Pages in topic:   < [1 2 3]
Is there anybody out there?
Thread poster: Matthias Brombach
Olav Karlsen
Olav Karlsen
Norway
Local time: 21:29
English to Norwegian
+ ...
Flexional languages Feb 4, 2017

Wojciech Matyszkiewicz wrote:

mikhailo wrote:

Just a question. How will this feature deal with flexion languages? For now filter in DVX does not support search of all wordforms.
Most of langs has a limited flexion (different sing.-pl. forms) but DVX does not support even them.

How could such defective function be very useful?



But it's already there. Set the minimum score to about 70% and enable fuzzy terminology match and voila.
There might be some false positives, but as far as I know, no CAT tool does better here. Fuzzy matches is the only choice to deal with terminology in inflected languages.


Limiting lookups to e.g. the nominative for nouns and pronouns, infinitive for verbs can be beneficial. The process goes much faster since the glossary size can be made smaller. After all, the translator is supposed to have the linguistic competence to correct the inclination missing. This is part of the competence the contractor pays for. If there are repetitive set phrases, the translator can send them to the TM to save time.


 
Wojciech_ (X)
Wojciech_ (X)
Poland
Local time: 21:29
English to Polish
+ ...
Right Feb 4, 2017

Olav Karlsen wrote:

Wojciech Matyszkiewicz wrote:

mikhailo wrote:

Just a question. How will this feature deal with flexion languages? For now filter in DVX does not support search of all wordforms.
Most of langs has a limited flexion (different sing.-pl. forms) but DVX does not support even them.

How could such defective function be very useful?



But it's already there. Set the minimum score to about 70% and enable fuzzy terminology match and voila.
There might be some false positives, but as far as I know, no CAT tool does better here. Fuzzy matches is the only choice to deal with terminology in inflected languages.


Limiting lookups to e.g. the nominative for nouns and pronouns, infinitive for verbs can be beneficial. The process goes much faster since the glossary size can be made smaller. After all, the translator is supposed to have the linguistic competence to correct the inclination missing. This is part of the competence the contractor pays for. If there are repetitive set phrases, the translator can send them to the TM to save time.


I fully agree. But some words e.g. in Polish, are so heavily inflected that we need the possibility to add synonyms to a glossary, because the root of a word may change as well.
Consider singular Polish "wieś" (village) and plural "wsie" (villages) - no pipe mark can help here, but only the option to add the plural as a synonym. Adding a sole "w|" would be ridiculous




[Edited at 2017-02-04 18:19 GMT]


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 21:29
Questions Feb 5, 2017

Wojciech Matyszkiewicz wrote:

There might be some false positives, but as far as I know, no CAT tool does better here. Fuzzy matches is the only choice to deal with terminology in inflected languages.


Can you give a few examples of Polish words that are recognised as belonging to the same stem, and that aren't recognised in other CAT tools that offer stemming (e.g. via Hunspell)?

Wojciech Matyszkiewicz wrote:

But some words e.g. in Polish, are so heavily inflected that we need the possibility to add synonyms to a glossary, because the root of a word may change as well.
Consider singular Polish "wieś" (village) and plural "wsie" (villages) - no pipe mark can help here, but only the option to add the plural as a synonym. Adding a sole "w|" would be ridiculous


Where would you want to add the plural as a "synonym"? At the source side? Is this possible in Déjà Vu? And, if so, it would probably be impossible to have Déjà Vu match the plural of the target term automatically.

The number of forms in Polish look very complicated for me (as a native Dutch, just one singular: ingenieur, one plural: ingenieurs):

Screen Shot 2017-02-05 at 12.30.50

Just out of interest: Would it be possible to define 1 stem for all singular forms and 1 for the plurals?


 
Epameinondas Soufleros
Epameinondas Soufleros  Identity Verified
Greece
Local time: 22:29
Member (2008)
English to Greek
+ ...
Correct term matching Feb 5, 2017

Accurate term matching makes for accurate quality assurance. It's not that the translator is lazy or incompetent. By allowing to enter RegEx expressions in the term base, one can be as detailed as one wants. If you only want the nominative, do it that way. But I want to be detailed and thorough, and I don't want incorrect flags about supposed QA errors.

[Edited at 2017-02-05 11:52 GMT]


 
Wojciech_ (X)
Wojciech_ (X)
Poland
Local time: 21:29
English to Polish
+ ...
What I meant Feb 5, 2017

CafeTran Training wrote:

Where would you want to add the plural as a "synonym"? At the source side? Is this possible in Déjà Vu? And, if so, it would probably be impossible to have Déjà Vu match the plural of the target term automatically.



i think each of us understands the idea of a "synonym" differently in a CAT tool. I'm talking about the option that I'm using in Trados.
SOURCE:
wieś
wsie

TARGET:
village
villages

Trados recognizes both "wieś" and "wsie". It's not just an additional info (e.g. a note) that Trados wouldn't recognize. Deja Vu doesn't have this option - you can only add a note, but this wouldn't be recognized when the term appears.
That's why I'm asking for such a feature in Deja Vu.


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 21:29
Thanks! Feb 5, 2017

Wojciech Matyszkiewicz wrote:

i think each of us understands the idea of a "synonym" differently in a CAT tool.



Yes


I'm talking about the option that I'm using in Trados.
SOURCE:
wieś
wsie

TARGET:
village
villages

Trados recognizes both "wieś" and "wsie". It's not just an additional info (e.g. a note) that Trados wouldn't recognize. Deja Vu doesn't have this option - you can only add a note, but this wouldn't be recognized when the term appears.
That's why I'm asking for such a feature in Deja Vu.


Thanks. Now, please back to your statement:


There might be some false positives, but as far as I know, no CAT tool does better here.


Can you give some examples please, where Déjà Vu does better in recognition of inflected terms than any other tool?


 
Wojciech_ (X)
Wojciech_ (X)
Poland
Local time: 21:29
English to Polish
+ ...
Yhm Feb 6, 2017

CafeTran Training wrote:

Wojciech Matyszkiewicz wrote:

i think each of us understands the idea of a "synonym" differently in a CAT tool.



Yes


I'm talking about the option that I'm using in Trados.
SOURCE:
wieś
wsie

TARGET:
village
villages

Trados recognizes both "wieś" and "wsie". It's not just an additional info (e.g. a note) that Trados wouldn't recognize. Deja Vu doesn't have this option - you can only add a note, but this wouldn't be recognized when the term appears.
That's why I'm asking for such a feature in Deja Vu.


Thanks. Now, please back to your statement:


There might be some false positives, but as far as I know, no CAT tool does better here.


Can you give some examples please, where Déjà Vu does better in recognition of inflected terms than any other tool?


Not quite understand why you want me to give you any examples, since I didn't mention that Deja Vu does better than any other tool. I simply stated, that AS FAR AS I KNOW, all CAT tools have more or less the same problem with inflected languages and fuzzy matching seems like the only sensible option available now.
Pardon me, if you misunderstood that.

I would also be happy to know, which tool does better in this department, although bearing in mind your nick, I can guess which one


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 21:29
We are here to learn Feb 6, 2017

Wojciech Matyszkiewicz wrote:

I simply stated, that AS FAR AS I KNOW, all CAT tools have more or less the same problem with inflected languages and fuzzy matching seems like the only sensible option available now.
Pardon me, if you misunderstood that.


I'm here to learn, and often that means that I have to ask questions. I indeed interpreted your statement that you had positive experiences with Déjà Vu (a great tool, that I've used for many years, until I migrated to Mac). I was hoping to learn from your experiences.

I'd say that any rule-based stemming would result in less loss (thus better recognition) than any simple cutting of letters or Levensthein algorithm.


I would also be happy to know, which tool does better in this department, although bearing in mind your nick, I can guess which one


I don't know which tool performs best. That's exactly what I was hoping to learn, since I don't work from a heavily inflected source language.

My assumption would be that Transit, omegaT and CafeTran perform best, since they all use rule-based stemming.

BTW: My nickname doesn't mean that I don't have a personal opinion about the tool I use or that I think that it's the greatest thing after chocolate. It's just a working tool.


 
Pages in topic:   < [1 2 3]


To report site rules violations or get help, contact a site moderator:

Moderator(s) of this forum
Pavel Tsvetkov[Call to this topic]

You can also contact site staff by submitting a support request »

Is there anybody out there?






Trados Studio 2022 Freelance
The leading translation software used by over 270,000 translators.

Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop and cloud solution, empowering you to work in the most efficient and cost-effective way.

More info »
Wordfast Pro
Translation Memory Software for Any Platform

Exclusive discount for ProZ.com users! Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value

Buy now! »