Hi, Yes, of course! That has always seemed a bit unnatural to me. It's harder to decide on the right source language lemma before translating than doing it after translation.
The ambiguity in the source language is in most cases not present in the target language. After translation you have will have an indication if the translation "makes sense" or not, this could be quite useful when choosing between two different translations due to ambiguity in the source language. But Apertium doesn't work that way. You just bet on one of the possible source lemmas before translation. And yes, ambiguity regarding how to translate that one and only source lemma is a more limited task. Can Apertium be manipulated somehow to translate all possible source lemmas into translation hypotheses for the whole sentence (instead of choosing just one source lemma)? The lemmatisation seems to do half of the job: displaying all possible lemmas, separated by '|'. I would like to continue one step further and translate all possible variants (analyses) of a sentence. An example: Apertium sv-da has some trouble to translate the sentence: 'ta en blå kon' (=take a blue cone) to danish. 'kon' might be the indefinite form of 'kon' (= cone) or the definite form of 'ko' (= the cow). We have: (kon→ kon<n>/ko<n>) Translating the whole sentence would give us: tag en blå kegle / tag en blå koen (= take a blue cone / take a blue the cow) Wouldn't that be quite revealing in many cases? In this case e.g. a statistical language model could easily separate the wheat from the chaff. BTW Apertium sv-da bets on the second option. Yours, Per Tunedal On Thu, Mar 3, 2016, at 21:53, Kevin Brubeck Unhammer wrote: > Per Tunedal <[email protected]> čálii: > > > If the constraint-based lexical selection module is used for a pair, I > > cannot see why it couldn't be used. The rules are already in place. All > > you have to do is to translate the ambiguous sentences and let the > > module select the best translation. > > > > The tricky bit would be to use this information backwards to choose the > > right lemma in the original language. I'm not savvy enough to figure out > > how to do it. > > The right source language lemma is already selected by the time lexical > selection runs. Lexical selection is about selecting the right *target* > language lemma. > > -- > Kevin Brubeck Unhammer > > GPG: 0x766AC60C > ------------------------------------------------------------------------------ > Site24x7 APM Insight: Get Deep Visibility into Application Performance > APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month > Monitor end-to-end web transactions and take corrective actions now > Troubleshoot faster and improve end-user experience. Signup Now! > http://pubads.g.doubleclick.net/gampad/clk?id=272487151&iu=/4140 > _______________________________________________ > Apertium-stuff mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/apertium-stuff > Email had 1 attachment: > + signature.asc > 1k (application/pgp-signature) ------------------------------------------------------------------------------ Site24x7 APM Insight: Get Deep Visibility into Application Performance APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month Monitor end-to-end web transactions and take corrective actions now Troubleshoot faster and improve end-user experience. Signup Now! http://pubads.g.doubleclick.net/gampad/clk?id=272487151&iu=/4140 _______________________________________________ Apertium-stuff mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/apertium-stuff
