mike gasser talk – 3/3/06
summary of cogsci thoughts on knowledge – knowledge inside, knowledge outside – sets up question of what it means to know and where we get our knowledge (good example of providing snapshot, addressing issues, and pointing out what Won’t be covered to cut off “well, what about…?” threads, as discussed in capstone class the other day. there are also a lot of little charts, well-sprinkled jokes, and a clear interest in the material.)
summary of issues of knowledge inequality
linguistic inequality on the internet
http://portal.unesco.org/ci/en/ev.php-URL_ID=20882&URL_DO=DO_TOPIC&URL_SECTION=201.html
summary: most of the internet is in english, and material in other languages is often limited in scope and potentially directed at tourists, not native speakers -> the knowledge divide is perpetuated and even furthered by the lack of linguistic representation online.
solutions
- teach everyone english
- find a way to make people write more things in other languages (would still not allow for everyone to read the same stuff)
- translate!!
//what barriers would still exist even if the linguistic barrier is reduced or eliminated?
//isn’t it still worthwhile to tackle?
translation can’t be centered at the word level because there are just too many possibilities in some languages (amharic is often given as an example) for english words. the mapping is nowhere near one-to-one. translations would be shitty.
translations should instead occur morphologically, at the sentence level.
how do we do that?
it’s quite hard.
we have to teach it.
developed a quecha -> spanish system using the sentence examples given in a dictionary. it doeesn’t work that well, but there are not that many sentences, and it shows promise.
we could use a wiki. maybe.
relationships between morphemes can’t be the whole enchilada, there are also factors relating to linguistic categories, but there wasn’t time to cover that in the presentation.
/*
i don’t understand all the descriptions of how it would work, but it’s an interesting topic
coming from the angle of someone who is more concerned with the work of developing technology that helps people express themselves on their own terms, not only linguistically, but culturally, artistically, technically, whatever, it feels a bit premature, or something. not really premature, i guess, but optimistic? something like – if only information was readable in lots of languages, people would read it and we would read what they write and everything would be hunky dory. i don’t buy that.
wait up… he’s talking about these questions, perhaps.
*/
“Qualms about machine translation”
-it doesn’t work (yet)
-there are encoding problems for character sets that could make it impossible to represent things in these languages even if they were translated (progress is being made here)
-it’s labor intensive
-requires large amounts of data
Basic rebuttal – ok, but it’s still a good thing to do, and we don’t have any proof that we Can’t do it, so we’re going to keep trying anyway
mike talks a bit about wikipedia, and about how their goals are very much in line with these goals, and they spend a lot of time working on translation questions. allying with them could progress things forward by leaps and bounds.
people raise some interesting questions about the inherent biases in the idea of wanting to put all this information online. the internet, and wikipedia, and these translation tools, are all products of the dominant culture that is also responsible for the preponderance of english online. should we presume that it’s a good thing to bring other languages into the mix? john paolillo brings up the point that, right now, languages serve as “firewalls” that particular memes have to break through before ideas are transferred between cultures, and asks whether that might not be a good thing. if everything were automatically translated into every languages, could potentiall harmful memes spread virally in ways that haven’t been possible before? and if we have languages that only have a very small written pool of knowledge, doesn’t that say something about their culture other than just that they are “ignorant”? will the value of the fact that only a few things would be natively recorded somehow diluted by giving them copies of newsweek in their native tongue? are we asserting the values of our culture in irrevocable ways even as we try to be inclusive? isn’t it condescending to act as if the greatest goal of other languages should be an interoperability with english?
the discussion of these issues is unresolved. time ends. people seem split on the matter. but mike seems to be sticking by the idea that it’s true that there are a lot of complexities to the issue, and it’s true that some things are worrisome, but it’s also worrisome that the people who Want access and who have things to say are hindered by these linguistic barriers, and this technology seems like a good that is worth further pursuit. i think it’s hard to argue with that. it is interesting work, and he clearly cares about it deeply. may we all be so blessed.