Project overview |
About 1.9 million people in Kenya and in diaspora communities around the world speak a language called Kipsigis. However, there are no online materials about Kipsigis for speakers or others interested in learning it--for instance, it's not on Google Translate. In this project, I worked with a cross-functional team to fill this gap by developing an online Kipsigis Living Dictionary with common Kipsigis words and phrases.
|
ChallengeThe Kipsigis community is large and scattered around the world, with very different needs and constraints. What's the best way to create user-friendly resources for this large and diverse community? What features are most important to users, and how can we ensure that these language resources are maximally accessible?
|
SolutionI worked with Kipsigis community members to identify their needs and constraints in a language resource. Based on this feedback, I assembled and led a team that developed an online Kipsigis Living Dictionary containing nearly 150 Kipsigis words and phrases alongside an audio recording for each entry. By involving community members in every stage of the project, from need finding to implementation, we created a dictionary designed for people all over the world with different backgrounds and varying resources.
|
Methods
|
Tools
|
Team
|
My role
- I conceived of the project by identifying the lack of resources about Kipsigis as an area for growth.
- I led need finding with Kipsigis community members.
- I collected language data for the dictionary with native Kipsigis speakers through 1-on-1 and group interviews.
- I created a prototype of the dictionary, involving Kipsigis community members throughout the process.
- I formed and led the team that developed and launched the current Kipsigis Living Dictionary.
My process
At the start of the project, I collaborated with 2 Kipsigis community members to identify their needs and brainstorm what information should be included in a resource about Kipsigis and how this information should be presented. We considered questions like:
Working with collaborators who were deeply familiar with the Kenyan cultural context allowed us to identify these user needs early in the project, which kept us focused on the tools that would be practical for our users from the get-go.
- Who will use this resource? We identified target users including speakers in Kenya, non-speakers in Kenya, speakers in diaspora communities around the world, and non-speakers outside of Kenya who want to learn Kipsigis. These users all have different needs, preferences, and logistical constraints, which guided our development of the resource throughout the project.
- What will users want to know about the language? For all users, a top priority was common Kipsigis words and expressions, like greetings and family terms, which would allow them to have a basic conversation in the language. A secondary priority was information about the grammatical structure of the language, so that people could learn to build their own Kipsigis sentences.
- What practical constraints will users face? Users--especially those in Kenya--face practical constraints including limited internet access, extensive multilingualism, and varying degrees of literacy. Most Kenyans access the internet on their phones, so our resource needed to have a well-developed mobile interface. It also needed to be available in English and Swahili to ensure that everyone could benefit from it, regardless of the language that they felt most comfortable speaking. Finally, it needed to have an audio component to be accessible to users with lower literacy.
Working with collaborators who were deeply familiar with the Kenyan cultural context allowed us to identify these user needs early in the project, which kept us focused on the tools that would be practical for our users from the get-go.
Once we identified what content should go into the resource, I conducted interviews with 2 native Kipsigis speakers to collect the relevant language data. I prepared an interview script with questions about how to say of common words and expressions, while also setting aside time in the interview for speaker-driven digressions, which often led to interesting insights about the language. I worked with speakers 1-on-1 and as a group, since individual and group interviews had different strengths. 1-on-1 interviews allowed me to get through questions efficiently and collect detailed data, while in group interviews, I was able to observe the speakers interacting with each other and speaking Kipsigis more naturally.
During these group interviews, my collaborators and I also discussed how the Kipsigis words and phrases that we collected should be spelled, since Kipsigis doesn't have an accepted writing system. Even though I made audio recordings and transcribed the expressions phonetically, we needed a way to write the language that would be accessible to users without knowledge of phonetic transcription. To achieve this goal, we used a range of methods for writing system development, including checking smartphone autocomplete to see how speakers write these words when texting.
During these group interviews, my collaborators and I also discussed how the Kipsigis words and phrases that we collected should be spelled, since Kipsigis doesn't have an accepted writing system. Even though I made audio recordings and transcribed the expressions phonetically, we needed a way to write the language that would be accessible to users without knowledge of phonetic transcription. To achieve this goal, we used a range of methods for writing system development, including checking smartphone autocomplete to see how speakers write these words when texting.
Next, my Kipsigis collaborators and I explored different tools for building our language resource. First, we considered making a physical dictionary that would be printed and shared with people in Kenya and posted online for others. We identified pros and cons of this option based on the user needs that we'd identified earlier in the project.
|
Pros
|
Cons
|
Given the cons above, we decided that a physical resource wasn't ideal for our project. Instead, we explored the Living Dictionary web app developed by the Living Tongues Institute. After speaking with the manager of the web app, we identified the pros and cons of this option.
|
Pros
|
Cons
|
Given that the pros outweighed the cons, we decided that this web app was the best way to make our Kipsigis language resource.
Next, I created a lightweight prototype dictionary containing about a dozen entries to familiarize myself with the software and to give my Kipsigis collaborators a sense of what the resource would look like. In these entries, I included a word or phrase written with the Kipsigis spelling determined by my Kipsigis collaborators, as well as a phonetic transcription and an English translation. I also included an audio recording, so that users could hear the pronunciation. After I shared the prototype with my Kipsigis collaborators and made some edits based on their feedback, I got the go-ahead from them and took steps to scale the language resource.
Next, I created a lightweight prototype dictionary containing about a dozen entries to familiarize myself with the software and to give my Kipsigis collaborators a sense of what the resource would look like. In these entries, I included a word or phrase written with the Kipsigis spelling determined by my Kipsigis collaborators, as well as a phonetic transcription and an English translation. I also included an audio recording, so that users could hear the pronunciation. After I shared the prototype with my Kipsigis collaborators and made some edits based on their feedback, I got the go-ahead from them and took steps to scale the language resource.
Through UC Berkeley's Linguistics Research Apprenticeship Program, I brought on 4 junior research to help bring the dictionary to life. Growing the project team enabled us to make faster progress implementing the dictionary and gave me valuable team management experience.
Under my supervision, these junior researchers learned key background about Kipsigis and built out the content of the dictionary. In just a few weeks, the Kipsigis Living Dictionary went from a prototype to a fully functional online dictionary with nearly 150 entries, 135 of which have linked audio recordings. I also worked with the junior researchers to draft an overview of the structure of Kipsigis, written in plain language accessible to a non-technical audience, which I'm currently editing and will add to the dictionary's "Grammar" tab shortly.
Under my supervision, these junior researchers learned key background about Kipsigis and built out the content of the dictionary. In just a few weeks, the Kipsigis Living Dictionary went from a prototype to a fully functional online dictionary with nearly 150 entries, 135 of which have linked audio recordings. I also worked with the junior researchers to draft an overview of the structure of Kipsigis, written in plain language accessible to a non-technical audience, which I'm currently editing and will add to the dictionary's "Grammar" tab shortly.
Dictionary entry with English interface
|
Dictionary entry with Swahili interface
|
Next steps
Here are some additional steps to continue improving the Kipsigis Living Dictionary, with a focus on evaluative research.
|
01
Now that the dictionary is live, I want to identify and track metrics to understand how real-world users engage with the site. How many dictionary entries do users visit? What features get used most or least? This information would inform the development of future features, enabling us to devote resources to the features that users care about. |
02
I would also like to conduct usability testing to understand how the dictionary meets the needs of users across social and geographic contexts. What sticking points do users face when navigating the site? How does the dictionary's usability differ across user groups, and how might we remedy those disparities? |
Skills
Through this project, I gained experience mentoring junior researchers, managing a collaborative project, and working with a cross-functional team including Kipsigis speakers, other researchers, and a website manager. I was involved in every stage of the project, but others played a role at only some points. Keeping the project running smoothly at these transitions required planning and clear communication on my part. This project also provided me with further experience identifying user needs, developing interview materials, and conducting 1-on-1 and group interviews.
