I’m in the process of translating a fourteenth-century text, Wang Dayuan’s 1349 Brief Treatise on the Island Barbarians (Daoyi zhilue 島夷誌略), and I do this by putting passages in Grok, and then checking its translations.
Today I was working on a section on “Jiaozhi,” the area of the Red River delta, when Grok gave me the following translation:
“Among the common people, sixty-seven cash are reckoned as equivalent to one tael of Zhongtong silver. The official rate is fixed at seventy” (民間以六十七錢折中統銀壹兩。官用止七十為率。)
I don’t know what “Zhongtong silver” (Zhongtong yin 中統銀) is, so I wasn’t sure if this was correct. In general, I know very little about the history of money and currencies, so this is something I always need to check.
The first thing I always do is to check a good online Chinese-Chinese dictionary (zdic).
What came up was not “Zhongtong silver” but “Zhongtong bills/notes” (Zhongtong chao 中統鈔), with the definition saying “Paper money issued during the Zhongtong 中統 reign period of the Yuan dynasty, in two types: jiaochao 交鈔 and yuanbao chao 元寶鈔.”
Ok, so from that I realized that “Zhongtong” meant a Yuan dynasty era, and looking that up, I find that it was from 1260 to 1264, about a century before the text I’m looking at was compiled.
However, I wasn’t sure why the text I’m looking at had Zhongtong “silver” when there was a kind of currency that was created in the Zhongtong era that was called “bills/notes.”
So, I asked ChatGpt, “What was 中統銀?”, meaning “What was Zhongtong silver?” It thought for 25 seconds, and then for some reason it responded in Chinese as follows:
1) As a conversion/pricing unit equivalent to “taels of silver” (銀兩)
In Yuan-dynasty materials—and in Yuan authors’ descriptions of foreign polities—transaction values are often expressed by converting them into “so many taels of silver.” In this sense, “one tael of Zhōngtǒng silver” (中統銀一兩) usually means “the value of one tael of silver as reckoned under the Yuan’s Zhōngtǒng system (or with Zhōngtǒng paper money as the benchmark).
A typical example is Wang Dayuan’s Dǎoyí zhìlüè (島夷誌略), in its entry on Jiaozhi (Giao Chỉ / Đại Việt), which says:
“Copper cash circulates for use. Among the people, sixty-seven cash are converted as one tael of Zhōngtǒng silver; for official use, they stop at taking seventy as the standard rate.”
That is: locally, copper cash is the circulating medium, but there is an exchange-rate conversion between copper cash and “one tael (Zhōngtǒng) silver” (private rate 67 cash = 1 tael; government rate 70 cash = 1 tael).
2) Possibly a corrupted/mistaken writing for “Zhōngtǒng paper money” (中統鈔)
Some scholars—especially those working with surviving paper-money artifacts and textual/edition collation—point out that what is commonly attested in the Yuan is Zhōngtǒng chāo (中統鈔; Zhōngtǒng Yuánbǎo Jiāochāo 中統元寶交鈔). In some documentary contexts, “Zhōngtǒng silver” (中統銀) may therefore be the result of a scribal or printing error in which 鈔 was miscopied/misprinted as 銀. Hence one must judge from context.
3) (Supplement) “Zhōngtǒng silver-currency goods” (中統銀貨): a proposed silk-based “currency” equivalent to silver
Japanese-language scholarship also mentions a view that, early in Qubilai’s reign when the Zhōngtǒng paper-money system was being implemented, there was a proposal or accompanying concept of “Zhōngtǒng silver-currency goods” (中統銀貨)—made of silk and treated as equivalent in value to silver.
(This claim usually appears in discussions of the design of early Yuan paper-money policy and the intended “reserve/redemption” mechanisms behind it.)
HOLY S%^&!!!!!
Not only does that answer every question that I had in my mind and more. . . it’s all backed up with links to reliable sources of information.
This didn’t happen a year ago. I’m not sure when this started, but it hasn’t been that long.
So, now when I’m working, what I’ll do is to have different LLMs open, and I’ll throw a question into one and let it think about it, and then I’ll ask another to translate a passage, and then I’ll check back to see how the first LLM responded to my question, etc.
Is this different from Googling for information?
To some extent, “no,” because I probably could have figured all this out without an LLM, but it would have taken much longer than 25 seconds, and it would have interrupted my work.
Now, it’s becoming a seamless process. I don’t necessarily follow up on all the links and information that I’m informed about, but that stuff all seeps into the back of my brain, and I know that it’s not only “out there” if I need it, but that I can search for and find it quickly again in the LLM that I’m using.
Sounds easy. One could also just read what the official history of the Yuan 元史 offers on 鈔法, or the exchange rates described in Yuan wenlei 元文類. Just trying to preserve some self-worth here faced with machine competition…
Yes, it took more than 25 seconds to find these references, about 35 secs to access the relevant databases, but then Chinese studies requires patience. And there is not much more on that specific currency.
If you have access to the Daoyi zhilüe with commentary (Daoyi zhilüe jiaoshi), you will find that in the description of Penghu (Pescadores?) the currency is referred to as Zhongtong qianchao 中統錢鈔 to which is added an informative discussion by commentator Su Jiqing 蘇繼廎. In the descriptions of Luohu 羅斛 and Wuye 烏爹 a Zhongtong chao 中統鈔 is mentioned, which may or may not be correct for Zhongtong yin. It is obvious, if we want to believe the Daoyi zhilüe (which has a somewhat sketchy transmission history) that some foreign people were still using “dated” currency in the middle of the 14th century.
Thanks for the comments!! Just to be clear, I’m writing these posts about AI to document the kinds of changes that I see happening, because I think the gradual nature of the development of LLMs and our use of them, will (like our gradual transformation from living and working in the analog to digital worlds) make changes feel second nature and we won’t really recognize how radically the way we work is being transformed.
A little moment like the one I just described – realizing that an LLM is not just explaining what something means, but now also identifying issues that one needs to consider and pointing to writings on those topics – is the type of thing where we usually just go “Wow!,” and then move on. And then a year from now, when we are so used to using LLMs to try to learn what the issues one needs to consider regarding something, we will forget that not that long before, this didn’t happen.
Your statement about the need for patience and how it’s not that hard to find this stuff in a database actually points to exactly what I’m talking about.
I started in the 1990s when the way to find information in sources required that one look through concordances. To do that, there was some kind of numerical system that you had to learn, as each character had a number associated with it, then I would have to go to the university library, because the concordances were in the reference section and could not be checked out, then I had to go through the concordances to look for where that term appeared, then I would copy that down by hand. Then I would look up in the library catalog the call number for that text (and some entries in the catalog were in Pinyin, but others were still in Wade Giles, so you had to know the tricks of how to search). I would go find it in the stacks (which might be on a different floor or sometimes even in a different building – I ran up and down stairs A LOT), and I would look through it to find the passage. I wouldn’t find it, so then I would go back to the concordance (which you couldn’t take out of the reference room) and look again, and I would see that it was actually in juan 11, not “juan 12” as I scribbled on a piece of paper. So, then, back to the text. . . And if there was some information there that was indeed important for my research, texts like the dynastic histories were “non-circulating” (meaning you could not check them out of the library), so I would either have to take notes by hand, or go get a copy card, put money on it, stand in line to wait to use the copy machine, or go from one floor to another looking for a copy machine that wasn’t broken or out of paper, make copies, go back to my office and use a 3-volume Chinese-Chinese dictionary that I had, and if there were things that I still could not understand, I would have to go back to the library to look them up in a better dictionary, like Morohashi’s Dai Kan-Wa Jiten, which, of course, was in the reference room and could not be checked out.
Now THAT took patience! What we do today only requires a tiny fraction of the patience (and physical effort!!) of what was required just 20 years ago. And yet, we got from that extreme-patience-necessary place in the analog age to our current minimum-patience-neccessary place in the digital age without much sense of ambivalence other than feeling a bit nervous that a digitized version of a text might have OCR errors.
I think that can partially be explained because it happened gradually. When I completed my PhD in like 2001, my adviser gifted me a set of DVDs of the Siku quanshu, but it was not searchable. . . At some point near the end of the 2000s, a graduate student installed a pirated version of some software on my computer so that I could search through the SKQS. But I would still then have to search using that software, and then find the DVD, load it, and go through the document to find it.
That also took patience, but less than was needed a few years earlier.
It’s really only in the past 15 years that databases, Ctext, etc. have all become common tools of the trade. However, because all of this happened gradually, and we adopted each new tool as it emerged, now people don’t think/remember/know anymore how recently things were different.
This is exactly what is now happening with LLMs, only that it is transpiring more quickly (and that might partially explain why people are ambivalent about it), and it’s what I am trying to document and point out.
I’m not saying that this could not be done before (just like it was possible to search for information in texts before the digital age). I’m just surprised and impressed that rather than simply telling us what it means (which, I think, is what it would have done a few months ago), it immediately identifies precisely the kinds of issues surrounding it that we need to consider, and those are precisely the questions that pop into one’s head when one conducts a search to try to figure out what the term means.
Further, simply seeing such information, even if you don’t need it, is helpful. And yes, that same thing happens when you search, but it’s just in a much more random and time-consuming way. And so now we can do this without disrupting our research even for things that we don’t need such information for to build an expanded back-of-the-brain body of knowledge. That’s what I meant when I said that I throw something into one LLM and then go back to work on something else and then check back to see what it says.
So, I can predict that this is something that more and more people are going to do, and like I said, not long from now, people will no longer remember that this was not possible just a little while before.
Finally, thank you very much for pointing out where else Zhongtong silver/notes is mentioned in the Daoyi zhilue!! I skipped Penghu and went right to Jiaozhi. And I do have the edition by Su Jiqing, but I prefer to try to figure things out by myself first, and then I consult what others have said later. I do this because it helps me learn, and I don’t like to be influenced by other people’s ideas from the start (In particular, I don’t want to see where they think places were located). But yes, I will consult that work later.
In general, I find that Chinese scholars have a problem in that they try really hard to link names in old texts with current place names in Southeast Asia, and that leads to a lot of distortions.