Writing sample in linguistics for a master's in comp ling?

[gulabjamun]

I am a recent engineering graduate in India, and I've been thinking about applying to computational linguistics master's programmes in the US. Some of these universities ask for a sample of academic work/writing sample. My problem is that I haven't done much writing, except for a long essay related to sociology/ancient Indian literature in an elective class. I decided to write something on my own in linguistics for applications, because I haven't taken any classes in linguistics at all (except for a couple of MOOCs) and this is the only way I can show them that I know (at least some) stuff.

I wrote about the unique features of Indian languages, in terms of their usage, orthography, grammar and pragmatics, that present difficulties in developing rules for machine translation (MT) involving them. I wrote about three languages that I know, and a couple of techniques used in some papers on the topic. Now I don't really know if this is acceptable, because it doesn't have a thesis statement or any particular argument, except that MT is not all that easy (yeah I know give me a medal). I mean it's like a glorified wiki article - is this in any way acceptable?

[humidz]

Your sample writing should provide evidence not of your general knowledge (that's what the transcript & the CV are for), but of your ability to do research. It sounds like you have a lengthy list of problems for machine translations. If you feel like you already have some ability to do so, my advice would be to look for whatever seems to be the most interesting one, to describe that problem in detail and to try developing a solution.

Generally, "languages that I know" does not make for a well justified sample, so unless there is sth specific to the language comparison that is interesting in itself, I'd recommend focusing on one language.

[gulabjamun]

I think what I've done is talk about grammatical 'divergence patterns in machine translation', along with the challenges faced in the machine translation of Indian languages with regard to Indic scripts, availability of language data and cultural differences. There's a lot of papers on divergence patterns in language pairs on the internet, and some of them talk about techniques to solve these problems. I know two Dravidian languages and one Indo-Aryan language so I'm trying to compare them with regard to their language families as well. I tried to go for more of an overview because I don't have that much to say about one specific feature; I can't do any research and I don't know that much. Also the universities don't really ask for anything specific like research in linguistics or anything, it's like a 'writing sample that shows your scholarly ability/writing proficiency' or any sample of academic work, since it's for master's programmes. That's why I'm kind of unsure as to what they're looking for/how much I have to do. But I suppose I could water the writing down and talk about solutions, like you said.

Edited June 30, 2016 by gulabjamun

[gulabjamun]

And thank you, humidz!