As GLAM (galleries, libraries, archives, and museums) institutions rapidly digitise their collections to increase reach, a critical gap remains in understanding how these initiatives serve linguistically diverse local communities, especially beyond assessing footfall metrics. Contemporary research suggests that digital humanities initiatives, especially those offering participatory engagement, can significantly enhance the engagement and accessibility of cultural heritage within Indian GLAM institutions (Gandhi, 2024; Bonney et. al, 2009). However, limited work actually utilises such approaches to assess the visitor engagement or accessibility in India. This paper presents a participatory action research (Vaughn & Jacquez, 2020; Cornish, 2023) that shifts the evaluation authority from heritage experts to heritage site visitors and community translators. We collected the artefact descriptions from a museum in Jodhpur (a cultural heritage city in Western India), which were available in Hindi and English. We then extended their accessibility to the native ‘Marwari’ speaking audience using a mix of three types of translation strategies: (1) human translations by native speakers, (2) machine translations using state-of-the-art Machine Translation (MT) systems, and (3) human post-edits of machine output by native speakers. We then evaluated the engagement of these respective outputs through a mixed-methods approach, presenting all three versions to local visitors of the museum (n=50) and receiving their preferences on a Likert scale, supplemented by a corresponding set of MCQs to validate the assessment.

Our methodology is grounded in two key conference themes: A) Co-creation and participatory digital humanities (Ma et al, 2025), and B) Mapping engagement and evidencing public value. By positioning community speakers as co-creators rather than passive consumers or data sources, and by utilising visitor preferences as a measure of "engagement value", we address the question: Does engagement in vernacular languages, within the context of heritage sites, enhance institutional and cultural accessibility? This finding is critical for understanding the public value of vernacular translations of cultural heritage site artefacts. The results of this study help depart from the technopositivist approach, which presumes machine efficiency without centring on the agency of the native and/or the intended audiences. Our work delivers future insights and directions for policymakers and researchers who wish to enhance engagement with heritage institutions for the vernacular audience. Besides, our findings also reveal significant limitations of commercial MT systems to perform desirably in cultural and heritage domains.