The 'Chinese' Theory: For and Against

[Jorge_Stolfi]

You talk about the Red Rooster more often than about these.

For two reasons. First because ROOS and parag f105v.32 are extreme outliers in terms of size, so there is no question of which entry each of them should they be matched to.  If SPS=SBJ, then  f105v.32 is the only possible transcription of ROOS, and conversely ROOS is the only possible "decipherment" of f105v.32.

Second, because ROOS is the only recipe that has eight sub-entries with the 主 keyword.  A few other recipes have three 主s, several have two, and the rest has one or none.  Thus the matching of f105v.32 to ROOS, with the 主 crib (still the best one I have) requires matching the sizes of nine gaps, rather than just four or less.

If I show an example of middling size with only three 主, like BLAR, you could reasonably say that, with 50+ parags of about the right size, many with more than three daiin-like strings, it would be very likely that one of them produced a decent match to the four gaps, just by chance.  (Even more so if you only consider the two internal gaps, as in @eggyk cokbook example above.) 

Refuting this expected counterclaim of yours would require detailed probability calculations -- that you would probably reject too, since they must start from a guessed model of "random" paragraph.

All the best, --stolfi

EDIT: strike out incorrect claim about @eggyk's experiment.

[ololololo]

- that you would probably reject too,

Oh, I'll clarify here. I do not reject the Chinese theory and recognize the comparison with the Red Rooster, there are indeed coincidences in size. It's just that, in my opinion, you draw very few useful and valuable conclusions from this, and also seem to focus on certain recipes, slowly moving on to new ones.

As I said, SBJ recipes are a useful direction for Chinese theory, rather than reasoning about what the Dictator said there and what the Author heard (this will become clear during the analysis)..

How many SBJ prescriptions have been checked in total? How many matches are there? Are there any discrepancies (for example, one SPS prescription turned out to be different from an SBJ prescription for some reason, or there are some discrepancies)?

[Jorge_Stolfi]

I was curious about this, so I looked for an English cookbook, and on my very first (!) try, I found a match with—I think—similar requirements. 3 daiin out of 6 daiin

Oh wow!  So the SPS could be a cookbook? 

Seriously, first I must commend you for choosing the right type of text, rather than a novel or theological treatise.  I wish everybody did the same in their tests.

Indeed the BLAR example by itself (or any other recipe with three or fewer cribs) is not a strong proof of SPS=SBJ.  It is only a modest confirmation of it.  The proof requires more exampleas and/or more cribs.

You are taking an SPS parag and looking for a (cookbok) recipe that matches.  I do the opposite, for various historical and practical reasons.  But I suppose it does not make a difference for the analysis of your example.

But a problem with your example is that your tolerance of ±5 EVA letters is quite a bit more than the errors I consider reasonable.   If you are measuring the positions from the centers of TAKE and DAIIN, you have ~66 positions out of 237 that count as "hits".   

That is, it is like you had 237 closed boxes, 66 containg a colored ball of one of 6 dfferent colors, 11 of each color; and you opened three of the boxes at random. The probability of getting a colored ball in the first try is 66/237.  If that happens, then the probability that the second box has a ball of a different color is 55/236.  For the third box, the probability of getting ball with a color different from the first two is 44/235.  

Thus the probability P of getting three balls of different colors is the product of those three numbers, which is ~0.012 or 1.2%.  If you repeat the exercise 100 times -- that is, if you scan 100 recipes with about that size -- you have 1- (1-P)^100 = ~70% chances of success.  With 200 tries, the probability of a success is more than 90%.  How many recipes are in that cookbook?

In my tests I expect that, in a correct match, most size errors will be less than ±3 EVA letters.  (To reduce the noise I measure N+1 gap sizes instead of N positions, but since the sum of the gap sizes is fixed, effectively I have N independent datums to match).  Then instead of 11 boxes for each color I have only 7.  On the other hand, if I count the main alternatives to daiin like kaiin and daiiin, there are 11 (not 6) sets of 7 colors, which raises P to ~2.5%.  On the third hand I penalize those alternative choices in the badness score, which is somewhat equivalent to reducing the number of boxes for those colors.  So it is not clear that I had as much chance of a meaningless coincidence with BLAR as you had in the reverse search.

Another question is how you picked the word "take" to be the anti-crib of daiin.  In my case, the pairing 主 = ~daiin was determined from the ROOS = f105v.32 matching; and I had no other viable choice there.  I then use that same crib for all other recipes; I don't have the choice of redefining 主 = ~chedy or something else for BLAR.   Can you say the same of your anti-thesis, "SPS=FOC"?

All the best, --stolfi

[Jorge_Stolfi]

 a big problem here is you clearly do not understand the significance of your own examples.

Sigh.

No, it is you who is not understanding anything of what I am doing and proposing, no matter how much I try to explain.  Do you even want to understand?

It seems clear to me that you do not care about the Chinese Origin theory, or the SPS=SBJ claim, or even about the Dictation Scenario.  It seems that your only goal here is to declare that I am completely wrong about... something, I can't really figure out what.  Prove me wrong on this...

For the billionth time:

I am providing those readings of SBJ recipes in Mandarin, Cantonese, Vietnamese, etc only to (a) make my posts a bit less forbidding to those readers who (like me) cannot recognize the hanzi from memory, or see the difference between 主 and 生, and to (b) give them a feeling for what a phonetic transcription of the SBJ would look like in some generic East Asian monosyllabic language.  

I do not use those readings in any way in my analyses and attempts to determine the correspondence between the SBJ and the SPS.  For that I use only the hanzi text of the SBJ and the Voynichese EVA text of the SPS.  I make no attempt to map either of them to sounds. 

The Chinese Origin theory (COT), the SPS=SBJ claim, and the Dictation Scenario are three distinct independent theories.

• The Chinese Origin theory says only that Voynichese is an un-encrypted phonetic transcription of some East Asian monosyllabic language using a phonetic script or shorthand.  The main evidence for it is the word structure and various statistics of the text, including many statistics (like the apparent absence of grammar) that have been claimed to exclude a natural language in the plain or with a simple substitution cipher.  The similarity of the Big Red Weirdos of You are not allowed to view links. Register or Login to view. to upside-down Chinese characters, and the parallel between the 24x15 division of Zodiac and the Chinese Solar Terms are not claimed to be main evidence; they are just two bonus bits of evidenciary candy. 
  

• The SPS=SBJ claim says that the Starred Parags section is a transcription of the Shennong Bencao Jing in a mostly one-hanzi-for-one-word fashion, minus the Nature, Alias, and Provenance fields, and some non-medical comments in some entries.  Specifically, that it is a transcription of the SBJ text that was quoted in white-on-black in a printed edition of the Zenghe Bencao that was available in 1400 CE.  This claim does not specify any particular language or encoding, except that each hanzi seems to be consistently mapped to the same Voynichese word -- at least for the some common words, and modulo some spelling variations of still undetermined cause.  In particular, it says that the hanzi 主 is usually transcribed as daiin or slight variations thereof. The evidence for this claim is the surprisingly accurate matches between the positions of 主 and other key hanzi in several recipes and the positions of daiin and other specific Voynichese words in some SPS parags.
  

• The Dictation Scenario proposes that the draft of the Starred Parags section (and possibly the rest of the VMS) was created by an Author who could not read Chinese characters but had some mastery of a spoken East Asian language, by taking dictation from a reading of the SBJ (and possibly other books) by a Dictator who could read the Chinese characters and say them loud in that same language.  The draft would then have been transposed to vellum in Europe, several years later, by an European Scribe. The purpose of this Scenario is only to show that the two theories are historically and practically possible; that is, there is at least one plausible way that they could have come to be true.  Therefore it does not need to be proved; it suffices to argue convincingly that it indeed could have happened.
  

Neither the COT nor the SPS=SBJ claim require the Dictation Scenario to be true, and the evidence for them does not depend on it.  But the Dictation Scenario, as a side effect, also provides explanations for some observations made in the exploration of those theories -- such as the large number of "errors" (variants) of spelling and word spacing.  And for why the VMS uses an original alphabet, and why the alphabet looks like that.

Strictly speaking the COT and the SPS=SBJ claims are logically independent: either one of them could be true even if the other is false.  But of course if the SPS=SBJ is true, the COT becomes much more likely, and vice-versa.

Consider that just a few days ago I introduced this quote from a linguist:

Again, that may be true, but it is totally irrelevant to the COT, the SPS=SBJ claim, and to the Dictation Scenario.  How scholars would normally read the SBJ aloud has no implications for any of them.  In the Dictation Scenario, the Dictator would not even have to know the "Classical reading"; he would only need to know (or guess) the general meaning of each hanzi, and say that in the local language that the Author spoke.  He may have deviated from the one-hanzi-one-syllable routine occasionally (like read 雄鸡 = "male chicken" as a single syllable meaning "rooster", or 血闭 as some specific syllable or compound for "amenorrhea" that did not include the syllable for  血= "blood"); but the evidence for the SPS=SBJ theory indicates that such deviations were rare.

I should stop here, but I cannot resist commentinig on the following quote:

But any literary Chinese text has a perfectly specific spoken form, composed of morphemes many of which occur in modern spoken Chinese and all of which are etymologically identifiable with morphemes that have occurred in spoken Chinese at some historical period:  there is a well-defined way of reading a literary Chinese text aloud (morphemes which are obsolete in spoken Chinese are given the pronunciations that result by applying subsequent sound-laws to the pronunciations the morphemes had when they were current in speech),... 

Yes, I am well aware that scholars in each country of the Sinosphere have a longstanding consensus about how to read Classical literature.  Like Egyptologists have a longstanding consensus about how to read hieroglyphs, and Sumerologists have the same about Sumerian.  And like linguists have a longstanding consensus about the lexicon and grammar of Proto-Indo-European.

But just being a long-standing widely adopted consensus does not imply that a hypothesis is true.  Especially in linguistics.  

All those readings may be supported by valid linguistic theories, but are ultimately a pile of arbitrary conventions.  Those scholars don't really know if the conventional pronunciation of the words, the meanings that are conventionally assigned to them, and the conventional parsing of the words into sentences are those that the authors actually meant, thousands of years ago.  The phrase that they all confidently read as "a stone mason" may have been actually "a crude statue" or "the cruel official" or "the stoning of a slave" or "the girl with hard heart that I passionately love but who does not correspond my love".   

And that is particularly true of Classical Chinese, because the Egyptian and Sumerian scripts were at least partially phonetic and could in principle record inflections and auxiliary particles that early Chinese script may well have omitted.  

If you think that this is a "blasphemous" claim, see the recent discussion about why Medieval depictions of Moses have horns...

But, again, this point has no implications for any of the three theories above, or fof my analyes and proof attempts.

[rikforto]

If that is the only plausible "written characters ---> dictation ---> voynichese" process, how much would the order of the broad content change? (depends on local language? did word order vary? (verb/noun/adjective))

It is going to vary a lot! Some of the languages that have been suggested are SOV languages, which should have been immediately ruled out because the cribs should be in VERY different places. (There's some acknowledgement of this from Stolfi at the link I highlighted above, but only after trying to force the issue by asserting a more literal translation is possible.) But there will be other differences that will rearrange and change things. A major typology difference through the region is some words have noun-like attributives and some have verb-like attributives, and that can move stuff around and eliminate or add words if translated to a language with a different approach. But, picking up what you're putting down, yes, it could, in principle, look something like the lines we see in the above graphs---especially if the unused nodes were way less prevalent.

I would take really solid cribs in that manner as "further study warranted" indicator, especially if we are only talking about SVO languages---which the Sinitic languages all are. That is, to a point, a agree with this:

I would imagine that although certain words would not be as perfect a match, their rough position within full recipes may not change that much, nor the amount of times many of the words would be said. "named/called" would still be near the start, for example. Unfortunately, this would likely make it impossible to ever construct a coherent 1-1 translation from voynichese to plaintext, unless we knew the exact language.

From both ends, I don't think that's what's happening here. On one end, I don't think the cribs are very solid---the texts are both manipulated, there are a lot of arbitrary choices, and the method is generating multiple matches per entry. And at the other end, plainly much stronger conclusions are being asserted. However, if someone wants to start this work afresh and try and unwind some of these assumptions, actually prove correspondence, and purge the theory of the flatly wrong assumption that all varieties of North and Southeast Asian languages are interchangeable at the syllable level, I'll hear the case afresh.

I actually think even if it dead ends, there might be some interesting things to learn from pinyin, jyutping, and chữ Quốc ngữ. (Even though it has different properties, hangeul is also an interesting case study as well as one of the best-documented premodern invented scripts and, while not completely tabula rasa, quite a leap from anything else around it.) And while I think there are huge hurdles to it being phonetic Classical Chinese, there are a lot of other possibilities. I don't think I've used the term "Indosphere" yet, but some of the "monosyllabic" languages are from there, and that's outside my wheelhouse but surely---surely?---there is an herbal tradition there? Do we have any Indosphere nerds on the forum?

[eggyk]

Neither the COT nor the SPS=SBJ claim require the Dictation Scenario to be true, and the evidence for them does not depend on it.  But the Dictation Scenario, as a side effect, also provides explanations for some observations made in the exploration of those theories -- such as the large number of "errors" (variants) of spelling and word spacing.

Aren't the justifications for removing certain sections of the SBJ based on interpretations of a Dictation Scenario?

If he could discern which areas to remove on his own, he must have been able to read the chinese characters (and as such had no use for a phonetic version). Therefore, at some level a dictator or "explainer" is necessary for explaining how the VMS author was able to remove the sections you believe were removed.

[rikforto]

This is not true:

Strictly speaking the COT and the SPS=SBJ claims are logically independent: either one of them could be true even if the other is false.  But of course if the SPS=SBJ is true, the COT becomes much more likely, and vice-versa.

They are mutually contradictory. There was no "local language" in 1400 that existed in one-to-one reflex with Classical Chinese on each syllable. You can not square SPS=SPJ (which matches to the syllable) with The Dictation Scenario (requiring a local language which does not exist in reflex this way to Classical Chinese). Again, this is not an issue of pronunciation standards or script typology, which continue to be a red herring here. The more you find that the text matches Classical Chinese, the less likely it is another language was being dictated. Again, I understand you are using the transcription in the characters, but again, transcribed any which way that text is in Classical Chinese.

And to be clear, there's no relevant linguistic reconstruction here; all of these languages, including period-appropriate pronunciation, are directly attested and in continuous transmission. Classical texts are understood about as well as any Greek or Latin text, and if you were trying to cast doubt on Latin herbals this way on this forum you'd be rightly pilloried. (And yes, I am aware this leaves some room for specific things to be unclear as in the Latin and Greek corpus, and I'm open to non-LLM evidence of specifics.) All of the reconstructions you are mentioning---controversial redundant affixes and Old Chinese pronunciations---are irrelevant by 400 CE, never mind 1400.

I also agree with Eggyk's observation that you are entwining SPS=SBJ, and hence are not logically independent in practice. This simultaneous entwining and contradiction is the heart of my objection!

[Jorge_Stolfi]

Aren't the justifications for removing certain sections of the SBJ based on interpretations of a Dictation Scenario? If he could discern which areas to remove on his own, he must have been able to read the chinese characters

There are other scenarios that could result in the VMS that we have, with SPS=SBJ and/or the COT.  Maybe the Dictator was a "Chinese" who visited Europe.  Maybe the Author could read Chinese characters, but decided to translate them into a phonetic script for the benefit of others (like his German-born heir) who could speak the language but did not know the characters.  Maybe Voynichese is a codebook cipher where each hanzi was mapped to a number in a funky number system, without going through the sounds.  

However, those alternative scenarios,and others I can think of, seem to be much less likely than my Dictation one, and some are not consistent with the statistics etc.  But maybe someone can think of some better alternative.

Anyway, if Voynichese is indeed a phonetic script, and the Author understood the spoken language, no matter how he got the phonetic representation he would have realized that those fields of each entry were useless -- if not before he started, then shortly into the project.  He surely knew the spoken words for "sweet", "sour", "warm", "another name", "grows", etc; and any doctor would have explained to him that those flavors and provenances are not literal ("excuse me, but how can rooster quills be sweet and warm?") but codes into Chinese medical theory -- and useless outside of it.

Even without knowing the language (written or spoken) the Author in theory could have recognized those fields because they are almost always at specific places in the recipe, and are introduced by specific characters: 味 = "flavor", 一名 = "another name", and 生 = "provenance,habitat".  To exclude those fields he would have to skip from the 味 (inclusive) to the next 主 (exclusive), and anything from the first 一名 or 生 to the end of the recipe.  Which is my case, in fact.

All the best, --stolfi

[eggyk]

Even without knowing the language (written or spoken) the Author in theory could have recognized those fields because they are almost always at specific places in the recipe, and are introduced by specific characters: 味 = "flavor", 一名 = "another name", and 生 = "provenance,habitat".  To exclude those fields he would have to skip from the 味 (inclusive) to the next 主 (exclusive), and anything from the first 一名 or 生 to the end of the recipe.  Which is my case, in fact.

If he didn't know the written or spoken language, he would not have been able to recognise those fields, even in theory. In order to remove them, he had to know what the symbol was and what the section represented. 

Anyway, if Voynichese is indeed a phonetic script, and the Author understood the spoken language, no matter how he got the phonetic representation he would have realized that those fields of each entry were useless -- if not before he started, then shortly into the project.  He surely knew the spoken words for "sweet", "sour", "warm", "another name", "grows", etc; and any doctor would have explained to him that those flavors and provenances are not literal ("excuse me, but how can rooster quills be sweet and warm?") but codes into Chinese medical theory -- and useless outside of it.

Why would the doctor explain that to him? The classical chinese text would not have provided an understandable flow of intelligible language when spoken character by character. Therefore I would assume such a flow would not have been understood by the VMS author as it was spoken, and that those sections wouldn't have been particularly more or less understandable than any others, so he would likely not have asked for an explanation. 

This would be different if the dictator read the characters and translated full sentences/concepts into the understandable local dialect which the VMS author understood. But in that case the structure of the SBJ content would be changed. 

If the concept was explained to him by the doctor, wouldn't writing it down still hold the knowledge of what it meant within the context of chinese medicine? Later on, the VMS author would speak out loud "is sweet, warm" and say "ah yes, sweet+warm means this in chinese medicine".

--------------------------------------------------------
Let me use an analogy close to my experience to perhaps get my point across: Let's say there's an englishman going to the netherlands. The englishman knows many dutch words, enough to have day to day conversations. Lets say for the sake of analogy that the englishman cannot read dutch text at all.  

One day, a dutch family all read a written dutch message from someone that they all seemingly find informative, provoking a strong emotional reaction. The englishman, curious, asks to be told what the message said so he can write the message in english for his english family back home. His family love a bit of gossip. 

But as his dutch isn't too great, he asks for it to be read word for word. It's going okay until: 
maar - but 
vanmiddag - this afternoon
kwam - came
de aap - the monkey
uit de - out of the
mouw - sleeve

"???" says the englishman. He knows all of the words, yet none of the meaning. 
"maar vanmiddag kwam de aap uit de mouw? They had a monkey?"
"nee nee, that is saying... uhh de secret.. intentie of him was then clear" says the dutchman
"Ah, the cat was out of the bag then!" replies the englishman
"Een kat?"

The analogy here is thus: the dutch saying in this scenario is the classical chinese section that makes no sense when directly spoken character by character. The meaning of the phrase isn't conveyed word by word, but rather as a block. 

Literally, it is maar+vanmiddag+kwam+de+aap+uit+de+mouw   ---->  "but this afternoon came the monkey out the sleeve"

but in reality it is being read as:
maar vanmiddag [kwam de aap uit de mouw]  ------> but this afternoon [came the monkey out the sleeve] -----> but this afternoon [the concept of a hidden motive being revealed]

The only reason the englishman knew to ask was because the dutch saying was nestled between content that he understood. The only reason the dutchman felt the need to explain was because the englishman knew to ask. In the chinese character by character dictation scenario, the entire content is comparable in intelligibility to that dutch saying. There wouldn't be sections he understood, and sections he didn't. He wouldn't understand the structure of anything, other than grasping a word here or there outside of its full context.

[Jorge_Stolfi]

If he didn't know the written or spoken language, he would not have been able to recognise those fields, even in theory. In order to remove them, he had to know what the symbol was and what the section represented.

He would need to just be able to recognize those hanzi.

I myself don't know any Chinese language and I can't read written Chinese.  For the latter,I would have to know the shape and meanings of at least a thousand hanzi, and thousands of compounds thereof.  But I know a few dozen hanzi from the time I studied Japanese.  I can recognize those keywords and easily delete those fields.  In fact, my program does that automatically, with only a couple of python lines.

The classical chinese text would not have provided an understandable flow of intelligible language when spoken character by character. Therefore I would assume such a flow would not have been understood by the VMS author as it was spoken, and that those sections wouldn't have been particularly more or less understandable than any others, so he would likely not have asked for an explanation.

A literal one-word-per-hanzi reading of that Rooster recipe into English would be

red male chicken flavor sweet  slightly warm main woman person burst inside drip down red white pour ... fatten fat grows plain marsh

Even with this mechanical reading the Author could tell that "flavor sweet  slightly warm" and "grows plain marsh" would be those standard non-interesting fields.

But a sensible Dictator would, at the very least, have used intonation to hint at the parsing.  That is, he would insert "oral punctuation", like modern editions of the SJB insert printed punctuation:

red male chicken: flavor: sweet,  slightly warm.  main: (woman person) burst inside, drip down, red white pour; ... fatten fat. grows: plain marsh.

This would be different if the dictator read the characters and translated full sentences/concepts into the understandable local dialect which the VMS author understood. But in that case the structure of the SBJ content would be changed.

Yes. I see some hints of that -- like the title 丹雄鸡 "red male chicken" becoming just two words poar.keeo -- but they seem rare and are not confirmed yet. (The poar.keeo may still be three words with a missing space.)

If the concept was explained to him by the doctor, wouldn't writing it down still hold the knowledge of what it meant within the context of chinese medicine?  Later on, the VMS author would speak out loud "is sweet, warm" and say "ah yes, sweet+warm means this in chinese medicine".

The meaning of those fields is quite complicated.  IIUC, the "warm" tag means that remedies derived from a rooster are supposed to have the general effect of moderately boosting the flow of 气, which is some invisible fluid that transports the nutritional value of food from the digestive center (stomach and spleen) to certain other organs.  It seems unlikely that the Author would have learned enough Chinese medical theory to understand that, and make use of that understanding.  

But, worse, for historical reasons there is usually only one 味 field per entry, even when the entry has sub-entries with very different indications, effects, and preparations.  Thus "officially" that "sweet, slightly warm" field applies not only to the rooster's meat abut also to its quills and to the white part of its poo.  That apparently makes sense within Traditional Chinese Medicine, because (IIUC) it assumes that all products derived from X will somehow inherit some of the healing attributes of X.  But the Author, even if he was just a silk and cloves merchant, may have seen that assumption as a silly ancient superstition...

Anyway, writing down those fields (or copying them to vellum) would have been a waste of time if the Author and/or the intended readers were not knowledgeable in Chinese medical theory (or did not believe in it).  The Author already had opted to omit all the additions and commentares that were appended to each entry in the Zhenge Bencao, including the "semi-sacred" additions made before 400 CE or so, which were printed black-on-white but in double size characters.  Obviously he did not have infinite time and money to spend in this project...

Let me use an analogy close to my experience to perhaps get my point across: ...
kwam - came
de aap - the monkey
uit de - out of the
mouw - sleeve
There wouldn't be sections he understood, and sections he didn't. He wouldn't understand the structure of anything, other than grasping a word here or there outside of its full context.

Yep, that is probably what happened.  

But even Chinese doctors versed in Traditional Chinese Medicine sometimes cannot understand words and expressions in the SBJ.  Like 鸡白蠹 = "chicken white grub".  Or like they (and the Author) superficially undertstand 鸡子可作虎魄 = "eggs can be turned into amber", but have no idea of what alchemical process is supposed to do that.  

So the Author would be confused like them, only over 70% of the text instead of 1%...

On the other hand, he would understand the overall structure of most recipes, because it is quite simple and regular.  Like that Englishman would understand or guess the  structure of most sentences, such as "coffee good is" or "has you today milk buy?".

All the best, --stolfi