An In-ear Chatbot For Language Acquisition

Interaction Design (2017 Spring) Team Project.

In-class Project
Interaction Design
Hedgie Choi
Jisu Lee
Lucie Benevise
Fiche is an in-ear device with an app accompaniment for intermediate language learners who are or will travel abroad to stay in a foreign country for at least a couple of months.
Project Output
  • Mobile application prototype
  • Promotional Video
  • Panel for summer show

1. Problem Space

  • The best way to learn another language is to speak with a native speaker.
  • This is especially true in learning how to listen and speak.
  • But many language-learners find it difficult to do so; they may not know a native speaker, or, even if they do, they may feel ashamed or afraid of making mistakes.

2. Concept Statement

Fiche provides users with the experience of speaking to a native speaker, without the potential humiliation.
Our in-ear device can be used to talk - out loud or through texting - to an AI in a foreign language.
Though the device can be used as a stand-alone piece, there is also an app component that gives users more control over their experience.

Understanding The User

Who are the users?

  • Users are in their 20s - 30s, and are comfortable using digital devices and apps.
  • Users are at least at a beginning - intermediate level.
  • Users are learning through or want to practice in a more immersive methods and environment.
  • Users are going to another country, and want to become comfortable speaking and listening before they go.
  • Another group of users are returning from another country or want to retain and improve the language they learned abroad


Judy Baker

  • University Student
  • Took an introductory class in French
  • has plans to visit France
Judy is a hardworking student in Atlanta, Georgia. She thinks French is a beautiful language and has taken an introductory class in French. She’s planning on visiting France during the summer holidays. Before she goes, she hopes to become comfortable carrying a simple conversation in French. She’s been trying to keep up her vocabulary practice, but she wishes she could practice speaking out loud and thinking on her feet.


  • Be comfortable speaking to a native speaker
  • Travel around France with ease
  • Improve her skills in her favorite language


  • She is scared of making mistakes in front of native speakers.
  • Her vocabulary is adequate but her ability to make sentences is poor.
  • She feels that her language knowledge is limited and lacks context.

User Research

We conducted two sets of structured, formative interviews. The first set of interview targeted university students who speak at least 3 languages, and have lived in a foreign country. The second set of interview targeted international students who are taking Beginning Korean I in Yonsei University. Both sets were analyzed through open coding.
Theme Conclusion Design Implications
What is not covered in class There are some things that are not taught in class that would be useful for everyday life. This provides justification for the app - it is a more practical, daily-use practice app that can teach you things that a textbook can't.
Pronunciation concerns Beginner users may not be able to use our service becuse their pronunciation or grammer is too incorrect. The discrepancy between the ability to understand and the ability to talk means that some users in our user group will only be able to listen to the chatbot, not reply to what it says.
Understanding spoken Korean Understanding is possible, even without a grasp of grammar, because all it takes is a few keywords and some context clues.
Speaking Korean On the other hand, speaking is much more difficult because you must know how to place each word in a grammatical sentence.
Speaking Korean (depending on speaking partner) All interviewees seemed to enjoy the free speaking session during the class. They are all able to understand each other bcause they have similar abilites and a shared pool of vocabulary. Fixed expressions can help situations feel confident speaking. Would there be a way to elicit "fixed expressions" from the "keywords" that the user says? This would allow beginner users to learn expressions that are immediately useful in day to day life.

Talking is very enjoyable as long as you are understood and you can understand what's going on. Perhaps the AI could pick up on what words the user is already comfortable with and incorporate them heavily in the conversation to help the user build confidence.
Conversational failure In the case of communication failure, most are quick to switch back in English because their conversation partner also speaks English. Perhaps we should include an option t switch back to English. Some of the interviewees, however, was motivated to talk to someone who doesn't speak English; so maybe it's more motivating to disallow them to use this function.
Conversation topics Interestingly, all interviewees seemed to divide the nature of their conversation into "daily" and "deep". Though they seemed to view "deep" conversation as good, daily converation also seemed necessary. The app should have two components - one for "daily" conversations and one for "deep" conversations. This function already exists, as the "immersive conversation" tab and "talk by topic" tab.
Conversation pauses & ends Unexpectedly, the interviewees claimed that when a topic was interesting enough, there was no lull in the conversation, even when it was hours long. Originally, we believed that the app and in-ear device should be used constantly through the day. It would talk to you and then pause now and then to simulate a real conversation. It seems that most users have no such pauses in conversation when talking with someone engaging.

Preliminary Sketch


Visual Identity

Promo Video

Panel for Summer Show