An interesting side-effect of obsessing over these correspondances between Proto-Indo-European (PIE) and Proto-Semitic (PSem) is that I've been noticing some potentially interesting and very minute details about Proto-Semitic pronunciation. There's one issue that's starting to get me excited involving the exact nature of “emphatic” stops.
From what little I've admittedly read on Proto-Semitic, my understanding so far is that emphatic stops are considered to have originally been either ejective stops (as in Amharic) or pharyngealized stops (as in Arabic). Regardless of which one they were, they apparently derive from the ejective stops of Proto-Afro-Asiatic, the ancestral proto-parent of the Semitic, Egyptian, Chadic, Cushitic and Berber languages. Interestingly, the Mid IE correspondances that I've identified so far seem to suggest to me that, rather than having ejectives stops, Proto-Semitic had pharyngealized stops as in modern Arabic. The reason why I think this regards equations like PSem *ḥāniṭu “ripening” based on the triliteral verb root *ḥnṭ “to ripen” (c.f. *ḥinṭu “wheat, barley”) and Mid IE (MIE) *xénda “to blossom” (> PIE *h₂endʰ-).
To an Indo-Europeanist or Nostraticist who may be simultaneously of the belief of both a Glottalic IE and a Glottalic Semitic, the two words may be associated only with some difficulty despite congruent semantics because PIE *dʰ is supposed to be underlyingly plain /d/ while Semitic emphatic *ṭ is presumed to be an ejective /tʼ/. However, I don't think this is the only rational option.
As stated earlier on my blog, I've come to the conclusion that MIE's inherited ejective stops had already deglottalized to stops with creaky voice, opposing the plain-voiced stops (i.e. The ones traditionally written with superscript “h”). In other words, I believe there was an intermediary stage lasting from the Mid IE period to well into Fragmenting PIE when the traditionally-described “plain stops” (or rather, the “ejective stops” of the Glottalic Theory camp) were in fact creaky voiced stops (i.e. half-voiced stops). Thus, even if PSem had ejective stops, MIE speakers should be expected to find more native approximates to these foreign sounds. And indeed, this appears to be the case from the examples I've shared so far in my continuously edited pdf of Semitic loans in PIE.
However, we still have problems equating these two aforementioned lexemes if one remains resolute in this adapted belief that MIE had creaky-voiced stops while PSem had ejective ones since it's hard to explain away such a phonetically implausible replacement of an entirely unvoiced ejective stop with a creaky voiced stop by any innocent speaker no matter how foreign they may be to the exotic sound of ejective stops. Yet, if we allow our minds to consider the possibility of pharyngealized emphatics in Proto-Semitic where concurrent voicing would still be possible, albeit delayed a moment after stop closure, then the mystery of the equivalence between MIE *d /d/ and PSem *ṭ /tʕ/ in the above example immediately disappears. In fact, elegantly so if I do say so myself (and I will because I'm playfully cocky that way). This can also explain what would otherwise be problematic correspondances regarding MIE creaky-voiced *g̃(ʷ) and PSem *q (if /kʕ/) as in the example of PSem *qawāmu “to rise up” vs. PIE *gʷem- “to come”.
From what little I've admittedly read on Proto-Semitic, my understanding so far is that emphatic stops are considered to have originally been either ejective stops (as in Amharic) or pharyngealized stops (as in Arabic). Regardless of which one they were, they apparently derive from the ejective stops of Proto-Afro-Asiatic, the ancestral proto-parent of the Semitic, Egyptian, Chadic, Cushitic and Berber languages. Interestingly, the Mid IE correspondances that I've identified so far seem to suggest to me that, rather than having ejectives stops, Proto-Semitic had pharyngealized stops as in modern Arabic. The reason why I think this regards equations like PSem *ḥāniṭu “ripening” based on the triliteral verb root *ḥnṭ “to ripen” (c.f. *ḥinṭu “wheat, barley”) and Mid IE (MIE) *xénda “to blossom” (> PIE *h₂endʰ-).
To an Indo-Europeanist or Nostraticist who may be simultaneously of the belief of both a Glottalic IE and a Glottalic Semitic, the two words may be associated only with some difficulty despite congruent semantics because PIE *dʰ is supposed to be underlyingly plain /d/ while Semitic emphatic *ṭ is presumed to be an ejective /tʼ/. However, I don't think this is the only rational option.
As stated earlier on my blog, I've come to the conclusion that MIE's inherited ejective stops had already deglottalized to stops with creaky voice, opposing the plain-voiced stops (i.e. The ones traditionally written with superscript “h”). In other words, I believe there was an intermediary stage lasting from the Mid IE period to well into Fragmenting PIE when the traditionally-described “plain stops” (or rather, the “ejective stops” of the Glottalic Theory camp) were in fact creaky voiced stops (i.e. half-voiced stops). Thus, even if PSem had ejective stops, MIE speakers should be expected to find more native approximates to these foreign sounds. And indeed, this appears to be the case from the examples I've shared so far in my continuously edited pdf of Semitic loans in PIE.
However, we still have problems equating these two aforementioned lexemes if one remains resolute in this adapted belief that MIE had creaky-voiced stops while PSem had ejective ones since it's hard to explain away such a phonetically implausible replacement of an entirely unvoiced ejective stop with a creaky voiced stop by any innocent speaker no matter how foreign they may be to the exotic sound of ejective stops. Yet, if we allow our minds to consider the possibility of pharyngealized emphatics in Proto-Semitic where concurrent voicing would still be possible, albeit delayed a moment after stop closure, then the mystery of the equivalence between MIE *d /d/ and PSem *ṭ /tʕ/ in the above example immediately disappears. In fact, elegantly so if I do say so myself (and I will because I'm playfully cocky that way). This can also explain what would otherwise be problematic correspondances regarding MIE creaky-voiced *g̃(ʷ) and PSem *q (if /kʕ/) as in the example of PSem *qawāmu “to rise up” vs. PIE *gʷem- “to come”.
Indeed an elegant solution.
ReplyDeleteI must say though, that I do have some problems with the semantics of 'To rise up' > 'to come', is there any precedent of such a semantic shift?
And there seems to be another problem with this word. I assume that you got the IE labio-velar from the combination of q+w.
But why then, didn't *h1ékuos get a labio-velar, but a series of a (palatal-)velar and a *w?
Of course, in *h1ékuos the *k and *u were already touching
*h1ékua=sa, while qawamu first had to undergo syncope (qawamu > qawému > qawéma > qwém- > *gʷém-). But that hardly seems like a convincing difference in phonetic environment to explain this. Although arguably, I guess it is possible. I'd just like you to confirm that that is what you had in mind ;-)
Phoenix: "I must say though, that I do have some problems with the semantics of 'To rise up' > 'to come', is there any precedent of such a semantic shift?"
ReplyDeleteNote that PIE *bʰeuhₓ- itself fundamentally means "to grow (from the ground)" and hence "to arise; to come up". This has then shifted to "to become". When something "arises", we may also say that it "comes about".
To boot, the very handy Encyclopedia of Indo-European Culture remains unspecific about which laryngeal this root contains. Considering my comparison to Semitic triliteral *bwʔ, it makes me wonder if I can get away with even more direct phonetics using *h₁, instead of *h₂ as I had been proposing. Thus perhaps *bʰeuh₁-?
Phoenix: "And there seems to be another problem with this word. I assume that you got the IE labio-velar from the combination of q+w."
Yes, but you have the details wrong. I propose that the unstressed pretonic syllable *qaw- in the Proto-Semitic word was already misheard as a labialized creaky-voiced velar *gʷ when it was borrowed into Mid IE. Also I propose that in Mid IE, *q didn't exist as a seperate phoneme and could only surface on the phonetic level in stressed syllables neighbouring *a as an allophone of underlying *k. Here then, *q(ʷ) would never be a possible MIE reflex of Semitic *q /kˁ/ without changing the core vowel to *a.
All right, nice that works for me. Only thing that puzzles me is that the Indo-Europeans seem to have loaned some very essential verbs. If all of these words come to be accepted as actual loans, the influence of the Semites on Indo-European must have been enormous.
ReplyDeleteYes, which brings us back to what I commented in March 2008: Semitic and IE in the Neolithic: How intensive was the language contact?.
ReplyDeleteThe distribution of ejectivs in modern Semitic also hints of the possibility of PS pharyngealized emphatics. The cross-linguistic evidence is that ejectivs are quite stable, certainly they're much more common than pharyngealized consonants. So where do we find the sound change to have not reached, or to have been reverted in? South Semitic, exactly the area in contact with other languages that contain ejectivs. It's not much by itself, but certainly supports the theory.
ReplyDeleteBut I would still expect to see a PS voiceless pharyngealized stop reflected as a MIE voiceless stop. You seem to be suggesting a "cluster" pronunciation with a separate, voiced pharyngeal (that, or your superscripts are broken). The problem with this is that AFAIK any actual PS clusters of such sort don't turn into emphatics!
Aluckily, there is evidence for a fully voiced pronunciation of emphatics from Arabic. In addition to ḍād and ẓā', I suspect this of qāf (to explain its dialectal reflexes as /g/, as well as the seemingly random palatalization in jīm). (Also note ḍād corresponding to `ayin in Aramaic.)
Even then, we may wonder if such sounds wouldn't be better approximated by MIE's own "voiced emphatics" (the creaky voiced stops)? A possible solution is that the Semitic de-ejectivization actually predates that in IE...
There's another possible problem with this version of the hypothesis, namely why the loss of creakiness common in IE affects no Semitic language, and most of them rather devoice the sounds again. But let's first hear if this makes any sense to you to begin with?
Tropylium: "But I would still expect to see a PS voiceless pharyngealized stop reflected as a MIE voiceless stop. You seem to be suggesting a 'cluster' pronunciation with a separate, voiced pharyngeal (that, or your superscripts are broken)."
ReplyDeleteOne may use superscript ayin, or another alternative is to use superscript "ɒ". At any rate, my notation is acceptable and used by other authors. The notation I'm using seems to be what's preferred on Wikipedia (not that this says much :P) but frankly, I quite like this symbol the best.
"The problem with this is that AFAIK any actual PS clusters of such sort don't turn into emphatics!"
I never suggested such a thing. Even if we ignore legal Proto-Semitic syllable structure, there's a phonetic difference between /kˁ/ and /kʕ/ anyway.
"Even then, we may wonder if such sounds wouldn't be better approximated by MIE's own 'voiced emphatics' (the creaky voiced stops)?"
That's precisely the equivalence I expect the most. However, loanwords can't be expected to show regular reflexes in the same way as inherited words and so I believe that the context of the sounds must also be examined on a case-by-case basis. Such is certainly the case of PIE *sweḱs "six" with its voiceless velar *-ḱ- when compared to its Semitic equivalent with word-medial voiced alveolar coronal *-d- (and I've explained the exact mechanics of this seemingly odd remapping before on this blog).
In the case of Semitic *ḥinṭu, *n neighbours emphatic *ṭ which could theoretically sound to a foreigner as though the following stop were voiced, regardless of the actual pronunciation of the native speaker.
"A possible solution is that the Semitic de-ejectivization actually predates that in IE..."
On the surface, that doesn't seem very objectionable at all. I'll have to ponder on the implications of that alternative.
"There's another possible problem with this version of the hypothesis, namely why the loss of creakiness common in IE affects no Semitic language,[...]"
Why would it? The influence seems to me to be one-way only. Only PIE was affected by Semitic, not vice-versa.
Chalk the "pharyngeal cluster" issue to dodgy fonts; they now display fine, but on the last time, they were displaying exactly as /kʕ/, not /kˁ/. :/
ReplyDeletePhonotactical reshaping is an option, sure, but aren't *nd and *nt just as valid in PIE as *ndʰ? OTOH I would however blame this, not the phonetically minute alveolar/dental difference, for the dθ > ks change.
And as for the emphatics' fate, no, I'm not after a sound change that radiates from IE to Semitic. Rather, if we want to explain these loans by a stage of Semitic with voiced emphatics, we'll then need a later stage of emphatic devoicing. In IE however, we have almost universally loss of creakiness insted. Rather different fates for rather similar sounds. I'd expect one outcome to dominate in both families, kinda like when you have palatal stops, you can expect them to eventually turn into sibilants of some sort. I can't really think why would pharyngealization cause voicelessness.
Tropylium: "Phonotactical reshaping is an option, sure, but aren't *nd and *nt just as valid in PIE as *ndʰ?"
ReplyDeleteYes, but you're looking for an absolutely *regular* rule and that's just not realistic.
Take Japanese for example. Many words have been loaned from English words ending in syllabic /r/. Sometimes the words drop the "r" (e.g. コンピュータ kompyuuta "computer") and yet sometimes the words keep the sound intact (e.g. カイゼル kaizeru "Kaiser").
Read Phonetics or Phonology: Asymmetries in Loanword Adaptations - French and German Mid Front Rounded Vowels in Japanese [pdf] by Katrin Dohlus of Kobe University discussing how languages remap foreign sounds. It's not as straight-forward as you seem to think it is and unfortunately, that can't be helped.
"Rather, if we want to explain these loans by a stage of Semitic with voiced emphatics,[...] we'll then need a later stage of emphatic devoicing."
Regardless, voice in pharyngealized stops is non-contrastive in Proto-Semitic (unlike Arabic) so it seems to me that musing on a theoretical "devoicing" of such voiced emphatics (as you suggest) which would thus lack voiceless counterparts is probably leading us astray.
Tropylium: "In IE however, we have almost universally loss of creakiness insted. Rather different fates for rather similar sounds."
Such is the wacky world of linguistics. Every language defines its own path, much like people.
"I'd expect one outcome to dominate in both families, kinda like when you have palatal stops, you can expect them to eventually turn into sibilants of some sort."
One can't predict too much of how any language will change. Depalatalization of palatal stops (i.e. /kʲ/ > /k/) is also just as possible as affricatization (i.e. /kʲ/ > /tʃ/). Gemination (i.e. /kʲ/ > /kk/) can be yet another eventual result. There are many possibilities so one shouldn't feel that there's only one fate for a linguistic feature.
Did you have a specific point in mind from that pdf? Its main points seem to be that loanwords can be adapted either phonetically or phonologically. I don't see how *nṭ > *ndʰ with traditional, or even glottalic PIE sound values would be either. There may be room for variation, but not anything goes.
ReplyDeleteAnd where exactly do you see us "going astray"? On a sidetrack, sure, but a potentially useful one.
As for sound changes are concerned, no, we can't "railroad the plot", but implausible changes should be avoided just as well. I don't like setting up odd diachronic changes by thin evidence any more than reconstructing unlikely phonological systems.
Turning palatalization into gemination, BTW, is completely ridiculous - does that really happen somewhere or are you picking that from something like the Germanic gemination-before-a-sonorant pattern?
"Sometimes the words drop the "r" (e.g. コンピュータ kompyuuta "computer") and yet sometimes the words keep the sound intact (e.g. カイゼル kaizeru "Kaiser").
ReplyDeleteThis is due to variation within English (between rhotic and non-rhotic dialects), not irregularity in the mapping of foreign sounds. Anyway, kaizeru is from German, not English.
Why does emphatic /t/ correspond to IE *dh but emphatic /k/ (i.e. /q/) correspond to IE *g(W)?
ReplyDeleteRob: "Why does emphatic /t/ correspond to IE *dh but emphatic /k/ (i.e. /q/) correspond to IE *g(W)?"
ReplyDeleteIn the case of the equation of PIE *gʷem- and Semitic *qwm, the *q would technically be the reflex of *q *and* the following *w in the initial unstressed syllable of the Semitic word.
Lameen: "This is due to variation within English (between rhotic and non-rhotic dialects), not irregularity in the mapping of foreign sounds. Anyway, kaizeru is from German, not English."
ReplyDeleteReally? No irregularity in the mapping of foreign sounds you say? So, do you dare suggest that foreign speakers hearing a language for the first time will reproduce the sounds faithfully then? That's absurd.
What about Japanese ファンファーレ fanfaare for English "fanfare" (instead of what would be expected *fanfaa based on the general pattern). Note also English loans in Cantonese that likewise show variable treatment of English word-initial aspirated stops.
Please understand that I do believe that there's a general pattern to look for in these interlingual loans but it also must be understood that things can't be expected to be regular all of the time.