Conversation design, voice interfaces and artificial intelligence are “fast-growing spaces within technology,” writes U-M alumnus Joseph Tyler (PhD 2012) in a recent social media post. Joseph works as a conversation designer for an award-winning start-up company, where he challenges himself to stay abreast of the latest technology trends and tools. On a visit to the Linguistics Department in November, Joseph discussed some of these trends with graduate students as well as the nature of his experience as a linguist in a career outside academia.
Current Position: Senior Conversation Designer
Joseph works for Sensely, a healthcare tech company in San Francisco that delivers multimodal, avatar-based conversational interactions to help people understand their health and connect them with advice and services.
Joseph’s main areas of expertise are speech perception/production and conversation/discourse. He works on multi-modal conversational interactions in mobile apps, dialogue design, usability testing, natural language processing (NLP)/speech and analytics, localization, and the voice-user interface.
“As a conversation designer within a startup, I wear many hats,” says Joseph. “This includes the core responsibilities of dialogue design, writing the words and building the flows.” Because a single conversational interface product actually includes many different paths and experiences for a user, part of Joseph’s job is to design these paths and the factors that condition what path a user travels through.
Linguistics in Action
During his Ph.D. in Linguistics at U-M, advised by Associate Professor Ezra Keshet, Joseph researched the relationship between speech (intonation, pausing, loudness) and discourse, i.e. how sentences fit together. His work has been published in the peer-reviewed journals Lingua and Discourse Processes.
Joseph says his linguistics background has served him well in his current role as a conversation designer: “My day-to-day job involves writing and revising conversational content, knowing that this could be delivered in text or voice modes. Linguists know that speech and writing are not the same. And while many people may have intuitions about what sounds good and what doesn’t, linguists are particularly adept at explaining why.”
Consider the turn-taking interaction between a human and a computer (as when a person responds to recorded instructions via telephone). It’s important that the user can tell when the computer’s turn is over, which can improve the overall experience. Linguistically, Joseph explains, in a sentence with a subordinate and a main clause, pre-posing the subordinate clause can make it obvious at the interclausal moment that the computer isn’t done talking. If the subordinate clause is post-posed, the user may think the computer is done after completing the main clause, and is then surprised when it continues talking.
Growing Market, Changing Technology
As part of his Linguistics background, Joseph has extensive experience conducting experimental research, surveys, quantitative and qualitative data analysis; he is also fluent in German and French. While Joseph is responsible for a fairly broad range of tasks, he often describes his current position this way:
“I sit at the intermediate place between engineering and the world. My job is to take engineering features (the conversation platform itself, specific new features, app builds) and assemble them into a deliverable conversation product.”
All in all, says Joseph, the demand for skills in conversation design is growing. “Personally, I’m excited to continue to develop my skills in conversation design and build products for conversational interaction,” he says. “It’s an active, growing market, where technology is changing fast. Much like my life in academic linguistics, there is always more to learn.”