I concur. Furthermore, Google is approaching machine translation from the product manager's perspective; instead, they should be trying to use applied linguistics with all the rule sets it entails. This is evident in their so-called "Chinese" translation, which may be close to the worst case scenario, as serial verb constructions and idioms are simply ignored.
A better example of Chinese machine translation would be a site like Adsotrans, where the translation is constantly improving. Cheers, Eric On Sun, Oct 19, 2008 at 8:17 AM, LUT polytech <[EMAIL PROTECTED]> wrote: > > I think the biggest problem of the translator services is that the > server gives always the same translation to a word regardless of for > example the homonyms of the word or for example in finnish we have > different kind of verbs in every personal pronoun (subject) and for > every tenses. Thats why for example this finnish translation is almost > totally gibberish. Only a very few translated sentences can be > understood without extra efforts. > > I'm sure that finnish is not even the worst case. In any language > translations between languages that have different kind of homonyms > cause this kind of major errors. > > I'd say this would be realtively easy to rehabilitate. It is very > important to use the right homonym of the word, also to identify the > predicate (personal) and tenses of any sentence. There sure is more > things that should be perceived in order to make an understandable > translation for a sentence. But the major collapses in finnish are due > to homonyms, predicates and tenses! > > I don't know what's the google's way to index the dictions but I'd see > this kind of indexing as a smart one. > > for every diction of a language there would be metainformation of: > > 1) all the homonyms for the diction > 2) all the synonyms for the diction > 3) the formulas for any tenses > 4) the formulas for any personals (predicates) > > > ITEM 1 > In order to be able to index any the different homonyms of a diction > there need to be a database of "categories of the talk" with headers- > titles-subtitles etc., for example... > 1. analysis > 1.1 technical > 1.1.1 science > 1.1.1.1 chemistry > 1.1.1.1.1 biochemistry > ...so that there is homonym for any category. The format for indexing > would be like > "hom01 EXCEPT; use hom02 for 1. AND 5.4; use hom03 for FOR 2. AND 3 > AND 4 AND 5" > Now: every internet site / web page need to be categoried. That can be > happened by the first user of a translated page or by Google's > automatic category_keyword_recognition. > > ITEM2 > In order to be able to index any the different synonyms of a diction > there need to be a database of "themes of the talk" for example... > 1. official > 2. playful > 3. angry > 4. chattering > 5. artistic > ...so that the translation would sound natural as the words are of the > same "style". > > ITEM3 > The correct tense of predicate can be figured out from the > exactsentence/earlier sentences. > > ITEM4 > The correct (personal) predicate of a sentence can be found from the > exact/latest sentence with subject. > > To get the most correct/natural dictions for translations Google > should purchase the indexes for all the dictions of a language from an > outer organisation. Such as wikionary community. > > Thanks for reading. > > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Google Translate" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/google-translate?hl=en -~----------~----~----~----~------~----~------~--~---
