Thread by @benjcartlidge

OK so are you ready for a live-tweeted reading of the new #Voynich MS article?

Not that it particularly matters, but the paper is written in very curious English: "...following his death, the manuscript’s custodian became his wife Ethel #Voynich (1864–1960)". I see what is meant, but this is not perspicuous prose. /1

"One scholar even produced a transcription of the manuscript that is entirely incorrect". Did s/he indeed? That's what you call chutzpah, my chums. #Voynich

"the manuscript was written in an extinct and hitherto unrecorded language as well as using an unknown writing system and with no punctuation marks" - not impossible. But 'extinct' languages are still languages; may be a crucial point. #Voynich

"Perhaps inevitably, and certainly ironically, the manuscript has revealed itself to be far more interesting and informative than imagined" - this is Alanis Morissette irony.

Right, now we come to it: lang of #Voynich arose from "a blend of spoken Latin, or Vulgar Latin, and other languages across the Mediterranean during the early Medieval period". Not clear what this means, but it can't mean 'Vulgar Latin'.

Nor can it mean 'Proto-Romance'; apparently the #Voynich lang. is not similar to 'Proto-Italic' - by which C. must mean Proto-Italian, as Proto-Italic is quite another matter (the 'parent dialect' of Latin, Oscan, Umbrian, South Picene etc.).

Sidenote (I realise I've been forgetting to count tweets but whatevs): Proto-Romance is not attested in texts, but it isn't a 'fiction' - it's a scholarly hypothesis that is trying to say something about the evidence Romance langs give us about their relationships.

I.e. there are criteria about what Proto-Romance should, or should not, look like. Quite complex criteria at that. No-one would reconstruct Proto-Romance without any unaspirated stops, or without any pronouns, or with intact Latin nominatives singular.

Oddly, C. says that the 'vowel symbols' are similar to (his) "proto-Italic", but that the 'consonant symbols' are different - but *symbols* are not the issue, *sounds* are: the symbols, in the #Voynich MS, are not like anything else (sort of the point...).

The confusion between sounds and symbols goes on. There are 26 letters in the #Voynich alphabet, apparently, plus diphthongs (but I think digraphs are meant) plus triphthongs (again, trigraphs are meant); some letters are silent, because pronunciation had changed.

But this is quite alarming: we are now assuming a potentially ENORMOUS grapheme inventory (plainly not every logically possible trigraph will surface but even so). If there's anything we can't explain using this, then something really must be up.

"the writing system of the manuscript can be apprehended once the grammatical rules are understood" - methodological nonsense. How are you going to write a grammar of a language whose writing system you don't understand? How can this possibly have been peer-reviewed?

Three e-vowels, <e>, <ee>, and <é>, but no explanation about what is really meant by this: is this graphic or phonetic variation, and if phonetic, is it also phonological? Two a-vowels, one spelling word-internal <a>, one spelling initial, final <a> and <a> after /p/!

Ah, now we understand - 'triphthong' means 'ligature'. But I've never seen a ligature system like this one. By combining the 'l' and 'e' signs in ligature you spell 'ele'. Do provide a parallel if you know of one, but this seems unusual to me.

"Figs 17 and 18 show the letter 'p' and the phoneme 'qu'" - this is not linguistics.

Even if the interpretation isn't right, it is interesting to see what the symbols of the #Vyonich MS look like in this way. But (should've offered disclaimer at start) I am no Voynich expert - my interest is really in the linguistic 'argument', such as it is.

The #Voynich language had sequences -el(l)e, -em(m)e, -ep(p)e and -eque, but not -en(n)e: the latter lacks a 'Latin root'. I imagine the roots of the other forms are going to be stated at some juncture? W/out an example of a lexeme it is simply impossible to evaluate this claim.

No upper/lowercase distinctions in the #Voynich MS. Reasonable enough, but I'd need to take another look myself.

So now we get to some examples, e.g. Figure 32. 'omor néna', 'killed baby'. This requires first that we accept the red blob drawing as an abortion: possible, I suppose. omor, cf. Rom. omor, 'to murder', néna, cf. Sp. 'néna', baby. But what is desperately odd about this...

...is that the forms are identical to the modern languages. That is totally at odds with everything we know about comparative linguistics. I could almost believe it if the proposed #Voynich form corresponded to *even* a *new* reconstruction of a Proto-Romance word.

I'm no Romanian expert, and I can't work out quite what 'omor' means but I think its the infinitive. But why is a Romanian infinitive a #Voynich past participle? If there is Romance lexis, there should either be Romance derivational morphology...

....or a clear explanation about why morphology is lacking (oh and there totally are ways you could get to this, but you'd need a whole other explanation with some known mitigating factors in play to make it convincing: very tricky in the face of a totally unique document).

Five women in a bath are labelled with 'temperaments'. Really? I'd've gone with names, I have to admit. Apparently one of the temperaments is 'golden bird', which I sort of like as a kind of Proto-Romance Scouse slang.

Right, well we now get a whole phrase: la naza éo eme ona oma nor nais t. On the basis of no fewer than 6 languages this phrase means 'The baby it's to acquire good growth as for normal birth'. Hot tip: if your translation doesn't make sense, it might be a sign.

Furthermore, the 6 'Romance' languages include - are you ready for this? - Greek. And that I think is a clue to how this has happened. The data for this interpretation are taken from the index to the Meyer-Lübke Romanisches Etymologisches Wörterbuch.

Because M-L's dictionary includes Greek loanwords into Romance, Greek is in the index - *not* because Greek is a Romance language. The Greek word is apparently 'oma' - feminine of homós, 'same'? Perhaps. 'eme' is supposedly from Latin emere.

M-L has been assumed to have the status not of an etymological, but of a synchronic dictionary. Thus it would not be encoding a history of linguistic change, but describing a real language. This is the assumption at the root of this catastrophic paper, I think.

To be honest, I have kind of lost the will to live now. I've skimmed through some of the other drawings with alleged evidence. #Voynich

I return briefly to the issue of 'néna' supposedly meaning 'baby'. From the drawing (Figure 32) it looks absolutely apparent that the first and third letters are different.

In other words, this interpretation has not really been arrived by a study of the script itself first. It's just been a leap into the abyss. If the MS is going to be deciphered, you would need to construct a full inventory of all the signs first.

"As a result, proto-Romance survived by vestigial fragmentation of its lexicon into the languages we see today." This doesn't mean anything, to my way of thinking. What is 'vestigial fragmentation'?

Right, now we've got another MS in play, which apparently proves that the writing system and language of #Voynich evolved from Romance. This seems to be because they sort of look a bit similar. Again, no real demonstration of this, just a picture.

A picture paints a thousand words? No, because the point at issue is the inventory of #Voynich signs. So it's incumbent to demonstrate their similarity, point by point.

The other MS uses consonants 'stylized and abbreviated in a remarkably similar way to the symbols' in #Voynich. OK - but tbh all I think this shows is that this is why we *do* think the Voynich MS contains writing: it does kind of *look* like writing!

The paper ends rather abruptly with something on the historical context, and I leave that for others to evaluate - indeed, there's probably a whole host of #Voynich threads going on right now.

To finish my own. A decipherment has to proceed palaeographically first of all, and there's evidence this hasn't. Then standard sequences of some kind have to be identified, especially looking for combinations of sequences that exhibit principled variations.

Incidentally, none of the trigraph/digraph readings seem to play any big role in the actual texts presented. Nor is the claim that many #Voynich words start with the d-symbol actually demonstrated. Maybe the sign C. interprets as d- is a grammatical form of some kind? No idea.

And that's the problem: C. hasn't constructed a systematic grammar of a language: he's taken an etymological dictionary and launched himself at anything he could find. You couldn't make a synchronic grammar of a language out of this stuff.

Indeed, many of the comparative bits of evidence show sound changes specific to other languages. If the #Voynich MS is in an unattested Romance language it should have its *own* sound changes, starting from Latin, not the sound changes of other langs.

And that's a wrap. Thanks for reading! Sorry if I've ignored any replies, I'll try to pick them up.