A voz feminina em assistentes virtuais: uma análise pelos estudos da linguagem

Abstract

This research aims to discuss the voice of virtual/digital assistants - computer systems that work as personal assistants run by Artificial Intelligence (AI) - from a perspective of Language Studies. In this study, we focus on the artificial voice of digital assistants such as Siri, Cortana, Alexa and Google Assistant. We assume that the voice, in Western society, is a sonic-linguistic materiality whose meanings are discursively constructed in history. In language, voice is a materiality of significant frequency, that is, it has fundamental frequencies that are materialized in the speech, which signify, produce identifications and position the subjects in social places of speech. The research questions that guide this study are: how do these voices work as a discursive element? What are some of the discursivities about them? Our hypothesis is that the production of virtual assistants of default artificial voice frequency with a gender considered feminine is part of a capitalist axiomatic with discursive constructions of coloniality/modernity identification. We are guided by Critical Applied Linguistics, French Discourse Analysis, post-structuralist and feminist philosophical studies, the decolonial epistemological proposal and specific research on voice and production of meanings in different areas. It is a qualitative and discursive research that presents two moments in the analytical process: in the first stage, we analyzed formal characteristics of the digital voices selected for the study using the free software Praat (experimental corpus); and in the second part, we analyzed other materials that address the voice and that are available on the Internet: the companies’ websites, reports about the natural voice and the artificial voice of these virtual assistants and the Equals report (archive corpus). Through the analyses, with the experimental corpus, there was a regularity in the frequency of the voices that presented an average - which is very close among all of them -, constituting, a discursive regularity. From the archive corpus, we found that the textual aspects pointed out about these voices are, in fact, discursive regularities that are found in the narratives invented for them, in a speech functioning digital assistance service, which is characteristic of the capitalist axiomatics. It is important for Language Studies to consider issues involving digital information and communication technologies (DICT), human-machine relationship and gender relationship.