This semester, one of the three classes I'm taking is a programming class meant for scientists, rather than for computer science majors. (I'm only taking three classes! What is this?)
This is pretty cool, because it means I'm with other graduate students, and also everyone realizes that they need more programming skills than they have.
Our final projects are all supposed to relate to our research, if at all possible. As a masters student in math, I don't have official research with my department currently, but I do have research interests through the disability side of things. I'm interested in cognitive interpretation, like what Neurodivergent K is describing here, and in treating disability-related communication barriers as translation problems. (Sign interpreters totally already do this, so this is not a new idea, not on its own.) In full generality, this would be a huge project and nowhere near appropriate for a semester, but by taking a smaller project, like applying one already existing translation-related technology to communication barriers similar to those I face, I can hopefully get somewhere this semester.
I'll be looking at Computer-assisted Translation (CAT), which already has software to support it. The idea behind computer assisted translation is that sometimes you need to translate a sentence, phrase, or communication similar to one you've needed to translate before. The software that is assisting the translation finds similar phrases that have been translated before, finds what their translations were, and suggests those translations.
Because the translation or interpretation that a cognitive interpreter is doing is between a "standard" dialect and the (non-standardized) communication patterns of a disabled person, we can't really draw on most already existing translation histories. However, relevant translation histories could be created. It may well be possible to give the software some translation history based on the interpretations of a human interpreter, and because certain communication traits are more common among people with particular conditions, it may be possible to create "starter" or "default" translation histories that come with the software. I think that including translations or explanations of common internet language uses would make sense, as one example -- many Autistic people, including myself, are echolalic, an plenty of us tend towards code mixing rather than code switching, which means we may well use internet language in contexts where it won't be understood.
I could also try to bring in comparable corpora, which is something I've been reading about in Comparable Corpora and Computer-assisted Translation. The idea behind comparable corpora is that when people are writing or speaking on a given topic in their own language, this reads and sounds different than translations from other languages do. By using texts on similar topics which were originally in different languages, we can have translations with less "translationese" in them. I think bringing in comparable corpora is unlikely this semester, but that I do want to incorporate it eventually.
Interfaces could also be modified to better match with use by people whose disabilities affect communication and who might not be experienced at translating documents between "standard" languages. (I have some translation experience for personal use, so I do know that this sort of experience is not mutually exclusive with communication disabilities.)
I found one computer assisted translation program, Virtaal, which is free and open source, written in Python, the language we're learning in the programming class. Because Virtaal is a many-file program with a graphical user interface, there's a lot going on in that program that I don't understand yet. My project is going to be based very heavily in learning to understand the code for this already existing program, which I'll then try to make some modifications or additions to over the course of the semester. I'd like to do even more with it later -- I want this software to exist already, and I want it on my laptop, fully functional, now. That's why I'm trying to build it!
This is pretty cool, because it means I'm with other graduate students, and also everyone realizes that they need more programming skills than they have.
Our final projects are all supposed to relate to our research, if at all possible. As a masters student in math, I don't have official research with my department currently, but I do have research interests through the disability side of things. I'm interested in cognitive interpretation, like what Neurodivergent K is describing here, and in treating disability-related communication barriers as translation problems. (Sign interpreters totally already do this, so this is not a new idea, not on its own.) In full generality, this would be a huge project and nowhere near appropriate for a semester, but by taking a smaller project, like applying one already existing translation-related technology to communication barriers similar to those I face, I can hopefully get somewhere this semester.
I'll be looking at Computer-assisted Translation (CAT), which already has software to support it. The idea behind computer assisted translation is that sometimes you need to translate a sentence, phrase, or communication similar to one you've needed to translate before. The software that is assisting the translation finds similar phrases that have been translated before, finds what their translations were, and suggests those translations.
Because the translation or interpretation that a cognitive interpreter is doing is between a "standard" dialect and the (non-standardized) communication patterns of a disabled person, we can't really draw on most already existing translation histories. However, relevant translation histories could be created. It may well be possible to give the software some translation history based on the interpretations of a human interpreter, and because certain communication traits are more common among people with particular conditions, it may be possible to create "starter" or "default" translation histories that come with the software. I think that including translations or explanations of common internet language uses would make sense, as one example -- many Autistic people, including myself, are echolalic, an plenty of us tend towards code mixing rather than code switching, which means we may well use internet language in contexts where it won't be understood.
I could also try to bring in comparable corpora, which is something I've been reading about in Comparable Corpora and Computer-assisted Translation. The idea behind comparable corpora is that when people are writing or speaking on a given topic in their own language, this reads and sounds different than translations from other languages do. By using texts on similar topics which were originally in different languages, we can have translations with less "translationese" in them. I think bringing in comparable corpora is unlikely this semester, but that I do want to incorporate it eventually.
Interfaces could also be modified to better match with use by people whose disabilities affect communication and who might not be experienced at translating documents between "standard" languages. (I have some translation experience for personal use, so I do know that this sort of experience is not mutually exclusive with communication disabilities.)
I found one computer assisted translation program, Virtaal, which is free and open source, written in Python, the language we're learning in the programming class. Because Virtaal is a many-file program with a graphical user interface, there's a lot going on in that program that I don't understand yet. My project is going to be based very heavily in learning to understand the code for this already existing program, which I'll then try to make some modifications or additions to over the course of the semester. I'd like to do even more with it later -- I want this software to exist already, and I want it on my laptop, fully functional, now. That's why I'm trying to build it!
Autistic computer scientist here with some experience in linguistic matters... this sounds like a very nice project. Good luck to you!
ReplyDelete