Introduction:
Language discordance, i.e., where the midwife and childbearing person do not share the same language, can trigger the need for language support and multilingual communication practices (MCP) with implications on pedagogy. Midwives from UHSussex NHS Maternity Trust have adopted various multilingual practices, including technology-based solutions, to provide language support¹. However, in the United Kingdom (UK), current language support operates sub-optimally, with significant variations in how patients can access interpreters,^2,3 as well as concerns about quality and trust⁴. This research is part of a larger interdisciplinary study⁵ between Midwifery and Interpreting Studies yielding information about encounters along the midwifery journey (antenatal, intrapartum, postnatal) and variations in MCP use. Ad hoc interpreting may occur, involving the use of a bilingual staff, family member or friend, or professional interpreters may provide in-person or remote interpreting services via telephone. In other cases, language support is provided by using English as a lingua franca or relying on language technology such as pre-translated apps or Google Translate (GT) speech-to-speech or speech-to-text tools.

Material and Methods:
The study adopted a Constructivist Grounded Theory (CGT) approach to the analysis of semi-structure interviews to investigate the experience of stakeholders, including 29 service providers (midwives from UHSussex NHS Maternity Trust) and 15 service users (i.e., language-discordant mothers who had their babies between 2018-2021). The analysis process involved utilizing inductive iterative coding on the dataset, which progressed through three stages - initial coding, focused coding, and theoretical coding - culminating in the development of an abstract theory. This paper will focus particularly on those voices who have been silenced or marginalized due to language support issues i.e., service users who may be entwined with social constructs like race and culture⁶. To conduct this study, a purposeful sample of service users was recruited in their respective mother tongues. The sample consisted of 15 participants, including 6 Arabic speakers, 2 Bengali speakers, 2 Polish speakers, 1 Lao speaker, 2 Thai speakers, and 2 Turkish speakers. Out of these participants, seven were primigravidae, and eight were multigravidas. The experience timeline covers three periods: pre-pandemic (January 2018 to 1st UK Covid-19 National Lockdown- (5 participants); during the pandemic (March 2020-December 2020 -5 participants); emerging from the pandemic (January 2021-December 2021 - 5 participants). Service users’ interviews were conducted from September to November 2022 via telephone (lasting up to 1 hour 35 minutes) with professional interpreters (Arabic, Bengali, Polish and Turkish). Thai participants were able to speak directly in their mother tongue without a professional interpreter since Thai is also the mother tongue of the researcher conducting the interviews. Verbatim transcripts in English of respondents’ answers were produced and analysed concurrently. Follow-up interviews of three participants were used to clarify the study data. Using CGT iterative analysis may provide insight into the rich service users' perceptions of technology for MCP within this diverse group.

Results:
During the CGT analysis of the service users' interviews, several codes (TC 1-6) were generated. Service users’ reported that the difficulties in arranging a professional interpreter (in-person or via telephone) could prompt the use of language technology (TC1). However, the use of GT was met with reticence due to concerns about miscommunication, especially when accents or dialects came into play, and the potential for it to be time-consuming (TC2). Participants acknowledged that relational interaction operates differently between midwife-woman. For instance, a human interpreter can understand emotions, yet participants still liked technology as a backup in language discordant scenarios (TC3). Participants also appreciated language technology because it could help childbearing people to feel confident during communication and generally expressed a preference to talk rather than write using such apps (TC4). Some childbearing mothers already use translation apps to check words in short conversations outside of maternity care, which could aid their personal learning of the host language (TC5). Finally, participants suggested that video calls could be an acceptable mode of connecting to a professional interpreter (TC6).

Conclusions:
The service user dataset provided early insights into practical and relational suggestions for language support in various situations along the maternity journey. Implications of technology intersection with multilingual communication and midwifery care is complex and requires careful consideration. While participants welcomed the use of a professional interpreter via technology (e.g., telephone or video) issues arise with by pre-translated apps and machine translation. Midwifery training perhaps warrants interdisciplinary practice which creates awareness of these technological differences⁷. For instance, incorporating machine translation should not be relied upon as the sole tool for multilingual conversation automation in midwifery, as it could lead to patients agreeing to something without full understanding and potential interactional difficulties⁸.Therefore, further research is needed around technology in language support and its appropriateness to all midwifery care. In turn, this can address midwifery education and training.