Investigating speech disfluencies in Spontaneous L2 Speech Production:A psycholinguistic study of Indonesian TikTok users

Hardiat, Shera Yanuar (2026) Investigating speech disfluencies in Spontaneous L2 Speech Production:A psycholinguistic study of Indonesian TikTok users. Investigating Speech Disfluencies in Spontaneous L2 Speech Production: A Psycholinguistic Study of Indonesian TikTok Users, ٦ (٢). ٧٣٤-٤٥٤. ISSN 2808-1099

Preview

Text Artikel.pdf Download (400kB) | Preview

Abstract

Speech disfluencies are natural yet meaningful features of spontaneous speech that reflect real-time cognitive processing demands during language production. In second-language (L2) contexts, disfluencies occur more frequently because speakers must simultaneously manage conceptual planning, lexical retrieval, syntactic encoding, and self-monitoring with limited linguistic automaticity. Despite growing research on L2 fluency, disfluency as a psycholinguistic phenomenon remains underexplored in self-initiated digital communication contexts, particularly on platforms such as TikTok, where young adult L2 speakers produce spontaneous, minimally edited English speech as part of self-improvement challenges. This study aims to describe the types of speech disfluencies produced by three Indonesian TikTok users in their spontaneous English speech and to explain the underlying psycholinguistic factors that trigger them. Employing a qualitative descriptive method, nine video samples from the ‘Day 1–30 Speaking English Challenge’ were analyzed through verbatim transcription and qualitative content analysis, drawing on Levelt’s (1989) speech production model and Kormos’ (2006) framework of second language speech processing. The analysis identified 227 disfluency tokens distributed across three major categories: pauses, comprising silent and filled pauses (171 instances total); hesitations, encompassing repetitions, false starts, truncated words, prolongations, and self-corrections (84 instances); and lexicalized fillers (9 instances). Five underlying psycholinguistic factors were identified as triggers: conceptual planning difficulty, lexical retrieval effort, syntactic and phonological encoding demands, working memory load, and self-monitoring behavior, each operating at distinct stages of speech production. While the findings are specific to this dataset and are not intended for broad generalization, they demonstrate that disfluencies in spontaneous L2 digital speech are not random errors but systematic indicators of real-time cognitive processing. The study contributes to the extension of Levelt’s (1989) and Kormos’ (2006) frameworks to digitally mediated speech contexts, with implications for psycholinguistically informed fluency assessment, language pedagogy, and the role of self-initiated digital speaking practice, such as TikTok speaking challenges, in fostering authentic L2 fluency development.

Actions (login required)

View Item