CoTalk

"Second Language Learning for Job Success."

Sponsor

Artefact

Role

Prototyper, Designer, Researcher

Team

Daisy Bao, Diandian Ding, Tong Wen, Matthew Eziashi

CoTalk was my capstone project for my MHCID masters program. During the final quarter of the program I was in a team with 3 other students and our goal was to go through the UX process and tackle a problem we had identified earlier in the research stages of the capstone process.





Problem Statement

Many companies are trying to diversify their workforce and hiring talent with strong capacities in professional fields, in order to meet demand. At the same time, more and more international students are coming to the US to pursue higher education and explore their future careers. Some employers hesitate to hire international students, especially those who speak English as a second language, because of their insufficient language skills, although they're trained professionals.

Even if they have professional knowledge, lacking English proficiency creates barriers for non-native speaking graduates to develop their careers and pursue their personal success in the US. How can we help non-native English speaking students who will be pursuing a career in the US become more proficient in English before starting their careers? This is not only a problem for students, but for companies as well. Assimilating into the workforce as easily as possible is the best scenario for both parties.









Background

According to the annual survey from the Institute of International Education, in 2015, nearly a million (974,926) international students were studying in the US, with more than half of them (52%) coming from Asia. Most of these learners are using English as their second language (31% Chinese and 7% Korean). At the same time, in some specific fields such as technology, science and global marketing, non-native speakers are becoming a crucial workforce for companies in the US.





Target User

In order to help us scope down the problem space, we chose non-native speaking international students, graduating from schools in the US as our target user. They're looking to build their careers in English speaking countries. These students have already taken formal English classes and have academic learning experiences. However, their lack of contextual language learning experiences and concern about their language skills could lead them to encounter difficulties in communication at the beginning of their careers.














Research

In order to have a better understanding of the problem space, and user’s needs, we performed secondary and primary research. Our secondary research activities included literature reviews, media scans, and a competitive analysis. we built a holistic knowledge base toward the problem space and our target users. Moreover, we also evaluated the business opportunities through the competitive analysis. Our primary research included interviews, and diary studies with subject matter experts and our target users. These methods allowed us to surface the deeper user needs and concerns to support our design process.









Literature Review

After a literature review on current research, we discovered contextual information is a crucial element of the second language learning process. Instead of knowing the definition of the words, knowing how to use the words, common slang, and grammar is more important. However, the majority of existing language learning applications provide very little context around content.(Culbertson, 1).

The learning curve described by Psychologist Hermann Ebbinghaus shows that new skills or knowledge can be acquired quickly at the beginning while the subsequent learning process becomes much slower. Education, gamification, and a persons motivation are also crucial variables determining the rate and efficiency at which a language is learned.







Competitive Analysis

We looked at the current landscape of language learning products, and analyzed their weaknesses and strengths. We then located areas of untapped potential. Professional Preparation, Networking, Crowdsourced Learning, and Domain Focus are components that current language learning products don’t effectively take advantage of. By including these in our system, we created a unique, focused product that’s perfect for our target users.











Subject Matter Expert Interviews

We performed two rounds of expert interviews based on their different roles in the language learning system. We interviewed teachers with decades of experience, to students studying in Seattle. Their insights gave us many different perspectives on the current landscape for our target users. We accounted for their opinions, pain points, and suggestions, scoping down our design.











Insights



Learning experiences need to be intrinsically and extrinsically rewarding.

According to our research, the initial learning motivation for most language learners is extrinsic. However, we aim to embed intrinsic motivation (internally motivated) in our design, since it can give users a feeling of accomplishment.

Keeping non-native speakers motivated is crucial.

At the beginning, learning motivations and goals are the triggers for non-native speakers to start learning. If motivation isnt strong enough at the start, they won’t enjoy learning because usually they can’t gain feelings of accomplishment.

An "intelligent machine" could avoid the social risk and provide the social interaction.

In most cases, the initial learning motivation for most language learners is extrinsic. However, we aim to embed intrinsic motivation in our design, since it can give the user a feeling of accomplishment, which can keep their motivation for learning high.

Improving the conversational and technical proficiency is important for non-native speakers

In a professional setting, being articulate is key to ensuring effective collaboration. To have better social interaction with other employees, non-native speakers should also be fluent. Lacking of sufficient verbal communication skills, the employee may decrease a team’s working efficiency.





Target User Interviews and Diary Studies

We recorded and documented interviews with participants who fit in our target user group. We had a structured interview guide, and made sure we all took notes during our respective interview sessions. We also did diary studies with participants, having them document their experiences as someone who learned English as a second language on a day to day basis. This gave us valuable insight in the real world scenarios that our target users deal with.





Insights










Ideation

At the beginning of the ideation phase, several design prompts were made based on the research findings for guiding the design direction. I made an experience map based on the data gathered from primary and secondary research. In our first brainstorming session, more than 50 ideas were considered. We analyzed these ideas by criteria like feasibility and practicality. We grouped ideas into several “idea buckets” for scoping. After a series of ideation and evaluation activities, six clear potential design directions were developed.

Both individual brainstorming, and group discussion were used frequently during the ideation process in order to encourage team members to think outside the box. It’s an effective way to gather different ideas from different individuals and stimulate new ideas through group conversation. We focused on the ingenuity of the idea, rather than thinking of the technical constraints. While some ideas may not have been entirely feasible, there are often valuable interaction dynamics in an idea that make it worth remembering. Feedback from instructors and peers was helpful for evaluating and scoping ideas.

I designed an experience map based on our research findings. Our research gave us insight into user’s behaviors, thoughts, and emotional status at each stage of their journey. Bridging the two major parts (the academic and professional setting) throughout the journey, the stages “Looking for jobs in the US”, and “Preparing the work with the company after accepting an offer” were the richest areas to create our solution.





Concept Refinement

We evaluated our 6 ideas by their desirability, usability, usefulness, possibility, and viability, allowing us to select three concepts. We made storyboards for each concept to understand each use case. We also outlined what types of users the concepts would be most benefical to, and created storyboards to visualize the step by step process of how the system would work.











Personas








Prototyping

We created High and Low Fidelity prototypes to put our ideas in the hands of users for feedback. Both prototypes were made with specific research questions, hypothesis's, and goals in mind. We also did a scenario mockup with participants, where we simulated the different types of feedback a target user would receive in our systems.

In our solution, experts are native English speakers, who are in the same domain as the student. For example, if a student's major was computer science, they would be able to connect with software engineers and people in the tech industry.





For the Low-Fidelity prototype, we made a paper prototype which we tested on a mobile phone with the POP mobile app. The focus of this prototype was on the connection, appointment, and conversation features. We tested the prototype with three participants who fit our target user. We also created reserach questions to focus our testing.



For the High-Fidelity prototype, we made detailed comps of the UI to showcase to users. In contrast to the low-fidelity prototype, this version was able to focus more on the look and feel of the system. We recruited participants who fit in our target user group for testing, and a brief follow-up interview. We also created reserach questions for this prototype to guide our designs.





Each team member did a card sorting exercise, where we had participants rank what information they would expect to see on an "expert's" profile. We created research questions focused on understanding if our information architecture was meeting our users needs.





Findings

Users are concerned about the quality of feedback from experts, and also concerned how to choose the right expert to talk to.

Participants preferred instant feedback (feedback in conversation with expert) over delayed feedback. (feedback after the conversation)

Users have concerns when they’re using messaging functions during the conversation. Using messaging features during a session with an expert may be difficult for some learners.

We should provide more flexibility around the appointment starting time. Everyone has different schedules, so ensuring that conversation sessions are convenient for both parties is essential.

English proficiency, professional domain, expert rating, and comments are the most important information for users to see on an expert's profile.

User flow throughout the application is clear. Each participant we tested with understood the flow of information as they navigated the application.









Iteration

Based on the findings from both prototypes and the card sorting activity, we developed a plan for our next iteration to incorporate in our final design. We changed the conversation experience, and added a few new features.

From our low fidelity findings, we changed the design of the "end call" button, and added a way to upload files. We got rid of the "report" that would be generated after a session, and instead added a short period of time after the session ends for users to ask questions and receive feedback.

From our high fidelity findings, we added an indicator so show how much time was remaining in their session with the expert. We gave users the ability to quickly "mark" areas in the conversation they found confusing. We added "English Proficiency", "Overall Experience", and a short description to experts profiles. Payment information for the experts was added to their profiles. We made expert profiles more welcoming by changing the wording, and using emoji's instead of stars for ratings. The section of our system devoted to crowdsourced feedback is called the "channel." Here any user of the platform is able to view others conversations (if they set them to public) and leave feedback. Users can comment, and upvote others comments, providing the original user with another dimension of critique. Comments are timestamped, and how much users help others will have an impact on their ranking.














Design

By creating a mobile application, we were able to reach the largest amount of students, and experts fluent in the English language. CoTalk is an online conversation practicing platform, designed to help students who learned English as a second language improve their professional English and prepare for their careers.





There a many ways of delivering feedback to a learner during a conversation. In order to uncover the best interaction model for providing feedback, we tested several scenarios. The 3 we tested were instant oral feedback, instant on-screen feedback, and delayed feedback, which would be done after the conversation session. We found that instant feedback is the strongest method for users, because being corrected in the moment reinforces learning. And based on our findings, we iterated the design of the conversation section.






Features






Connect


Users are able to connect with others in the same job domain.They can check experts' English level, domain, and rating in the connection list. Users can schedule conversational practice sessions with their connections.

Conversation


Based on the converasation topic, the learner will practice while the expert provides instant feedback. Users can leave marks and provide detailed feedback after the conversation section. Both participants evaluate each other after the session.





Channel


The "Channel" has published conversation records, and is a contextual learning forum. Users can leave questions, comments and also react to others comments when replaying the conversation.





Profile


Users are able to check their rankings and learning progress in the dashboard. Personal learning suggestions will be provided based on the learning data collected during their conversation sessions.





Video

We wanted the video to have a personal feel, and focus on one character through their experiences. Because we had done the research on our target user, it was easy for us to craft a scenario in which a character would benefit from using our sustem. The video was made using Adobe Premiere Pro and After Effects for the motion graphics. We storyboarded the video as a group beforehand, scouted the perfect locations around campus, and utilized green screens on phones to make compositing the UI's in After Effects simple.