Language Patterns and Behaviour of the Peer Supporters in Multilingual Healthcare Conversational Forums

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 963–975 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 963 Language Patterns and Behaviour of the Peer Supporters in Multilingual Healthcare Conversational Forums 1Ishani Mondal, 1Kalika Bali, 1Mohit Jain, 1Monojit Choudhury, 2Keshet Ronen, 3Millicent Ochieng, 3Kagonya Awori, 3Jacki O’Neill 1Microsoft Research India, 2University of Washington, 3Microsoft Africa Research Institute, Kenya {t-imonda, kalikab, monojitc}@microsoft.com Abstract In this work, we conduct a quantitative linguistic analysis of the language usage patterns of multilingual peer supporters in two health-focused WhatsApp groups in Kenya comprising of youth living with HIV. Even though the language of communication for the group was predominantly English, we observe frequent use of Kiswahili, Sheng and code-mixing among the three languages. We present an analysis of language choice and its accommodation, different functions of code-mixing, and relationship between sentiment and code-mixing. To explore the effectiveness of off-the-shelf Language Technologies (LT) in such situations, we attempt to build a sentiment analyzer for this dataset. Our experiments demonstrate the challenges of developing LT and therefore effective interventions for such forums and languages. We provide recommendations for language resources that should be built to address these challenges. Keywords: Multilingual Conversations, Healthcare, Peer support groups, Code-mixing 1. Introduction In recent years, various online chat applications such as WhatsApp and WeChat are being increasingly lever- aged for establishing peer support groups beyond the formal healthcare system (Karusala et al., 2021; Bhat et al., 2021). Exchange of peer support is critical for attaining improved physical and mental well-being as it helps the individuals deal with uncontrollable and emo- tionally crippling life events. Prior work on peer sup- port forums are limited to studies of social connection and engagement patterns (Sharma et al., 2020a; Sad- eque et al., 2015), and modelling of user behavior (Hos- seini and Caragea, 2021; Sharma et al., 2020b). We are not aware of work that quantitatively studies and char- acterizes the linguistic patterns of interaction in such forums. In this work, we conduct analysis of the language and communication patterns in two WhatsApp based healthcare forums in Kenya, which were used by youth, living with HIV, for discussing health, lifestyle, per- sonal and cultural issues, and connect to other in- dividuals with similar needs and challenges. While there was a significant use of English in the conver- sations, the users were mostly tri-lingual, speaking En- glish, Kiswahili and Sheng (Sheng is a Kiswahili and English-based cant). We also observe frequent code- mixing between pairs or all of the three languages. The objectives of this work are threefold. 1) We wish to understand the linguistic patterns of in- teractions in such forums that fortify peer support and create trust and bonding. 2) In particular, we want to study the relationship be- tween the choice of language, and (a) linguistic accom- modation, (b) conversational intent, and (c) sentiment. 3) We would also like to understand if we can build positive technological interventions for such forums with the …

Content truncated to 500 words