Are Ambiguous Conjunctions Problematic for Machine Translation?
Maja Popović, Sheila Castilho · 2019
The translation of ambiguous words still poses challenges for machine translation.In this work, we carry out a systematic quantitative analysis regarding the ability of different machine translation systems to disambiguate the source language conjunctions "but" and "and".We evaluate specialised test sets focused on the translation of these two conjunctions.The test sets contain source languages that do not distinguish different variants of the given conjunction, whereas the target languages do.In total, we evaluate the conjunction "but" on 20 translation outputs, and the conjunction "and" on 10.All machine translation systems almost perfectly recognise one variant of the target conjunction, especially for the source conjunction "but".The other target variant, however, represents a challenge for machine translation systems, with accuracy varying from 50% to 95% for "but" and from 20% to 57% for "and".The major error for all systems is replacing the correct target variant with the opposite one.but 1 but 2 en We wanted to go to the beach, We didn't want to go to the hotel, but we went back to the hotel.but to the beach.de Wir wollten zum Strand gehen, aber Wir wollten nicht zum Hotel, wir sind zurück zum Hotel gegangen.sondern zum Strand.es Queríamos ir a la playa, No queríamos ir al hotel, pero hemos vuelto al hotel.sino a la playa.sr (hr) Hteli smo da idemo na plažu Nismo hteli da idemo u ali smo se vratili u hotel.hotel nego na plažu.en She will not come but she can call.She will not come but call.es No va a venir, pero puede llamar.No va a venir, sino a llamar.de Sie kommt nicht, aber sie kann anrufen.Sie kommt nicht sondern ruft an.sr (hr) Neće doći ali može da zove.Neće doći nego će zvati.and 1 and 2 en The walls and the door are white.The walls are white and the door is black.hr (sr) Zidovi i vrata su bijeli.Zidovi su bijeli a vrata su crna.en I studied for the whole day and I I studied for the whole day and I learned a lot.didn't learn anything.hr (sr) Učio sam cijeli dan i svašta naučio.Učio sam cijeli dan a ništa nisam naučio.en Years passed, and he came back.Years passed, and he still hadn't come back.hr (sr) Prošle su godine i on se vratio.Prošle su godine a on se još nije vratio.en Who is this and what is he doing here?And what is he doing here?hr (sr) Tko je to i što on radi ovdje?A što on radi ovdje?