Something that many programmers have to do, consciously and subconsciously, is think abstractly.  Some have defined intelligence as the ability to think or reason abstractly.  Abstraction occurs from specification design, all the way to the actual code construction.

I bet many of us have run into some kind of problem in a program where we realise that perhaps one set of data was incorrectly or unnecessarily related to another.  Sometimes the reasons for this are related to a lack of understanding of the data that's being dealt with, sometimes it's just oversight.

Something I see happening all the time is the first problem: lack of understanding.  This presents itself very often as text encoding problems: “I just want the standard 8-bit ASCII!” is heard often.  The easy solution is giving someone a quick primer in Unicode and different encodings.

However, if someone grew up in English, and only uses English, their thoughts regarding the abstraction of language versus text can be quite limited.  Perhaps they took a year or two of Spanish or other similar language, so they know that grammar structures can change around.  But even with Western European languages, the relation of written versus spoken language is somewhat similar -- at least there is a letter-based alphabet.

I think it should be mandatory for students to learn another alphabet.  It's not needed that they understand a language behind it.  Simply writing English in a foreign script can be a great mental excercise.  Abstracting written language from alphabets is a good thing to know of. 

Also, I believe that anyone learning another script or language should do so not only on paper, but use a computer with different inputs configured.  Being able to read and write isn't too useful when you're stuck on a computer and you don't know how to use the IME.  I can't remember when I used a pen last (except my digitizer).  And who is going to have paper pen-pals?  Nowadays, it's easier and more fun to get online IM-pals or email-pals.

A simple example is my Chinese Hangman program.  In Hangman, I'd be tempted to take the incoming keystroke and add that to the guess -- one letter at a time, just like the paper game.  In concept, that works fine for Chinese -- a one character guess.  In practise, the problem is that to get that character, many keystrokes or perhaps even characters could be written.  For me, I use the Korean word 과일 (Gwa-il) and then convert to Hanja (Chinese characters).  My keystrokes are: [Right Alt][r][h][k][Right Ctrl][2].  The right alt switches to Hangeul, rhk are: ㄱ ㅗ ㅏ, which combine to form 과.  Right control tells the IME to list Chinese characters for words with the current syllable, and 2 is the number from the list that corresponds to fruit.  The end result: 果.  Note to everyone who is trying to grab control keys and stop their normal usage for some funky functionality in their own app: You're screwing with someone's input in a very annoying way.

In less two weeks, someone can learn a simple phonetic alphabet and how to use an IME.  At least well enough to type a few simple things in, and get a feel for how input might be entered.  However, the lessons learned are going to be there adding another automatic “what if...” case while coding or designing, and hopefully avoid some flaw.