Ghost words revived in Allatius

The canon (patchwork though it is) of Greek lexica that I described a while back has a fair representation of German scholarship: Lust Eynikel & Hauspie, Bauer Danker Ardnt Gingrich, Trapp.

The oddity is that German scholarship wasn't represented there for the Classical period. Yes, LSJ is a major work, and DGE is more comprehensive still (if only I would live to see it completed). But it's odd that the Germans didn't corner the market in Ancient Greek lexicography in the 19th century, when they ruled Classical philology.

As it turns out, Wilhelm Pape's Handwörterbuch der griechischen Sprache (revised by Max Sengebusch in 1880) is competitive with LSJ, with 99,000 headwords against LSJ's 116,000. Given that LSJ, published in 1940, covers an additional 60 years worth of papyri and inscriptions, Pape's dictionary has nothing to be ashamed of.

I've recently had occasional to compare it against the other lexica of Greek: I keep trying to fill in gaps in lexical coverage hither and thither—for reasons that should be obvious to the regular readership I've been neglecting. Of those 99k, 545 lemmata are absent from the other lexica, but still turn up in the TLG corpus. That's a lot of lemmata, given that dictionaries have been making a point of filling in each other's gaps (Trapp in particular).

There are a few reasons for Pape's 545 not to have been recorded anywhere else. Pape has not picked up the allergy to Late Antiquity that LSJ did. This allergy has meant that pagan late antiquity is the least well covered period in Greek lexicography. As more inscriptions and papyri turned up, LSJ displaced late entries with the newly found earlier entries. With these late entries attributed to no author more specific than "Eccl." or "Byz.", LSJ wasn't desperate to hold on to them, to begin with.

Pape has the same dismissive "Byz." attribution for its late entries; but it has not undergone the same kind of cull—to the benefit of the Greek Anthology and Oribasius.

And, to my surprise, of the scholiasts. The scholiasts are something of a discomfort to lexicographers. Scholiasts provide their own definitions of Classical words, so Classical lexicographers care about the words the scholiasts use. But scholiasts explain Classical words to post-Classical audiences, by using decidedly post-Classical vocabulary. Which means that documenting the scholiasts puts Classicists in the Mediaeval Greek business. (Not always wisely, as seen with LSJ's mishandling of στοίχημα "wager".)

There is a smaller group of lemmata in Pape which the other lexica overlook for different reasons—and which by rights should not be turning up in a Classical corpus at all. The text of the Classical canon is described by lexicographers following standard editions; but the manuscripts they work from had a lot more variability than the lexicographers now need to account for. Our modern edition of Pindar has established that Pythian Ode 8.74 reads
πολλοῖς σοφὸς δοκεῖ πεδ’ ἀφρόνων

to many he seems wise among fools

So we are not interested that in some mediaeval manuscripts, the last two words πεδ’ ἀφρόνων "with the non-prudent" were run together as πεδαφρόνων "of the after-prudent"—or, as Pape has it:
πεδά-φρων, ον formerly appeared in Pind. P. 8.74, and was explained as Aeolic for μετάφρων, someone who is wise later, after the deed; Böckh writes it as πέδ' ἀφρόνων.

Once Böckh settled that πεδάφρων did not exist in Pindar, πεδάφρων ceases to be of interest for Classicists.

But the people who read those mediaeval manuscripts thought πεδάφρων existed. The scholia are commentaries on mediaeval versions of the Classical texts, and reflect the mediaeval understandings of those texts. If the scholia describe these misreadings, they perpetuate them, even after we have cleaned up the source text (to our best judgement). So our Thucydides 8.91.3 now reads
ἦν δέ τι καὶ τοιοῦτον ἀπὸ τῶν τὴν κατηγορίαν ἐχόντων, καὶ οὐ πάνυ διαβολὴ μόνον τοῦ λόγου

This was no mere slander, there being really some such plan entertained by the accused.

Somewhere along the line διαβολὴ μόνον "mere slander" was metanalysed as διαβόλιμον ὄν "being liable to slander". The scholiast accordingly tries in the margin to make sense of the word they saw in the main text. Our main text is emended, but the margin still counts as text in the Greek corpus:
διαβόλιμον ὄν: ἤτοι καὶ οὐκ ἔχοντος τοῦ λόγου διαβολὴν (?), οὐδαμῆ εἶχεν ὁ λόγος διαβόλως (?)
being liable to slander: namely, not having slander in speech (?); the speech was in no way slanderous (?)

(You can thank Karl Hude, who edited the scholia, for the incredulous question marks.) Thomas Magister also recorded the word in his "Selection of Attic nouns and verbs"—although he was reluctant to go so far as to commend it:
Διαβόλιμον Θουκυδίδης λέγει τὸ διαβεβλημένον· καὶ οὐ πάνυ διαβόλιμον ὂν ἀπὸ τῶν Μεγάρων τὴν Σαλαμῖνα παραπλεῖν. σὺ δὲ διαβεβλημένον λέγε.
liable to slander is what Thucydides calls something slandered. "Without it being liable to slander that they should sail by from Megara to Salamis". But you should say "slandered".

You should also not trust Thomas' reading of Thucydides; the sailing from Megara to Salamis is from Thucydides 8.94.1, a few pages on, and the mission talked about as διαβόλιμον was heading to Eetonia near Piraeus.

The mediaeval manuscripts were not just read by the Byzantine scholiasts and lexicographers, though. The manuscripts—and the scholia and lexica explaining them—were how the Byzantines learned the Classical language, on which they modelled their own literary language. That means that διαβόλιμον is of interest in the study of later Greek. Sure, it's a ghost word, like dord, and it was never part of any spoken form of Greek.

But if the text of Pindar that Byzantine writers read said πεδάφρων, then writers would assume it to be real enough: no less real than μετάφρων, which they could have coined in their own, Atticist learnèd Greek. No less real, for that matter, than embiggen.

The main reason these ghost words were taken up was that the ghost words did make sense on their own: they could be analysed according to the rules of Greek compounding. If you know how Classical Greek works, you already know that διαβόλιμον really does mean "liable to calumny", because -ιμος is a real suffix (cf. ἀγώγιμος "carriable"); and once you know that πέδα is Aeolic for μετά, you know that πεδάφρων must somehow mean "after-mined" (cf. ἔκφρων "out of one's mind"). The misreading of διαβολὴ μόνον or πεδ’ ἀφρόνων is not alien to spoken language after all; it's the same metanalysis that gives us adder or newt—or helipad or pollute. So it may look like Byzantine Greek writers were asking for this kind of bogosity, by slavishly relying on flawed manuscripts for their models; but what they came up with was not that different to what spoken language has done elsewhere.

And so it is no surprise that πεδάφρων shows up in the TLG too, even if it is not in the TLG's Pindar. It turns up in someone who has read Pindar in those corrupted manuscripts, and is happy to show his reading off. But this is not a scholiast or lexicographer. This time, it's a reader who is producing literature of his own: Leo Allatius, in his poem Hellas, written in the 17th century. (A couple of centuries before Böckh.)

My Ancient Greek isn't as good as I might make it seem, and Allatius' run-on syntax doesn't make it easier; but this is how I think he uses πεδάφρων—in the same genitive πεδαφρόνων as in the pre-Böckh Pindar.
ΕΛΛΑΣ γάρ εἰμι, τέκνον, ΕΛΛΑΣ, ἧς κλέος
ἄσβεστον ἔργοις τιμίοις πεπραγμένον,
διελθὸν ἔπτη πρός τε γαῖαν, καὶ διὰ
πόντου πέρασσε νυκτός, ἠδ’ ἠοῦς λάχος,
ἠδ’ εἴ τι ποῦ ἐστι λαιά, κἀπιδέξια
συμφραδμόνεσσι τηλόθεν γ’ ᾠκισμένον,
πεδαφρόνων γὰρ οὐδὲ μικρόν μοι μέλει,
ὃ ΓΑΛΛΙΚΟΙΣ στήθεσφι τοῦδ’ ἄχρις χρόνου
ἄφθαρτον ἐμπεφυκὸς ἀγλαΐζεται. (vv. 196–204)
For I am Hellas, child, Hellas, whose inextinguishable
glory, carried out in honourable deeds,
is past, has crouched down to the ground, and has
passed through the sea of the night: this is the doom of the dawn;
but it is also what is there in a place, that is dwelled in
left and right by counsellors from afar,
(for I do not care the slightest for those wise after the fact):
a thing which is splendid in Gallic breasts
growing incorruptible to this day.

Pindar is not the only Classical author whose fractured words turn up in both Allatius and Pape. ναυβάτης "ship-walker" turns up in several Classical sources—Aeschylus, Sophocles, Euripides, Herodotus, Thucydides, Xenophon). The most obscure source is Lycophron:
καὶ τὰς Ἐρεμβῶν ναυβάταις ἠχθημένας προβλῆτας ἀκτάς (vv. 827–828)
and the jutting shores of the Erembi, abhorred by mariners

The TLG's edition of Lycophron dates from 1964; Pape in 1880 records ναυάτης as Lycophron's reading, and also has it as a variant reading in Euripides, IT 1380. Our Lycophron and Euripides have now been cleaned up to read ναυβάτης; Allatius' had not, so that his poem has the first appearance of ναυάτης in the TLG corpus:
Ἀργοῖς ἐρετμοῖς πόντος, εἰ καὶ μαίνεται,
κλύδωνί τ’ οἰχθείς, ψάμμον ἐκβράζει βυθοῖς,
βύκτας ἀέλλας ἡμερώσας ναυάτης,
οἰηΐοις τε νηὸς ἰθύνας δρόμον,
πρὸς ὅρμον ἔλσεν ἀσκηθὴς σόον σκάφος. (vv. 233–237)
Though the sea rages, and throws up sand from the depths,
departing with slow oars and through the billow,
the mariner calms the blustering whirlwind,
and straightens his course with the ship's rudder,
to shelter the vessel unscathed in its anchorage.

(If you go searching the TLG for ναυατ-, btw, ignore Ναυάται in Germanus I, Narratio de haeresibus et synodis ad Anthimum diaconum 48. His Nauatae predate Allatius by eight centuries, but they are Novatianists, a sect, as you would expect in an anti-heretical text.)

The misreading of ναυβάτης as ναυάτης is unsurprising: they sound the same in Modern Greek, [naˈvatis]. Unlike πεδάφρων or διαβόλιμον, ναυάτης doesn't mean anything on its own in Greek. It was taken up because it still looks like another real word of Greek, ναύτης "shipper = sailor", which has survived into the Modern language. Copyists simplified ναυβάτης into ναυάτης, and then figured that Euripides was just using ναύτης with an extra α in the middle. The Ancients did strange this like that, after all, didn't they?

(The Baroque did strange things too, just in a different way.)

Outside the lexicographers and scholiasts, there aren't many instances of such rejected readings turning up in literary Greek. ἠπίαμα "cure", as defined in Pape, is a metanalysis of Herodotus 3.130 ἤπια μετὰ τὰ ἰσχυρὰ προσάγων "supplying both mild and strong (medicine)"; the misquote turns up in Constantine Porphyrogenitus De virtutis et vitiis II p. 11, and Suda, delta 442; but that doesn't count as new usage. The run-in εὐ ναιόμενος "well-dwelling" of Iliad 14.255, which Pape allows as a single verb, turns up as a distinct verb not just in commentators, but also in Galen (Kühn 18b p. 763). But that's it. (I thought I saw an instance in Gregory of Nazianzen's aping of Homer, but I can't find it.)

That may mean that Allatius' Greek is more derivative than his forebears; I have my doubts, given how Byzantine literary culture worked. Since I did not track down words already defined in Lampe and Trapp, it's likelier that any earlier such repurposings had already been dealt with there.

… And so I return to blogging. Missed it.


  1. Would ναυάτης be glossed 'a student of Na'vi'?

    I don't understand what pollute is doing in this post. The OED calls the por- in polluere an ablaut variant of pro, and I suppose you could call it metathesis, but how is it metanalysis?

    (Say, didja know that Rotuman has systematic metathesis? In certain syntactic constructions, the final vowel of a morpheme is lost, but umlauts the vowel of the preceding syllable into a copy of itself. The exact outcome depends on the identity of the two vowels.)

    And welcome back!

  2. Dear Nick,

    some may well perceive this as something between an esoteric question and an inside joke, but let me heat things up a little bit by asking: have you, during your lexical travels and quests, ever encountered the term λεξικοκυλισμένος?


The Other Place (opɯcɯluklɑr)

Powered by Blogger Widgets