Skip to main content
18 events
when toggle format what by license comment
S Oct 31 at 6:38 history suggested Segorian CC BY-SA 4.0
Spelling, punctuation
Oct 8 at 18:17 answer added Sir Cornflakes timeline score: 0
Oct 8 at 9:19 review Suggested edits
S Oct 31 at 6:38
Jan 16, 2019 at 15:00 history tweeted twitter.com/StackLinguist/status/1085552359730753536
Jan 16, 2019 at 7:27 comment added Probably @UserKa That's for sure
Jan 16, 2019 at 7:23 comment added UserKa @Probably Maybe that's because google translate use direct translation between Czech and Slovak. But there are no official documents that indicate which pairs use direct translation.
Jan 16, 2019 at 7:18 comment added Probably As a Czech, I've often translated Slovak and I've never spot Google Translator would make an error.
Jan 15, 2019 at 11:15 history edited Adam Bittlingmayer
edited tags
Jan 15, 2019 at 8:46 comment added abarnert @UserKa My point is that much of the research, and much of the actual production work, involves English. If you specifically want/need to know about Turkic languages, ask about them of course. But if you're interested in the generic question "does translating related languages make translation easier?", you're almost certainly better off asking about pairs with English than pairs with Turkish, for practical reasons. (And that's even before I thought about the bridging issue in Adam Bittlingmayer's answer, which makes that much more true.)
Jan 15, 2019 at 8:32 comment added UserKa @abarnert I am not a linguist, and therefore, under one language family, I mean such a language pair like Ukrainian / Belarusian, Uzbek / Kazakh, Tatar / Bashkir, etc. I know that Hindi and German are in the same language family, but this not this way. as close as Uzbek / Kazakh. I also know that there are still some discussions about which language to include or not for a particular family.
Jan 15, 2019 at 8:15 comment added abarnert @AdamBittlingmayer Which languages in question? I already paralleled Turkish/Azerbaijani with English/German, two Germanic languages. I was under the impression that Kipchak is significantly farther away. Would Turkish/Chuvash be a better comparison for English/Hindi, or are even they closer than that? Should it be, say, English/Frisian/German :: Turkish/Azerbaijani/Chuvash instead of English/German/Hindi? (I don't think that changes my point at all, even if so, but still, interesting to know.)
Jan 15, 2019 at 8:01 comment added Adam Bittlingmayer @abarnert The languages in question are much more closely related than IE languages, a better comparison would be Germanic languages.
Jan 15, 2019 at 6:36 answer added Adam Bittlingmayer timeline score: 8
Jan 14, 2019 at 21:58 comment added abarnert Anyway, I know Google has published some results on the accuracy of GNMT and its predecessor. For example, this blog post says that (where the metric is apparently native speakers evaluating translations subjectively on a scale of 0 to 6) English/French scores 5.43, while English/Chinese scores only 4.3, although they don't give the source for that.
Jan 14, 2019 at 21:53 comment added abarnert Also, given how much machine translation work involves English, I suspect you'd be a lot better off looking for comparisons between, say, English/German, English/Hindi, and English/Turkish than, between say, Turkish/Azerbaijani, Turkish/Kipchak, and Turkish/something-not-Turkic-at-all. Would those English-based statistics be acceptable?
Jan 14, 2019 at 21:49 comment added abarnert I suspect "same family" as a binary distinction isn't very useful. German, Spanish, and Hindi are in the same family, but a lot more distantly related than, say, German, Swedish, and Dutch. (Not to mention that sometimes two closely-related languages can have major differences in grammar, while sometimes a broad family with dozens of languages can be grammatically very consistent.)
Jan 14, 2019 at 17:20 review First posts
Jan 15, 2019 at 9:15
Jan 14, 2019 at 17:18 history asked UserKa CC BY-SA 4.0