A sociophonetic analysis of female-sounding virtual assistants

<p>As conversational machines (e.g., Apple's Siri and Amazon's Alexa) are increasingly anthropomorphized by humans and viewed as active interlocutors, it raises questions about the social information indexed by machine voices. This thesis provides a preliminary exploration of the relationship between human sociophonetics, social expectations, and conversational machine voices. An in-depth literature review (a) explores human relationships with and expectations for real and movie robots, (b) discusses the rise of conversational machines, (c) assesses the history of how female human voices have been perceived, and (d) details social-indexical properties associated with F0, vowel formants (F1 and F2), -ING pronunciation, and /s/ center of gravity in human speech. With background context in place, Siri and Alexa's voices were recorded reciting various sentences and passages and analyzed for each of the aforementioned vocal features. Results suggest that sociolinguistic data from studies on human voices could inform hypotheses about how users might characterize conversational machine voices and encourage further consideration of how human and machine sociophonetics might influence each other.</p>

Name thesis:degree_name

Master of Arts (MA)

Level thesis:degree_level

Open Access Thesis

Discipline thesis:degree_discipline

English Language and Literature

Year dc:date.available

2022

Author dc:creator

• Allen, Alyssa

Contributors dc:contributor

• Eric Acton, PhD
• T. Daniel Seely, PhD

• conversational machines
• HMC
• language and gender
• sociolinguistics
• sociophonetics
• Linguistics

Repository record dc:identifier

https://commons.emich.edu/theses/1171

OAI identifier oai:identifier

oai:commons.emich.edu:theses-2543