Tuesday, October 17, 2006

Primordial Alphabet Soup

I was listening to an interesting story on the radio tonight about an effort in India to produce keyboards that can type in Hindi. (Gosh, that's exciting! Tell us more!)

The problem is that Hindi is a phonetically written language. The "alphabet" of Hindi consists of over 100 symbols. Urdu is similar. To type in Hindi, you must either invent a very large keyboard, or learn a very complex style of typing which involves multiple keys per symbol.

Pictographic/ideographic writing styles are even worse, with literally over 10,000 symbols (many a complex combination of a base of symbols that numbers merely in the 100s to thousands). Now we're talking a really large keyboard. No laptops for these folks.

When writing by hand and ink was in use, this may not have mattered much. I assume that writing speeds are relatively similar across writing systems. (Although I may be wrong, I don't think a Japanese writer takes significantly longer to write "Can you read me now?" than an English writer would).

But with the advent of computers, this is changing. Computers represent an environmental evolution selection mechanism, weeding out written languages which require "too many" symbols. Inefficient writing styles (at least for the purposes of producing input that a computer can process) are weeded out of the population, with only the most efficient remaining. This is why most of the countries with less efficient symbolic representations are adopting "standard" keyboard sizes of a little over 100 keys.

In attempts to further reduce keyboard sizes and increase portability, there are a number of approaches. Most involve either "chording" (the simultaneous pressing of multiple keys) or the familiar "texting" approach on western cell phones, where each key represents multiple symbols that you reach by multiple clicks of the key. Neither is very consumer friendly.

If there were a language that had a written alphabet of only 12 characters (the number of keys on a standard cell phone keypad), I bet it would win the race. (Most languages, like English, have 40-50 phonemes, so 12 is really pushing it).

I wondered if whether advanced speech recognition in computers could eventually nullify the alphabet race. In theory, native speakers could dictate their text and the computer would handle the symbols. Like writing speeds, most speaking speeds are within the same order of magnitude of each other, conveying information at a similar rate. There would no longer be the evolutionary range of variation to make selection meaningful.

But on reflection, I don't think speech recognition will put the alphabet evolution on hold. There are too many ways where keying information into a device is more efficient, or at least more circumspect. And as long as there are keyboards, the alphabet evolution will continue.

May the most efficient alphabet win.

0 Comments:

Post a Comment

<< Home