Queer, Nonbinary, or Ambiguous? Rethinking Voice Labels through Queer Theory in HCI

Paper Title

Queer, Nonbinary, or Ambiguous? Rethinking Voice Labels through Queer Theory in HCI

Publication Info

• Topic area: Voice design in Human-Computer Interaction (HCI) informed by feminist and queer theories.
• Keywords: voice design, queer theory, nonbinary, ambiguous voices, gender-neutral, synthetic voices, inclusive design, gender-expansive, feminist HCI, performative gender.

Background and Problem

• Problem / challenge: Voice technologies predominantly rely on binary gender norms, limiting representation and reinforcing stereotypes. Existing labels like "gender-neutral" or "ambiguous" fail to fully capture the experiences of gender-diverse communities.
• Significance: Inclusive voice design is critical for affirming queer and nonbinary identities, challenging binary thinking, and promoting ethical and socially responsible technology development.
• Motivation and related work: Prior HCI research has explored "gender-neutral" voices but often reproduces binary assumptions. Terms like "ambiguous" and "nonbinary" have emerged as alternatives, but their implications and community preferences remain underexplored. This paper builds on feminist and queer theories to address these gaps.

Solution

• Proposed approach: A framework for using "queer" and "nonbinary" as more accurate alternatives to "ambiguous" in voice design, informed by feminist and queer studies and community perspectives.
• Novelty:
  1. Introduction of "queer" as a conceptual tool for voice design, emphasizing fluidity and resistance to binary norms.
  2. Community-informed survey exploring nonbinary participants' preferences for voice labels.
  3. Practical guidelines for when to use "queer", "nonbinary", or "ambiguous" labels in voice technologies.
• Procedure and key techniques:
  • Theoretical grounding in feminist and queer studies, including Butler and Barad's concepts of performative and relational gender.
  • Survey of 18 nonbinary participants to examine preferences for labeling synthetic voices.
  • Recommendations for reducing gendering in voice assistants and designing queer/nonbinary agents.

Results

• Concrete findings:
  • Out of 72 voice labels provided by 18 nonbinary participants, 39 included gender-expansive terms like "nonbinary" (N=9), "any gender" (N=7), and "mix of masculine and feminine" (N=5).
  • Participants preferred labels that affirm fluidity and multiplicity, such as "nonbinary" and "any gender", over binary descriptors.
• Advantage over baselines: The study highlights the inadequacy of "ambiguous" and "gender-neutral" labels for representing nonbinary identities, advocating for community-informed terminology.
• Experiments / evaluation:
  • Anonymous online survey with nonbinary participants recruited via Prolific.
  • Participants labeled synthetic voices created from gender-expansive speakers, focusing on American English speakers.
  • Limited racial and cultural diversity in the sample constrains generalisability.
• Limitations and future work:
  • Small sample size and limited racial diversity restrict broader applicability.
  • Future research should include qualitative methods like focus groups and interviews to deepen understanding of community preferences.
  • Greater collaboration with queer and nonbinary communities is essential for ethical and inclusive design.

Summary

This paper examines how feminist and queer theories can inform voice design in HCI, emphasizing the need for inclusive and precise labeling of synthetic voices. Through a survey of nonbinary participants, it identifies preferences for labels like "nonbinary" and "any gender" over "ambiguous", highlighting the importance of community-informed terminology. The authors propose practical guidelines for reducing gendering in voice technologies and designing queer/nonbinary agents, advocating for participatory methodologies and interdisciplinary collaboration. This work contributes to ethical and inclusive voice design that affirms gender diversity and challenges binary norms.