Vernacular Polytonic is Absurd: Nolo Contendere
To start with, I agree with the position that applying the polytonic to Modern Greek is capricious and arbitrary and a blockage for learners. Triantafyllidis was the linguist Kaplanis cited (with tildes for circumflexes) in his testimony on the "Trial of Accents", and he reiterated the objections I had summarised, with further examples and cases. The particular instance Triantafyllidis was gleeful to latch on to was, the Academy couldn't get its story straight on when to circumflex, either in its successive recommendations, or within a single recommendation. This made for an unlearnable system.
Nor does learning polytonic have many clear benefits to the literariness of your Standard Modern Greek. People brought up in the monotonic can still read polytonic texts passively, even if they can't produce them. The circumflexes don't give them much etymological or morphophonological benefit at all, the way that historical spelling does in Greek or English (more so in English than Greek, I suspect). The aspirates do help people work out etymologies a bit; but to be frank, these days the initial <h>'s are almost as accessible to Greek students via English as via polytonic.
So cudgelling 21st century Greek to an accentuation that died out two millenia ago is inherently kind of silly. And inasmuch as the language of 1200 is structurally the same as the language of 2009, at least in its accentuation, it's as silly to cudgel 1200 Greek into polytonic. Vernacular Greek from 1200 is just as legible in monotonic, and less bother to write down, especially since people have less recall, and less patience to learn, the shifting polytonic rules for the vernacular. (That's why people go to software now—and don't have enough familiarity with the polytonic in their wetware, to proofread what the software has come up with.)
There's a minor linguistic catch, and a major extralinguistic catch I'll come back to. The minor linguistic catch is, that Greek was seldom purely vernacular as written down, particularly before the Cretan Renaissance. Of course, because of the Church if not diglossia, spoken Greek was probably not always purely vernacular either—though I think the orthodox opinion, that the written vernacular had more archaic Greek than did the spoken vernacular, has not yet been disproven. And the closer we get to Ancient Greek, the more linguistic sense the polytonic makes. This holds not only for the Greek of 400 BC Ancient Athens, but also for the learnèd Greek of 1400 Constantinople, written in an approximation of Ancient Greek: not because those Constantinopolitans would have read it out loud with pitch accent and /h/, but because what they wrote was in a language where those distinctions mattered, even if they themselves didn't realise it. Learnèd Greek was, after all, a written language.
But just as polytonic goes through gradiations, as people tried to apply it more practically to the Modern Language, so too are there degrees of fit of the monotonic to increasingly archaic Greek—or at least, to different levels of mixing of Ancient and Modern Greek. Straight monotonic will be just about fine with the Cretan Renaissance. Or at least, any tweaks to make standard monotonic fit better to 1700 Cretan don't need to move towards the polytonic, for the texts to be more easily readable or accurate to their phonology: it's morphologically odd stress accents like έρχομέστα that you need to accomodate, not written pitch accents. But earlier texts, which mix in more datives and futures, at least in formulaic use, need some tweaking of the accentuation to deal with them. Kriaras was an aggressive promoter of the Demotic, and starting to use monotonic in an Early Modern dictionary in 1977 was a decidedly activist gesture; but he did accent monosyllabic relativisers, and readmitted the iota subscript.
If we go further in the relative mix of Ancient and Vernacular, we move from a text that can basically be monotonic, with the occasional archaism that monotonic can accommodate, to a text that should basically be polytonic, with the occasional vernacularism. That occasional vernacularism could stick out macaronically, by switching between polytonic and monotonic: people haven't really done that to date, but it wouldn't be the end of the world, and actually, it would send out an interesting message. (More on the messages the accentuation sends, later.) Or, they could leave the whole text in polytonic, because after all, there is a convention for polytonicising the Demotic, and it won't be absurd in small doses of the occasional sentence.
So when Anna Comnena or George Cedrenus cite a sentence of the vernacular in an archaic text, we won't monotonicise the remaining five hundred pages to accommodate the one sentence. When John Cantacuzene cites vernacular letters to the Sultan of Cairo, in an otherwise archaic text, we won't monotonicise the remaining text either. We could monotonicise the letters, because after all they do stick out from his text as being in a separate language. Or we could just shrug for the sake of consistency, and put them in polytonic too. We might want to use 1960 polytonic rules and not 1300 Ancient rules, because those are vernacular texts; of course, Cantacuzene didn't use 1960 rules, and then the question is whether we're allowed to normalise what Cantacuzene and his contemporaries wrote. We probably are allowed to, and we do normalise manuscripts a fair bit. Two polytonic norms in the same text is somewhat odd, though, and a bit more glaringly anachronistic than just switching to monotonic, or accenting the vernacular letters like Cantacuzene would have done, in Ancient fashion.
The real problem is when the mixture is more equal. There aren't many such texts, but they do exist: The Mass of the Beardless Man (Spanos) is the clearest example. It's a parody of the mass, so it has an ancient substrate; but there is a lot of vernacular Greek piled on top of it. It spills over, Queneau-style, into parodies of other genres, including saint's lives and wedding contracts; and those other parodies have more or less vernacular in them. How do you accent a text like that? Eideneier used polytonic, because he doesn't do Early Modern monotonic. But with a basically archaic departure point, I think that's the correct thing to do. Not only for strictly linguistic reasons, but also for the "extralinguistic" reasons I go into later.
How broad the vernacular canon is is going to depend on who's asking, and to what end. A literary canon is going to be on the narrow side, and is going to hesitate about John Camaterus. (It will probably hesitate about the Proto-Bulgarian inscriptions as well.) A linguistic canon is going to grab whatever it can, especially the further back one goes. The TLG lemmatiser has to apply quite different rules to the vernacular, and refuses to confuse Ancient Greek with the vernacular to avoid specious ambiguity. So it casts its net as wide as anyone has. Camaterus is in, though I did draw the line at Michael Panaretus, whose vernacularisms did seem to be just lexical.
All things being equal, texts in the same tradition should have the same orthography, especially as they are increasingly machine-searched rather than eyeballed. Not every text does invent its own orthography from scratch, and refusing to emend is indeed cowardice (ατολμία) rather than respect, given that the spelling and accentuation of the manuscripts is so chaotic. The recent edition of the Historia Imperatorum by Iadevaia is particularly egregious in its failure to normalise. There are multiple manuscripts, so she's hardly preserving a single manuscript's reading, diplomatically. The result is oddball to read, and unsearchable on computer, to no real benefit. The audience who care about the original spelling mistakes is much smaller than the audience who want a readable and searchable text. Since we have had no edition of the text at all, we did not need the first edition ever to be an unsearchable mix-and-match transcription. And the audience that does care about the original spelling mistakes isn't served anyway, since the different spellings of the different manuscripts have not been noted.
Even if there was just one manuscript, that's not what we do with texts. If the texts are on papyrus or stone, reflecting the contemporary pronunciation, we tend to leave them alone—although we do normalise the local alphabets to the late Classical norm. But the extended ancient literary texts, which have come to us via mediaeval scribes, are normalised to what we believe correct Ancient Greek was. (We know Ancient Greek better than the mediaeval scribes, right?) And for most contexts and audiences, spelling normalisation is more important than preserving spelling eccentricities. The audiences that do care about spelling eccentricities... well, they can get photographs. (That's a pretty elitist take, I have to say: only university scholars of good standing have had access traditionally, and they used to have to schlep over to Paris or Mt Athos to do it. The interwebs, and the possibility of digital reproduction, may eventually help here.)
I will note about normalising accentuation, that the Byzantines get the circumflex and acute mixed up where a Modern polytonicist might hesitate—in the penult, when the accent depends on the stem, rather than the ultima where it is a matter of grammar, or the antepenult, where it's always meant to be acute. We don't see ἆνθρωπος in their editions, and you don't see τῆς καλής that often; but you'll see κύμα for κῦμα routinely. So the learnèd Byzantines' confusion in the penult does not reject the polytonic, the way the random accentuation of the vernacular could be argued to. It just illustrates why the polytonic wasn't such a good idea by then. Still, editors do not often fix those mediaeval acutes: they do allow the texts to reflect the Byzantine confusion. It led Maas not to bother circumflexing anything at all. And it disrupts the need for consistency, which all other things being equal is desirable.
For my part, I have routinely had to add alternate stems to the TLG lemmatiser, to deal with these "wrong" accents, because the stems were not being accented with ancient quantity. But I do not believe this means these basically archaic texts should be monotonicised. And I don't believe those texts should have been normalised to Ancient quantity either. It's a minor enough deviation (affecting just lexicon and not morphology), that it doesn't disrupt the basics of polytonic, the way the vernacular hesitations do. And it tells us something about Byzantine learnèd Greek.
And as far as the Early Modern vernacular is concerned, yes, editors should Sin No More with their orthography, and settle on a norm, all other things being equal. But there is a 150-year backlog of editions of Early Modern Greek texts, predating any normalisation of Demotic spelling, and until they're all redigitised and respelled by someone, we will keep having to deal with heterogeneous orthography. (The TLG is not going to be doing that respelling: it's a bit too interventionist to fit in their mission statement, and doing the conflating in the lemmatiser rather than the texts is the more hands-off way of doing it.)
And as long as we're stuck with heterogeneous spellings—let along accentuations—the case for normalisation in editions is less pressing. It is pressing for a dictionary, or other words of secondary scholarship; and Kriaras has made its choices. They haven't reprinted the first volumes that had appeared in polytonic though, or the next few volumes that appeared in Kriaras' version of monotonic, before the 1982 standard was settled. Consistency is a good thing, but it's not so urgent that we drop everything else that we're doing.
The extent to which editors have intervened with the language of the Early Modern texts, to make it more digestible to modern eyes, has not been yelled about enough. Granted, the source manuscripts are messy enough that you have to do something to make sense of them. And the scribes do often enough display signs of stupidity—or if you like, neglect; there are textual variants that just don't make sense, and you can confidently say they don't make sense if an alternate reading does. (Assuming of course you're well-informed enough to make a sound judgement.) But that makes you mistrust everything the scribe writes: not just their spelling and their wording, but their morphology and their metre. But just because many a scribe had a tin ear does not mean that Early Modern metre is identical to the modern Political verse: it isn't, and many an editor has systematically edited away anapests, not realising that the mediaeval verse allowed them.
The same goes for morphology, for the phonology of loanwords, or indeed for the integrity of the texts themselves. Kaplanis singles out Alexiou's edition of the War of Crete; there's also plenty to look askance at in his Escorial Digenes—even though it was precisely Alexiou's interventionism that first made the Escorial Digenes legible, and revived scholarly interest in it. And as to the linguistic anachronisms of the edition of the War of Troy... well, suffice it to remind people that, no, this edition does not prove θα was being used in 1400: it does indeed still date from 1700; and if you're using the text for linguistic research, make sure you're checking the original readings in the app crit. (You should of course be doing that in general.)
What Kaplanis says is common sense, and is practice elsewhere: normalise the spelling as you will, leave the phonology alone. Normalising the accentuation of vernacular texts from 1200 is not going to do any real violence to the linguistics of the texts; nor will normalising the spelling, especially because Modern spelling is still basically historical. So you don't have to grapple with the issue of when gemination died out in Standard Greek—and not Cyprus; Greek is still spelled as if gemination survives everywhere.
Still, there's something uncomfortable about the spelling normalisation. I came up with a much too uncharitable phrasing when I was trying to explain the issues to a friend: "We respect 17th century Cretan. That's why we publish it in 21st century orthography." Of course, that's because we have set up a categorial opposition between the three practices surviving in use: Ancient, Monotonic, and Demotic Polytonic. And we deride the lack of unity of Demotic Polytonic. We have no constituency pushing for 17th century norms: people want to see texts through a normalisation someone is using now, and Kaplanis says so explicitly—"we aren't publishing these texts for a hypothetical 17th century scholar". Not that there was much of a norm in the 17th century, to judge from the autograph manuscripts. (But what about the printers? If anyone was going to impose normalisation, it wouldn't be the authors.)
But not every language's philologists do the same thing: we modernise Shakespeare's spelling, but not Chaucher's, Montaigne's but not the Chanson de Roland's. In fact, in a roundabout way, using the monotonic for Ptochoprodromos and Cornaro is doing the same thing as is using the polytonic for them, in the opposite direction. They are both assertions of cultural continuity. And they should be seen as ideological statements, not just linguistic statements.
Arbitrariness of norms
Kaplanis argues that the two 17th century autographs in Greek script are as chaotic as the scribes before them, so they weren't taking the polytonic seriously; and he also wonders whether, given the option of monotonic rather than the absurdities of historical accentuation, they wouldn't have jumped at the chance. The third manuscript is in Latin script, as appears to have been the norm in Crete. It may be that the Veneto-Cretans had no idea about Greek script, although given that the Orthodox church survived Venetian rule, I doubt it. It may be that it was an ideological choice, aligning with Venice and Western learning rather than antiquity; that is more than likely, given how Ancient Greek deities are Italianised in the texts. But it also relieved them of all the absurd choices of Greek script: not just the accentuation, but the iotacisms and the omegas and the impossibility of distinguishing /d nd nt/ in their multitudinous Italian loanwords, and all of it.
Latin script is not a legitimate choice for anyone publishing these texts now, just like IPA isn't a legitimate choice. But it's not linguistics determining that choice. You can write Greek in Italian orthography, and that's exactly what Foscolo did. Whether we put in circumflexes or omegas when we reprint his work, we're imposing our own normalisations on his text. Not that we shouldn't be normalising it: there are good reasons to do it, primary being that we shouldn't have to transform our readership into Veneto-Cretans to read these texts today. But an omega/omicron distinction is no more intrinsic to Foscolo than a circumflex/acute distinction. Doing the former and not the latter is not motivated by linguistics, or being responsible to his text. It's being responsible to our audience.
Which is what I meant above. If you spell Foscolo with the fully Ancient paraphenalia, you assert that he's Their text—the ancestors', not Ours. If you spell Foscolo in monotonic, you assert that he's Our text, and that our language isn't the same as Theirs. Not *too* dissimilar, though—it's just the accents and a few endings that have been brought into line. But autonomous. If you really wanted to break with the past completely, you would follow the 1920 Soviets and have a phonetic Greek alphabet. Or you could print Foscolo as you found it, in Italian orthography.
But with monotonic, you're compromising between what we say and what the Ancients wrote. It's less of a compromise than the Demotic polytonic, which was unstable and unlearnable and deluding itself that the pitch accents of 200 BC still mattered. But it's still pretending that the 200 BC distinction between omicron and omega matters. Demotic polytonic ultimately dealt with the absurdities of having to choose by defaulting to the unmarked acute when in doubt, which makes it look compromised; but the historical orthography persisting under monotonic also now tends to less marked choices when the etymology is murky or a new transliteration: αβγό now not αυγό, Μπρίτνι Σπιρς not Μπρίτνυ Σπηρς; and Ρήγκαν became Ρίγκαν within my lifetime. The difference between implementing historical orthography with polytonic and implementing it with monotonic is a matter of degree. It's less absurd, sure, but it still forces you to rote-learn: IPA, it is not, and the argument is not about linguistic correctness, because noone is advocating IPA here.
Difference of degree is fine when you're choosing an orthography, precisely because this is not purely a linguistic issue. The compromises the orthography makes are counter to what Psichari, in his neogrammarian rigour, grumbled about the way Demotic was going: "linguistics admits no compromises". Inasmuch as linguistics is a unitary algebra of phonemes and morphemes, no, it doesn't. But Psichari's neogrammatically correct Demotic did not prevail: Standard Modern Greek has compromises with the learned language aplenty, enough to make its phonology absurd. The spelling of Greek is just as compromised, and the Purists needled Psichari because he wouldn't take the logical next step there. And both have happened because language does not only go the way it lists as an abstract system: languages are spoken by people and societies, and it serves their ideologies. That too is what language is about, even though it's something linguists usually abstract away from methodologically.
Language serves ideologies on paper no less than it does in soundwaves—probably more so. There's nothing linguistically wrong with the Gothic or Glagolitic scripts, but noone particularly wants to learn how to read them, and the Germans and Russians feel entitled to claim them as their heritage, by making them readable to their fellows, in Latin and Cyrillic. They tinker with the transliteration enough to be linguistically responsible: hvair had to be added for Gothic, and djerv for Cyrillic. But the concern was to make Wulfila legible to Bismarck, not to Alaric. It's the readership of one's fellows that matters.
That argues for Early Modern monotonic on the basis of familiarity and readership, and that will become a more and more compelling case as less Modern Greeks can use the polytonic. But as long as polytonic can be read passively, it is no deal-breaker—just as monotonic Homer in snippets in the press is no deal-maker.
It's still true that the polytonic is more of a rallying point than the spelling—that editors, especially non-Greek editors, are happy to use contemporary spelling but not the monotonic. After all, it is a more obvious visual cue than subjunctive -ῃ, and a more obvious visual break from Learnèd Greek. And it's true that there's nothing scholarly about polytonicising the texts. I don't believe it's true that there's something massively more scholarly about monotonicising them, though: there was a conventionalised polytonic orthography by 1960 just as there is a conventionalised monotonic orthography today, and you could choose either as the basis for consistency (which realistically won't happen anyway). You're making more work for yourself as an editor by choosing thirty-year-old squiggles and increasingly inaccessible norms; but you're not absolved of all normalisation and decision-making by going monotonic.
The real choice comes down to, whose texts you want to assert they are. That's the generations-old Greek debate about their past, a discomfort that did not finish when Puristic Greek was abolished in '76. In the case of non-Greek editors, though, I think things are far simpler than Kaplanis makes out in his psychoanalysis. Monotonic asserts "they're Our texts". A German has less motive and investment to say so than a Greek—even a German as progressive and monotonicist in his approach to Modern Greek as Hans Eideneier. He has less reason to dispute what has been the default thinking, that anything before 1800 is not "Ours", but "Theirs"—even if its morphology and lexicon are close to contemporary. Kaplanis suggests the prestige of polytonic might be able to get texts printed more easily in Hesperia (as some Greeks chuckle to call the West). But that is certainly going to be less and less of an argument, as it is trumped by the annoyances of polytonic online. (They are surmountable with minimal effort, and Kaplanis could really have done better than the tildes in his online text; but it is more work to type polytonic than monotonic, and I don't bother to myself unless there's a real point to it. So Lord Krum goes to monotonic.) At any rate, I don't think classicists look at Chortatzes, with its mixture of Italian and Cretan affricates, and think any better of it just because it's polytonic.
Do I have a conclusion? Not really. Polytonic is not linguistically correct for most Early Modern texts (though not all). Monotonic is only more correct, it is not a phonemic transcription from God. The unthinking choice of polytonic is not scholarly, but habit and ideology. But unless you leave Foscolo in his Italianate Greeklish, it's all ideology. It would be nice if what was the dominant faction in 2000 acknowledged that it was habit and ideology, not scholarliness and respect, that dictated their choices. Not that there's anything wrong with habit and ideology. At all. Don't fret: the monotonic will slowly prevail in this domain too—hopefully applied intelligently to the pre-modern language. But honestly, given the morass of legacy 1870 spellings in the backlog of editions in the corpus, a couple more polytonic editions aren't going to make things worse.
And to be honest, for my work with the TLG, it'll be Kechagioglou's monotonic, not Eideneier's polytonic, that will stick out as the inconsistency, and make me put in extra hours. Not that this means Kechagioglou shouldn't have monotonicised. It just means it's all messy, and I was going to have to put in the extra hours anyway.
Like commenter Sapere Aude said over at the Magnificent Nikos Sarantakos' blog: «Άβυσσος το spelling αυτού του weird lingo…»