Bryn Mawr Classical Review

M. H. Crawford, Imagines Italicae: A Corpus of Italic Inscriptions (3 vols.). Bulletin of the Institute of Classical Studies supplement 110. London: Institute of Classical Studies University of London, 2011. Pp. xxii, 610; x, 688; x, 344. ISBN 9781905670307; 9781905670352; 9781905670369. £276.00.

Other editors: W. M. Broadhead, J. P. T. Clackson, F. Santangelo, S. Thompson, M. Watmough and computing by E. Bissa and G. Bodard.

Reviewed by Benjamin W. Fortson IV and Michael Weiss, University of Michigan; Cornell University (fortsonb@umich.edu; mlw36@cornell.edu)

Version at BMCR home site

This is a monumentally important work for Italic epigraphy and linguistics. 'Italicae' in the title means 'Sabellic'; all known inscriptions in Sabellic languages are included with the understandable exception of the Umbrian Iguvine Tables. The value of the book resides primarily in the many new readings based on autopsies (for each of which the date(s) and relevant editor(s) are given), and the (usually) clear and abundant photographs (with more to be placed on the Web). A strong additional focus is on providing detailed archaeological information and on correcting countless erroneous previous accounts of the locations and circumstances of the finds. (An especially nice example of this is the discussion of Teate Marrucinorum 2 pp. 229f.; at times, though, these efforts can feel nitpicky.) Their presentation of each inscription includes the following rubrics: description of the object; date; last known location/autopsy; discovery, archaeological context, and later history of the object; epigraphical account; text and apparatus; translation; and bibliography. Of course, not all of this information is available for every text.

The editors' overarching interest in the material, as stated on p. 5, is not just in the languages but in the communities. This leads them to include, for example, Greek and Latin coin legends from Oscan-speaking areas, and a Latin inscription like Trebia 1, where plaga refers to a local Umbrian political entity (cf. Livy 9.41). This interest does not extend to sociolinguistic investigations of the texts (p. 3: "the alleged insights […] seem to us pretty banal") but does include the alphabets and their histories (discussed in some detail in the Introduction).

The emphasis on the objects, their communities and alphabets, and their archaeological setting is reflected in the one major drawback of the work, especially for linguists: it is organized purely geographically, and only the alphabet of each inscription is regularly identified by name—the language is frequently left unidentified. To be sure, we cannot always name the language with certainty (as e.g. for Capua 35, as they discuss in detail, p. 448); but these features make it very difficult to pull together all the texts in a particular language, for which Helmut Rix's Sabellische Texte (ST) is still necessary. Texts in different languages and alphabets rub shoulders with one another in the same section (South Picene, for example, is not differentiated from later forms of Sabellic in the same region). Also inconvenient to use are the indices at the end, which separate out items by semantic categories and, most irritatingly for a nearly 1700-page work, lack page numbers (only the inscription is referred to, labeled according to geographical location, and to find it one has to search through the [not alphabetically arranged] table of contents for the relevant geographical area). The absence of a unified word index—not unusual for epigraphical publications— only adds to the cumbersomeness.

The book is in general well-produced, though the typography is unsophisticated, and the practice of having each inscription begin on a new page results in a lot of white space. Happily, the editing was careful and thorough, as is clear from the impressively low number of typos.

We devote the rest of this review to a selection of details, focusing on interesting new findings as well as interpretations that we find dubious. The usual disclaimer applies: the carping should not detract from the work's great value and the critically important foundation it lays for future research on these texts. For ease of reference, ST numbers are also supplied.

Volume 1

p. 85, Umbria 1 = Um 32–3: The inscription tutas 'of the people' is dated to 300 BCE and thus gives us a firm terminus ante quem for the Umbrian monophthongization of the diphthong *ou.

p. 90, Umbria 3 = Um 38: A quite late inscription, dated 50 BCE, and found in Hungary. Froehner's Hongrie is not to be emended to Ombrie contra a suggestion of Lejeune. For inscription B uaria perhaps read uarea, i.e. the scribe omitted a stroke from || (correctly written on side A).

pp. 96–7, Tadinum 2 = Um 7: This 4th-century inscription from Tadinum in the editors' reading provides a new instance of eitupes, a verb form comparable to Iguvine eitipes (V a 2, 14) 'they have decreed' and South Picene eitipes (Falerio 1, p. 183). A third instance of the spelling with p makes it likely that the second half of the verb form is from the root ap- 'take', not hab- (Weiss 2010:303). On the other hand this new form also complicates the claim that Iguvine -ipes is from a perfect stem *ēp- comparable to Latin (co)ēpī.

p. 145, Interamna Nahars (not in ST): pap̣ṛịṣ is strange if 'Papirius' is the right interpretation. The -r- in the latter is rhotacized from *-s- according to Cicero; *papisios, however, would have undergone syncope well before rhotacism in Umbrian, and -ps- would not have rhotacized. The inscription dates to 250 BCE or later, so more than a century after Latin rhotacism; a borrowing of the Latin name is thus conceivable, but only if the form was "adjusted" by a secondary round of syncope.

p. 156, Satricum 1 = VM 1: We do not agree with the editors' assessment that the "the language is Latin".

pp. 159–60, Sabini (?) 2 (not in ST; Ve 513): This inscription dating to 500-400 BCE, if correctly read as paq{q}is blaisiís, would be a very early example of the nominative singular of a gentilic in -iís. This spelling is surprising at this date (see Weiss 2010:302) and the supposed samekh ( = í) is not well formed.

pp. 163–5, Forum Novum 2 = Um 2: The editors look favorably on the idea that skerfs = Lat. scirpus and refers to the cord that attached the flask to someone's neck. We find the supposed connection with scirpus very problematic. The form scurfus cited by Walde and Hofmann (2:496), which is sometimes cited to support the idea that the root-final consonant was originally an aspirate, is nonexistent.

pp. 173 and 176, Capena (?) 2 and 5 (not in ST; Ve 359 d): o(uies) nauies eco is translated as 'I (am) O. Navius', but perhaps a better translation would be 'I (am) of O.N.' with nauies a gen. sg. The same goes for k sares esú 'I am of K. Sarus'. The genitive makes better sense in a possessor inscription, and how could sares be a nominative of an o-stem? Similarly staties Vestini, Marrucini, Paeligni, Marsi (?) 2, p. 211 is better taken as a genitive.

p. 187, Falerio 3 = Sp AP 5: The interpretation of the form noúínis as Nōnius < *nowenios is attractive but it is difficult to explain the use of í for a short e when e is used for the same sound in the same inscription (peteronis = Petronius). Perhaps <*nowēnios?

pp. 190-1, Asculum Picenum 1 = Sp AP 1: The form audaqum is read and interpreted as 'bold'. If this is correct, then it casts some doubt on the traditional connection of Lat. audāx and audeō with auidus, since the suffix -ido- is from *-idho-. Thus a Sabellic cognate would be expected to be †aufak-. The form údiíns is tentatively interpreted as 'have laid' but SPic. does not otherwise have evidence for a secondary 3rd pl. ending -ns. Instead we find -h.

pp. 196-7, Interamnia Praetuttiorum 1 = Sp TE 5: The editors offer a new reading meít{t}istrúí which looks like a magister-type form *meit- is-tero-. Could it be connected with meitims in the same inscription and could that point to an analysis of meitims as a superlative? We find the interpretation of nemúneí as a form comparable to Lat. nēmō 'no one' phonologically suspect. panivú can hardly be an ā-stem nom. sg. since there is no evidence for the rounding of final -ā at this early date.

p. 218, Furfo 1 (not in ST): The form deias 'goddess' gen. sg. is now attested on an inscription from Todi in Umbria too (Manconi and Prosdocimi 2011:425). These two forms seem to suggest that *deiwyās became deiias in Sabellic.

p. 226, Aufinum 1 = Sp AQ 2: On the famous "Warrior of Capestrano" the editors suggest that the form rakinevíi is to be broken as raki nevíi and that nevíi is naevio. But monophthongization of ai to e at such an early date (550 BCE) is unlikely. Contrast praistaínt on Interamnia Praetuttiorum 3 from 500-400 BCE.

pp. 229–33, Teate Marrucinorum 2 = MV 1: On the Bronze of Rapino, which gives the text of a sacred law of the Marrucini, the forms ferenter and feret are interpreted as a 3rd pl pres. pass. indicative and 3rd sg. pres. act. indicative, as is traditional. We wonder if these might not be old short-vowel subjunctives from an originally athematic fer-, which would fit the context better. Note the tendency of Oscan to use the subjunctive where Latin would use the imperative that the editors themselves comment on at p. 1441.

pp. 234–5, Teate Marrucinorum 3 and 4 = Sp MV 7 and 6: These two inscriptions seem to offer the nom. pl. (salaus < *salaōs < *salawōs 3) and sg. (salas, 4) of the stem salawo- 'well' (Lat. salvus), suggesting that w was deleted in some part of the paradigm, presumably before a back vowel, as in Latin. But we are not sure the grammatical interpretations are correct or that either form shows loss of -w-. salaus is surely the same as σαλαϝς Cosilinum 2 (pp. 1356-7), which is followed by singular ϝαλε; the editors render this as '(may you be) well. Farewell' (with no indication of number!), just like salas vali in TM 4. We wonder whether salas is simply erroneous for salaus and all three of these are just nominative singulars syncopated from *salawos < *salwos (with vale gapped in TM 3), addressed to the reader/passerby. There is no real reason for a plural in TM 3 anyway, nor any clear independent evidence that w was lost before a round vowel in Oscan. On p. 516, Cumae 13 = Cm 18 s(is) salavs offers another clear nom. sg.

pp. 237–8, Teate Marrucinorum 6 = MV 8: The form leexe cannot be separated from the 2nd pl. lexe on Corfinium 6 and is not a 2nd sg. as the editors suggest. peis is probably not a nom. sg. of Lat. pius but the nom pl. of the relative pronoun equivalent to OL quēs.

pp. 257-9, Interpromium A and B = Sp BA 1: The editors object to the interpretation of úlúgerna as an ablative indicating the place of origin and prefer to interpret the form as a nom. sg. fem. adjective agreeing with and understood 'helmet'. We see no reason why úlúgerna could not be an ablative indicating a place of origin in the ablative. On Interpromium B the editors are correct that it is hard to interpret erimenú spolítiú as a dedication of a helmet taken as booty from Spoletium dedicated to a divinity erimeno-, but their suggestion that erimenú is a fem. nom. sg. with rounding of final *-ā seems unlikely for this early a date (325-250 BCE).

pp. 267–8, Corfinium 6 = Pg 9: The editors read the first line as [pat]ir pracom[.] cippiim ecu(f) and translate 'Her father [sets up] a ??? prayer (and) a monument.' But it is unlikely that pracom can be equated with precem and taking cip̣p̣ịịm with Lat. cippus is difficult. The pp would have to be the result of Latin phonology (the so-called Iuppiter rule) and thus the word would have to be a recent loanword, but then how to explain the declension shift? The authors also accept a suggestion of Peruzzi's that uiđad means 'widowhood', but this is impossible because the word for 'widow' had a voiced aspirate (cf. Ved. vidhávā) and this would have yielded a Sabellic f, not any segment which might have become đ.

pp. 303–4, Sulmo 3 = Pg 4: This inscription, which survives only in two manuscript copies (one in Wolfenbüttel, one in Bologna), has the interesting sequence pid sei d(e){.}di(d) bratom pam p(e)perci sefei i{.}nom suois cnatois, which the editors translate plausibly as "because she (Minerva) has given her (Pacia) the favour which she (Pacia) has begged for herself and her children." The sequence sei is taken, after Rix, as miscopied for sefei but actually that does not make sense. The subject of dedid is Minerva and a reflexive bound within its clause would have to refer to Minerva, but that is evidently not correct here. Could it be that sei is correctly copied and is semantically equivalent to Latin non-reflexive ei 'to her' and is a trace of the old anaphoric non-reflexive dative *soy/ *sey seen in Old Hittite -še, etc.?

pp. 359–61, Minturnae 1 = Ps 10: Text B on this impasto bowl is clearly Latin. N.B. 1st sg. acc. med not Sabellic meom.

pp. 430–1, Capua 27 = Cp 28: prúfts, translated correctly as 'they set up', should be expanded as prúf(at)t(en)s, not prúft(en)s.

p. 497, Cumae 4 bis (not in ST): The divine name restored as pid[ieí] d[iv]iúí cannot be Fidius Dius. The editors do not seem to bring ζωϝει πιζηι on Potentia 11 (Lu 35), pp. 1380-1, into the equation, since they leave the second word untranslated. But it is likely that πιζηι and pid[ieí] are essentially the same word.

p. 508, Cumae 9 = Cm 13: The editors assume for this and the other similar curse tablets that the accusative personal name is the object of a gapped verb 'I curse' and translate tíf[ei at the end as 'for you'. In fact, the gapped verb is 'I give, consign' and tíf[ei is 'to you', i.e. the avenging or underworld spirit that is to carry out the curse (cp. the famous Plotius curse tablet, do tibi frontem Ploti, Proserpina Salvia, etc.).

Volume 2

p. 615, Pompei 1 = Po 54: This inscription, which is incomplete in ST 54, gives us the word kúsúl representing Lat. consul. Normally ú is used for the reflex of *o, while *ō is represented by u(u). Presumably the secondarily lengthened o of cō(n)sul was qualitatively more similar to the vowel of Osc. ú. The second ú probably indicates that Oscan borrowed the word when it was still archaic cō(n)sol.

pp. 662–3, Pompei 27 = Po 19: Very useful and interesting discussion of the mensa ponderaria, with analysis of the units.

pp. 665–7, Pompei 28 = Po 80–4: The editors endorse the interpretation of vaamunim as vadimonium and this brings up interesting phonological issues. If this is correct, as seems probable, the form vaamunim shows that a *wafemōnio- not only syncopated the second syllable, as expected, but also eliminated the f with compensatory lengthening of the preceding vowel.

p. 674, Pompei 32 = Po 72: If the (resuscitated) reading nee[rúm is right, the -ee- is correctly said by the editors to be taken over from the nom. pl. *neer. This would contain secondarily lengthened ē from *ners < *neres just like Umbrian nom.pl. frateer. Contrast nom. sg. niir Cumae 8 = Cm 14 with inherited *ē.

pp. 685–6, Pompei 38 = Po 55: The connection of mefíta<n>aís with Mefitis is difficult since Mefītis has a long i. Connection with mefio-, i.e. *mefyetānā- may be preferable.

pp. 702–3, Pompei 46 = Po 66: This text from a brothel, Markas, is the latest dated Oscan inscription. The wall into which it was scratched was put up after 72 CE on the basis of a coin of that date impressed in the plaster!

p. 728, Pompei 67 = Po 63: The editors offer the reading Sabinís = Sabinus, but a form Sabinís for an o-stem nom. sg. is unexpected. The image shows a second cross-stroke through the putative í and Rix and Vetter were probably correct to take this stroke as a cancellation mark, i.e. the correct reading is probably Sabins. This really would be the Oscan equivalent of Sabinus, a form actually attested in same building (Casa dei Cuspii) in Latin. This is a nice example of the value of including the images along with the transcription.

pp. 793–5, Pompei 114 = tPo 17: The text of stamp B aísívu XII is interpreted as 'of the 12 gods' not implausibly, but the idea that aísívu is a contamination of ais- 'god' and the last two consonants of deív- 'god' is highly implausible. Perhaps the expected gen. sg. of the u-stem *aiswom > aisiwom by anaptyxis? The fate of *-sw- in Oscan is not otherwise known.

pp. 830–1, Pompei 139 = Po 71: The interpretation of this graffito as adíuhbans, an Oscan spelling of Lat. adiuvans, if correct, offers an interesting representation of Latin /w/ by Oscan b. Compare Verus written as Berus in the Pompeian Latin graffito CIL IV.4380, among other examples.

pp. 887–92, Abella 1 = Cm 1: The editors offer a fairly conservative translation of the Cippus Abellanus (with no rendering for the unclear terms prupukid and sverruneí). They do offer a new and convincing interpretation of slaagí- as the end points of land (Gk. eschatia). This meaning actually somewhat improves the Joseph (1982) etymological connection of slaagí- with Gk. λήγω 'cease, stop'.

pp. 907–8, Nuceria Alafaterna (?) = Um 34: The inscription on a helmet found in Bologna is attributed to Nuceria on the basis of the text reh: nuvkri, but the attribution seems shaky to us. The editors consider the possibility that reh may be abbreviated for the Oscan equivalent of Lat. rectio in the sense of 'region', but a more straightforward interpretation would be the ablative of rēs. reh would be a possible spelling for rē, i.e. 'from Nucerian property.' This would mean, however, that the inscription was not Oscan.

pp. 1125–30, Bovianum 119 = tSa 27: The sequence kerí bení in the tile reading kerí bení súm cannot be for Cerrinus Bennius. Modified í is not an exptected spelling for either ī (as in Kerrīneís) or y (as in Bennieís).

pp. 1156–8, Terventum 8 = Sa 4: The form seemuneí which occurs in this inscription and also in Terventum 8 is intrepreted as equivalent to Lat. Sēmōnī, but the first syllable has the wrong vowel for a true cognate: we would expect síímuneí, as in fíís[nam] also in Terventum 8. Is seemuneí a Latin loanword or does it have some different origin?

pp. 1174–6, Terventum 18 = Sa 7: The image printed here clearly shows that akdafed not aíkdafed is the right reading, but what is akdafed? Prosdocimi's analysis ak-dafed where dafed = 'gave' (1996:263) is quoted with approval and is supported by the existence of δαfενς (Lucania, Brettii or Sicilia 1), but remains mysterious.

p. 1183, Terventum 22 (not in ST): A new inscription reading kúnsíf deívúz is interpreted as consens *divitus, but this makes no sense. deívúz< *deiwot-s must be a t-stem derivative of *deiwos and the phrase must be compared to the Lat. di consentes.

pp 1239–40, Aufidena 3 (not in ST): This new inscription offers some interesting verb forms: úpstúst is interpreted not implausibly as opsita est in the meaning 'buried' and angítust as 'is declared' (better perhaps 'was declared'). The interpretation of these two forms as past participles plus prodelided forms of íst is much preferable to the recent suggestion of La Regina (2011:439) that these are active pluperfect.

The evidence of úpstúst supports a separation of the Lat. form obsitus from the past participle of obserō. Lat. obserō simply means 'to sow (seeds), plant (plants)'. Obsitus on the other hand means 'covered', e.g. with rags (Ter. Eun. 236 pannis annisque obsitus, Haut. 294 pannis obsita). The probable appearance of this meaning in Oscan suggests it is old. Perhaps the -situs of obsitus 'covered' is the same as that in po-situs? The 'cover' sense would come from the preverb, cf. obditus 'covered, wrapped', oblinō 'cover over by smearing', etc. This inscription also provides the form pústí equivalent to the proto-form of of Lat. post (cf. Old Latin poste) and the unclear verb form avzsed.

pp. 1260–3, Anxanum Ortona 1 = Sp CH 1: We think the form kdúíú = Lat. clueō is better translated 'I am called' rather than 'I am famous' and we doubt that múreís can be a dative plural.

Volume 3

pp.1309–10, Lucania Brettii or Sicilia 1 = Lu 26: δαfενς is interpreted as preverb da- plus aorist of *dheh1-, but if that is the case, how does this square with ak-dafed?

p. 1311, Lucania or Bretti or Sicilia 2 Stamp (not in ST): This stamp offers the form δειϝιν(ο), an exact match for Latin dīvīnus. Also in Capua 8 (Cp 15) deivị[nas if correctly restored.

pp. 1328–31, Buxentum 1 = Lu 62: This legal text is one of the more important finds of recent years and offers a number of interesting new forms. If hαfειτουδ means habētōd as is likely, then it is unlikely that ιπειδ is a form of the same verb. The form ϝαfουστ if related to Lat. vādō would confirm the traditional etymological connection with the family of English wade. The probable gen. pl. πονδιουμ points to the retention or restoration of ō in the genitive plural. Contrast the acc. pl. ending in -ομ.

p. 1360–1, Numistro 1 = Lu 4: σουϝεν μεδδικεν 'in his term of office' is a nice example of doubled -en, showing the latter's grammaticalization as a desinence (cf. Fortson 2010); contrast Buxentum 1 (Lu 62), if correctly restored, σουf μεδεικα[τεν, though σουf could be abbreviated.

p. 1371, Potentia 6 = Lu 9 (part): The commentary on restoration of [τρει]μωνωμ is beside the point. Since Umb. tremnom need only mean something like 'house' (tabernaculum) there is no need to see a reference to augury in its restoration here. So the unlikeliness of an augural reference is not an argument against restoring it. The editors are probably right not to combine this fragment with Potentia 8.

p. 1379, Potentia 10 = Lu 7: The editors follow Untermann in taking διομανα[ς] as = Lat. dominae, but there was no palatalization of dentals before o like there was before u.

p. 1393, Potentia 18 (not in ST): This new inscription furnishes a second example of the adjective μαμερτοι 'belonging to Mars', cf. μαμερττοι Lucania / Potentia 20 = Lu 28. In the index (p. 1628) the editors interpret this as the dative of a yo-stem *μαμερτις, but that should produce *μαμερτ(τ)ιοι (cf. mamertiuí Acerra 1). From the same locale comes a morphologically similar form ϝενζαι 'belonging to Venus' < *wen-s-ā-. We either have here rare examples of bare o-stem relational adjectives, or a regional loss of postconsonantal yod in certain positions; speaking against the latter is the preservation of the yod in διωϝιιας Potentia 10 = Lu 7, διοϝιοι 40 = Lu 13, and generally in gentilicia.

pp. 1437–45, Bantia 1 = Lu 1: The editors confirm that the final r of loufir 'or' in line 8 is clear and this confirms the existence of t-less 3rd sg. middle forms in Sabellic.

p. 1465, Breig 1 (not in ST): βρειγ, probably a place-name, is taken by the editors as continuing a root *bergh- 'hill', but the real root is *bhergh- (*bherǵh-) and the comparison is phonologically impossible.

Additional comments by the authors on Imagines Italicae may be found on the Bryn Mawr Classical Review blog.

References

Fortson, Benjamin W. IV. 2010. "Reconsidering the history of Latin and Sabellic adpositional morphosyntax." AJP 131:121–54.
Joseph, Brian. 1982. "Oscan slaagí-." Glotta 60:112–4.
La Regina, Adriano. 2011. "Castel di Sangro, Aufidena." Studi etruschi 74:436–42.
Manconi, Dora, and Aldo L. Prosdocimi. 2011. "Todi." Studi etruschi 74:425–8.
Prosdocimi, Aldo L. 1996. "Appunti sul verbo latino (e) italic." Studi etruschi 61:263–312.
Weiss, Michael L. 2010. Language and Ritual in Sabellic Italy: The Ritual Complex of the Third and Fourth Tabulae Iguvinae. Leiden: Brill.