MonoTrans: Collaborative Language Translation

The Human-Computer Interaction Lab at the University of Maryland has developed MonoTrans, which uses a combination of collective human judgements and human-machine partnership for language translation. According to the research team's website, their goal is a "framework for translation that exploits imperfect technology and limited human abilities in tandem to achieve capabilities neither can achieve alone." One interesting aspect the project is that the human participants need only know one language.

I participated in a translation project using MonoTrans at the International Children's Digital Library via a link provided by Ben Bederson.  Things were set up so I could work on a series of very short tasks at my complete discretion.  There were two kinds of tasks.   In one case, I was shown a sentence from a children's book followed by a proposed paraphrase.  My job was to edit the paraphrase as needed so that it would convey the same meaning as the original sentence.  In the other case, an initial sentence was followed by three potential paraphrases, and my job was to pick the best one.  Each task took only a few seconds, and I could complete as many or as few tasks as I wished.

My experience in this project was minimally collaborative.  I had no awareness of other people doing work similar or complementary to my efforts.  However, I did assume that other such work had or would take place.  That assumption was what made my experience collaborative.

My participation was driven by three goals:

1. I wanted to learn more about the MonoTrans project.
2. I wanted the MonoTrans research effort to be fruitful.
3. I wanted a book that children might enjoy reading to be available in multiple languages.

My first goal was not directly collaborative; I did not believe that anyone else who might participate in the translation project was doing so in order to help satisfy my individual curiosity.  My other two goals were directly collaborative; I knew that my efforts alone would not be sufficient to achieve these goals, but I assumed there would be other participants who shared the same goals.

Collaborative efforts often involve individual goals as well as shared goals, and the pursuit of  individual goals can have a positive, neutral, or negative effect on the success of the collaboration.  In the case of the translation project, for example, knowing more about the MonoTrans project might allow me to do a better job on translation tasks.  On the other hand, spending time learning about the MonoTrans project might take away from the time I would spend on translation tasks.  Achieving positive synergy between individual goals and shared goals can be a stepping stone to the success of collaborative work.

In my earlier post on Collaboration and Mutual Knowledge, I suggested that work is not truly collaborative unless the participants know (or believe) it is collaborative.  In the case of my participation in the MonoTrans project, such knowledge was key.  While I might have completed one or two short tasks just to satisfy idle curiosity, I would not have spent several hours on such tasks without the belief that others were committed to shared goals and working towards them.

Those involved in the management of collaborative projects would benefit from effective tools and techniques for  (1) establishing knowledge of shared goals and efforts towards them, and (2) achieving positive synergy between individual goals and shared goals.