Utilizing voice technology, like Amazon’s Alexa, Apple’s Siri, and Google Assistant, can streamline daily tasks and boost productivity. However, it is common to encounter errors in speech recognition and understanding during interactions with these devices. Users often modify their speech patterns, adopting a louder and slower register known as technology-directed speech.
Research on technology-directed speech has primarily focused on mainstream varieties of U.S. English, overlooking speaker groups that face consistent misunderstandings by voice technology. In a study published in JASA Express Letters by researchers from Google Research, the University of California, Davis, and Stanford University, the impact of technology-directed speech on African American English (AAE) speakers was examined.
AAE speakers are frequently misunderstood by voice technology, leading to higher rates of automatic speech recognition errors. This linguistic discrimination can have significant repercussions on various sectors that rely on voice technology, including healthcare and employment.
To address this issue, the researchers engaged with Black users to understand their emotional, behavioral, and linguistic responses when interacting with voice technology. An experiment was designed to observe how AAE speakers adjust their speech when communicating with a voice assistant compared to a human counterpart.
The study involved 19 Black or African American adults who had encountered difficulties with voice technology. Participants were asked to pose questions to a voice assistant, a familiar person, and a stranger, with their speech patterns analyzed for differences in speech rate and pitch variation.
The results indicated that AAE speakers exhibited a slower speech rate and less pitch variation (more monotone speech) when interacting with voice technology. This suggests that individuals develop specific communication strategies when engaging with technology to enhance understanding due to disparities in speech recognition systems.
Aside from AAE speakers, other groups, like second-language speakers, also face challenges with voice technology. The researchers aim to broaden the scope of language varieties explored in human-computer interaction studies to address existing barriers in technology and ensure inclusivity for all users.
