On nominalisations ending in -εία

A post on Greek spelling. You've been warned.

The spelling of the noun ending -εία vs. -ία had come up a few months ago on the Magnificent Nikos Sarantakos' blog, as an orthographic bedevilment. Modern Greek writers feel ἀμηχανία (awkwardness) about how to spell the ending, and they'll be reassured to know the Byzantines felt the same ἀμηχανεία.

The story goes like this:

• Ancient Greek has an ending -ία, used to form abstract nouns from verbs and adjectives. It corresponds to -ness, and when it is borrowed into English, it shows up as -y. So:
  ἄμνηστος /ámnɛːst-os/ "unremembering, forgetful", ἀμνηστία /amnɛːst-ía/ "forgetfulness, amnesty" (because I'm forgetting your crime).
  ἁρμόζω /harmó-zdɔː/ "I fit", ἁρμονία /harmon-ía/ "joint, suture, harmony" (because the parts fit together).
  
• When -ία is attached to a verb ending in -ευ- /ew/ in Proto-Greek, the result is spelled -ε-ία /ewía/ > /e.ía/ > /éːa/. That is to say, it's the same /-ía/ suffix, attached to /ew/. But because /w/ did not stick around in Greek, /ewia/ ended up pronounced /éːa/, and was distinct from the normal /ía/ ending.
  
• So: δυνάστης /dynást-ɛːs/ "master", δυναστεύω /dynasté-wɔː/ "be lord over", δυναστεία */dynast-ewía/ > /dynást-éːa/ "lordship, dynasty".
  εἴρων /éːrɔːn/ "dissembler", εἰρωνεύομαι /eːrɔːné-womai/ "play dumb, use understatement, make fun of", εἰρωνεία */eːrɔːn-ewía/ > /eːrɔːn-éːa/ "dissembling, understatement, mockery, irony".
  
• So there was a rule on how to form these nouns. If a verb ending in -ευ- was involved, it ended in -εία. Otherwise, it ended in -ία. So ὀνοματοποιέω /onomatopoi-éɔː/ "make up a name", ὀνοματοποιία /onomatopoi-ía/ "onomatopoeia". It isn't ὀνοματοποιεύω, so it's not ὀνοματοποιεία. Which is just as well: onomatopoeia has enough vowels in it already.
  
• Problem number #1 with spelling these nouns was, by the time of Christ, ει and ι were pronounced identically. So the subtle etymological differentiation was begging to be undone.
  
• Problem number #2 was, -ία was not only applied to verbs and adjectives. It could also be applied by analogy from other abstract nouns, even though there was no underlying verb to derive it from. So μαντεύω "to tell the future" gives μαντεία "telling the future". People starting making up nouns of different ways of telling the future: ἡλιομαντia "by the sun", ἡμερομαντia "by the date", κυνομαντia "by dogs", λυχνομαντia "by lamps".
  
• How do you spell these? Do you spell them like the simple noun μαντεία? Or do you say that there is no such verb as ἡλιομαντεύω or λυχνομαντεύω, so you should use the simple -ία spelling? LSJ chooses the analogy: -μαντεία.
  
• Similarly, λάτρις "hired servant" > λατρεύω "adore" > λατρεία "adoration". When people made up a word for "adoration of idols, idolatry", /eːdɔːlolatría/, they did not go through a verb εἰδωλολατρεύω. So how were they supposed to spell it? Like λατρεία "adoration"? Or should they instead derive it from εἰδωλολάτρης "adorer of idols", which would make it -λατρία? This time, conventional spelling did not go with the analogy, and decided to spell it εἰδωλολατρία, deriving it straight from εἰδωλολάτρης. Which is more plausible etymologically.
  
• But that brings you to the unfortunate situation that, whenever you spell a word in -ía, you need to know the derivational history of Greek—whether a corresponding verb in -ευ- has ever turned up or not. This is of course ludicrous, and people were thoroughly confused. LSJ chose analogy for λυχνομαντεία, but notes that the papyrus the word turns up in spells it λυχνομαντία. LSJ has εἰδωλολατρία, but a verb εἰδωλολατρεύω does show up, in Eusebius, Athanasius, and John Chrysostom; and the TLG has 342 instances of the εἰδωλολατρεία spelling versus 1038 of the εἰδωλολατρία spelling.
  
• To complete the confusion, Problem #3. Byzantines couldn't see why verbs ending in -εύω produced nouns ending in -εία, but verbs ending in -έω didn't also keep the epsilon. I mean, they both had epsilons in them; and Byzantines didn't know or care about the prehistory of */w/ in Greek. So they started spelling with -εία words the ancients had never spelled with -εία. Remember "amnesty"? There isn't just an adjective ἄμνηστος "forgetful"; there's also a derived verb ἀμνηστέω "to be forgetful". And if ἀμνηστέω exists, that's reason enough to start spelling "amnesty" with an epsilon, as ἀμνηστεία.

The thread over at Sarantakos' included the host's melancholy observation that this conundrum itself was reason enough to long for a phonetic spelling reform. It probably won't come to that, but it *is* enough for Greeks to turn to their linguists, and plead with them "give us a rule we can follow!"

(The thread also triggered Sapere Aude's immortal snark "άβυσσος το spelling αυτού του weird lingo".)

The guesswork of "whether a corresponding verb in -ευ- has ever turned up or not" is not such a rule. (Has there ever been a verb ἀγγελολατρεύω "to venerate angels"? No peeking!) Universally spelling such nominalisations with just -ία is also a non-starter: λατρία does look like wholesale spelling reform. The sensible compromise is, the simpler alternative when in doubt. (Which applies to a lot of contemporary Greek spelling.) That means compound nouns like εἰδωλολατρία always get spelled with an iota, no matter what verbs Eusebius came up with.

Not that I'm going to bother initiating correspondence with whoever's running Greek spelling these days. (It sure ain't the Academy of Athens. Education Ministry, I guess.)