Appropriate Accommodations: Speech Technologies and the Needs of Older Adults

Older adults interacting with speech technologies may benefit from a range of communicative accommodations. Potential accommodations include volume, intonation, and sentence structure, to name a few. This paper reviews the literature on human communication with the elderly in search of recommendations for speech interface design. We conclude that spontaneous human behavior cannot easily be taken as a guide in designing communicative interactions for older adults. Due to substantial variability in the population, successful accommodations are largely dependent on the specifics of speaker and situation. Regrettably, humans are frequently not attuned to these specifics, and inappropriate accommodation is often perceived as patronizing. Speech technologies present an opportunity to offer accommodations appropriate to the specific communication needs and social values of individual users. Acknowledging the limitations of using research between human communicators to inform the design of speech interfaces, we offer considerations for further research on appropriate communication technologies for older adults. Assistive Technologies for Seniors To a large extent, work on assistive technologies for seniors focuses on monitoring the state of the older adult in the home. Projects using various sensor technologies seek to inform family caregivers about aspects of the older adult’s health, if the adult is not eating regular meals, for example, or spending a great deal of time in bed (Mynatt et al. 2001; Haigh et al. 2003). Other interfaces give the feedback directly to the older adult in situations where the stove has been left on or a daily medication has not been taken (Pollack et al. 2002). Monitoring technologies may ultimately be able to provide cognitive assistance for a wide range of activities. In the future, an aware home may recognize that a common sequence of events has been only partially completed and begin prompting the user to complete the remaining steps. The overarching goal of this work is to support “aging in place,” allowing seniors to retain their autonomy by remaining in their own homes. Supposing that speech interfaces will become more prevalent in applications that provide cognitive assistance, it is wise to consider how they should be designed. Among humans, communication is facilitated by an exchange of linguistic and prosodic adjustments between speakers. Some adjustments take place as a subconscious reaction to features of the conversational partner’s speech, but other adjustments seem to be based on an estimate of the partner’s communication needs. When considering the design of speech technologies, there are a range of features that could potentially accommodate the specific communication requirements of the elderly population. This paper reviews the literature on human communication with the elderly in search of recommendations for speech interface design. We conclude that spontaneous human behavior cannot easily be taken as a guide in designing for older adults. Due to substantial variability in the population, successful accommodations are largely dependent on the specifics of speaker and situation. Regrettably, humans are frequently not attuned to these specifics, and inappropriate accommodation is often perceived as patronizing. Speech technologies are an opportunity to offer accommodations appropriate to the specific communication needs and social values of individual users. Acknowledging the limitations of using research between human communicators to inform the design of speech interfaces, we offer considerations for further research on appropriate communication technologies for older adults. Human-Human Communication One of the most robust findings from the literature on communication with the elderly is that adults do not speak to the elderly the way they speak to adults their own age. In reviewing the literature on this phenomenon, we will look first at what constitutes this difference in communication behavior. We will then consider several functions of this behavior from the perspective of both the speaker and the listener. Finally, we will review a general model of this phenomenon.