The Sabellic accusative plural endingsand the outcome of the Indo-European sibilants in Italic

Размещено на http://www.allbest.ru/

The Sabellic accusative plural endingsand the outcome of the Indo-European sibilants in Italic

Blanca Maria Prosper

This work aims to clarify the phonetics and phonology of sibilants and sibilant clusters in the Italic languages, and will specifically attend to the outcomes of /ns/ and /rs/ in different positions. The structure and meaning of a number of Sabellic words and sentences will be reanalysed and reinterpreted, with a special focus on Oscan and one of its dialects, Marrucinian. An appendix containing a novel interpretation of the new «Opic» inscription of Niumsis Ta- nunis is included.

Keywords: Indo-European linguistics; Sabellic languages; Ancient Italy; general phonetics.

Б. М. Проспер.Окончания множественного числа аккузатива в сабельских языках и развитие индоевропейских сибилянтов в италийской ветви

Цель настоящей работы -- прояснить особенности фонетического и фонологического развития сибилянтов и кластеров, содержащих сибилянты, в италийских языках, с особым упором на рефлексы сочетаний /пб/ и /гб/ в различных позициях. В статье предлагается новая интерпретация для целого ряда сабельских слов и предложений (в первую очередь относящихся к оскскому языку и одному из его диалектов -- маррукинскому). В приложении дается новое прочтение «опийской» («до-самнитской») надписи Ниум- сиса Тануниса.

Ключевые слова: индоевропейское языкознание; сабельские языки; древняя Италия; общая фонетика.

Introduction The Spanish government has (again) explicitly refused to fund this work.

phonetics phonology italian oscan marrucin

The consonant system of the Italic branch of Indo-European is still subject to debate. To begin with, we have to rely, as usual, on written materials. These consist of comparatively few and often fragmentary or barely readable remains, some of which show defective spellings. Many words are attested only once, and our paradigms are often incomplete, so that new documents, however short, often come as a surprise and scramble pieces in our puzzle that we had thought were reasonably well fitted. In addition, there is no consensus as to which languages may be classified as «Italic» in the genetic, not geographical sense. In this, as in former works, I shall assume that Latin, Venetic, Oscan, Umbrian and South-Picene go back to a single ancestor which can be labeled as Proto-«Italic», and whose origins are somewhere in central or Western Europe. When or how this unity was broken is on the whole unknown, but I further take it that there have been at least two waves of advance that brought speakers of Italic into the Peninsula that gave them its name. One of them, which I dare call Latino-Venetic, may have been the earliest people of Indo-European ancestry to traverse the Alps and leave us unequivocal traces of their language and culture, albeit the existence of other substrate populations has been occasionally claimed. Over time, the impact of other Italic populations, which may be globally termed «Sabellic», severed the former continuum into two halves and settled down in Central Italy, where it underwent further dialectalisation See on this depiction of the events Bonfante (1988).. Thereupon, an indeterminate number of mostly prosodic and phonetic innovations took place that affected variable areas of the Italic territory, in this way blurring the original genetic relatedness of the extant dialects beyond recognition.

This work aims to recover some of the common traits of Italic and its subgroups from a different perspective than usual. As we are going to see, both the mechanic reconstruction of a Proto-Sabellic stage and the tacit assumption that secondary, contact-induced phenomena -which changed the appearance of dialects even after thay have acquired a personality as individual entities- are scarce or dubious, may have given us a strongly biased vision of the relative chronology of Italic sound shifts. What is more, sound changes have been traditionally taken for granted exclusively on the grounds of the apparent cognacy of forms containing them. Interestingly, however, they have never been justified in purely phonetic terms, albeit they are often based on a single example. General linguistics, including contact linguistics, general phonetics and what we know about the typology of sound change, will accordingly play an important role in my arguments. As a result, I shall not only take issue with the widely accepted views on the evolution of consonant clusters containing sibilant sounds and the resulting relative chronology, but also reexamine the way the different writing systems were used, the attribution of specific forms to a paradigm or word class, the plausibility of their etymologies and external connections, and the interpretation of the syntactic structure of the texts in which they occur.

1. The Oscan accusative plural endings and the form iaf Unless otherwise indicated, all Sabellic texts follow ST. All texts in the Latin alphabet are in small capitals, and texts in the national alphabets are in italics.

Two identical Oscan forms, namely Marrucinian IAFC (MV 1, Rapino, Chieti, c. 250 BC) and south-Oscan i af (Lu 62, Roccagloriosa, Salerno, probably 3^rd C. BC) are universally held to continue the acc. pl. fem. of the anaphoric pronoun *eia(n)s. The alternative previously defended by Wallace (1985: 100, fn. 16) considered this and other forms as suggestive of the mixed ancestry of Marrucinian, which purportedly displayed Umbrian phonetic traits. In fact, this has always been exclusively predicated on the existence of U. eaf `eas' (Um 1 Ib 42).

When the Roccagloriosa inscription was uncovered, the existence of the Oscan form became suddenly problematic not to say contradictory, since the finding places of the Rapino and Roccagloriosa inscriptions (respectively a lex sacra and a legal text concerning theft), are separated by more than 300 km. The assumption of a Proto-Sabellic, that is to say, not specifically Umbrian generalisation of -f as the acc. pl. ending has since become, if often not professedly, essential to salvage both the attribution of Marrucinian to the Oscan dialectal constellation and the interpretation of iaf and iafc as full matches of U. eaf. To my knowledge nobody has put forward a theory that questions their accepted etymology or their cognacy with the Umbrian pronoun.

The uniform, but in principle unexpected spelling i- is apparently a minor inconvenience for the established etymology *eia(n)s, but it is nonetheless surprising that this pronoun is also transmitted in Roccagloriosa B, l. 9, in the nom.-acc. pl. neuter as Јiok (cf. the formally similar fem. sg. iuk in Cm 1, Abella). In fact, it stands to reason that iaf has been written in this way because it begins by i-. While it is true that in the Roccagloriosa document we apparently find (few) cases of <i> alternating with <Јi> and <Ј> to render /e/ from /e:/ and /i/ when flanked by consonants See McDonald et al. (2012: fn. 2). (in fact all of them contain the pronoun nig, m5), when <i> precedes a vowel it stands for etymological i-: ioufЈxou5 `iubeto' (B8), faKxiЈ5 (uninterpreted) (B9) and novSioup `of weights' (B11). Conversely, there is no case in which i- is spelt in any other way. What is more, e- in hiatus, going back to *ei- with Proto-Italic loss of intervocalic -i-, is never spelt <i> in southern Oscan, witness pЈia{ia}va[a (Lu 39, Anxium) and Јiok above. The same can be said of Rapino, where iafc, ioves and iovia are in all likelihood rendering i-. There is consequently every reason to call into question the idea that Oscan iafc/iaf is bisyllabic and begins by /e/.

Before we proceed to account for Oscan iafc/iaf, some words are in order regarding their alleged etymology. In one of his most influential works, Rix (1986) offered two different explanations of the evolution leading to the attested forms of the accusative plural masculine and feminine in Sabellic, which deserve some comment in view of their universal success.

According to his first scenario, all but the consonant stems show a uniform Proto-Sabellic acc. pl. ending -s with compensatory lengthening of the preceding vowel. Accordingly, Rix starts from the following Ursabellisch endings: 1^st-дs,2^nd-os, -i-stems -Is, consonant stems e. g. *-r-ns (>-r(e)f). Thereupon, Umbrian generalised the ending -f of the consonant stems by a simple proportion. Oscan, in contrast, apparently conflated both allomorphs, -s and -f, into *-fs. This proceeded through two stages: first, -s was added to -f in the consonant stems (allegedly under the influence of the dat.-abl. pl. -fs, which is most incredible since it would have ipso facto driven the inflection into unnatural syncretism). Second, -fs was generalised to all the stems, ousting -s. Finally, -fs became -ss by regressive assimilation. Why this uneconomical crossing should have taken place escapes me. I believe it to be nothing but an ex post facto explanation, with the particular circumstance that there is no trace of -f left in Oscan, and that the posited -fs would give very marked codas in the case of the consonant stems usurs and aginss (ex hypothesi presupposing an unattested stage *-rfs and *-nfs). The evolution in the feminine form must then have been *-дs >>*-дfs > -ass, attested in viass, etc. The other Oscan pronominal ending -af we are discussing remains unexplained.

Let us now take a look at the second scenario, according to which the Proto-Sabellic outcome -f is regular in more inflectional types. Rix starts from the following Proto-Italic endings: 1^st*-дs, 2^nd*-ons>-of, -i-stems *-ins >-if, consonant stems *-r-m (>-r(e)f). Thereupon, -f was generalised in Umbrian, whereas Oscan, again, followed a different path: It is fundamentally the 1^st declension ending *-дs (and presumably the diphthongal stem *g^uouns>*bos) that triggered the spread of -s to the rest of the forms, giving rise to 2^nd *-ofs, consonant stems *-r(e)fs, and then, in a completely unnecessary turn of the screw, was contaminated by these and corrected into *-дfs, eventually yielding -ass. It should be noted that Rix believes compensatory lengthening to have taken place at the stage in which the cluster -ns still existed, which thereupon evolved into -f. I find this incredible, in spite of the phonetic account that he offers in passing some lines before (1986: 586): «Primдres und sekundдres -nsergaben dann schon ursabellisch-f, wohl ьber -np>-nf (oder eher, was hier nicht zu diskutieren ist, ьber -nz -nц -nf > -&)», which tiptoes around the problems of the nasal-to-fricative transition I shall review below, as well as the unlikely voicing of fricatives after nasals, and offers no parallels for the proposed changes.

Again, the pronominal ending -af was disregarded by Rix, probably under the tacit assumption of Umbrian influence or appurtenance, which as we have seen is no longer acceptable. Firstly, there is no explanation for the fact that the pronoun shows a different ending from the noun. Secondly, the phonetic side cannot be right: if one gives any credit to Rix's first scenario, the pronominal ending -af would have come into being by way of substitution (in analogy to the consonant stems) followed by preservation of -f in the pronoun as opposed to addition of -s to the noun stem, which is not the tenor of Rix's account and makes no sense. Following his line of reasoning, by contrast, it is impossible to come to terms with our pronoun, since the ending -s has been pervasively replaced by -fs. This, in turn, means that *eдfshas unaccountably lost its -s instead of evolving into teдss. Things fare no better for Rix's second scenario, which demands that pronominal -дsfirst contaminated the rest of the paradigms, then was changed into -дfs,only to lose its final distinctive -s in this particular pronoun only. As a consequence, Rix's strenuous achievement of a uniform acc. pl. ending is distorted by an unexplained, non- phonetic turn of events which arbitrarily introduced an otherwise unattested plural ending⁵. This account is generally and rather uncritically accepted as far as I know (e.g. WOU: 356, Tikkanen 2011: 33), and not refuted, at least in its general lines, by Clackson (2013: 28-29), who simply remarks that, in view of

IAFC/iaf, the process must have been more complex than envisaged by Rix. One does not easily see, however, how it could be still more entangled without calling to question the basics of historical morphology.

Finally, the acc. pl. fem. ekass, ekask `these' (Po 1, 14, Pompei) discredits the idea that iafc and iaf contain an acc. pl. ending beyond any shadow of doubt. There is a sequence fitias estas amgenas in South-Picene (Sp AP 3; Iltal. I: 183, Falerio 1) which could be a phrase in the acc. pl. fem. including a pronoun estas and thus revealing another Sabellic case of pronominal -as, but this cannot be proven beyond doubt (Dupraz 2012: 38).

As a consequence, the very idea that the Oscan accusative plural feminine ever contained -f is a matter of personal belief. It is based on a number of requirements, none of which is unassailable: a) the input for the whole process must be *-ns > -f This change goes first in the diachronic sequence of shifts leading to the attested forms, and consequently necessitates a primitive, unitary output; this is placed in a canonical Proto-Sabellic stage in Rix's rigidly genealogical conception of language relatedness; b) the secondary shift *-nts >-nss>-ns> -f is shared by all the Sabellic languages and therefore must also be situated as early as Proto-Sabellic; c) all this happened before final vowel syncope, and in this way the nagging obstacle of a potential phonetic merger of -ns with postsyncope -n(V)s can be circumvented.

As implied above, Rix's ideas on the history of the Oscan accusative plural endings are in his view borne out by the abl. pl. *-fos>-fs> -ss, but the comparison is off the mark. To begin with, there is no single case of an acc. pl. preserving the cluster -fs which, as we have seen, is an essential intermediate stage in both variants of his reconstruction. Next in Rix's argument comes a subtle phonetic fallacy: while the sequence -ss in the abl. pl. is obviously the product of assimilation, it is not directly indicative thereof, since in all likelihood it is synchronically marking a tense sibilant of whatever origin in word final position. To account for the fact that *-ns# gave rise to a tense sibilant [s:] in Oscan and [f] in Umbrian and South-Picene, we have to take a fresh look at the phonetic reasons behind the divergent evolutions. But first of all, I shall try to make sense of iaf.

2. Oscan iaf revisited

What the forms iafc and i af have in common is the fact that they precede a verb in the 3^rd person singular, and that this verb is in both cases in the future tense, respectively feret⁶ and KAonouax (see below 3. for a more precise description of the syntax of these texts). The only Oscan form that could conceivably end up in -af and, additionally, could agree with such a verb form in person and number, is an active present participle bearing the suffix -nt-, a category thus far attested in Oscan kunsif (Iltal. II: 1183, Pentri/Terventum 22), the unclear statif (Sa 1, Agnone), and possibly in the disputed form staef `standing' (if <*sta-e-nt-s), as per Rix (1986: 596), WOU: 697, but recently read as ta<v>ffud (Iltal. II: 895, Capua 22). The outcome -f is shared by Umbrian, as in zeref `sedens' (Um 1 Ia 25, 33, 34), restef `re-stans, restoring?' (Um 1 Ib 9), kutef `in silence (Um 1 Ia,b, passim). I consequently trace iaf back to Proto-Italic *iant-s, and further identify this form with L. iens, Ven. ia.n.t..s. (personal name, Vicenza, LV: 124), Skt. yant-, Gk. lu>v, possibly Hitt. iiant-`sheep', ultimately from IE *hii-ent-s, *hii-nt-.

Athematic root participles are a dying category in Italic, and Latin preserves only a few, like the lexicalised forms cliens (see Fortson 2017: 842), as well as probably parens (EDLIL: 445446). A number of them is preserved in the onomastic corpus of Venetic, for instance the per- If it goes back to an athematic form *feresti like U. ferest (Um 1 IIa 26) with weakening of /s/ in coda position. Alternatively it may be a 3^rd sg. present indicative.

sonalname in the dative La.n.te.i. (Padua). In their edition of this text, Marinetti et al. (1994: 189-190) mention similar names, like Vants and lants, but offer no account of their make-up beyond the purely descriptive information that they contain «Ca + -nt-». In fact, La.n.te.i.may continue a root participle *ulant-, from *ulH-nt- `ruling' A sequence #ulV- was simplified early in Italic, as transpires from Ven. leno (LV: 12, 14, 25, Este).. Any attempt to come to terms with the unexpected phonotactics of this and similar forms containing *-l/rH-n- has proved futile in the past: the expected outcome of *ulH-nt- is *ual-ant-. As in the case of *hH-nt- `through', reflected in L. trans, U. traf `through' and OIr. tra `so, indeed', we probably have to reckon with early laryngeal resyllabification into *ulH-nt-, *trHnt-, etc. Sergio Neri (Munich) has kindly made the point to me that he would reckon with the inversion of Linde- man's Law in this case, by which an unstressed epenthetic vowel *CRHV->*CdRV- would be syncopated in a tri-syllabic form, giving *CRV-: e. g.dat. *ulh₂nt-йi>*udlnt-йi>*ulnt-йi. Cf. Neri (2019: 50). In addition, La.n.te.i.cannot possibly be a Celtic form, or it would have been adopted in Venetic as fflant-. Finally, Vants, Vanta, etc. (personal names, LV: 9, 53, 78, Este) come from *g^u(e)h.2-nt- `going', attested in Gk. pag. Whether all these onomastic remnants of aorist participles reflect synchronically living verb forms is impossible to ascertain, but on the most likely assumption they were already archaisms, and their systematic replacement by other forms that conduced to a drastically reduced system with one single active participle for all stems had begun long before our first attestations.

It has recently been proposed (Prosper 2018a) that the universally accepted idea that all the Italic languages except Latin and Venetic have remodeled the endingless nom. sg. -o (> -u) of the masculine nasal stems as *-o-ns somewhere down the line is erroneous: This is not the case in South-Picene, where in my view we find a nasal stem panivu `monument' (and not **panivuf) that agrees with an adjective meitims `most gentle, best' in the nom. sg. masculine (Sp TE 5, Penna Sta. Andrea) A kind reviewer reminds me that the ascription of SP. panivыto the nasal stems is not assured. However, even if the following form meitims were not a superlative adjective as I contend, but a noun, both a credible attri-bution of panivыto another paradigm, or an alternative syntactic analysis that invites one to consider its inclusion in a different word class, are lacking. At all events, the burden of the proof still falls on those postulating the addi-tion of -f to nominatives of nasal stems in Proto-Sabellic.. It is not proven that Umbrian partakes of this analogical extension, either, since the relevant forms show no ending: cf.tribrigu `trinity' (Um 1 Va 9) and karu `meat' (Um 1 Va 24, 27, Vb 4). In my view, the Oscan nom. sg.in-uf is definitely not the outcome of earlier *-ns, but simply the product of the spread to nasal stems of the ending -f, which could only go back to *-nt-s. Moreover, O. -uf is only attested with absolute certainty in one document from Campania, and only in *-(t)i-o, -(t)i-n- stems: fruktatiuf `use, enjoyment', uittiuf `use' and tribarakkiuf (Cm 1 A21, B14, A11-12, Abella). In other words, the spread of -f is comparatively late and contributes nothing to the resolution of the chronological problems of the acc. pl. endings. This morphological innovation, whatever its causes, is not even certain to encompass the whole of the Oscan territory: if Marrucinian BABV is a nasal stem, it has not undergone this innovation, either (see immediately below). In turn, we have to reckon with the possibility that the intrinsically unstable sequence *-nt-s was restored in the individual dialects. But, if what we actually find is the direct, expected Sabellic outcome of *-nt-s, this would only mean that this particular sequence, which was unique in containing a complex coda with a segmental /t/ and word-final /s/, became -f early on. In spite of Rix's convoluted arguments, there cannot have been a Proto-Sabellic evolution *-nts > *-nss > *-ns > -f (which actually goes counter to universal phonetic tendencies) and consequently it cannot be detrimental to my starting point that the Oscan and Umbrian acc. pl. endings have no common source *-ns > -f in any inflectional stem.

Now to the semantics of iaf: if we had to translate this sequence of present participle + verb in personal form into a modern European language, we would have to resort to a verbal hendiadys or pseudo-coordination by which two consecutive verbs, the first of which is often a verb of movement which becomes grammaticalised, come to express one single state of things, and no longer a sequence of separate but related events. The first verb comes to be superfluous and only emphasises the voluntary nature of the action, and thus underscores the subject's decision to do something: the subject is animate and acting both intentionally and immediately. In essence, the whole construction refers to a single event For instance, the dictionary of the Real Academia de la Lengua Espanola (ed. 1992) includes this as a secondary meaning of coger `to get, catch': «unido a otro verbo por la conjuncion y, decidir y cumplir inmediatamente la accion significada por este».. But by using it, they speaker may even be going so far as to decry this event as too daring or even paradoxical. In modern languages, both verbs are coordinated and used in personal form and agree in tense and mood, as in `he goes and says' (a variety of expressions of this kind is also quite common in English: see Stefanowitsch 2000), Sp. `coge y se va', It. `prendo e me ne vado'.

This construction is usually, though not universally, and not exclusively, ingressive, and at all events the inclusion of the first verb form is only intended to modify the inner aspectual- ity of the second. On the pragmatic side, it has taken on some vulgar nuances. On the other hand, the process of grammaticalisation is often not completed, as transpires from the fact that an array of introductory verbs can be used according to the degree of expressiveness, and that the superfluous verb may even take a direct object. Besides the Western European languages, this construction is well known in the Balkans, specifically in Medieval Greek, in which constructions with uiavw and apyiC^ are found. In some languages like Rumanian, the construction can be paratactic and the conjunction can be omitted (see Wagner 1955). By contrast, the introductory verb has been fully grammaticalised in Arabic, to the point of becoming a particle, as in the ingressive particle of verbal origin qam `get up' and its dialectal variants, which modify the aspect of the main verb and can be freely translated as `suddenly' or `without delay' (Naim 2016: 349-353).

The participial construction illustrated by Oscan iaf + 3^rd pers. verb is, mutatis mutandis, the hypotactic equivalent of this structure, and is known to have enjoyed currency in Ancient Greek: for instance, Coseriu (1936: 53) drew attention to the «pleonastic use of Aapwv» attested from Homer onwards; he even thought that the modern paratactic construction can be traced back to the hypotactic one, which conveys a similar meaning: cf. ті p'ou Aapwv ЈKXЈivag `why didn't you go and kill me?' (Soph., O.T.1391), poAu>v AapЈ `come and take them (the weapons)' (attributed to Leonidas at Thermopylae by Plutarch, Ap. Lac. 225c), Ј^ayyЈAA' iu>v 'go and tell' (Soph., O.C. 1393), ooqv кат' аитту uppiv ЈктЈІоаіт' iu>v 'how great a hybris he had gone and avenged on them' (Soph., Ai. 304), Tqv Јpqv Kiorqv iu>v / ЈuAAapЈ `go and take my basket' (Ar., Eq. 1211-1212), etc.

Since we lack anything near a real corpus of Italic texts containing this or similar expressions, we are at a loss as to how far the process of grammaticalisation had progressed, whether other auxiliaries carrying different pragmatic nuances could be employed, or, at the opposite end of the scale, whether iaf had been fossilised into an ingressive particle. Finally, Greek influence from Magna Graecia that could have contributed to the development of this structure cannot be ruled out In point of fact, since the hypotactic construction is alien to Latin syntax, Coseriu points to Late Greek as the model for the paratactic construction in the Romance languages, especially as regards the Italian dialects and Rumanian..

3. The text architecture of Roccagloriosa and Rapino

Line 7 of Roccagloriosa has been plausibly segmented as [-?-ои]отаитiaf кЛопоиот[-?-] by Rix (Lu 62), who in this way privileges one of the alternative segmentations considered by the original editors (see Poccetti et al. 2001: 220-222). They had correctly favoured this analysis, but called for prudence regarding the interpretation of iaf as an acc. pl., in view of the spelling (for expected teiaf, see above), and because it crucially differs from the Oscan synchronic ending -ass and consequently has to be regarded as a remarkable archaism (whatever that means since, as we have seen, no theory of the diachronic evolution of this ending, however elaborated, has been able to accommodate this form thus far)¹². The translation offered by McDonald et al. (2012: 32) reads: `he shall have [-?-] or he shall have stolen them (fem. pl.)'.

My own, alternative translation would provisionally run as follows: `whoever/if somebody should [...] or should intentionally steal [...]' (with a DO in the lost part). In the preceding section of the text there is no single form to which an anaphoric pronoun iaf could refer, but this can be put down to the unknown number of letters that have been lost. At all events, if the two perfective futures are paratactically coordinated and stand in immediate proximity, and further precisions that would separate them and hamper comprehension are lacking, the anaphoric reference to the DO does not seem indispensable. One thing seems clear to me: the active participle underscores the intentionality of an illegal action, which for this very reason is not simply an action, but the commission of an offence. In essence, the participle points to premeditation, and the legal consequences require that the action has been completed, which explains the future perfect.

The bronze tablet of Rapino is conducted in Marrucinian, plausibly defined as a northern Oscan dialect. It raises many different questions relating to ritual and even to the divinities involved, and its reading is far from clear, since all we possess are drawings and the original document was lost in WWII. In contrast, its syntax seems to be comparatively unproblematic for current scholarship, not least because the transmitted text has been subject to a number of «improvements». I shall argue, however, that the accepted translation is the product of a chain of misconceptions affecting the central part of the document, and that a more elegant account can be reasonably defended. It is impossible in a single study to do justice to all the previous interpretations, and my analysis will zero in on the problems posed by lines 5-10. This part, which arguably depicts the second stage of the ritual, pivots around the form iafc, which, as should be clear by now, has been unanimously interpreted as a feminine pronoun in the acc. pl. The text reads (according to Mommsen 1846):

AISOS PACRIS TOTAI / MAROVCAI LIXS / ASIGNAS FERENTER / AVIATAS TOVTAI / ⁵MAROVCAI IOVES /

PATRES OCRES TARIN/CRIS IOVIAS AGINE / IAFC ESVC AGINE ASVM / BABV POLEENIS FERET /

¹⁰REGEN[--] PEAI CERIE IOVIA / PACRSI EITVAM AMATEN/S VENALINAM NI TA[G]A NIPIS PED/і SVAM

As in Roccagloriosa, the spelling is iafc, and not +eafc, ¹³ and it precedes a 3^rd person singular future verb form. In Rapino at least, the construction might be fully grammaticalised, which would account for the use of enclitic -k (possibly also present in Roccagloriosa if we had The alternative segmentation LafK Aonouo (Tocco 2001), which straightforwardly identifies the alleged DO with Marrucinian IAFC, can probably be disregarded, but see immediately below. See more recently McDonald et al. (2012), with a discussion of former views, the interpretation of Roccagloriosa as a legal text concerning theft and its consequences. Anyway, raising of /e/ in hiatus after early loss of intervocalic -i- could justify this rendition, since the whole document has <I> for /e/. to read iaf(K) kAououot with omission of the first -k), which served an emphatic or demarca- tive function that reinforced the syntagmatic unity of participle and the verb As in English `now he goes and says', Sp. `y/pues va y le dice'.. And then, iAFC/iaf(K) would have become little more than an adverbial, synchronically marking ingres- sive aspect. In that case, it would be no different from the external aspect marking in Arabic dialects. As implied above, it is difficult to say whether we are speaking about a syntactic calque from the Greek dialects of Magna Graecia but this possibility cannot be rejected out of hand.

The sentence iafc esvc agine asvm babv poleenis feret is variously translated, e. g. as `eas hac pompa *asum Babo Polenius fert' (Bottiglioni 1954: 331), and more recently as `these, in the ceremony of these, the Babu, the poleenis, take [them] to roast [them]' (Dupraz 2012: 193), and `at the festival of these (divinities) babu polfenis (or babu and polfenis) will carry these (perhaps pieces of meat) for roasting (asum) (Weiss 2010: 136-137, fn. 58). Naturally, the only reason to view esvc as a genitive plural form is the prior assumption that we have already found the DO of feret. But, if iafc is ruled out as an acc. pl., the only forms that qualify to replace it in the unoccupied syntactic slot are esvc and asvm.

Some words are in order as to the intrinsic plausibility of identifying esvc as a genitive plural form, which crucially hinges on the identification of its stem. it is usually held to be an anaphoric pronoun going back to *eiso- (see Dupraz 2012: 195-196). Here is where we find the first obvious obstacle: There must have been either monophthongisation, by which /ei/ yields /e:/ contrary to expectations, or a misspelling, by which <i> has been simply omitted. Neither assumption is especially problematic, since l. 1 reads totai for tovtai, but both require special pleading, and a solution that respects the extant text is preferable.

Secondly, the text is said to refer to a number of divinities mentioned two lines above, which justifies the genitive plural esvc. But this is by no means an unassailable idea, either. in fact, the only divinity mentioned in direct connection to the ritual is iOViA, an epithet that can refer, as traditionally contended, to Demeter, but also to the goddess who was her child by iuppiter, that is to say Persephone. The only possibility of finding more than one divinity to which esvc could refer is to coordinate iovias with ioves patres and make both depend on AGiNE; this is syntactically uncompelling and makes the syntax of the whole text intractable, and it is intriguing that iuppiter is never mentioned again. As we are going to see, ioves patres depends on TARiNCRiS. Accordingly, the reference is to one, not several divinities. This is independently suggested by the corollary iovia pacrsi which, despite the repeated attempts to correct it into ioviai, means nothing but `may iovia be favourable' What is more, the preceding part reading REGEN[--] PEAi CERiE may also refer to her as a Cerean divinity, since she is Ceres' daughter, but the ending is difficult to assess (see Rocca 1997; Prosdocimi 1997). The readings differ substantially at this point, and we have ST:REGEN[Ei] PEOi CEREi iOViA PACRSi; Utah:REGEN[Ei] PiOi CER<Ei>iO- ViA<i>PACRSi (accepting the conceivable but unwarranted correction to CEREi iOViA<i> by Meiser 1987: fn. 51). This problem lies beyond the scope of this work..

Thirdly, even under acceptance of the above premises, esvc is undeniably superfluous from a pragmatic point of view: The reader does not need to be reminded of the destinataries of the festival, ex hypothesi iuppiter and iovia, nor does their festival need to be distinguished from another one in such a short inscription and in an otherwise unambiguous context. in sum, the translation «in the festival of these» entirely relies on the premise that esvc must be a genitive plural because iafc must be an accusative plural because it must anaphorically retrieve ASiGNAS.

i would naturally derive esvc from *ekso-. According to Dupraz, there is no single anaphoric example of this pronoun. As we have seen, this compels him to accept a monophthong- isation of *eiso- for the Marrucinian form (2012: 200), a possibility gainsaid by eitvam. Note that the idea that esvc agine is a phrase with both forms agreeing in the abl. sg., meaning `in this festival' vel sim. (see recently Rocca 1997) ails from the same problem: either the pronoun has undergone monophthongisation or its anaphoric nature is not guaranteed Note the problems surrounding the origins of the ending -<E> of AGINE: if it were an ablative, we would expect +AGINVD. And if it was a locative ending (whatever the synchronic value), the locative PEDI in the last line would be unaccounted for. Accordingly, this is an instrumental -e from IE *-eh<sub>1</sub>, as reconstructed for the Umbrian ablative -e (Rix 1994: 26), Tikkanen (2011: 29). According to Meiser (1987: 112) it is down to the alleged South- Picene substrate. On the most economical assumption, it is suggestive of a Common Sabellic ending *-e, only later replaced in Oscan by the thematic ending, but preserved in peripheral Oscan varieties (the same would apply to Pael. AETATE). One can hardly resist the temptation of identifying the protagonist of this ritual with Greek Baubo, also at-tested since Hellenistic times as Babo, originally an old woman who, according to the myth, managed to make the distressed goddess Demeter laugh. This character is also found in the form Babo on inscriptions as the receiver of cult beside other divinities (IG V: 12, 227), and still present nowadays in fertility folk rituals held in Thrace and Greece, where she (or he, since it is often a man in the guise of a woman) carries a basket or a cup. The appearance of a character with mythological associations is easier to explain than an otherwise unknown personal name, de-void of the slightest hint of an Italic ancestry and unexpected in a ritual prescription.. But then, this may not be an anaphoric pronoun. It could be a text deictic ablative *eksod-k(V) `with regard to the above' > `accordingly' or even `thereupon, from there on', in agreement with the sense reconstructed by Dupraz for *ekso-.

In turn, the hitherto non-negotiable interpretation of iafc as an acc. pl. introduces distortion in the following section of the text, since no other accusative is expected to occur after iafc. In order to bypass this objection, asvm has been understood as a supine *ad-tu-m `in order to roast', which expresses the purpose of the action and goes back to the accusative of an action noun, of which IAFC is the DO: `in order to roast them', `for them to be roasted'. By this expedient, the problem inherent in the lack of agreement vanishes. The resulting construction is then compared to similar Italic ones, like L. ad visitatum Ciceronem or U. anzeriato avef `in order to watch the birds'.

All this is unnecessary if we start from the idea that iafc agrees with the subject of the sentence, babv, and asvm is a substantivated participle *ad-to- in the acc. sg. neuter that naturally takes its place as the DO of feret. If asvm is a match of U. aso (Um 1 VIb, 50) and L. as- sum `roasted', the meat is not taken to be roasted at the feast; on the contrary, once the meat has been roasted, it is taken to be offered to the divinity. Nonetheless, the meaning of U. aso is unclear. The sentence reads erihont aso destre onse fertv, translated by Weiss as `the same person should carry [...] on the right shoulder'. Weiss opens up a number of interpretations for ASO, some of them unrelated to roasting (which is not favoured by the general context), and even proposes a meaning `axe' (*akso-). The Umbrian and the Marrucinian sentences look amazingly similar in that Umbrian erihont (a hapax) is the subject, ASO the DO, fertv (fut. impv.) vs. feret a verb that indicates that someone should bring something, and destre onse vs. poleenis the way or instrument in which it should be carried. In the second case, poleenis stands in the abl. pl., and then, contrary to previous assumptions, it is not a nom. sg. of an adjective or family name in *-iios agreeing with the subject babv¹⁷.

There is little more that can be said about this with any degree of certainty, except that both ASO and asvm are accusatives, both occur as a DO, and both could designate the same thing. To recap, I translate the sequence as `the babu shall thereupon go and carry the roasted? (meat) at the feast in arms/trays? (as an offering) to [...]'.

Let us now take a look at the immediate antecedent of this sentence. tarincris is nearly unanimously believed to be the gen. sg. of a place name in apposition to OCRIS or a divine name agreeing with IOVIAS. Meiser (1987: 113, fn. 50) stated that there is no paradigm of which this form could be the genitive and corrected into tarincriis, which is entirely unwarranted Cf. MV 1, Utal. I: 231, WOU: 735..

What is more, the resulting syntax has never been cogent. These are some of the advanced translations: Pisani (1953: 115): «Prosiciae ferantur auspicatae civitati Marrucinae Iovis patris ocris, Tarincris Ioviae pompa» (una localita chiamata Tarincris Iovia). Vetter (1953: 153) does not fare better when he translates «non prosectae accipiuntur a marruco populo Iovis patris arcis Tarincris Ioviae causa». More promising is Bottiglioni's approach (1954: 331), who translates «Iovis patris ocris Tarincribus Ioviae pompa». More recently, Utal. I: 232 is both noncommittal and forced: `the portions [...] by the pronouncement (?) of Jupiter the father - of the Tarincrine Mount - (and of) Jouia, are brought'.

The correction of TARINCRIS into TARINCRIIS has proved, if anything, detrimental to the global comprehension of the text. tarincris can be taken at face value as an appellative noun in the ablative plural, and the sequence ioves / patres ocres tarin/cris iovias agine can be translated as `in the festival of Iovia, (held) at the top of the hill (= acropolis) of Iuppiter'.

This is consistent with the fact that Italic cult sites were often situated in peaks, especially those related to Iuppiter. tarincris `high-mount?' is a compound similar to L. mediocris `middle- mount' > `average', and may go back to *tarn(o)-okri-fios. While the expected ending as a result of final vowel syncope and assimilation is -iss, it has passed largely unnoticed that this document simply does not use double spelling, and consequently does not note geminates, nor, as in this case, the difference between tense and lax sibilant sounds. Medial syncope would have had the same effect as in O. patensins<*patnse-<*patna-se-, but the resulting form would have undergone context-sensitive raising *tarepkri->taripkri-, and is in this respect comparable to asignas (-sipn-<*-segn-). The form Tadinates occasionally brought to bear on this matter (under the assumption that its base might have been rhotacised in Marrucinian) is in all likelihood unrelated.

The nearest cognates of tarincris are ^BToch. tarne `summit, peak', Hitt. tarna `head, skull', from *tr(H)-no- (see DTB: 298, EDHIL: 845-846). DTB quotes a suggestion by Craig Melchert to the effect that both forms could be grouped under a single preform *tr_oH-no-. In that case, the Italic form would become their full match under the modified version of the «palma»-rule, according to which *CfH.CV- yields Italic *Car.CV- (see Prosper 2020). tarincris consequently meant `high-mount?' and designated the summit of a hill devoted to Iuppiter in which the rites were performed.

4. The fate of Indo-European sibilants in Oscan: an overview

At this point, it is necessary to revisit the Oscan phonemic system, which must be established starting from documents attested in three different alphabets and is fraught with indetermi- nacies and apparent contradictions, especially as regards sibilants (cf. Lejeune 1970; Zair 2016). Even factoring out scribal errors or hesitations in the rendition of geminate segments, the extant testimonies are compatible with the following scenario:

a) Oscan had a lax phoneme /s/ that was contextually realised as a voiced [z] (in intervocalic position, in coda position before a sonorant or a lax/voiced obstruent, and in postsyncope onsets after a sonorant) or as a voiceless [a] (in coda position before a tense/voiceless obstruent, in auslaut, or in internal post-syncope onsets after voiceless stops). The latter variant may have been prone to aspiration and eventual effacement.

b) In early Oscan, /s/ stood in contrast with a tense /s:/ which mostly went back to clusters. As implied above, I provisionally assume that the phonemic contrast was one of lax vs. tense, and additionally suggest that it arose when the first intervocalic clusters including a sibilant, like IE *-t^st- From now on, I shall use this notation for any IE clusters containing a dental sound in immediate contact with /t/, in my view yielding the same outcome. and *-ts-, became /s:/ (to which we may add *-ks-, *-ps-> /s:/ As in Vestinian OSIINS (MV 2), with a four-stroked allograph of <S> reflecting a tense sibilant, < PItal.*op-s- `made', and O. OSINS <*op(i)-si- `adsint' (Lu 1, Bantia); cf. Rix (1993: 336).). The process as a whole may have started in Proto-Italic, since /s/ already had a voiced intervocalic allophone at that stage Cf. Untermann (1968), Stuart-Smith (2004: 91). This contrast cannot be equated to that of simple vs. gemi-nate obstruents or sonorants, however. We cannot assert that what I represent as /s:/ actually behaved like a clus-ter and closed the first syllable in intervocalic position, or that the contrast was one of length rather than, for in-stance, a combination of other feaures, like aspiration or voice. In some contexts the intervocalic sibilant must have been weakened and dropped earlier. Cf. Weiss (2017); Prosper (2018a) for superlative forms, and now Prosper (2019a) for these and other instances of Venetic and Italian Gaulish *-VzV- >*-VhV-> -V-.. Unlike Latin, Oscan had lax sibilants in word-final position, so that, for instance, the suffix of an -s-stem contained a phonemic lax sibilant in every case of the paradigm.

c) The contrast /s/ - /s:/ was neither consistently spelt nor, in all likelihood, existent in all positions from the beginnings of written Oscan (for instance, it was certainly neutralised in initial position). It was mostly rendered <s> vs.<ss> in the Oscan alphabet, <Z>/<o> vs. <o>/<oo>in the Greek alphabet, and <Z> vs. <S> in the Latin alphabet.

d) As far as epigraphy in the national alphabet is concerned, the contrast is attested in medial and final position, both intervocalically (cf. fluusai `to Flora', Sa 1, Agnone, vs. essuf `there', Sa 4, Pietrabbondante), in onsets after sonorants (patensins `aperirent', Cm 1, Abella, vs. kenssurineis, Cp 24, Capua), in codas (fiisnu `temple', Cm 1, Abella, vs. kerssnais `dinners', Cp 31, Capua; fust`will be', Cp 31, Capua, vs. ess-kazsium `access', Cm 2, Campania), and in word-final position (as in aasas `altars', Sa 1, Agnone, vs. feihuss `walls', Cm 1, Abella, meddiss `magistrate' < *-k-s, Cm 6, Nola). There are several examples of <h> for final /s/ before word- initial <s>-, namely upsatuh sent `operati sunt' (Si 4, etc., Teanum) and puiieh sum `cuius sum' (Cp 41, Capua). Final -<s> is also sometimes omitted in the nom. sg. of proper names. This is undoubtedly indicative of lenition If this is a valid criterium in order to establish an etymology, it follows that the recently uncovered nestruis on a tile cannot possibly go back to the comparative corresponding to the superlative nessimas `next, closest' (as per La Regina 2017). It would have given Sabellic *nessistro-, matching the superlative form *nedzizVmo->*nessimo-, to judge from SP. meitistrui, matching the superlative *meitizVmo->*meitimo->meitims `most gentle, dearest', or the Lucanian place name Noupioxpwv (Ptol. 3, 1, 65), Numistro (Livy, Ab Urbe 27, 2), from *nomh]_-is-tero-. It may alter-natively be traced to *nes-tero- `our', L. noster (the root vocalism was modified in analogy to *ues-tero- `your'), a possibility already considered by WOU: 499. The same probably applies to nistrus (Cp 37, Capua). In MINSTREIS (< *ministero-), syncope of both the second and third syllables, which may have operated at different stages, neither compromised the semantics nor gave rise to undesirable phonotactics. Additionally, we have to reckon with anal-ogy, e. g. *mais / *maistro- `more' vis a vis *mins / *minstro- `less', so the processes involved are not entirely clear..

e) We have few instances of IE intervocalic /s/ in the Greek alphabet, and they show contradictory spellings: ЈiCi5o|a, ЈiOЈi5o|r (Lu 5, 11, Rossano di V.), fAouooi (Lu 13, Tricarico), as opposed to the gentilic Kwooavw (Utal. III: 1475-77, Petelia 2), etc. It should be noted that <Z> is used in medial position (Utal. III: 1485, Caulonia 2, see below 5.4) long before the only text containing it in initial position. The epithet accompanying the dative C^F^i `Jove' in Lu 35 is niZ,r\i. Lejeune (1970: 310) reconstructed a stem *pid-es- «du nom de la source, cf. grec niSaZ», which is most uncompelling The rest of the hypotheses about this name presuppose errors in the anlaut or declensional type and can be outright rejected (see WOU: 562).. By the same token, one could propose an -i-stem action noun *kueis-i-, *k<sup>u</sup>is-ei- `survey' (from *k^ueis- `auf etw. achten, wahrnehmen', LIV: 381) which eventually became an agent noun, and relate this formula to Iuppiter Quirlnus (<*k<sup>u</sup>isi-hin-o-) or even to the Quirltes, a name that designated the citizens of Rome in peacetime duties, when they were seeing to their own or public business corresponding to their capacities (if one is allowed to reconstruct *k<sup>u</sup>isi-i-t- `survey-goer'; the multifarious, often untenable associations put forward for this name since ancient times lie beyond the scope of this work). In that case, <i> would stand for /e/, as is sometimes the case.