Generalised use of να in Early Modern Greek

I've been reluctant to write this post for a couple of reasons:

1. It requires dropping a moderate amount of linguistic science;
   
2. I'm not prepared to do either the research or the bibliographic survey to back it up;
   
3. It's probably already been worked out by the Grammar of Mediaeval Greek people.

If the latter is the case, then it won't be the first time Notis is more up to date than I am. But I'll write, and I'll see who bites.

In a previous post, I mentioned that I'd like to do a detailed linguistic survey of the 1420 letter by Manuel Chantakites, and how its language differs from contemporary standard Greek (and for that matter contemporary Cretan). Readers remarked that the letter was remarkably readable. But there is one feature of the text that, although readable, is emphatically absent from contemporary Greek. It has to do with how να is used.

You can no longer say sentences like this:

• ωσάν ήκουσα να γυρεύγεις της νύφης σου νοίκια, "when I heard that you are asking for rent from your daughter-in-law" (Manuel is not physically present)
  
• ήμαθα και ήλεγες να εβγάλεις την νύφην σου από τα σπίτια της, "I have learned you've been saying that you would kick your daughter-in-law out of her houses"

The phrases would now be:

• σαν άκουσα πως/ότι γυρεύεις από τη νύφη σου νοίκια
  
• έμαθα πως έλεγες πως/ότι θα βγάλεις τη νύφη σου από τα σπίτια της.

I'm going to work through why this is an oddity, and what I think has happened. I am also going to tie it in to another expression which is just as impossible in Modern Greek:

• Τότε να είδες τα βουνά, να είδες και τους κάμπους / να είδες και τα δάσητα, να είδες τας λαγκάδας, "You should have seen the mountains and the prairies! You should have seen the forests and ravines!" (Entertaining Tale of Quadrupeds 103-104)

Modern Greek would use after να not the perfective past (the aorist indicative, historically), but the imperfective past (the imperfect indicative, historically):

Τότε να έβλεπες τα βουνά, να έβλεπες και τους κάμπους / να έβλεπες και τα δάση, να έβλεπες τις λαγκάδες

And I'm going to claim these differences are related.

It will take a lot less time to say this if I presuppose you're all linguists, which you're not; I'm going to give the opaque summary here, and the semantics lesson under the fold.

Early Modern Greek να appears to have been used more broadly as a complementiser than its current distribution associated with events and dubitatives, contrary to its purposive etymology and indeed the further restriction on factual complements in Modern Cretan. This may be associated with a temporary generalisation of να to more "indicative" contexts, e.g. as a future marker, and its use with the historical indicative aorist in optative as well as evidential contexts.

Understood? No? Well, let's walk through this.

In Modern Greek, να is the main subjunctive marker; historically it is derived from ἵνα "in order to". να has displaced the Ancient infinitive, so verbs with να appear where Ancient Greek used the subjunctive, the optative, or the infinitive. One of its main functions, just like the infinitive's, is to introduce complements of verbs: phrases that are the objects of verbs. ("I like to watch", Μου αρέσει να κοιτάζω, φιλῶ βλέπειν).

But whereas the Ancient infinitive could introduce a broad range of complements, να is more restricted. Like the English infinitive, it mostly introduces events, things that happen; and unlike a lot of "indirect discourse" Classical infinitives, it does *not* introduce facts or propositions, pieces of knowledge. Those instead are introduced by ότι or πως. (Some complements also have που or και or nothing, but I don't want to get distracted.) So:

• Άρχισα να διαβάζω, "I began to study"
  
• Θέλω να διαβάσει, "I want him to study"
  
• Σε διατάζω να διαβάσεις, "I command you to study"
  
• Λέει πως διαβάζει, "He says he's studying"
  
• Νομίζω πως διαβάζει, "I think he's studying"

In fact, by switching between να and πως, you change whether a fact or an event is involved, and thereby you change the meaning of the verb. Which is pretty common across languages:

• Τον άκουσα να βρίζει, "I heard him swear" (I perceived an event)
  
• Άκουσα πως έβριζε, "I heard that he swore" (I perceived a claim = proposition, and not an event)
  
• Του είπα να διαβάσει, "I told him to study" (I spoke an event—which is reintepreted to: I commanded an event)
  
• Λέω να διαβάσω, "I'm considering studying" (lit. "I say that I should study": I spoke an event I will carry out—which idiomatically means I'm thinking aloud about doing an event, I'm making a commitment)
  
• Ξέρω πως διαβάζω, "I know that I'm studying" (I know a fact)
  
• Ξέρω να διαβάζω, "I know how to study" (I know an event—I know how to make the event happen)

There is an exception in Modern Greek where να can introduce a proposition. That means putting a proposition in a grammatical context which doesn't fit propositions; and the context is expressing doubt about the proposition:

• Πιστεύω πως διαβάζει, "I think that he's studying"
  
• Πιστεύω να διαβάζει, "I *think* that he might be studying"

You can't do that with more certain verbs: ??Ξέρω να διαβάζει, "I know that he might be studying". (Works in English, not in Greek.) And that exception makes sense because of comparable constructions, where the να-clause introduces an inference, rather than a certain fact—i.e., again, a proposition:

• Πρέπει να διαβάσει, "He must study" (an event that must happen)
  
• Πρέπει να διάβασε, "He must have studied" (a proposition that must be true)
  
• Να διαβάζει; "Might he be studying?"

The doubting clauses and the inferences are all tied together by the tenses they allow. Normally, να introduces a present imperfective or present perfective—the historical present subjunctive and aorist subjunctive. It can introduce the present imperfect—the historical imperfect indicative—in contexts much like the English conditional and the Classical optative: wishes, counterfactual conditions, future-in-past:

• Αχ να διάβαζε! "If only he'd study"
  
• Αν διάβαζε, θα πρόκοβε "If he'd study, he'd get ahead"
  
• Μου είπε πως θα διάβαζε "He said he'd study (= was going to study)"

I'm treating the future marker θα as an extension of να: it is derived from θέλω να "I want to" (like English will), and has similar tense restrictions.

The past perfective—the historical aorist indicative—is normally too "indicative" to appear with να. As it turns out, it never appears with events:

• *Άρχισα να διάβασα
  
• *Θέλω να διάβασε
  
• *Σε διατάζω να διάβασες

It only appears with propositions put in doubt, and inferences (which are intrinsically doubtful):

• Πιστεύω να διάβασε, "I think he might have study"
  
• Πρέπει να διάβασε, "He must have studied"
  
• Να διάβασε; "Might he have studied?
  
• Μου είπε πως θα διάβασε "He said that he will have studied (= must have) studied)"

So this is a reinterpretation of να, away from events and into doubtful propositions. We could speculate that the shift of πρέπει "must", from obligation to inference (deontic to epistemic), came first, because it would explain the shift from event to proposition; but the Ancient subjunctive also conveyed this sort of doubt as a main clause. Typically introduced by μη, but then the Modern equivalent is typically introduced by μήπως:

• μὴ ἀγροικότερον ᾖ τὸ ἀληθές εἰπεῖν: μήπως να 'ναι πιο αγροίκο να πούμε την αλήθεια: "it might be too rude to speak the truth"
  
• εἴπωμεν ἢ σιγώμεν; να μιλήσουμε ή να σιωπήσουμε; "should we speak or keep silent?"

So it could also be an inherited pattern. The tense usage though is an innovation: there's nothing in Ancient Greek to suggest using an "indicative" tense here.

The distribution of ότι/πως vs. να in complements—ότι/πως for facts and propositions, να for events—is pretty stable across Greek dialect. I don't know of any instances where να picks up more than doubtful propositions. In Modern Cretan at least, even that exception is rolled back: "He must be studying" is Πρέπει πως διαβάζει, just like Νομίζω πως διαβάζει and Ξέρω πως διαβάζει (I think/I know that he is reading). So in Cretan, the analogy has gone the other way: analogy has switched "must" from events to propositions, and analogy shifts its following complementiser from the event marker to the proposition marker.

And if you're historically minded, you can find justification for the distribution in the etymology of να: its ancestor ἵνα "in order to" introduced events, things you intended to do. The restriction of the past perfective is also neat, and this all suggests a clearcut system in place since the beginning of Early Modern Greek.

The examples at the start of the post (remember?) do not. The verse from the Quadrupeds shows the past perfective used with wishes, where Modern Greek would use the "optative" past imperfective:

• Τότε να είδες τα βουνά, να είδες και τους κάμπους / να είδες και τα δάσητα, να είδες τας λαγκάδας, "You should have seen the mountains and the prairies! You should have seen the forests and ravines!"

But that breaks the neat association between the indicative tense and doubt. The wish it introduces is still a proposition that hasn't happened, but doubt is not the point of a wish, and Modern Greek doesn't make that extension. Which means that the Modern restriction wasn't always in place: Early Modern Greek didn't distinguish between wishes and doubts, both could take a past tense after να, imperfective or perfective.

Chantakites' letter points to another breakdown. I've seen similar breakdowns in the Chronicle of the Morea, with ότι να used where Modern Greek uses just ότι (on which more below); but Chantakites' letter is more persuasive because its Greek really does look more unforced. And what Chantakites does is break down the Modern distinction between perceiving an event, with να, and perceiving a fact (second hand), through πως. Manuel is miles away from Crete, but

• ωσάν ήκουσα να γυρεύγεις της νύφης σου νοίκια, "when I heard that you are asking for rent from your daughter-in-law"

In Modern Greek, you can only use να after "hear" if you're directly perceiving an event: σαν σε άκουσα να γυρεύεις από τη νύφη σου νοίκια, "when I heard you ask(ing) for rent from your daughter-in-law". And you'd have to insert the direct object, σε "you": the syntax of the dependent clause is different, its subject is raised to become the object of the main verb, expressing perception ("I heard him ask" = "I heard the event in which he asked", "I heard him while he was asking"). Whereas what Chantakites said treats να just like ότι, with no raising of the subject.

What's going on? I think there's a hint in the other instance:

• ήμαθα και ήλεγες να εβγάλεις την νύφην σου από τα σπίτια της, "I have learned you've been saying that you would kick your daughter-in-law out of her houses"

The letter has some instances of what was to become the Modern Greek future construction, εγνωρίσει θέλω "to know I want" > θέλω να γνωρίσω "I want to know" > θα γνωρίσω "I will know" (and Modern Cretan να γνωρίσει θέλω). But that was the start of the future construction: until then in Early Modern Greek, as I've mentioned in earlier posts, the future was expressed with να.

• Michael Glycas: ὁ βασιλεὺς φιλάνθρωπος καὶ νὰ σὲ συμπαθήσῃ, "the kindly emperor's gonna pardon you"
  
• Continuators of Scylitzes: Ἐγὼ σὲ ἔκτισα, φοῦρνε, καὶ ἐγὼ νὰ σὲ χαλάσω, "I built you, oven, and I'll destroy you."

You might argue this construction is still there in Chantakites:

• Και εθάρρουν εις εσέναν … να φλογοτομάς το αίμα, "I trusted in you even to bleed your own veins" ("I trusted in you that you would bleed your own veins"?)

And you can also read the other instance of an odd να that way:

• ήμαθα και ήλεγες να εβγάλεις την νύφην σου από τα σπίτια της, "I have learned you've been saying that you would kick your daughter-in-law out of her houses"

This could be like the idiomatic Λέω να διαβάσω Modern construction: "I'm saying for me to read = I'm thinking aloud that I should read = I'm considering reading." Sounds odd to use it outside the first person, but it's possible. But it's likelier that this use of ήλεγες is literal: Manuel has heard reports of his father speaking, and να introduces what he was speaking, as a proposition rather than a figurative event: it is equivalent to Modern έλεγες πως, "you have been saying that…".

The Modern Greek would continue: έλεγες πως θα βγάλεις τη νύφη σου, "you have been saying that you will kick out your daughter-in-law". With a future tense construction. But θα βγάλεις used to be να εβγάλεις, with the old future use of να. So in 1100, this would have been: έλεγες ότι να εβγάλεις τη νύφη σου—with the ότι να combination you can see in the Chronicle of Morea:

• Ἐνταῦτα ἀπήρασιν βουλὴν ὅ<τι> νὰ ἀπελθοῦσιν, "there they made the decision to depart" (H 145)
  
• Κι ἀφότου ἐστερεώσασιν ὅτι νὰ τὸ πληρώσουν, "and after they affirmed that they would pay for it" (H 180)
  
• ὑπόσχεσιν τοῦ ἐποῖκαν / ὅτι νὰ βάλουσιν βουλήν, "they made him a promise that they would decide" (H 8546–7)
  
• (Jean de Catavas): Κι ὅποιος ἰδῇ ὅτι νὰ τραπῶ ἢ τίποτε δειλιάσω, / ἐχτρὸν τὸν ἔχω τοῦ Χριστοῦ, νὰ μὴ μὲ σφάξῃ εὐθέως, "and whoever sees that I turn or lose courage at all, I shall hold him to be an enemy of Christ if he does not kill me immediately" (H 4755–6)
  
• Τὸ ἀκούσει ὁ Ροῦσος ντὲ Σουλῆ ὅτι νὰ γένῃ ὅρκος, "when Rousseau de Sully heard that there would be an oath" (H 7927)

The first three phrases can be expressed in Modern Greek as either ότι θα, a future proposition, or να, an event (commitment): you can decide that you will do something, or decide to do something. If it's an event, ότι looks redundant: "they made the decision that to depart" (Modern πήραν απόφαση να αποχωρήσουν). That kind of redundancy can happen in Early Modern Greek, but I'm sympathetic to reading it instead as the old future: "they made the decision that they will depart" (Modern πήραν απόφαση ότι θα αποχωρήσουν).

The last two instances are verbs of perception, just as we saw with Manuel Chantakites hearing that his father was seeking rent. De Catavas' clause involves direct perception: Modern Greek allows both να and ότι, and would prefer να normally; but the absence of raising ("sees me turn", μὲ ἰδῇ (ὅτι) νὰ τραπῶ) suggests that the clause is really being introduced by ότι, and να is a future marker. The future sense is clearer in Rousseau de Sully's clause, which involves indirect perception: de Sully can't hear people making an oath, because he is agreeing for them all to take the oath that has just been proposed. So he really has heard "that there will be an oath", and that's what the Greek says: ὅτι νὰ γένῃ ὅρκος, Modern ότι θα γίνει όρκος.

Now the point of all that was, it is difficult to work out whether these clauses express events or propositions: the future να turns up with propositions, and makes them sound like events. The sequence ότι να sounds like two complementisers next to each other, and it would be easy to drop the first as redundant, or as a zero-complentiser. (Είπε πως θα παει "He said that he would go", Είπε θα πάει "He said he would go", which in Early Modern Greek would be Είπε να υπάγει.)

If it's difficult for us, it was difficult back then too. So it would be possible for the future (ότι) να in those contexts to be reanalysed as a propositional complementiser: Τὸ ἀκούσει ὁ Ροῦσος ντὲ Σουλῆ ὅτι νὰ γένῃ ὅρκος, "when Rousseau de Sully heard that there would be an oath" could lead to ωσάν ήκουσα να γυρεύγεις της νύφης σου νοίκια, "when I heard that you are asking for rent from your daughter-in-law". The reanalysis may not have generalised, and it certainly hasn't left any traces behind—especially given how Cretan has since got rid of the propositional να in Πρέπει πως διαβάζει, "He must be studying".

But it suggests that the distribution of να as a complementiser was messier, messed up by the use of να as a future marker—which pushed it towards more "indicative" meanings. The neater modern distribution of να, along with the neater distribution of past tenses after inferential να, may have have resulted from a subsequent cleanup of the language—speakers reimposing order on their use of να, rather than inheriting it.

Thank you if you've made it this far; this is properly a paper, but I'm not likely to write it, and it may already be a solved problem. I'm pretty sure it hadn't been worked through in the literature I read for my thesis. But given the spasmodic status of Early Modern Greek linguistics until fairly recently, that's not saying much.