The Elements of Voice First Style

Chapter 17. Language Design

Let us reiterate our core driving design principle that we have been advocating throughout this book: voicebots are not human and, therefore, the humans who will be interacting with these voicebots should not be forced, tricked, or even “encouraged” to speak to these voicebots as if they, the voicebots, were human. We saw in Chapter 11 how this principle informs the types of prompts to craft—that is, what the voicebot should say. In this chapter, we touch on the other side of the equation: what language we should expect, and want, the user to speak in order to deliver an effective voicebot.

On “Naturalness”

The notion that the user should speak “naturally,” as we have previously noted, commits a basic essentialist double error. First, it assumes that there is indeed one way of speaking that is “natural,” while other ways are “unnatural”; and second, that this “natural” way is in some sense superior to the “unnatural” ways—and all of this naturalness exists and is easily accessible at all times in all circumstances. But in reality, we speak differently when we are addressing, for example, a close friend, a colleague in a professional setting in normal times, a colleague in a professional setting in a time of crisis, a child, our boss, a person who is hard of hearing, or someone who is not fluent in our language. Moreover, we also don’t expect the people we speak with to all speak back to us in that purported “natural” way of speaking. Instead, we modulate ...