Curious glossary behaviour
Thread poster: Tony M

Tony M  Identity Verified
France
Local time: 18:57
Member
French to English
+ ...
Feb 25, 2009

I am using a registered version of WF 5.51t3 [< correction!] with Word under XP, and am only just really coming to grips with it now as I am using it on a 'real' job (FR > EN).

(I did once try a later version, but it screwed things up so royally, I was obliged to revert!)

I am making a great deal of use of the 'glossary' feature (TM is of little use for this job), but it keeps behaving oddly. Can anyone please tell me if this behaviour is 'normal' or a known problem, and if there are any things I could do to solve or at least mitigate the problem? My glossary is not particularly large, only around 22k.

1) Sometimes, words that I know perfectly well are in the glossary do not come up as placeables (even though I may have used them just a few segments before. This document does not have weird formatting, and I have checked very carefully for any syntax anomalies, however minor.

2) Where more than one translation exists for one particular source term in the glossary, it sometimes works as expected, i.e. the glossary list pops up, and I can use the arrow keys to select which option I want to use, then press 'return' to paste the term into my target segment. However, sometimes (and it always seems to be with particular terms) the list pops up, but not in the right place: either right at the beginning of the alphabet, or a few words away from the wanted word. In these circumstances, if I now use the arrow keys to move down to the wanted translation, pressing 'return' merely enters the term from the source segment (as if I had used 'paste source' just for that one word)

3) Where I have a word that always recurs in parentheses, I have entered it as such in the glossary, and although it appears alright at the head of the list, it is never offered as a placeable; for certain texts this is a real pain!

4) glossary 'fuzzy' behaviour is turned on, but seems very erratic! Sometimes, it will offer a gloaasry entry for a word that is quite a long way from the actual source word, but at other times it will miss something very simple, like a plural -s or a feminine -e

For example, it will offer me the translation of 'document' where the source word is 'documentation' (logical enough, and not a problem); but then it can't seem to cope with the difference between 'émis' and 'émise' (much more of a nuisance!); it also seems to only work on the end of words, so, for example, it fails to recognize « l'arrêt » as being related to « arrêt » — these are particular problems with FR syntax, of course, but very irksome and time-wasting! Particularly since the TU fuzzy matching algorithm seems much less fussy!

Any enlightenment would be much appreciated!

[Edited at 2009-02-26 08:44 GMT]


Direct link Reply with quote
 

Gerard de Noord  Identity Verified
France
Local time: 18:57
Member (2003)
German to Dutch
+ ...
Update and try again Feb 25, 2009

Hi Tony,

WF 5.1 sounds like 2006, WF 55.1 would be 2007. My current version is 5.56.

I'd never add terms within parentheses to the glossary. That's unnecessary and could harm term recognition (the latter is my gut feeling).

I believe I've read fuzzy term recognition has improved too, due to Finnish and Russian preasure groups, but do you really need it?

Cheers,
Gerard


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 17:57
French to English
+ ...
With Gerard Feb 25, 2009

Yes, I'd be tempted to update to a newer version too - I'm using 5.52zd at the moment. I did try a later one but it did something odd (can't remember what now!), so I reverted to this one which seems to work fine.

I must admit I don't use fuzzy recognition either - it brings up far too much rubbish for my liking, whereas the straight recognition will often bring up plurals or words after apostrophes anyway.

I agree that occasionally words aren't recognised in one segment but are in the next - no idea why! Sometimes reorganising the glossary helps.

Have you tried the TM and Glossary Companions? I find these invaluable if there are multiple entries for particular terms in the glossary as all of these are shown on the companion screen, rather than having to scroll down the drop-down menu. You should be able to activate these from Pandora's Box. I have them on a separate screen, but if you have a wide screen there should be room for them on your main screen anyway.

Good luck!


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 18:57
Member
French to English
+ ...
TOPIC STARTER
Thanks, Claire and Gerard! Feb 26, 2009

After my earlier experiences of updating versions in mid-project, I am naturally a little wary this time — especially since the 5.51t3 version I'm using is not so very old as all that. And the glossary seemed to work fine on previous jobs...

I really just wanted to know if this was normal behaviour or a known bug, or if there is something specific going wrong here that I might be able to fix.

I have tried re-organizing the glossary (frequently, in fact), and I noticed this morning that the first time I encountered one of these 'rogue' terms (and it does always seem to be just one or two specific terms that give the problem) it worked fine; but after that, it started screwing up again.

Terms in brackets, Gerard:

In this particular instance, it is a term that always only ever appears in brackets, and it is a real nuisance when you are in a segment containing a whole nice string of placeables to suddnely have to break out of your rhythm to type the opening and closing parentheses. Unlike certain other punctuation marks, I don't see why parentheses in particualr ought to pose a specific problem, I'd have thought the system could treat them just like any other character.

As for skipping terms, it is a real nuisance! Sometimes, I just assume that I must be mistaken, so I enter it all over again; then when I check the glossary, I have loads of entries for the same term — and still it sometimes seems to get missed!

And fuzzy matching, Claire — I don't find that the 'rubbish' it throws up is particularly irksome, but it is such a waste of time having to manually enter sometimes four different declensions of a FR adjective, for example; and wirth verbs, sometimes just having the root offered aa a placeable is a great help, where the word is long and tricky to type; any eventual conjugation is easy enough to add on the end.

Thanks for the tip about the companions; I don't know how to use PB, but i'll look into that one, it sounds as if it could be just what I need.

Learning all the time!


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 17:57
French to English
+ ...
Fuzzy matching Feb 26, 2009

Maybe it's because my other language is German that I find fuzzy maches so irksome - with long compound nouns, you really do get all sorts of weird and wonderful suggestions thrown up - I couldn't stand it so I disabled it!

It should be fairly straightforward to update Your Wordfast version even in the middle of a project without ill effects. Just in case, make sure you keep a copy of the template you're using in a separate file, then just follow the download procedure online. It's much easier than it used to be and should just update automatically. If for whatever reason you don't like the new version, you can easily go back and click on your old template to load that again.

Re PB: again that's really easy and definitely worth looking into. If you scroll down the list (under Setup / PB) you'll see

Wf_Companion=TM(OnTop),Glossary(OnTop)

The underscore means the comand is disabled. If you remove the underscore the command will be enabled and you should see two companion windows when you next open your document. PB also has lots of other glossary rules which will help your fuzzy matching - probably best to read the manual about those because I'm not sure I use them, although I do have

GloStrip=able,ible,ed,er,ies,ing,ion,ness,ism,ful,eed,ied,s,e

enabled, which I think means that plurals are shown as well. There's probably a French equivalent of this too. Ditto the FR commands that automatically switch continental decimal points to English or vice versa - invaluable!


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 17:57
French to English
+ ...
Brackets Feb 26, 2009

Re brackets, again there's a command in PB (under Placeables) that allows you to treat items in brackets (of any kind) as placeables - this is probably why it doesn't copy the brackets themselves as part of the placeable, although I agree that can be annoying. I translate for an agency who send me 2-column Déjà Vu files to translate with codes between curly brackets. It would be much easier if I could just transfer the whole bracket and code unit rather than just the code and then type the brackets by hand. Maybe someone else will know of a way of doing this?

Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 18:57
Member
French to English
+ ...
TOPIC STARTER
Thanks, again, Claire! Feb 26, 2009

All those comments are really helpful, Claire!

One of the troubles I find with the Wordfast (and indeed, any other similar!) manual is that it is too much to assimilate if you jsut read through it — and indeed, often pretty meaningless until you actually try it out on a 'real' project; but on the other hand, if you just 'dip into it', it is hard to look things up if you don't know what they're actually called.

I do find 'get you started quick user guides' are helpful.

Thanks to your encouragement, I have investigated Pandora's Box now, and what was previosuly meaningless gibberish suddenly seems to make some sense, now that I know what 'tags' and 'segments' are, at least! Here is a chunk of text which explains just what I wanted to know (but didn't know where to look!) — it follows on of course from what Claire has mentioned above.

I quote directly from the W/F manual:

(see the FTR section for an introduction on Fuzzy Terminology Recognition)
This command complements the glossary's FTR algorithm defined above by providing language-specific endings that should be stripped in order to get a term's root, or stem.
If the ending of a word is present in the list of most common endings (here, ful,ing,ness,es), then this ending is stripped in just one iteration, and the FTR rate is set to 1 - even if more than one letter was stripped. In other words, both GloStemmingRule=5,4,2 and GloStrip=ing,ness would get from the inflected multiplexing to the stem multiplex and hit a match in the glossary for multiplex. But GloStemmingRule=5,4,2 would give a rate of 3 (a rather poor match rate) while GloStrip=ing,ness would yield a better rate of 1. GloStrip, of course, is always tried first.
If the language-specific endings are too many, you can have multiple GloStrip= entries in PB - they will all be taken into account.
It is recommended to use both GloStrip and GloStemmingRule.


It took me several readings to understand what that was all about, but now I have grasped the idea, it's all very helpful!

And while I was at it, I also found a few other potentially useful things!


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 17:57
French to English
+ ...
Information overload Feb 26, 2009

Yes, I know exactly what you mean - it's all just too much to take in when you're first starting out! I know I've said this before, but I did find it really helpful attending a course at this stage. You've mastered the basics, but it's the fine tuning that really makes the program come into its own - like the context search, PB, attributes, etc - all of which seem very complicated when it's just you and the manual!

Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 18:57
Member
French to English
+ ...
TOPIC STARTER
PB and brackets Feb 26, 2009

Claire Cox wrote:

I translate for an agency who send me 2-column Déjà Vu files to translate with codes between curly brackets. It would be much easier if I could just transfer the whole bracket and code unit rather than just the code and then type the brackets by hand. Maybe someone else will know of a way of doing this?


As I understand it, that's exactly what the PB comand is meant to do, isn't it — to copy the contents and brackets as a placeable unit?

PlaceableBetween=[],(),{} Instructs Wordfast to consider text comprised between [ and ], ( and ), { and } as placeables. You can specify other sets of two characters. Separate pairs of characters with commas. It is not recommeneded to use PlaceableBetween= or PlaceableBetween=&; with tagged files.

...but it doesn't make it clear if the brackets themselves are included!

However, the following command should work for what you need, I think:

PlaceableContains=#@=+: Instructs Wordfast to consider words containing the characters appearing after the equal sign as placeables.
Here, for example, the I+YOU expression would be considered a placeable.

So surely if you include { and }, it will consider those + any word contained within them (codes don't usually have spaces, do they) as a placeable.


[Edited at 2009-02-26 11:03 GMT]


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 18:57
Member
French to English
+ ...
TOPIC STARTER
Information overload (cont.) Feb 26, 2009

Yes, Claire, that's eaxctly the problem!

I totally take your point about courses, but there are a number of practical difficulties for me, like distance, cost, and time. I use W/F so little, and have already invested in purchasing my registration, I consider the return on investment to be marginal at present, and if I invest an even huger sum of money going on a course, I will never be able to recoup that.

But apart from that factor, I also find most such courses too intensive — again, the 'information overload' comes into play; perhaps I'm just getting old and my brain is addled, but I no longer find I can assimilate such a mass of information in the context of a short course. Especially if it is not in the context of wokring on a 'real' project.

What I find most immensely helpful is just this sort of interchange, where at just the moment I discover the need for something, a 'mentor' like your good selves comes along and explains in small chunkcs what to do and where to look for further info. It makes such a big difference to have someone who is familiar with the system to say "Yes, i've been there and found the same problem... and this is the solution I found"

I know I'd have got on a lot faster with W/F is someone had just sat down and explained to me the basic principles, it took me ages to discover them really quite by accident...

Maybe someone should write a Beginner's Guide to getting started — with in-document hyperlinks, it is so easy to link across to the more detailed articles to 'find out more'; if I could afford to take the time out of my day job, I'd do it, just for the fun of it! As a former teacher, I enjoy writing training materials.


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 17:57
French to English
+ ...
Brackets again Feb 26, 2009

No, I've tried both the commands: placeable between {} and placeable contains {} and it only highlights the number, not the brackets. Perhaps the numbers command takes higher priority. I may post it as a query on the Wordfast group and see what they say.

I see what you're saying about courses - all I can say is that my investment in both the initial subscription and the courses (two - a general beginners' introduction to CAT tools and an intermediate wordfast course) has repaid itself many times over in vastly increased productivity and improved consistency/quality control/glossary maintenance and retrieval. The intermediate course in particular was a small group of maybe 8 of us, of varying standards, and it was precisely the kind of interaction that you mention - although we did cover a lot of material, the tutor was happy to be sidetracked into indivudual queries and make it relevant to each person's needs and we all helped each other based on our discovereis and experience so far. I certainly came away feeling exhausted, but I felt much more confident about exploring different tools that I hadn't even known existed before.

This isn't a sales pitch, by the way, just my impressions! Maybe you could try some of the webinars that are advertised?

Good luck, whatever - it really is worth persevering.


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 18:57
Member
French to English
+ ...
TOPIC STARTER
Brackets (contd.) Mar 2, 2009

Claire, I've just done the experiment myself, and pasting

PlaceableContains={}

into PB certainly works for me for transferring codes like {256}

...always providing, of course, the brackets are not isolated from the code by spaces (which i don't think they usually are)

If this fails to work for you, then I can only assume you must have some other conflicting command in your PB


Direct link Reply with quote
 

Claire Cox
United Kingdom
Local time: 17:57
French to English
+ ...
Brackets yet again Mar 3, 2009

Thanks Tony,

It definitely doesn't work for me - no idea why not. It never has, not with any of the different upgrades. I've experimented with various versions of "Placeable contains or between" but to no avail. It's not a huge problem as I usually copy the whole source segement in such cases and then overwrite the rest. It would be interesting to know why though.

I appreciate you experimenting on my behalf though!

All the best,

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 »

Curious glossary behaviour

Advanced search


Translation news related to Wordfast





LSP.expert
You’re a freelance translator? LSP.expert helps you manage your daily translation jobs. It’s easy, fast and secure.

How about you start tracking translation jobs and sending invoices in minutes? You can also manage your clients and generate reports about your business activities. So you always keep a clear view on your planning, AND you get a free 30 day trial period!

More info »
Déjà Vu X3
Try it, Love it

Find out why Déjà Vu is today the most flexible, customizable and user-friendly tool on the market. See the brand new features in action: *Completely redesigned user interface *Live Preview *Inline spell checking *Inline

More info »



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