WORDLE

A speech habit tracking app.

Background

During my exchange term in Finland, my final project in my HCI course involved conceptualizing and prototyping a self-quantifiable smart-watch application. The specifics were not given; it was up to us to come up with an interesting idea that tracked something about a user. As a team of four (Faisal El Hussein, Edward Brown, Luca Lüthi), we came up with a word-tracking application.

Tracking health (or not)

When brainstorming, our first thought was to perhaps focus in health. Or maybe not, because self-quantifiable applications tend to be in the health or personal care field.

After pinpointing a desire to do something different, we decided to focus on an education-centric application, that could track users’ speech habits. Because the goal of a self-quantifiable application is to passively (or consciously) track habits to discern a pattern, the goal of our application was to encourage users to track something less common but very much focused on everyday life.

Assignment Goal

Design a smartwatch interface that collects information about a user’s speech habits, therefore allowing a discernible pattern to be observed and changed if so desired.

Concept

We did not have to build the actual application, but conceptualize the idea and then design an auxiliary smartwatch application interface.

Highlights

Click here to view the inVision prototype.

  • Tracks unfavourable speech habits, such as filler words, profanity, or overused vocabulary.

  • Continually collects audio and sends it to the user’s phone, so that the word can be interpreted and tallied in the database.

  • Drunk mode can be activated for periods of casual speech, so the user can still wear his/her smartwatch without ruining statistics.

Process

Research

Personas and scenarios were developed for our project.

  • Will, 45, Manager

    Will is a 45 year-old manager who often gives presentations at work. He’s been told that he uses a lot of filler words when he loses his train of thought. Will wants to improve his articulation to increase his professionalism. The workplace also has a lot of non-native English speakers who have trouble conversing with those who mumble and constantly use filler words. Will would like to lead by example and potentially inspire others to work on their communication skills.

  • Sarah, 28, Office worker

    Sarah is a 28 year-old office worker who wants to reduce her slang usage at work. However, during a night out, she wants to be able to speak casually with her friends without ruining her statistics in the app. In this situation, she can activate drunk mode to separate word usage from the ordinary data set.

  • Mikita, 21, International student

    Mikita is an international student from Japan who wants to develop his English-speaking skills. His vocabulary is limited and he finds himself using the same words all the time. He is able to find synonyms in the app for overused words and find areas of language that he could be researching to flesh out his communication skills.


A number of ideal use cases were developed, such as:

  • the user speaks without a heavy accent/dialect

  • the user wears the watch constantly, especially in social settings

  • the smartwatch is consistently within speaking distance of the user and can pick up on his/her voice

  • the application can recognize filler words, such as “um”, “uh”, and “like”

Iterations

Apart from the paper prototype, two main iterations were done.

User Research & Testing

We went through two rounds of user testing. The first evaluation commenced after producing the paper prototype to evaluate our interface design. From this first round of feedback, we considered design heuristics and sought peer reviews in class.

Test 1

Users were asked to perform three tasks: two easy ones and one harder one.

  1. Add a new word to your pinned list.

  2. Compare yourself to the average user regarding filler word 1, and return to screen one.

  3. Protect your statistics for speaking casually.


From these tasks, which were performed individually, we understood that Android users consistently had trouble with gestures, such as swiping back. However, with a few small hints, they were able to complete the tasks. Despite the application’s intent to be for Apple users, we decided to add a visible back arrow to accommodate both users.

In general, all users appreciated our intuitive and lean interface, and even the idea itself.


After evaluating our idea using Nielson’s heuristics and understanding user needs, we decided that:

  • a loading screen would be highly useful

  • display for voice input should be animated, making the system status visible and engaging

  • put in a back arrow for ease of navigation

Test 2

We chose to gather analytics for our second set of tests. We gathered 5 testers (the recommended number for pattern detection) and asked a series of questions, such as their age, gender, and quantifiable questions such as rating ease of navigation, recoverability from mistakes.

One issue observed, however, was the attention of participants as they completed the survey. The scales were rated from left as strongly agreeing to right as strongly disagreeing, with neutral in the middle. However, when reading the question “The interface is cluttered”, many participants spoke their thoughts aloud about how clean it was, yet rated it as strongly disagreeing.

Other improvements could have been made, especially in terms of testing. The InVision prototype testing was done on the computer, but there were occasionally glitches where the application would freeze. However, it was discovered after that if opened on a smartphone, the user experience was a bit smoother and could have led users to make different decisions on their evaluation of the interface. Furthermore, there was only one issue regarding the interface itself. Nearly every person interviewed mentioned that the speaker symbol in the list screen was confusing. They were expecting the smartwatch to spell the word. We could have improved this by using a more appropriate icon, e.g. a microphone.

Most of our audience were non-native English speakers, who cound the application useful. However, many users also said they would not use the application, if it was on the market, due to lack of access to a smartwatch and constantly checking their statistics; while the application ran in the background, it required conscious work and effort to use the app as intended.

Data

Reflection

The assignment itself was very interesting to interpret and complete. There are things I would’ve done differently, such as provide more depth to the prototype, or at least flesh out logistics a bit further. But in general, I’m satisfied with the team that I had, the research we conducted, and the research paper we wrote and turned in.

After completing this, I learned more about user testing and gathering quantifiable data to build a solid understanding of my users and for iterative purposes. The testers were selected at random, but were from a batch of foreign exchange students, many of which were not native English speakers. I felt that while this was a niche audience, it turned out to work to our advantage, because many of our peers were interested in improving their language skills.