New Phrygian metrics and the formula
ALEXANDER LUBOTSKY1. The pioneers of the Phrygian studies (Sayce, Ramsay, Solmsen, Calder) often tried to find traces of hexameters in New Phrygian (NPhr.) inscriptions.1 Since it later appeared that the NPhr. malediction formulae do not normally conform to the hexameter rhythm, the search for metrical patterns has gradually come to an end. At least, in the publications of the last fifty-sixty years I was unable to find a single reference to the metrical organization of the NPhr. texts.
Nevertheless, I believe that NPhr. malediction formulae were often metrical. If we look at the usual protasis of these formulae, we see that it has a dactylic rhythm: ()() ()/ `whoever brings harm to this grave...' (–∪∪ –∪∪ –∪∪ –∪∪). Dactylic rhythm is also often found in the apodosis: or (∪∪ –∪∪ –∪∪ –∪) `let him be cursed by Zeus'. In the present article I shall analyze some NPhr. inscriptions from the metrical point of view and discuss the consequences for the interpretation of the formula.
2. Before we turn to the inscriptions, let us first look at some features of the NPhr. phonological system that are metrically relevant.
2.1. Brixhe (1990: 98) has argued that NPhr. had no opposition between long and short vowels. He assumes a system of five vowels2:
1As an example, I can cite the following words of W.M. Ramsay (1905: 85): “It is suggested to me by Professor
Sayce that the commonest formula was originally two hexameters:
It is certain that numerous traces of metrical arrangement are seen in the inscriptions (as Solmsen remarked).”
2The numbers of the NPhr. inscriptions refer to: Haas 1966: 114-129 (Nos. 1-110), Brixhe 1978a: 3-7 (No.
111-114), Brixhe – Waelkens 1981 (No. 115), Brixhe – Neumann 1985 (No. 116), Laminger-Pascher 1984: 35 (No. 117), Mitchell 1993: 186, fig. 33 (No. 118), Brixhe – Drew-Bear forthcoming (No. 119-125). I am very grateful to Cl. Brixhe and Th. Drew-Bear for putting the manuscript of their last-mentioned article at my disposal.
I use the brackets as follows:
In the phonological analysis: {} = grapheme , /o/ = phoneme o, [a] = sound a;
/i/ {}: ,
{}/{} before vowels: (), ()3 /u/4 {}: ,
{} (rare): 62, 25, 15, 120
{}/{}/{}/{} before final nasals: , , 62, /e/ {}: ,
{} (rarely) in open syllables: 25, 25, 1035, /o/ {}: ,
{}/{}/{}/{} before final nasals: , , 62, 14 /a/6 {}: , ,
Brixhe interprets {} as a variant spelling of /o/ (p. 96). As the omega is consistently used in the dat.pl. ending - < *-is, e.g. `among men and gods', it is safer to assume a separate phoneme /o2/ (= //?). This ending is spelled with {} three times only ( 6,7, 39)7.
Another moot point concerns the monophthongization of the diphthong ei. Brixhe (1990: 77ff.) assumes that this diphthong had already developed into in Old Phrygian (OPhr.), which further yielded NPhr. , but this scenario seems implausible to me. In final position, we have the following attestations of the dat.sg. of the consonant stems:
-: (± 40×), (1×) -: (18×), (1×), (16×), ()8 (1×) -: (7×) -: (1×), (6×), (1×) -Ø: (13-14×)
3This word, which refers to the monument or its part, is not inflected (but cf. 120, which is a more
probable analysis than , given by the editors [Brixhe and Drew-Bear, forthc.]), which may point to a borrowing. The variants with and without initial - are reminiscent of Gr. `hard, hard metal, stone', MLat.
diamas `diamond'. It is conceivable that / means `stone' and is related to these words.
4From *u (-) and * ( < *H
1eitd). As far as the etymology of - `grave' is concerned,
Meister 1909: 317, fn. 2 assumed a connection with Gr. , whereas Brixhe – Neumann 1985: 172 suggested a `croisement entre le phrygien keneman et le grec ?'. Since the meaning of - is the grave proper (cf.
Lubotsky 1993: 129f), both of these etymologies are unconvincing. I would propose a connection with Gr. `to scratch', cf. for the semantic development Germanic *greb- `to scratch, dig' (e.g. OE grafan `to dig, grave, engrave,
carve, chisel' and grf `cave, grave, trench').
5[] 5 is most probably due to engraver's mistake for (Brixhe, pers. comm.). 6From long and short a and from * ( < *dheH
1-k-).
7I do not think that -, a variant of the dat.pl. ending, can be explained from *-oisi (pace Brixhe 1990: 96). I
would rather assume a contamination of - and *-. The ending is moreover rather uncertain: 75 is most probably a mistake for T, so that the only clear example of this ending is 92 [] [...].
8Written , cf. Lubotsky, forthc.
Brixhe (1990: 79) concludes that `-EI est la graphie historique, dont la persistance a ete favorisee par l'equivalence de EI et de I dans le systeme graphique utilise; -I represente une gra-phie phonetique; -E reflete probablement la neutralisation de l'opposition /e – i/ en finale...' It is possible that - is a `historical' spelling, but then one of a recent history: monophthongization in OPhr. is out of the question, cf. OPhr. dat.sg. materey, eveteksetey 9. Moreover, 3sg. impv. `may he become' is consistently spelled with -, while acc.sg. part. occurs 10x with -- and only once (58) with --. On the other hand, always appears with --.10
These facts receive a plausible interpretation if we assume that the development ei > {//} was operative at the latest stages of NPhr. It first occurred after , so that there is not a single instance of **. As I have indicated elsewhere (Lubotsky forthc.: fn. 23), 11 occurs only before consonants, whereas is limited to the position before vowels. We may conclude that NPhr. still had a diphthong /ei/, which in final position was prone to monophthongization.
The consistent spelling of other NPhr. diphthongs indicates that they, too, were not monophthongized. In addition to /ei/, we may assume the following four diphthongs:
/ai/ {}: , , , {}:
/oi/ {}: , , , /au/ {}: , /eu/ {}: ,
In final position, the diphthong (dat.sg. of the -stems) tended to lose its second element, cf. (), (), etc. It is unclear whether NPhr. had the diphthong /ou/, which at any rate would be undistinguishable in writing from /u/.
Accordingly, NPhr. had at least six (or seven, depending on the status of {}, see fn. 11) vowels and five diphthongs. Brixhe' assumption that NPhr. vowels had no length contrast seems possible, but is unverifiable.
2.2. As far as the consonants are concerned, only two aspects require short comments. (1) Geminates are probably simplified in NPhr. (cf. Brixhe 1978a: 14), which accounts for the variants /, /, /, /,
9This is a decisive argument against the derivation of OPhr. devos from *deiuo-, defended by Brixhe 1990: 76f. 10There is only one exception, viz. (19), found in a peculiar inscription, the engraver of which has left
out the middle part ( /[] <...> []).
11The phonemic status of {} is unclear. It stands for prevocalic /ei/ in (31), (116), cf. also
(58). In (98) nom.pl. `parents', - most probably reflects *-eies, the generalized nom.pl. of the i-stems, as in Latin. Also (42, 87, 118, 120) is likely to be nom.pl., but (86) is certainly a dative (sg. or pl.). In , -- stands for -- (cf. ). Cf. further on Lubotsky 1989: 82f.
etc. on the one hand, and unetymological geminates in (44, 53), etc., on the other. Forms like , with geminates represent historical spellings. We therefore must reckon with the possibility that the geminates could have been used as a metrical license in order to create long vowels by position.
(2) The phonetic value of NPhr. is unclear. Greek of this period had already become [z]. However, - in < *dhghemelis seems to be a product of palatalization (since we
assume that Phrygian is a centum language) and may stand for [dz] or some other double consonant. Spellings like (39, 113) may then be interpreted as an effort to write an affricate, although for occurs in Greek inscriptions, too (Brixhe 1978a: 7).
3. We have seen above that the NPhr. malediction formulae often have dactylic rhythm. If the NPhr. vowel system had no length contrast, a classical hexameter based on alternation of long and short vowels is not what we expect. The function of long vowels can only be taken by vowels which are long per positionem and by diphthongs. Therefore, in the absence of long vowels, the best approximation of the hexameter was a dactylic verse consisting of five feet plus –× at the end.
The usual protasis of the NPhr. malediction formulae, viz. () () ()/ `whoever brings harm to this grave...' consists of 4 dactyls (–∪∪ –∪∪ –∪∪ –∪∪).12 In order to make a full hexameter, we need yet another foot followed by –×. And indeed, the protasis is often enlarged with an addition like (18, 26), (86) `...or to the monument (?)', ' () (14), (87), () (112), ' (120) `...or to the stone (?)', ʮ (4) `...or to its sepulchral chamber', (73) `...or to this stele (?)', [] (92) `...or to the place', etc. None of these additions produce a correct line, but it seems plausible that the authors of these inscriptions at least intended to produce something like a reasonable hexameter. The best attempt was () (112), ' (120), but it seems premature to assume that this was the “original” text.
4. The majority of the apodosis formulae end in () `let him become accursed', which has the metrical structure of the end of a hexameter line (∪∪ –∪∪ –×). What then was the beginning? A reasonable guess is that the line started with a formula, which is found more than twenty times (for the meaning see the next section). The following variants of this formula are attested:
(3, 6, 97, 113, 119) (21, 103), (124)
12We have to assume that the geminates --, -- were restored due to a metrical license (cf. 2.2).
[ ] or ]()()(5)13 (96) (112) [] [][] 14 (42) [] [...] (92) [ ] (7), [] (39), (118) (40) (63, 93), [] (4), []() (121) [ ] ' (25) (75) (62)
Which variant is the opening of the metrical apodosis? It seems certain that the normal Phrygian expression was or () . First of all, the only asyndetic expression is . Secondly, the order – contradicts the Behaghel's Law (“das Gesetz der wachsenden Glieder”). Thirdly, “gods” being more important are likely to be mentioned before “men”, so that we expect the order `gods and men' and not `men and gods' (cf. Lat. dis hominibusque, di hominesque). It thus appears that the unexpected order – must have been due to metrical reasons, whereas the order – was the current expression, which explains why it is found in so many inscriptions. These observations are further confirmed by the fact that the commonest variant, viz. (with minimal spelling differences), conforms to the dactylic rhythm (–∪∪–∪∪–∪)15 and is thus likely to be the opening of the metrical line (as already suggested by Sayce, see fn. 1).
In order to make the second “hexameter” line complete, we need two syllables in the middle. These two syllables are most probably / `by Zeus'16, the formula () `let him be accursed by Zeus' occurring thirteen times (plus 4 times () ). We may thus conclude that the second line of the original “hexameter” was 17 () . This apodosis is indeed attested in two inscriptions, viz.:
13For the final - see Ramsay 1905: 79ff, Brixhe 1978b: 1, fn.2, Lubotsky 1989: 81.
14The group may mean something like `jointly, altogether' (cf. mhi huwedri `assembled gods', which
are often invoked in the Lycian curses). We may analyze it as the preposition plus dat.sg. (Gr. ?). Also (120) may refer to `the whole family, the family altogether'.
15Assuming that - represents a double consonant ([dz], [zz], [zd]?) (cf. 2.2).
16For this Phrygian god and the analysis of the apodosis formulae see Lubotsky 1989 (esp. 84f).
17Note that the dative ending (< *-ei) counts here as long, which indicates that the metrical line was composed when
it still was a diphthong or, at least, a long vowel.
6. T [] 97. <> T [ ]
The same apodosis may also be present in inscription 3, the second line of which is damaged. According to Ramsay's copy (1887: 387), the traces of the second line seem to correspond to 19 [][ ]20 T. The inscription thus reads: []
[][ ] T
.
Inscription 113 is broken exactly at the most interesting place. The editor of the inscription (Haspels 1971 I: 321) gives the apodosis as [ ]?, but at the photo of the impression (II: 619), I cannot identify the last letter. It seems conceivable that the apodosis was T [ ].
Accordingly, among the five occurrences of the formula , three or four are followed by T . Only the apodosis of inscription 120 ( ) does not have the middle piece T.
5. What is the meaning of the curse T ? There are two current explanations of the formula in the literature. The one, given for the first time by Ramsay (1905: 107f.), interprets this expression as `(among) gods and men', whereas the other, which eventually goes back to Gustav Meyer (cf. Calder 1911: 207), takes the words to mean `by heavenly and chthonian gods'. A good discussion of these alternatives is given by Calder (1911: 206ff.). A priori, the `gods and men' interpretation is preferable. First of all, an asyndetic expression `heavenly and chthonian gods' is hardly feasible. Secondly, there are syntactical problems: one has to assume that the word for `gods' is missing, standing for `heavenly (scil. gods)' and for `chthonian (scil. gods)'. Thirdly, the Hesych gloss Φ is easier to account for starting from a noun meaning `man' than from an adjective `chthonian'. Finally, corresponds with Gr. < *dhH
1sis
18I wonder if the correct reading is rather , considering the fact that the endings of the two
members are generally identical. Ramsay was the only one who saw and copied this inscription. In 1887 he wrote (p. 389): “The letters of this text are so faint and worn, and the stone was in such a bad position -- it lies with its face towards the ground, and we dug a hole below it to enable me to see the letters obliquely -- while I was copying it, that in places where I did not know the formula, I could only guess at the reading”. In a later publication (1905: 107), however, he says that the reading is certain, but an engraver's error is probable.
19Or , cf. 53. [], 62. . The segmentation is unclear. Brixhe (1979b: 180) opts
for () /, but in view of inscription 3, where we again find a geminate, I would prefer () /.
20Ramsay (1905: 106) mentions that this part contains 12 letters.
and not with Skt. devais < *deiuis, as the monophthongization of *ei does not occur in the middle of the word (see above).
However, there are two weighty arguments against the `gods and men' interpretation, which were put forward by Calder (o.c.). First, in the Greek inscriptions of Asia Minor, the violator of the tomb is often devoted to , but never to `gods and men', “and, if this were the meaning of the commonest Phrygian formula, it would be certain to occur on Greek epitaphs of Phrygia” (p. 208). This argument was the main reason for Heubeck (1987: 74ff.) to opt for the meaning `heavenly and chthonian gods'.21 Nevertheless, as Haas (1966: 92ff.) has indicated, Greek inscriptions with `gods and men' do exist. He found one inscription from the island of Teos and one from Diokaisarea (L. Robert, Villes d'Asie Mineure, 21962, 330), which reads ... . Moreover, I think Haas is right in stressing that the Phrygian malediction formulae continue an indigenous Anatolian tradition, and it is only by chance that we may find the same expression in the Greek inscriptions. We shall return to this point below.
The other argument against `gods and men', mentioned by Calder, concerns the fact that we find constructions both with and without : “In the cases in which is preceded by , we could understand it to mean `(let him be accursed) when among gods and men', i.e. both when alive and when dead (for the dead were deified in Phrygia and were called ...). But this meaning does not suit the simple dative, for with the dative, corresponding to with the dative, must mean `devoted to'; and it was only to the gods that wrongdoers were devoted, not to men” (p. 207). First of all, the expression hardly means `when among men and gods', but rather `amidst men and gods'. Furthermore, Calder's argument presupposes identical syntactical constructions in Phrygian and Greek formulae, which is an open question. We cannot even be sure that () really means `devoted to' and not, for instance, `condemned'22, which can be construed with a dative `by gods and men'.
6. In an article dedicated to the problem of NPhr. (Lubotsky 1989), I hesitatingly proposed to see in the curse a mixture of two formulae, viz. () and (p. 84). The main reason for this analysis was my assumption that means `by mortals and gods'. I therefore faced the problem that the sentence seemed to contain three
21His other consideration, viz. that in 5 and 6 is dat.sg. and refers to the chthonian deity M, is highly
questionable for both palaeographic (see above, fn. 18) and morphological (dat.sg. of the o-stems ends in -)
reasons.
22If Phrygian had Lautverschiebung, which, in my opinion, is a serious possibility (e.g. NPhr. T = Gr. ), we
may even consider etymological connection of () with Greek - `judge', - `to condemn', PIE *deik-.
co-ordinated members in the dative and only two times . Now we know that is likely to represent the original apodosis, and it becomes clear that I was wrong. My starting point must be adjusted: does not mean `by mortals and gods', but `among mortals and gods' as already indicated by Ramsay (1905: 107). The Phrygian preposition and preverb is etymologically related to Gr. , which in the construction with a dative means `among, between'. It follows that must be interpreted `let him be condemned / accursed by Zeus among men and gods'.
7. Recently, when reading Hawkins' article about Luvian inscriptions where the Luvian goddess Kubaba – the prototype of Phrygian Kybele – is mentioned (Hawkins 1981: 162), I was struck by a passage from the inscription Karkami A 3 (line 4):
wa/i-sa- | DEUS-na-za | CAPUT-ta-za-ha | zi-na-na | (DEUS)TONITRUS-ta-ti-i | (LOQUI)ta-tara/i-ia-mi-sa i-zi-ia-ru
“(and) let him be made accursed by Tarhunzas before gods and men!”
The Phrygian and Luvian curses are practically identical (even Tarhunzas = T), and in spite of the fact that these expressions are separated by a time span of more than a thousand years, we can hardly avoid the conclusion that they continue the same tradition. The Luvian formula provides an important argument in favor of the analysis of the Phrygian curse suggested here and adds a new aspect to the study of Phrygo-Anatolian relations.23
References Brixhe, C. 1978a: Etudes neo-phrygiennes I. Verbum 1,1, 3-21.
Brixhe, C. 1978b: Etudes neo-phrygiennes II. Verbum 1,2, 1-22.
Brixhe, C. 1979a: Etudes neo-phrygiennes III. Verbum 2,2, 177-192.
Brixhe, C. 1990: Comparaison et langues faiblement documentees: l'example du phrygien et de ses voyelles longues. La reconstruction des laryngales (Bibliotheque de la Faculte de Philosophie et Lettres de l'Universite de Liege, fascicule CCLIII). Liege-Paris, 59-99.
Brixhe, C. 1994: Le phrygien. Langues indo-europeennes, ed. Fr. Bader. Paris, 165-178.
23There can be little doubt Phrygian had an Anatolian substratum. The influence of the Anatolian languages can be
seen, for instance, in Phrygian onomastics, cf. Mamutas, , (Brixhe 1994: 175f.). Also many of the
terms for parts of the grave and the monument are likely to be borrowed from an Anatolian source. For instance, dat.sg. (73. ---) clearly refers to a (part of the) monument and may represent a borrowing from an Anatolian language (cf. Lydian mruwaa-/murwaa- `stele', maybe
also Lyd. mola-, muwe~nda-, muwe~si-, Hieroglyphic Luvian malwa- `votive (stele)'), as already proposed by
Haas (1966: 80).
Brixhe, Cl. – Drew-Bear, Th. forthcoming: Huit inscriptions neo-phrygiennes. Proceedings of the Phrygian conference in Rome. [[= Frigi e Frigio. Atti del 1° Simposio Internazionale. Roma, 16-17 ottobre 1995. A cura di R. Gusmani, M. Salvini, P. Vannicelli. Roma (Consiglio Nazionale delle
Ricerche), 1997, 71-114.]]
Brixhe, Cl. – Neumann, G. 1985: Decouverte du plus long texte neo-phrygien: l'inscription de Gezler Ky. Kadmos 24, 161-184.
Brixhe, Cl. – Waelkens M. 1981: Un nouveau document neo-phrygien au musee d'Afyon. Kadmos 20,
66-75.
Calder, W.M. 1911: Corpus inscriptionum neo-phrygiarum. JHSt. 31, 161-215.
Haas, O. 1966: Die phrygischen Sprachdenkmler. Sofia.
Haspels, C.H.E. 1971: The Highlands of Phrygia. 2 vol. Princeton.
Hawkins, J.D. 1981: Kubaba at Karkami and elsewhere, Anatolian Studies 31, 147-176.
Heubeck, A. 1987: Phrygiaka I-III. KZ 100, 70-85.
Laminger-Pascher, G. 1984: Beitrge zu den griechischen Inschriften Lycaoniens, Osterreichische
Akademie der Wissenschaften, Phil.-hist. Klasse, Denkschriften, 173. Band (Ergnzungsbnde zu den Tituli Asiae Minoris, nr. 11). Wien.
Lubotsky, A. 1988: The Old Phrygian Areyastis-inscription. Kadmos 27, 9-26.
Lubotsky, A. 1989: New Phrygian and . Kadmos 28, 79-88.
Lubotsky, A. 1993: New Phrygian . Kadmos 32, 127-134.
Lubotsky, A. forthcoming: New Phrygian inscription No. 48: palaeographic and linguistic comments. [[=
Frigi e Frigio. Atti del 1° Simposio Internazionale. Roma, 16-17 ottobre 1995. A cura di R.
Gusmani, M. Salvini, P. Vannicelli. Roma (Consiglio Nazionale delle Ricerche), 1997, 115-130.]] Meister, R. 1909: Die olischen Demonstrativa , , und die Partikel () im Phrygischen. KZ
25, 312-325.
Mitchell, St. 1993: Anatolia. Land, Men and Gods in Asia Minor, I. The Celts in Anatolia and the Impact of Roman Rule. Oxford.
Ramsay, W.M. 1887: Phrygian inscriptions of the Roman period. KZ 28, 381-400.