Wednesday, 7 March 2012

如果經過不要錯過 ! (Well, I'm Impressed Anyways)

Recently, my wife found and installed on my 'iPhone' a small application that allows me to enter text into various applications (email, instant message, Facebook) in a variety of type faces.  Amongst the available options are Cyrillic (don't speak Russian), accented Latin characters (très utile si vous avez un ami qui parle français), and most amazingly, full-form Chinese characters.

I'm old enough to remember the days when Asian character sets for products like Aldous (now, Adobe software, I think) PageMaker had all sorts of problems using "double-byte" characters to display properly.  Later, Windows OS allowed for alternate character sets using more or less the same platform.

Entering the text required one to enter a phonetic spelling called "pin yin", which resulted in the display of various character choices, and then to select the desired character.  It was slow, cumbersome, and often resulted in text that would get confused when downloaded to your printer, which had to decode the non-native characters.

Needless to say, they were not "True Type" fonts.  (If you're younger than 30, look it up).

Today, my phone allows such an approach, of course.

The really amazing part, to a tech dinosaur like me, anyways, is that one can actually enter the characters by tracing them onto the touch-screen of the phone.  The software will attempt to recognise the word you've "written" with finger tip or stylus, and then embed it.

That in itself is impressive - there are literally thousands of Chinese characters, many with tiny differences between them (for example compare the two words 精 and 情.  The left ("jing") and the right ("qing) mean totally different, unrelated things).  The software does not always get them right, of course.  But I find it is far better than the handwriting recognition software that I encountered on my Palm Pilot a decade ago, which could not reliably detect the difference between the number "4" and the letter "P".

(confession: I've handwriting that fairly could be called "awful.")

I am not sure how this has been accomplished - a simple read is that the fuzzy logic has improved enormously.  Additionally, Chinese characters, unlike English letters, are comprised of small pieces, and there is a logic as well as a frequency/pattern in their use and construction.  For example, the quite complicated word for "bay" ("wan") 灣 is comprised of not fewer than five larger bits - the "water" radical "shui" (氺) represented as the three lines to the left, two instances of the word "mi" (糸) - threads - that "tie together" the character "yan" ( 言) the character "yan" (words), all mounted above the word "gong" (弓), or "bow."


灣 = 氺 + 糸言糸 + 弓

I am not sure that that is how native speakers learnt to read and write, but as a foreigner who is strongly influenced by patterns (my day job is making mathematical models), it made sense to me as I sat with my 遠東漢英/英漢詞典 ("Far Eastern Chinese-English/English-Chinese Dictionary) twenty years ago.

It's an impressive marriage of new technology and a 4000 year old cultural artefact.

Post a Comment