@inproceedings{sharma-etal-2026-determinants,
    title = "Determinants of Hesitations and Repetitions in {H}indi Spontaneous Speech",
    author = "Sharma, Eashani  and
      Arun, Ishita  and
      Husain, Samar",
    editor = "Voigt, Rob  and
      Warstadt, Alex  and
      Feldman, Naomi  and
      Linzen, Tal",
    booktitle = "Proceedings of the Society for Computation in Linguistics 2026",
    month = jul,
    year = "2026",
    address = "San Diego, CA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.scil-main.7/",
    pages = "59--71",
    ISBN = "979-8-89176-412-5",
    abstract = "This study investigates the factors that predict disfluencies in Hindi spontaneous speech. In particular, we probe the influence of lexical, syntactic, phonological, and prosodic factors on two kinds of disfluencies, namely, hesitations and repetitions. These disfluencies are probed through both the nature of linguistic factors as well as through the source (preceding vs. following word) of these factors. Our results show that hesitations and repetitions pattern differently during spontaneous speech. Hesitations increase due to lexical, syntactic, as well as articulatory features from both preceding and following words. On the other hand, repetitions arise mainly due to lexical and articulatory factors of the upcoming word. Further, while previous research (e.g., Bell et al., 2009; Dammalapati et al., 2021) on English highlights the importance of upcoming difficulty on disfluencies, our results suggest that previously encountered difficulties can also lead to an increase in disfluencies. This suggests that language typology (SVO vs SOV) can play a critical role in determining the planning process and thereby affecting the distribution of disfluencies in a language. Together, these findings highlight the need for increased cross-linguistic research to understand the nature of incrementality and monitoring of the production system cross-linguistically."
}