(How does Nikos Sarantakos put it? Οι όμορφες διακρίσεις όμορφα καίγονται. Nice distinctions burn down nicely.)
The phonology of the prefixes, on the other hand, was the same as that of prepositions used as distinct words. This phonology was not particularly subtle, but it did have rules, which made sense in ancient phonology—although some prepositions dodged them. These rules have not survived into their much-attenuated modern vernacular counterparts.
The first rule is Elision: if a preposition ends in a vowel, that vowel is lost before another vowel. So μετα-μορφῶ /meta-morpʰɔ̂ː/ "I transform", but μετ-εμόρφωσα /met-emórpʰosa/ "I transformed"; ἀντί-θεσις /antí-tʰesis/ "opposition" but ἀντ-αλλαγή /ant-allaɡɛ́ː/ "exhcange", ἐπὶ τόπου /epì tópu/ "on the spot" but ἐπ’ ἀγροῦ /ep aɡrû/ "in the country". A couple of prepositions never followed that rule: περί, πρό, and in later Greek ἀμφί.
The elision rule persisted for the rest in learnèd Greek, and indeed in 19th and early 20th century coinages: μετεκλογικός "post-election", μεταπελευθερωτικός "post-liberation" (World War II). But it was abandoned in the vernacular's prefixes (whose meaning has changed a lot):
|μετ-εῖπον /met-eîpon/ "I spoke amongst"||ματα-είπα /mata-ˈipa/ "I said again"|
|παρ-έτρωγον /par-étrɔːɡon/ "I nibbled"||παρα-έτρωγα /para-ˈetroɣa/ "I was over-eating"|
|ἐξ-ίδρωσα /eks-ídrɔːsa/ "I perspired"||ξε-ΐδωρσα /kse-ˈiðrosa/ "I stopped sweating"|
|ἐξαν-έρχομαι /eksan-érkʰomai/ "I come forth from"||ξανα-έρχομαι /ksana-ˈerxome/ "I come again"|
If anything, its the next vowel which can get deleted in Modern Greek: ξαναέκανα or ξανάκανα /ksana-ˈekana ~ ksaˈna-kana/ "I did again". And contemporary coinages, however well-educated, also no longer bother with elision.
- "Anti-national", which was coined in 1825, has elision: αντ-εθνικός. Usage has respected this: 9400 instances in Google of αντεθνικός vs. 93 of unelided αντιεθνικός. (Not least because the word got a lot of use out of the Colonels' regime.)
- "Anti-nationalist" on the other hand is very much a 20th century notion, and accordingly it lacks elision: αντι-εθνικιστικός (Google: 8:760).
- The World War II coinage "post-liberation" has elision, μετ-απελευθερωτικός (Google count: 9:1)
- But the Google count for "post-colonial" is 5:40 against elision, μετα-αποικιακός.
The second rule is aspiration: if a prefix ends in an unaspirated stop (once its vowel is lost), and it goes in front of a rough breathing, it becomes aspirated. In Ancient phonology, while /h/ existed, that is just common sense:
- μετά + ἐμόρφωσε /metá + emórpʰɔːse/ > μετ-εμόρφωσε /met-emórpʰɔːse/
- μετά + ἵστημι /metá + hístɛːmi/ > /met-hístɛːmi/ > μεθ-ίστημι /metʰ-ístɛːmi/
- ἀπό + ἄνθρωπος /apó + ántʰrɔːpos/ > ἀπ-άνθρωπος /ap-ántʰrɔːpos/
- ἀπό + ἥλιος /apó + hɛ́ːlios/ > /ap-hɛ́ːlion/ > ἀφήλιον /apʰ-ɛ́ːlion/
- ἐπὶ αὐτοῦ /epì autû/ > ἐπ’ αὐτοῦ /ep autû/
- ἐπὶ ἡμῖν /epì hɛːmîn/ > /ep hɛːmîn/ > ἐφ’ ἡμῖν /epʰ hɛːmîn/
The aspiration is concealed somewhat in the alphabet, because /pʰ tʰ kʰ/ are written as single letters, <φ θ χ>. But writing /met-hístɛːmi/ as μεθ-ίστημι is exactly what you'd expect to write using the Ancient Greek alphabet—so long as <θ> is pronounced something like /t-h/.
By the time of Christ, initial /h/ has disappeared, and /pʰ tʰ kʰ/ were on the way to their modern pronunciations as /f θ x/. So aspiration was no longer phonological common sense; it was becoming orthographic voodoo.
If a compound had survived from antiquity into your spoken language, you kept pronouncing it as an aspirated compound, even if you could no longer take it apart. That's how ἀφίημι /apʰíɛːmi/ < /apó + híɛːmi/ "I let" has ended up, after much analogical reformulation, as Modern αφήνω: noone would even begin to think of prying /af-/ apart from /-ino/, because there is no such verb as /-ino/. If you were making up new compounds in the vernacular, the fact that there used to be an /h/ there is meaningless. In any case, with elision gone for prepositional prefixes, you wouldn't get the chance to apply aspiration.
If you were writing in learnèd Greek, on the other hand, you still tried to apply aspiration to your compounds. But applying aspiration has now became a matter of rote memorisation: you had to remember which words used to start with an /h/ (and were still written with a rough breathing), and you would switch the final consonant of the prefix because the grammar books said so, not because changing a /p/ into an /f/ made any phonological sense to you. Because the spelling of Greek also retained rough breathings, correct spelling in polytonic Greek involved memorising tables of words starting with rough breathings—as everyone educated in Greece before 1981 still remembers.
Your correspondent is among the youngest people alive to whom ᾅδης ἅγιος ἁγνός "Hades, holy, pure" rings a bell. Or at least, he should be; but moral panic has brought the teaching of Ancient Greek forward; so "Hades, holy, pure" have now merely shifted from primary to secondary education.
So aspiration was a rule of Ancient Greek, that made no sense in the native language of the people still writing in Ancient Greek. That provides an opportunity for the rule to break down: for those writers to forget to aspirate their prefixes.
There are occasional such instances in Byzantium; and they are brought up in polemics against the polytonic. (Nikos Sarantakos in the linked piece cites ἀντυπουργήσειν in Constantine Porphyrogenitus and ἀντυπενεχθείσας in Gregory of Nyssa.)
One of the only linguistic arguments made for polytonic accentuation is that it enables you to do correct aspiration in compounds. The counterargument is that noone is using aspiration in compounding any more: peals of laughter would greet *ἀνθηλιακός "sun protection" or *ἀνθισταμινικός "anti-histamine". And if you want to passively make sense of a compound like ἀφήλιον "aphelion", there's little point memorising that ήλιος "sun" takes a rough breathing. It's more useful to deduce the rough breathing of ήλιος from ἀφ-ήλιον—or else, to do your rote memorising of /h/'s where it will have more of a pay off. Like learning English, and picking up Helium and heliocentric from there.
If the Byzantines slipped up in failing to aspirate compounds, we also expect them to do the opposite: to aspirate compounds where there is no justification in Ancient Greek. That is hypercorrection. Hypercorrection happens when speakers are trying to apply a rule in a language variant that is alien to them (such as the formal variant of their language). Speakers can't readily apply the rule, because the conditions of the rule don't make sense in their native version of the language. So they lurch for the more formal-sounding option from the rule, in the hope that it will pay off. English-speakers end up saying "between you and I", because they can't understand why they have to say "it is I" instead of "it's me".
Unsurprisingly, hypercorrect aspiration happens in Byzantine Greek. After all, it happens even now in Greece; Nikos Sarantakos mentions a teacher from Drama Prefecture so confident in his command of Ancient Greek (which he inflicted on primary school students), that he spoke against καθ’ επάγγελμα εκκλησιομάχοι, "professional anticlericalists". That's επάγγελμα, as in ἐπ-άγγελμα /epáɡɡelma/, and the prefix /ep(i)-/ has never begun with an /h/.
Hypercorrection in aspiration does not happen in Byzantine Greek as frequently as I'd expected; I've found maybe a hundred instances in the TLG. But it does happen often enough to show that not everyone was confident about how well they'd memorised their "Hades, holy, pure" tables.
- So when John of Gaza writes καθ’ ἧμαρ "by day" (Anacreontea 6.93), the καθ’ leaves no doubt that he has got the breathing of ἦμαρ wrong. He's trying to sound elevated by using a Homeric word; but he hasn't got the Homeric aspiration to go with it.
- When Ephraem the Syrian (or rather, his early Greek translators) write μεθέπειτα instead of μετέπειτα "afterwards" (De paenitentia, Frantzoles p. 81), they are making the same mistake as the teacher from Drama: /ep(i)-/ in /met-ep-eita/ has never begun with an /h/.
- When Paul of Aegina writes ἀφ-ουροῦντας "urinating away" (Epitomae medicae, 126.96.36.199), he's forgetting that the word is "urinate", not *"hurinate" (or, given how Ancient Greek relates to Indo-European, *"surinate").
So much for aspiration. Having gone through the rule violations in the TLG corpus, I wondered whether the rule for /ek-/ was also violated. That rule is: Movable /s/: ἐκ /ek/ becomes ἐξ /eks/ in front of a vowel. So ἐκ-βάλλω /ek-bállɔː/ "I throw out" but ἐξέβαλον /eks-ébalon/ "I threw out", ἐκ τοῦτου /ek tûtu/ "from this" but ἐξ αὐτοῦ /eks autû/ "from that".
But what is the likelihood that Byzantine writers would get the rule wrong? For a rule of Ancient Greek to be violated, the conditions for the rule have to be inapplicable to later Greek; that was indeed the case with aspiration, once aspiration had disappeared in the spoken language. Vowels, on the other hand, still existed in later Greek; so the rule for /ek/ going to /eks/ was still learnable by Byzantines, with reference to their spoken language.
The rule itself also needs to have been no longer applied in the vernacular, so that speakers trying to apply it to Ancient Greek could go astray: the vernacular would need to have used /ek/ before vowels as well as consonants (if indeed /ek/ survived at all), for writers to get the /ek ~ eks/ rule wrong in Ancient Greek.
As it turns out, ἐκ as a preposition did survive, in at least some dialects of Greek, as αχ or οχ, before both vowels and consonants. That does suggest that the alternation between /ek/ and /eks/ broke down in some locations, but not that it had globally broken down: most dialects of Greek don't have a reflex of /ek/ at all. The prefix /ek-/ has also survived in the vernacular, again without the alternation of /ek/ and /eks/. But in the vernacular, it's /eks-/, not /ek-/, which is in universal use.
Now that's an odd development: speakers should have a devil of a time trying to pronounce compounds like /eks-vrakono/ "to strip naked, to un-pants" or /eks-xtenizo/ "to tussle hair, to uncomb"; so how did /eks/ end up taking over?
Here's how: with verbs, Modern Greek has always preferred the aorist stem to the present, as the more regular formation. Unlike the present stem, the aorist stem ends in only a few, predictable consonants—mostly /s/. So the present was usually remodelled to line up with the aorist, with a few simple rules. (That's where the Modern language got the idea that as many present stems as possible should end in /n/.)
The beginning of the aorist is regular too: the indicative aorist always began with an augment, which means it always began with a vowel. So while the present tense alternated between /eks-/ and /ek-/, the aorist indicative reliably used /eks-/: ἐκβάλλω :: ἐξέβαλον, ἐκβάλλετε :: ἐξεβάλατε, ἐξαγανακτῶ :: ἐξηγανάκτησα, /ek-bállɔː/ :: /eks-ébalon/, /ek-bállete/ :: /eks-ebálate/, /eks-aganaktɔ̂ː/ :: /eks-ɛːɡanáktɛːsa/.
In Modern Greek, unstressed initial /e/ was dropped; that made unstressed augments optional. So ἐκτένιζα /ekténiza/ "I combed" became χτένιζα /ˈxteniza/, and /eks-ekténiza/ could be metanalysed as /eks-eˈxteniza/ > /ekse-ˈxteniza/. The initial unstressed /e/ of the prefix /eks-/ was also dropped; so the prefix became /kse-/. With its /-e-/ no longer considered an augment, the /kse-/ prefix could be applied to the present as well as the aorist: ξεχτένιζα /kse-ˈxteniza/ "I uncombed", ξεχτενίζω /kse-xteˈnizo/ "I uncomb".
The prefix applies to nominalisations as well as verbs; so learnèd εκκίνηση "starting point" has the vernacular counterpart ξεκίνημα "beginning". The ancient meaning of "movement out of" is retained with older compounds, such as ξεκινώ < ἐκκινέω "to move out from = to set off, to begin". In productive use, though, it usually means undoing an action—like we saw with "uncomb" and "unpants". The prefix even applies to nouns, though in a more ad hoc way; there is an old rebetiko song called Ο Ξεμάγκας, "the un-mangas, someone who has rejected being a mangas". (In particular, giving up on hashish and bouzouki music.)
(You'd think I was looking for excuses to include YouTube videos in linguistics posts, or something...)
So, because /eks-/ survived in the vernacular as /kse-/, we don't have strong evidence that /eks-/ was strange to the vernacular as a prefix, and that speakers would be going vernacular by using /ek/ before a vowel. On the other hand, while /eks-/ survived precisely because it preceded a vowel, it's not like Ancient Greek offered any counterexamples with /ek/ preceding a vowel. So there would be little reason for writers to think that /ek/ before a vowel was a more proper way of writing—i.e. a plausible hypercorrection.
The /ek ~ eks/ alternation may have been on the way out; but any instances of /ek/ before a vowel in Byzantine writing, I conclude, would be infrequent; and they would more likely be thinkos, than harbringers of vernacular influence.
So, I did a search among unrecognised words in the TLG corpus, substituting /eks/ for /ek/. It's a first cut; if Trapp's Lexicon has already registered such compounds with /ek/ before a vowel, such a search would not find them. Nonetheless, while I just predicted that there wouldn't be many instances of /ek/ before a vowel, I was surprised to have found exactly one such instance.
That one instance is from the pen of Leo Choerosphactes. We have already bumped into Leo in this blog, as the target of the rage of Constantine of Rhodes, which resulted in the longest adjectives of Byzantine Greek.
Choerosphactes' instance is from a letter to Tsar Simeon I of Bulgaria, with whom Leo was conducting diplomatic negotiations—sometimes from a Bulgarian prison cell. It is a short letter, but it could have been a lot shorter. Apparently, Simeon had sent Leo a letter, in which Simeon wrote an untruth which Leo found more telling than the prosaic truth. In other words, Simeon wrote a joke; and Leo reacts with all the awkward wordiness of someone puzzled at the invention of humour. Or lying.
G. Kolias. 1939. Léon Choerosphactès, magistre, proconsul et patrice; biographie—correspondance (texte et traduction). Texte und Forschungen zur Byzantinisch-Neugriechischen Philologie 31: 77–129. Letter 12.
Τοῦ αὐτοῦ, Συμεὼν ἄρχοντι Βουλγαρίας.
Θαῤῥεῖς πρὸς ἀλήθειαν, ἀρχόντων ὁ ἀληθέστατος, θαῤῥεῖς πρὸς τὰ λίαν ἐπαινετά· θέλεις δὲ καὶ τὸ σὸν οὐχὶ ἐν ἴσῳ τοῦ τῶν ἄλλων πιστεύεσθαι, ναί, καὶ τὴν δοκοῦσαν κατὰ παιδιὰν ἄρνησιν, ὁμοίαν εἶναί τε καὶ νομίζεσθαι τῆς ἑτέρων ἀληθοῦς κατανεύσεως. Δείκνυται οὖν ἡ ἐπιστολὴ ἀληθεύουσα πράγματι, κἂν δοκῇ τῷ γράμματι ψεύδεσθαι, ἵν’ ᾖ ὑπὲρ τὴν τῶν ἄλλων ὑποπτευομένην ἀλήθειαν ἡ σὴ σοφῶς ὑποπαίζουσα καὶ κωμικευομένη ψευδομυθία. Καὶ τοῦτο θαυμαστόν, καὶ τοῦτο γέμον φιλανθρωπίας, ἵνα, εἰ δοκῶν ψεύδεσθαι ἀληθεύεις, ἐκαληθεύοντός σου τίς ἔσται ὁ πιστεῦσαι δυνάμενος, ὡς ψεύσῃ πώποτε; Ὢ ψεύδους ἐγγραμμάτου, ἀληθείας ἐμπράκτου γέμοντος! Οὕτω δοκῶν ψεύδεσθαι ἀληθεύεις, καὶ ἀληθεύων οὐκ εἰς ψεῦδος αὖθις ἀποκλίνεις. Ἔῤῥωσο.
By the same, to Simeon Lord of Bulgaria.
You have the courage for truthfulness, most truthful of lords; you have the courage for matters most praiseworthy. And you want your writings not to be believed equally to others'; yea, nor do you want your denial, seemingly in jest, to be just like others' truthful assertions, or even to be thought as such. So your letter is shown to be truthful in reality, even though it seems to be lying literally. Thus your sagely playful and antic telling of lies surpasses others' supposed truthfulness. And this is a marvel, and a thing full of humanity: if you tell the truth while seeming to lie, then who can believe you would ever be lying, if you should stray from the truth literally tell the truth? Oh what a literate lie, full of truth in practice! Thus do you tell the truth when seeming to lie, and you do not deviate back into lies when telling the truth. Farewell.
The verb ἐκαληθεύω should be ἐξαληθεύω; but no such verb has been attested in Greek. If the rule about /eks/ was ever going to be violated, it would be in a verb such as ἐξαληθεύω, which was newly coined by the author. Any writer of Greek would have imbibed hundreds of verbs prefixed with /eks-/, from both the Classical language and his own vernacular; he'd be unlikely to get the prefix wrong on a verb he was already familiar with. Coining a new verb, he would momentarily been thrown into unfamiliar territory. And unfamiliar territory is where thinkos are likelier to happen.
The verb ἐξαληθεύω is not attested; but ἐξαληθίζομαι is. LSJ cites it from the Etymologicum Magnum, compiled a couple of centuries after Choerosphactes; but the Etymologicum is citing Photius, a generation older than Choerosphactes:
Διαπορεύεται δὲ τά τε ἄλλως περὶ θεῶν τοῖς Ἕλλησι μυθολογούμενα καὶ εἴ πού τι καὶ πρὸς ἱστορίαν ἐξαληθίζεται.
And he goes through all the other myths about the Gods told by the Greeks, and whatever is "out-truth-ised" in history. (Bibliotheca, Codex 239 p. 319a)
The difference between the -εύω and -ίζω suffixes is slight: having the condition or activity of X, vs. doing the action of X. ἐξαληθίζομαι, by that token, should mean something like "to act 'out from' about truth". In a compound, /ek/ has the following meanings:
Out, from, off, away (cp. ἐξελαύνειν drive out and away); often with an implication of fulfilment, completion, thoroughness, resolution (ἐκπέρθειν sack utterly, ἐκδιδάσκειν teach thoroughly)
So "to act about truth thoroughly", which makes more sense than "to act 'out from' about truth". What sort of actions can you carry out with truth? Telling it; so ἐξαληθίζομαι should mean "to tell the truth thoroughly". And LSJ's gloss of ἐξαληθίζομαι is indeed, "to be truly recorded".
ἐξαληθεύω in turn should mean something like "to have a condition 'out from' about truth", or "to do an activity 'out from' about truth". What condition or activity is associated with truth? Again, telling it. In fact, ἀληθίζω is glossed in LSJ as ἀληθεύω (scroll to the end, Perseus' LSJ has conflated ἀληθίζω with ἀληθής), and ἀληθεύω is glossed as "to speak truth". ἀληθίζω and ἀληθεύω are synonyms. ἐξαληθίζομαι and ἐξαληθεύω should also have been synonyms: Choerosphactes' verb should have meant the same as Photius'.
[EDIT: The following paragraph is incorrect—see comment #1]
From the context, clearly it doesn't. Choerosphactes has gone back to the literal meaning of /ek/, "out from, away": his ἐξαληθεύω is "to do an activity of being away from the truth", i.e. "straying from the truth". It's not the sense that occurred to Photius. And in fact, that too shows that Leo was in unfamiliar territory, when he coined the verb: he didn't use the most obvious sense of /ek/ in context, but the most literal. It was literal enough to make him forget to modify the prefix /ek/, and to leave it, ill-fitting, as ἐκαληθεύω.