Tusom2021

Tusom2021 Is Based Upon Field Recordings That Were Made In 2004 And Several Passes Of Transcription And Correction Starting In 2004 And Continuing Intermittently Up To 2021, When A Final Version Was Completed. Recording Was Carried Out Using An Inexpensive Lavalier Microphone Attached To The Analog Microphone Jack Of A Laptop Running The Audacity Software Package. Words Were First Elicited And Transcribed With No Recording. The First Author, Who Had Graduate Training In Phonetics And Field Linguistics, Elicited An Item From The Speaker, Then At- Tempted To Imitate It Until The Speaker Confirmed That It Was Correctly Produced, Then Transcribed It. The Words Were Then Divided Into Tranches. Each Tranche Was Recorded In One Session. For Each Item In Each Tranche, The First Author Provided An English Word And The Speaker Responded With An East Tusom Translation, Repeated Three Times. The Transcriptions And The Recordings Were Subsequently Aligned Manually, Using Praat Textgrids. The First Author Then Made A Pass Over The Transcriptions. Subsequently, The First Author Made A Partial Second Pass, Correcting Transcriptions Of Tone. The Second Author Made A Global Pass, Checking All Transcriptions For Correctness And Consistency. Final Consistency Checks Were Done Using A Combination Of Automatic And Manual Methods (Using Python Scripts To Extract All Tokens Transcribed With A Particular Phone, Then Using Auditory Inspection And Acoustic Analysis Using Praat To Winnow Out Distinctions That Had Been Inconsistently Noted In The Previous Pass, Such As Voicing In Plosives).