Abstract
The aim of this study is to illustrate a methodological innovation in the analysis of museum texts by examining the linguistic features that support accessibility and comprehension in a corpus of English-language exhibit labels. Set in the M+ art museum in Hong Kong, the research employs a robust mixed-methods approach, integrating ethnographic reception studies with corpusassisted discourse analysis (CADS).
The methodology involved several stages. First, a specialised corpus was constructed, comprising 68 authentic exhibit labels (34 pairs of ‘disputed’ versions) produced by museum professionals between 2023 and 2025. Each pair represents alternative drafts reflecting internal debates among curators about which version should be displayed, ensuring the corpus captures real-world communicative tensions and choices.
To assess audience reception, an on-site survey was conducted with museum visitors, stratified by educational background (undergraduate and postgraduate). Inspired by prior reception studies in museum contexts (Chmiel and Mazur 2012; Leung 2018), the survey instrument included sections on sociodemographic data, perceptions of exhibit labels as a genre, and direct preferences between paired label versions. This design enabled the identification of ‘preferred’ and ‘less preferred’ texts within each educational group. Subsequently, the corpus was divided into subsets according to these preferences. Using the UAM Corpus Tool (O'Donnell 2009), a comparative keyness analysis was performed between the preferred and less-preferred sets for each educational group. Concordance analysis was then applied to interpret keywords in context, revealing patterns in nominalisation, clause complexity, domain-specific vocabulary, and evaluative language.
Results reveal marked differences in linguistic preferences by educational background. Postgraduate visitors favour labels with higher nominalization, complex noun groups, and embedded clauses (e.g., frequent use of possessive markers like ‘its’, conjunctions like ‘which’, and prepositional phrases), reflecting a tolerance for abstraction and density of information. In contrast, undergraduates and visitors with less formal education prefer labels featuring tangible, domain-specific terms (e.g., ‘design’, ‘craft’), contextual markers (e.g., ‘Thai’, ‘Japanese’), and evaluative adjectives (e.g., ‘new’, ‘traditional’). These labels employ more relational verbs in the present tense (‘is’), demonstratives (‘this’), and pronouns (‘it’), supporting straightforward and factual communication.
This methodological triangulation, combining quantitative corpus analysis with qualitative audience reception, enables a nuanced understanding of how linguistic choices in exhibit labels affect accessibility for different visitor profiles. The approach demonstrates the value of small, principled corpora and experimental survey design for applied linguistic research in heritage communication.