Software | Lenguajes de especialidad, corpus lingüísticos y lingüística inglesa aplicada a la ingeniería del conocimiento

Dr Begoña Bellés Fortuño
Universitat Jaume I, Spain

Facultad de Letras, 19 June, 19:00

Sala Jorge Guillén, Facultad de Letras

Abstract (Español)

La transcripción médica se originó a principios del siglo XX con los taquígrafos. La adopción de la tecnología de reconocimiento automático del habla (ASR, por sus siglas en inglés) con un solo hablante se generalizó en el sistema sanitario a mediados de la década de 1990, facilitada por la proliferación de las tecnologías ASR y de procesamiento del lenguaje natural (NLP, por sus siglas en inglés) (Chiu et al., 2017). Además de su papel crucial en la asistencia a los dictados de los médicos, estas tecnologías han sido objeto de estudios relacionados con la documentación clínica (Hodgson y Coiera, 2015) y el examen de pacientes con afasia (Abad et al., 2013; Jamal et al., 2017) o implantes cocleares (Zhao et al., 2020). Más recientemente, la globalización y los movimientos migratorios han repercutido en la forma en que se producen las conversaciones en el ámbito médico entre pacientes y profesionales sanitarios. En el caso de las deficiencias auditivas, las personas no son capaces de oír los sonidos adecuadamente. Dada la escasez de personal humano, existe una necesidad acuciante de tecnología ASR que pueda agilizar la transcripción de las citas médicas.

Este estudio pretende diseñar y desarrollar una app, PhonicSoft, para la mejora de la comunicación clínica y la optimización de los tratamientos médicos en pacientes con hipoacusia. Para lograr este objetivo, la aplicación se probará con un grupo preseleccionado de pacientes con hipoacusia. El estudio, llevado a cabo en la Comunidad Valenciana, tiene en cuenta la condición bilingüe de la región, donde el castellano y el catalán son lenguas cooficiales y deben tenerse en cuenta a la hora de desarrollar el software PhonicSoft. Se realizarán entrevistas a lo largo de todo el proceso para evaluar sus puntos débiles, así como la satisfacción de los pacientes. La participación puede contribuir a mejorar y ampliar las funcionalidades de la versión beta de la app. Esta tecnología ASR puede ser una herramienta aplicable a otras patologías dentro de la hipoacusia, como la afasia. De esta forma, se puede mejorar la comunicación médico-paciente en entornos clínicos ya que favorece la progresión de la calidad, la comprensión y la inclusión de los pacientes con diversidad funcional auditiva, tanto en la consulta médica, como en otras intervenciones clínicas o entornos.

Abstract

Medical transcription originated in the early 20th century with stenographers. The adoption of single-speaker Automatic Speech Recognition (ASR) technology became prevalent in the healthcare system around the mid-1990s, facilitated by the proliferation of ASR and Natural Language Processing (NLP) technologies (Chiu et al., 2017). In addition to their crucial role in assisting doctor dictations, these technologies have been the focus of studies related to clinical documentation (Hodgson & Coiera, 2015) and the examination of patients with aphasia (Abad et al., 2013; Jamal et al., 2017) or cochlear implants (Zhao et al., 2020). More recently, globalization and migratory movements have had an impact on the way conversations in the medical field between patients and healthcare providers take place. In the case of hearing impairment, individuals are not able to hear sounds adequately. Given the scarcity of human staff, there is a pressing need for ASR technology that can expedite the transcription of medical appointments.

This study seeks to design and develop an app, PhonicSoft, for the improvement of clinical communication and the optimisation of medical treatments for patients with hearing impairment. To achieve this goal, the app is to be tested with a pre-selected group of patients with hypoacusis. The study, carried out in the Valencian Community, is concerned with the bilingual condition of the regional area, where both Spanish and Catalan are co-official languages and should be taken into consideration when developing the PhonicSoft software. Interviews will be conducted throughout the whole process to assess its weaknesses as well as the patients’ satisfaction. Participation can contribute to improving and expanding the features of the beta version of the app. This ASR technology can be a tool applicable to other pathologies within hypoacusis, such as aphasia. In this way, doctor-patient communication can be enhanced in clinical environments since it favors the progression of quality, the understanding and inclusion of patients with auditory functional diversity, both in the medical consultation, other clinical interventions or environments.

References

Abad, Alberto, Pompili, Anna, Costa, Ángela, Trancoso, Isabel, Fonseca, José, Leal, Gabriela, Farrajota, Luisa, and Pavão Martins, Isabel (2013). Automatic word naming recognition for an on-line aphasia treatment system. Computer Speech & Language 27: 1235-1248.

Chiu, Chung-Cheng, Tripathi, Anshuman, Chou, Katherine, Co, Chris, Jaitly, Navdeep, Jaunzeikare, Diana, Kannan, Anjuli, Nguyen, Patrick, Sak, Hasim, Sankar, Ananth, Tansuwan, Justin, Wan, Nathan, Wu, Yonghui, and Zhang, Xuedong. (2017). Speech recognition for medical conversations. arXiv preprint arXiv:1711.07274

Jamal, Norezmi; Shanta, Shahnoor; Mahmud, Farhanahani; & Sha’abani, MNAH (2017, September). “Automatic speech recognition (ASR) based approach for speech therapy of aphasic patients: A review”, in AIP Conference Proceedings (Vol. 1883, No. 1). AIP Publishing.

Hodgson, Tobias, and Enrico Coiera. (2016). Risks and benefits of speech recognition for clinical documentation: A systematic review. Journal of the American medical informatics association 23: e169-e179.

Zhao EE, Dornhoffer JR, Loftus C, et al. (2020). Association of patient-related factors with adult cochlear implant speech recognition outcomes: A meta-analysis. JAMA Otolaryngol Head Neck Surg 146(7): 613-620. doi:10.1001/jamaoto.2020.0662

Begoña Bellés Fortuño, PhD, is a Senior Lecturer in the Department of English Studies at Universitat Jaume I (UJI, Castellón), Spain, where she lectures English Studies degree students as well as in the degree of Medicine and supervises MA projects in the Secondary Education, Vocational Training and Language Teaching master’s degree. She is currently the Director of the Interuniversity Institute of Modern Applied Languages (IULMA), a research institute that serves as a bridge between language in the different specialised domains and society. She is a member of the GENTT research group which main areas include the analysis of the Discourse of Medicine and clinical setting communication encounters focusing on the analyses of multilingual communication in the fields of Medicine. She has participated in several projects with the GENTT group analysing multilingual communication as it takes place in the bilingual region of Comunidad Valenciana in patient-doctor encounters as well other projects focusing on the improvement of the Informed Consent (IC) genre. Some projects include: Creation of resources for teaching communicative skills in medical and healthcare fields funded by the Valencian Government and Creation of multilingual resources for improving doctor-patient communication in Public Health Services funded by the Spanish Ministry of Science, Innovation and Universities. She has also directed the project Development of online multilingual resources for improving the Informed consent as a multimodal communication act, funded by Universitat Jaume I R+D+I programme.
Member of associations such as AESLA (Asociación Española de Lingüística Aplicada) and vice president of AELFE (Asociación Europea de Lenguas para Fines Específicos), where she actively participates and is the coordinator of two research panels, Discourse Analysis and Genre and Discourse, respectively. More recently, also a member of EALTHY (European Association of Language Teachers for Healthcare) bringing the 4th English for Healthcare conference for the first time to Spain at Universitat Jaume I in October 2019 (https://www.englishforhealthcare.com/). Begoña is the Language Value journal Editor-In Chief and one of the executive directors of IBERICA journal. She has reviewed articles for JEAP, System or Language and Communication among other journals. Recent publications include the article Multimodality in medicine: How university medical students approach informative leaflets, SYSTEM Vol. 77 (2018), Evaluative language in medical discourse: A contrastive study between English and Spanish university lectures, LANGUAGES IN CONTRAST 18(2) (2018), the chapter Popular Science Articles vs Scientific Articles: A Tool for Medical Education in Ordoñez-Lopez, P. & Edo-Marzá, N. (2016) Medical Discourse in Professional, Academic and Popular Settings, Multilingual Matters or the article Accommodating the Syllabus to Visually Impaired Students in the English Language Classroom: Challenges and Concerns, INTERNATIONAL JOURNAL OF ENGLISH STUDIES 21 (1) (2021). In the educational side, she runs an educational project with a multidisciplinary team of content lecturers called CLHIOS which aim is to introduce CLIL/EMI pedagogies in the university classroom within different disciplines of the STEM.

–Corpus Linguistics and the Analysis of L2 Spoken and Written Texts
Dr Kris Kyle, University of Oregon
26 October, 19:00 (Madrid-Paris-Brussels-Berlin time) Zoom registration link

Productive lexical proficiency has been an important topic in applied linguistics for over 25 years (e.g., Crossley et al., 2011; Kyle & Crossley, 2015; Laufer & Nation, 1995). During this time, word frequency measures have played a dominant role (Laufer & Nation, 1995). While word frequency is undoubtedly important, a number of recent studies have demonstrated that lexical proficiency is most accurately modeled when multiple lexical and lexicogrammatical features are used (e.g., Kim et al., 2018; Kyle et al., 2018;). In this talk, an overview of selected measures of lexical proficiency at the word (e.g., concreteness, contextual diversity, lexical access, etc.) and lexicogrammatical (i.e., n-gram, dependency relations and verb-verb argument construction strength of association) level is provided. The use of these features is then highlighted in two learner corpus research studies.

Kristopher Kyle is an Assistant Professor in the Department of Linguistics at the University of Oregon. His research interests include second language acquisition, second language writing, and second language assessment. He is addresses these topics using corpus linguistic methods through the adaptation and development of natural language processing tools.

Coordination: Prof Pascual Pérez-Paredes