What is segmenting? Learning the technical skills of online translation at Viki (Part 1).

As interesting as an academic post on translation programs would be (though that’s coming, no worries), I thought value could be added to this survey by including the voice of an individual who does segmenting work and ask them about their experience. Below is an interview with a Viki volunteer who runs a segmenter training program. For a great introduction to Viki, please refer to Tessa Dwyer’s earlier post on this survey.


Viki specializes in audiovisual translation (primarily Asian dramas) and volunteers translate approximately ten minutes of video at a time. However, before volunteers can enter their translations, the text-boxes for the translations are carefully edited by volunteers who time each one by hand. The volunteers who do this technical work are called segmenters. Segmented episodes are then sent to bilingual volunteers who can translate the piece into over 100 languages for the site.


This work is essential to Viki’s success, but may go unnoticed by the general viewing public. Volunteers train and network for positions on the most most popular “live” channels. Many volunteers turn to volunteer-created segmenting training programs to learn these technical skills. These training programs help volunteers gain the technical and and cultural knowledge to edit videos to match dialogue with .01 seconds of accuracy. After graduating, volunteers receive badges, which they can display on their profile pages as certifications of their craft.


This interview ran longer than I planned and will publish over two days for this survey. The first part of the interview will be about the interviewee’s own experiences learning to segment and the second on the choices she’s made establishing her own training program and comments on segmenters’ working relationship with Viki.


What is Segmenting?

Segmenting is the first step, laying out the foundation for the entire video to be subtitled . If a foundation is really poor, some subtitles may run three lines, causing too much of the video to be covered by text,  or some may have one word and quickly flash off the screen . We need to create a good foundation that is comfortable for the audience to enjoy what they are watching. That is how we can actually share knowledge. If it’s not enjoyable or unpleasant it's going to drive people away.


Where did you find out about Viki?

I heard a song on YouTube that I really liked by 831 Band called “The Best Ending.” I decided to see what it was all about and it was actually the theme song of the Taiwanese-dubbed version of Rooftop Prince from GTV. So I checked out what Rooftop Prince was all about. The name is kind of cute. I saw a clip of it, the one where Park Yoochun and all his followers were eating omurice like crazy on the floor. I thought that was interesting and I wanted to see the entire drama. I Googled it and Viki was the only place I could find it. I decided to watch it and from there I started watching more and more shows and that got me interested in more of the Asian dramas.  


When did you start getting interested in segmenting? How did you find out about it?  

When I started watching I realized that you could volunteer on the site and become a team member. At first I was only a subtitler for Chinese dramas. Then  I realized that some dramas were not segmented, so I could not subtitle those dramas. So I felt the need to segment. I used an  old segment timer that Viki had. We could actually segment and type in our subtitles right then. But then Viki got a new Beta timer. It was half a page of wave and the video was really small. I wondered how I was ever going to segment in the future. That scared me away for a few months. Then they said it got implemented and I decided to open the timer and I noticed that the wave was much smaller and the video much bigger so I could actually see what was going on to segment. I still had a lot of flaws. I felt like that was inadequate and I needed to take a class.


You started with Seg101 (one of several training programs that volunteers have built). Could you tell me a little bit about how you were trained in segmenting?

After I saw the new timer I decided that it was time to get some training if I wanted to be accepted on channels in the future. So I filled out Seg101’s application form and I was accepted into their program. I had the choice to go through the traditional program or the express program, but they said they had no space in Express. With the new timer, many of their teachers went away, but I was ultimately accepted [in Express]. I went through the videos. The first practice video was just 7 segments and I got through it. When I reached the final videos I realized that my segmenting was not that great. They actually passed me by express standards, but not the traditional standards. The longer training expected a higher level of perfectionism, 95% to graduate. So, it wasn't perfect in their eyes, but it was good enough that I could practice more after graduating. I had potential.


I found you through a forum focused on helping segmenters to improve their skills. It’s been active for the past 2 years. What prompted you to start that forum?   

In my training in Seg101, they used to have a discussion forum and also clear directions on how to segment on the old timer. But when I did my program I didn't have any guidance. So, people could segment however they like, but if people don't segment well and they are on a team and ask for their advice [Channel Managers] would give it to them. [But Channel Managers] don't really want to tell people the truth unless they ask for it. I realized that it was time for me to write something for everyone.


So I decided to just write up something so people could have a head start so when they feel as though they couldn't do it on their own. They could decide for themselves whether to seek further training or whether my material was helpful so that they could work on fan channels. Fan channels don't require any training. For me, training is highly recommended, but it is up to the volunteer.


What makes for a great segmenter?

Definitely whether they are able to work with other people. If parts are assigned to only certain segmenters then you don't have the equal rights to segment. We should all have equality if we could make it there. When it's uploaded then it's based on first-come-first-serve. I just realized that being able to be a game player, being able to do your job well, being able to get along well with others are all volunteer job qualities. Also whether you're able to, basically whether you have the passion to segment and you're willing to accept constructive feedback to improve along the way.


Over time we would hope that you would have the speed so we could get [an episode] segmented in one hour . That's an average of 30 minutes per person on their part, assuming that they’ve entered their parts at different times. Then within an hour we could get it to the subbers. Many of the subbers are actually from New York because a lot of Korean immigrants live on the East Coast. It's really important to give them the opportunity during lunch hour to help us sub it. Then by night time, around evening, it will be great for editors to do it and then we have the entire job done in a day.

So, basically, a good segmenter must get along with people, cooperate, work independently, do their job. It’s teamwork for Viki. It’s not a hero thing. 


I realize my subject line may sound like a WoW reference, but I'm thinking more of how trades and crafts were once learned.  

The apprenticing / training, and earning one's way into a group via some repeated basic skills until one can prove ones chops / earn a badge, is reminiscent of the old way of controlling a market or a skill --  and an essentially communal / self-policed system at that.  

That said, the segmenting seems to be almost reminiscent of the industrial revolution with the shift away from a worker having control from start to finish of a project to a repeated single action in a factory.  

Looked at in labor terms this is an intriguing blend of pre-capital / early capital  models and is especially fascinating as we apply it to Viki's fan-based for-profit model. 


In my head I keep associating this kind of work with my own experiences with academia. But you're right, it is very much a pre-capital guild system (just like academia). 

In general, these models, often built on gifting of time and knowledge seem to work rather well in side of for-profit models. I think the big thing is that they mostly work in the same spaces, but with different currencies. 

Thanks for this fascinating insight into the inner workings of Viki and the crucial role played by segmenters! Can't wait to hear more.

This interview is really interesting for the way that it highlights the fairly chaotic, grassroots vibe that persists on Viki despite its professionalisation. The Seg101 training program, for instance, is something that volunteers have initiated and built. Also, this segmenter comments on the fact that directions, guidance and advice are only given if people seek it out and that otherwise people can segment however they like. 

I'm also interested in the way that geography, location and time zones play such a critical role in the fansubbing processes, despite the supposedly 'deterritorialised' nature of online spaces. Food for thought...

Thank you for giving us some background information in Viki, segmenting, and subtitling. It sounds like a great practice to know to get to know others within the community, to engage in new languages and media is as widely acknowledged by the general public. Dr. Dwyer, I was also interested in how different locations and time zones determined how the fansubbing was done and how they were all able to work together when needed. I can imagine there are those that are well versed within segmenting and fansubbing compared to others and I was thinking about how that worked.

Yes, its really interesting to hear about how these collaborative online communities actually work in practice. Mizuko Ito (2012) has written on this subject in relation to anime fansubbing and there has been quite a bit published on the working methods of Italian and Brazilian fansubbing groups also. 

Really interesting to see Viki from the perspective of a Segmenter. It had me thinking that this is such a unique and fascinating form of composing/writing, one truly worth the attention of rhetoric, composition, and communication scholarship. Again, so many angles to unpack with this work, and so few seem to know about it. 

Add new comment

Log in or register to add a comment.