Wordfast mixes up numbers in TM matches Thread poster: David Umpleby
|
|---|
I work with Wordfast 5.0n and Word 2000 under XP. When translating a text from DE to EN, such as,
"Die Färbemittel enthalten sowohl die Komponenten A1 als auch die Komponenten B1 bevorzugt in einer Menge von 0,005 bis 10 Gew.-%, vorzugsweise von 0,1 bis 5 Gew.-%."
Wordfast can offer partial TM hits (>75%) or 100% matches as follows with any of the mistakes):
(First is correct)
The dyes comprise both the components A1 and also the components B1 preferably in an am... See more I work with Wordfast 5.0n and Word 2000 under XP. When translating a text from DE to EN, such as,
"Die Färbemittel enthalten sowohl die Komponenten A1 als auch die Komponenten B1 bevorzugt in einer Menge von 0,005 bis 10 Gew.-%, vorzugsweise von 0,1 bis 5 Gew.-%."
Wordfast can offer partial TM hits (>75%) or 100% matches as follows with any of the mistakes):
(First is correct)
The dyes comprise both the components A1 and also the components B1 preferably in an amount of from 0.005 to 10% by weight, preferably from 0.1 to 5% by weight.
(Second: retains commas in numbers)
The dyes comprise both the components A1 and also the components B1 preferably in an amount of from 0,005 to 10% by weight, preferably from 0,1 to 5% by weight.
(Third: may or may not retain commas in numbers and mixes numbers around)
The dyes comprise both the components A1 and also the components B1 preferably in an amount of from 10% to 0.005 by weight, preferably from 0,1 to 5% by weight.
Another mistake is that Wordfast may or may not retain superscripts or subscripts in partial or 100% TM matches.
As I translate a lot of German chemical texts, this is frustrating...
Does anyone have a suggestion to correct this problem? ▲ Collapse
| | | |
Gerard de Noord 
France
Local time: 20:35
Member (2003)
English to Dutch
+ ...
| You may not like this but | Nov 18, 2007 |
Hi David,
You may not like this but Wordfast doesn't retain superscript and subscript in the TM. The 2 in H2O will be a normal 2 in the TM and will be returned without formatting.
Isolated numbers aren't entered in the TM as such, they will be replaced by placeholders. Wordfast does a rather smart triangular comparison to try to match the source segment. The results are surprisingly good for normal translations but I can understand the limitations of Wordfast when trans... See more Hi David,
You may not like this but Wordfast doesn't retain superscript and subscript in the TM. The 2 in H2O will be a normal 2 in the TM and will be returned without formatting.
Isolated numbers aren't entered in the TM as such, they will be replaced by placeholders. Wordfast does a rather smart triangular comparison to try to match the source segment. The results are surprisingly good for normal translations but I can understand the limitations of Wordfast when translating chemical or mathimatical texts.
False 100% matches against a green background should always be reported to Yves. He's very serious about them. But in your profession you should always enable DifferentNumbers=99 in PB.
Regards,
Gerard ▲ Collapse
| | | |
Heinrich Pesch
Finland
Local time: 21:35
Member (2003)
Finnish to German
+ ...
| Cannot expect too much | Nov 19, 2007 |
Formatting is not preserved in Wf-TM. Also bold, italics etc. are lost. One thing you must keep in mind is additional spaces. If the source text says "95m" instead of "95 m" and you add a leading space to the unit as it should be, Wf will produce "95mm" next time when it comes to this occasion and show green background and 100% match.
I don't know what Trados would do, but Trados-TM preserves formatting, so you should try it out, or one of other more expensive tools like DV, SDLX or Transi... See more Formatting is not preserved in Wf-TM. Also bold, italics etc. are lost. One thing you must keep in mind is additional spaces. If the source text says "95m" instead of "95 m" and you add a leading space to the unit as it should be, Wf will produce "95mm" next time when it comes to this occasion and show green background and 100% match.
I don't know what Trados would do, but Trados-TM preserves formatting, so you should try it out, or one of other more expensive tools like DV, SDLX or Transit.
Cheers
Heinrich ▲ Collapse
| | | |
Samuel Murray
Netherlands
Local time: 20:35
Member (2006)
English to Afrikaans
+ ...
| This and that about Wordfast | Nov 19, 2007 |
David Umpleby wrote:
Wordfast can offer partial TM hits (>75%) or 100% matches as follows with any of the mistakes):
(Second: retains commas in numbers)
My WF never recognises commas in numbers. Could it be something to do with your German MS Word (I don't mean having a German GUI but it being a version intended for use in German)? I use some kind of English MS Word.
(Third: may or may not retain commas in numbers and mixes numbers around)
The mixing of numbers also happen to me. That is why I have the following enabled in PB:
DifferentCase=99
DifferentNumbers=99
Another mistake is that Wordfast may or may not retain superscripts or subscripts in partial or 100% TM matches.
Superscript and subscript are font formatting issues, like bold, underline, etc. The WF memory doesn't remember formatting. WF gives the target segment the same formatting as the first character of the source segment.
As I translate a lot of German chemical texts, this is frustrating... Does anyone have a suggestion to correct this problem?
Well, WF regards any word with numbers in it as a placeable, and when copying placeables, WF retains formatting. So try grabbing the h2o using Ctrl+Al+left or +right, and then copying it to the cursor position using Ctrl+Alt+down.
I use placeable copying an awful lot... also useful for glossary work and software translation of nontagged text.
| | |
|
|
|
Samuel Murray
Netherlands
Local time: 20:35
Member (2006)
English to Afrikaans
+ ...
| Numbers in the TM | Nov 19, 2007 |
Gerard de Noord wrote:
Isolated numbers aren't entered in the TM as such, they will be replaced by placeholders.
One might think this, but no... WF does in fact put isolated numbers in the TM (if you mean isolated in a sentence). Then... WF just regards numbers in the TM as placeholders, so to speak. Open up a TM and see for yourself.
| | | |
Samuel Murray
Netherlands
Local time: 20:35
Member (2006)
English to Afrikaans
+ ...
| WF and numbers inside "words" | Nov 19, 2007 |
Heinrich Pesch wrote:
If the source text says "95m" instead of "95 m" and you add a leading space to the unit as it should be, Wf will produce "95mm" next time when it comes to this occasion...
Yep, those spaces can be annoying. Another weird (but entirely expected) behaviour of WF's number matching attempt I encountered recently was this:
SL: Today is the 18th of November 2007.
TL: Vandag is 18 November 2007.
SL: Today is the 19th of November 2007.
WF proposes 75% match: Vandag is 19thovember 2007.
Thankfully in this case both "18" and "November" are placeables (as I have Placeable=FirstCap in PB), otherwise it'd drive me insane!
| | | |
In one of my TMs, I get "Table 2" as a 100% match for Table + any number, however often I correct it (yes, I do have DifferentNumbers=99 enabled). As I know about the problem I just make sure I look out for it when I use that TM, but should I report it to Yves, as Gerard says?
| | | |
Heinrich Pesch
Finland
Local time: 21:35
Member (2003)
Finnish to German
+ ...
| Just for the record... | Nov 20, 2007 |
You can save a lot of work if you preformat your text prior to translation. Then you can avoid misformatting by Wf in many cases. Missing leading spaces with units are better fixed befor you start translation, also listings (bullets).
| | | |
Login to reply/comment 
Translation news related to Wordfast
