We have long known that [p] and [f] cannot be "normal" characters, in the sense that the overwhelming majority of their occurrences are in the very restricted positions of the first lines of paragraphs, titles, or other text that is somehow marked as prominent. In the non-initial lines of regular text paragraphs, they occur very rarely.
The most obvious hypothesis is that they may be alternate forms for the similarly shaped "gallows" characters [k] and [t]. However, this hypothesis has the problem that it dramatically shrinks the size of the Voynich character inventory, which is already very small as it is, and makes it very difficult to conceive of a system whereby all of any language's consonant phonemes could possibly be represented. Furthermore, I believe it was Currier who first made the point in the 1970's that [p] and [f] occur in distinctly different environments than [k] and [t]: The clearest example I recall is that [p] and [f] almost never occur before [e], whereas [k] and [t] very often do.
Well, I happened to check the statistics for the glyph [d], and I was reminded that it also very, very rarely occurs before [e]: only 1% of all [d] glyphs in the ms are followed by [e]. This is similar to the same statistic for [f]: 0.8% of all [f] glyphs are followed by [e]. The proportion for [p] is even lower.
Likewise, the very frequent words and sequences with [d] also tend to be relatively frequent with [f] and [p] (though not to quite the same extent). We all know [-daiin]; I find that [-paiin] (44 tokens) and [-faiin] (14 tokens) are relatively common as well, considering that [p] and [f] are not all that frequent themselves. So it is with [-dar-], [-par-] (50 tokens), and [-far-] (28 tokens); and with [-dal-], [-pal-] (47 tokens), and [-fal-] (15 tokens).
Of course this correspondence does not exist in *all* contexts of the glyph [d]. Most blatantly, the ubiquitous Currier B sequence [-edy] has no frequent counterpart for [p] and [f], although perhaps it is worth noting that [-epy] and [-efy] do occur 12 and 11 times respectively.
All of this leads me to consider the following hypothesis:
* The glyph [d] represents a distinct letter from [p] and [f], but in non-initial lines of regular text paragraphs, the scribe simply chose to write all of them as [d].
In this scenario, [d] would still have its distinct environments where it represents its primary letter value, where [p] and [f] would not necessarily occur at all. But in many environments, we would see a certain correspondence between the occurrences of [d], [p], and [f]. This is indeed the pattern we observe in the statistics I noted above.
=====
So what would be the logical letters / phonemes to participate in such a substitution?
I suggest this system could make sense if the glyph [d] represents a value like /u/~/v/, and the glyphs [p] and [f] represent labial consonants like /b/, /f/, or /p/.
In many plausible candidate languages for the Voynich ms, the sound change known as "betacism" has caused /b/ and /v/ to be hardly distinguished from each other at all. Spanish is probably the most famous example, but the process has also occurred in Catalan, southern Occitan dialects, many other Iberian Romance languages and dialects, Neapolitan, Maceratese (Macerata, Italy), as well as in Medieval/Modern Greek and Ancient Hebrew. On Spanish store signs, etc., in New York City, one finds these two letters randomly alternating for each other all the time. Wikipedia notes the clever medieval Latin saying, "Beati hispani, quibus vivere bibere est" ("Fortunate are the Spaniards, for whom living is drinking").
It would not be at all surprising to me if the scribe found it inconvenient to write the fancy gallows character ([p] or [f]) for /b/ in non-initial lines of text, where there is less room above the line to draw the elaborate glyph. It would have been quite a simple fix to just write the glyph [d] representing /v/ instead, which may have sounded almost the same to the author. I notice that on the very first page You are not allowed to view links.
Register or
Login to view. , [p] and [f] occur frequently in non-initial lines. Perhaps the scribe tired of drawing them in cramped spaces and made an ad hoc decision to substitute the almost identically sounding glyph [d] as he wrote more and more pages of text. This substitution process may have begun as simply writing /v/ for /b/, and from there it was extended to writing /v/ for the phonemes /f/ and /p/ as well.
The letter representing the sounds /u/~/v/~/w/ has a distinctive place in many languages: sometimes a vowel, sometimes a consonant, sometimes a glide. It does not fit so neatly in the "series" of other groups of phonemes. Likewise the glyph [d] is distinctive in the Voynich script, and it does not belong to or pattern with any other series or group of glyphs.
I considered the glyph [d] as the phonemes /u/~/v/ while taking a cursory look at the idea of Old Occitan (see my post about the Old Occitan troubadour cryptogram poems in the Pre-Modern Cryptography forum). My first idea was to read [daiin] as the indefinite article "un", but it could just as readily represent other common words, syllables, and suffixes such as "-um", "van", "-ban", "ven", or "ben", not to mention "fin" or "vin". Of course there would probably have to be some kind of cryptographic element to make a Romance language possible as the language of the Voynich ms text at all. I do note that writing all of the phonemes /b/, /f/, /p/ the same as /v/ throughout most of the text is a simple form of cryptography, whether it was originally intended in that way or not.
It also occurs to me that the [d] = /u/~/v/ idea could also explain another curiosity of the text, the sudden and ubiquitous appearance of the [-edy] suffix throughout the Currier B sections of the ms, in contrast to the almost complete absence of this suffix in the Currier A sections. As is well known, the glyph [y] looks like the very common medieval Latin ms abbreviation symbol that represented the suffix "-us". Perhaps the Currier A scribe simply used the glyph [y] in the same way, as the suffix "-us", but the Currier B scribe only used [y] to represent the letter "s" alone. In that case, the Currier B scribe would need to add a character before it to represent the "u": with my hypothesis here, this character would have been the glyph [d]. This explanation would account for the otherwise strange discrepancy between the Currier A [-y] suffix and the Currier B [-dy] suffix.