The metathesis of iota is well known in greek, where it occurs especially in verbal continuous (present) themes constructed with the suffix –y(o)-. Thus, *φαν-j-ω, τεν-j-ω, κερ-j-ω became φαίνω [I show], τείνω [I stretch], κείρω [I cut]. The same type of metathesis occurs in Daco-Rumanian in some words, like pâine [bread], mâine [tomorrow], câine [dog] from pâñye < lat. panem, mâñye < lat. mane, câñye < lat. canem, or in the oltenian subdialect in pronounciations like veichi (lit. vechi) < vulg.lat. vetlu, oichi (lit.ochi) < vulg.lat. oclu etc.

I found this phonetic phenomenon also in a series of Thracian names. Working on Herodotus' testimonies, I observed that the name of the modern river Isker was, at his time (5th century BC), Ὄσκιος[1]. I thought at first it was a mistake, but contemporary Thucydides insured me it wasn't, for in his histories (Pelop.War 2.96.4) he wrote: μέχρι τοῦ Ὀσκίου ποταμοῦ, where the river's name is again Ὄσκιος (a form that we could transcribe phonetically as /oskyos/). But from the I century BC on, we only meet the forms gr. Οἶσκος/lat. Oescus (phon. /oyskos/) and their further phonetic and graphic variants: Ὕσκος, Ἴσκος, Escus, Iscus a.s.o, every one of them with its own history which I shall discuss somewhere else. We can therefore conclude that between the 5th BC and 1st AD in this name took place the metathesis of /y/ (iota). I first explained this phenomenon as a celtic influence, given that they invaded the ancient Balkan Peninsula in the same timespan (4th century BC). However, in certain cases, as you will notice, the metathesis must be much earlier.

The phenomenon I am trying to present could be reduced to the formula: CVCy (+vowel) > CVyC [where C=consonant, V=vowel, y=iota](type *φανjω > φαινω). Since the resulting context Vy means in fact a diphtong, we would look for Thracian names whose variants present the alternance vowel/vowel+y in their root: a/ay, e/ey, o/oy and u/uy. But before searching for these alternaces in the Thracian language remains, we must state a few methodological principles ment to help us avoid some traps of the Greek and Latin transcriptions of the Thracian names. There is no monographical work upon the way the Greeks and Romans used to transcribe barbarian names and words. In its absence researchers usually take into comparison the actual forms of Thracian names, and this is a great methodological mistake. Let me give an example: if we are interested in the etymology of a name such as Ἄθως (the mountain peak in the Pallene), we would normally search for etymons like *at(h)ō or *adhō or other variants. But if there was any work to draw our attention upon the fact that [θ] might reflect, at least in the Thracian area, not a dental of the type /th/ or /þ/, but merely a local sound close to /č/ or /ts/ or another inbetween (see the Thracian Palatal), we would rather prefer as etymon the IE root *ak' "sharp" most suited for an oronym. There are many such 'graphic conventions', as I called them, and they have their own phenomenology (for example: they changed in time, they were imposed by the authority of some known writers etc.). These conventions need to be monographically analized and presented in order to become an instrument for further studies. Since it is still missing and I cannot refer it, in the following I shall present those conventions which touch to our subject.

1-after the first century BC the diphtong [ει] becomes the usual graphic form for transcribing not only the actual diphtong /ey/, but merely a long /ī/. This is a true convention, since Latin authors constantly render this sound in the same names by /i/ and not by /ei/. As a result, in many cases we wouldn't know for sure where does [ει] transcribe a genuine diphtong /ey/ or a long [ī].

2.-the Greek diphtong [αι], as well as the Latin [ ae], are very often used purely graphically (as hypercorrect) instead of the bear vowel /ę/ (open). Here again, we have no mean to prove the existance of a real pronounciation /ay/ (except for some particular cases, of course).

3.-the diphtong [οι] /[oe] seems to be more stable, so we can rely more often on its phonetic value /oy/ in both languages.

4.-a special mention must be made concerning the Greek letter [υ]. We know that its original value was /u/, and that at a certain time it got palatalized, its old value being assumed by the former diphtong [ου] (something very similar to what happened in French). There is yet no positive arguments to identify this new palatalized value, which the adepts of the "Erasmic pronounciation" of old Greek spell as /ü/. In fact, analyzing the mutual transcriptions of Greek and Latin names, we find that this letter often stands for a diphtong which could be phonetically noted by /yu/, /iw/, /uy/ or /wi/ (I recall now the procopian toponyms Πάλυρος = lat. paliurus [namely palyurus] –where [υ] = /yu/ - and Κυντόδημος, Κυντῶν for Quintode(ci)mus and Quintorum – where [υ] = /wi/; in Appendix Probi we can find uir non uyr, where uyr must have been pronounced /wúyr/ a.s.o.). As a consequence, we might admit that in some names the alternance [ου]/[υ] or even [ο]/[υ] can reflect phonetic variants with /u/ (/o/) and /uy/.

 

Other examples of thracian metathesis.

The gloss κινούβοιλα is quoted by Dioskurides as a Dacian equivalent of βρυωνία λευκή, a plant whose scientific name is in botanics bryonia alba. In Romanian it is also called "cireaşa câinelui" (dog's cherry), where cireaşă (cherry) [and also măr (apple)] only indicates a round and small fruit, while the dog (or wolf) shows that the plant grows in the wild. It corresponds, following Georgiev, to the lit. šun-obuolas hound's apple, from an ie *kwōn-abōlo "id.". This seems to be the best explanation among those given to this name up to now (much less probable Decev 550 and Russu 98). Its correspondent is, at Pseudo-Apuleius, discopela (Dacian) and dinupula (Bessian), both probably corrupted to some extend but still containing the merely genuine element -opela/-upula. For this vocalic evolution (ie *a > dacian *o, at least in this context) also stands the name Ὀβουλήνσιοι by Ptolemy, undoubtly a Greek transcription of a Latin ethnicon *Obulenses, from a place name *Obula "the Apple(s)". The Dacian gloss is nevertheless an adjective and not a noun. The [ι] in the second component's root (-ουβοιλα) appears as a result of a metathesis in an adjectival form *ubolya > *uboyla "apple-like".

Pokorny: the Baltic shows clear traces of the very rare ie *ābōl, G. Sg. *ābeles, in the l- declention. The extended degree of the suffix appears in the words meaning "apple": ostlit. obuolỹs, lett. âbuolis, westlit. óbuolas, lett. âbuols (o-St.) from ie. *ābōl-; the normal degree occurs most in word meaning "apple tree": lit. obelìs (fem. i-St.), lett. âbels (i-St.), âbele (ē-St.) from ie. *ābel-; but apr. woble f. (*ābl-) "apple", wobalne (*ābolu-) f. "apple tree"

Οἰτήνσιοι, the name of a tribe in Moesia Inferior by Ptolemy must be, as other names quoted by the Geographer in this area, a Greek adaptation of a Latin Oetenses. We cannot know the original Moesic name of this tribe, but we can assume that it was also a plural whose root we can get by leaving the lat. suffix -enses apart: *Oyt-. Let us, just hypothetically, presume a form *Oytoi. We are interested in fact only in the root *Oyt- which can be very well explained from *uyto- < *utyo- < Utus = the name of the actual river Vid in Moesia Inferior. So *Utyoi (or Utyensi), become *Uytoi (= Oetensi) and ment "those by the Utus (river)". The alternance of aperture o/u is common for many languages, including TDM.

Ροιμος + variants (Roimus, Ruimus etc.) and Ῥοιμη- in compounds (Ῥοιμητάλκης, Ῥοιμηζευης, Ῥοιμηλητικη) constantly alternate with Ῥομ-: Ῥομεις, etc. Very interesting is the form Romaestas: it seems that in this case the suffix -y(a) was added to an already extended theme *romast- and so it was the second vowel of this new theme which developped the diphtong: *romast-ya- > *romayst-.

Coisstobocensis and Κοιστοβώκοι. The common transliteration of the name of this tribe is Κοστοβώκοι, both in litterary and epigraphical sources. There are yet two of its most trustful attestations which show the diphthong [oi] in the first syllable. Ptolemy, whose information comes most of the times from direct sources of latin and dacian character, gives us the form Κοιστοβώκοι in two different places (in 3.5.9 and 3.8.3), so that we could reject any copying error. The second, an inscription from Rome, raised by the nephews of Pieporus, the very king of this nation, says Ziai Tiaiti fil. Dacae uxori Piepori regis Coisstobocensis. This last form, which obviously shows the care of the dedicants to render as exactly as possible the genuine spelling, must be considered as the best and taken as reference. Relying on it, we may transliterate this name as *Koyšto-bōk-(ai). I shall discuss somewhere else why did I transcribe [ss] by [š] and what could be the etymology of this name.

Αἵμος/Αἵμον, the name of the Haemus mountain, the modern Balkan (turkish), bulg. Stara Planina. In my opinion it comes from an earlier, but also indo-european, linguistic layer, characterized by another phonetic evolution, a consonantic mutation *k > h (or *k- > h- only in intial position?). To the same language belonged the name Hebros/Ἕβρος, which came from Kebros through the same phonetic change of /k/ to /h/. Haemon would come from a theme *kaymon, perfectly explainable from i-e. *kamon "stone" (cf. sl. *kamen- and litt. akmuõ "stone"). Its derivative form, *kamyon, becomes *kaymon (>Haymon) through the same metathesis.

Moesia/Μυσία could also be explained as a metaphonic form. There are here two issues to discuss. The first is the nature of the diphtong expressed by [υ] in greek, but by [oe] in latin. It transcribed, in both scripts, a native /úy/ (*múysia). The second issue is the pronounciation of the group -si+vowel-, which merely transcribes a consonant like /š/, for two reasons: 1. we don't find the form *Moesa/*Μυσα expected after the depalatalization of /s/ (*Musya would become *Muysa), which indicates another function of the [i] 2. there are other variants of the name which support such an interpretation: Mussaticus and Musiaticus etc. So, the epichoric pronounciation of the name must have been *muša (rendered by the homeric Μυσία, when [υ] was pronounced /u/) and then, with the metathesis of iota, *muyš(i)a (in classical greek Μυσία and in latin Moesia, Musia), where the preservation of iota indicates that the consonant remained palatal.

Aizisis < *azyi- < * az' "goat". Ie. has *aig-, *ag, * kag- and others yet. This is a thracian-greek isogloss.

Καρπούδαιμον. Following Ptolemy, it was a town in the regions of the two –dama: Uscudama and Sadama. This is why it can be very well explained through the evolution *Karpudama (possibly the same as Tarpodizus) > *Karpudamyon > Karpudaymon, by metathesis.

Ἀνάδραιμος quoted by St.Byz. 90,12: Ἀμφίπολις, πόλις Μακεδονίας κατὰ Θρᾴκην ἥτις Ἐννέα ὁδοὶ ἐκαλεῖτο. κεκλῆσθαι δὲ Μυρίκην και Ἠϊόνα, Ἀμφίπολιν δὲ διὰ τὴν περίρροιαν τοῦ Στρύμονος, ὡς Θουκυδίδης τετάρτῃ. ἐκαλεῖτο … καὶ Ἀνάδραιμος. With some reserve ("wenn"), Decev (p.17) admits that the name is Thracian and that it could be the original translated by Ἐννέα ὁδοὶ "Nine roads". It is hard to decide wether it is so indeed or it would come from gr. ἀνά + Drama, the name of the town nearby. In any case, it is a compound from ana+draym(os) where the second formant is a derivative from dram-yo-s. 

 

Other ideas:

Κοιτυς and even Κοτυς < Kotyos

-keila (if it is phonetical indeed and not a transcription of -kīla) < *kel(l)-ya < attested -cel(l)a "source, spring"

Κοῖλα, Cuila < Colja (there exist forms with Col-, Cul-)

-cf. also Mihailov La Langue..., p.55

-plant name διέλλεινα is also an adjectival derivative, from a *déllenya.

[I shall add here every example I would find in my researches, in order to write an article on this phenomenon]

 

 

 

 

 

 

Notes

 

[1] In 4, 49 occurs the form Σκίος [καὶ ὄρεος Ῥοδόπης Σκίος ποταμὸς μέσον σχίζων τὸν Αἷμον] which stands, undoubtly, for Ὄσκιος as a result of a well known copist mistake: the initial [O] was taken as an article and then separated from the rest of the name.