Cracking the Oracle Bones: The Yinxu Oracle Bone Inscription Digital Database (Part 2)

In the second part of our exploration of the Yinxu Oracle Bone Inscription Digital Database 殷墟甲骨文數據庫 (YOD), we’ll continue our walkthrough and assessment of the database’s main features, turning to the “Original Text Search” (原文檢索), “Oracle Graph Input Assistant” (甲骨文輔助輸入), and “Oracle Bone Dictionary” (甲骨字典) functions. As well as discussing the broader implications of these features for future oracle bone studies, I’ll also suggest some practical research scenarios wherein each might be used to its full advantage.

Original Text Search:

While many oracle bone databases permit searches using modern equivalents for oracle bone graphs (Academia Sinica, ChAnT), only the YOD allows the user to input the graphic structures of oracle bone graphs as they appear in inscriptions. The “Original Text Search” feature allows users to filter results by inscription collection, diviner group, and inscriptional themes using a combination of “and” (並且), “or”( 或者), and “not” (非) functions. This new development in digital oracle bone research is facilitated by two other significant innovations unique to the YOD – the oracle bone font (see part one for a discussion) and the “Oracle Graph Input Assistant.”

As the only database with an integrated oracle bone font, the YOD is the only database capable of an original script search function. At the same time, the developers have also drawn on previous scholarship’s attempts to systematize bone graphs in designing their input interface.

The YOD’s purpose-built “Oracle Graph Input Assistant,” accessed by clicking the box to the right of the search field, opens a pop-up window subdivided into three panels – 1. “Classifier” (部首), 2. “Classifier Sub-Character” (部首屬字), and 3. “Facsimile” (摹寫) – which collectively enable identification and input of oracle bone graphs (see below).

Like modern Chinese characters, oracle bone graphs can be broken down and organized by their graphic components. The YOD developers identify and tabulate 163 such components accounting for nearly every graph in the database (excluding 48 graphs listed under “other” (其他), 247 combined graphs, and 13 numerical symbols). This list represents an updated version of Shima Kunio’s 島邦男 systematic graphic component index featured in his 1967 publication, Inkyo Bokuji Sōrui 殷墟卜辞綜類. Like the YOD input assistant, Shima Kunio’s index enabled the reader to locate inscriptions containing a particular graph by looking up a graph’s components in a separate tabulated index at the back. For added convenience the YOD “Classifier” panel also provides quick access to numerical signs, the 22 gangzhi (干支) day names, 48 graphs unaccounted for by the classifier components, and 247 “combined graphs” (合文) – that is, graphs combining two or more graphs into one (e.g. “small bovine” xiao niu.) The second window, the “Classifier Sub-Characters” window, displays bone graphs (or their modern equivalents, if known) containing the component selected from the first window. Finally, the “Facsimile” window displays a selection of variant bone graphs, one of which the user chooses to input into the search field.

To understand how the assistant works, it might be helpful to walk through the stages of a plausible research scenario. Let’s imagine that we are reading a collection of newly published inscriptions (i.e., outside of the database) and we come across a graph we do not recognize  – []. If we want to search the database for this graph, we’ll need to determine its graphic components and cross-reference these with those provided in the “Classifier” window. In this case, we notice our graph is made up of the [] and [] components. Whereas Shima Kunio’s Inkyo Bokuji Sōrui listed oracle bone graphs under each and every one of their graphic components such that the reader could find a graph by looking up any of its components, the YOD indexes each oracle bone graph under just one of its graphic constituents. Since I have not yet been able to determine the conventions the editors used when deciding to list a graph under a specific classifier, I had to use a bit of trial and error to find the classifying component. In any case, checking both [] and [], we discover the [] graph has been assigned to the [] classifier. However, we only discover this by clicking through the graphs listed in the 2. “Classifier Sub-Character” window for [] until we locate the graph under its modern equivalent header – zun 尊, “to venerate; to raise in veneration”.

The bottom-right window displays several variants for zun 尊, one of which closely resembles our original graph. Double-clicking this graph inserts it into the search field. We then have the option to narrow our search using the filters discussed earlier – in this case, we’re interested in all the inscriptions containing our graph, so no filters are necessary.

Overall, the process is relatively intuitive and, as a bonus, provides a quick and convenient way to check whether a bone graph has a known modern equivalent. Note that the “Original Text” search will return only inscriptions containing our specific variant of zun 尊 (i.e. []) as opposed to every inscription containing the word zun 尊. To view the latter, we would need to input zun 尊 into the “Interpretation Search” feature discussed in part 1.

While this should not be taken to mean the “Original Text” search is only useful for researching specific variants, its main applications are probably limited to the following scenarios:

  1. Researching a specific orthographic variant of an oracle bone graph.
  2. Researching an unknown oracle bone graph.
  3. Researching a known oracle bone graph with no known modern equivalent.
  4. Researching a known oracle bone graph with a known, non-standard modern equivalent.

By way of example, let’s take up scenario four above – “researching a known oracle bone graph with a known but irregular modern equivalent.” Suppose we come across the graph [] in a recently published collection of inscriptions and we want to investigate its meaning and the contexts in which it appears. The printed publication provides the transcription , which is not typable into the “Interpretation Search” using input systems currently in use. We can get around this by using the oracle graph input assistant to locate the graph under its designated classifier. Looking at our graph, we notice it has three graphic components we can cross-reference against the classifier window: a cross shape closely resembling the [] classifier, a  [] (bu 不) component, and a [] (shan 山) beneath them. After some trial and error, we locate the graph under [] (shan 山).

We then choose the first oracle bone graph from the three variant forms provided in the lower right window and hit enter. Since we haven’t specified any filters, this will perform a full-database search.

These results reflect all the inscriptions in the database containing our chosen variant. If we were only interested in this specific form of the graph, we might stop here. But what if we want to find every inscription containing the word {}? Since the database does not allow users to copy and paste from the results window or the input assistant into the “Interpretation Search” (see part 1), we must instead open the graph in the bone viewer window by clicking one of the search results:

From here, we can highlight the transcribed graph in the “interpretation” (釋文) field. A pop-up window presents several options: “copy” (複製), “search” (檢索), “note” (筆記), “cite” (引用), and “oracle bone dictionary” (甲骨字典). 

Clicking “search” performs an interpretation search of the highlighted graph in a new window. As you can see below, this time we get five results instead of three.

The added examples are highly fragmentary, but we note that our graph is usually preceded by yu 于 “to/in/at”. At this point, we might wish to consult the in-built oracle bone graph dictionary for some help. To do so, we must return to the bone viewer window, highlight the character, and select “oracle bone dictionary.”

The Oracle-Bone Dictionary:

The dictionary opens in a new tab displaying the modern equivalent (if known), the variant bone graph forms, the Chinese pronunciation (if known), and a definition accompanied by some example inscriptions. According to the dictionary, our graph denotes a “place name” (di ming 地名) of unknown pronunciation. At this point, we can leave off our example scenario and have a closer look at the dictionary (the other scenarios listed above generally follow the stages outlined here).

In addition to the above information, more exhaustive dictionary entries provide (multiple) graphic explanations, as well as interpretations of recurring phrases containing the graph.

The editor of the YOD, Chen Funian 陳福年 has compiled detailed definitions for 3905 graphs – more than enough to develop a basic oracle bone graph vocabulary. When I began research into oracle bones back in February, I often used the built-in dictionary in tandem with the digitized, fully-searchable version of Xu Zhongshu’s 徐中書 “Oracle Bone Graph Dictionary” (甲骨文字典).

Overall, I think the YOD dictionary achieves exactly what it sets out to provide – an invaluable and authoritative resource for oracle bone literacy. The convenience afforded by a quick pop-up dictionary window is hard to overstate and allows for a kind of “learning on the go.” Later on, of course, the user may choose to consult it alongside Matsumaru Michio 松丸道雄 and Takashima Kenichi’s 高嶋謙一 “Synthetic Index for Interpretation of Oracle Bone Inscriptions” (甲骨文字字釋總覽) and Yu Xingwu’s 于省吾 “Jiagu wenzi gulin” (甲骨文字詁林) which summarize the divergent interpretations of various oracle bone scholars.

By providing a workable and intuitive digital interface for bone graph input, the YOD has opened up innovative research avenues into bone graph orthography. At the same time, it’s pretty clear the heuristic and research value of the “Original Text Search” and input assistant features also extend to more general inquiries into graph meaning and context. From the perspective of developer insights, the YOD underscores the contingent design relationship between fonts and search capacities. While we still await the verdict on the official Unicode bone graph font, bone font developers will likely need to agree on a set of basic graphic components to facilitate input and classification in the future. 


Shima Kunio 島邦男, ed., Inkyo bokuji sōrui 殷墟卜辭綜類 (Tokyo: Kyūko, 1971).

Matsumaru Michio 松丸道雄 and Takashima Ken-ichi 高嶋謙一, eds., Kōkotsumoji jishaku sōran 甲骨文字字釋綜覽 (Synthetic Index for Interpretations of Oracle Bone Inscriptions) (Tokyo: University of Tokyo Press, 1994).

Yu Xingwu 于省吾, ed., Jiagu wenzi gulin 甲骨文字詁林 (Beijing: Zhonghua shuju, 1996). 

2 thoughts on “Cracking the Oracle Bones: The Yinxu Oracle Bone Inscription Digital Database (Part 2)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s