references.
Data consist of 159 five-minute audio files with high-quality human transcriptions and access to the automatic LENA diarization (see below for diarization details). There are three files from each of 53 families. Each five-minute segment was excised from a daylong audio recording collected in the family’s natural or home environment using the LENA system.
Thirty-seven of the recordings are from a family with a toddler with mild-to-moderate hearing loss (HL) and no other documented disability or disorder (11 boys, 26 girls), and sixteen are from families with children who are typically developing (7 boys, 9 girls). The children with hearing loss had a documented better ear pure tone average (BEPTA) hearing loss of 47.6 dB (SD=12.1 dB) and speech intelligibility index (SII) of .76 (SD=.14). Recordings are from children who averaged 29.8 months of age (SD=2.8 months), and age did not vary significantly by hearing status group. The socio-economic status (SES) of the pooled families, estimated on a 12-point scale of the achieved education of the mother, averaged 7.81 (SD=2.36, range=4-12). SES did not significantly vary by hearing status group or sex of the child. All families were traditional, two-parent households. All but two families in the study identified as “white/Caucasian,” with one family in the HL group identifying as “black/African-decent” and one family in the typically-developing group identifying as “mixed race.”
Each of the 53 daylong recordings was subjected to the LENA processing algorithms, outputting an estimated count of conversational turns for each five-minute segment in the day, evenly divided into 12 segments for each clock hour. Each segment was then rank-ordered by number of conversational turns determined by the LENA algorithms, and the three highest-ranked, non-adjacent segments were excised. Transcriptions of each five-minute segment was completed by a trained transcriptionist and verified by a second trained transcriptionist. Disagreements in the transcription were assessed by a panel of three trained judges until consensus was reached. Transcriptions were completed and aligned with the audio source in the CHAT format and checked for compliance with the CHAT conventions. Diarization generated from the LENA system, the .ITS file, is provided for each of the 53 whole-day recordings. The CHAT file for each five-minute segment includes the exact onset and offset times of that segment in the header information. To obtain the LENA diarization record, the approximate onset time in the .ITS file must be located, then excised as needed. Note that the onset and offset times of the extracted waveform will not exactly align with the .ITS file because segment boundaries for the .ITS files are arbitrary according to the determination of the algorithm while five-minute segments are predetermined as having exactly 12 in each hour.
Supported by the following: NSF/SBE-RIDIR: 1539133, 1539129, 1539010; NIH/NIDCD: R01DC009569, DC009560-01S1; WSU Seed Grant: 124172-001; Washington Research Foundation
Usage Restrictions: Email Mark VanDam to discuss how you plan to use these data and for citation requirements. Please advise Mark VanDam (firstname.lastname@example.org) of publications/presentations to append to the list of referring scientific work below.