Overview
Before starting this project, my team was given the challenge of coming up with a way to help Duolingo leverage community to make up for the limitations with AI chatbots. We aimed to find out what tools would help language learners reach oral proficiency and use that data to inform our solution.
Over the course of three weeks, I worked alongside 3 other UX designers to complete the following:
Collect and synthesize data on the social side of language learning through user interviews and competitive analysis
Establish a persona, journey map, problem statement, how might we statement, user flow, style guide, and some grayscale low-fidelity wireframes
Add fidelity to the existing wireframes, create new high-fidelity pages to improve clarity and ease of use, connect the frames to produce a working prototype, and run the prototype through usability testing
Research
User interviews
As a team, we recruited 7 interviewees, all adults learning languages that they do not already consider themselves fluent in. We came up with a script based on a specific research objective: to understand how community and social interactions can help people become proficient or fluent in a language through interviews.
After collecting our data, transcribing it onto sticky notes, and creating an affinity map to see what themes existed, we came up with a few main takeaways:
Language learners learn best when practicing with family
They need opportunities to practice in their everyday life
They value social connections when learning languages
They want to find a language learning partner who meets their specific criteria
What went well: the data we collected gave us really valuable insight into what helps our users learn and what pain points they experience when learning a new language
What can be improved: although all of our team members reached out to a wide network of language learners, we unknowingly collected data mostly from heritage speakers. This caused our data to represent a more specific demographic than we were aiming for. In the future, ensuring the sample is more varied will be a priority when recruiting.
Competitive analysis
To gather additional data, Nancy, one of our team members, looked at other language learning tools. Comparing Duolingo to some of the other top language learning apps emphasizes its lack of social features available. The competitor who leaned the most heavily into the social side of language learning was Hello Talk, which had a number of social features.
What went well: comparing to competitors gave us insight into what social features already exist and are available on other apps.
What can be improved: when looking to compare specifically social features, it would have been beneficial to compare Duolingo to other apps that focus more on social aspects. Some of the competitors do not provide very helpful insight into what social learning tools are out there.
Focusing in
Persona: introducing Sam
Sam is a 27 year old extrovert who is learning Vietnamese using Duolingo. She wants to practice speaking Vietnamese with people the same age as her who are at around the same level. She lacks opportunities to practice applying her learning in everyday life.
Keeping this persona in mind, we defined a problem statement that expresses what we need to solve and why: Sam is plateauing in her language learning and needs social engagement in order to build proficiency in the language she is learning
With the problem clearly defined, we began thinking about solutions. We asked ourselves, “how might we use a social element in Duolingo to supplement Sam’s Vietnamese skills?”
What could be improved: because the data gathered from interviews mostly focused on the learning journeys of heritage speakers, our persona followed suit. While aiming to create a solution that helped the average language learner, we mistakenly focused on one specific group. We did return to the persona to remove the heritage speaker aspect from her story.
Journey map
Our journey map follows Sam as she attempts to speak Vietnamese to a family friend and is embarrassed when the friend notices her lack of speaking skills and speaks to her in English instead of Vietnamese.
Design
User flows
Our solution was expressed in two user flows.
Flow 1: the first flow followed the user as they looked for a language learning partner, filtered by preference, sent a friend request, awaited a response, and scheduled a lesson with their new friend.
Flow 2: the second flow followed the user as they attend their lesson with their language learning partner. They connect to a voice call, take the lesson with their partner, and have the option to schedule another lesson.
Sketches and low fidelity wireframes
In order to get as many ideas out as possible, we all made quick sketches of what we envisioned the new feature looking like. From there, we discussed the features that appeared in the sketches that best addressed the problem we had defined. We chose the sketches that we collectively decided were the most intuitive and would mesh well with the app.
The final sketches provided a basis for greyscale wireframes that established margins consistent with the existing app as well as a general layout for the higher fidelity frames.
High fidelity frames
Sticking to the existing color scheme and typography style that makes Duolingo so recognizable, we brought our frames from low fidelity, greyscale frames to frames that would fit right into the colorful Duolingo app.
Deliver
Prototype
With all of our high fidelity frames made, we were ready to connect them to make a working prototype. The prototype follows the two user flows we focused on, starting by adding a friend and then taking a lesson with them.
Feel free to explore the prototype below
Usability testing
To test the usability and intuitiveness of our prototype, we recruited 6 testers to run through it. They were given 2 tasks:
Find a new friend and add them then schedule a lesson with the friend
Attend the scheduled lesson
4/6 users were able to complete task 1
6/6 users were able to complete task 2
4/6 users said they would use the new feature
The users who ran through the prototype were able to give some key feedback. There were various parts that they found confusing, though a lot of the confusion came from issues with the script. Some of the feedback had to do with being unfamiliar with new icon, and some with Figma limitations, but the rest gave us a guide to what our next steps should be
Next steps
Add more safety features to ensure the social part of the app is a safe and comfortable space
Work on the accessibility issue that arises from the use of white text on a green background. Because this goes against the current branding, it was not applied in the first round of design.
Do more usability testing with each new iteration to make sure we are moving in the right direction