Here an article I wrote as it appeared in the April/May 1999 issue of MultiLingual Computing:

Top 10 Internationalization Errors

1. Assuming that all letters lie between a-z or A-Z.  (For example, the Danish alphabet is abcdefghijklmnopqrstuvwxyzæøå.)
2. Assuming that all languages only use one character to represent a letter.  (Spanish uses "ch" and "ll" as two distinct letters - for example, "llama" is pronounced "yama", and collates between "loma" and "mañana".)
3. Assuming that a letter only represents one character.  (German sorts "ß" as "ss".)
4. Assuming that all characters can be converted to upper or lower case.  (Chinese, Japanese and Korean characters do not have the concept of case.)
5. Assuming that words are separated by spaces.  (words in most Asian languages are not usually separated.)
6. Assuming that sentences are read from left to right, then top to bottom.  (Japanese is usually, but not always, written from top to bottom, then right to left.)
7. Assuming that sentences are read in only one direction.  (Arabic has a bi-directional writing system, in which sentences are mostly written from right to left, with the exception that numbers are written from left to right.)
8. Assuming that there are twelve months in a year.  (The Israelis have an thirteenth month every leap year, and their months have only 29 to 30 days.)
9. Assuming that the year 1998 means the same thing to everyone.  (India and Thailand's calendars number from Buddha's birth. Right now, in Japan, it is the year 10 Heisei; in Moslem countries it is the year 1418, and it is the year 5759 according to the Hebrew calendar.)
10. Assuming that time zones are multiples of one hour apart.  (Newfoundland, Canada, has a time zone which is half an hour different from the mainland, and Guyana's time zone is offset by 45 minutes from its neighbors.)