Mobile menu

Wf 5.52zd various problems, is it just me?
Thread poster: Tony M

Tony M  Identity Verified
France
Local time: 21:36
Member
French to English
+ ...
Nov 30, 2007

I'm a relative beginner with W/F, I've only just paid for the registered version in order to use it for the first time in earnest on a 'proper' job. I'm using Word 2003 under XP.

It seems to me that I'm encountering certain problems that I didn't have with earlier versions (I've been dipping in and trying it out every now and again for ages!)

Because of my inexperience, it would be immensely helpful if someone could tell me if I'm just doing something wrong, or if I'm expecting too much — or maybe if it's a quirk or an issue.

1) Sometimes, I have validated a segment, and then in the light of the next segment, decide I would like to concatenate the 2 and start the segment again. But apparently, once a segment has been validated, there is no way I can find that will allow me to expand it.

2) A related problem: if I have a proposed segment that is too long, I can highlight the bit I want and use 'Force Segment' — but sometimes, it's the reverse, and nothing I can do will seem to expand the segment right out so that I can then highlight the bits I want. No, I am not running up against a LR or HR... and I have tried playing around with the segment delimiters in use.

3) Consistently now, I am finding that old 100% or near matches from the TM work OK, but new segments don't seem to be being added to the TM (remember, I have the registered version, and am only at around 1200 TUs) — as a result, I don't get any matches, and have to keep re-typing the same text — rendering the whole expensive exericse a bit pointless! However, after a seemingly random number of tries typing the same text, for some reason it suddenly seems to wake up and do its job. Is there some kind of issue with how long it takes for segments to reach the TM?

4) Is anyone else as irritated as I am by the way W/F overwrites the contents of the clipboard under certain circumstances? It seems to be after validating an edited fuzzy-match segment, but I'm not sure about that.

5) When using 'translate until no match', is it usual for W/F to display '100%' for segments where there is a small change — for example, words going from l/c to ALL CAPS, or an initial cap, or the punctuation at the end of a segment? These tiny but imprtant details seem to get glossed over sometimes... and yet others, it gives me a fuzzy match for segments that are to all intents and purposes 100% identical (and yes, I have checked most carefully!).

6) There seems to be something wrong with the way this version handles figures; this morning, I was using 5.51 and it coped OK with figures (only had problems with words!), and now I've 'upgraded', it's the figures that are giving me some grief — though not all the time...

Well, I think that's quite enough for tonight — any help on any of the above would be much appreciated, before I tear out the rest of my hair and throw the whole lot out of the window!

PS: I perhaps should have added that I am working in a rather large Word document, in a table exported from an XLS.

[Edited at 2007-11-30 19:15]

[Subject edited by staff or moderator 2007-12-10 13:52]


Direct link Reply with quote
 
Charlie Bavington  Identity Verified
Local time: 20:36
French to English
A relative Luddite writes.... Nov 30, 2007

I can't help with most of your issues.
As a former programmer, I only use 'puters for stuff which it is stupid not to - I don't trust 'em at all (similarly, I once knew a bloke who worked for Rolls Royce engines who absolutely refused to fly anywhere). So I don't go in for that auto-translate or placeables malarky.

However, with respect to segments, I think you're right, once it's done, it's done. But what you can do is end the translation session and frig it manually, turning the toggle protection off while you do it.
So say you've got:
{Fr seg 1}0{Eng seg }
{Fr seg 2}0{Eng seg 2 - WIP, and you realised this would all look better as one unit}

Turn off the protection. Delete the "{ at the start of Fr seg 1, delete "}0 {Eng seg 1}" . Ditto seg 2.

You should be left with the text looking as if you had stopped translating immediately before Fr seg 1.

Toogle protection back on.

Do anything you need to do to ensure that Fr segs 1 & 2 will get treated as one (e.g. replace full stop with ; - it doesn't matter about the punctuation in the source text in this document anyway, you've still a copy of the orginal source to refer back to if need be, right?)

Stick the cursor immediately before Fr seg 1, and hit ctl + alt +down to start translating. You should find segs 1 & 2 become one seg. You may find, depending on their relative size and your settings, that a partial match from the old seg 1 is proposed.

Hope that's clear

Re: not finding matches.
I've noticed this. I've also noticed that, if I open the TM in Excel, these appear to be marked FR-FR (i.e. source lang code) in about column 7 or 8. You can clear this crud away and no harm appears to result and WF finds them again as matches.
I only discovered this a few days ago - no idea why it does it !


[Edited at 2007-11-30 20:18]


Direct link Reply with quote
 

Lori Cirefice  Identity Verified
France
Local time: 21:36
French to English
A few ideas Nov 30, 2007

>1) Sometimes, I have validated a segment, and then in the light of the next segment, decide I would like to concatenate the 2 and start the segment again. But apparently, once a segment has been validated, there is no way I can find that will allow me to expand it.

I have never tried that - are you sure you really need to concatenate (I had to look that word up

>2) A related problem: if I have a proposed segment that is too long, I can highlight the bit I want and use 'Force Segment' — but sometimes, it's the reverse, and nothing I can do will seem to expand the segment right out so that I can then highlight the bits I want. No, I am not running up against a LR or HR... and I have tried playing around with the segment delimiters in use.

I hardly ever need to expand/reduce a segment, only when things get cut off at the wrong place due to punctuation (ie : Mr. Smith) In all the versions of WF that I have used, I've never had any trouble with segmentation. Is there anything particular about your text that you should need to expand/reduce ?

You probably already know this, but on the setup "general" tab you can modify the way WF segments.

3) Consistently now, I am finding that old 100% or near matches from the TM work OK, but new segments don't seem to be being added to the TM (remember, I have the registered version, and am only at around 1200 TUs) — as a result, I don't get any matches, and have to keep re-typing the same text — rendering the whole expensive exericse a bit pointless! However, after a seemingly random number of tries typing the same text, for some reason it suddenly seems to wake up and do its job. Is there some kind of issue with how long it takes for segments to reach the TM?

I have noticed this in older and recent versions as well, though not often enough to really annoy me. Try reorganizing your TM every once in a while. Usually it happens on "recent" TUs for me too. If this is happening consistently, just go back to an older version that worked well for you.

>4) Is anyone else as irritated as I am by the way W/F overwrites the contents of the clipboard under certain circumstances? It seems to be after validating an edited fuzzy-match segment, but I'm not sure about that.

Not sure I understand what you're describing...

>5) When using 'translate until no match', is it usual for W/F to display '100%' for segments where there is a small change — for example, words going from l/c to ALL CAPS, or an initial cap, or the punctuation at the end of a segment? These tiny but imprtant details seem to get glossed over sometimes... and yet others, it gives me a fuzzy match for segments that are to all intents and purposes 100% identical (and yes, I have checked most carefully!).

No - WF usually picks up on very small changes (especially punctuation). Not sure about caps though ? Maybe you should go back to your older version. I've been happily using 5.50a here, although I have a 5.52something version installed on my other computer as well, I prefer this one.


>6) There seems to be something wrong with the way this version handles figures; this morning, I was using 5.51 and it coped OK with figures (only had problems with words!), and now I've 'upgraded', it's the figures that are giving me some grief — though not all the time...

Arrgh - no idea !

>Well, I think that's quite enough for tonight — any help on any of the above would be much appreciated, before I tear out the rest of my hair and throw the whole lot out of the window!

Oh dear - you need a break !

>PS: I perhaps should have added that I am working in a rather large Word document, >in a table exported from an XLS.

Should not make any difference ...


I think that your segmentation woes may just be that you're not used to the way WF segments ? However your matching and number handling issues seem more serious, I suggest going back to an older version and see if that improves. Also try the yahoo support group.


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 20:36
French to English
+ ...
A few options Nov 30, 2007

Hi Tony,

Oh dear, I know that window-throwing moment feeling only too well..... Not usually with Wordfast, however, which I've found to be generally straightforward and the Yahoo support group is brilliant at solving any problems!

Re your first point: Have you tried doing Alt Delete, to take you back to the previous segment and then either expanding that segment or copying the text you want into the source and proceeding from there?

Point2: If expand/reduce segment don't work, I just manually remove the text I want and paste it below so Wordfast then treats it as a new segment (or paste it in if the break has occurred before the end of the sentence). I usually have this in documents that have been converted from pdf, so the breaks aren't in the right place. It's a pain, but not insurmountable.

Point 3: I've just upgraded to 5.52zd and haven't experienced any problems with TM recognition. As Lori suggests, reorganising might help.

Point 5: The longish (12,000 word) job I've just done certainly picked up on sentences that were in lower case in the contents list and upper case in the text itself - identical but for the case and displayed them as fuzzy. Similarly, commas, colons etc at the end all trigger fuzzy not identical, as they should.

Re matching figures, I think it depends how fuzzy the match is - for segments with the same number of figures, it seems to cope fine, but if there are different figures, e.g. section 2, 25g 14 MW or whatever, but the fuzzy only comes up with 2 of these numbers, you'll end up with them out of sync - that's what I've found anyway. Again, I did have problems a while ago with a previous version (no idea which one now), which did put in wrong nos., but, like your job, that was a huge table of invoices with lots of similar figures, converted from a pdf, and I just learnt to be extra careful with it. Normally, it's fine.

Hope this helps - and do persevere! I've just attended the recent intermediate Wordfast course in Nice and found it really useful. I'm by no means an expert, but I do feel more confident about its capabilities and more able to cope if I go wrong!


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 21:36
Member
French to English
+ ...
TOPIC STARTER
Further comments... Dec 1, 2007

In the light of some of your most helpful comments above, I paid particular attention during this evening's translation session, and noted a few specific points:

3) 'missing' TUs
I tried validating a segment, then opening the TM as a TXT file; just sometimes, the segment is not visible (it does not appear with the wrong language codes); after one or more further attempts, it appears, but there is a spurious carriage return where the missing segment should have been.
Merely saving the document sometimes (but not always) provokes the entry of the segment into the TM, as does SOMETIMES (but not always) re-organising the TM.

I feel more than ever convinced this must be a bug!

False 100% matches
I get a weird and wonderful mix of displays: sometimes I get a grey box but with 100% displayed, sometimes I get a grey box with 100% matched text in it, but it still says 0%, etc.

More worrying is the way it sometimes gives me a 100% match with as much as one whole word extra ; while other times, it won't even offer so much as a fuzzy with just oen word different.

All in all, I get the impression that I'm wasting a huge amount of time here!

The reason for the odd segmentation is because I'm doing some work with comma-separated lists of attributes, where I want to use commas as a segment delimiter, but just occasionally, I want to expand the segment to form a bigger, more logical chunk. Also, there are lots of figures with ( ), +, and full stops marking the end of abbreviations like dim. All of this means that I am constantly having to reconstruct segments into more manageable chunks — not least, because the word order is often different between FR and EN.


Direct link Reply with quote
 

Heinrich Pesch  Identity Verified
Finland
Local time: 22:36
Member (2003)
Finnish to German
+ ...
I use 5.50za without problems Dec 1, 2007

Its always a risk using the latest version. But I find my current version ok. If you cannot find it from the wf-site, send me a mail and I'll send it to you.
When combining two segments you should first press cntrl+alt+X (empty source segment), then alt+up, alt+down and then alt+PgDn or use forced segmentation.

Cheers
Heinrich


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 21:36
Member (2006)
English to Afrikaans
+ ...
Some answers Dec 1, 2007

Note: I haven't checked if I remember the shortcuts correctly.

Tony M wrote:
1) Sometimes, I have validated a segment, and then in the light of the next segment, decide I would like to concatenate the 2 and start the segment again. But apparently, once a segment has been validated, there is no way I can find that will allow me to expand it.


Usually you should be able to go Alt+up to the previous segment, and then Alt+PgDn to expand it.

If that doesn't work (and sometimes it doesn't), press Alt+Del, which will unsegment the sentence and end your session. Then select both sentences and press Alt+Shift+down to get both in one segment.

2) A related problem: if I have a proposed segment that is too long, I can highlight the bit I want and use 'Force Segment' — but sometimes, it's the reverse, and nothing I can do will seem to expand the segment right out so that I can then highlight the bits I want.


Go Alt+del to end the session and unsegment the segment. Then highlight all the way up to where you want to segment, and press Alt+Shift+down to start again.

3) Consistently now, I am finding that old 100% or near matches from the TM work OK, but new segments don't seem to be being added to the TM...


Go Alt+Del or Alt+End to end the session. Open the WF control panel, and reorganise your TM.

4) Is anyone else as irritated as I am by the way W/F overwrites the contents of the clipboard under certain circumstances? It seems to be after validating an edited fuzzy-match segment, but I'm not sure about that.


I don't have that problem. My clipboard stays intact.+

5) When using 'translate until no match', is it usual for W/F to display '100%' for segments where there is a small change — for example, words going from l/c to ALL CAPS, or an initial cap, or the punctuation at the end of a segment?


It is not a bug, it is a feature. But if you don't want the feature, go to PB and enable stuff like DifferentCaps=99 (can't remember the exact name, check the user manual). This will force such instances to display 99% instead of 100%, which means your autotranslate will stop at that segment.

6) There seems to be something wrong with the way this version handles figures; this morning, I was using 5.51 and it coped OK with figures (only had problems with words!), and now I've 'upgraded', it's the figures that are giving me some grief — though not all the time...


Any WF version is always a beta version. To find out how bugfree it is, check for how long a version has been the latest version. If not long, then use an older version.


[Edited at 2007-12-01 09:35]


Direct link Reply with quote
 
xxxmediamatrix
Local time: 15:36
Spanish to English
+ ...
Beta... Dec 1, 2007

Samuel Murray wrote:
Any WF version is always a beta version.


That's true of the vast mjority of softwares on the market - including stuff like Windows.

And especially so for softwares for 'niche' markets - like CAT tools - where the developers are unable to put the necessary resources into full-scale testing and debugging under all working conditions likely to be encountered in the 'real world'.

The worst 'culprits' in this respect tend to be open source softwares. In some cases - MySQL, for example - the most-recent stable version for Windows is over two years old, even though there have been numerous new major and minor releases since then.

MediaMatrix


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 21:36
Member
French to English
+ ...
TOPIC STARTER
Beta.. but... Dec 3, 2007

Huge thanks to you all — it's great having this wonderful support, and I'm only sorry if I am asking stupid questions, or ones that have been asked many times before.

I do know the Wordfast group on Yahoo!, but part of the trouble is, there is such a vast welter of information out there, it is sometimes hard to know exactly how to wedd through it all to find the answer — especially with the pernennial problem of any 'help' system: "how can I ask for help on something when I don't even know what it's called?"

Special thanks to Samuel for his recent contribution, and the specific answers, which I have been able to implement; just knowing how things are supposed to work helps a lot!

Beta version or not, I still find it odd that some very blatant and immediately obvious new bugs should get through; I'm not talking about obscure situations here, but the most obvious function of a CAT tool: to save and propose translated segments.

Another day's use has proven beyond any doubt that some (but not all) segments are not being saved into the TM — even though they are ostensibly similar, and the exact same segment may eventually get saved later on. Re-organizing the TM does NOT consistently help, though sometimes saving the document AND re-organizing seems to have an effect.

Is there any way of manually forcing the saving of a TU? I've wasted so much time today having to constantly retype identical segments many times, until eventually they decide to save.

I've also noticed that sometimes a TU is offered, perfectly correct, but with a grey background and marked as 0% — this seems to be a sort of 'halfway house' situation between not being saved at all, and being recognized properly.

The clipboard issue is also fairly consistent, though still seems to show a degree of randomness. That is to say, if I copy something to the clipboard (e.g. some special character, or a recurring proper name), it is available for me to paste into my document for quite some time, until suddenly it gets overwritten by an unrelated segment; I can't quite work out when this is happening, but it seems to be related with editing / validating a segment that has been a fuzzy match and so needed editing.

I'd be very interested to hear if other people have had similar experiences? And whether they are confined to just the latest release, or have been there in earlier ones too?


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 21:36
Member (2006)
English to Afrikaans
+ ...
Did you report the bugs? Dec 3, 2007

Tony M wrote:
Beta version or not, I still find it odd that some very blatant and immediately obvious new bugs should get through...


Bugs get through because people don't report them.

Is there any way of manually forcing the saving of a TU?


There is a PB option that makes the "Quick Clean" not clean, but only add TUs to the TM. Perhaps you can Quick Clean regularly...?

Or... how about downloading an older version?


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 21:36
Member
French to English
+ ...
TOPIC STARTER
Bugs... and cleaning Dec 3, 2007

Samuel Murray wrote:
Bugs get through because people don't report them.


No, I didn't report them yet, simply because I wanted to be sure they were bugs first — and I find it odd that no-one else seems to have observed them?

Perhaps you can Quick Clean regularly...?


Sadly, I have a very large document, and that would just take TOO much time


Or... how about downloading an older version?


I think I may have to resort to that, although once again, it is all wasted time with my deadline looming...


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 20:36
French to English
+ ...
Wf Companion Dec 3, 2007

hello again Tony,

Just a thought, but have you tried enabling the Wordfast Companions in Pandora's box? Whilst this won't solve all your problems, it will at least show you what's actually in the TM and what % of fuzziness it is whilst you work? I'm working on a pdf conversion at the moment, which doesn't always show the right colours, even though I've changed background fill to none from white, but I notice that the TM companion does show the right colour. I have both companions open on my second screen and find them invaluable for showing all the glossary and TM entries at a glance.

Claire


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 21:36
Member
French to English
+ ...
TOPIC STARTER
Great tip, Claire! Dec 3, 2007

Thanks, Claire — that's a great tip, though I must admit, I'm still very nervous of the idea of opening Pandora's box!

But sadly, it still doesn't address my fundamental problem; I already know that my TUs aren't going into the TM; what I don't know is what to do about it!

And the bad news is, I've gone back to an earlier version, and it still doens't work! There must be something curious about my document or my configuration that's causing it, but all I know right now is that I have wasted about a day trying to sort this job out, and have more or less ended up translating it manually anyway, in which case, W/F just gets in the way and slows me up!

Oh woe is Tony, with great gaping holes in his TM!


[Edited at 2007-12-03 22:27]


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 20:36
French to English
+ ...
Short segments? Dec 3, 2007

Are the segments it's not finding very short? Wordfast is notoriously bad at recognising short segments - don't know why. The only other thing I can think of might be to change the fuzzy threshold under Setup. I use 50 rather than the recommended 75 and it doesn't seem to bring up too much rubbish.... However, if you're saying it's not finding 100% matches, that won't be the answer.

As you say, it might just be that particular file. I suppose you could try copying all the text and inserting it in a new document, but it's all messing around and not getting on with the job....

Don't let it put you off - it really is worth it in the end!

Bonne chance!

Claire


Direct link Reply with quote
 


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


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

Wf 5.52zd various problems, is it just me?

Advanced search


Translation news related to Wordfast





memoQ translator pro
Kilgray's memoQ is the world's fastest developing integrated localization & translation environment rendering you more productive and efficient.

With our advanced file filters, unlimited language and advanced file support, memoQ translator pro has been designed for translators and reviewers who work on their own, with other translators or in team-based translation projects.

More info »
WordFinder
The words you want Anywhere, Anytime

WordFinder is the market's fastest and easiest way of finding the right word, term, translation or synonym in one or more dictionaries. In our assortment you can choose among more than 120 dictionaries in 15 languages from leading publishers.

More info »



All of ProZ.com
  • All of ProZ.com
  • Term search
  • Jobs