The effect of interaction language on preferences for communication repair strategies in Digital Voice Assistants (DVAs): a comparative study

As digital voice assistants (DVAs) become increasingly prevalent, understanding user preferences for communication repair strategies across various language contexts is crucial. Previous research has focused on conversational user interfaces, but seldom on how interaction language and user characteristics such as prior experience and computer self-efficacy influence these preferences. This study quantitatively explored the impact of interaction language on repair strategy preferences in DVAs using a pairwise comparison method (N=99). Findings indicate that while apologies and direct repeat requests are favoured by both native Arabic (AL1) and non-native English (EL2) speakers, EL2 speakers show a stronger preference for confirmation strategies due to linguistic limitations. Additionally, explanations for breakdowns are more favoured by native speakers; however, qualitative insights reveal that the preference of this strategy depends on context and its timing within the interaction sequence. These results underscore the need for tailored repair mechanisms to enhance effective and personalized DVA interactions.