Speech interfaces are growing in popularity. Through a review of 68 research papers this work maps the trends, themes, findings and methods of empirical research on speech interfaces in HCI. We find that most studies are usability/theory-focused or explore wider system experiences, evaluating Wizard of Oz, prototypes, or developed systems by using self-report questionnaires to measure concepts like usability and user attitudes. A thematic analysis of the research found that speech HCI work focuses on nine key topics: system speech production, modality comparison, user speech production, assistive technology & accessibility, design insight, experiences with interactive voice response (IVR) systems, using speech technology for development, people's experiences with intelligent personal assistants (IPAs) and how user memory affects speech interface interaction. From these insights we identify gaps and challenges in speech research, notably the need to develop theories of speech interface interaction, grow critical mass in this domain, increase design work, and expand research from single to multiple user interaction contexts so as to reflect current use contexts. We also highlight the need to improve measure reliability, validity and consistency, in the wild deployment and reduce barriers to building fully functional speech interfaces for research. Author Keywords Speech interfaces; speech HCI; review; speech technology; voice user interfaces Research Highlights• Most papers focused on usability/theory-based or wider system experience research with a focus on Wizard of Oz and developed systems, though a lack of design work • Questionnaires on usability and user attitudes often used but few were reliable or validated • Thematic analysis showed nine primary research topics • Gaps in research critical mass, speech HCI theories, and multiple user contexts

Conversational agents promise conversational interaction but fail to deliver. Efforts often emulate functional rules from human speech, without considering key characteristics that conversation must encapsulate. Given its potential in supporting long-term human-agent relationships, it is paramount that HCI focuses efforts on delivering this promise. We aim to understand what people value in conversation and how this should manifest in agents. Findings from a series of semi-structured interviews show people make a clear dichotomy between social and functional roles of conversation, emphasising the long-term dynamics of bond and trust along with the importance of context and relationship stage in the types of conversations they have. People fundamentally questioned the need for bond and common ground in agent communication, shifting to more utilitarian definitions of conversational qualities. Drawing on these findings we discuss key challenges for conversational agent design, most notably the need to redefine the design parameters for conversational agent interaction.

Humanness is core to speech interface design. Yet little is known about how users conceptualise perceptions of humanness and how people define their interaction with speech interfaces through this. To map these perceptions n=21 participants held dialogues with a human and two speech interface based intelligent personal assistants, and then reflected and compared their experiences using the repertory grid technique. Analysis of the constructs show that perceptions of humanness are multidimensional, focusing on eight key themes: partner knowledge set, interpersonal connection, linguistic content, partner performance and capabilities, conversational interaction, partner identity and role, vocal qualities and behavioral affordances. Through these themes, it is clear that users define the capabilities of speech interfaces differently to humans, seeing them as more formal, fact based, impersonal and less authentic. Based on the findings, we discuss how the themes help to scaffold, categorise and target research and design efforts, considering the appropriateness of emulating humanness.

Limited linguistic coverage for Intelligent Personal Assistants (IPAs) means that many interact in a non-native language. Yet we know little about how IPAs currently support or hinder these users. Through native (L1) and non-native (L2) English speakers interacting with Google Assistant on a smartphone and smart speaker, we aim to understand this more deeply. Interviews revealed that L2 speakers prioritised utterance planning around perceived linguistic limitations, as opposed to L1 speakers prioritising succinctness because of system limitations. L2 speakers see IPAs as insensitive to linguistic needs resulting in failed interaction. L2 speakers clearly preferred using smartphones, as visual feedback supported diagnoses of communication breakdowns whilst allowing time to process query results. Conversely, L1 speakers preferred smart speakers, with audio feedback being seen as sufficient. We discuss the need to tailor the IPA experience for L2 users, emphasising visual feedback whilst reducing the burden of language production. CCS Concepts • Human-centered computing → User studies; Natural language interfaces; Accessibility design and evaluation methods.

The assumptions we make about a dialogue partner's knowledge and communicative ability (i.e. our partner models) can influence our language choices. Although similar processes may operate in human-machine dialogue, the role of design in shaping these models, and their subsequent effects on interaction are not clearly understood. Focusing on synthesis design, we conduct a referential communication experiment to identify the impact of accented speech on lexical choice. In particular, we focus on whether accented speech may encourage the use of lexical alternatives that are relevant to a partner's accent, and how this is may vary when in dialogue with a human or machine. We find that people are more likely to use American English terms when speaking with a US accented partner than an Irish accented partner in both human and machine conditions. This lends support to the proposal that synthesis design can influence partner perception of lexical knowledge, which in turn guide user's lexical choices. We discuss the findings with relation to the nature and dynamics of partner models in human machine dialogue.