On my wife’s Swedish side there are many surnames that I entered using ‘special characters’, as was correct. Unfortunately they only come up when searching the index when you input them with the correct special character and do not come up using the character without the accent. To make matters worse many ceased using the special characters a few generations ago which makes it a real pain sometimes (wish I’d never bothered!). Example screenshot below.
Is it possible to have an option to ignore accents when searching the index, please, so ‘Stahl’ and ‘Ståhl’ for example both come up when searching ‘Stahl’?
Until there is such a feature, try using the wildcard characters: underscore for 1, percent for any number. For example, “st_hl”. Of course it’s not ideal as it can return results other than those you want.
For this one, the å is not really an accented letter. Rather, it is a letter in its own right. But RM uses a special collating and comparing sequence that treats the å as if it were an a. So in this one special case, RM already does what you want. Whether treating the å as if it were an a for collating and comparing is a good idea or not is an interesting question without a very good answer. However, RM does not in general ignore accents for sorting and searching, especially when the accents are really accents rather than being marks that define a separate letter in its own right.
Sorting and searching in languages other than English is an extremely non-trivial issue, and I’m quite certain I don’t really have very many of the answers as to how it ought to be done in a perfect world. For example, my understanding is that French treats all its accents and cedillas as accents and cedillas and not as different characters. I have no idea what that means for things like sorting and searching French names, nor what that means for things like spell checking dictionaries in French. Also, I have been told that phone book sorting in Quebec is different than phone book sorting in France, so there can be a cultural component to this stuff in addition to there being a language component. (I don’t know if this difference in phone book sorting between Quebec and France is really true, or if it’s just an urban legend of some sort.)
My favorite and most confounding example of sorting and searching problems is the letter å in Norwegian names. In the typewriter days, typewriters in Norway often did not include the letter å. So it was typed as a double a, viz. aa. So the surname of Våge might be rendered as Vaage. To a Norwegian, Vaage has only four letters not five, with the aa representing a single letter. The confounding part is that to this day, some people with that surname will prefer and demand that their surname be written as Vaage. Software then has to know when to render the name as Vaage and when as Våga, but the same software has to search and sort the name exactly the same no matter which way it is rendered.
Actually, I had not seen this article. It’s amazingly comprehensive. I was already familiar with many of the issues, but there are many more issues in the article that I was not familiar with.
The trouble is, I don’t now how the issues and their solutions from the article translate into actual databases in the real world. For example, I wouldn’t have a clue how to make the same genealogy Web site make sorting and searching for a German user work according to German collation and for a French user according to French collation. I understand the concepts, but I have not a clue how to do that in the real world.
