In this post, I’d like to share my experience initializing a crowd-subtitling project for the Endangered Alphabets Project.
The Endangered Alphabets Project is a registered 501(c)(3) nonprofit that supports endangered, minority and indigenous cultures by preserving their writing systems. Its founder and president Tim Brookes is a world authority on endangered writing systems both as a scholar and an artist. His book Endangered Alphabets is the only book on the subject.
I first got in touch with Tim when I saw his tweet recruiting interns and emailed him to apply. There was no job description, so I supposed he was open to any kind of support and told him about my various skills related to localization. After a few email exchanges and a meeting with him, which my friend Celia also attended, he requested a globalization and localization strategy.
Given Celia’s and my expertise in subtitling and the timely beginning of Tim’s Open House Zoom series, we decided to develop a crowd subtitling project while localizing this series.
To engage volunteers with various experience and skill levels, we started by looking for a collaborative platform. It didn’t take us long to settle with Amara, which Celia had used for TED projects. An award-winning subtitling platform, Amara has a free plan and is easy to use for both project owners and participants. However, it did not really have a project management feature and had to be integrated with another platform.
Celia and I had worked with Teambition and Trello and quite liked both, but we decided to use Notion because it offered far more than that with its tons of templates. For example, our welcome message and guidelines are pages. The linked Task Board was created using a Task List template.
With the platforms chosen, we moved on to the most time-consuming step in the preparatory phase—actually setting up the infrastructure, namely establishing processes and creating guidelines.
With the platforms chosen, we moved on to the most time-consuming step in the preparatory phase—actually setting up the infrastructure, namely establishing processes and creating guidelines. The key was to balance engagement and quality. To maximize participation, we had to consider different scenarios and came up with corresponding guidelines. For example, a volunteer might want to help with either transcription, or spotting, or both. Whatever task a volunteer committed to, they might want to work on Amara or another tool of their choice. While high expectations would be unrealistic, we need to set a bottom line to at least ensure any volunteer’s contribution would do more good than harm. For example, we asked volunteers to mark anything missing or they were unsure about so that it would be easier to spot and solve the issues.
Only when those factors had been taken into account did we start to get technical and utilize our expertise and experience in subtitling. Some clauses in the guidelines were adopted from Netflix’s style guides, which said each line should not exceed 42 characters. We changed the number to about 70 because we were not going to make English subtitles but use the transcript for translations into other languages.
The last thing we thought of when building the infrastructure was a Google account dedicated to the project to minimize use of personal accounts of either me or Celia, which would ensure better information management and smoother successions.
After we finished building this infrastructure, we launched the project as a Globe project to invite MIIS TLM colleagues to be our first volunteers. While our goal was to open it up to real crowd, starting with colleagues offered higher accountability and smoother communications.
Jacqueline Xu 