o-vocatives: Analogical Account, IV: What Henrich said

I’ve finally taken the time to read Günther Henrich’s 1976 thesis on the spread of the -o vocative and -o genitive in Greek.

My blog series has been something like 15 pp written off the cuff, with minimal research. Henrich’s is 270 pp of meticulous historical and dialectal research. He has orders of magnitude more detail than I provided, and his tracing of developments is much more fine-grained than I could have done. (Of course. After all, I spent a couple of evenings on this, and he spent several years.)

It’s an impressive piece of work, all the more so because it was done in 1976; and I’m always impressed with the kind of work people used to do with index cards instead of computer corpora. (My own PhD was at the tail end of that.) I haven’t been well disposed the recent Bible Code-like attempts by Henrich to read cryptogram signatures into the works of the Cretan Renaissance; but the quality of research in this work has more than made up for that.

And I’ll allow myself the gratification that he hasn’t contradicted me. The developments I posited, like the neuter for Manolio(s) “Manuel”, were confirmed. The claims of analogical factors that I’ve made are all there in his thesis: formality, penult stress, proper names, all given with quantitative justification. (Of those original third declension nouns like ɡerɔːn that have switched to the second declension, and which are the core of the o-vocative, 3/4 are penult-stressed, and 2/3 are proper names.)

He also a quirky-looking derivation for the o-vocative (and its related o-genitive), which seemed implausible to me at first, but which has grown on me:

	nom.	gen.	acc.	voc.
Ancient	ɡerɔːn	ɡerontos	ɡeronta	ɡeron
Early Mediaeval (ca. 1000): Indeclinable	ɣeron	ɣeron	ɣeron	ɣeron
Later Mediaeval; Cretan	ɣeros	ɣero(n)	ɣero(n)	ɣero(n)
Modern	ɣeros	ɣeru	ɣero	ɣero

In other words, the old third declension in -ɔːn did not switch wholesale to the second declension (-os, -u, -o, -e); it moved one case at a time, and the vocative (subject to less analogical pressure than the genitive, as a more rare case) is the contemporary battleground. That means that Cretan, with its o-genitive corresponding to the o-vocative, is archaic rather than innovative, which is consistent with the o-genitive being more common in Mediaeval Greek. (That also makes more sense because it’s harder to think of the vocative or accusative influencing the genitive, which is what would be required if the o-genitive were innovative.)

Because Henrich had a lot more old data to work from than I did, and because he’s taken more of an old school historical linguistics approach, he spends a lot more time on how the phenomenon got started than how it spread. Not only does he analyse how the original nom. ɡerɔːn, gen. ɡerontos, voc. ɡeron gave rise to the o-vocative; he also introduces a lot of pathways for the o-vocative to have been continually imported into the language, mainly through borrowings (proper names from every Balkan people there ever was, and both proper names and common nouns from Venetian/Italian).

It is a subtle game to establish the relative importance of the factors that bring about language change: the initial reanalysis or borrowing, that brings the change into being, versus the subsequent analogy that generalises it through the language. My own opinion is that Henrich trusts etymology too much: I’m not convinced all those Albanian and Bulgarian names were imported complete with an o-vocative from the very beginning, and without the powerful analogy of proper names already in place, I don’t think the loanwords would have lined up so promptly.

And not all of Henrich’s etymological accounts have the same explanatory power. For example, he thinks that the o-vocative is particularly popular after velars, because it avoids the switch between palatal and velar, which makes the morphology less transparent (that’s how he explains [kakurɣos ~ kakurʝe] “criminal”, [likos ~ lice] “wolf”); but those allophonic switches happen in verbs and nouns constantly, and native speakers are not aware of them at all.

He also posits -ios > -os as a class of nouns that takes the o-vocative: that accounts for eɣiptios > ɣiftos “gypsy”, ɣeorɣios > ɣiorɣos “George”, and several diminutives like -akos > -akios. But that phonological simplification does not particularly explain why the vocative -ie would switch into -o. In fact the laundry lists of reanalysed and borrowed names that take the o-vocative (names formed from verbs! names formed from adverbs! Albanian diminutives! Slavic diminutives!) have a simpler underlying explanation—which he does mention in passing: novel name forms in general will tend to take the analogically spreading form, since the conservative force of the inherited e-vocative is not present there to push it back. All because the o-vocative is now the unmarked strategy for penult proper names (as he himself concludes, already in 1976: “all penult-stressed proper nouns ending in –os can potentially take an o-vocative.”)

And where the o-vocative pushes into previously unavailable terrain (like common nouns), I think semantic similarly counts for a lot more than phonology in how the analogical spread works. The vocative kakurɣo after all shows up next to the vocative ðolofono “murderer”, which has no velar—but a lot semantically in common with the main contemporary class of common nouns taking the o-vocative (murɣos “uncouth”, bufos “fool”, rufianos “informant”).

On the other hand, I probably underestimated the contribution of loanwords: there certainly appear to be a disproportionate number of Italian words among the common nouns that do use the o-vocative, and the dialects where the o-vocative has prospered most (including extending to antepenults) have been the dialects with the longest Italian contact, Cretan and Heptanesian. The fact that Nicolò was indeclinable in Cretan Greek notary documents up until 1500 (appearing as ŋikolo even in the nominative) shows that the o-vocative was certainly going to be reinforced by the influx of Italian names. I had guessed as much, but I hadn’t realised how much that extended to common nouns as well (like say filiotsos < figliozzo “godchild”). And many of the contemporary common nouns that take an o-vocative can be traced to Italian (kamarotos, kouniaðos, bufos, rufianos, tsarlatanos).

The other overall thing I noticed about our different accounts was that too much detail gets in the way of the big picture. That’s not necessarily a fault in an historical account, of course; but analogy does not work by appeal to etymological detail; it works by squinting the eyes, and blurring all the detail into large, arbitrary generalisations. Sort of like I did with my account. Which, after all, was seeking to explain not what had happened in the 11th or 16th century, but what generalisations explain what has been taking place over the last century.

It’s interesting to see both what hasn’t changed and what has. Renaissance Cretan had anticipated a lot of the developments taking place now in Standard Modern Greek (since the o-vocative was more advanced there); the modern military slang neo “newbie”, for example, turns up in a tragedy around 1600, and Phoebus (which Cretan dramatists had only heard of via Italian) gets an o-vocative, like any Italian name: Φοίμπο fibo. (That’s not the same thing as o-vocatives now spreading to learnèd names; but that’s how it gets started. In fact, a much more modern play, Petros Markaris’ translation of Brecht’s Life of Galileo, makes a point of having scholars in the play use the vocative ɣalilee, while commoners use the vocative ɣalileo.)

On the other hand, Henrich still found very little evidence of o-vocatives spreading to antepenult names, let alone common nouns; that’s something that has changed in the last four decades.

It’s also frustrating to realise the examples you don’t come up with if you’re thinking off the top of your head, instead of doing research. I wouldn’t have made the connection between my name, Nikos, and the Mediaeval name Nikōn; it’s too hard coded in my head that Nikos is short for Nicholas, but that doesn’t mean it can’t have initially been the regular development of Nikon. But I should certainly have realised that the Grim Reaper, Charon < Charos, was a canonical example of the shift of -ɔːn to the second declension (and its o-vocative is attested as early as the late 15th century, in Emmanuel Limenites’ Plague of Rhodes. Which, I notice, was recently published in a popular edition by one Günther Henrich.)

Oh, and Philip Newton? Good catch on the e-vocative in γέρε being associated with adjectival use. Henrich argues that the o-vocative does not stick with adjectives, because there is much more analogical pressure to align adjectives with the well-established first and second declensions, since adjectives decline in all genders and numbers. In fact, it’s why third declension adjectives like askʰɛːmɔːn “misshapen” did not keep an o-vocative when they switched to the second declension: nom.masc.sg asximos, voc.masc.sg asxime “ugly”.