Abstract
As the population continues to age, nursing homes will increasingly play a key role in caring for dependent individuals. To enhance the well-being of the elderly, it is crucial to focus on the language skills used during care interactions. However, issues such as the taboo surrounding dependency, scandals involving private nursing home management, the pressure for caregiver efficiency, and the variety of care contexts make monitoring these skills challenging. One way to address this is by collecting in situ data, supervised by language researchers and caregivers specialized in elderly care. This is the approach we have followed: the data collected was then analyzed using machine learning models to provide caregivers with crucial insights for improving care outcomes. Our research highlights the importance of specific factors in language-based interactions, especially in varied care situations. Notably, we emphasize the careful use of humor and the impact of caregiver experience on the success of care sessions. Consequently, we advocate for caregiver training that is grounded in real-life practice, focusing on context adaptation, active listening, and dialogue with residents.