The possibility that the characters [o, a, y] are vowels is the first proposal I would like to make regarding the sound values of the Voynich script. I am not the first to assign vowel qualities to these characters, as many theories have done likewise to all or some of them. I hope that one difference with earlier proposals is the basis on which I claim these are vowels. Whereas other theories have assigned vowel values due to the similarity of characters in other scripts or through reading individual words, I hope that I can back up my beliefs with a stronger linguistic argument.
Vowels are important to languages because they typically form the heart of a syllable. Almost every word, and every syllable in a word, will have a vowel. Indeed, the presence of a vowel (or, exceptionally, a consonant which can act as one) defines a syllable. A syllable is a string of sounds attached to—both before and after—a vowel. The vowel is also the pivot of the syllable with the consonants which come before obeying different rules from those which come after. Most languages have simpler structures after vowels than before, to the extent that some languages will not let any consonants at all come after.
The initial argument for [o, a, y] as vowels is simply their ubiquity. Of the one hundred most common Voynich words ninety–seven have at least one [o, a, y], and the three which do not are all single characters. In the manuscript as a whole, most pages have only a handful of words without these characters. (This is not including the null value which I believe occurs when is deleted.)
Conversely, gallows characters [k, t, f, p] occur in fewer than half the one hundred most common words; [ch, sh] in only a third; and [d, l, r, s] in just under two thirds. No other set of characters comes near to the ubiquity of [o, a, y], meaning that they are the best match for this aspect of vowels.
But further than this, I hope I have shown in the low level structure of a word, that when we analyze Voynich words by splitting them up according to Primes (that is [o, a, y] but also with the null value) we see a fairly regular section structure which can easily be described. These sections, as I have called them up to now, are really syllables. The characters which come before or after the Primes obey set rules and show the same kind of difference in complexity that natural languages show.
This is an important fact to stress and one which should help us immensely when we look further at the potential values of other characters. Not all ways of looking at the structure of Voynich words are equally good, but having [o, a, y] as vowels is particularly successful. It shows us a structure which is simple, regular, and bears the marks of being genuinely linguistic. These speak in its favour and, I hope, give it further support.
As to the specific values of [o, a, y] I can only guess at this time, so take the following as speculative. I spoke not long ago about the sounds we might expect in the Voynich language, and three vowels is really the minimum possible. The most likely layout is maybe one which maximizes contrast, such as /a, u, i/—this is the inventory of Classical Arabic (along with distinction for length, which might be encoded by another character).
The relationship between [a] and [y] could then be one of a differing vowel quality, such as height or backness. That might make [a, y] match /u, i/, with different environments changing the backness of the vowel (though which is which is unknown). However, would then almost certainly be a low vowel of /a/. I am wary of being so specific, but the logic runs to that conclusion. There may even be support for [a, y] as high vowels from the potential values of sequences, which I will discuss at a later time.