I won't do that this post either, because instead I want to look at how not just the semantics, but the syntax of the expression changed. Like the semantics, I will claim that the syntax of this expression, like that of so many others, changed through an iteration of reanalysis and extension: hearers reinterpret an expression in an ambiguous context, assigning it a new structure—and then they extend that new structure to novel contexts, where the old interpretation could not be used. The reanalysis brings the new structure into being, but only the extension makes the structure visible, as an unambiguous addition to the language.
So it was with the semantics of the expression—through sentences where the Ottoman warcry was reinterpreted as an expression of bloodymindedness, as I will try to walk through. So it is was, at least for the most part, with the syntax. There is a change at the end where the extension may not have been out of an ambiguous context, but a more abstract jump—which is perfectly possible, but not what I prefer. But let's get there slowly.
A good deal of the change in the syntax of αμέτι μουχαμέτι had already happened in Ottoman Turkish. Originally, of course, ümmet-i Muhammed was just a noun phrase, and we see that meaning literally in (1, 5):
- Τον κύριο παρακάλεσαν να κάμη μερχαμέτι, Και να τους κάμη ολουνούς του Μουχαμέτ’ ουμμέτι "They begged the Lord to show compassion, and to make them all Muhammad's nation".
Ottomans addressed groups of Muslims as ümmet-i Muhammed, which makes the expression a vocative. The salient example of this, of course, is the warcry to Muslim troops. I claimed (2) was the first example of this, but the syntax is darker than that. But the warcry is clearly there in (4, 6, 8, 9, 13)
- ιμέτι, Μωχαμέτη, ’Σ τον ντιν ντουσμάνη σήμερα να κάμωμεν γαϊρέτι, "Nation of Muhammad! Today against the infidel enemy we shall show endurance."
Extension can't be usefully shown here, because this is still a noun phrase. The next function is too close to disentangle as well: as an exclamation. When troops are being addressed, as in (4), it is ambiguous whether ümmet-i Muhammed is an exclamation or a vocative. The extension happens when there are no troops being addressed, so the expression no longer makes sense as a term of address.
That extension is clearly what Ali Pasha used, when ümmet-i Muhammed is an exclamation of surprise or dismay, not an address to troops (3):
- Ορίστε, λεγ’ Αλήπασας, ιμέτι Μουχαμέτη, Χαΐρι δεν εχούμε ’μεις εφέτ’ απ’ το ντουβλέτι, "'See,' Ali Pasha says, 'By the Nation of Muhammad! We're not seeing any joy from the government this year!'"
The late exclamations of dismay attributed to Ottomans (14, 15) also fall under this class. We would also see the warcry as an exclamation instead of a vocative, if it were used in addressing someone other than troops, or in a way otherwise incompatible with vocatives.
- In an example like "Ümmet-i Muhammed! Today you will die, infidels!", ümmet-i Muhammed would not be used to address Muslims.
- Kolokotronis, by nominalising the cry, is treating it as something said, which is more consistent with an exclamation than an address (7): και με το αμέτ μουχαμέτ ώρμησαν κατά των ιδικών μας, "with an Amet Moukhamet they rushed onto our men".
- The song on Xopateras (11) treats Ümmet-i Muhammed! as the same kind of phrase as SubhanAllah "Glory to Allah!", and it is not followed by a command.
- The song on Tryfitsos (16) joins Ameti mou Khameti to the next sentence with "and"—which is inconsistent with a vocative: αμέτι μου χαμέτι και ο Τρυφίτσος είν κιοσές που μας-ε παίζει μπέτι, "Ameti mou khameti, and Tryfitsos is the one who is firing at our chests." That cannot be interpreted as "Oh Nation of Muhammad!—And Tryfitsos is firing at us."
The Greek expression, with its notion of obstinacy, clearly came from the warcry and not the cry of despair. What's critical for the next reanalysis is that the expression is an exclamation—which allows it to turn up in a broad range of contexts. In several instances of the warcry, the exclamation introduces a commitment to do something:
- (4) Nation of Muhammad! Today against the infidel enemy we shall show endurance.
- (6) Nation of Muhammad! Fortify the place boldly.
- (13) Nation of Muhammad! We will burn them down.
If Greeks no longer understood ümmet-i Muhammad, and distorted it to αμέτι μουχαμέτι, the exclamation could be reinterpreted as an adverbial phrase: "definitely, at any cost, come hell or high water"; so "at any cost, we will burn them down!" It would help such an interpretation along, that one of the phrases likely to have been conflated with ümmet-i Muhammad was μα το Μουχαμέτη "by Muhammad"—which also can be used to expressed commitment.
The adverbial interpretation helps when the expression turns up in indirect speech, where the vocative cannot, and even the exclamation is problematic. That indirect speech use seems to me to motivate (2) and (10)—(10) more clearly than (2):
- (2) τους Τούρκους όλους λάλει· όλοι τους – ουμέτι Μουαμέτη – ας δράμουν στην μητρόπολι "Tell all the Muslims: all of them—Nation of Muhammad—should run to the cathedral":
- Interpreted as apposition: "all of them, namely the Nation of Muhammad, should run": noun phrase
- Interpreted as indirect speech, quoting exclamation: "Tell all the Muslims: 'Nation of Muhammad! All of you should run to the Cathedral'" A vocative could not be so quoted: "Troops! All of you should run to the Cathedral!" cannot be rendered as "Tell the soldiers, all of them—??Troops!—should run to the Cathedral." So ümmet-i Muhammad no longer has vocative force.
- Interpreted as apposition: "all of them, namely the Nation of Muhammad, should run": noun phrase
- (10) Όλοι τους ωρκισθήκανε αμέτι Μουχαμέτη, στο Μεσολόγγι να εμβούν, να κάμουν κιαμέτι "They all swore, 'Nation of Muhammad!', to enter Messolonghi and cause havoc." It is impossible for 'Nation of Muhammad' here to be a vocative: this is quoting the Muslims' oath, "Nation of Muhammad! We will enter Messolonghi and cause havoc!", with 'Nation of Muhammad' treated as an exclamation.
But while (10) makes sense as quoting Muslims crying "Nation of Muhammad!", embedding an exclamation in indirect speech is a very odd thing to do. Any exclamations in indirect speech should be coming from the speaker, not those quoted. An exclamation like "onwards!", though, can be reanalysed as an adverbial phrase, "crying 'onwards'"—or "as if crying 'onwards'"—or, for that matter, "at any cost".
That interpretation is more comfortable with (10): "They all swore, 'Nation of Muhammad!', to enter Messolonghi and cause havoc" > "They all swore, crying 'Nation of Muhammad!'", or "They all swore, in a 'Nation of Muhammad!' way", or "They all swore, at any cost to enter Messolonghi."
We have now arrived at the Modern Greek expression, which is adverbial. We have also arrived at the meaning of the Modern Greek expression, which derides obstinacy: αμέτι μουχαμέτι means, ultimately, "in such a bloodyminded way, you'd think he was urging Muslims into battle". But for that intepretation to be possible, people had to be using "Nation of Muhammad!" in Greek in indirect speech, and no longer as a vocative.
By (12, 17, 19) we clearly have extension to a context where both the vocative and the exclamation are impossible, and indeed to a context where there can be no reference to Muslim troops at all. This can be straightforwardly read as adverbial use, in the modern sense:
- (12) αλλά ο Γαρδικιότης είναι Αμέτ Μουαμέτ κατά του στραβού, "but Gardikiotis is bloodymindedly against the blind man."
- (17) ο Αλικιάδης είχεν απόφασιν, “Αμέτ Μουαμέτ”, να βάλη τη δουλειά εμπρός, "Alikiadis had decided, bloodymindedly, to go ahead with the venture."
- (19) ήρθε αμέτι μουhαμέτι να μαλώση, "he came bloodymindedly intending to fight."
The expression did not stop as an adverb, however. Recall that αμέτι μουχαμέτι is used in the following ways in Modern Greek:
- as an adverb generally: τα 1.280.000 ευρώ πρέπει να γίνουν, αμέτι μουχαμέτι, ποδηλατόδρομος, "The 1.28 million euros must be spent come hell or high water for a bicycle path" (αμέτι: 11700 hits on Google)
- θέλει αμέτι μουχαμέτι accusative NOUN OR να VERB, "he wants, ameti moukhameti, NOUN/ to VERB" (θέλει/ήθελε αμέτι: 122 hits on Google)
- βάλθηκε αμέτι μουχαμέτι να... "he has set himself ameti moukhameti to..." (βάλθηκε αμέτι: 4 hits on Google)
- το έχει αμέτι μουχαμέτι να... "he has it ameti moukhameti to..." (έχει/είχε αμέτι: 11 hits on Google)
- το έχει βάλει αμέτι μουχαμέτι να... "he has set it ameti moukhameti to..." (έβαλε/'βαλε/έχει βάλει/είχε βάλει αμέτι: 3021 hits on Google)
Using αμέτι μουχαμέτι with θέλω "to want" can be explained with an adverbial meaning: "he wants, bloodymindedly, a raise"; Η Ντόρα ήθελε αμέτι μουχαμέτι να πάει για τις υπογραφές, "Dora wanted, bloodymindedly, to go get the signatures". And that use of θέλω can be traced back to quotations of the warcry: αμέτι Μουχαμέτι, θα μπούμε στο Μεσολόγγι "Nation of Muhammad! We shall enter Messolonghi" > ήθελαν «αμέτι Μουχαμέτι» να μπούνε στο Μεσολόγγι "they wanted 'Nation of Muhammad!' to enter Messolonghi" > "they wanted, bloodymindedly, to enter Messolonghi."
The same holds for βάλθηκε, "he set himself/herself": Υπουργός τότε ο πολλά βαρύς και όχι Βασίλης Κοντογιαννόπουλος […], ο οποίος βάλθηκε αμέτι μου χαμέτι μου να μεταρρυθμίσει την Παιδεία, "The minister at the time was the dour-as-black-coffee Vasilis Kontogiannopoulos […] who set himself, bloodymindedly, to reform Education" (blacksad)
But the main use with a verb is with and το έχει βάλει "he has set it"—with το έχει "he has it" (18) trailing far behind:
- (18) το είχεν αμέτ Μωαμέτ, να γίνη δήμαρχος, "he had it amet Moamet to become mayor."
- Ο Υπουργός Δημοσίων Έργων […] το έβαλε αμέτι μουχαμέτι να τον εκτρέψει, "The minister of Public Works has set it ameti moukhameti to derail him."
These instances are not adverbs: ??"he had it bloodymindedly to become mayor", ??"he set it bloodymindedly to derail him".
What these expressions do sound like though, is (το) είχε σκοπό να..., το 'βαλε σκοπό να... (έχει/είχε σκοπό: 1,021,000 hits on Google; έβαλε/'βαλε/έχει βάλει/είχε βάλει σκοπό: 135,600 hits on Google), "has/sets purpose to = intends to". In other words, αμέτι μουχαμέτι is behaving, in its usage after βάζω and έχω, like the noun σκοπός, "purpose"—which means that αμέτι μουχαμέτι has been reanalysed as a noun.
Other nouns have already fitted into this template: έχει/είχε καημό να "sorrow = yearning to" (4340 hits), έχει/είχε βάλει καημό να (2 hits); έχει/είχε άχτι να "spite to" (534 hits), έχει/είχε βάλει άχτι να (4 hits). There would be a simple reason of analogy for αμέτι μουχαμέτι to fit this noun template: it looks like a neuter noun. In fact, αμέτι μουχαμέτι > ümmet-i Muhammed looks like the neuter noun άχτι "spite" < ahd "oath, promise". And άχτι is primarily used with the verb έχω "have" and βγάζω "take out".
Unlike σκοπός, καημός, and άχτι, the expression αμέτι μουχαμέτι is mostly associated, so Google has it, with βάζω "set"—particularly with the perfect το έχει βάλει (1940 of the 3020 hits). One explanation for that is another analogy—with βάζω όρκο "to set an oath = to swear" (and not *έχω όρκο): ümmet-i Muhammed was understood to be an oath, explicitly so in (10). But that does not explain by itself why βάζω "set" was associated with an adverb "bloodymindedly": "he set it bloodymindedly" doesn't quite make sense.
(17) provides one ambiguous context which explains this reanalysis: ο Αλικιάδης είχεν απόφασιν, “Αμέτ Μουαμέτ”, να βάλη τη δουλειά εμπρός, "Alikiadis had a decision, bloodymindedly, to go ahead with the venture" (adverb) > "Alikiadis had a decision, an 'Amet Mouamet', to go ahead with the venture" (noun). In other words, αμέτι μουχαμέτι was reanalysed as a noun in apposition with "decision"—another word for "decision". The punctuation, if it is Papadiamantis', supports that—αμέτι μουχαμέτι was still more like an interjection than an adverb, and its separate intonation makes it look like apposition.
Maybe, but the reanalysis looks unconvincing to me: appositions are a bit too learnèd to be plausible here. The best I can come up with is the construction έβαλε να... "he set out to": έβαλε να φτιάξει καφέ, "he set out to make coffee". Combined with αμέτι μουχαμέτι, the construction becomes έβαλε, αμέτι μουχαμέτι, να φτιάξει καφέ, "he set out bloodymindedly to make coffee", which can be reanalysed as "he set a purpose to make coffee". Once that reanalysis happened, the clitic object could be added in, by analogy:
- έβαλε σκοπό να φτιάξει καφέ, "he set a purpose to make coffee" = "he intended to make coffee"
- το 'βαλε σκοπό να φτιάξει καφέ, "he set it a purpose to make coffee" = "he intended to make coffee"
- το 'βαλε όρκο να φτιάξει καφέ, "he set it an oath to make coffee" = "he swore he'd make cofee"
- έβαλε, αμέτι μουχαμέτι, να φτιάξει καφέ, "he set out, bloodymindedly, to make coffee" > έβαλε αμέτι μουχαμέτι να φτιάξει καφέ, "he set a bloodyminded intent to make coffee"
- το 'βαλε αμέτι μουχαμέτι να φτιάξει καφέ, "he set it a bloodyminded intent to make coffee"—by analogy with το 'βαλε σκοπό and το 'βαλε όρκο
We may not need to find an ambiguous sentence to explain το 'βαλε αμέτι μουχαμέτι. It may merely be an analogy straight from το 'βαλε όρκο "he set it an oath" and το 'βαλε σκοπό, with αμέτι μουχαμέτι sounding like a noun to replace όρκο and σκοπό, but as a conceptual leap rather than a shift latent in context.
At any rate, αμέτι μουχαμέτι is acting like the noun σκοπό "purpose" in the βάζω/έχω … να construction—but the noun analysis has not undergone further extension. αμέτι μουχαμέτι is not used in the contexts that σκοπός is:
- Ο σκοπός είναι να δούμε νέες ιδέες, "the intention is for us to see new ideas"; *το αμέτι μουχαμέτι είναι να δούμε νέες ιδέες
- Έχει καλό σκοπό, "he has good intentions"; *έχει καλό αμέτι μουχαμέτι
- Η θεωρία του σκοπού της ζωής, "the theory of the purpose of life"; *η θεωρία του αμετιού μουχαμετιού της ζωής
- Έρανος με σκοπό να αγοραστεί δορυφόρος, "a fundraiser with the intention of buying a satellite"; *έρανος με αμέτι μουχαμέτι να αγοραστεί δορυφόρος
Using αμέτι μουχαμέτι as the noun "purpose" in general sounds absurd for now. But in Turkish, "he set it Nation of Muhammad to make coffee" sounds absurd. (Try it out: Kahve yapmak için, ümmet-i Muhammed o yerleştirilir. It would be less absurd, if I actually knew any Turkish...) Syntactic change is possible, if people see a point in it; and we can't rule out future extension of the construction, along the path it has started.