the central italic languages in the period of roman expansion

32
THE CENTRAL ITALIC LANGUAGES IN THE PERIOD OF ROMAN EXPANSION’ By ROBERT COLEMAN 1.1 It is generally agreed that the Ancient Italic languages divide into three groups: one comprising Latin (La) and Faliscan (Fa); the second Oscan (0s) and Umbrian (Um); the third - sometimes called Sabellian2 - comprising the languages of the Sabini (Sa), Aequi (Ae) and Volsci (VO),~ all adjacent to the Latini (La), beyond this group those of the Marsi (Ma) and Paeligni (Pa), and away on the east coast those of the Vestini (Ve) and Marrucini (Mr). It is with this third ‘Sabellian’ group that the following study is chiefly concerned: in particular with questions of affinity between individual languages in the group and others either inside or outside the group. The geographical distribution of the relevant languages is indicated approximately on the map (Fig. 1). 1.2 Widely differing answers have been given to these questions, as a glance at the table below (Fig. 2)4 indicates: Buck and Pisan? apparently found insufficient evidence for subdividing the group, which they labelled respectively as Sabel- lian and Dialetti Minori. Conway, Poultney, Maniet and Durante all differ not only in their subdivision of the group but also in assigning external connections for the various subgroups. Con- way’s Latinian subgroup, for instance, included also Praenestine, which most scholars would regard as an idiosyncratic dialect of Latin,‘ while the assignment of the Pa-Mr-Ve trio to Oscan, as in Conway’s, Maniet’s and Durante’s classifications, places them in a continuum with the Oscan dialects of the Frentani or of Bantia or Capua. 1.3 Mention of the Frentani raises a further question. The seven linguistic units in Central Italic are defined by the tribal divisions recognized by the Romans, but there is no guarantee that these corresponded. The language used at Antinum, for instance, well inside Marsian territory, is often taken to be a dialect of Voscian

Upload: fsdnc

Post on 02-Feb-2016

42 views

Category:

Documents


5 download

DESCRIPTION

Coleman 1986 Ancient Italic languages divide into three groups: one comprising Latin and Faliscan ; the second Oscan and Umbrian; the third - sometimes called Sabellian - comprising the languages of the Sabini, Aequi and Volsci, all adjacent to the Latini,

TRANSCRIPT

Page 1: The Central Italic Languages In the period of Roman expansion

THE CENTRAL ITALIC LANGUAGES IN THE PERIOD OF ROMAN EXPANSION’

By ROBERT COLEMAN

1.1 It is generally agreed that the Ancient Italic languages divide into three groups: one comprising Latin (La) and Faliscan (Fa); the second Oscan (0s) and Umbrian (Um); the third - sometimes called Sabellian2 - comprising the languages of the Sabini (Sa), Aequi (Ae) and Volsci ( V O ) , ~ all adjacent to the Latini (La), beyond this group those of the Marsi (Ma) and Paeligni (Pa), and away on the east coast those of the Vestini (Ve) and Marrucini (Mr). It is with this third ‘Sabellian’ group that the following study is chiefly concerned: in particular with questions of affinity between individual languages in the group and others either inside or outside the group. The geographical distribution of the relevant languages is indicated approximately on the map (Fig. 1).

1.2 Widely differing answers have been given to these questions, as a glance at the table below (Fig. 2)4 indicates:

Buck and Pisan? apparently found insufficient evidence for subdividing the group, which they labelled respectively as Sabel- lian and Dialetti Minori. Conway, Poultney, Maniet and Durante all differ not only in their subdivision of the group but also in assigning external connections for the various subgroups. Con- way’s Latinian subgroup, for instance, included also Praenestine, which most scholars would regard as an idiosyncratic dialect of Latin,‘ while the assignment of the Pa-Mr-Ve trio to Oscan, as in Conway’s, Maniet’s and Durante’s classifications, places them in a continuum with the Oscan dialects of the Frentani or of Bantia or Capua.

1.3 Mention of the Frentani raises a further question. The seven linguistic units in Central Italic are defined by the tribal divisions recognized by the Romans, but there is no guarantee that these corresponded. The language used at Antinum, for instance, well inside Marsian territory, is often taken to be a dialect of Voscian

Page 2: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 101

Aurunci ( A

Carnpani (CA j

Samnitc\ (SM) Mor\r ( M A ) l lcrnici (111:) A rqur ( A t )

Sahrrir(SA) Piccni (PI) Vc,.>~rnt (VE) f'aulfgrir (PA)

Marnrc I P I I ( M K j Frentani (FRj

I1 I 0 ? ( I ill JII 50 L I I I I J

kin

Figure I

rather than of Marsian (see 85.6). Moreover some of the tribal regions covered large areas and it is by no means certain how far the evidence of, say, a single inscription from one place like Velitrae can be justly taken to represent the language of a popula- tion as widespread as the Volsci. In practice I have eschewed extremes of scepticism on these matters and have adopted only one self-denying principle, namely to reject all data that are either epigraphically ambiguous or have no generally agreed linguistic interpretation.

Page 3: The Central Italic Languages In the period of Roman expansion

102 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

Volscian

Oscan

Sa Ae

Ma

vo Pa Mr

Ve

Sabellian Umbric

Volscian

North Oscan- Oscan Oscan Sabellian

Conway [Vetter I Poultney IManiet

I I I

Durante

Oscan

Umbrian

Oscan

Figure 2

2.1 The Paeligni inhabited an area east of the Fucine Lake. They were allied to Rome at the end of C4’ and remained loyal until the Allied Rebellion of 91-87 B.c., when their principal town Corfi- nium was the rebel capital. Unlike their western neighbours the Aequi, who had received large Latin colonies at the end of C48 and were therefore in all probability Latinized relatively early, the Paeligni retained their language epigraphically until C1 .9 Of the surviving inscriptions most are from Corfinium (V 21 1-215, Po 209),” Sulmo (V 202-210,’’ Po 211-14) and Superaequom (V 216,217). The earliest (V 210b) may be as old as C3, the latest (V 209) is from c.50 B . C . Nearly all are short: the longest (V 213), from early C1, has thirty-six words.

2.2 [1]12 A number of Pa lexemes have parallels in Oscan and Umbrian but not in Latin; thus aisis ‘to the gods’ (originally Etruscan: Suet. Aug. 97); cf. 0 s AISUSIS (abl.) ‘sacrifices’, Um ESUNES (abl.) ‘rites’. cusnar ‘old man’, medix ‘magistrate’ and solois ‘to all’ were all specified as Oscan in Antiquity.I3 ecuc ‘this’ (fem.nom.) has the stem of 0 s EKAK (< *ekrI.m-ke) rather than 0 s exa-c, Um esu (< *ek-srI.d). ip ‘there’ = 0 s fp, Ma (ce-)ip, herentus ‘Desire’ (V 213) and puclois ‘to the children’ show lexemes attested elsewhere only in Oscan. fume1 ‘slave’ is cognate with La famulus, which is said to be Oscan in origin but may have been

Page 4: The Central Italic Languages In the period of Roman expansion

COLEMAN - T H E CENTRAL ITALIC LANGUAGES 103

pan-Italic,“ as was the denominative verb *opes-& reflected in upsaseter, the sense of which, ‘to make’, however, agrees with Oscan and Umbrian (cf. #2.4[12] and 4.2). A few words are found in Umbrian, not Oscan, but these also occur in other languages; e.g. bium ‘fresh spring’ (?) = Urn bia, Mr bea; et ‘and’ (V204) = Um and La et as against 0 s i ~ i ~ ; pacris ‘propitiatory’ = Um pacer, Mr pacris. The accidents of survival make firm inference hazar- dous, but it is remarkable that in so small a lexical corpus a positive Oscan connection should be so discernible.

2.3 Paelignian phonology shares a number of items with Oscan and Umbrian, e.g. the treatment of original labio-velars in pam and bratom (= La quam, gratum) and of the voiced dental aspirate in loufir ‘free’ (cf.n. 2 5 ) , the backing and raising of *cl and *d in prismu, ptruna (= La prima, Petrbnia) and the syncope exhibited in graex (= La Graecus: Po 208) and ponties, popdis, with *-ilos > -iis > -ies; cf. 0 s P ~ N T I I S , f e o n q , Um TETEIES.

The reduction of consonant clusters shows a less tidy pat- tern of comparison. Medial *ks is reduced, as regularly in Umbrian and sometimes in Oscan. Thus usur ‘women’ < *uksores; cf. 0 s USURS (acc.:) but La uxores. In final position both original and secondary ks remain: sacaracirix ‘priestess’ (= La *sacratrix) and graex. Oscan and Umbrian tended to reduce both;” e.g. 0 s MEDD~SS (< *-iks), Urn uas (<*wkos). , but paradigmatic analogy seems to have operated in 0 s TUVT~KS, Um FRATREKS (< *-ikos) and there may have been dialectal variation in Oscan, e.g. M E D D ~ S S

(< *-ikes V 115 Nola)I6 but ~ E ~ G E L E (V 196 Messina), though these could represent diachronic rather than diatopic variation. Paelignian in any event is more conservative than either language. Medial * k t shows the same pattern of comparison as medial *ks: in sat0 ‘revered‘ beside 0 s S A A H T ~ J M , Urn SAHTA (acc.fem.) < *sa(n)kt-; cf. La sanctus. uicturei in the phrase Herclei uicturei (V 217), like 0 s V ~ K T U R R A ~ ‘to Victoria’ (Po 16, Bovianum Vetus). is recent enough to be a Latinism.”

Palatalization is attested in petiedu, uibdu (V 213 = La Petiedia, Vibidia), where the local letter 43 is generally intepreted as [dz] or [d3] from *[dj], and rnusesa (V 204 = La Mussedia) where -sa represents the further shift of [dz] to [z]. popdis (V 212 = La Popidius), musedi (Po 216 = La Mussedi(us)) reflect the [di]

[2]

[3]

Page 5: The Central Italic Languages In the period of Roman expansion

104 TRANSACTIONS OF T H E PHILOLOGICAL SOCIETY 1086

variant ”-dios with syncope of *o. It seems that initial [dj] > [j], as in Latin and Marrucinian; e.g. iouiois (dat.) ‘belonging to Jove’, cf. Mr ioues, La Iouis (also late 0 s 1irv~i for earlier DtljvEi). The palatalization *[nja] > [pa] is seen in petruna (V 215 = La Petro- nia) and perhaps anaes (V 214), if this is the same name as aniaes (V 215). There is, however, no trace of any change of articulation in the velar stops; e.g. ecic (< *&id-ke), Loucies.

By contrast Umbrian has palatalization of * k before e, i and ;+vowel, e.g. Sesna, TICIT (= Lat cena, decet) and P u R r I N c u s ‘you will have offered’ (< *pordink&.s-s) and of n before ;+vowel in

no trace of it in dental stops before [j]. The only Italic” parallels for the latter in this period are in Marsian martses (<*Mar@eis; cf.

and one dialect area of Oscan. At Bantia (mod. Banzi) we find bansae ‘in Bantia’, and in initial position zicolom ‘day’ (<*dieko- l o m ) and at nearby Rossano di Vaglio LoFqt (<*dl:ouei ‘to Jove’) beside GtwFyc’ (Po 169, 167). Bantia also has famelo and herest (<*famefja and herjest, attested in Umbrian), which pre- sumably indicate palatalization of [l] and [r] before [j]; in which case the absence of any examples of [p] from [nj] would be merely accidental. Bantian meddixud (<*medodik@d) is comparable to Umbrian, but contrasts with Paelignian treatment of ki+vowel.

The pattern of Paelignian palatalization” is thus closer to Mar- sian and Sabine - so far as the meagre evidence from these languages goes - and especially Bantian Oscan than it is to Umbrian. But the parallel is far from complete and there is no justification for regarding Paelignian as the epicentre for South Italic palatalization.*” In any event the intervening Samnite dia- lects of Oscan show no sign of palatalization, and the phenomena discussed above are better treated as independent developments, whether due to substrate or other causes.

The loss of final d after long vowels and of final m, frequent in Umbrian but almost unknown in Oscan, occurs sporadically in Paelignian. Thus oisu (abl.) < *oissad (cf. La Gsu) and on V 213 dida ‘may she give’ < *didad beside fertlid ‘fertilely’. -m is con- sistently written in V 203 (e.g. p a m , inom) but omitted in V 204: sato < *sa(n)ctom. V 213 again shows variation between pracom,

SPINA, SPINIA ‘COlUmll’ and RUPINAM, RUPINIE (Cf. La Rubinia); but

OS MAMERTTlAfS with -=I- for [tj]), Sabine C f U u S u S (< *Claudios)

[4]

Page 6: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITAI.IC LANGUAGES I05

cerfum and deti <: ”deiueti(o)m; also praicime < “praidiciom- en , with loss of -n , as often in Umbriari but never in Oscan. It is clear that the same causes that brought about the loss of these consonants in Umbrian and Latin operated here also, but there is no necessity to assume either a diffusion specifically from Umbrian or - for the retention of -m - the dominant influence of Latin orthography.”

Phonological parallels specifically with Oscan include [5] the retention of diphthongs, though there are a few apparent excep- tions, notably in the treatment of ei and oi.

(i) ei: sefei ‘to himself‘, eite ‘go’. If des and deli (V 213, 214) are cognate with La diues ‘rich’, then they attest at least for Corfinium a monophthongization”2 comparable to that of Umbrian and Faliscan.

(ii) oi: oisa (abl.) ‘having been used’, empratois (dat.abl.) ‘ordered’. But uisis and lifar are problematic. In view of Mr aisos and Ma esos (norn.pl.), aisis is unlikely to be a con- sonant- or i-stem dative (< “~isifs).~~ The inscription seems to belong to the last part of C2, when the La thematic dat. pl. was already monophthongized to -is; but in what is otherwise a very un-Latinized text (V 204) the assumption of a Latinism is implausible. So better take -is here as parallel to La -is and Um -ir (< -cr < -eis < -ois) in uerir, etc. lifur (V 213) can hardly be a Latin l~anword , ’~ but it might just be a contami- nation of a native “loufur (cf. 0 s L O V F R E ~ S ‘of Liber’) or ”leifar by the La Liber.2s To take it as an impersonal subjunc- tive ‘let it be pleasing’ (< *lubh-u-r)2h presupposes Latin interference, since the change of ir to i here cannot be assumed for any other Italic lang~age .~’ More probably (assuming the form is not an epigraphic error) it reflects like aisis the monophthongization of ei followed by raising, the older form being preserved in des alongside i t . Parallels are regular in Latin (Iiber etc.) but rare in this position in Umbrian; e.g. irer but erer, for which cf. 0 s E ~ S E I S .

(iii) ai: aisis and yraicime. aetatu ‘life-time’, which occurs with pruicime on V 213, and anaes ‘Annaeus’ represent the same shift in the diphthong as La ae and 0 s ~ i , ue.*’ For iii see [7] below.

Page 7: The Central Italic Languages In the period of Roman expansion

106 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

(iv) au: pfauties (= La Plautius). (v) ou: loujir and foucies (= La Lficius). Uranias 'of Urania' on

Anaptyxis, which is a familiar Oscan characteristic, is well attested in Paelignian. unacetu (V 214) 'to Angitia' beside anacta (V 207) and anceta (V 211), all < *ungeta- (cf. 0 s ANAGTIAI < *ungetia- and for the anaptyctic vocalism generally 0 s ARAGETUD < *argent6d); salauatur (V 215) < *salucit6r. V 213 offers two notable instances: pristafalucirix < *prist@itrix 'assistant priest- ess'; cf. 0 s STAFLATAS (fem.nom.pl. 'set up'); and sacaracirix < sacracrix,2y attested in Marrucinian, < *sacrcitrix. For the initial syllables cf. 0 s S A K A R A K L ~ M < *sacratlorn and for the final syl- lables 0 s A A D I R I ~ S < *Atrius, zicolom < * d i C k l ~ m . ~ "

[7] Divergent from both Oscan and Umbrian is the ci reflex of -at- in dat.sg. minerua, anacefa, cerria. Most of the examples are

from Sulmo, all are relatively recent. Although it is found in the Latin of S.E. Latium and of other areas, where it contrasts with the forms current in the local languages, there is no ground for assuming Latin influence, and the change, which must postdate the raising and backing of older 1I in pvisrnu, etc., may have been widespread in the Central languages.32

Another divergence from Oscan and Umbrian, shared with Marrucinian (cf. [6]), is [8] the change *tr > *cr reflected in sacaracirix. The [k] here could have been by assimilation to the preceding [k] before (or after) anaptyxis,33 but this is less likely.

2.4 In Paelignian morphology, beside features shared with both Oscan and Umbrian, like nom.pl. puus 'who' < *kW6s, 3 pl. -ns in coisutens, there are a few that are divisive.

Abl.sg. aetate (V 214) agrees with the Um KAPIRE and La cupide, lege against 0 s ligud. The inscription is recent enough to permit the assumption of Latin influence, but there are no other obvious Latinisms in the remaining eleven words. uetutu (V 213) is either abl. (cf. 0 s ligud) or more likely acc. (cf. 0 s tunginom, Um ABRUNU). aerate seems to point to an Umbrian affinity, but 0 s -ud looks like a recent innovation and may have replaced the cognate of aetute, viz. * A ~ T A T ( ~ ) ; so the divergence from Oscan is not quite conclusive.

V 213 is sufficiently recent to be a Latinism. [6]

* - .31

[9]

Page 8: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 107

The i-stem nom.pl. pucris, formed by analogy with nom. -as and -6s, is paralleled in Mr pacris, 0 s A ~ D I L I S (Pompeii, Sam- nium) from -is as against T R ~ S (Capua) from *-&, which is also reflected in La trds, Um pacrer.

A dental formant appears in the perfects locatinfs) ‘they put out to contract’ and coisatens ‘they made arrangements for’. Parallels occur probably in Vo sistiatiens, certainly in Mr urnatens,

paradigm and have no parallel in Umbrian or Latin. The impf.subj.pass. upsaseter < *opescIseter (cf. La oper-

aretur) indicates that Paelignian like Oscan and Latin did not distinguish primary and secondary inflections (cf. 0 s SAKARATER

(indic.), SAKA<RA>H~TER (subj.) with Lasacrdtur, sacrttur but Um indic. HERTER, subj. EMANTUR).

2.5 Paelignian thus shows agreement with Oscan in some of its phonology, [5] and 161, as well as in its lexicon, [l], and a strong agreement in its morphology, [lo], [ 111 and [ 121. Of the remaining items [2] is equivocal, [3] is shared with one area of Oscan, [7] with other Central languages and dialects, [4] and [9] with these and also Umbrian.

3.1 The Marrucini, settled in the region between the Adriatic and the eastern slopes of the Apennines, were already allied to Rome before 300 B.C. but like their kinsmen the Vestini and Paeligni joined in the Allied Rebellion of 90 B.C. There is one substantial inscription (V 218), dated c . 2 5 0 B.c., from Rapino near the border with the Oscan-speaking Frentani. To the brief V 219 can be added Po 204-206. The remains are meagre, some of the data are obscure or equivocal, but a few characteristic features emerge.

3.2 In the lexicon [ l ] some items have cognates in Oscan and Umbrian, e.g. aisos ‘gods’ (cf. §2.1), esuc ‘of these’ (gen.) or ‘thus’ (abl.), reflecting either *eiso- ( 0 s EISUN-K, Urn ERU) or *ekso- (0s exuc, Um toutai ‘to the people’. !Specifically Oscan connec- tions are exemplified in agine ‘on account of’ (cf. 0 s acc. pl. AGINSS ‘actions’), eiruarn ‘money’ and on Po 206 ecan ‘this’ (fern. acc.sg.). Specifically Umbrian are beu ‘fresh spring’ (Po 206),

[lo]

[ll]

0 s DIUUNATED and PRUFAITENS. All the forms belong to the %stem

[12]

Page 9: The Central Italic Languages In the period of Roman expansion

108 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986 pacris ‘propitious’, also attested in Paelignian, ocres ‘of the moun- tain’ andpedi ‘anything’ (cf. Um PIRE but 0 s P ~ D U M ) . There is thus much less of a bias towards Oscan here than in Paelignian.

3.3 A number of phonological items are shared with both Oscan and Umbrian, e.g. the palatalization of *n before *! in k i n a (Po 205); the syncope of *-iios in dies ‘other’ (see 02.2); and the treatment of original labio-velars in pedi (< *k”id-i), which also shows the hesitation between e and i resulting from the raising of *[e.] seen in ni, Iixs. The raising itself is paralleled in 0 s ni, Iigud and the graphic indecision in Urn screhto, screihtor (< *skri-; cf. 0 s s c r i f t ~ s ) . ~ ~ Here original [i.] remains, as in pacris, uenalinam.

Parallel with Oscan is [2] the anaptyxis and syncope in salaus (Po 205) and with further syncope salas (Po 204); cf. 0 s salaus < *salauos < *saluos. However, sacracrix (Po 204, 205), unlike Pa sacaracirix, salauatur ($2.3[6]), 0 s S A K A R A K L ~ M , is not affected; nor is asignas ‘cut portion^'^' (< *ad- or amf- + sek-na-), for which cf. 0 s oeyovo (Po 175), S E G A N A ~ E D (Po 21), A K E N E ~ but Um acnu, etc. So the connection with Oscan here is only partial. It is worth adding that the change [el > [i] in asignas recalls the raising before [g] in Um cringatro and especially La signum (see 35.3151). More clearly Oscan is [3] the conservation of final -m and -s in eituam, ioues etc.

Against these can be set [4] the Umbrian treatment of final -t: pacrsi < *pacrisit, ta[g]a < *tagut; cf. Urn si, FACIA. However final *-nt loses the nasal, as in Oscan: thusferet ‘they receive’; cf. 0 s SET and SENT, Um only sent.37

Some items do not match neatly with either Oscan or Umbrian. Thus [5] the variation in the treatment of diphthongs. On V 218 beside eituam we find ioues, patres, ocres reflecting the gen.sg. *-eis (cf. Um matrer but 0 s MAATRE~S), unless these represent PIt *-es, which is reflected in Latin and Faliscan. On the same inscrip- tion we find toutai, maroucai beside totai. The remarkable graphic variation in a single text suggests that ei and ou had already been monophthongized, but - in contrast to Umbrian - recently enough for the digraphs to be retained as an archaism. While the retention of the diphthongs in the neighbouring Frentanian dialect of Oscan (e.g. MAATRE~S, K A ~ L ( I S ) , L I ~ V F R E ~ S ) may have affected spelling

Page 10: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 109

convention, there is no close source of influence for the monoph- thongal forms.

The situation with ui is even more obscure. aisos is clear enough, but the forms of the a-stem dative pose a problem. V 218 has not only toutai, peai, rnaroucai but also iouia and - on the usual interpretation - cerie. -ai, -a and -e cannot of course belong to the same diachrony. If -ai = [ail, then -e will represent its monoph- thongal reflex (cf. Um GERFIE against 0 s K E R R ~ A ~ ) , and iouia must either be meant for iouiai, with -i omitted at the edge of the bronze, or less likely be an intrusive form (from Paelignian?). If on the other hand -ai here = [a.i], then -a could be its normal reflex and cerie either an intruder (from Umbrian?) or more likely a locative, reflecting *-ai or *-ei. In view of the -6 dative in Paelignian (§2.3[7]), it is certainly tempting to take the second option, but the question must be left open. For the possibility that *oi also > 0 see n. 41.

iafc ‘these’ (fem.acc.pl. < *e&ins-ke) nicely illustrates the phonological ambivalence of Marrucinian. For it has the raised initial vowel of 0 s ionc, ~ A K , etc. but the Umbrian reflex of -ns; cf. Um eaf, EAF with os ASS.^^

A few phonological items are not paralleled at all in either Oscan or Umbrian. Thus [7] cibat (Po 205) < cubat, if it is not an inscriber’s error, recalls nothing geographically closer3’ than La fiber < lubet and may similarly have taken place first in com- pounds. [8] rnedix (Po 206)40 shows retention of final [ks], as in Latin and Paelignian (§2.3[2]). [9] The change of *tr to cr in sacracrix (Po 204, 205) is paralleled only in Paelignian (§2.3[8]).

3.4[10] There are a few morphological items that are shared with both Oscan and Umbrian: nom.pl. -6s in aisos41 and -6s in asignas; -er in 3 sg.pres.pass. ferenter. patres also belongs here, if it reflects +is, but divides Marrucinian from both Oscan and Umbrian if it reflects -es (cf. [5] above). 1111 The cons. stem abl. agine provides a connection with Umbrian, but also with Latin and Paelignian (92.4191). On the other hand two items associate Marrucinian with Oscan: [12] nom.pl. pacris and [13] the dental formant in 3 pl.pf. amatens ‘they decided’, both also found in Paelignian (§2.4[10],

[6]

[Il l) .

Page 11: The Central Italic Languages In the period of Roman expansion

110 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

3.5 Of the decisive phenomena considered above, [2] and [3] show agreement with Oscan in phonology, [12] and [13] in mor- phology, but only [4] and [ll] respectively agree with Umbrian. It is interesting that Paelignian shares four of these - [2], [ll], [12] and [13] - together with two of the three phonological features - (81 and [9] - that are not found in either Oscan or Umbrian. The only exclusively Marrucinian item, [7], is of uncertain status, and the remainder show agreements in both directions, as epitomized in iafc.

4.1 From the Vestini the material is very meagre indeed: three short inscriptions, V 220, 22742 and Po 207,43 all dating from the period 250-100 B.C.

4.2 A number of Vestinian items agree with both Oscan and Umbrian: the denominative verb *opes-& ‘to make’ in &ens (Po 207) ‘they made’ (cf. 0 s UUPSENS, Um osatu ‘let him make’, La operari ‘to be busy’); the change of *-iios to ils in Ebdies (Po 207) < * E p i d i o ~ ; ~ ~ the raising of 6 to z i in dun0 < diinorn ‘gift’ (cf. 0 s D ~ N ~ M , Urn flus a flu sure < *flos%ri (cf. 0 s FLUUSAf); the reflex of [g“] in brat ‘thanks’ (cf. 0 s brateis, La gratus); the reduplicated pres. didet ‘he gives’45 (cf. Um subj. dirsa, 0 s fut. didest). All these are shared with Paelignian also (see 032.2, 2.3 init.). Iffifeltares is the correct reading on a Latin inscription from Furfo (58 B.c.), then the intervocalic f points to a non-Latin origin46 and the word can reasonably be attributed to Vestinian.

4.3 The only secure item that is shared with Oscan against Umbrian is [ l ] the retention of -s- in flusare. On the other hand a number of items agree with Umbrian. Thus [2] the reduction of ps resulting from syncope in oSens < *Gpesenns; cf. 0 s UPSED, Um osatu; [3] the change Vns > Vs in mesene; cf. Ma nouesede < *nozynsedc?s, Um UZE, aseriatu beside onse, ANZERIATU and MENZNE. The use of the same suffix -en- in the Vestinian and Umbrian forms of this word is inconclusive without Oscan evi- dence. But if locatival functions had already been taken over by the intrusive *-dd in Oscan consonant stems (with figud replacing loc. *Zig from *Zi?gi), just as the ablatival functions had been by the loc. sg. in Latin and Umbrian (cf. pede, PERI), then [4] rnesene does indeed align Vestinian with Umbrian rather than O ~ c a n . ~ ~

Page 12: The Central Italic Languages In the period of Roman expansion

COLEMAN -- THE CENTRAL ITALIC L.ANGUAGES 111

The presence of the diphthong in poimunien ‘in the sanctuary of Pomonus/a’ does not exclude an Umbrian connection, since the word derives like Um PUEMUNES ‘of Pomonus’ from *pouim-. No data are available for the wider treatment of diphthongs in Vesti- nian. flusare could be either cons. stem abl. (< *loc. -i; cf. [4]) or i- stem loc. (< -*ei; cf. Um UKRE beside abl. U K R I < -*id).

4.4 Two items appear to distinguish Vestinian from both Oscan and Umbrian. [5] Final *-s has been lost in data, assuming that brat. data < *brcitcs *datcis (nom.) ‘thanks given’; cf. Ma nouesede but Pa dutas, 0 s scriftas, Um U R T A S . ~ ~ [6] herclo ‘to Hercules’ has the thematic declension, as in 0 s H E R E K L I ~ , Paelignian La hercolo. This is not in itself conclusive since we have no compar- able forms in Umbrian. But the dat.sg. -6,4y also attested in iouio. is probably derived like La -6 direct from PIt -6i, possibly from its Oscan reflex -oi, as in HrjRTrif (cf. La [oil > [Q-] (> [u.]) in curare etc.), but certainly not from its Umbrian reflex -e, as in p ~ p l e . ~ ”

4.5 Of the five items enumerated only [l] points towards Oscan, and even this could be interpreted as reflecting an original Umbrian dialect that diverged before rhotacism occurred in that language. [6], which like [5] is an independent development that could have started equally well from either Oscan or Umbrian, also implies an early date for divergence, depending as it does on the presence of -oi in the parent language. Items [2], [3] and especially [4] agree with Umbrian; [3] and [S] with Marsian.

5.1 For the Marsi also the data are very meagre: two short inscriptions, V 225 and Po 222, dating from 300-150 B.c., and seven dialect Latin inscriptions included in V 228 and Po 219-224, ranging in date from c.280 (V 228a) to 150 B.C. V 223, whose attribution is disputed, is treated in 45.6.

5.2 In the Marsian lexicon esos ‘gods’ has exact cognates in Pa aisis, Mr aisos, and the root is represented in Oscan and Umbrian (see §2.2[1]). Umbrian cognates are extant for pacre ‘propiatory’ or ‘propitious’ (cf. Um pacer, Pa and Mrpacris); pesco ‘offering’ < *perk- (cf. Um PEKSKLUM, 0 s P E S T L ~ ~ M ‘temple’). pucle[s ‘to the sons’ (Po 222) is paralleled in 0 s PUKLIJI, Pa puclois. nouesede ‘of the nine seats’” turns up also in Sabine and Latin (see 47.3[5]). If

Page 13: The Central Italic Languages In the period of Roman expansion

112 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986 ceip = ‘here’ on the Latin V 228a, then it will be a Marsian com- pound of ce-, as in La ce-do ‘give me’, 0 s cebnust ‘he will have come here’, and the reflex of *ikwe, as in 0 s ip ‘there’, Pa ip. It is striking in so small a lexical sample how many items are shared with one or more of the other Central languages as well as with Oscan and/or Umbrian.

5.3 Turning to the phonology of Marsian, (11 martses (< *mar- riais abl.pl) on V 228a shows palatalization comparable to that of Paelignian, Sabine and the Oscan of Bantia (cf. §2.3[3]). No test forms occur in Marsian itself and the other Latin inscriptions show no palatalization - Vetius, Atiedius (V 228) and Cisiedio (Po 218); but La Marsus ‘Marsian’ beside Martius ‘of Mars’ suggests that the palatalization is genuinely Marsian.

Dat. sg. herclo, from an o-stem form, as in Oscan, Paelig- nian and Vestinian, is a direct reflex of PIt. -0i or less likely of the -oi attested in Oscan. It cannot be in diachrony with Um -e (see 64.4). On the other hand, the treatment of *ei and *ai recalls Umbrian. Thus [3] ioue ‘to Jove’, iouies pucle[s; cf. Um IUVE, VERES ‘gates’ (dat.pl.); [4]’esos < *ah- ; cf. Um ESUNES, Ae state but 0 s AISUSIS, Pa aisis, Mr aisos. Both monophthongal reflexes occur in the Latin inscriptions also, e.g. uecos (nom.sg.), iue (< *iouei) and probably queistores with ei for [e.] from [ail.

[5] seino and seinq. (for seing.) ‘statue’, on two late C3 Latin inscriptions (Po 223,220), < *sek-nom. La signum is derived thus: [segno] > *[seqno] > [seqnu] a.013 [siqnu]; cf. seignurn (CIL I’ 42 Nemi) with ei for [el, singnijer (VI 3637), dgnurn (I2 2096a, S. Umbria) with I for [i].52 seing(no) would fit easily into this diachrony; seino would do so only on the hypothesis of a dialectal reduction of the cluster [qn] to [n.], for which cf. sinnu (CIL IX 2893, coastal Frentani). Reduction of the relevant clusters seems to have occurred in Umbrian: *ko(n)-gnigatos > KUNIKAZ ‘kneel- ing’; cf. *nigrom > niru ‘black’ with [nigro] > [niyro] > [ni-ru]. So seino may reflect a Marsian treatment of vowelSgn cluster that is at least partially comparable to both Umbrian and Latin. 171 the change of Vns to Vs in nouesede links Marsian with Vestinian and Umbrian against Oscan (see §4.3[3]).

5.5 In the absence of criteria1 morphological items the affinities of Marsian must remain obscure. Apart from [ 5 ] all the items listed

[2]

Page 14: The Central Italic Languages In the period of Roman expansion

COLEMAN - TIIE CENTRAL ITALIC LANGUAGFS 113

above are shared with one or more of the Central languages, [ 11 with an Oscan dialect as well, [2], [3], [4] with Umbrian.s4

5.6 It is appropriate now to look at the disputed V 223.” This inscription, dated c. 150 BC., is from Antinum, which lies towards the north end of the Liris valley, down which the Volscians prob- ably came in their C4 migration southwards. Though separated by a mountain range from the chief area of Marsian settlement and 3-4 hours journey from Lucus (Luco), it was well within the borders of what the Romans regarded as Marsian territory.s6 By this date it is most unlikely that any earlier Volscian settlement that may have survived there would have retained its language, but substrate or interlingual effects on the Marsian spoken there must be allowed for.

The inscription shares with Volscian (V 222), Aequian (V 226) and possibly Marsian (V 224a) the ‘Umbrian’ onomastic formula, personal + paternal + gentile names.” [ l ] The form of cetur ‘censor’, if it reflects *kens-tor (cf. 0 s KEENZSTUR) with [e.nst] > [e.nt] > [e.t], is paralleled for the last stage by 0 s AKAGETUD, Um USTETU and Volscian La lubetes (I2 1531). All the other data in V 223 have parallels elsewhere. Thus for [2] -ijs from *-izos in pucuies ‘Pacuuius’ cf. Ae pomposiies, Vo tufunies; for [3] medis with [s-] from *[ks] cf. Ae meddiss, 0 s MEDD~SS.

as in Ma esos, Vo declune. The same word together with dunom and cetur shows [5] raising of earlier *6,59 as in Vo declune, estu but apparently not in Marsian, if esos in V 225 is nom. or voc.pl. Two possible explanations are to hand. The first is that the variation is dialectal; with uesune etc. at Antinum perhaps due to Volscian influence or common substrate. Alternatively, since esos is unusual among the neighbouring languages, all of which raised original *6, esos nouesede (V 224) might be dat.pl., with -0s < 4 s . However, nouesede < *-iss < * - i J y is hard to accept,60 unless the final sibilant has been omitted at the edge of the plaque.

Of these five items [4] agrees with both Marsian and Volscian, [3] with Ma against Vo, [5] with Vo but probably not Ma, [2] with Vo but without a comparable item in Ma, and [l] cannot be tested against either. The linguistic criteria, as far as they go, cannot resolve the dilemma, and if V 223 is to be assigned to

Dat.sg. uesune shows [4] the monophthongization of

Page 15: The Central Italic Languages In the period of Roman expansion

1 14 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

a (Volscianized) dialect of Marsian, it must be mainly for geographical reasons.

6.1 The Aequi, who had in early C5 been as far East as the Alban hills in Latium, received Latin colonies in Carseoli (Carsoli) and Alba Fucens (Avezzano) at the end of C4 and were quickly Latinized.61 Only one Aequian inscription survives, V 226,62 dating from C2, and there are two dialect Latin texts, CIL I2 1831 (late C3) from the area of Cliternia near the border with the Sabini and V 228h from Alba somewhat later.

6.2 The Oscan magistral title meddiss is recorded but there is nothing else to indicate Oscan connections. The unrhotacized -s- of pomposiies, in the ‘Umbrian’ onomastic formula po. ca. pompo- siies ‘Pontus Pomposius son of Gaius’ is inconclusive in view of Um asa and esono. Reminiscent of both Oscan and Umbrian are the reduction of *-ks to -ss in meddiss and the syncopated form -iies for *-iios; cf. Pa ponties (§2.3). Umbrian monophthongization is paralleled in state ‘to Stata’ (< *-riz), and in the dialect Laplostru ‘cart’ (standard Laplaustrum); cf. Um TUTE (< *-cli), toru (< tau-, cf. La tauros). Aequian Latin has priuata, itur and dat.sg. albsi (abbrev. for albensi) p ~ t r e , ~ ~ showing both i and P from ei, as in Umbrian and in the Roman Latin of this period.

6.3 There is a predominantly Umbrian flavour to this list but, since most of the items are phonological, this does not entirely rule out the possibility of Oscan a f f i l i a t i ~ n . ~ ~

7.1 The Sabini were believed in Antiquity to be of Umbrian origin,65 but it was also from their territory in Central Italy that the Samnites were said to have begun their southward migration.66 The Samnites called their new country SAFINIM (V 149,200 G 2),(j7 which like Gk Saunitai < *saPn < *sabhn-; cf. La Samnium, Samnit2s of the same origin. Sabini is the Latinization of the Sabines’ name for themselves, *saPin6s (< *sabh-).68 The link with both Umbrians and Samnites invites the hypothesis that Sabine was an important linguistic bridge between Umbrian and Oscan, the language of the Sabelli (see n. 2).

7.2 Unfortunately the linguistic evidence is again meagre. There are no Sabine inscriptions; V 227, classed as Sabine by Vetter, is

Page 16: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 11s

probably Vestinian, V 512 Tiburtine Latin. The few pre-imperial Latin inscriptions from the region (CIL I2 385, 1847, 1848, 1861) yield very little. Thus [l] dat. feronia (< *-ui) may be non-Roman Latin rather than an intrusion from the local language (see §2.3[7]). [2] que (nom.) and nuges (abl.) show the 2 reflex of *ei, also found in Umbrian, Marsian, Volscian and Faliscan.6’ It is possible that in the last three of these the monophthong was subsequently raised, as in Latin, to [i.], but the substitution of 2 for lcertainly seems to have been a Latin dialect phenomenon in Cl.7“ [3] heicei ‘here’ < *hei-ce and suauei (< *sufiui(s)) reveal [el as an allophone of /e/ in final position, for which there is no exact parallel .71

7.3 Loanwords and glosses72 yield a few certain data. Thus 141 nerio glossed by Gellius 13.23.7 as uirtus, contains a root attested in 0 s NIIR ‘magistrate’, Um nerf (acc.) ‘leaders’ but not in Latin.

The phonological shape of some of these words is instructive. Thus [5] nouensiles ‘of the nine seats’ or ‘newly installed’ (see n. 51) turns up as Ma nouesede and La nouensides. It was regarded in Antiquity as a Sabine loanword in Latin. The Ild variation in initial as well as intervocalic position (e.g. solium, lacrurna) is generally accepted as Sabine in origin. There is no exact parallel for the change [d) > [I], but cf. Urn ZEKEF, serse < *seden~.’~ [6] La dirus ‘dreadful’ is recorded (Serv. ad Aen. 2.235) as coming from the Sabini et Umbri. The cognates Av. dva2&i, Gk deidb etc. indicate an original “dueiros, which would yield La *bir- but Um *dir-; cf. purdito < *por-duitorn ‘offered’ beside pres. purdouitu ‘he is to offer’.74

If, as seems likely, we are to infer from Var. L. 5.68 that [7] sbl ‘sun’ was of Sabine origin, then the otherwise puzzling fact that *sduel (cf. Gk dtlios etc.) 3 La *saul or * s d is accounted for, since Sabine like Umbrian, Aequian and Faliscan had 6 for original au; cf. the pronunciation plostrum for plaustrurn used by the emperor Vespasian, a Sabine from Reate (Suet. Vesp. 22.1).

Varro, who was himself probably from Reate, reports (L . 5.97) [8] Sa fedus for La haedus. The monophthongization of original ai was in this period common to Umbrian, Faliscan, Marsian and Volscian and is also found in the Latin dialects of these areas. 191 The derivation of fedus from *ghaidos is supported by Goth guits.

Page 17: The Central Italic Languages In the period of Roman expansion

116 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

PIE *gh- > h- everywhere else in Italic save Faliscan; e.g. foied ‘today’ < *ghddigd. Whereas the change ai > e is recent, the complex sequence of changes affecting the inherited voiced aspir- ates began in PIt. and the agreement of Faliscan and Sabine in this item must therefore be more ancient, whether due to an earlier period of unity or to prolonged c ~ n t i g u i t y . ~ ~

Safircus for La hircus ‘goat’ is also cited by Varro, ibid. The relation between this word and hirpus, which is variously r e p ~ r t e d ’ ~ as the Sabine or the Samnite word for ‘wolf’, is very unclear. Either both fircus and (h)irpus are Sabine, reflecting different roots, or fircus is Sabine and (h)irpus Samnite Oscan. The two words would then be either cognates, derived from PIE *gherk” - which would imply that Sabine was dissociated from the p-languages of the Osco-Umbrian group - or reflexes of different roots, whether IE or not. [ll] -irc- here, if it does reflect -erk-, recalls 0 s crniricatud, Praenestine La rnirqurios.

One final instance of Sa f - = La h-77 is fasena for harena ‘sand’ (Varro up. Vel.Long. 7.69), which also shows [12] retention of intervocalic s as in Oscan. Festus’s report that the Sabines nusum dicebant (P.Fest. 8L) must refer to a period before the change of au to 0, and if this was mid C4 or earlier, then the testimony is inconclusive, since Latin and other rhotacizing languages, includ- ing perhaps even Umbrian, could still have had -s- then.78

In P.Fest. 456L: scensas (v.1. sensas) Sabini cenas dicebant [13] the Sabine word cited looks like a garbled version of something like Um Sesna. Within Italic only Umbrian and Volscian show palatalization of the velar stop before the sequence front vowel + consonant. Finally the Roman Claudii traced their descent to a Sabine Attius Clausus, as Livy (2.16.4) apparently called him. At the date of his migration to Rome, 504 B.c., his name must have been written Klaudios, whence its Latin form; but (141 Clausus is attested in a Sabine Latin inscription of 150-50 B.C. (1nscr.It. 13.65). This palatalization of *[djJ (< [di]) recalls Paelignian, Mar- sian and the Oscan of Bantia. In contrast to [12] it does not correspond to Umbrian palatalization.

7.4 These data present a complex picture. Only [l] and [2] have any Latin connections, and these are with dialect Latin, in which

[lo]

Page 18: The Central Italic Languages In the period of Roman expansion

COLEMAN - TIIE CENTRAI, ITAI.IC LANGUAGES 117

the possibility of Sabine and other substrate or bilingual interfer- ence undermines their independent testimony. While [ 141 links Sabine with Paelignian and the Oscan of Bantia, a large number of the items dissociate Sabine from both. Of these [2] and [5]-[8] are shared with two or more of the quartet Umbrian, Faliscan, Mar- sian and Volscian. [ I31 associates Sabine exclusively with Umbrian and Volscian; [9] and perhaps [lo] with Faliscan. On the other hand [11] agrees with both Marsian and Oscan against Umbrian and Faliscan. So the picture is complex as well as fragmentary.

8.1 The Volsci occupied a large and never very unified area, and the consequent tendency to dialectal variation must have been enhanced by the numerous Latin colonies, including Cora (503 B.c.), Velitrae (494 B.C.)’~ and Norba (492 B.c.). Velitrae provides the only certainly Volscian inscription, V 222, a substantial but obscure text of early C3.’” Two further inscriptions have some- times been claimed as Volscian: V 221 from Tarracina on the Volscian coast and V 223 from Antinum. Both are rejected here (see 555.6 and 8.6), but another possible recruit to the minute corpus is proposed in $8.7. Finally there are a few Latin inscrip- tions with dialectal features that may relate to the local language: CIL I2 359,360 from C2 Norba, I’ 1531 from mid-C2 Sora, I2 151 1 from early C1 Cora. V 222 must provide the basis for any signifi- cant view of Volscian. The data that follow are all from this inscription, unless otherwise indicated.

8.2 In the lexicon the meaning and therefore the etymology of some items are still disputed. There is general agreement, how- ever, that [ l ] esaristrom ‘expiation’ contains the Etruscan root azs- widely attested in Italic (see $2.2[1]); that uesclis is cognate with Um VESKLES (abl.) ‘vessels’;81 and that sepu is abl. of the participle cognate with 0 s sipus ‘knowing’ (< *sFp-; cf. La supere). Finally medix (nom.pl.) attests the Oscan magistral title and toticu (prob- ably abl.) the Oscan and Umbrian root tout- ‘community’. These all exclude Latins2 without pointing towards a distinctively Oscan or Umbrian connection.

8.3 A number of phonological items are secure. Some of these are shared with both Oscan and Umbrian. Thus [2] the syncope of

Page 19: The Central Italic Languages In the period of Roman expansion

1 18 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

*-ilos in tafanies and of [3] *-kes in nom.pl. medix, with retention of the resulting cluster (see 82.3 init. and 121); [4] the raising of *6 in declune ‘to Declona’ and estu ‘let it be’. These are also found in other Central languages, especially Paelignian (see 82.3 init.).

A near parallel to [5] the fronting of U in bim ‘cow’ is found in Umbrian, which, though it has bum, also has sim (< *s@ < *sUm ‘pig’).

[6] The monophthongization of diphthongs is very reminiscent of Umbrian. Thus “‘ai and *ei both > e: e.g. deue ‘to the god- d e s ~ ’ ; ~ ~ cf. Urn deueia (adj.) but 0 s D E ~ V A ~ ; sepis ‘if anyone’, where se < *sei or *mi; cf. Lasei quis, Um SVEPIS, 0 s S V A ~ P ~ D , from*suai-. *ou > o in toticu ‘public’; cf. Um todcom-e (< *toutico-) but 0 s toutico. locina like iunone on CIL I2 359 (cf. loucina, iunonei on 360) seems late enough to be significant dialectally. The emperor Augustus, who was brought up near Velitrae, is reported as still using dombs for domUs (< *domous) ‘of home’ (see Suet. Aug. 1.6, 87.2).

In line once again with both Oscan and Umbrian are [7] the labial reflex of *k” in sepis and [8] intervocalic f in tafanies where *bh or *dh would be reflected by La 6 , d . However, [9] the retention of final m in statom, bim is more an Oscan characteristic, while [lo] the loss of final d after long vowels, in toticu, estu (< *-bd) andfq ia (< *-ad) ‘let him sacrifice’ recalls Umbrian; cf. Um

tation of arpatitu is uncertain, it is generally taken as an imperative with [11] ar- from ad-, which recalls Um ARPELTU (< *adpeldetGd; cf. La adpellitb) but also La arfuise and perhaps arbiter. In the obscure [21] asif it is clear that -f < *-as, a distinctively Oscan- Umbrian change, even if the morphology cannot be more closely specified.

The most striking item of phonology is [13] the palatalization apparently represented by the 3 that is transcribed as E in f a ~ i a . If this reflects *[kja-1, then parallels can be cited in Um FACU

beside FACIU ‘to make’, Bantian 0 s meddixud (< *-kjdd). If it reflects *[kia-1, then only Um FACIA and TICIT are relevant. Either way the palatalization follows a different pattern from Paelignian (Q2.3[3]). The shortage of data prevents more detailed compari- son, but it looks once again as if the phenomenon was independent in the different languages.

poplu, FACIA but 0 s dolud, FAKIIAD. Although the precise interpre-

Page 20: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 11’)

8.4 In morphology [14] rnedix reflects the nom.pl. *-es that is common to Oscan, Umbrian and the Central languages, while [lS] the infinitive ferom has the regular Oscan-Umbrian form; cf. Um AFERUM. On the otl-ier hand [16] estu ‘let it be’ agrees with 0 s escud against Um futu.

More difficult is [15] sistiatiens ‘they have set up or decreed’. This is usually taken as an inscriber’s error for sistuttens, a dental perfect like Pa coisatens, 0 s PR~FATTENS, formed from a present sistii- (cf. Gk histiimi etc.), with the reduplicative vowel general- ized to replace *sest-, ’stest- or whatever in the perfect; cf. La didici < *dedecei : *didcsc6. The confusion of ti and tt is more plausible, given that V 222 follows Latin writing conventions, since at this date geminates were not noted in Latin. But the first ti cannot be so explained, and would be easier to account for - by graphic anticipation - if the second ti were genuine. Analysis as a denominative from *stati- (cf. 0 s STAT~F) would give a nice parallel to La statuere : status; but reduplication is improbable in a denomi- native and there is no other perfect-marker in the Emen- dation to statiattens would meet both these objections but would presuppose even more drastic miscopying. It has also been sug- gested that ti could itself be a perfect formant, but this is unlikely to have been reflected by 0 s f t , since prevocalic *[ti] > *[tj] > [ttj] generally in Oscan; cf. MAMERTTIA~S.’~ We are left then with the assumption of a double inscriber’s error as the least unsatisfactory analysis.X6

8.5 Many of the items listed agree with both Oscan and Umbrian: [l] in lexicon, [2]-[4] in phonology, [7], [8], [12], [14] and [lS] in morphology. Agreement specifically with Umbrian is shown in [ 5 ] , [ 6 ] , [lo], [ll] and especially [13], all phonological and together with the ‘Umbrian’ onomastic order8’ responsible for the distinctly Umbrian appearance of V 222.88 On the other hand item [9] and the morphological [16] and [17] agree with Oscan and are important enough just to tip the balance in favour of Oscan affinity. But the closeness of the verdict makes Volscian a typical member of the Central group.

8.6 Turning now to the disputed text V 221: the gentile name Cloif(is) shows retention of the diphthong as in Oscan. Although it is possible that oi survived in Volscian of this period, the general

Page 21: The Central Italic Languages In the period of Roman expansion

120 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

picture of monophthongization revealed by [6] above makes this unlikely and the use of the ‘Oscan’ onomastic formula here, in contrast to V 222, favours an Oscan attribution.8’

8.7 The Lapis Satricanus, found in the 1977 excavations on the site of the temple of Mater Matuta at Satricum,’” contains an inscription dating from 525-475 B.C. which runs Iteisteterai pop[io- sioualesiolsiosuodales mamertei. This has generally been taken as Latin but the attribution is improbable. Satricum itself lies between Velitrae to the North and Antium to the West, both of them for long periods Volscian towns. It seems to have changed hands between Volscians, Latins and Romans many times before it was finally destroyed by the Romans in 347 B . C . ~ ’

The first eleven letters have been variously divided.’* For instance aid]e iste terai ‘a temple (acc.) in or for that land’; en aid]e iste terai ‘in the temple here of Terra’; eiste terai ‘this (neut.acc.) in the land’. None of these is acceptable. en aid] requires more space than appears to be available. Moreover in Latin of this date we should not expect (i) the loss of -m in aidem or of -d in *eistid, assuming that a cognate of Um este ever existed at all in Latin alongside “istod; (ii) tera- rather than *tersa- (> La terra-, 0 s tsra-); (iii) the gen.sg. -as already replaced by -at especially in a religious dedication; (iv) monophthongization in *istei or *istai. It seems better therefore in spite of the unusual sentence-initial position to take steterai as a verb ‘they have set up’ preceded by h]ei ‘in this place’ or something of the sort.y3 The form cannot of course be reflected in La stetsre, which perhaps < -2ri (cf. dederi on I2 37), but could itself be in origin a middle *-2rai (cf. Ved. tasthire beside act. tasthzir), as 1 sg. steti is, at least o n the most likely hypothesis (< *stestai; cf. Ved. tastht). There is of course no parallel 3 pl. formation anywhere else in Italic.

Even more unLatin than steterai is popliosio ualesiosio ‘of Pub- lius Valerius’, a form of the thematic gen.sg. that is reflected in Gk -oio, Ved. -asya and most closely in Fa kaisiosio ‘of Caesius’. Reflexes of -osio have sometimes been sought in the La pronomi- nal US.^^ But *kwmio (cf. Ved. krisya) +s would normally have given La *curius, not cui(i)us, since dissimilatory loss of the first s, though possible, is somewhat ad hoc, and the change *-os!:os >

Page 22: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 121

*-oiLos cannot be justified by *magicis > mai(i)or, *pedi6s > pei(i)or. In fact -ius can be connected with -osio only on the assumption of drastic and unparalleled reshaping.9s

It is remarkable that within a radius of 100 km. from Rome no less than three separate forms of the thematic gen. sg. are attested: -osio in Faliscan; -t in Latin Barbati, etc., for which cf. Gaul Segomari, and Ven louki (beside uoltiio < *-esto); and the bor- rowed i-stem -eis in 0 s GAAVIE~S ‘of Gaius’, Um uoiszener ‘of Volsienus’. Now the thematic declension was relatively conserva- tive in Italic: it alone retained a distinct voc.sg. and provided the base for the retention of a distinct abl.sg. in other noun paradigms, as well as exporting its own acc.sg. -om to consonant stems in Oscan and Umbrian. That there should be such divergence in one case alone is therefore all the more surprising. We can only surmise that both -i and -mi0 reflect PIt forms distributed, say, between nouns and pronouns, rather as La -i and -ius came to be. ”

But even if we can believe that Latin of around 500 B.C. had a nominal gen. -osio, which disappeared without traceq7 within the next 250 years, whence came the -5 that replaced it? Certainly not from the pronouns, which are unlikely to have had -I at this date and then replaced it by -ius, especially if this was remodelled from -osio itself. Moreover the reason for such a substitution is hard to

The only known language in which popliosio ualesiosio would be at home is Faliscan.

suodales ‘companions’ by contrast could well be Latin, the earlier form of La sodales in fact, whether we derive it from *swedh- (cf. Ved. svadhu, Gk Cthos) or *swed- (cf. Boeot. Gk whedibtus) .99 The corresponding forms in Oscan and Umbrian would have been respectively *suefal- or *sued& and *sofar- or *sodal-.

Mamerfeei ‘to Mars’ has exact parallels in 0 s pap&et&i (Po 177)’00 and Mamers, reported by Festus 150L, and in Sa Marners (Var.L. 5.73). La Mamor, which occurs with the variants Marma, Marmor, Marmar on the hopelessly corrupt Carmen Arvale (I2 2 ) , is less likely to reflect Mumer than *Mauor or some such archaic vocative of Mauors.

It is not easy to see how an inscription consisting of one obscure

Page 23: The Central Italic Languages In the period of Roman expansion

122 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

word whose morphology is apparently unparalleled and three others that would have been at home respectively in Faliscan, Latin and Oscan adds up to a Latin text of around 500 B.C. or any other period. A contemporary Latin version would have looked more like hei steteri poplii ualesii suedales mauortei. There is, however, nothing incompatible with a Volscian attribution: ei could well have been monophthongized later, the form of suodales, if it really is Latin and not Volscian, can be accounted for if Poplios Valesios was a Roman visitor,"" and mid. steterai and act. sistattens could well be from doublet forms of *st&, whether or not they were distinguished semantically (cf. the Latin doublets sistere and trans. -stare and the variation between active and middle in La suspicere, suspicare, suspicari). Since the inscription is not in Latin, there is a strong geographical probability that it is Volscian. If so, then the complexity of Volscian's relationships is further enhanced by the -osio gen.sg. and its quintessentially Central status (Q8.6) confirmed.

9.1 Even on the very limited evidence that we have surveyed here it is clear that the seven languages do not form anything like a monolithic block. '02 There is in the first place the broad division into Oscanic and Umbrianic. Defined in purely synchronic terms as sharing certain of their distinctive characteristics with Oscan and Umbrian respectively, these labels are capable of two separate but not mutually exclusive diachronic interpretations. They may imply that the languages so classified derive from dialects of Oscan or Umbrian that have become detached from their original lan- guages and de-/eloped independently. Alternatively their shared characteristics may result from common substrate - as, say, between Umbrian and Umbrianic - or prolonged bilingual con- tacts or population mixtures: all of them realistic possibilities in a relatively small geographical area with considerable population movements in the period 800-400 B . c . , followed by prolonged and relatively stable settlement. Given these last three possibilities, we may also envisage an earlier single Central Italic language, derived from a third dialect in proto-Oscan-Umbrian and therefore sharing characteristics from the start with both Oscan and Umbrian. This language would be reflected to a greater or lesser extent in all seven of the extant central languages.

Page 24: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 123

9.2 Paelignian (52.5) is clearly the most Oscanic of the group. I t is, however, distinct from the neighbouring dialects of Oscan attested among the Frentani and the Samnites. If it is not by origin a Central language that has undergone Oscan influence, then it must be derived from an Oscan dialect that became detached from the language and thereafter developed independently over a rela- tively long period. ' 03

Marrucinian ($3.5) , though still within the Oscanic half of the spectrum, shows more affinities than Paelignian with Umbrian and the Umbrianic languages of the central group. As such it repre- sents descriptively something approaching the norm for the group as a whole, but the characteristics that it shares specifically with Paelignian point to a common origin or prolonged bilingual contact.

Volscian ($23.5) also belongs in the Oscanic half on morphologi- cal grounds - most notably the presence of the dental perfect, which it shares with Paelignian and Marrucinian. It may be derived from an Oscan dialect that has separated from the rest of Oscan at an early date and been subject thereafter to Umbrianic influences or it may with Marrucinian represent the last traces of a third linguistic group that was from the start distinct from both Oscan and Umbrian. If the Volscian attribution of the Lapis Satricanus is correct, then Faliscan as well as Umbrian connections must also come into the reckoning."'4

Starting from the Umbrian end: Sabine (67.4) shares much more of its phonology with Umbrian than with Oscan, and a number of these items are also found in Marsian and Volscian within the Central group and Faliscan outside it. The loose unity of these four languages, based as it is on phonological criteria alone, may be due merely to geographical contiguity. The absence of morphological evidence of the kind that enabled us to separate Volscian from the Umbrianic subgroup counsels caution. Hut at least we can say on the evidence available that Sabine is not to be placed in some 'Latinian' subgroup but towards the Umbrian end of the Central spectrum.

If Aequian ($6.3) was originally a dialect of Oscan, it must have been subject to considerable Umbrianic influences. It is better therefore to place it within the actual Llmbrianic group from the start. But this and the alternative hypothesis that Aequian

Page 25: The Central Italic Languages In the period of Roman expansion

124 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

together with Volscian, Marsian and Marrucinian“” derives from a third dialect of proto-Oscan-Umbrian distinct from both Oscan and Umbrian can only be tested if more substantial material including some criteria1 morphology is found. In any event the ‘Latinian’ label is again unjustified.

Similarly with Marsian (45.5): conclusions based on phonology must be tentative but they point not to a ‘Latinian’ but to an Umbrianic affinity.

Finally there is Vestinian ( $ 4 . 9 , which on the phonological and very slender morphological evidence available appears to be within the Umbrianic half of the spectrum. The alternative hypothesis is that it is derived from an Oscan dialect that has undergone Umbrianic influences. The parallels with the non-con- tiguous Marsian may well be relevant to the earlier history of the two languages.

9.3 These c!assifications differ in a number of ways from those set out in Fig. 2 (41.2). Sabine, Aequian and Marsian are still grouped together but Vestinian is now added to them to form an Umbrianic quartet. Volscian joins Paelignian and Marrucinian in the Oscanic subgroup. These conclusions are summarized in Fig. 3. The relative geographical positions of the languages are indicated by common boundary lines: thus Marsian is adjacent to Volscian,

Figure 3

Page 26: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 125

Aequian and Paelignian, etc. The spectrum of affinity runs from the top to the bottom, with the thick line marking - more emphati- cally no doubt than the evidence warrants - the division between the Umbrianic and Oscanic subgroups. The vertical ordering of the languages indicates the spread from most Umbrianic at the top to most Oscanic at the bottom: viz. Sabine; Aequian, Marsian and Vestinian; Volscian and Marrucinian; Paelignian.

The tentative nature of the conclusions must be stressed. The picture could be changed quite radically by any one of three events: (i) a revaluation of the specific data from which the affini- ties are inferred; (ii) new and generally accepted insights on the numerous controversial items that have been deliberately omitted from consideration here; (iii) the discovery of a single substantial inscription or a handful of small ones containing some criteria1 morphology from any of the seven languages but especially from Sabine, Marsian and Volscian. An advance in any one of these areas might help us to resolve the most important question of all: whether, as implied in Fig. 3 , the Central group really does divide more or less neatly into two or we should instead posit a third language over against Oscan and Umbrian but sharing character- istics with both, out of which eventually came a central core comprising, say, Volscian, Marsian, Vestinian and Marrucinian, with the rest of the seven in various relationships with this core and with Oscan and Umbrian. What is most unlikely is that new information would actually reduce the diversification that we can already see among the Central Iralic languages.

Emmanuel College, Cambridge

NOTES 1. The earliest colonies outside Latium (Velitrae and Norba in Volscian terri-

tory) date from the early C5 and Rome controlled the whole of Italy south of thc Arno by the mid-C3. The epigraphic data on which this paper is based belong with few exceptions to the period 3Os-50 u.c..

2. Sabdli, etymologically connected with Sahini (see P7.1), was used in Anti- quity of Oscan-speakers gcnerally. The first recurded occurrence of the name is relatively late: Varro ap. Philarg. ad. Vg. Aen. 2.167. See Dcvoto 103-4, Durante 817.

Page 27: The Central Italic Languages In the period of Roman expansion

126 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

3. The Hernici have left nothing and are assumed to have been illiterate until their language was replaced by Latin.

4. cf. Conway 1897.l.xvi-xviii and 233-369; Vetter 140kh4; Poultney 1959.9; Maniet 524; Durante 792-813.

5. Buck 3; Pisani 112-25. Pulgram 134 similarly leaves the group undivided, cautiously opining that they are ‘perhaps . . . more closely related to Oscan than to Umbrian’, but with an important proviso (see n. 38).

6 . The status of Praenestine within the range of Latin dialectal differentiation will be treated in a subsequent paper elsewhere.

7. Liv. 9.45, 10.3; Diod. 20.101. For their origins see Devoto 110. 8. At Alba Fucens in 304 B . c . , Carseoli in 302 or 298 (Liv. 10.1.2, 3.3). 9. The linguistic relevance of the repatriation of Paeligni in 177 B.C. from the

Latin colony at Fregellae in Volscian territory (Liv. 41.8.8, 8.12, 9.9) is impossible to assess.

10. References to Poccetti’s collection (Po) are always indicated, those to Vetter (V) only where specification is important.

11. V 202 and 203 survive only in MS. copies. 12. The principal items discussed in each language are enumerated in square

brackets. 13. Var. L.7.29, P.Fest. 41 L; P.Fest. 110 L, Liv. 23.35.13, the title occurring

also in Ae and Vo; Fest. 372 L, 384 L. 14. P.Fest. 77 L. fumulus is re-formed fromfumul (Enn. A . 313). Um F A M E ~ I A S

(= La familiue nom.pl.) suggests that the absence of fume/ from Urn may he accidental.

15. As also Aequian and probably Marsian (516.2, 5 . 6 [ 3 ] ) . 16. On V 153 (Bovianum Vetus) T ~ V [ T ~ K ] S or T6V[Tk]S is possible. 17. It is hard to make anything of the apparent contrast between cerfum (V 213),

with which cf. Um FERFE (see Poultney 1959.277). and cerriu (V 206), with which cf. 0 s K ~ K R ~ , always assuming that they both directly reflect a form with *-rs-.

18. From just outside Italic there is Venetic IIUVANTSAI beside IUVANTIIOI (Lejeune 1974 nos. 58 and 9, etc.).

19. Which may or may not include medial [j] > [dg], depending upon the analysis of af ded. See Maniet 558.

20. Pisani 1954.117-19. 21. Porzio-Gernia 153-5. 22. Assuming that diues < *deivet-. The supposed etymological connection with

diuus (Var. L.5.92) does not take us far. 23. Umbrian unlike Oscan and Latin kept the two stems distinct: cf. FRATRUS,

sacris with 0 s TEREMN~SS, SAKRISS and La fratribus, sacribus (beside o-stem sacris). 24. Vetter 149. .25. The relation of the theonym to *a , leudhero- ‘free’, *lubh- ‘desire’ and */eib-

‘pour’ is notoriously obscure, as is the derivation of La liber ‘free’. Neither leiberei (CIL I* 614), where the diphthong is probably authentic, nor loebesum (P. Fest. 108 L), where it probably is not, provides a normal reflex of p.It. *leu&-, for which cf. loufir below.

26. Pisani 1964.116. 27. The vocalism of 0 s NFSSIMAS, Um nesimei, though relevant, is not corro-

28. No need therefore to follow Poccetti (Po 208) in taking Gruex as a Latin borative.

import.

Page 28: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES 127

29. For the view that this is the true Pa form and sacuracirix a false archaism see Lazzeroni 1976.39 1-2.

30. La examples include pdculurn < *potlorn, gubernaculurn < *-&/om. La *sacratlorn > *sacruculum, not sacellurn (< *sacra-lorn the diminutive), with which the 0 s word is often equated.

31. O n which see Lazzeroni 1964. 32. For Marrucinian see 93.3[5], for Vestinian n. 49.The evidence in both is

tenuous. 33. The change *rl :, kl reflected in Um P I H A K L U , La piaculurn is not relevant

since the motivation here is the velar articulation of 1. It is not certain that Oscan shared in this change: the k in SAKARAKLUM might just he due to the preceding k and PE.:ESSL.I~M (Aufidena) appears to be derived directly from PESTLUM (Bovianum Vetus) without an intervening stage *-skl-; contrast Um pesclrr, perscler, if these reflect *perk-sdo, not ‘perksk-lo-. For tr cf. 0 s FKAPKUM, Urn FKATKU. etc.

34. For derivation from gen. *eis6rn-ke see Vetter 154; from abl. *ekscid-ke Pisani 1964.120,

35. Thc possibility olgraphic confusion between I1 = e and I = i must always be reckoned with at any rate on V 218. regenfail, usually taken to mean ‘for the queen’, could then represent riginai (i *rc?gindi). But the meaning is uncertain, and the word could be a gerundive (< “regendui), with perfectly normal vocalism.

36. This word must have had wider currency, since it is recorded for Latin: mignae xeia pEetl;opEva ‘portions of meat’ (CGL 2.24.6).

37. There is no certain evidence for or against intervocalic rhotacism. The derivation of esuc is uncertain (see [ I ] ) ; asurn < *assom ’a coin’ or *arsurn ‘to burn‘ (supine); the Umbrian cognates of aisos have -s-; e.g. ESUNU ‘divine‘ (fem.nom.1, esono ‘rite‘ (neut.nom.).

38. Pulgram 134 and 147 notes this important exception to his cautious generali- zation on the Central group (see n. 5 ) .

39. For the change * U > i in Umbrian see Poultney 1959.37. It is not clear that 0 s castrid is relevant; see Buck 41.

40. Although perronr with -6-not-u- may he a Latinism. and perhaps even fec also (Po 206; cf.Os fefirc-), it seems unlikely that the Latin form of a distinctly ‘Oscan’ official title would have been used in this region.

41. Assuming that aims pacris is not dative: < *aisois *pacrifs. If (oil > (0.1. a connection with Oscan i s possible since 0 s ui could represent a midway stage [oil; but Umbrian is excluded, since e - the Urn reflex of (oil. cannot be derived by way of [Q.] (cf. 84.4[6]).

Vetter to the Sabini. Pisani (1964.121) is non-committal. For V 221 see 98.6. 42. From Scoppito: assigned by Conway (1.2.59) to the Vestini, by Buck and

43. aidiles. with d not f f rom *dh, is Latin hut the rest of the text shows no sign of La influence; cf. 0 s K v a i s s r u K , Um KvEsrRETiE in contexts that are otherwise unLatinized.

44. The final of uerio ‘Vettius’ (V 220) is too uncertain graphically to provide a counterargument.

45. Pisani (1964.121) interprets as perfect with di- for de- from the present; but we should then expect either *dided or *dide.

46. CII. I* 756. The most plausible analysis (Ribezzo 1923) gives a cognate to La *fid&litares ‘those pledged’. Other probable Vestinian influences can be seen in the inscription; e.g. a/& for alius,flusare (cf. [ I ] below) and perhaps the uncertainty in distinguishing qu and c in hoiusque.

Page 29: The Central Italic Languages In the period of Roman expansion

128 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

47. cf. Pa aetate in 12.4[9]. 48. If the phrase were acc., which is less likely, then an Oscan connection would

be excluded (cf. 0 s *brafiss * d a m s ) ; but not necessarily an Umbrian one; cf. Um *brat i ( f ) *data( f ) .

49. Although dat.sg. -u is widespread in dialect Latin, Feronia in CIL I’ 1848 (Aquila) could perhaps reflect the native Vestinian form. For the Paelignian evidence see 62.3[7].

50. The different reflexes of *oi exhibited in pople and KUKATU. parallel to La populi (nom.pl. < e < -ei < -oi) and curaturn, correlate with original word-accent position.

51. < *noyen-sed- referring to the augural fernpla. See Letta-D’Amato 45-6. The alternative analysis ‘noy-ensed- ‘newly installed’, while it suits the Latin contrast between di indigifes ’native gods’ and di nouensides, is harder to accept in a Sabine or Marsian context.

52. Allen 71-3 interprets the tall i as [ I ] , but this makes the Romance develop- ments (see n. 53) harder to explain.

53. There is no need to assume palatalization of g here or in seino. much less palatalization of n in the latter, though this is a possibility in dialectal La sinnu, given the subsequent general shift of La [ignu] to [igu] reflected in It segno, Spseria, etc. See Poultney 65, Poccetti 165-6, Marchese 215.

54. esos does not exclude intervocalic rhotacism, since Umbrian, which is not- ably rhotacizing, has ESUNU, etc.

55. See Conway 1.269, Kerlouegan, Pisani 124, Radke 797f., Letta 1972.33-4. 56. Kerlouegan 288 notes that Antino still belongs to the Marsica, the border of

which is in fact half way between Antino and Sora. 57. If pe.ui.p[] = ‘Pettius P[ ] son of Vibius’. 58. Marsian La inscriptions have dat.sg. -e , as in Vo, Ae, Um, and -a, as in Pa;

e.g. angitie (Po 221) but actia (V 228a, if this is the right reading); hence it is impossible to decide whether *-ui > Ma -a or > (*-ai >) -e .

59. Marsian La inscriptions all have o: dono, Aplone (= Roman donurn, Apol- lini) - except 1’ 392 (Ortona), which has uesune.

60. The derivation from *-sedois (Ribezzo 1930.77) is implausible in the absence of other Italic evidence for a thematic derivative from the verbal root sed-.

61. Livy 9.45, 10.19. 62. Conway 2.532 and earlier editors were sceptical about its authenticity. 63. For the variation within a concogdant phrase cf. iunone seispitei rnatri (CIL I*

1430) from Lanuvium in Latium. 64. Ogilvie 130 regarded the Aequi as a branch of the Oscans but without

referring to linguistic evidence for the assertion. 65. Dion.Hal. A.R. 2.49, citing Zenodotus. 66. Var. L. 7.28, 29, Gell. 11.1.5, Strabo 5.4.12. 67. cf. S A F I N ~ ~ S etc. on three CS inscriptions from South Picenum; see Marinetti

1985. 68. It is assumed that Sabine had the fricative reflex of -hh- rather than, as in

Latin, the occlusive one; but the evidence is admittedly not substantial, e.g. the equation of the Sabine river name Farfarus with La Fabaris (Serv ad A m . 7.715).

69. cf. La priui, hic, diui with Urn PKEVE, Fa hec, Vo deue. 70. See Cic. de Or . 3.48; Var. R. 1.2.14, 48.2, citing ueha, uella, speca for uia,

uilla, spica.

Page 30: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC 1,ANGUAGES 129

71. L:t here, sihe, qume for heri, sihi, quasi (Quint. 1.4.8 and 7.22) show a reverse phenomenon: li] > [el.

72. For detailed lists see Vetter 362-78 and Bruno. Many of the words, e.g. cafus and crepusculurn, contribute nothing to the present investigation.

73. For nouensiles itself see the ancient tcstimonia cited by Conway 1.357. On dil doublets in Latin sec Leumann 128.

74. Not all the 0 s and Um examples cited are conclusive. Urn difue for * h i - bhuiom ‘folded double’ could he due to dissimilatory loss of u . Similarly 0 s AKKATUS for *ad-uok&os ‘advocates’ could he due either to loss of u followed by syncope of o or to a succession of syncopes: *-duok- > *-duk- > *-dk.

75. Fa he ‘here’ ( V 332) beside J? (V 339a), if it is not a Latinism. must be a dialectal variant - the sites are 8 or Y km apart ~ and this would he relevant to the contiguity explanation.

76. See respectively Serv. ud A m . 11.785; P.Fest. 93L; Irpini. 77. It is difficult, as usual. to know who precisely the anriqui w’ere in Festus’s

fueduni untiqui dicehunt pro huedo, folus pro holere. fosrim pro hoste, forlium pro hosria (74L). The etymology o f liostiu is unknown; infolus andfostis initialf- < PIE *gh-. as in fordrum for hordeurn, also attributed to nritiqui (by Quintilian. 1.4.14). A Sabine origin for these f-doublets seems probable.

78. Ernout 73 1. claimed Sabine as the source for at least some Latin words that preserved intervocalic .s; c,g. nasus.

79. Both said to have been Latin towns at an earlier date: Plin. N a f . 3.63, Catofr 58P and Dio 45.1, Dion. Hal. A . R. 6.42, Liv. 2.30.14.

80. For the fullest discussions since Vetter sce Untermann 1956 and Radke 779-96.

81. cf. Urn VESTIKAII! ‘let him pour a libation’. The vocalism resists connection with La uasculurri ‘vessel’.

82. couehriu may reflect *ko-uir-io- ‘gathering of men’, parallel to La cciriu < *ko-uir-iri-; but derivation from *ko-ueghe-sio- ‘bringing together’ is phonolo- gically easier.

83. Which indicates that the La dat.sg. loucina on CIL I’ 360 (Norba) cannot be due to Vo influence.

84. So Wallace 96 against Radke. 85. See Untermann 125, whose preferred derivation of 0 s [tt] from *[tw]

encounters the same difficulty; cf. D F K R V I A K ~ M < *dekuiariorn. It is moreover unlikely that a finite tense formant could have been created from a verbal noun suffix like -fi- or -at-, since even in periphrastic use with auxiliaries such nouns would hardly have acquired specifically perfective aspect.

86. The relcvance of the obscure sesf.a.plens ‘set up’ from a lost Parlignian inscription V 202 is uncertain.

87. mu.ca.faJunie.~ ‘Maracus Tafanius son of Ciaius’. 88. Hence Poultney 113-14 scts Volscian apart from the other Central languages

as being more akin to Umbrian than to Oscan. 89. Vctter inexplicably assigns it to Vestinian., Conway 1.269 more plausibly t o

Volscian. With the ‘Oscan’ formula sfuris cloil .~. ‘Statius Cloelius son of C.‘ cf. thc ‘Umbrian’ one in n. 87.

90. For a detailed analysis of the text see De Simone. 91. Liv. 7.27.5-9. Diod. 14.102.4. See Ogilvie 332. 92. See, beside D e S h o n e , Peruzzi 348, Prosdocimi 189, etc.

Page 31: The Central Italic Languages In the period of Roman expansion

130 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 1986

93. cf. De Simone 74 f . 94. See Leumann 289, Giacomelli 142 f. 95. The derivation -osio > *-oiio > *-oii > -i (Pisani 1964.3445) is of course

preposterous. 96. The possibility of a similar distribution in Faliscan is doubtful so long as the

existence of Fa (as distinct from Faliscan La) -iis questionable. This objection also applies to the argument (Untermann 178 f., De Simone 83), that Faliscan had gen. -i, possessive -osio, which is further weakened by the absence of any relevant parallel for such a functional distinction. For further discussion see Bonfante.

97. Unless it is reflected in Mettoeo Fufetioeo or whatever the correct restoration is of the corrupt Ennian phrase cited at Quint. 1.5.12. It might not have been beyond the experimental audacity of Ennius to import something like Mettosto- que Fufetiosio or an Homerized version of it, Mettoioque Fufetioio, to characte- rize the dictator of Alba, whose name anyway is distinctly unLatin; cf. *Fudetius or ‘Hubetius.

98. The suggestion (Peruzzi 346) that after rhotacism *-orio would be insuffi- ciently distinct in ordinary speech from adjectival -6rios is unconvincing in itself and entails an improbably recent date for the replacement.

99. The first, accepted by Walde-Hofmann, is doubted by Mayrhofer (s.v. .rvadha); the second is favoured by Benveniste 1.328 f.

100. From Rossano di Vaglio; see Lejeune 1972.403. MAMERTTIA~S was already attested at Capua (V 84) etc.

101. Most historians seem content to identify him with P.Valerius Publicola, suffect consul in 509 B . C .

102. On the diversified ‘continuum’ represented by the seven see Wallace 99. 103. cf. Von Planta 1.20, Lazzeroni 1976.389. The latter’s earlier suggestion

(1964.73 f . , followed at least in general terms by Campanile 10.5 f., 117), that apart from the ‘archaizing’ V 213 Paelignian was Latinized at an early date, is not convincing. Of the data cited both palatalization and anaptyxis are found outside V 213 (see S2.3[3] and [6]), there are no a-stem datsg. forms in V 213 to set beside -u elsewhere (02.3[7]), and cerfum and uetute are inconclusive (see n. 17 and S2.4[9]).

104. The status of Faliscan vis-d-vis Latin, Umbrian and the Central languages will be the subject of a later paper elsewhere.

105. For the grouping of Aequi, Marsi and Volsci see Letta 1972.35.

REFERENCES (Abbreviations as in L’Annte Philologique)

ALLEN, W. S . , 197g2. Vox Lurina: A Guide to the Pronunciation of Classical Latin,

BENVENISTE, E.. 1969. Le Vocubuluire des Institutions Indo-Europkenes, Paris. BONFANTE, G. , 1978. ‘La nuova iscrizione di Satricum e il genitivo in -osio’, R A L

BRUNO, M. G., 1961-1962. ‘ I Sabini e la loro lingua’, RIL 95,501-44; 96, 413-42,

BUCK, C. D . , 1929‘. A Grammar of Oscan and Umbrian, Boston. CAMPANILE, E . 1978. ‘La diaspora italica’, in La Culturu Iralica (= Orientamenti

Cambridge.

33, 269-72.

565-640.

Linguistici 5 ) , Pisa.

Page 32: The Central Italic Languages In the period of Roman expansion

COLEMAN - THE CENTRAL ITALIC LANGUAGES

CONWAY, R. S., 1897. The Italic Dialects. Cambridge. DE S i M o N E , C.. 1980. ‘L’aspetto linguistico’, in C. M. Stibbe and others, Lapis

DEVOTO, G.. 1967’. Gli Antichi Italici, Firenze. DIIRANTE., M . , 1978. ‘I Dialetti Medio-Italici’, in A. L. Prosdocimi (ed.), Lingue e

Dialetti (= Popoli e Civilta dell’Italia Antica, vol. 6), Rome, 789-824. ERNOUT, A , , 1909. Les Eliments Dialectaux du Vocabulaire Latin, Paris. GIACOMELLI. G., 1963. La Lingua Falisca, Firenze. K ~ R L O U ~ G A N , F., 1958. ‘Le parler d’Antinum: Marse ou Volsque?’ REA 60.280-9. LAZ.ZERONI, R., 1964. ‘11 dativo “sabellico” in -a’, SSL 4, 65-86. LAZZERONI, R. , 1976. ‘Differenze linguistiche nel territorio dell’Abruzzo e del

Molise in epoca Italica’, in Scritti in onore di Giuliano Bonfante, Brescia.

LPJEUNE. M., 1972. ‘Inscriptions de Rossano di Vaglio 1972’, R A L 8.27, 399-414. LEJEUNE. M. , 1974. Manuel de la Langue Vbnete. Heidelberg. LF.TTA, C . , 1972. I Mursi e il Fucino nell’Antichita. Milano. LEVA, C. , 1976. ‘Dialetti italici minori: Marso’, SE 44, 275-81. LETTA, C . and D’AMAI-0, S., 1975. Epigrafia della Regione dei Marsi, Milano. LEUMANN, M., 1963. Lateinische Laut- und Formenlehre (= Leumann-Hofmann-

MANIET, A, , 1972. ‘La linguistique italique’, in Aufstieg und Niedergang der

MARCHESE, M. P., 1978. ‘Marso seino = latino signum’, SE 46, 213-21. MARINETTI, A., 1985. 1,e iscrizioni sudpicene, I : Testi, Florence. OGILVIE, R. M.. 1970’. A Commentary on Livy Books I-5, Oxford. PERUZZI, E., 1978. ‘On the Satricum inscription’, PP 33, 34650. PISANI, V., 1954. ‘Palatalizzazioni osche et latine’, AGI 39, 112-19. PISANI, V., 1064. Le Lingue dell’ltalia Antica olrre il Latino (= Manuale Storico

P o c c ~ ~ ~ r , P., 1979. Nuovi Documenti Italici (= Orientamenti Linguistici 8 ) , Pisa. PORZIO ( ~ E R N I A , MARIA L., 1974. ‘Contributi metodologici alio studio del latino

POULTNFY, J . W., 1951. ‘Volscians and Umbrians’, AJP 72, 11-3-27. P O U L T N ~ Y , J. W . , 1959. The Bronze Tables of Iguvium, Baltimore. PROSDOCIMI, A. L., 1979. ‘Studi sul Latino arcaico’, SE 47, 173-221. PULGKAM, E. , 1978. Itulic, Latin, Italian, Heidelberg. RADKE, G . , 1961. ‘Volsci’, Real-Encyclopadie 9A1, 773-827. RIBEZZO. F., 1923. ‘FIFELTARES’, RICZ 7, 180. RIBEZZO, F.. 1930. ‘Roma delle origini, Sabini e Sabelli’, RIG1 14, 59-99. UNTERMANN, J . , 1956. ‘Die Bronzetafel von Velletri’, I F 62, 12S35.

VETEK, E., 1953. Handbuch der italischen Dialekte, Heidelberg. VON PLANTA. R. , 1892-97. Grammatik der oskisch-umhrischen Dialekte,

131

Satricanus, ’s-Graven hage, 7 1-94,

1.389-99.

Szantyr. Lateinische Grammatik I) , Miinchen.

rdmischen Welt (ed. I H . Temporini), 1.2, 522-92.

della Lingua Latina vol. 4), Torino.

arcaico. La sorte di M e D finale’, M A L 17.4.

U N I t K M A N N , J . , rev. GIACOMELLI, 1964. La Lingua Falisca, GCA 216, 171-82.

Strasshurg. WALLACE, R . , 1985. ‘Volscian sistiatiens’, G1 63. 93-101.