|
 |
Forum Member with most language families | ||
Tags: Polyglot | ||
Share with:
| ||
 Language Learning Forum : Polyglots |  | |
| Sprachprofi Nonaglot Senior Member Germany learnlangs.comRegistered users can see my Skype Name Joined 6266 days ago 2608 posts - 4866 votes  Speaks: German*, English, French, Esperanto, Greek, Mandarin, Latin, Dutch, Italian Studies: Spanish, Arabic (Written), Swahili, Indonesian, Japanese, Modern Hebrew, Portuguese | Message 25 of 41 01 October 2010 at 7:27pm | IP Logged | |||||
|
The idea of statistical machine translation is mainly proposed by programmers lacking a
background in linguistics, yet its superiority hasn't been proven. Let's say that if Google with its access to unheard-of amounts of texts and funding can't make it work, then we don't stand a chance. Instead, the UNIKOM system is based on a hybrid approach and more recent research results. The statistical approach hasn't seen serious innovation since 1995, that's eons in a science as young as computational linguistics. As for earlier attempts to utilize Esperanto, these were hobbyist endeavors for the most part, suffering from lack of expertise and/or lack of funding. They never advanced far enough to be able to exploit Esperanto's unique advantages. 3 persons have voted this message useful | ||||||
| Raчraч Ŋuɲa Triglot Senior Member New Zealand Joined 5614 days ago 154 posts - 233 votes Speaks: Bikol languages*, Tagalog, EnglishC1 Studies: Spanish, Russian, Japanese | Message 26 of 41 03 October 2010 at 12:07am | IP Logged | |||||
I totally agree with you on this about statistical MT, I would never expect statistical MT to approach human like translation, no matter how fast it is, unless mixed with some grammar/syntax rules. I even doubt hybrids. The old adage still apply I think: "Garbage in garbage out". What is needed is a fresh approach that would really work. The problem is, this is big business, so no way we will see the innards of these current approaches discussed publicly, and no way for theoretically unsound models to be ridden of their bugs. It's like opening Google or Microsoft search technology to competitors. I am also a hobbyist linguist(?), but I doubt if Esperanto can be a viable intermediate language; its probably the wrong "sieve" or "net" to catch meaning distinctions for a lot of reasons, that's why I prefer a new well-thought out conlang that approaches a more "universal grammar" by considering the full continuum of language typologies. But do give Esperanto a try and let us know if it worked. If it does not work, you know more now how to enhance your model and give it another try. I wish you success. 1 person has voted this message useful | ||||||
| Journeyer Triglot Senior Member United States tristan85.blogspot.c Joined 6664 days ago 946 posts - 1110 votes Speaks: English*, Spanish, German Studies: Sign Language | Message 27 of 41 03 October 2010 at 9:27pm | IP Logged | |||||
|
I have a fellow Peace Corps friend who speaks languages from three families.
Indo-European Polish (native) English (practically native, learned at a young age) Spanish Japanese (Japanese-Ryukyuan) Mandinka (Mande, learned in The Gambia) However, when I have thought of language families, I've always thought of the subfamilies. Rarely I find people who seem to speak outside of two or three language subfamilies. 1 person has voted this message useful | ||||||
| clumsy Octoglot Senior Member Poland lang-8.com/6715Registered users can see my Skype Name Joined 4974 days ago 1116 posts - 1367 votes Speaks: Polish*, English, Japanese, Korean, French, Mandarin, Italian, Vietnamese Studies: Spanish, Arabic (Written), Swedish Studies: Danish, Dari, Kirundi | Message 28 of 41 06 October 2010 at 12:05pm | IP Logged | |||||
|
me: Polish, English = Indo-European
Korean = Altaic/isolated Japanese = Japonic(?) or Altaic, or maybe even Austronesian. Chinese - Sino-Tibetan They (Asian ones) share a lot of vocabulary though. 1 person has voted this message useful | ||||||
| Doitsujin Diglot Senior Member Germany Joined 5116 days ago 1256 posts - 2363 votes Speaks: German*, English | Message 29 of 41 06 October 2010 at 1:22pm | IP Logged | |||||
Actually, Google is not very secretive at all about their technology. If you want to get a general idea about how their technology works, watch their presentation. There's nothing secretive about the technology either, because many companies use customized versions of the free Moses MT toolkit What you consider bugs are inherent weaknesses of the statistical MT approach. I.e. statistical MT will never be 100% perfect, but neither will be a rule based MT system. 1 person has voted this message useful | ||||||
 Iversen Super Polyglot Moderator Denmark berejst.dk Joined 6499 days ago 9078 posts - 16473 votes Speaks: Danish*, French, English, German, Italian, Spanish, Portuguese, Dutch, Swedish, Esperanto, Romanian, Catalan Studies: Afrikaans, Greek, Norwegian, Russian, Serbian, Icelandic, Latin, Irish, Lowland Scots, Indonesian, Polish, Croatian Personal Language Map | Message 30 of 41 06 October 2010 at 2:21pm | IP Logged | |||||
|
I'm tempted to cut this thread up in two, giving the discussion about Google translate it own thread.
1 person has voted this message useful | ||||||
| Cainntear Pentaglot Senior Member Scotland linguafrankly.blogsp Joined 5807 days ago 4399 posts - 7687 votes Speaks: Lowland Scots, English*, French, Spanish, Scottish Gaelic Studies: Catalan, Italian, German, Irish, Welsh | Message 31 of 41 11 October 2010 at 10:06am | IP Logged | |||||
++disagree Esperanto has no "unique advantages". In MT, you can create your own interlingua, and you can recreate all the advantages of Esperanto without exposing itself to any of its weaknesses. And your own demonstration of Esperanto's "advantages" undermined itself, because you can and did make the English words "lightning-man" and "rain-place". Esperanto is designed for human use so is too small and limiting to use as a machine translation interlingua. I know you have to put up with a lot of Esperanto-bashing here, but right now it looks like you're being somewhat fanatical. This is not what Esperanto is for. PS. Iversen, I think that's a good idea. I personally find the machine translation far more interesting than the language families thing, and it could run for a while. Edited by Cainntear on 11 October 2010 at 10:08am 1 person has voted this message useful | ||||||
| Volte Tetraglot Senior Member Switzerland Joined 6235 days ago 4474 posts - 6726 votes Speaks: English*, Esperanto, German, Italian Studies: French, Finnish, Mandarin, Japanese | Message 32 of 41 11 October 2010 at 10:12pm | IP Logged | |||||
What you're missing is that those are normal Esperanto words that you can feasibly find in existing dictionaries, and that an affix system (which, again, is used in normal dictionaries) can be quite helpful. Given the current size and quality of Esperanto dictionaries, this is a large boon for small teams. It's only irrelevant if you want to invest tens of thousands of hours, or perhaps more, getting to the same point with augmented-English or new-interlanguage dictionaries. Esperanto's "unique advantage" isn't that it's perfect for this - it's far from it. It's that its regularity and the size of dictionaries already available for it are an unmatched combination. I'm not convinced that using Esperanto as an interlanguage is a good approach, but it does have significant merits over any of the proposals I've seen people come up with in response to it. 1 person has voted this message useful |
|  | 
Printable version
|
| Copyright 2024 FX Micheloud - All rights reserved No part of this website may be copied by any means without my written authorization. |
