Winter's Law in Balto-Slavic, "Hybrid Theory" and phonation - Part 2

(Continued from Winter's Law in Balto-Slavic, "Hybrid Theory" and phonation - Part 1)

So given the problems and confusion as I previously mentioned in Part 1 of this rant, I want to try out my latest revelation to foster open discussion. It pretty much sounds like my old view, but now enriched with extra details.

Let's first assume that Winter's Law exists in some form or another, despite the academic bickering on the exact details. Let's also assume that everything in PIE began with ejectives, and yet that ejectives only existed in some stage of Pre-IE, not PIE itself. Finally, let's assume that whatever phonetic qualities PIE *d and *g had, they were in some way recent derivatives of ejectives. Here below, I'll use the dental stop series of *t, *d and *dh as examples of the general 3-way contrast of PIE stops between what is traditionally described as "voiceless", "voiced" and "voiced aspirated".

Now, while Glottalic Theory explains that PIE's purported ejectives had mostly eroded into implosives or "preglottalized" phonemes in almost all dialects without explaining why exactly voiced *d was the overwhelmingly typical end result, my Hybrid Theory position has been that these stops were already voiced in PIE. Yet unlike Traditional Theory, I deduce that *d could not have simply been "plain". Rather, I feel that *dh should be considered "plain" for the same reasons as Glottalic Theory. Up to now, I've described this contrast as a three-state distinction of voice. I presumed that the difference between *d and *dh must have been in regards to the onset of voicing (VOT), the former being "semi-voiced" (as "d" is spoken in English) and the latter being "fully voiced" (as "d" is spoken in French). I've looked to languages like Thai and Korean as real-world examples of similar phonological constructs.

However, I have an even more accurate definition of *d and *dh that doesn't involve voicing as a marked feature at all. Instead, let's ponder on the possibility that *d was not merely "semi-voiced" but also had a marked phonation.^[1] "Phonation" describes the way in which a sound is voiced and basically measures how loose or stiff vocal chords are as a sound is being pronounced. Stops in English, French and most other languages are described as "modal" (a.k.a. "plain voicing"). This is the default. However there are stops which are "creaky" or "breathy" as well. Creaky stops lie on the "stiff" end of the spectrum and sound just as they're described, while breathy stops are positioned on the slack side. Modal stops are in the middle, neither too stiff nor too slack. So let's explore the implications of a PIE model with creaky *d (i.e. "stiff" /d/) and modal *dh (i.e. plain ol' /d/).

In relation to PIE phonology, creaky voiced stops may seem like the "semi-voiced" stops I'm looking for because of the accompanying crackle in the throat that intermittantly allows air through a flapping narrow passage in the throat. Think of the sound as a voiced /d/ hidden behind a stream of glottal stops produced many times a second that turns this voicing on and off, giving it a distinctive, rumbling pitch. The weakening of ejectives to creaky voiced stops is a perfectly natural development^[2] that only involves a marginal slackening of the glottis to allow some air through. The passage of air allows the possibility of voicing to exist afterall, which is not possible with the complete glottal closure of ejectives. No passage of air, no voice.

So if PIE *d were a creaky voiced stop, this would have implications for how we interpret Winter's Law in Balto-Slavic. If we take Kortlandt's account of it at face value, we would be expected to come up with a voiceless sound to explain the genesis of rising tone before *d. Yet, what if Kortlandt is wrong about Winter's Law as Miguel had surmised in the quote that I cited in Part 1? There's no guarantee that he's correct and there is no clear consensus on the issue so far, so let's assume instead that Kortlandt's formulation of Winter's Law is hiding a different reality, that Winter's Law is a rule that governs pretonic syllables only^[3]. Now, we're not required to be bound by rules of tonogenesis because a syllable preceding a stressed syllable with falling or high-level pitch might very well develop a rising accent. How? Considering that the unstressed syllable would be relatively lower in pitch, the transition from the unstressed syllable's low pitch to the high pitch of the onset of the next syllable will unavoidably cause some degree of rising tone. With a shift in the position of stress, this non-distinctive rising tone in the unstressed syllable may eventually become distinctive from the pre-existing default accent of falling or high-level pitch.

If this is the more correct formulation of Winter's Law, then all we need to explain is the source of lengthening. However, this is now a piece of cake since creaky stops are known to lengthen preceding vowels by way of their accompanying laryngealization.


NOTES
^[1] Yallop/Fletcher, An Introduction to Phonetics and Phonology (2007), p.53 (see link).
^[2] Fallon, The Synchronic and Diachronic Phonology of Ejectives (2002), p.285 (see link): "Or an ejective may undergo laryngealization in which the glottalic component bleeds into creaky voice so that eventually the whole segment becomes voiced and laryngealized, and eventually simply voiced." This seems to be precisely the evolutionary path taken by PIE: Pre-IE ejectives turn into creaky voiced stops in PIE which turn into plain voiced stops in most dialects of PIE. Since Glottalic Theory insists on ejectives in the PIE stage itself, it adds an extra phonetic step that isn't necessary to describe the data.
^[3] See Shintani, On Winter's law in Balto-Slavic, Apilku 5 (1985), p.273-296. The claim is that Winter's Law only operates on unstressed short vowels.