o-vocatives: Analogical account, Part II

So far we have accounted for:

• M1: Bisyllabic common nouns that used to be third declension: ˈɣeros “old man”, ˈðjakos “deacon”. (Ancient ɡérɔːn, diákɔːn).
• M2: Bisyllabic truncated, informal given names: ˈɣjorɣos, ˈnikos, ˈðimos (corresponding to the formal forms ɣeorɣios, nikolaos, ðimitrios) “George, Nick, Dimitri”
• M3: The trisyllabic (truncated) name aˈlekos “Alec”
• O2: Bisyllabic formal given names (which are used in both formal and informal language): ˈpavlos, ˈpetros, ˈstavros, ˈmarkos “Paul, Peter, Stavros, Mark”

We accounted for M3 in passing: the bisyllabic restriction on o-vocatives does not appear to have been particularly important; so the class of bisyllabic truncated informal given names in M2 was straightforwardly extended to trisyllabic truncated informal given names. The grammarians have puzzled over aˈlekos as an exception; but the economic way of stating the rule for given names was not about them being bisyllabic (although it’s understandable that grammarians would prefer that criterion, as more linguistically clear cut); it is about them being vernacular and familiar.

So the features that define the o-vocative so far are: GIVEN name, PENULT-ACCENTED, TRUNCATED and VERNACULAR/FAMILIAR. We have just seen that the BISYLLABIC feature was quickly ignored.

The names in O3, which have escaped grammarians’ notice, have dropped the TRUNCATED feature, as indeed O2 have: avɣuˈstinos, maˈrinos, puˈlikos, leoˈnikos “Augustin, Marinos, Poulikos, Leonico”. These names have both e-vocatives and o-vocatives, just like O2. O2 keeps the e-vocative because they are FORMAL as well as VERNACULAR; O3 have taken up the o-vocative because they too are PENULT-ACCENTED, GIVEN, and particularly VERNACULAR: they have no Hellenic heritage, and you’d be hard put to find an Orthodox saint’s name behind them. They keep the e-vocative as well, because they are not as typical of the category of o-vocatives: they are not BISYLLABIC, and unlike M3 they are not TRUNCATED, nor especially FAMILIAR.

What is promoting the o-vocative for those words appears to be that they are disconnected from formal names as used by the Orthodox church, just as the truncated forms like nikos are. Poulikos appears to be a (Pontic?) diminutive of Paul, which means it’s in the same bucket as Alekos and Nikos. There are a couple of St Marinus commemorated by the Greek Orthodox Church (Marinus the Elder, born in Tarsus, martyred in 304; Marinus born in Rome, and a Roman senator, late 3rd century), and Augustine of Hippo is officially an Orthodox saint, although not popular among Orthodox theologians; but it is fair to say they are obscure saints in Orthodoxy (unlike St Marina), so there would likely have been less occasion to hear e-vocatives from them. Leonico is Ladino, and Sephardic Jews would not have an equivalent to liturgical Greek to promote the e-vocative. (The minute commenter Leonicos hellenised his name from leoˈnikos to leˈonikos, he noted, he also hellenised its vocative.)

Friend to this blog commenter Diver Of Sinks raised roˈðolfos “Rudolf” as a counterexample: a trisyllabic given name that does not have an o-vocative. “Rudolf” is clearly not a Hellenic name, nor a name commemorated in the Orthodox church; so the e-vocative isn’t pushing back on the o-vocative because of Koine Greek. But it doesn’t need to. Greek speakers recognise that roˈðolfos is a learnèd loan into the language: the -lf- cluster is enough to indicate that (cf. puristic aðelfos vernacular aðerfos “brother”), and the fact it’s German, inflected, and with a [ð]. Linguistically, roˈðolfos is not German: it’s learnèd. It might as well be Ancient Greek linguistically. And speakers know that learnèd names do not have o-vocatives.

The names in M5, ɣjanaˈkos, ðimiˈtros, ˈnikoljos, manoˈljos “John, Dimitris, Manuel” are given names, vernacular, and familiar, all like O2. But they violate the pervasive restriction of the o-vocative to penult-stressed forms: they are stressed on the final syllable. The analogy with familiar forms has proven stronger than the phonological constraint on stress; but given how pervasive the phonological constraint is elsewhere, one would think that it has had some help to push through. My own speculation is that these forms are based on Italian diminutives: Niccolò, Manolò—and hearing Niccolò, Manolò as vocatives would have reinforced the analogy to use o-vocatives with them when they were taken up in Greek, even though they were the only such instance to use the o-vocative on the final syllable.

The words in M4 and O4, ending in diminutive suffixes like -akos, reflect a different and more far-reaching breakdown, that of the restriction of o-vocatives to given names. In the 1941 Triantafyllidis grammar, those diminutive suffixes are the only instances where the o-vocative shows up in surnames (O4) and common nouns (M4b); so it is reasonable to assume that this class of words started off as given names (M4a), and then extended to surnames and common nouns with the same suffixes.

At the beginning, presumably, came diminutive given names like ɣjorˈɣakos, nikoˈlakos “George, Nick”; they call in the same class as O3, and we just say puˈlikos as a similar name with a slightly different diminutive suffix, and both of them share the feature FAMILIAR with the core class of O2. The o-vocative then would have spread to given names that ended in -akos without any diminutive meaning (kiriˈakos “Cyriac” < “of the Lord”—though you could argue that the suffixes are ultimately related).

And once the o-vocative was associated with diminutive suffixes in given names, like Poulikos and Nikolakos and (seemingly) Kiriakos, it is straightforward analogy to extend the o-vocative to other words ending in the same suffixes: surnames like Dimitrakos and Sarantakos, and common nouns like anθropakos. In the case of the vernacular diminutives, there was not a strong enough pushback by an e-vocative, even if they are common nouns: the vocatives are universally anθropako, filarako “puny little man; buddy”.

Surnames however as they are used in the Modern Greek State are formal Greek, which the State has had its hand in, and speakers know they are formal Greek: they retain Puristic phonology, and Puristic inflections. Because they are formal, and because they are a more recent category, speakers will tend to use e-vocatives for them more than for given names; which is why the o-vocative is still optional with surnames like Sarantakos and Dimitrakos—and is not optional when those same forms are used as given names.

The more recent trend in surnames, reported in Katsoudas’ grammar, is a new rule (O5): etymologically transparent surnames take e-vocatives, etymologically opaque surnames take o-vocatives. There’s an obvious analogy behind that rule too: o-vocatives are characteristic of proper names, therefore e-vocatives are characteristic of common nouns; and if a surname is identical to a common noun, then the analogy with the common noun is pressure to retain the e-vocative. In the absence of that countervailing pressure, the default pattern takes over; and it allows us to see that there has been a markedness shift in names since 1941. In 1941, only diminutive-looking surnames took the o-vocative. Now, the o-vocative has become the default for surnames.

Just as the o-vocative has in fact become the default for given names: the given names that retain the e-vocative have in reality become just those names where the e-vocative is reinforced from Koine or Ancient or Puristic Greek. Which would explain the hitherto sporadic instances reported for given names with o-vocatives accented in the antepenult, like ˈmario (O6): a Romance, unchurchlike given name like in O3, and the stress is no longer a barrier to the o-vocative. For that matter, just like highly Hellenic names like liˈkurɣos “Lycurgus”: penult-stressed like Nikos and Marinos, but Ancient Greek, which should have blocked the o-vocative from ever appearing. Yet Panos in Lowercase reports that he has heard the vocative liˈkurɣo, which suggests that that constraint, too, is starting to break down. And there will be more: commenter Manousos reports an o-vocative for anˈðronikos “Andronicus”, which is not only learnèd (complete with unvernacular -nðr- cluster), but accented on the antepenult—breaking both the restrictions to date. (“Speakers know that learnèd names do not have o-vocatives”, I said above with relation to roˈðolfos “Rudolf”. Not so much any more.)

These analogies are spreading at different rates for different people; it causes me some glee to note that Nikos Sarantakos himself is clearly on the conservative side when it comes to most of these, as well as the extensions of the o-vocative to common nouns. That doesn’t make him a fuddy-duddy blocking language progress. (Or at least, that in itself doesn’t. 🙂 Nor for that matter is guaranteed that all names will end up dropping the o-vocative: the last two decades has seen a resurgence of archaising purism in Greek, and the o-vocatives could yet be stigmatised once the purists actually notice them. But it does mean that the system of Greek vocatives is a lot more in flux than anyone seeking to codify Greek grammar can be comfortable with.

More in flux than anyone trying to decline words consistently can be comfortable with, in fact, and they clearly aren’t comfortable. The use of o-vocatives is unsteady, and strongly influenced by context, because people’s language instinct is uncertain about how to form the vocative for any of these. Thence the spontaneous debates about the proper vocative of “Augustin”. Thence forms like ðimo verikio, where the o-vocative on the etymologically opaque surname confounded Giannis Haris; but if the rule around O5 is new and unsteady to begin with, the mere presence of an O2 o-vocative in ðimo will be enough to undo it.