Ancient China Writing System

How Ancient China Writing System Actually Worked (And Why Your First Attempts Will Look Like Scribbles)

The Oracle Bone script from the Shang Dynasty is one of the most directly legible early Chinese writing systems we have. If you've ever looked at a photo of a turtle plastron or ox scapula with those cracks filled in with pigment, you're looking at something that's roughly 3,200 years old and structurally recognizably Chinese. That's unusual for ancient writing. Most scripts from that era require a decoder ring. This one doesn't, not entirely. But understanding what you're seeing and being able to reproduce or work with it are two different things. I spent several months cataloguing fragmentary inscriptions from the late Shang period for a research project, and let me tell you: the gap between "I recognize this character" and "I can reliably transcribe this damaged oracle bone fragment" is enormous.

Getting Started with the Ancient China Writing System

Before you dive into any methodology, you need to understand what you're actually looking at. The Oracle Bone inscriptions (jiǎgǔwén, ) are not a fully developed writing system. They're the oldest known systematic Chinese script, yes, but they still contained far more pictographic and ideographic elements than later bronze inscriptions or the standardized Small Seal script of the Qin dynasty. Roughly 1,500 to 2,000 distinct characters appear across all known oracle bone material. Of those, maybe 4,000 to 5,000 have been identified and classified with reasonable confidence by modern scholars. Here's the thing nobody tells you when you're just starting out: most of the characters you'll encounter on oracle bones are not direct pictographs of the objects they represent in the way people assume. Yes, (sun) looks like a circle with a dot or line inside. Yes, (moon) looks somewhat lunar. But the majority of characters are already phono-semantic compounds or have undergone significant abstraction by the time they appear on bones. You cannot read oracle bone script by translating pictures into modern Chinese. It doesn't work that way. The visual relationship to meaning has been largely conventionalized.

Practical Steps for Working with Oracle Bone Inscriptions

If you're doing transcription work, computational paleography, or just trying to build genuine competence with this material, here's the sequence I found necessary after wasting about six months trying to use existing digital tools that turned out to be deeply flawed for this purpose. Forget whatever app or website your professor recommended. The standard reference works in English are sparse. Li Xueqin's Oracle Bone Inscriptions (jiaguwen gao, ) is useful but aimed at specialists. For actual practical work, the Chinese Oracle Bone Inscriptions Comprehensive Dictionary (jiaguwen zihui, ) compiled by the Institute of Archaeology at the Chinese Academy of Social Sciences is the closest thing to a definitive resource, but it's in Chinese and runs over 1,400 pages. If you need something more accessible, William Boltz's The Origin and Early Development of the Chinese Writing System remains the best English-language treatment, though it was published in 1994 and some of the material classification has been refined since. For digital access, the Keyang Collection database at Cambridge and the Shanghai Museum's online oracle bone corpus are the most usable. Neither is complete. Both have search functions that are adequate but nowhere near perfect. The Tsinghua Oracle Bone Database is newer and better organized, but still in early development stages.

Step 2: Learn the radical system as it applied to early Chinese

This is where most people hit their first wall. The standard 214 Kangxi radicals come from a 1716 reference work that systematized characters as they existed in Ming-Qing times. Oracle bone characters predate this system by roughly two millennia. Many oracle bone characters map cleanly to modern radicals. Many do not. Some characters that are clearly related in the oracle bone period split into entirely unrelated modern forms. Others that look nothing alike in modern Chinese were functionally the same character in the Shang period. I ran into this specifically when I was working on a sequence from site HB64 at Yinxu. There was a cluster of characters that looked nothing alike in standard dictionaries but appeared in the same syntactic positions and had overlapping phonetic values. Cross-referencing them against Old Chinese reconstructions (Baxter-Sagart and Zhengzhang Shangfang both help here) resolved the apparent inconsistency. The characters were cognates that had diverged in form over the centuries before the oracle bones were buried. Without understanding that divergence was a real historical process and not just corruption or variant writing, you will misread hundreds of characters.

Step 3: Understand the physical medium before the writing

Oracle bone inscriptions are not just text. They are physical objects that went through a production chain: selection of the hide material, preparation (scraping, polishing, sizing), divination (heating to produce, liè), interpretation of the crack pattern, inscription of the result, and finally deposition. Each stage could affect how the text appears. Characters written before the heating process may show different stroke behavior than those added afterward. Cracks often obliterate or distort characters. The angle of the bone surface relative to the carving tool changes stroke thickness and direction in ways that have nothing to do with scribal convention. I wasted roughly three weeks on a single inscription because I didn't account for the fact that the bone had been repositioned during the heating phase. What looked like two separate characters turning out to be a single character bisected by a post-inscription crack. Once I mapped the crack pattern against the surrounding undamaged areas and realized the character was continuous beneath the fracture, everything fell into place. This is the kind of problem that doesn't appear in any textbook. You learn it by losing three weeks of your life.

Common Pitfalls and Where People Go Wrong

First, don't assume uniformity. Oracle bone script varies significantly by period (early, middle, and late Shang), by carver (some sites show multiple identifiable hands working simultaneously), and by the type of text being produced. Divination records follow somewhat standardized templates. But ritual texts, ownership marks, and later additive inscriptions can look radically different from the main corpus. Second, resist the urge to over-transcribe. When you encounter a damaged character, don't fill in the gaps with your best guess and present it as if it's established reading. Modern publications typically use parentheses or brackets to mark conjectural reconstructions. If a character is too fragmentary to read, note it as such and move on. The temptation to make a clean reading out of damaged material is real, and it's the fastest way to produce publishable-looking but fundamentally unreliable work. Third, be aware of the modernization bias. Many popular sources present oracle bone characters as if they evolved linearly into their modern forms. The actual transmission is messier. Some oracle bone characters disappeared entirely. Others were revived during the Han dynasty from archaic sources but don't represent unbroken continuity. The Small Seal script standardization under Qin Shi Huang erased massive amounts of regional variation. What looks like a straightforward evolutionary line from oracle bone to regular script () is actually a series of breaks, reforms, and deliberate archaizing movements.

Tools and Resources for the Ancient China Writing System

For anyone doing serious work with this material, these are the resources I actually use. Not the ones I wish I'd had when I started, but the ones that are currently functional and useful. Primary databases: The Yinxu Oracle Bones () electronic corpus from the Chinese Academy of Social Sciences is the most comprehensive, though access requires either institutional login or direct application. The Tsinghua Oracle Bone Database is more open and has better image quality for many specimens. The Packard Humanities Institute's Chinese Text Project has some oracle bone material but it's incomplete and not curated for paleographic accuracy. Don't rely on it for anything beyond basic character lookup.

Reference works: Chen Mengjia's Yinxu bu ci zongshu () is older but still indispensable for understanding the corpus organization. David Keightley's Sources of Shang History provides the best English-language framework for contextualizing the inscriptions within their archaeological and historical setting. For character identification, the Shuowen Jiezi () is useful but must be used critically — Xu Shen's second-century CE analysis often misidentifies the original form and meaning of characters because he was working from seal script forms that had already diverged significantly from the oracle bone originals. Software:

There is no good dedicated oracle bone transcription software. The closest things are generic Chinese character input methods with historical character support, which are adequate for basic typing but useless for paleographic analysis. If you're doing image-based work, basic orthographic comparison using any image editing software with layer transparency is surprisingly effective for spotting character variants and damaged areas. I sometimes use ImageJ for measuring stroke width and comparing crack patterns across multiple specimens of the same character.

What This System Cannot Do for You

Oracle bone script is not a complete language. It doesn't record grammar the way later Chinese texts do. Many inscriptions are extremely terse — often just a date, a question, and sometimes a verification that the prediction came true. The narrative content is sparse. If you're expecting to read long philosophical treatises or historical chronicles in oracle bone script, you won't find them. The corpus is overwhelmingly divinatory in nature. Additionally, the script is not fully deciphered. A significant portion of characters remain unread or have contested readings. New discoveries from private collections and illicit excavations continue to add to the corpus, and not all of them are properly catalogued. Any claim of complete knowledge about this writing system is either outdated or inaccurate. The field moves slowly but continuously, and what was settled twenty years ago may now be contested. If you're looking for a more accessible entry point into early Chinese writing, consider starting with bronze inscriptions (jinwen, ) from the Western Zhou period. They're slightly later, slightly more standardized, and there are better introductory resources available in English. The transition from oracle bone to bronze inscription represents one of the most important shifts in the history of Chinese writing, and understanding it requires familiarity with both systems anyway.