Localizing/Translating website via neural translation

Hello everybody,

I hope that you are all safe and healthy.

Today I would love to hear your thoughts about a subject that has been bothering me for a long time now.

I am running a website with tons of content in English (hundred thousands of words actually) and I definitely want to translate/localize it, for multiple obvious reasons. With such crazy amount of content though, hiring an agency is not really an option (no, it definitely is not). A few months ago I stumbled across neural translation stuff and decided to go with weglot.

While the automatically generated translations are definitely far from great, they are significantly better than the usual auto translations. Still, proof reading and fixing hundred thousands of words is going to take forever (if we only talk about one added language) and that, while the tool costs money anyways.

My question and maybe some of you have experience there: Can those (neural) automatic translation even hurt my site instead of actually helping it? I mean, I want to localize in order to take advantage of my (great) English content by also getting visitors from Google who search in their native languages. And then, I am starting to get worried about other sites copying my content by translating it, which apparently, has already happened at parts.

Eventually though, if the "not great" automatically generated translated content would even hurt me, I would rather deactivate it all.

I appreciate your help and thank you very much in advance.

Quote:

---

Originally Posted by efialtis

Hello everybody,

I hope that you are all safe and healthy.

Today I would love to hear your thoughts about a subject that has been bothering me for a long time now.

I am running a website with tons of content in English (hundred thousands of words actually) and I definitely want to translate/localize it, for multiple obvious reasons. With such crazy amount of content though, hiring an agency is not really an option (no, it definitely is not). A few months ago I stumbled across neural translation stuff and decided to go with weglot.

While the automatically generated translations are definitely far from great, they are significantly better than the usual auto translations. Still, proof reading and fixing hundred thousands of words is going to take forever (if we only talk about one added language) and that, while the tool costs money anyways.

My question and maybe some of you have experience there: Can those (neural) automatic translation even hurt my site instead of actually helping it? I mean, I want to localize in order to take advantage of my (great) English content by also getting visitors from Google who search in their native languages. And then, I am starting to get worried about other sites copying my content by translating it, which apparently, has already happened at parts.

Eventually though, if the "not great" automatically generated translated content would even hurt me, I would rather deactivate it all.

I appreciate your help and thank you very much in advance.

---

Hi,

I've personally tried a few translation API's and different kind of translation tools, I came to a conclusion:

• none of the tools I've used are 100% good (or 100% perfect). They have flaws in translation.

Maybe I'm a bit of a 'perfectionist'.

Example:

If you are a native English speaker, and want to translate from English to Hungarian. You'll need to double-check the translated content with a professional Hungarian translator - after the automatically generated translated software/tool finished the "translation job".

Tested tools:

• DeepL.com (best one I've used)
• Catapult
• Google Translate

In my opinion there are a few years ahead (which we need to wait) until AI translation becomes more advanced.

From my testing... with 1 particular website.

https://www.gpwa.org/forum/images/im...om/kzCjVcV.png

Thank you very much for sharing your personal experience, I appreciate that.

Regarding your screenshot: That is a very significant difference, which is awesome. Unfortunately, I will have no chance of proof reading and revising each and every bit of our content since the amount is so huge.

Thanks!

DeepL have Worpress Plagin?

WOW, really impressive!!!
what is the total investment/cost that made you get to that point?

Quote:

---

Originally Posted by gambling_star

Thanks!
DeepL have Worpress Plagin?

---

You missed the point that he made
none of the tools I've used are 100% good (or 100% perfect). They have flaws in translation.

Rick
Universal4

Quote:

---

Originally Posted by Jacob-itai-s

WOW, really impressive!!!
what is the total investment/cost that made you get to that point?

---

Investment:

- DeepL API = about 152.57 EUR (please check screenshots below).
- Content was free, because one of my friends helped me with the content.

Notes:

- this particular website I'm "talking" about is in a different niche. It is not a casino website.
- it is EMD (exact match domain) website.
https://www.gpwa.org/forum/images/im...om/oRdeWra.png
https://www.gpwa.org/forum/images/im...om/o8ZcqNh.png

https://www.gpwa.org/forum/images/im...om/LorCTmu.png

I see that the topic founder's site has introduced new automatically translated content.
I wanted to ask what are the effects in retrospect?

DeepL is amazing and its API. The translation is so easy.
I would never choose to translate a page through google translator. However, DeepL translates X times more accurately and is amazing.

However, I'll be honest that I don't use it, but sometimes I wonder if it's not a cool option, especially since the technology has advanced a lot. Despite the fact that I do not use it and my site is not multilingual I am curious about this issue.

That's why I'm refreshing the topic because I'm curious to know how the use of DeepL has developed in your company since I see that the site is multilingual currently.

Quote:

---

Originally Posted by dannyx

I see that the topic founder's site has introduced new automatically translated content.
I wanted to ask what are the effects in retrospect?

DeepL is amazing and its API. The translation is so easy.
I would never choose to translate a page through google translator. However, DeepL translates X times more accurately and is amazing.

However, I'll be honest that I don't use it, but sometimes I wonder if it's not a cool option, especially since the technology has advanced a lot. Despite the fact that I do not use it and my site is not multilingual I am curious about this issue.

That's why I'm refreshing the topic because I'm curious to know how the use of DeepL has developed in your company since I see that the site is multilingual currently.

---

Yes, we have launched our German and Spanish versions a while ago and the translations are indeed based on Deepl. FWIW, Deepl is doing a pretty good job partly but it is not like it is great all the time, on the contrary. We have been editing our Spanish & German content for months and are not even close to getting it all done. Means you will still have a lot of costs.

As a conclusion though, it definitely helps in cases where you have large sites that simply cannot be translated entirely all at once.

Nb. Our traffic has also been increasing even though it takes time for Google to pick it all up and rank one accordingly.

DeepL reportedly does a great job with German.
Deep is better than many human translators. This is the opinion I have sometimes encountered on translator forums.

However, a lot depends on the language pair and I wouldn't use it for Spanish, I'm surprised myself but even though it's popular in my opinion Deep doesn't do well with EN-ES, I'm not sure if I'm right.

It's cool that you have traffic (I checked with Ubersuggeest and in German you already have practically as much traffic as in English) Are you planning to add new languages? However, do these editions consume so much work that these are enough?

Google translate, as of my knowledge is OK for English to Spanish.

Also keep an eye out for slang words when working with Deepl or any other translator. Slang words that have been translated into another language usually have no meaning. For example, translating "piece of ****" by a word into any other language would not make much sense. The same another way around from other languages to eng. One or two slang words can easily change meaning of material you're translating.

DeepL API up to 500,000 words per month is free. So anyone can test it.
I know it works well with Translate Press and Weglot plugins, of which TranslatePress is better.
They recently added Indonesian language.

Interestingly, when translating into some languages, you can lose traffic.

For example, by translating into Russian or French. Having a site in English you can have inputs from these countries and by translating the site the institutions of these countries will block our site. So by translating you will drop traffic from those countries, well unless some people use VPN.