User:Ivan Štambuk/Serbo-Croatian/Google translate

Google translate lists different options for Croatian and Serbian language. But does it provide different translations? I tested on the following excerpt of relatively simple English text, taken from Dorothy and the Wizard in Oz:


 * The space underneath the roof, where they stood, permitted them to see on all sides of the tall building, and they looked with much curiosity at the city spread out beneath them. Everything visible was made of wood, and the scene seemed stiff and extremely unnatural.


 * From their platform a stair descended into the house, and the children and the Wizard explored it after lighting a lantern to show them the way. Several stories of empty rooms rewarded their search, but nothing more; so after a time they came back to the platform again. Had there been any doors or windows in the lower rooms, or had not the boards of the house been so thick and stout, escape could have been easy; but to remain down below was like being in a cellar or the hold of a ship, and they did not like the darkness or the damp smell.

Here are the comparative results:

The expected differences are in the Ijekavian : Ekavian pairs (i.e. in one letter): gdje: : gde:, vidjeti: : videti:, djeca: : deca:.

There is one word wrong in the Serbian translation: Ijekavian bijeg: has its proper Ekavian pair beg:.

There is only one other difference, the Serbian translation has drveta where Croatian has drva, for the translation of "wood". The lemma form of this word is, and it's the same in both of the standards (being an inherited word). However it has the dual inflection: the usual thematic, and the consonatal t-stem, the latter one only with the sense "tree in growth" (drvo can mean both "wood", and "tree"). The preposition of (translated as od) binds the genitive case, which is drva and drveta respectively.

So it appears that the Google Translate engine uses the same algorithm to translate both of these "langauges". It also apparently uses the same lexical database, with items that are different among the standards tagged as "Serbian" and "Croatian" respectively. Sometimes these are not even properly tag (as in the case of bijeg).

Note also, since Serbian is standardized in both Ekavian and Ijekavian variety, the only difference among the translation would be in the abovementioned inflection of drvo.