Hacker News Reader

Comment by nonsince | Hacker News Reader

Where do you sort the Dutch "ij"? The obvious answer is "between 'ii' and 'ik'" but it's actually just the print representation of "ÿ", a letter that is essentially only used in freehand nowadays. So, do you sort it in place of "ÿ", when in the Dutch locale? What if you have a borrowed word that happens to have "ij" in it, like "hijack" (which really is used in Dutch)? In practice, you sort it as if it's English (i.e. between "ii" and "ik"), but that leads to confusion because when capitalising you treat it as a single letter. The titlecase form of "ijzer" (iron) is "IJzer". I guess the correct way would probably be whatever they do in dictionaries and phone books, but I'm an expat so I have no idea what ordering those use.

View on HN

"ij" is usually sorted between "ij" and "ik" by Dutch dictionaries. The odd exception (even for Dutch speakers it seams odd) are phonebooks, because for historic reasons the surname of family members is sometimes written with "y" and sometimes with "ij". Titlecase (IJzer) is not really a problem, because there are no loan words that start with "ij" (like hijack). Very, very rarely "ij" is treated as a ligature, and counted as one letter. So, for the most part you should follow the English sorting rules, and you need one exception for handeling title case. Unicode contains a ligature "ij", which should basically not be used.

English has simply abandoned the letters thorn and eth (and others), which was largely the result of the rise of printing (IIRC).

It's interesting to me that this happened far enough back that the only time we tend to consider 'th' to be letterlike is when we need to explain pronunciation — e.g. when reaching a kid to read

Based on other cases, I guess the "unicode-style" solution would be to add a new codepoint DUTCH CHARACTER IJ OR ÿ which is cased and sorted accordingly.

Of course then you'd have to teach people to use that character when writing instead of just typing "ij". And one day, there will be someone who, for stylistic reasons, needs tight control over when it's rendered as "ij" and when as "ÿ"...

It already exist: U+0132 and U+0133.

Her is a link with some nice information: http://www.uazone.org/multiling/euroml/annex02.html

Wikipedia claims that Dutch dictionaries sort 'ij' as if it were two characters, but telephone directories sort as if it were 'y'.

Based on precedent, I'd expect the Unicode Consortium to declare it the radical of a Chinese character.

(it does exist in Unicode, but as a ligature -- U+0132 for majuscule, U+0133 for minuscule -- and decomposes to the two code points for 'i' and 'j', with use of the ligature discouraged)

Reply on news.ycombinator.com