3 posts tagged “attic greek”
I'm on a bit of a Greek spree at the moment, because next week I'm starting Homeric Greek and historical grammar of Greek, so I'm just getting back into it.
Today I'd like to talk about Greek geminates and especially the absence of those in the spoken language.
Two common geminates especially drew my attention, namely: Attic ττ and Ionic σσ, which actually have the same root, look for example at Attic θαλαττα 'sea' vs Ionic θαλασσα 'id.'.
In normal analyses we would just say they were the lengthened form of τ and σ respectively. But I have reason to doubt this. To understand this, we must understand what exactly is the origin of these geminates, for this I'll have to consult both PIE and Greek verbal flection. I will focus on Attic Greek, but as far as I know replacing every ττ with σσ should yield you the Ionic counterparts.
I'll state three present tenses of three verbs, and their respective verbal stems.
φυλἀττω φυλακ-
τἀττο ταγ-
ταρἀττω ταραχ-
As you can see the underlying stem is always a velar. it is never a dental. These present stems were formed with a PIE present suffix *i(e/o) Simply put these are the shifts that took place from PIE to Greek
*-k-i(e/o)- > ττ(ε/ο)-
*-g-i(e/o)- > ττ(ε/ο)-
*-gʰ-i(e/o)- > ττ(ε/ο)-
Root of these PIE velars can actually also be palatovelars and labiovelars, but for now this'll be complete enough. now there is really no good phonetic reason why a [k] would change into a [t] after a [j], and especially not a [t:]. Although, when one of the two consonants is lost, there often is compensatory lengthening, this is notable not as common when we speak of clusters with [j] or other semi vowels for that matter. often [j] palatises the preceding consonant and then disappears itself. Then what would <tt> stand for? Well obviously the Greek writers were not in the possession of a letter for [c], so they had to use the alphabet they had, apparently to Greek ears [c] sounded closer to [t] than to [k], so they modified the letter <t> by writing it twice.
Later [c] probably shifted to an affricate [tʃ], which gives a lot more credibility to the shift ττ > σσ in Ionic. It's almost unheard of that a languages has this shift [t:] > [s:] but not [t] > [s]. Hebrew comes to mind though that had [t] > [θ] and [tt] > [t]. But this is not quite the same shift. With the 'doubling is modifying' hypothesis, also σσ is nicely analysed as [ʃ].
The only mystery that remains though is how did the voiced 'g' loose it's voice, It's not an obvious shift. The answer doesn't lie in g, but actually in the suffix -i(e/o)-.
Another series of presents with their respective stems:
κὀπτω κοπ-
βλἀπτω βλἀβ-
ῥἰπτω ῥιφ-
As you can see, also the labial stems underwent this devoicing when followed by [j]. Here [j] was also assimilated to a stop, rather than the approximant it was. They only thing we can conclude from this is that [j] was not actually voiced, or at least not in this environment, else it would be incredibly hard to explain it's devoicing traits of the letter. Maybe /j/ was more like [ç]. Another possibility is to assume analogy took place, but I myself don't think this is a very likely instance of analogy.
These labial stem verbs though, do have some proof that /j/ might was able to change into a stop after another stop, which has some influence on my statement that <tt> cannot represent a longer version of the consonant, for /j/ isn't likely to affect length. Nevertheless the actually phonetic value of <tt> is very well disputable because [t] is not likely to come from [j], or proposed value [ç] for that matter.
So to conclude: I believe <tt> nor <ss> had the phonetic values [t:] and [s:] respectively, but rather [tʃ] and [ʃ]. To explain the odd devoicing you can either assume the /j/ had a true phonetic value of [ç] (at least, in some environments) or you could assume analogy has taken place.
Thank you for your time, and feel free to burn my theory down to the ground.
Today I was thinking about classical Greek dialects, and especially what happened to the 'waw' or digamma 'ϝ' which had a likely phonetic value of [w] it comes directly from a consonantal form of *u (some might transcribe *w, I don't) in Proto-Indo-European.
Most people familiar with classical Greek will not be familiar with the digamma, because it disappeared before the time that the bulk of Greek texts were written. The sound law it followed was very straightforward ϝ > Ø, it just disappeared in every context. Good example is the root νεϝο- 'new' which became νεο- ('New' is also related to the same PIE root *neu-).
This is all good, except there is one very strange chronological problem I bumped into. I'll be talking in particular about the Ionic and Attic dialect. These two dialects are particularly close, but there is one big difference, the Attic dialect tended to blend vowels together, where Ionic kept the original contrasts.
A good example for this is the word for 'light' which in Ionic is φάος but in Attic contracts to φῶς. Now there's also a typical contraction in Attic of εο > ου. But νεος does not contract to **νους. The reason why this does not happen can only be explained by assuming that these contractions took place before the ϝ was dropped. If the ϝ is there, the two vowels aren't adjacent so they can't contract to another vowel. That much makes sense.
But that these contractions took place before the dropping of ϝ is a bit of a problem. The most distinguishing part of Attic in comparison to Ionic is the contraction of vowels. But apparently after there has been so much dialect differentiation between the two, the sound change apparently still managed to act out upon both of the dialects. I'm not saying this is impossible, but you often see, that once dialects start to diverge, they tend to not come back together again, and especially not for such a fundamental shift.
Another problem is, that vowel contraction in Attic seemed to be an active process, even after ϝ was dropped. Especially in Attic poetry you can see that contractions are split up again on the go to support the meter. The above mentioned φῶς for example, is found in Hippolytus in both it's contracted as non-contracted form.
If such a sound law of contraction is still active although ϝ was lost, you'd expect that the sound law will also have effect on words that have lost their ϝ, after all, now two vowels are touching, and contraction is an active process. The fact that Attic poets were able to reconstruct the non-contracted forms without making mistakes, even shows that they were quite aware of the underlying phonemes, although this is not at all obvious in most forms.
How do we solve this anachronistic problem?
The only thing I can really think of, is that Attic had indeed merged all it's mergings, and the ϝ stopped it from happening because the contractions took place before the ϝ drop. φάος probably wasn't an existing word in Attic at all, but since the poets were of course well educated, they may have had knowledge of the Ionic dialectal words, and might have even had some insight that contractions had taken place. Thus only the more educated people had knowledge, and insight to be able to use non-contracted forms. These non-contracted forms were probably no where near different enough to be completely misunderstood by the Attic civilisation, especially not in the sparse use it had.
This still isn't completely satisfactory, because I could've sworn that ϝ > Ø took place way before there was any strong dialectal variation, but I guess my memory refuses to serve me properly.
Sorry for the late update, and sorry for the rather inaccessible update. I have a word-per-word glossed Classical Greek text from Lysias. I'm sorry to say though, that for now, it's in Dutch, since I use it as my homework in my Greek classes. But I figured it'd be better than updating nothing at all. So here it is.
