2.1 Transcribing Disordered Speech: Towards an Educational Resource with Actual Footage

Elisabeth Boyer & Elena Babatsouli, PhD, University of Louisiana at Lafayette

Phonetic transcription is a vital skill for speech language pathologists (SLPs). Phoneticians and SLPs traditionally use the International Phonetic Alphabet (IPA) as their primary tool for transcribing typical or atypical speech to guide research and clinical assessments. Nevertheless, the IPA is inadequate for use in the clinical context as it does not encompass an inventory of symbols for atypical speech sounds in the context of communicative disorder. The Extended IPA (extIPA) and the Voice Quality Symbols (VoQs) chart are available for the phonetic transcription of disordered speech, but they are under-utilized in clinical practice worldwide. This is because most university training on phonetic transcription for SLPs relies on a single undergraduate phonetics class (Shriberg et al., 1997), in which the focus is on typical productions of adult speech in the students’ native language (e.g., Ball et al., 2021). Also, existing phonetics manuals/workbooks lack relevant exercises and accompanying audio/visual examples for use as training material by students or practicing clinicians (Ball et al., 2009). With the exemption of few tutorials (Ball et al., 2009, 2010; Rutter et al., 2010), there are no educational resources for the instruction of extIPA and VoQs utilizing actual live data.             It is common knowledge that there are perceptual limitations in the identification of non-familiar sounds (Best et al., 2016; Flege, 2021), and this has a direct impact on SLPs’ ability to phonetically transcribe unfamiliar sounds, often also relating to how awareness and responsiveness to diversity may affect clinical practicum success (Babatsouli, 2021). Boyer et al. (2023) have embarked on creating a student research experience (SRE)-supportive curriculum that enhances firsthand, evidence-based training in phonetic transcription of atypical speech, utilizing actual footage from a variety of communicative disorder contexts in children and adults (e.g., articulation disorder, apraxia, dysarthria). The present study reports on preliminary results of this project focusing on child stuttering. The data come from FluencyBank English Voices-CWS Corpus (talkbank.org); data excerpts, transcriptions, and extracted segments are archived in Phon (Hedlund & Rose, 2020). Phonetic transcriptions were carried out by two trained phoneticians and interrater transcription reliability was at 80%. Analysis of the data highlights speech sound examples representing various extIPA and VoQs symbols. The disordered speech productions are evidenced in the context of targeted segments produced as clicks (bilabial, glottal and alveolar), laryngeal stops, and variants of laryngeal trills, percussive /p/, often in combination with tongue protrusion and/or jaw shifts, ingressive airflow, and a variety of voice qualities, such as whisper, falsetto, breathiness and allegro speech. The extracted segments are archived as part of the larger project aiming to establish an online showcase corpus, the basis of a practice manual that aligns examples and respective exercises to actual speech. The paper also outlines the organization of the corpus and identifies missing tools necessary for digital transcriptions. The study has significant scientific, educational, and clinical applications, as the ultimate educational resources benefit university coursework, clinical practicum, and speech-language research in communicative disorders at large.