I don't agree with everything in the article or blog however. In particular, Zeke claims that Minoan loans in Greek that surface with the characteristic -nthos ending show that "it was unlikely that the Minoan language was like the Japanese", that is, in terms of phonotactic rules. This is derived, I believe, from a misunderstanding about the two languages.
Concerning the still uncertain theory that commonly identified words ending in Greek -nthos (ὑάκινθος, ἐρέβινθος, πλίνθος, etc.) come indeed from a specifically Minoan source, this may only imply a Minoan termination in *-inta, a sequence of syllables that is perfectly natural in Japanese syllabics where syllable-final -n is the only allowed coda consonant, as in 三 san 'three' and 一番 ichiban 'first, best'.
If, to the contrary, Minoan phonotactic rules mirror those of modern Japanese so closely, one may then wonder if Minoan Linear A actually dropped word-final -n in writing since such a rule would be a perfect source for the Linear B rule to likewise omit all of its more expansive set of coda consonants (eg. Linear B ko-wo for Mycenaean *kórwos 'boy'). As we can see, a rule like this in Minoan is minor and self-explanatory if there is only /n/ allowed in syllable codae, even more so if there is no phonemic contrast between a vowel-plus-nasal sequence and a nasalized vowel, whereas the same rule in Mycenaean produces the orthographic train wreck with which specialists must struggle.
Also, on the topic of PA-I-TO and its identification in both Linear A and Linear B as 'Phaistos', I'd like to suggest an alternative explanation that avoids inconsistency with the above observations. Putting aside all supposition, the important facts here are: 1) the Greek name shows medial -st-, 2) Linear A precedes Linear B, and 3) there is no doubt that Phaistos was a Minoan city. Facts therefore show us that Greek Φαιστός can only rationally come from a Minoan name. Yet if the Minoan name is written in Linear A as PA-I-TO just as in Linear B, how do we reconcile the inevitable consonant cluster!? Simple: We avoid taking the sequence -st- at face value and explore other possibilities in line with the aforementioned phonotactic restrictions. Namely, there is the overlooked potential that Greek -st- is metathetical and was meant to, albeit inaccurately, represent Minoan /t͡s/. From this suggestion, it might be extrapolated that the syllable TO was always pronounced /t͡so/ (merging therefore with ZO in spelling perhaps?). Strangely, Japanese too shows lenition of dental plosives neighbouring back vowels (ie. specifically, the high back unrounded vowel u). Are we seeing a mirror reflection? This hypothesis achieves the congruence we desire: Minoan *Paito /'p(ʰ)aj.t͡so/ > Mycenaean *Φaistó-.
What sleeve are you pulling this metathesis *ts > /st/ from? I thought Proto-Greek *ts (from pre-Greek *kj, *kʰj) normally yields /ss/ or /tt/, depending on the dialect.
ReplyDelete(Independantly, [tso] being the regular realization of /to/ doesn't fly as a free assumption either — a non-close back vowel is about the least likely to trigger affrication.)
If you want a nonsegmental feature interpretable as a cluster, one possibility that doesn't impose on the estabilish'd sound laws of Greek is preaspiration: /pajto/ [pʰajʰto]?
Still, I have to agree that given unwritten coda nasals, extending that rule to include unwritten coda /s/ is a preferrable solution.
(BTW: is it within your power to enable HTML tags such as <sup> in comments?)
Tropylium: "What sleeve are you pulling this metathesis *ts > /st/ from?"
ReplyDeleteThe same sleeve which bases its conclusions on reality. It's an uncontestable fact that loanwords from a particular foreign language, while showing regular sound rules between the donor and recipient language *overall*, need not in any particular case conform to these tendencies. Deviations from the norm need to be understood on a case-by-case basis and they're more common than you appreciate. Tendencies cannot be misused as steadfast rules.
Case in point, the Japanese triplet sutoraiku, sutoraiki and suto, all derived from a single English word 'strike' (Maynard, Japanese communication - Language and thought in context (1997), p.67). Evidently, to fully understand the reasons for this triplet, it's insufficient to merely know the *tendency* of English word-final stops to acquire a following -u in Japanese.
"Independantly, [tso] being the regular realization of /to/ doesn't fly as a free assumption either — a non-close back vowel is about the least likely to trigger affrication."
Then you'll be hard-pressed to explain the source of Japanese affricates, ts and z then. They surface neighbouring a [-front, +high, -labial] vowel (ie. u). I'd be delighted to know what your solution to this conundrum might be since it assuredly has nothing to do with [+front] environments or palatalization.
Given my explanation in the preceding paragraph, we would be naive to expect perfect sound mapping. While there is likely a tendency that a sound like Minoan *z /ts/ surfaces in Classical Greek as -ss-, such a tendency cannot be used to deny other possible mappings.
In the specific case of PA-I-TO, we certainly could choose to assume a Minoan cluster *-st-, thereby also assuming more complex syllabic rules. However, this is but a single case where that motivates us to this overcomplicated desperation. Furthermore given all appearances, Linear A was most likely made for Minoan and the rules of the script would simply be too unnatural if Minoan did not have highly restricted phonotactics to begin with. That way the scarcity of Minoan coda consonants perfectly explains why coda consonants weren't written when Greek took the spelling conventions to heart. If Minoan had coda *s or allowed onset clusters, we lose a sensible reason for the over-restrictive rules against all these very things in Linear B!
A sensible compromise is obvious: TO and TU might be automatically lenited in similar fashion to the development of Japanese dental affricates from their respective plosives neighbouring a [+back] vowel. With neutralization of to/zo and tu/zu, we don't require a Minoan cluster to explain later Greek -st-, thereby preserving the most logical phonotactic pattern of the language given all available evidence.
By the way, I can do nothing about Blogger's paranoid restrictions on HTML, sorry. For superscript symbols, try unicode symbols instead. Googling for "Online IPA keyboards" helps tremendously.
ReplyDeleteEgad, I missed an important issue here:
ReplyDelete"If you want a nonsegmental feature interpretable as a cluster, one possibility that doesn't impose on the estabilish'd sound laws of Greek is preaspiration: /pajto/ [pʰajʰto]?"
To pursue this end, I'd recommend explaining the rules behind this alleged preaspiration, particularly since in your example [j] and [tʰ] are not tautosyllabic. Why or how then does aspiration cross syllable boundaries? You seem to create more questions than you explain.
If Minoan had coda *s or allowed onset clusters, we lose a sensible reason for the over-restrictive rules against all these very things in Linear B!
ReplyDeleteThis is a point; but against *st, not for *ts. And appealing to variation in substitutions of foreign sound patterns does not help you to estabilish any expectation for an unmotivated substitution of a nativly-occurring sound *ts by a different one, *st.
But continuing with the third possibility of preaspiration… on the basis of this single example, there are a number of possible explanations for the distribution, for example after vowels and semivowel. This would need further investigation of /st/ clusters' distribution to state with any certainty however, but so would your idea of /st/ corresponding to Linear A T specifically before back vowels.
I don't understand your question of aspiration "crossing syllable boundaries"? [ʰt] would seem to remain firmly in the 2nd syllable.
"Independantly, [tso] being the regular realization of /to/ doesn't fly as a free assumption either — a non-close back vowel is about the least likely to trigger affrication."
Then you'll be hard-pressed to explain the source of Japanese affricates, ts and z then. They surface neighbouring a [-front, +high, -labial] vowel (ie. u).
Not at all. Close vowels naturally encourage a release with the tongue closer to the palate, ie. with more friction. Back vowels in general however, as you propose, do not. (Notice also that the common feature in Japanese /i/ and /u/, ie. the vowels inducing affrication, is precisely [+high].) I would like to see an example where /o/ too induces affrication, if you have one?
Granted, allophonic preaspiration is not quite normal either (and were it phonemic, it would not be any less of an assumption than a phonemic *ts or *st would), I'm only bringing this possibility up to point that *ts and *st are not the only viable options here.
Actually, here's a fourth that I think may be the most phonetically plausible yet: perhaps Minoan /j/ was devoiced before voiceless consonants, and this [ç] was in this case substituted by /js/.
Tropylium: "Not at all. Close vowels naturally encourage a release with the tongue closer to the palate, ie. with more friction.
ReplyDeleteThat's correct.
"Back vowels in general however, as you propose, do not."
No, I never proposed that [+back] was the cause, as you can see when you reread what you misread.
However, I admit that I may have misread your comments as well. Previously I was explaining that friction neighbours a [-front] vowel in Japanese in response to your statement that "a non-close back vowel is about the least likely to trigger affrication" but I realize that I overlooked the important word "non-close" which is evidently against my proposal of /to/ being affricated, not against /tu/.
Alright, so maybe we're on the same page afterall.
Now I admit that I'm fooling around with some very tentative ideas here as I try to grasp Minoan better but I can think of one way that TO could indeed be fricativized. It could be fricativized if, say, written O represents actual /uo/, a diphthong with a closed onset. For symmetry, E would then subsequently represent /ie/.
But regardless of my clever trickery, I think I may have to let this idea go anyway. The odds aren't looking good as I think this through more. Alas!
1) Are you aware that the Baybayin writing system of the Philippines regularly omitted all final consonants? The initial Spanish attempts to add a vowel-canceller to the system failed. Thus, dropping code-final consonants does not need much of an incentive.
ReplyDelete2) /st/ has a strong tendency to behave as a single consonant. For example, the best attested initial consonant cluster in Tiberian Hebrew is /št/, as in _štayim_. Thus 'Phaistos' need not be as great a problem as you thought.
Thanks, Richard. I was aware of the Baybayin script but not of this detail.
ReplyDeleteI just dropped by after a long absence to see what you might have posted recently and my eye was caught by the unexpected sight of the word "Baybayin". This is an area I have been researching in detail over the past three years. Perhaps a bit more about the script's non-spelling of coda consonants would be of some use.
ReplyDeleteThis is actually inherited from South Sulawesi, where both Bugis and Makassarese have no coda consonants other than a nasal with predictable realisations in different contexts (a nasal homorganic with the following consonant or [ŋ] word-finally), and a stop with equally predictable realisations (first half of a geminate, or [ʔ] finally and — in Bugis — before a voiced stop). In these languages, there is only a three-way contrast syllable-finally, unlike in Tagalog and most Philippine languages where almost any consonant can appear in coda position.
The restricted three-way contrast in South Sulawesi languages is the likely explanation for the fact coda Cs were not represented in spelling (apart from a series of rarely used NC letters introduced into the script at a later period). This was the original argument proposed by Harold Conklin for a likely (immediate but not ultimate) Bugis origin for the script, now reinforced by three further lines of evidence.
Interestingly enough, there is consistent evidence in three documents from Pampanga province northwest of Manila that the variety of the script evolving there had, by the early 1600s, begun to develop a set of indigenous conventions for representing coda consonants. This was completely unrelated to the '+ kudlít' virama introduced by a Spanish priest in the 1620 Ilocano catechism. So it appears that a desire to better match spelling to the phonetic content of the language spontaneously gave rise to these innovations. Had the script continued in use instead of being replaced by Latin script, I imagine that these innovations may well have spread and been adopted elsewhere, much as certain styles and letter shape variants appear in the same time period in widely separated locations.
A different case that may relate to your thesis here is the use of Arabic script to write Swahili. In that case, the nasal component of NC clusters was systematically not spelled, nor were glides in CG sequences. This is another case, then, where spelling of consonants was restricted to the "core consonants", leaving out (relatively predictable) cluster members of higher sonority where other information was already available from the spelling.