Determine the behavioral impact of unequal AI support on minority-language use

Determine the magnitude and duration of the effect by which uneven AI support changes users’ language choices, including whether inferior support in a minority language increases switching to a dominant language.

Background

The paper defines functional exclusion in terms of the tasks users can successfully complete through a language, including speech recognition, translation, conversation, and knowledge retrieval. Inferior performance may increase interaction costs and encourage users to switch to a dominant language, although the paper distinguishes this proposed behavioral effect from language loss or cultural disidentification.

The authors explicitly leave the magnitude and duration of this effect unresolved. Establishing it would require empirical analysis that accounts for language proficiency, task requirements, and users’ prior language preferences.

References

The proposed effect is behavioral: uneven support may change language choice even when cultural identification remains stable. This extends the domain perspective on language shift to the practical affordances of AI systems, while leaving the magnitude and duration of the effect open to investigation.

— From Algorithmic Marginalization to AI-Mediated Re-Centering: Can Culturally Grounded AI Bring Hakka Language and Culture Back into Mainstream Society?  (2610.07672 - Chang, 6 Oct 2026) in Section 2.2, “Representation and functional exclusion”

The case is therefore analytically useful without serving as a demonstration of all proposed relationships. Historical research supports the importance of structural changes in language use. The effects of contemporary AI remain to be examined through designs that connect model behavior with actual user practices and resource circulation.

— From Algorithmic Marginalization to AI-Mediated Re-Centering: Can Culturally Grounded AI Bring Hakka Language and Culture Back into Mainstream Society?  (2610.07672 - Chang, 6 Oct 2026) in Section 4.1, “Historical language shift and contemporary AI environments”