1. Forum
  2. >
  3. Topic: Welsh
  4. >
  5. Welsh Immersion Hub / Hwb Tro…


Welsh Immersion Hub / Hwb Trochi Cymraeg

If you upload an article to translate, please post it in this thread.

To upload an article, follow the following instructions: https://www.duolingo.com/comment/15558075

August 9, 2016



The last one about Oceania is actually cy > es (Welsh to Spanish).


"Cymreag ysgrifenedig" CY>>EN (136 words) on Wikipedia topic.


I think I cannot decide if this is grammar or literature, and also I see there is no English version on Wikipedia.



also "Wales" en > cy

You also may want to add that when you to an en > cy article, when editing the value, they need to do cy} instead of en}


Submitted: "Medieval Welsh literature" ENCY (281 words)


I think this article also needs some assistance in the Welsh version (on Wikipedia) and the immersion topic is Art and literature.


Submitted: "Brittonic languages" EN>>CY (561 words)


I always use articles about Linguistics. I wish there was a new topic to submit these articles, but it is on "Wikipedia" (Should be included in Literature?).


Does anybody know how to overcome the Duobot's insistence that its (incorrect) translation into bad English is right and your (correct) one is wrong? I have bumped into it a couple of times now...


Duobot is pretty nutty on some of its translations. Some languages work better than others and I would imagine that Welsh would be one of the worst for Duobot, with the Welsh grammar being so different from English. I am unaware of how to change anything for Duobot. How is Duobot's insistence on translations causing an issue? If you upload a translation, the Duobot suggested translation goes away, no matter how similar or dissimilar your translation is to Duobot.


A couple of times I have put in a translation, and Duo popped up to say that it is wrong (when it is right) and then refuses to accept my translation. If I then look at the Duobot attempt, it has one or more fairly major errors, which explains why mine looks so different.

I wondered it it was worth putting in a translation closer to Duobot's, getting it accepted and then going in and editing it to be correct...

Edit to add... Tried that, and Duo wouldn't accept its own translation either!


I have never seen that happen!


That could happen with some languages (perhaps not with the most used by the immersion community). I tried with Indonesian immersion (and I have also practiced a lot on the English-for-Indonesian course) and the sentences can be shortened using the correct grammar (for example, I am a man is translated basically with two expressions, Saya adalah seorang pria, the formal, and Aku adalah seorang pria, the informal or familiar, with saya and aku as pronouns and pria, simply one of the multiple words used for man or male; now the grammar is very different from English so adalah is not the verb to be, but the particle indicating equals to, then this particle is used to match the verb but it has a different meaning itself; secondly we have the classifier seorang, se- is a prefix indicating one and orang is a noun that means person or human and also the Indonesian/Malay root of the word orangutan, but seorang is not meaning a human being, or a person, because a person is seseorang, so the literal meaning of seorang is one individual for the class of human/person and seorang can match with the article a in this sentence; additionally, Indonesians can solve this using only two or three words, lacking firstly adalah, or adalah and seorang the most of the time, then Saya pria and Aku pria mean I am a man and the grammar is absolutely right but the immersion machine can reject this lacking when there is another lacking here or there in the whole sentence and it can think that is lacking many verbs or articles; maybe this problem is not occurring with a short sentence but with a long one including the phrase I am a man).

The example above is very simple, and surely it is included in the database because the sentence is already in the basics of the course. But I think there are many other words not included yet. And perhaps the immersion is absorbing new words through the translations marked as correct by users or when the words are the most used in the translations.

I would like to know if the Welsh sentence has been rejected because it is long or for some different glitch. What would be the reason then?

I think the immersion machine, alias DB, can reject a shortened construction or time phrases and long expressions not included in the database of a new language, but (I repeat) I THINK this is occurring mainly with long sentences because the machine can count a limited number of lacking words. So, when we pass the limit the correct translations are rejected. I have also worked with some immersion to Latin just to test it, and the dictionary hints shows many English words mixed with Latin words and the translation to Latin is rejected always if I use only Latin words. Then I think no one can complete all the translations of the text using correct sentences.

Now to solve this, we will need an extra work for every new course team and the developer team of Duo to clean and fix the database for new languages. And I am guessing if this would be perfectly possible soon. =))

Learn Welsh in just 5 minutes a day. For free.