define 'word' boundaries for better word count in Trados

Hi,

Today I⁠ got a small file from a client using MemoQ. When it comes to the word count, it showed that Trados had 140 words instead of the 123 from MemoQ.

It turned out, that placeholders were defined badly thus variables between square brackets did not get converted and this was a difference of 5 words. So there’s still a difference of 12 words (10%) between the two.

When I had a closer look it showed that there were about 6 EU standards referenced in the text and Trados counted EACH number and string between and sourrinding the '/' as 1 word.

Is there a way to define these via regex as 1 word so they get counted as 1 word? The same would go for quite long Standard names (e.g. DIN 2137-1:2018-11 or even longer ones) or badly formatted URLs and some other strings.

Just to make sure: No, it’s not an option to convert them to tags or untranslatable text as they need to be adapted (e.g. adding word joiners and no-break spaces to avoid wrapping)

I  know you can adapt some missing word count settings via TM settings (language resources e.g. ’count as word if words contain’) but I can’t see any option to add regex or similar to define rules like above or am I⁠ missing something?

Best regards,

Pascal



clarification of additional settings
[edited by: Pascal Zotto at 2:14 PM (GMT 0) on 20 Jan 2025]
emoji
  •  

    I can't offer an easy fix, but some insight into the complexity of that which you are trying to achieve: https://multifarious.filkin.com/2012/11/13/wordcount/ (You might remember that blog post.)

    As a tiny practical contribution, if you define a regex that matches EU standards and convert them (or the problematic part thereof) into tags, your word count would decrease as a tag is counted as one word AFAIK.

    Daniel

    emoji
  • Hi ,

    yes, but as I said in the first post tags are not an option as I⁠ need to adapt them in translation. This especially goes for EU and other standards/regulations aso.

    emoji
  • Is there a way to define these via regex as 1 word so they get counted as 1 word? The same would go for quite long Standard names (e.g. DIN 2137-1:2018-11 or even longer ones) or badly formatted URLs and some other strings.

    Just to make sure: No, it’s not an option to convert them to tags or untranslatable text as they need to be adapted (e.g. adding word joiners and no-break spaces to avoid wrapping)

     

    I understood that the numbers were the problematic part – do they also change in the translation, such as  "2137-1:2018-11" in the above example?

    emoji
  • Hi ,

    the numbers themselves should not change (although it happened that a client mixed up numbers of a standard and I had to correct it) but I need to replace and add other characters:

    e.g. (the brackets only show which char was added/replaced)

    DIN 535/2137-1:2018-11 > DIN(narrow nbsp)535/(word joiner)2137(non-breaking hyphen)1:(word joiner)2018(non-breaking hyphen)11

    emoji
  •  

    It's not worth it for such a small document, but if you have this case more often, you might want to do your replacements in the source (with CleanUp Task e.g.) and then make them into tags (I hope that is possible, but should). Once you've made them into tags their content can't be altered anymore - as I am sure you know.

    emoji
  •  

    Yes, I get these quite often in larger texts too but as I said several times by now: TAGs are NOT an option as I⁠ need to be able to change parts of the string AND I need them to be counted as 1 word: say DIN 535/2137-1:2018-11 = 1 word. So yes I know they can’t be altered ^^ and it would also give a wrong word count as tags are not counted during analysis.

    I’m rater looking for a way to define word count as regex for DIN standards to be taken as 1 word. Even saying string between spaces is 1 word which would then still be closer to the correct word count (at least for these strings, but it would then result in problems with other strings).

    But I guess this is not possible at the moment.

    emoji
  •  

    I tend to agree with you... I'd also like a feature like that in Studio.  But it's not there.  A workaround might be to use the Cleanup Tasks app - https://appstore.rws.com/Plugin/23 - or the Data Protection Suite - https://appstore.rws.com/Plugin/39 - as you could use regex with these and add the batch task in a custom project prep (if you have the Professional Version of Studio) when you create your projects.  With the Freelance it's just an additional step to run the batch task after preparing your project and then run the analysis again. 

    Paul Filkin | 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
  •  

    Yes, but the problem is that I don’t need tags of these at all (they would even be counterproductive), so I don’t see how the plugin (which I⁠ use regularly in my professional version) would help me with that problem?

    emoji
  •  

    What I meant (see above) and how I understand is that you modify the DIN standard notation as needed in the source, using CleanUp Task. Then you "protect" the numerical part of the standard (e.g. "535/2137-1:2018-11") as a tag. This would result in the standard being counted as one word, and in being written the way you need it. I agree with Paul, for special use cases like yours it might be desirable to be able to define word boundaries freely, but since that is not an option at this point, this looks like a viable workaround.

    Screenshot of Trados Studio interface showing a segment with the text 'DIN 5352137-1:2018-11' highlighted in purple and a warning message indicating 'DIN' counts as one word, tag as none.

    emoji


    Generated Image Alt-Text
    [edited by: RWS Community AI at 2:08 PM (GMT 0) on 27 Jan 2025]
  • I know, but you always miss the point where I explained several times already that I MUST be able to replace the hyphens and add special space characters. So tags are NOT an option at all.

    emoji
1 2