Hi David,

On this issue try running a word count batch task to refresh this statistic.

I'd be interested to hear about all the other bugs too.

Regards

Paul

Thanks Paul

I did this but got an error message.

Other problems I've had recently include;

Object reference not set to an instance of the object

Enumeration not finished ....or something like that

Concordance doesn’t work on Swedish characters treats ä as a for example

F3 doesn't work on .target

I have problems with Studio not opening the correct memory too.

I will keep a log in future as my memory is not as good as a Studio memory! :)

Hello David,

These first two sound like file specific issues so I'd be happy to take a look if you still have the files?

Concordance...do you get a result like this?

I can search for "Wäs" and get an 88% on "Was" or "Was" and get an 88% on "Wäs".  This is deliberate and as expected.  What do you see that makes you think this is incorrect?

F3 will search source from source, or target from target, and both these work fine for me.  Other combinations might be:

Ctrl+F3 : Always search source

Ctrl+Shift+F3 : Always search target

On Studio opening the incorrect memory... I'm happy to take a look through a webex as this is likely to be the result of a TM being in your settings that is being picked up.  Certainly Studio will only pick up what you tell it so I think it will be something simple if you can show me what's happening?

Regards

Paul

Re concordance: why would it be deliberate Paul? It's a completely different word with a different meaning? How is that correct?

For example, if you searched for 'well' and the program returned 'wall', how is that of use?

Here is an example: I've searched for väl while means 'well' and the program returns both väl and val which means 'selection'.

(just for clarification in Swedish, ä is a totally different letter to a)

Sometimes you can get several hits with the 'wrong' word and not the word you want at all.

I get problems with the memories when I switch projects. I like to vary my work. In my experience Studio sometimes opens the wrong memory, i.e. the last one used rather than the one associated with the project.

My F3 does not work on target. I will try and customize this later.

I will try and contact you the next time I get one of the other error message.

Regards

Re concordance: why would it be deliberate Paul? It's a completely different word with a different meaning? How is that correct?

For example, if you searched for 'well' and the program returned 'wall', how is that of use?

Here is an example: I've searched for väl while means 'well' and the program returns both väl and val which means 'selection'.

(just for clarification in Swedish, ä is a totally different letter to a)

Sometimes you can get several hits with the 'wrong' word and not the word you want at all.

I get problems with the memories when I switch projects. I like to vary my work. In my experience Studio sometimes opens the wrong memory, i.e. the last one used rather than the one associated with the project.

My F3 does not work on target. I will try and customize this later.

I will try and contact you the next time I get one of the other error message.

Regards

Could I just add that in my example 'väl' could be a prefix or part or another word that is useful to me. These words will appear below hits for 'val' in terms of percentage.

Hi David,

This is deliberate because this feature, which is a strongpoint of concordance in Studio or Trados, allows you to find potential useful phrases even when they are mispelled.  If you are searching for single words like this then of course, assuming they are spelt correctly the result won't be much help.  But the scoring does help to discern things as they are presented in scoring order.

The way it works is that Studio "normalizes" a string character-by-character, which basically means that characters which have diacritical marks (accents, tremas, carons, and other "decorative elements") are mapped to the base character.  For example, your Swedish "a umlaut" is mapped to plain "a".  This is done so that even if you search for "a umlaut", you still get hits on good search results without the diacritical mark, or the other way around.  The purpose is to increase recall (leverage) and is something many users expect.

The other point is that concordance often produces better, or more useful results when you search for phrases as this may put more context around what you are looking for.  Certainly concordance is more than a simple word search and it is designed to be this way.

Regards

Paul

I'm sorry Paul in the case of Swedish, it simply does not apply. The Swedish ä is a different letter letter altogether- a is not the base character just because it looks similar. If in English you were to search for 'hit' and get 'hat' or 'hot', how on earth does that help you in any context whatsoever? It is not a strongpoint, it is a failing, and extremely irritating.

Regards

David

Hi David,

I'm stumped then.  I would have thought that had you searched for 'hit' and you had none in your TM then the fuzzy value would tell you immediately there were none.  So if you then didn't even want to see the fuzzies why not increase the percentage value used for concordance search and then you won't get them?

I understand they are of no value in this case, but even in your screenshot they are not presented as a 100% match.

What am I missing here?

Regards

Paul

Hi again,

On reflection, perhaps you are thinking the match value refers to the entire segment rather than the highlighted word itself?  This is not the case because concordance is not doing the same thing as TM lookup.

Apologies if you're clear on that one... I was just wondering so thought I'd check just in case.

Regards

Paul