Uta HagenBot: Designs in Embodied Interaction*

The Problem

As computers move more and more into diverse areas of our lives, we are interacting with them less in terms of keyboards and mice. Computers will soon be responding to more “natural” embodied interactions, such as responding to gestures and voice. Can I gain insights into the design of these new interfaces via the framework of embodied interaction?

My Role

Designer

Ten Dollar Words

Embodied Interaction. What does that even mean? It's the concept that computer interaction might improve if designers look at the way people interact with tools in the real world. Also, that people already interact with the world in tangible ways which could translate into better ways to interact with computer interfaces.

I decided to look at voice interfaces and see if I could create an application that helps actors memorize their lines. What kinds of interactions do actors have when they work on their memorization? Time to go look!

Creating an Ethnography

The first thing I did was to assemble an Ethnography of rehearsal. Another ten dollar word! Ethnographies are in-depth studies of human behavior. I conducted five sessions and observed each actor memorize their lines with the help of an line coach. I noted the kinds of interactions each had with the coach and the physical script. Did they cover lines as they spoke so they could test their memorization? What did they say to their line coach? What kind of shared vocabulary did the pair have? I sketched interactions, snapped photos and took copious notes.

A sketch of two actors rehersing side by side. One holds a script.

Fig. 1 - Andrea used her close proximity to be able to glance back and forth at the script—an example of embodied interaction.

Storyboards, User Journeys and Personas, Oh My

This discovery period allowed me to understand the specific ways that actors actually learn their lines. I then went into a brainstorming and analysis mode. I created a storyboard to illustrate my ideas of how Uta HagenBot might work.

A storyboard of how Uta might work.

Fig. 2 - My epic storyboard.

I also created several User Personas. These are not just fictional users, they are based on a data-driven persona method. I plotted my participants on several spectra. I then created the personas from groups of similar participant behaviors. This way I know my personas avoid stereotypes and represent real characteristics from the research.

Charts that show spectra that I used to create personas.

Fig 3. Persona spectrum. No “Mid 20’s Mike” or “Donny Deciders” here.
A user persona named Sajel.

Fig 4. Sajel—now there’s a name for a persona.

I also created a User Journey of the Actor’s path toward memorization:

A user journey of how Actors memorize their lines.

Fig. 5 - This is how memorization goes.

OK, Does Any of This Work?

So far, it's all been research. Will this research translate into an effective design? To find out, it was time to do some user testing. I created a high fidelity prototype of one of the flows of running lines. I got a short section of script and set up a web page that listens for actor’s spoken lines and reads their cue lines aloud. The prototype also displays what the actor actually says so that they can compare it to the script.

A screenshot of the Prototype.

Fig. 6 - The prototype—shoutout to The Bear by Anton Chekhov, translated by Paul Schmidt.

I then tested the prototype with two actors. I had them memorize the scene and give feedback about using the prototype. The prototype was buggy, but the actors liked the idea of a digital assistant and we all feel the design has promise.

Lessons Learned

  • Embodied interactions function best when they are designed with a bit of “looseness.” In the real world, people expect a wide range of similar actions to produce a satisfactory result. Computer interactions would do well to mirror this looseness.
  • Let the framework emerge from the ethnography.I think this is a really important idea. I have encountered researchers who rail against forcing a framework to the data, fearing that doing so can only serve to introduce bias. I think a very lightweight framework might be useful in getting analysis started, but letting ideas emerge from the data seems like the right way to do it. Similarly, don't just rearrange data during analysis—have opinions. Often I have worked on affinity diagrams that result in generic categories like "Technology Used" or "User Suggestions." While it may be useful to group various research results for processing, I think it's worth it to really think about the trends in the data, and see what the “Technology Used” is saying or what all of the “User Suggestions” have in common.
  • Testing is your friend. Nothing about my project was as helpful as putting a prototype in front of the first user. Even that simple, buggy test was helpful. As soon as your project works, show it to someone and see how they react. It's scary, but worth it.

The Final Product

Check out the Uta HagenBot prototype.

Take a look at my final report about Uta!

Back To My Full Portfolio

*For non-actors, Uta Hagen was a German-American acting teacher who wrote the seminal Respect for Acting. It's a very funny joke, trust me.