Computer: Translation: Partial
Disambiguation prompt   (+5, -2)  [vote for, against]
Prompt to disambiguate for a better translation

Have you ever gone to, for instance, a Web page in very poor English which you can tell was machine-translated from the Japanese?

I am guessing that when English is translated into a foreign language, similar problems arise.

My solution to many of the problems: When you input a text into a program for translation, the program should prompt you for features which exist in the target language, but might not exist in yours. For instance, a Japanese-to-English program should request that the Japanese user mark which nouns are to become plurals in English. (Many times, the machine translation does not take care of this, probably because it is a machine and cannot use the contextual clues which a human would.)
-- juuitchan3, Jul 19 2002

An alternative ? http://www.esperanto.org/
All verbs are regular ! Hooray ! [8th of 7, Jul 19 2002, last modified Oct 21 2004]

Universal Networking Language http://www.unlc.und...0Specifications.htm
A language for unambigeous representation of ideas. [flicken, Oct 21 2004]

The problem is that many web translators are used by native speakers of the target rather than the source language, wh don't have access to the contextual clues because they don't speak (read) the source; hence the need for a translator.

However, for authors wanting to cleanly translate their text into a target language, I see no reason that this can't be done; the problem then arising that they can't validate the output because they don't have the knowledge of the target language, otherwise they would probably do the translation themselves .....

But I will award a point anyway for a well-meaning and useful suggestion.
-- 8th of 7, Jul 19 2002


I've been pondering a way for improving over the status quo regarding this problem, but I just don't know if it makes enough sense to post here.
-- beauxeault, Jul 19 2002


beaux, since when has that mattered?
-- waugsqueke, Jul 19 2002


Or more generally you could mark up your text to indicate the function of ambiguous words, allowing it to be translated into a variety of languages. Problems arise not only from grammatical issues (like the use of "the" translating from languages that don't have definite or indefinite articles) but also from vocabulary, when you're using words that have multiple translations.

E.g. with markup in square brackets "Or you[=impersonal pronoun*] could mark up["mark up" synonym=label] your text to indicate the function of ambiguous words, allowing it[antecedent="text"] to be translated into a variety of languages. Problems arise not only from grammatical issues[=topics, not children] (like the use of "the" translating from languages that don't have definite or indefinite articles["article" used in context of grammar]) but also from vocabulary, when you're using words that have multiple translations."

*distinct from 2nd person singular or 2nd person plural, this might be translated into "on peut" in French.
-- pottedstu, Jul 19 2002


A colleague once remarked, "why don't we all just speak English?"
-- Grog, Sep 17 2002


This sounds like it might be a good idea, if the disambiguation prompts were designed to by usable by non-speakers of the target language. In many cases, the most natural wording in the original language would be ambiguous to a non-speaker (or even sometimes to a speaker!) but alternative wordings would be possible which--while awkward--would not have such ambiguity. Having the translator offer several versions of a sentence and asking which most closely matches the meaning of the original could be helpful.

Two more parting thoughs: (1) for translators which are to be used by non-speakers of the ORIGINAL language (e.g. those built into Google etc.) it may be useful to have the translator supply footnotes listing alternative meanings. (2) I don't know about other languages, but in English certain words may have several entirely different meanings and context is not always sufficient to resolve them. Consider:

In the morning, I read the paper on my porch and saw some firewood in the yard.

Depending upon following sentences, the sentence could be in the present tense, talking about chopping wood, or it could be in the past tense, talking about merely having seen the wood. Pronunciation would make it obvious, but in written form the sentence above could be read either way. How should a translator treat it?
-- supercat, Sep 18 2002


You might want to look into the Universal Networking Language (UML). It's an on-going project to produce a computerize notation that can be automatically translated into many different languages. That way, the document would only have to be translated once: into UML. After that, there is enough information in the notation to be able to translate that text into any language.
-- flicken, Apr 12 2003



random, halfbakery