Pronunciation of rough breathing after delta

[katalogon]

http://aoidoi.org/articles/GrkIambicComp-23-Apr-06.pdf

There is something to do with pronunciation in tragic that has been bothering me for a while. If you look at example 3.9 on page 7 of Chad's notes:

σφάζει δ' ὁ δοῦλος, / βοῦν τιν' ὥς, τὸν δεσπότην.

The second foot is δ' ὁ δοῦ. I'm wondering how to pronounce the δ' ὁ. On pages 54 and 55 of Allen, it is clear how to do something like this if the consonant preceding the aspirated vowel is a voiceless plosive (such as τ) - it would just be converted into its aspirated counterpart, θ. However, δ does not have an aspirated counterpart in Greek.

One thing that I could do is pronounce the δ' ὁ as a voiced aspirated plosive (dho) as in Sanskrit, but that would mean introducing a phoneme which is foreign to Greek. To avoid that phoneme, you would have to introduce a schwa between the δ and the ὁ to keep the aspiration out of the plosive, and the effect would be to approach an anapaest.

A final possibility would be to devoice the ὁ, similar to the idea of devoicing in ῥ.

Does any one have any thoughts on this?

A similar problem (for me anyway) is with the fourth foot, τιν' ὥς.

[Vasile Stancu]

For the pronunciation of δ, I have borrowed the sound used in Modern Greek, i.e. ð; therefore, I would pronounce δ' ὁ like ðho.

[mwh]

There was a thread on this very question by Sean—“Pronunciation of δʼ ὅτε.” I’d link to if I could remember how.

[katalogon]

Thanks! I found it, Nov. 10, 2019. Exact same question!

[jeidsath]

It would be useful to find a example of Grassman's law being kept/violated in a compound with a supposed aspirated δ. Maybe something with δρ? But I don't think that conceals an aspiration (εδρα would be a violation, if so).

Anyway, this morning (Palm Sunday), my wife was singing "No Tramp of Soldiers' Marching Feet" (by the great Cambridge-educated Timothy Dudley-Smith, who can be found on Youtube discussing his hymns). She got to the line:

"...and all the ransomed host proclaim..."

And my oldest interrupted:

"You said 'ransom toast'!"

Apropos, I thought. Write that sound confusion in Greek and it's "δ' ὁ" and "θο". In English at least they are equivalent. For the Greeks though, probably not, or they would have written it with the θ (at least some of the time).

[Menoeceus]

One thing that I could do is pronounce the δ' ὁ as a voiced aspirated plosive (dho) as in Sanskrit, but that would mean introducing a phoneme which is foreign to Greek.

The voiced aspirate pronunciation [dh] would be an allophone, not a new phoneme. The term 'phoneme' implies that the sound that distinguishes one word from another, but this phenomenon occurs between words. It is not uncommon for a languages to be pronounced with allophones that are not phonemes, particularly in connected speech. In many varieties of American English, for example, the letter 't' is pronounced as a glottal stop (denoted [ʔ] in the International Phonetic Alphabet, a catch in the throat sound) at the end of a word -- but as an alveolar flap (denoted [ɾ], similar to the single 'r' in Spanish, although produced a bit differently) before another vowel. So 'the cat sat on the log' is pronounced 'the ca[ʔ] sa[ɾ] on the log' in connected speech. Neither of these sounds are phonemes in English, because the meaning does not change if one does pronounce an actual t-sound; English speakers hear them both as t-sounds. In careful or non-connected speech, speakers will produce the t-sound: 'the ca[t] sa[t] etc.'

A connected issue that occurred to me is the pronunciation of οὐδείς/οὐδέν. Here's what Threatte has to say:

39.01 ούθείς, ούθέν, etc.

The spellings ούθείς (μηθείς), ούθέν, ούθενός, etc. first appear in a state decree of 378/7 and the forms with θ are the only ones in use by the end of the fourth century; in II2 885.5 read ούθέ?]ν and in II2 1299.8 μηθενός, cf. add. to IG II2 and no. 95 infra. That there can be no question of confusion between ούδέ and οΰτε here is shown by the exclusive use of ούδεμία, etc. in the feminine (cf. nos.74, 117, etc. with ούδεμία alongside ούθείς). One decree of 330/29 (nos.27, 76) has both ούθενός and μηδένα.

The original spellings with δ begin to reappear in the first century B.C. and are normal throughout the Roman Period. But forms with theta occur in both of the texts of regulations for the cult of Menes Tyrannos set up by Xanthus of Lycia, of which only the first contains peculiar non-Attic spellings (s. Ip.; four cases, nos. 122—123), and once in an eph. cat. of the late second century (cf. no. 124).

The final δ in this compound joined so closely with the following aspiration that it became an aspirate stop; the value of θ here was probably [d'] rather than [t'], since θ = δ + [h] in these words has developed like δ elsewhere to produce [δ] in MG, while θ has undergone a different development to [θ]. The reintroduction of ουδείς in the Roman Period is probably a characteristic learned revival of the original orthography.

There is probably one example of ούθ' for ούδ' before a following aspiration (no. 125 infra), and μηθαμοῦ occurs once on a defixio of ca. 300 B.C., but otherwise the use of theta is confined to ούδείς/μηδείς.

These words are sometimes written separatim, especially in the fifth century, cf. Syntax.

I don't think I quite agree, since this would in fact involve a phoneme /dh/≠/th/. I can imagine a few possible explanations though:

1. that Attic did in fact have voiced aspirates as marginal phonemes -- or at least a voiced dental aspirate. English also has at least one marginal phoneme: the voiced dental fricative in 'the,' 'this,' 'they,' etc. (Many English speakers are not aware that this is a different th-sound than the one in e.g. 'thin.') This voiced dental aspirate may only have occurred in compounds of ούδέ/μηδέ; similarly the Egnlish voiced dental fricative only appears in a few function words.
2. that the allophonic [dh] came to be pronounced as [th] in the compounds; later spellings with δ could reflect pronunciation after masc. and neut. were remodelled on analogy with the feminine ούδεμία and by decomposition into the component parts ούδέ, εἷς/ἕν. An English example of such remodelling is the pronunciation of the days of the week. My New Englander grandmother still says 'Sundy,' but her children say 'Sunday' ('sun' + 'day') etc.
3. that the aspiration would simply be lost by progressive assimilation.

It would be useful to find a example of Grassman's law being kept/violated in a compound with a supposed aspirated δ.

Threatte also has information on aspirate dissimilation, sections 38.01 and 38.011 on pp. 449 ff. I don't see anything there about supposed voiced aspirates. But I think to answer katalogon's question a compound won't help. We need a word with an aspirate in the penultimate syllable and a voiced plosive in the ultima immediately followed by a short vowel that is elided before an aspirate, no?

[mwh]

Just one simple point, then, or two. (1) The fact that aspirated δ is not alphabetically represented doesn’t mean it didn’t exist. The correlation between alphabet and phonology, while closer than in most languages, can never have been perfect, and (2) was certainly not stable over time.

[jeidsath]

But I think to answer katalogon's question a compound won't help. We need a word with an aspirate in the penultimate syllable and a voiced plosive in the ultima immediately followed by a short vowel that is elided before an aspirate, no?

I don't agree. Here are two Grassmann's law violations, if we can take δ as dho from compounded -ἡγέω:

ἐφοδηγέω
καθοδηγέω

Of course, these are some pretty out-of-the-way words. And who knows if their spelling reflects pronunciation? But maybe others can find other examples.

***

If we want to reason from this very insufficient data, I'd speculate that voice overcomes aspiration in Greek where they come into contact (compare Verner's law). This would be the opposite of English, where aspiration overcomes voice (see my "ransom toast" example above).

***

EDIT: It turns out that the above examples are from ὁδηγός or ὁδηγέω, which are more common, and show the (theoretical) Grassmann's law violation just as well.

[Hylander]

The dissimulation of voiceless aspirates known as Grassmann's law operated at an early stage of Greek. At some point it ceased to be operative, at least uniformly, and analogy came into play to form passives such as εχυθην.and Attic words such as πανταχοθεν. (I find this in Lejeune, Phonétique historique du mycénien et du grec ancien, sec. 45.)

ἐφοδηγέω and καθοδηγέω and οδηγέω itself are clearly compounds and were probably put together more recently than the time when Grassmann's law applied to voiceless aspirate,. So I'm not sure whether they tell us anything, one way or another, about voiced aspirate occlusives, even assuming such sounds, which are of course not reflected in the orthography, existed. Voiced aspirate occlusives are reconstructed for Proto-Indo-European, but they became voiceless in Greek.

One other factor that may (or may not) be relevant: the general tendency of Greek towards psilosis. Classical Attic was an exception, but there's evidence that even the "original" language of the Homeric poems was psilotic. We can't tell from manuscript texts because the texts were to some extent Atticized by scholars.

But even in Attic, one wonders whether aspiration/rough breathing followed by a voiced consonant simply disappeared. The fact that our manuscript texts are marked with a rough breathing over the vowel, e.g., δ' ὁ, isn't conclusive, e.g.., because this may simply reflect the usual spelling of the Attic word ὁ. Contra, there is ουδεις, μηδεις.

We'll probably never know. Αnd as mwh wrote, small details of pronunciation like this may have differed over time or by social class or even from individual to individual. As well-suited as Greek orthography is to the ancient language, we can't hope to recover all of the many fine details of articulation that must have characterized ancient Greek pronunciation, just as there are many such details of articulation that even the IPA barely captures.

[Menoeceus]

The dissimulation of voiceless aspirates known as Grassmann's law operated at an early stage of Greek. One other factor that may (or may not) be relevant: the general tendency of Greek towards psilosis. Classical Attic was an exception, but there's evidence that even the "original" language of the Homeric poems was psilotic. We can't tell from manuscript texts because the texts were to some extent Atticized by scholars.

[...]

We'll probably never know. Αnd as mwh wrote, small details of pronunciation like this may have differed over time or by social class or even from individual to individual. As well-suited as Greek orthography is to the ancient language, we can't hope to recover all of the many fine details of articulation that must have characterized ancient Greek pronunciation, just as there are many such details of articulation that even the IPA barely captures.

The point about Grassmann's law's dating is absolutely correct and should have occurred to me. My earlier suggestion was mistaken; definitely Attic inscriptions would not show evidence for, e.g., δ' ὁ affecting a previous word. I think ultimately the realization of this elision (i.e., over a word boundary) is probably unknowable unless we have other evidence such as misspellings (or rather "misspellings" since it's at least misleading to talk about an ancient Greek *orthography*), comments from grammarians, or transliterations by non-Greek speakers. Sadly (almost) all of our good evidence, by which we can make solid arguments about the evolution of Greek phonology, is in the papyri from a much later period.

(compare Verner's law)

Just wanted to add that I don't quite get what there is about Verner's law to compare with this. Verner's law has to do with the interaction of voicing and stress in Proto-Germanic.

[Hylander]

On rereading Menoeceus' quote from Threatte, I wonder about the pre-4th c. Attic epigraphic spellings ουδεις/μηδεις shifting to ουθεις/μηθεις. Does the earlier spelling reflect preservation of the voiced articulation with aspiration in Attic speech, with later devoicing? Or does the earlier spelling simply reflect a consciousness of the history of the words as compounds, not a graphic representation of a spoken voiced articulation, the later spelling being a concession to an articulation that had existed earlier in the spoken language?

Does the Modern Greek development of δ in ουδεις/μηδεις as a voiced, not voiceless, dental fricative reflect a continuous spoken articulation of the δ as voiced from the 5th c. and before through the Roman era (despite the ουθεις/μηθεις spelling), or a shift from an unvoiced aspirated articulation (or, perhaps by that time, already an unvoiced fricative articulation) to a voiced unaspirated articulation under the pressure of the reversion of the standard orthography from θ back to δ?

And should we be replacing ουδεις/μηδεις in our texts of Plato and Demosthenes with ουθεις/μηθεις?

[mwh]

On a minor point, surely elision was routinely effected?

[Paul Derouda]

Since there was no letter representing the voiced aspirated dh sound, wouldn’t it even have been possible that ουθεις/μηθεις represented a voiced stop sound? I’m not saying that it necessarily did, just that there are several possibilities and the uncertainties are many, as Hylander earlier implied.

[jeidsath]

I am under the impression that Grassmann's Law chronology is more complicated and somewhat different than Hylander or Menoeceus suggest. Further, the time periods or regions when it first had an effect, when it accomplished most of its language change, and when it stopped having any effect are all different. (And likely came and went.) Here, we would be concerned only with the last: when it stopped having any effect.

One reason to think that Grassmann still has an effect late is that Post-Homeric aspirated perfect forms seem to obey it just fine. In fact, by the Koine period, everything is broadly in agreement with Grassmann, with some exceptions.

Another reason is this stolen image of a cursed 7th century jar (Italy) that will strike anyone who unspoilers it θυφλος. But why doesn't θυφλος show up again later if it's a real form?

Spoiler

Show

ΤΑΤΑΙΕΣ ΕΜΙ ΛΕϘΥΘΟΣ ΗΟΣ Δ ΑΝ ΜΕ ΚΛΕΦΣΕΙ ΘΥΦΛΟΣ ΕΣΤΑΙ

And as far as exceptions go, they aren't so surprising. Analogy could have easily worked the other way for something like "πανταχόθεν."

Although Hylander makes the claim that οδηγ- compounds are too late for Grassmann's, he doesn't give any sort of evidence for it. In fact, they show up in tragedy, starting with Aeschylus.

So we're back to the very weak evidence that a single (theoretical) violation of Grassmann's Law provides. And it's very weak. But it's a data point where we don't have many others.

***

Verner's law comparison. It was a very minor point. It's an example of how voice after unstressed vowels in an Indo-European language beat out fricatives/aspiration, even though in modern English, aspiration apparently overcomes voice in the "ransomed host/ransom toast" confusion.

[Hylander]

No one until this discussion on Textkit has ever suggested that Grassmann's law applied to dissimulation of voiced aspirates because no such consonants exist in Greek. Grassmann's law applied to dissimulation of voiceless consonants, and it's pure speculation to think that something like Grassmann's law might have come into play in marginal situations where voiced occlusives came into contact with an initial aspiration due to elision or compounding -- especially since the orthography is at best ambiguous as to what happened in this situation, which is what started this discussion.

The effects of Grassmann's law continued to be observable in later Greek-- but not universally -- after it ceased to be operative synchronically because Greek words, including verb forms, had already been shaped by Grassmann's law in the earlier period when it was still operative. But analogical leveling eliminated the effects of Grassmann's law in some case in later Greek, e.g., φαθι, εχυθην, and, at the same time preserved its effects in words where a sound change eliminated the second aspirate, e.g. πεισομαι (<*φειθ+σομαι). You might want to read up on Grassmann's law in Sihler or, if you read French, Lejeune, whom I find easier to follow than Sihler. Those are the sources I rely on for information about the pre-history of Greek.

An isolated form on an archaic vase -- a form that apparently violates Grassmann's law -- doesnt prove anything one or another. It could be a feature of an non-Ionic epichoric alphabet or just an error.

[jeidsath]

It's not exactly "pure speculation", at least not more than any of the rest of the speculation in this thread. It's a reasonable inference.

You may wish to look at Filip De Decker's 2016 "Another attempt at a chronology of Grassman's Law" (source of my aspirated perfects example) for a different view than Sihler's, and also take a look at Plath's 2001 "Nochmals zur Datierung des grassmannschen Gesetzes im Griechisches" (source of my θυφλος example).

...because no such consonants exist in Greek...

A "consonant" is not a letter that someone writes down. It's the contact of the upper and lower elements of the vocal apparatus during speech. If the vocal chords are activated and breath is forced out at a moment of near total constriction, it's a voiced aspirate. Either this occurred at times during Greek speech (the suggested "dho" above) or it did not (my argument, again emphasized to be an extremely "weak" data point, from the lack of obedience to Grassmann's Law in οδηγ-).

[Hylander]

I'm not sure we're actually in agreement on much if anything.

There's no evidence that the ancient ancient Greek phonological system included voiced aspirated occlusive phonemes, and I think it's safe to conclude, with just about everyone else, that it did not.* It's possible that when a voiced occlusive came into contact with aspiration as a result of elision or compoundng, the occlusive was -- as a matter of phonetics -- articulated with aspiration. My point is that the evidence on this, as I think you recognize, is at best inconclusive (actually, I think, non-existent in the case of compounding).

I fail to see what Grassmann's law has to do with this. Grassmann's law is formulated in terms of dissimulation of voiceless occlusives -- and this is because the ancient Greek phonological system (as opposed to the phonetic,/i] acoustdical realization of ancient Greek) didn't include voiced aspirated occlusive consonants.

But in fact, the examples you cited -- the archaic vase (involving graphemes apparently representing voiceless aspirated occlusives, which would in fact be subject to Grassmann's law) and οδηγεω and its compounds --as you noted -- don't observe any dissimulation. But again, I don't think οδηγεω and its compounds offer anything one way or another, not just because the Greek alphobet doesn't represent voiced aspirates, but also because those words are transparently compounded. Even if they were formed from οδος and ηγεομαι when dissimulation of aspirates was still operative and applied to voiced occlusives articulated with aspiration, it seems to me entirely possible that they would escape dissimulation because native speakers would have been conscious of their history.

I'm sure there are many arguments about the timing of Grassmann's law, and perhaps dissimulation of aspirates continued to occur, if only on a sporadic, non-systematic basis, at later periods. But there's evidence that it was not operative, at least not systematically, in classical Attic.

*It's true that Linear B didn't distinguish aspirated, unaspirated and voiced consonants at all but we know, based on our knowledge of both PIE and ancient Greek, that they must have been distinct, so that's an example of a Greek writing system that didn't reflect important phonemic distinctions. But in the case of Linear B we do have evidence that those series of consonants were distinct in the Mycenean era, namely, PIE and ancient Greek; otherwise Linear B could never have been deciphered. In the case of ancient Greek, there's no evidence that the phonology included glottalized consonants, like Georgian or Amharic, and I think we can safely exclude the possibility that Greek had such consonants.

[jeidsath]

I'm not sure we're actually in agreement on much if anything.

That seems to respond to something that was not in my post.

I fail to see what Grassmann's law has to do with this...

Do you mean seriously that you don't understand, or are you trying to imply rhetorically that I'm incomprehensible and unreasonable? Obviously -- I think it would be the obvious conclusion about my viewpoint, anyway -- I see Grassmann's law as a tendency to the suppression of coincident aspirations during a speech expression, and not a strictly phonological law. I could present arguments for or against this view, but I don't think it justifies a stance of incredulity in someone coming across it.

...the archaic vase...

I believe that that you may have lost track of the course of the argument here. The vase came up only secondarily, in the discussion of Grassman's chronology, and does not bear on the initial point. If you recall, chronology came up because your initial objection was that Grassman's law operated so far back in pre-history that it was never active in the time when οδηγ- compounds were formed, and that I was therefore talking nonsense (why these rhetorical shots are necessary, I don't understand). You have retreated from that position now, quite rightly, and accepted that Grassmann's could well have acted much later. (That's all I would maintain myself.) My argument was not, if you recall, that my data point proved anything, but only that it was a minor piece of data in a situation where we had few.

[Hylander]

I actually meant to write "I'm not sure we're actually in disagreement." I was trying to be conciliatory, but somehow it came out wrong. My intended point was that there's not much evidence on how a voiced obstruent followed by a word beginning with aspiration might have been articulated, a situation that could only occur as a result of elision (which I think was a point mwh made) or possibly compounding, and I think you agree with that.

I apologize if you took offense, Joel.

[jeidsath]

I am sorry for interpreting it as a swipe and being offended. I should have stopped and realized that wasn't meant.

[ido66667]

I take such cases as simply indicating the gradual disappearance of the /h/ phoneme. Rather than postulating a new allophone or a marginal phoneme, it may have been the case that psilosis started operating rather early, even in classical times. The process could have been gradual. Perhaps at first /h/ was lost only in these cases, where you have elision. Later it may have spread to all mid-sentence words, and finally completely.

Remember that breathing was added later on, by Hellenistic grammarians and scribes, possibly after it was already lost in popular speech. We also know that they were pretty zealous about it, even in psilotic dialects such as Ionic (you can find both ἱκνέομαι and ἀπικνέομαι in the same texts, for instance). δ’ ὁ could very well have been pronounced /do/ and later on the scribes indiscriminately added rough breathing to all occurrences of the article.

Otherwise, why would /h/ not block elision, like the other proper consonants?

[Hylander]

/h/ doesn't block elision in the case of voiceless occlusives, and the breathings or /h/ are apparently preserved in the aspirated articulation of the occlusives. But you're generally thinking along the same lines as I was in a previous comment.

[mwh]

Does aspiration ever block elision, whether in Greek or in Latin? Apparently not, if verse is anything to go by, or only in exceptional circumstances.

[Hylander]

Yes, of course, aspiration never blocks elision in Greek. In pointing out that aspiration doesn't block elision in the case of voiceless occlusives, I was responding to ido6667's suggestion that if aspiration were retained in pronouncing δ(ε)+ /h/ο, the aspiration would block elision "like any other consonant". It seems to me that this suggestion is probably not correct, given that, at least orthographically, /h/ does not block elision in situations where a word ending in an unvoiced unaspirated occlusive + vowel is followed by a word beginning with /h/ + vowel -- and yet aspiration doesn't go away: it's retained as a feature of the unvoiced consonant, as e.g. τε /h/ο > θ'ο.

And of course /h/ doesn't "make position", so in that respect too /h/ doesn't behave like any other consonant in Greek (or Latin).

Sorry, I don't have breathing marks on my tablet.

[jeidsath]

I found a bit of discussion from Iliad scholia of internal word aspiration after vowels ι, υ, and the liquid λ, normally unmarked in our texts. Take it for what it's worth, but it seems to indicate that aspiration occurred in places where it was not necessarily written down.

(15.705b) Ap. H. (?) σχόλιον· ἐπιθετικὸν γὰρ ὂν δασύνεσθαι ὤφειλεν, ἀλλὰ τοῦτο, εἰ σύνθετον ἦν, ἐγένετο ἄν. νῦν δέ, ὡς ἐν τοῖς ἄνω (sc. in scholio a) εἶπεν, παραγωγὴν ἔλαβε τῆς αλος κατὰ παρολκὴν κειμένης. Διονύσιος δὲ ὁ Ἁλικαρνασσεὺς ἐν τετάρτῳ Ἀττικῶν ὀνομάτων οὕτως ἔφη· „ΦΙΛΙΠΠΟΣ μὲν τῆς μέσης δασυνομένης τὸ προσηγορικόν· οὕτως γὰρ καὶ ἐν Τηρεῖ Σοφοκλέους ἀναγινώσκομεν· ‘Ἥλιε φιλίπποις Θρῃξὶ πρέσβιστον σέλας’. Φίλιππος δὲ ψιλῶς ὁ Μακεδών. καὶ ὠκύαλος μὲν δασέως ἡ ναῦς, ψιλῶς δὲ ὁ ἀνήρ· ‘ὦρτο μὲν Ἀκρόνεώς τε καὶ Ὠκύαλος’ (θ 111).“ ἐναντίως δὲ ὁ Ἡρωδιανός. A

Herodian:

705. ὠκυάλου: ἐψίλωσαν τὸ α, παραγωγὴν δεξάμενοι καὶ οὐ σύνθεσιν· καὶ οὕτως ἡ παράδοσις ἐπείσθη. καὶ τάχα ἐπεὶ μηδὲν ἄλλο ἢ τὸ ὠκὺ ἐκ τῆς λέξεως σημαίνεται. ἡ δὲ ἁλός γενικὴ ἐν τῇ συνθέσει ἐπὶ μὲν κυρίων ψιλὸν ἀποφέρεται τὸ πνεῦμα, Εὐρύαλος, Ἀμφίαλος, Ἀστύαλος· ἐπὶ δὲ ἐπιθετικῶν τὸ ἴδιον τηρεῖ πνεῦμα «πολλαὶ ἐν ἀμφιἅλῳ Ἰθάκῃ» (Od. α 395) «ἀγχίἁλόν τ’ Ἀντρῶνα« (Il. Β 697), ὕφαλος, ὥστε καὶ κατὰ τοῦτο ἐλέγχεται τὸ ὠκύαλος ψιλούμενον. A.

Despite the disagreement on ὠκύαλος derivation, both seem to agree on a general rule that internal word aspiration is kept for adjectives, but dropped for names.

[Lucus Eques]

Very lovely to find this discussion a couple years after the fact!

I have often discussed this topic with others, so I appreciated seeing all the points of view presented. I used to be of the mind that the most appropriate choice was just deaspiration for δ’ ὁ [do], and I had never found the ουθεις for ουδεις in Threatte to be particularly convincing since, if Classical Attic lacked a /dʰ/ phonemically, it would make the most sense to render δ’ ὁ unaspirated voiced [do], and certainly not [tʰo].

But my mind has been slowly changed by the best Attic speakers, who, in addition to immaculate pronunciation, advocate aspirating the voiced stops. I used to politely oppose their point of view, but recognizing further evidence in Latin has persuaded me otherwise.

You may know that Latin authors regularly demonstrated internal aspiration in transcriptions: Pyrrhus is one example (though this is actually a voiceless R [r̥] it shows the Roman desire to represent the contemporary Greek sound), and Polyhymnia for Πολύμνια or, indeed, Πολυύμνια is especially demonstrative that, while usual Ancient Greek orthography does not show internal aspiration, it is in fact there even in the 1st century BC Attic the Romans heard and sought to represent in Latin orthography. I used to believe this was mere slavish orthographic imitation, but the regularity of the practice convinces me that this is Greek as the Romans actually heard it (at least in contemporary Athens), and is not a mere etymological spelling.

Something worth considering also, is that there is more than one glottal fricative [h], the voiceless variety, but also the voiced [ɦ]. This is a very likely candidate for the H in words like mihi in Latin: [miɦi], since its voicing permits a smooth transition to mī as a common alternative. The voiced glottal fricative [ɦ] actually occurs quite a lot in spoken English, often in how people say "ahah," especially when not especially excited in the saying.

The verb συνί̄ημι is a good place to use the voiced glottal fricative: rather than [sy.n̥ʰǐː.ɛː.mi] (note that the [n] has been rendered voiceless to [n̥] by necessity since the voiceless [h] is in the superscript), I recommend [sy.nʱǐː.ɛː.mi]. Since it's hard to see, the two variant characters in IPA are [n̥ʰ] vs. [nʱ]. I recommend [sy.nʱǐː.ɛː.mi] (with the ɦ) as nasals tend to voice 'weaker' sounds when they are able, such as σμῑκρός being [z̠miː.krós].

Since this is a reasonable sandhi, it stands to reason that a voiced glottal fricative could be added to a delta with no issues: [dʱ]. Indeed, [dʱ] is IPA for the consonant sound in Sanskrit and Hindi, not [dʰ], which would be impossible since choosing the voiceless glottal fricative would devoice the [d] to [t].

Given how readily Classical Attic retains aspiration where it is orthographically possible, such as καὶ οἱ > χοι, I think the best recommendation is to keep aspiration (voiceless or voiced glottal fricative) everywhere, even where the orthography cannot show it, thus δ’ ὁ = [dʱ]. In this recent episode of my podcast I attempt just that:

https://youtu.be/Wc6enX_Wo5k

To that end, I also now write συνἵ̄ημι etc.

[mwh]

Luke, I’m glad to see you writing the internal breathing on συνἵημι. Literary papyri often do that too. (Of course that doesn’t mean that aspiration was actually effected.) The quantity of the iota varies, however, and I don’t suppose that even in 5th-century Attic speech it was consistently long.

[Lucus Eques]

Luke, I’m glad to see you writing the internal breathing on συνἵημι. Literary papyri often do that too. (Of course that doesn’t mean that aspiration was actually effected.) The quantity of the iota varies, however, and I don’t suppose that even in 5th-century Attic speech it was consistently long.

Pardon my late reply! Would you be able to point me to some of these literary papyri? I have decided to write the internal etymological smooth breathing as a rule, as you can see here: https://docs.google.com/document/d/1z3z ... sp=sharing

I would love to point to the literary papyri you have in mind to justify my unusual (but I think paedagogically useful) spelling choice.

[mwh]

I don’t think you’re right to say it was once the norm in writing ancient texts. But you’ll be happy to see εξανἕστηκεν in P.Oxy.3151 (Sophocles) complete with breathing and accent, and with a breve on the alpha for good measure. In such texts lectional aids tend to be used sparingly, for clarity and disambiguation. My own preferred practice, a little more radical than yours, is to use rough breathings but not smooth ones.

[jeidsath]

The more I read, the stronger my impression becomes that the early copyists were far more interested in distinguishing ambiguous forms (like -ἕστηκεν and -ἔστηκεν) with their annotations than in indicating pronunciation.