Translation technologies and crisis management: exploring the challenges in developing Pidgin English-English bilingual resources for integration into machine translation systems

Abstract During public health crises, rapid dissemination of accurate information in the languages spoken by vulnerable populations is a matter of linguistic justice and, in many cases, a determinant of survival. Although machine translation (MT) systems constitute a scalable solution for multilingual crisis communication, low-resource languages such as West African varieties of Pidgin remain effectively excluded from these technologies owing to the absence of standardized orthography, terminological resources, and parallel corpora. This study examines the obstacles developers would face when integrating COVID-19 terminology into MT systems for Pidgin English. Drawing on a corpus of over 20,000 Cameroon Pidgin English tokens scraped from Facebook at the height of the pandemic and analysed with Sketch Engine, we investigate how speakers rendered key public health concepts, identify the predominant translation strategies, and document the resulting challenges for terminology compilation and downstream MT integration. The analysis focuses on orthographic variation, domain specificity, annotation consistency, preprocessing requirements, and ethical considerations surrounding community consent and representation. The study highlights linguistic and technical barriers preventing Pidgin-speaking communities from benefiting from automated crisis-translation tools and underscores the urgent need for community-driven terminological resources as a prerequisite for inclusive MT development in low-resource settings.