The Topography of Disfluency in Spoken Language: A Scoping Review

Speech disfluencies have been widely studied across speech-language pathology (SLP), linguistics, and engineering. However, the parameters used to describe disfluencies vary considerably across studies, partly because disfluencies are conceptualized differently across disciplines - either as observable speech features or as indicators of underlying speech production processes. As a result, the full range of documented disfluency types remains unclear, and cross-study comparisons are often limited. This scoping review systematically synthesizes existing research to map all documented disfluency types, clarify cross-disciplinary and cross-linguistic inconsistencies, and provide an overview of existing classification frameworks.

Following the PRISMA-ScR guidelines for scoping reviews, online databases (PubMed, Ovid Medline, APA PsycInfo, Linguistics & Language Behavior Abstracts: LLBA, Alexander Street Literature, IEEE Xplore, ACM Digital Library) were searched for papers published from 2000 to 2023, with relevant pre-2000 references included if cited in eligible studies. A total of 69 studies met the inclusion criteria, including 30 from SLP, 16 from linguistics, and 23 from engineering. Disfluency types were grouped into seven broad categories: repetition, prolongation, revision, pause, filler, incomplete production, and nonverbal behaviors. Inconsistencies in terminology and definitions were prevalent across fields, and most frameworks were English-centric, with limited adaptation to other languages. Only 29% of studies applied classification systems to non-English languages, including Mandarin, Cantonese, Spanish, and Turkish.

Terminology and definitions of disfluency types lack consistency across SLP, linguistics, and engineering. Developing a unified, cross-disciplinary, and cross-linguistic classification framework may facilitate more consistent research practices and improve the comparability of findings across studies. Such efforts may also support future clinical and technological applications involving speech disfluency analysis.