Fabla: A voice-based ecological assessment method for securely collecting spoken responses to researcher questions

This article reports on the validation of Fabla, a researcher-developed and university-hostedsmartphone app that facilitates naturalistic and secure collection of participants’ spokenresponses to researcher questions. Fabla was developed to meet the need for tools that (a) collectlongitudinal qualitative data and (b) capture speech biomarkers from participants’ naturalenvironments. This study put Fabla to its first empirical test using a repeated-measuresexperimental design in which participants (n=87) completed a one-week voice daily diary via theFabla app, and an identical one-week text-entry daily diary administered via Qualtrics, with diarymethod order counterbalanced and randomized. A pre-registered analysis plan investigated (1)adherence, usability, and acceptability of Fabla, (2) concurrent validity of voice diaries (vs. text-entrydiaries) by comparing linguistic features obtained via each diary method, and (3)differences in the strength of the association between linguistic features and their knownpsychological correlates when assessed by voice vs. text-entry diary. Voice diaries yielded morethan double the mean daily language volume (word count) compared to text-entry diaries andreceived high usability and acceptability ratings. Linguistic markers consistently associated withdepression in prior research were significantly associated with depression symptoms whenassessed via voice but not text-entry diaries, and the difference in correlation magnitude wassignificant. Wordcount-adjusted linguistic patterns were highly correlated between diarymethods, with statistically significant mean differences observed for some linguistic dimensionsin the presence of these associations. Fabla is a promising tool for collecting high-quality speechdata from participants’ naturalistic environments, overcoming multiple limitations of text-entryresponding.