A Greek Papyrus in Armenian Script.

Danny Bate writes for Greek City Times about a papyrus dated to between the 5th and 7th centuries:

The text is like a phrasebook – it’s key vocab and phrases for Armenians living in Hellenic Egypt! Although it’s one of the earliest sources for the Armenian alphabet (created c. 400 AD), the text doesn’t contain one word of Armenian. Instead, it’s lines and lines of Greek. There’s everyday words (like parts of the body), helpful phrases and even conjugations of common verbs! […] The text is so important for our knowledge of Greek. By using Armenian letters, its author didn’t have to follow the norms and archaisms of Greek writing – they were free to spell more accurately. The window into the Greek of that time and place that it gives us is incredible. For example, the consonant /h/ is consistently spelled (as in “hipar” ‘pony’), while the use of the Armenian letter Բ shows that /b/ hadn’t shifted to /v/ in this Greek.

The papyrus doesn’t have any specific name beyond simply ‘the Armeno-Greek papyrus’. It’s held in the collection of the National Library of France (BnF 332). For more information, Clackson 2000 (additional notes 2002) is the leading paper on it, and where I got the picture.

Clackson 2000 is A Greek Papyrus in Armenian Script (Zeitschrift für Papyrologie und Epigraphik 129 [2000]: 223–258):

This article concerns a papyrus containing Greek in Armenian script which is housed in the Bibliothèque Nationale in Paris (inventory number BnF Arm 332). Portions of the text and photographs have already been published, but I present here the first full edition and commentary. My edition differs substantially from previous readings of the text which did not recognise that the text, as mounted, was misaligned.

Fascinating stuff; thanks, Trevor!

Comments

  1. And although she may have studied with an expert Alexandrian philhellenian,
    I can tell that she was born Armenian!

  2. Er, that was in response to a comment that the poster apparently deleted.

  3. David Marjanović says

    [b] and even [h] intact in the 5th to 7th century! I would not have guessed.

  4. J.W. Brewer says

    I am fascinated by the explanation in Clackson 2000 that the MS was first discussed in scholarly literature in the 1890’s but then by the 1930’s someone who wanted to write about it couldn’t actually lay hands on it and it was not actually located again as a physical artifact (slumbering in the collection of the Bibliotheque Nationale) until 1993. So the MS missed the entire Cold War, and I can’t look to the disparate views of collaborationist Armenian SSR hack scholars versus intemperately-extreme anti-Communist emigre scholars to know what I should think of it?

  5. David Marjanović says

    Maybe not [b]. β is rendered b a few times, but αβα is rendered awa both times. The letter v is never used, though. Maybe we’re looking at the otherwise hypothetical [β] stage.

    However, δ comes out as t at least twice, both times between two vowels! It’s d otherwise, though. Endnote 18 of the paper says there are also cases of d for τ and one of k for γ, which “shows the confusion of voiced and voiceless consonants, widespread in Egyptian Greek as evidenced by the papyri (Gignac I 79).”

    Apparently the letter ê hadn’t been invented yet; it’s never used. η is rendered e or i at random. …or not; endnote 10 implies e is regular before nasals.

    Likewise, ô isn’t used, and ω is rendered o most of the time, but ow, i.e. /u/, four times.

    ει is consistently i, αι is consistently e.

    υ is written iw, i.e. /y/, except for two that are rendered i and at least two that are rendered ow.

    οι is variously written iw, e or ew. Apparently the merger of /y/ and the short-lived /ø/ hadn’t happened yet! This is further supported by ὀνύχιον being rendered as ewniwkʰ with, well, umlaut. (Similar assimilations occur with other vowels.)

    Numerous a/e confusions, also a few between other vowels.

    ψ, ξ are consistently pʰs, kʰs, as in Greek loanwords in Armenian.

    x and (perhaps less surprisingly) f are never used; χ, φ are consistently kʰ, pʰ. θ is . “Note that in Armenian loans from Greek the letter β is not transcribed by Armenian v before the 8th century, nor is Greek χ transcribed by Armenian x in loans before the 10th century (Thumb 1900: 412f).”

    h occurs, but is apparently almost always omitted from the definite articles and sometimes in other unstressed syllables.

    The Greek is all Egyptian, not Pontic (geographically closest to Armenia).

  6. the confusion of voiced and voiceless consonants, widespread in Egyptian Greek as evidenced by the papyri

    Presumably under the substratum influence of Coptic, which doesn’t distinguish voicing in general.

  7. David Eddyshaw says

    x and (perhaps less surprisingly) f are never used; χ, φ are consistently kʰ, pʰ. θ is tʰ

    (Sahidic) Coptic*, too, uses Greek phi, chi, theta as ligatures** for /ph/ /kh/ /th/ (and has a quite separate, Demotic-derived, letter for /f/.) The Coptic uses for the Greek aspirate letters are in general hard to square with other evidence from that period. I wonder if Egyptian Greek was particularly conservative phonologically? Or even pronounced with a strong, distinctively Egyptian, accent, maybe …

    *Bohairic uses them for aspirates, opposing them to the (unaspirated) glottalised /p/ /t/ /k/, which are not distinguished from the unglottalised stops in Sahidic orthography (though the actual distinction itself was presumably pan-Coptic.)

    ** That this was not just some mere graphic convention, moreover, is shown by the fact that Greek thalassa “sea” was reanalysed as t-halassa, with the Coptic feminine sg definite article t, and a corresponding indefinite uhalassa etc

  8. What is expected time of the change of aspirates to fricatives?
    I looked in Wikipedia, Koine_Greek_phonology, it has

    These changes seem widely attested from the 2nd century BC in Egyptian Greek, and in the early 2nd century AD in learned Attic inscriptions; it is therefore likely that they were already common in the 2nd century BC and generalized no later than the 2nd century AD.

    Interpretation is more complex when different dating is found for similar phonetic changes in Egyptian papyri and learned Attic inscriptions. A first explanation would be dialectal differences (influence of foreign phonological systems through non-native speakers); changes would then have happened in Egyptian Greek before they were generalized in Attic.

  9. I must admit I looked at this from the perspective of Arabic (or Semitic) borrowings: ك vs. ق vs. خ vs. ح…
    But I thought that the change is later with regional variotion…

  10. David Eddyshaw says

    As I say, difficult to square with other evidence …

    Origen’s Hexapla transcriptions of Hebrew give rise to similar problems: indeed, the convention is quite similar, with the Greek aspirates being used for the non-glottalised /p/ /t/ /k/ and Greek pi, tau and kappa being used for the glottalised stops.* At least there, it’s possible to argue that Origen (or his sources) were using obsolete conventions rather than simply trying to match sound for sound; the consonant transcriptions are pretty much the same as in the Septuagint.

    Arabic usually renders Greek tau, kappa with pharyngealised stops. There is of course no pharyngeal p to represent pi with, but Ethiopic uses its (very rare) glottalised p there.

    * In particular, they are not distributed in accordance with the bgadkpat spirantization of the unglottalised stops. Quite a lot of sources seem confused on this point, but it’s quite obvious when you look at the data.

  11. That this was not just some mere graphic convention, moreover, is shown by the fact that Greek thalassa “sea” was reanalysed as t-halassa, with the Coptic feminine sg definite article t, and a corresponding indefinite uhalassa etc

    OK, that’s very cool.

  12. Yes, when I wrote /k q x ḥ/ I meant specifically /k q/.

    /x ḥ/ are mentioned because they حappen (unlike حurricanes in حempshire, حereford and حertford).
    (as I was googling the line, I found a baseball team Hartford Hurricanes:/).

    (And yes, uhalassa is Beuatiful absolutely!)

  13. Why do we assume the author was making accurate phonetic transcriptions of contemporary Greek speech? Given the author’s knowledge of Greek and literacy it seems likely the author was literate in Greek as well. He may have created equivalences that reflect conservative “correct” pronunciations he had learned “in school” rather than the way the average Alexandrian was actually speaking.

  14. David Marjanović says

    The oldest FILIPPVS is said to be mid-2nd century; on a wall in Pompeii there’s a PILIPPHVS – obviously an aspiration-free Italian couldn’t remember where the aspiration went and decided to put it on the already louder, because longer, of the two options.

    Quite a lot of sources seem confused on this point, but it’s quite obvious when you look at the data.

    Yeah, a minute on Wikipedia is enough. …except I can’t find the Wikipedia article that had a sentence or two of examples. It’s not “Hexapla”, and it’s not “Secunda (Hexapla)” either.

  15. David Marjanović says

    Given the author’s knowledge of Greek and literacy it seems likely the author was literate in Greek as well.

    No, why?

    It’s a ridiculous idea in today’s culture to learn to speak a written language without also learning to write it. But Antiquity was different. There’s also, for example, a Roman-age papyrus from Egypt that teaches you to speak Latin, but not to write it – it’s entirely in Greek letters. (But it goes well beyond the “tourist phrasebook” level.)

    He may have created equivalences that reflect conservative “correct” pronunciations he had learned “in school”

    That would actually be remarkable enough. Few conservative peevers are going to be able to pronounce [h] when they and their teachers didn’t grow up with it for centuries. That takes Vedic-level obsession, and outside of that kind of religion that’s just not sustainable (without recordings). So if such things as [h] were obsolete features that weren’t in common use anymore, they must have died out just slightly earlier, within living memory or not much beyond that.

    Also, the pronunciations aren’t as conservative as they theoretically could be. Most obviously, αι is consistently transcribed as e, and οι is not rendered oi a single time. This is not Pericles speaking. And that’s before we get to the Coptic-influenced [d~t] and [g~k] confusions, or the complete lack of indication of where the stress goes (Armenian has fixed last-syllable stress), let alone which pitch accent it is.

  16. Really cool post. The thing with [h] is that it rarely gets directly written even in well spelled papyri. Even in the Classical Period after the adoption of the Ionic Alphabet it is routinely omitted in inscriptions in some places, although we can be sure it was generally pronounced; so one might well wonder whether instances of its omission in later documents do in fact occasionally belie a somewhat more vigorous presence in the spoken language. Obviously the evidence still clearly points to its loss as a phoneme, but a text such as this (and I would agree that there is no hint of any formal Greek written education on the part of the author) could suggest that in some pronunciations or registers the sound [h] did persist at least sub-phonemically beyond the period we would usually allow for.

  17. Also, slightly off topic, but hopefully not too outrageously: I’m currently plagued by this question I can’t find the answer to anywhere; are the venerable ranks of Hatters able to enlighten me, please?

    I’m currently thinking about whether Ancient Greek /s/, generally assumed to be a ‘normal’ [s], could have been retracted. Modern Greek /s/ is retracted; is there any evidence for when this change occurred (if it really did occur)?

  18. Andrew Dunbar says

    Did Armenian phonology not also change during these fifteen centuries? If it did wouldn’t this affect which Greek sounds the Armenian letters represent? If it didn’t isn’t that remarkable in its own right?

  19. I’m currently thinking about whether Ancient Greek /s/, generally assumed to be a ‘normal’ [s], could have been retracted. Modern Greek /s/ is retracted; is there any evidence for when this change occurred (if it really did occur)?

    An excellent question; now you’ve got me curious.

  20. I would have supposed that Greek /s/ became retracted to contrast better with theta after it became a fricative. Cf. Castilian.

  21. Roberto Batisti says

    It seems likely that the retracted pronunciation of Greek /s/ goes back to Ancient Greek, and, in fact, perhaps all the way back to PIE, as argued in 2010 by Aurelijus Vijūnas.

  22. Very interesting. An excerpt:

    When the sibilant systems of the modern Indo-European languages are compared with the corresponding system in the protolanguage, the following observation can be quickly made: in the modern languages that possess a postalveolar s, in most cases it is their only sibilant, and this corresponds well to the early Proto-Indo-European phonological system, which contained only one sibilant phoneme, too. On the contrary, the sibilant systems of languages with a dentialveolar or alveolar s are more complex. The languages from the latter group normally also have the prepalatal sibilant [ʃ], which is of dialectal origin, and can reflect a series of different things, depending on the language. Purely on structural grounds, from this distribution one could draw a preliminary conclusion that the languages with a single sibilant could have preserved a more archaic state, and, accordingly, would be more likely to reflect the original pronunciation of the Proto-Indo-European *s.

    Essentially the same claim was made by A. Martinet nearly half a century ago in his Économie des changements phonétiques (1955: 236f.). Having studied the articulation and the systems of sibilants in a number of Indo-European languages, Martinet concluded that for simple economical considerations, languages with a single sibilant would have it retracted, whereas the languages that have the more familiar variety of the hissing s and the shibilant š would show some kind of innovation. The reason for this innovation would have been the urge for “maximal differentiation”, and it would have pushed the articulation of the two sounds apart (Martinet, loc. cit.).

    Although Martinet’s claim was both fascinatingly simple and based on actual evidence, it has been largely neglected by Indo-Europeanists, and is nowhere mentioned in standard Indo-European handbooks. Likewise, nothing is mentioned in handbooks about all other studies of medieval sibilants, produced both before and after the appearance of Martinet’s Économie. In the following paragraphs, I would like to discuss these studies in a brief way.

    Makes sense to me (he goes into all the evidence thoroughly in the rest of the paper).

  23. David Eddyshaw says

    Martinet concluded that for simple economical considerations, languages with a single sibilant would have it retracted

    It’s surely by no means a general phenomenon. Kusaal has only /s/ /z/, both of which vary a lot in realisation but are only retracted before /i/ and /u/, and not always there. /s/ is often dental, and non-initial /s/ is often realised as [h] (/h/ is marginal as a phoneme.)

    I remember being very struck by the retraction of Modern Greek /s/ when I first heard the language. It’s very evident acoustically.

    http://www.beyond-the-pale.uk/cavafy.mp3

  24. David Marjanović says

    Oh, I didn’t click “post” on this comment a few hours ago…

    ===========

    did persist at least sub-phonemically

    If it persists in specific words, it contrasts with zero and is therefore phonemic. That’s what this papyrus indicates.

    I’m currently thinking about whether Ancient Greek /s/, generally assumed to be a ‘normal’ [s], could have been retracted.

    I bet it was retracted all the way back to Proto-Indo-European. That’s the default for languages that don’t contrast /s/ with some sort of /ʃ/, and found today e.g. in Finnish or Dutch.

    Before /tʃ/ became /ʃ/ in French, the French /s/ was retracted as well. This is seen in English words like push < pousser, and apparently in dialects that merged /s/ into the new /ʃ/ instead of moving them apart.

    Basque, Old Spanish and Middle High German have/had three-way contrasts of “normal” /s/, retracted /s/ and /ʃ/. In Spanish, the retracted /s/ is inherited from the Latin /s/. In MHG, it’s inherited from the Proto-Germanic */s/. Basque is different in that it had a /s/-/ʃ/ contrast before it developed the retracted /s/; it came from the “normal” one in certain environments plus from Romance loans.

    ============

    …and I didn’t know about the papers by Vijūnas or Martinet.

    Meanwhile:

    It’s surely by no means a general phenomenon. Kusaal

    Spanish outside Spain (excepting, roughly, Argentina where there’s a new /ʃ/) and in part of Andalusia is in the same situation. I wonder if sub- and adstrates like Nahuatl and Quechua are to blame; note the spelling Cuzco, where the non-retracted [s] is made explicit.

    and non-initial /s/ is often realised as [h]

    I wonder if that’s made easier by retraction. Though that is not likely to blame in Lower Bavarian (where initial /s/ merges into /h/).

    I remember being very struck by the retraction of Modern Greek /s/ when I first heard the language. It’s very evident acoustically.

    Yes, and next to /u/ some people produce nothing less than a full retroflex. I heard someone talking on the phone once and was torn between “this is obviously Greek” and “this obviously can’t be Greek” for a few minutes…

    I’ve also heard sueño with a strongly rounded retroflex in (a radio quote from) the Spanish parliament. RUKI is apparently not that hard to achieve.

  25. I heard someone talking on the phone once and was torn between “this is obviously Greek” and “this obviously can’t be Greek” for a few minutes…

    I’m sure I’ve told this story before, but the first time I heard Modern Greek was in a diner in New Haven, sitting with my roommate Paul Cardile (whose family was from Lucca); we spent an embarrassing amount of time arguing over whether it was a weird dialect of Spanish (me) or a weird dialect of Italian (him) before asking the guy behind the counter and discovering the mortifying truth.

  26. David Eddyshaw says

    Western Oti-Volta apart from Boulba has undergone the changes *c -> s and *ɟ -> z (a merger, as /s/ /z/ already existed.) This is probably relatively recent, and may even be areal, given that there are (probably) no other common innovations setting off all of the non-Boulba languages from Boulba, away over in Benin. [Gabriel Manessy also cites some stray non-Boulba WOV forms with /c/ /ɟ/, but I can’t verify all his sources, and the ones I can check are erroneous.]

    So conceivably, most-of-WOV might be explainable-away on the grounds that its one-sibilant-position status is recent (as Vijūnas tries to do with other exceptions.) [Dagbani has developed secondary new s/ʃ and z/ʒ contrasts, but I haven’t seen any detailed analyses of its phonetics.]

    I’m not sure if Proto-Oti-Volta actually had palatal fricatives to go with the stops, though. I haven’t found enough comparanda to be sure, so far.

  27. David Marjanović says

    the changes *c -> s and *ɟ -> z

    These can hardly have happened in a single step – unless they were borrowed (i.e. spread areally to WOV from elsewhere), in which case phonetic plausibility largely goes out the window anyway.

    So if this went [c] > [tʃ] > [ʃ], we’d have a stage with a /s/-/ʃ/ contrast before /ʃ/ merged into /s/, and likewise for [ɟ] > [dʒ] > [ʒ] > [z].

  28. David Eddyshaw says

    True.

    (In Boulba, /c/ remains unchanged, but *ɟ has become /c/, in line with the Atakora-Sprachbund unreasoning hostility to voiced stops and fricatives; the Eastern Oti-Volta languages got rid of *ɟ by shifting it to /j/ instead.)

  29. “We ton’t fafor foicet stops ‘rount here, putty…”

  30. To the extent that /s/-debuccalization is more likely when the sibilant is retracted or laminal, which seems like a reasonable thing to believe (cf. modern Spanish), that would argue for such a pronunciation in Proto-Greek.

  31. In fact I came across Vijūnas’ paper last year, which is what got me wondering and reading around the related literature in the first place. It’s very compelling stuff, but as he himself observes, the retracted quality of a single sibilant in a given language is still best thought of as a tendency rather than a rule as such. I was thinking rather along the lines of more solid evidence, such as transcriptions of Greek words in inscriptions and graffiti in languages such as Coptic, Arabic, or Indic, but haven’t really found any good papers on this. For example, one might hope for proof of a Greek retracted /s/ in occasional confusions with Coptic /ʃ/, but apparently this doesn’t happen; then again, it’s surely not automatically the case that speakers of one language will confuse their /ʃ/ with the retacted /s/ of another, as has certainly happened in borrowings between European languages. There are some purely local developments of rhotacism in Ancient Greek, which must strongly suggest a retracted /s/ in at least these areas. Interestingly, and surprisingly back on topic, Classical Armenian had /ʃ/ (according to its wikipedia entry), and yet there is no trace of it in this papyrus.

  32. Roberto Batisti says

    Getting back on topic: the Greek-Armenian papyrus is discussed in a brand new article by Ron Kim, in the context of the Old Armenian spellings ⟨v⟩, ⟨w⟩ and ⟨ow⟩.

    Unfortunately, the paper is paywalled, but I am making my way through the print edition.

    In the relevant paragraph, Kim concludes that since β is regularly rendered with բ, it “appears that the sound represented by ⟨v⟩ was already labiodental [v] in Armenian by this time. As a result, the composer of the Greek-Armenian papyrus identified the bilabial [β] of Egyptian Greek with Armenian /b/ except in the environment a_a, where it was perceived as closer to the glide or approximant /w/.”

  33. the changes *c -> s and *ɟ -> z

    In this respect, WOV perfectly matches Sylheti, where chacha ji “respected uncle” becomes sasa zi. More evidence for the great Volta-Ganges sprachbund.

  34. @DM: I probably should have written ‘…had marginal phonemic status’. But naturally there must have been a period before it dropped altogether when most speakers were using it in a fairly haphazard fashion, and its use was no longer contrastive. Even in this papyrus there is an example of the same root with and without initial [h] in ‘hodeuomen’ vs ‘odon’.

  35. By the way, for Arabic and Latin readers (and also lovers of early medieval correspondence):

    A papyrus in Arabic and Latin in Latin letters.
    photo
    the text (MS Word).

  36. David Marjanović says

    Interestingly, and surprisingly back on topic, Classical Armenian had /ʃ/ (according to its wikipedia entry), and yet there is no trace of it in this papyrus.

    Maybe Egyptian Greek had actually fronted its /s/ by equating it with the Egyptian /s/ as opposed to the Egyptian /ʃ/…

    In the relevant paragraph, Kim concludes

    TOTAL EXONERATION!

    But naturally there must have been a period before it dropped altogether when most speakers were using it in a fairly haphazard fashion

    I don’t think that’s natural. I rather imagine a period when [h] only surfaces for emphasis – followed by a stage where this use spreads to words that never had a /h/, but that’s not on this papyrus.

    Even in this papyrus there is an example of the same root with and without initial [h] in ‘hodeuomen’ vs ‘odon’.

    Could this be a sandhi phenomenon? What’s the preceding letter in these cases?

  37. David Marjanović says

    the text (MS Word).

    Oh, fascinating.

    The writing is upright and shows a slow ductus – no ligatures are to be found, except for three cases in which a ligature ti occurs, with the shape of a ‘reversed Beta’ (l. 6, gratiia; 12, bitiiti; 13, atiimtuạ; verso, l. 1, aṭiualla). It probably marks an assibilated sound /ts/.

    Footnote:

    “About this ligature, see ANTONIO CARTELLI, MARCO PALMA, L’evoluzione del legamento “ti” nella scrittura protobeneventana (secoli VIII-IX), in Pierre Lardet (ed.), La tradition vive. Mélanges d’histoire des textes en l’honneur de Louis Holtz, Turnhout, Brepols, 2003, pp. 35-42.”

    Also:

    If orthographical peculiarities such as co- for quo- and iscrib- for scrib- (or inscrib-) are quite common among early medieval Latin written evidence, the form binne mi (venit mihi, line 5) and tu si sanum (tu sanus es, line 6) are very close to the Romance languages, and recall the best-known earliest Romance texts like the Italian inscription of Commodilla’s catacomb, the “Veronese riddle”, the “Placiti Cassinesi” and many others. […] From the typological point of view, a quite similar (but distinct) context can be found in Greek witnesses’ subscription of Italian papyri. In these documents the notary dictates some Latin formulae to a person who can use only Greek writing, usually a merchant or a banker coming from the East: in transcribing what he hears, this person reproduces deviations that probably come from the spoken level, like ed espontaneo (et spontaneo), ed eis relicto es (et eis relectum est), and veienti cattur (viginti quattuor).

    Context:

    On the other hand the presence of Latin around Jerusalem becomes less puzzling if one looks at literary sources of the early 9th century: for instance, the commemoratorium de casis, a sort of account of church and monasteries around Jerusalem, composed for Charlemagne shortly after 807 AD, reports that in the Mount of Olives there were three churches and several monks “qui sedent per cellulas, eorum qui graeca lingua psallent XI, Georgiani IIII, Syriani VI, Armeni II, Latini V, qui Sarracenica lingua psallit I”. The exceptional combination of languages and script of Papyrus BL 3124, as long as no other similar exemplar comes to light, might therefore be explained in connection with some fortuitous circumstances like the passage, or the presence, of a western monk or a merchant in an Arabic speaking region.

    Every /k/ and /q/ is spelled c, no matter if i follows; e is used, o is not. The Arabic is already postclassical in some ways.

  38. @DM: It’s ‘pws hodeuomen’ and ‘me (ie ‘moi’) odon’. But there are also examples of final s before a dropped h and h after a final vowel. My ‘haphazard fashion’ was of course just British vagueness for ‘likely a mix of sandhi phenomena interspersed with sociolectic and analogical variables that I’m currently too lazy to try to read up on’. We do actually have ‘haurion’ (tomorrow) in this papyrus, where the form with h is not etymological -but possibly quite well established as it crops up elsewhere; there are other examples of words denoting times or seasons picking up h, such as ‘(h)etos’ (year) (thus ‘analogical’, above).

    “Maybe Egyptian Greek had actually fronted its /s/ by equating it with the Egyptian /s/ as opposed to the Egyptian /ʃ/…” -That’s an interesting idea, and it could nicely explain the apparent lack of sigma-shai confusion in Greek words transcribed into Coptic; although the seeming lack of confusion elsewhere (e.g., arabic) between these sounds (if it really is the case; I’m not sufficiently informed on this) is still potentially troublesome.

  39. David Marjanović says

    The Arabic change of [ɬ] to [ʃ] may have been late enough to prevent confusion there; Sibawayh seems to have described a likely intermediate stage, [ç].

Speak Your Mind

*