Our emotion-conveying technology

 kw: book reviews, nonfiction, writing systems, scripts, philology

The study of writing is not really the study of language; so says Silvia Ferrara in The Great Invention: A History of the World in Nine Mysterious Scripts. The book was translated from Italian by Todd Portnowitz. The author leads a collaboration, INSCRIBE, for Invention of SCRIpts and their BEginnings. They use a synergistic stockpile of methods, including deep learning, to tackle the remaining undeciphered scripts of the world.

My next-in-age brother and I, when very young, were enamored by a Table of Alphabets in the family encyclopedia. We have both retained interest in the subject, though he took it much farther than I, becoming a professional calligrapher in several script traditions including Japanese and Egyptian. As a job, that doesn't pay well, so he became a professor of art history. I went the IT direction, and retained the interest as an occasional hobby. I have always been fascinated by the multitude of non-Latin scripts. The endpapers of Gideon Bibles have John 3:16 translated into many languages, primarily in non-Latin scripts such as Sinhalese. Here is a smattering of lovely aperiodic crystals that I extracted from Google Translate:

The source statement is a bit hyped, but it refers to something in the last chapter of Greatest Invention: while lists of "great inventions" based on surveys with many people have "writing" some 30 places down the list (making the book's title rather ironic), several great inventions further up the list would not have been possible without writing. Yet the author has shown that writing really wasn't a necessity for carrying on a civil society, and several cultures came and went without having a writing system. Although it occurs to me that perhaps the lack of record-keeping doomed those that vanished. And not all "writing" can be committed to paper or papyrus (or clay tablets): the Peruvian quipu, or knotted string, recording system seems to be a full-featured "script". It's like a 3-dimensional alphabet.

In the translated texts above, only the latter 3 are alphabetic. Mandarin is not purely ideographic as is commonly thought, but is a combination of ideographs and phonic elements. Japanese and Korean are similar, but Japanese has two other writing systems that work together with the "square" Chinese-style characters called kanji ("Chinese writing"). These are hiragana and katakana, which are pure syllabaries. In a syllabary a consonant (or two) combines with a vowel, all represented by a single character (glyph). Hiragana characters are used to add inflections (Chinese has no inflections, but Japanese has plenty) and numerous words that have no kanji; katakana characters are used to phonetically spell out non-Japanese words. That makes Japanese the most complex writing system I know about.

Amharic and Gujarati are syllabaries of a style called abugida; this is explained briefly in the book. Hebrew, while it is called an alphabetic script, is semi-syllabic because although a few vowels are included as glyphs, most vowels are either understood in context or are explicitly indicated by dots and other small diacritics attached to consonants. Note that this Hebrew sample includes no diacritics.

The author's particular obsession is the undeciphered scripts. Some were clearly scripts conveying real languages. Others are more mysterious, such as the Voynich Manuscript; a part of one page is shown here. Philologists are still very divided over whether there is a language behind its unique set of characters, or if it is a hoax. The latter view is supported by the many illustrations that portray plants and animals that are not and never were.

I never figured out what the "nine mysterious scripts" are. The author discusses numerous writing systems, including many scripts that have yet to be deciphered; also many that are known, even though they are now defunct. It will be interesting to learn if the INSCRIBE project cracks any of them. Perhaps "deep learning" will help.

By the way, "deep learning" is actually "shallow but very wide learning". If the human mind is like the Mississippi River, deep and powerful, the best "deep learning" AI systems are like the Wind River of Wyoming: half a mile wide, half an inch deep, and prone to flowing uphill when the wind is right.

"Latin is a dead language,
it's plain enough to see.
First it killed the Romans,
and now it's killing me!"

Speaking of dead languages, my brother and I, ages 11 and 8, tried our hand at Babylonian cuneiform. We carved styluses from chopsticks. We used red clay from a stream and let our scribed-on tablets dry in the sun for a couple of days. Then we tried to fire them by putting them in the oven, set to broil. Apparently they weren't dry enough, because they exploded! Curious boys do curious things. Luckily, the oven wasn't damaged.

Whatever the nine scripts really are, this is definitely a book worth reading.