Apache OpenOffice (AOO) Bugzilla – Issue 59054
Correct Dutch contractions are wrongly corrected - smarter spell checking needed?
Last modified: 2013-02-24 20:43:10 UTC
In Dutch, combinations of two nouns must be written as one word (contraction). For instance: - 'spell checking' translates as 'spellingcontrole' - 'project analysis' translates as 'projectanalyse' A large number of these contractions (ie. 'spellingcontrole') are in the language file. No problems there. However, some contractions (ie. 'projectanalyse') are not in the language file, yet they are correct Dutch words. When you write 'projectanalyse' in OOo, it is marked as incorrect. In fact, OOo suggests the incorrect 'project analyse'. Compared to MS Word: The spell check of MS Word seems to 'understand' these Dutch rules and allows certain unknown contractions. When you write 'projectanalys' in Word, it is marked as incorrect, but Word has no suggestions. This proves that Word does not know the word 'projectanalyse'. However, when you write 'projectanalyse' in Word, it is recognized as a correct word. This smart way of dealing with unknown contractions would be a welcome enhancement to the OOo Dutch spell check.
Need a dictionary development effort to handle Dutch compounds with Hunspell, new spell checker of OOo (http://hunspell.sourceforge.net). Björn Jacke has experience with German compounds, I have with Hungarian ones.
Hi Németh, Count me in! Is there a forum or mailinglist where Hunspell can be discussed in English (the Hunspell forums listed on sourceforge are less than lively...)? Best regards, Simon.
Hi Simon! I think, we can use this issue for discussion. BTW. Please, take also a look to the Issue 58558 (my belated Christmas surprise for the Dutch :) ). If you can add detailed description about Dutch discretionary hyphenation, we will made better Dutch hyphenation for OOo. Best regards, Laci
I will study the Hunspell documentation to see how this can be implemented for Dutch. By the way, I understand that Hunspell supports the NOSPLITSUGS flag in the affix file. This was in fact conceived because many Dutch users would accept a suggestion such as "project analyse" for "projectanalyse", while in fact the latter word was correct (just not present in the dictionary). Unfortunately this was only implemented the Myspell of OOo 1.1 but not in that of OOo 2.0.
Nemeth->Simonbr: Oh, I have already understood the NOSPLITSUGS... :) Thanks.