• Replies 4 replies
  • Subscribers 125 subscribers
  • Views 1570 views
  • Users 0 members are here
Options
Related

Comparison Language Weaver and DeepL for EN-DE translations

Daniel Hug

I've been using RWS's Language Weaver and DeepL's stock engine in parallel for about a year now. Thought it would be interesting to share my experiences.

For complex sentences without defined terminology, DeepL is still performing better than LW. (I did not make use of DeepL's custom dictionaries as I use Studio 2021 and don't have the DeepL subscription level that would be required.)

For simple to average sentences with defined terminology and an established tone of voice, LW outperforms DeepL. The time savings are significant.

(The third MT I use is OpusCAT MT, which is free, adaptable and runs on my machine. It often outperforms DeepL with simple sentences or phrases where it had sufficient training data. It's free - no cost at all, so I hesitate to compare.)

The ability to train LW and use the dictionary for defined terminology is a winner. Marketing and related texts tend to avoid overly complex sentences or phrases. For this use case LW is ideal and a huge time-saver.

This is not the result of a controlled study, just my observations from a year of using both.

Daniel
emoji
Rate translation
×
Suggest better translation
×
Moderator UI
×
  • Get AI Suggestion
Parents

  • Thank you  for your observations.

    While you mention 2021, do take note that for 2022 SR2 we have a new Language Weaver Provider App  <- documentation included

    While the output remains based on Language Weaver NMT services, there are a a few new features that I believe will improve and better enable the use of dictionary/terminology given you say

    Unknown said:
    For simple to average sentences with defined terminology and an established tone of voice, LW outperforms DeepL. The time savings are significant.
    Unknown said:
    The ability to train LW and use the dictionary for defined terminology is a winner. Marketing and related texts tend to avoid overly complex sentences or phrases. For this use case LW is ideal and a huge time-saver.

    What comes to mind:

    • The new app now supports and ability to add terms to a selected dictionary that is managed with your Language Weaver Portal
    • We made subtle changes to how the managed dictionaries are being listed and how you enable what you need
    • We also support your instance of the app having a custom name, so you can differentiate NMT output based on custom parameters. Example NMT with our without dictionary

    In the near future we want to highlight those dictionary/term entries so as a user you can see how your terms are being applied. 
    Possibly perform a QA based on terms not used.

    If you can think of other missing features we should consider, please do reach out.

    Many Thanks

    Lyds

    Lydia Simplicio | RWS Group

    _______
    Design your own training!
    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

    emoji
Reply

  • Thank you  for your observations.

    While you mention 2021, do take note that for 2022 SR2 we have a new Language Weaver Provider App  <- documentation included

    While the output remains based on Language Weaver NMT services, there are a a few new features that I believe will improve and better enable the use of dictionary/terminology given you say

    Unknown said:
    For simple to average sentences with defined terminology and an established tone of voice, LW outperforms DeepL. The time savings are significant.
    Unknown said:
    The ability to train LW and use the dictionary for defined terminology is a winner. Marketing and related texts tend to avoid overly complex sentences or phrases. For this use case LW is ideal and a huge time-saver.

    What comes to mind:

    • The new app now supports and ability to add terms to a selected dictionary that is managed with your Language Weaver Portal
    • We made subtle changes to how the managed dictionaries are being listed and how you enable what you need
    • We also support your instance of the app having a custom name, so you can differentiate NMT output based on custom parameters. Example NMT with our without dictionary

    In the near future we want to highlight those dictionary/term entries so as a user you can see how your terms are being applied. 
    Possibly perform a QA based on terms not used.

    If you can think of other missing features we should consider, please do reach out.

    Many Thanks

    Lyds

    Lydia Simplicio | RWS Group

    _______
    Design your own training!
    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

    emoji
Children