Thoughts on character classification
Most of what follows is the product of several discussions I have had over the past few weeks with Naxi scholar Dr Li Jingsheng 李静生, former researcher at the Dongba Culture Research Centre here in Lijiang.
When we think of Chinese character classification, I guess it’s normal to think mostly of the preface to Xu Shen’s 许慎 Shuowen Jiezi 说文解字.
Also during the Eastern Han however, Ban Gu 班固 (in the Han Shu 汉书) and Zheng Zhong 郑众 (in the Zhou Li 周礼) laid out their own (very similar) categories for Chinese characters. Interestingly, where Ban Gu and Zheng Zhong put the first category as ‘pictographic’ 象形, Xu Shen’s first category is that of the ‘simple ideogram’ 指事, such as one, 一, or up, 上. In which order you categorise the characters reflects, at least in part, how you perceive the script to have developed over time.
Sinologist extraordinaire (国学大师) Zhang Taiyan 章太炎 believed that Xu Shen was correct in his order, with such numerals as 一 being a simple extension of tying ropes in knots to remember stuff.
He qualified his belief by saying (and I paraphrase) ‘the thought processes of all human beings progress from the simple to the complex’. Looking at oracle characters and present day equivalents, we can clearly see how ‘horse’ 
has become 马 and ‘tiger’ 
虎, simplifying over time.
However ‘one’ is already a relatively abstract idea, and there is a case for suggesting that Ban Gu’s order – pictographic first – better reflects the development of the script. If we look at the Naxi Dongba scriptures, we can see that there are indeed logographs for numerals, but there are examples in earlier scriptures where numbers are represented by repeated use of the noun in question. There are quite a few examples of this, but for reasons of brevity I’m only going to detail one here. In a Shu [Naxi god of nature] sacrificial text we find the following panel, which means ‘four door-guarding ghosts’, where ‘four’ is represented by repetition, not a simple ideogram.
Indeed, evidence from Naxi scriptures seems to suggest that representation of numbers developed first from basic pictographic representation, to an abstract numeral. I’m not sure if there is any evidence of this in the oracle characters, but I’d be very interested to find out. Anyone?
I’m not sure this is quite what you’re looking for, but Terry Jones’ The Story of One might make an amusing distraction.
Discussions of this sort get bogged down if one isn’t clear from the start about the distinction between words of a spoken language on the one hand, and written representations of them on the other. Furthermore, we have to distinguish writing — written representations of specific utterances — from other graphic forms, like drawings.
Numbers may or may not be difficult abstract concepts, but that’s irrelevant to writing. By the time the Chinese invented writing, there was certainly already a word for ‘one’ in Chinese. Numerical concepts preceded the invention of writing by many thousands of years. The issues then is how the WORD ‘one’ was represented in writing, not the development of the abstract concept.
In pre-literate cultures, it is normal to represent numbers by repeating a drawing of something, or using an equal number of some object. But at the point where spoken language is being written, all that is necessary is to come up with a conventional representation of the words that already exist in the language.
As we know from the development of writing in the four instances in which it was created from scratch — Sumerian cuneiform, Egyptian hieroglyphs, Chinese characters, and Mayan hieroglyphs — the earliest written forms are derived from pictographs (whether “zhishi” type or “xiangxing” type). We see representations of words for numbers and words for physical objects appearing together on the earliest written documents. The next step was the discovery of the rebus principle, which permitted all words (including grammatical words) to be represented.
The oracle bones reveal a fully developed writing system, capable of representing any contemporary Chinese utterance. Number words are therefore written with distinct characters, just as all the other words are.
Xu Shen and the other Han dynasty philologists were classifying characters according to the structures they saw in their contemporary writing system. There is no reason to think that their classifications are relevant to, or had any insight into, the formative period of the script, of which they were completely ignorant.
Sima, that’s a neat documentary. Thanks for the link!
Zev, many thanks for clearing up the issues involved.
Some Naxi scholars may take issue with your statement “We see representations of words for numbers and words for physical objects appearing together on the earliest written documents” – as far as Naxi is concerned.
This is because they point to evidence of repetition being used in early texts (in combination with examples of the rebus principle), replaced by distinct numerals in later ones. From Li Jingsheng’s 《纳西东巴文字概论》 (云南民族出版社, 2009) : 数词在东巴文中有由象形文字向指事字发展的趋势。假如意音性质的文字有一个一般发展规律的话，我们从东巴文的这些例子中可知原始汉字象形字的产生应早于指事字。 p20
Personally, I’m not convinced by the ‘evidence’, as it could even be reduced to being a stylistic issue, and things are made more fuzzy by the lack of a concrete timeframe for the development of the Naxi scripts and the texts themselves.
It’s my understanding — which I could very well be wrong about, as I have a very limited exposure to Dongba — that early Naxi texts aren’t exactly writing as we would define it linguistically, in that it did not specify precise utterances, but rather gave a general outline for what would be said. Have you seen anything that would suggest that that might be true?
Fang Guoyu 方国瑜 specifies three kinds of scripture: 1) where characters are used as mnemonic aids (以字记忆), 2) where characters are used to replace whole sentences (以字代句), and 3) where charaters are used for words (以字代词).
In the third case, each spoken word is represented by a character, but loanwords are extensively used. Fang Guoyu doesn’t provide exact dates for the scriptures he uses as examples for each category, however the examples he uses for category three are both listed as twentieth century sources. I would hesitantly agree, as you say, that the earliest Naxi texts all belong to the first category.
There are examples of secular Naxi texts which are linguistically ‘complete’, such as diary entries and accounts, but these all date from the twentieth century also. This may be because no importance was attached to such secular texts and thus they were not preserved.