REMARKS ON THE MORPHOLOGY, PHONOLOGY, AND ORTHOGRAPHY OF HISP.-CELT. LVGVEI AND RELATED MATTERS

Brent Vine for helpful discussion. All opinions expressed herein are my responsibility alone. — Epigraphic abbreviations: Round brackets ( ) indicate characters left out by the engraver; square brackets [ ] indicate characters which are restored; the underdot ̧ indicates characters which have been damaged or which are no longer clearly legible; angled brackets ‹ › indicate characters which replace those erroneously incised by the engraver; the tie-bar ̂ indicates ligatures; the pipe | indicates line breaks. 1 Untermann 1997, pp. 624–625 allows that the first token could be read as LVGVEŞ, though the syntactic analysis of the inscription by Ködderitzsch 1985, pp. 213 leaves no doubt in my mind that it should be read as dat. sg. LVGVEI. EMERITA. Revista de Lingüística y Filología Clásica (EM) LXXIV 1, enero-junio de 2006 pp. 77-88 ISSN 0013-6662

This leads us to expect *-ou¾ ei¾ as the dative singular desinence in Hispano-Celtic u-stems, as attested by Skt.-ave, OLat.-VEI (later -uÍ), and OChS -ovi 4 .A number of scholars have treated the discrepancy between attested LVGVEI and expected *lugou¾ ei¾ , explaining it via morphological, phonological, and orthographical factors.In this brief reässessment of a problem whose analysis is rendered more difficult by the fragmentary nature of the Hispano-Celtic corpus, I offer some commentary on previous research and introduce some new elements into the discussion.

II. OTHER RELEVANT FORMS
3. The principal relevant epigraphic comparanda for LVGVEI from ancient continental Europe are as follows (see also Stifter 1997, pp. 216-217) 5 : 6 I regard the writing of ‹C› for /™/ not as an error by the engraver or evidence that we are dealing with a different etymon, but as a token of quasi-phonetic orthography in view of the fact that /™/ is normally realised as [k] phonetically in the modern Celtic languages.I intend to publish on this matter in the future.

7
The reading is after Untermann 1980, p. 375 (AE 1912, 12) There is little doubt that (a) is an inherited Celtic form, and it is possible that (b) is, as well, though a Latinised form would not differ from the inherited nominative plural save in the vowel quantity of the desinence, which is not captured by the orthography.The remaining forms all occur in Latin contexts, however, and have been Latinised in various ways.(c) preserves the inherited vocalism of the suffix, but has adopted Latin vocalism in the desinence 8 .(g) and (h) go further and affix the normal Latin u-stem dative plural desinence -ibus to the Celtic stem. 9(e) and (f) reflect the common development of -/us/ > -/o/ word finally in the Latin of Hispania (Carnoy 1906, pp. 179-180) and demonstrate the differential treatment of inherited */ou/ as /o / or /u / attested in this regional Latin, as well (Carnoy 1906, pp. 63-64).It is probable that (d), too, attested in Transalpine Gaul, evinces a Latinised monophthongisation of the inherited stem vowel.Among these Though I note that it is not certain that there are any phonological geminates that could have been recorded in the inscription.Only the first person plural verb COMEIMV, which has been connected by Ködderitzsch 1985, p. 217 to *mei¾ -'change' andby Meid 1994, p. 389 to *h1ei¾ -'go', appears to be at all possible.11 My view is very substantially the same as Schmidt's, though with some added details which are presented in § §8-10.
12 This idionym has often been understood to be a nt-stem, but Tir ´iCanTos ´ (Bot.IV A3) now guarantees that the genitive singular desinence of this stem class bore o-grade vocalism.Other attested forms of this idionym are nom.sg.s ´TenionTes ´ (MLH K.1.3iv 2) and Latinised gen.sg.STENIONTIS (HEp.3, 191).All are consistent with classification of the idionym as an i-stem.SŢENIONTE¸ and s ´TenioTes ´ represent monophthongisations of dat.sg.*-ei¾ and gen.sg.*-ei¾ s, respectively.For -e-for expected -i-in the final syllable of nom.sg.s ´TenionTes ´, one can compare ar ´e-(e.g., MLH K.6.1)< *ari-; one might also compare EMERITA (EM) LXXIV 1, enero-junio 2006 pp.77-88 ISSN 0013-6662 forms, (c), (g), and (h) are key in that they demonstrate that the inherited form of the suffix was preserved in the weak stem.

III. PREVIOUS SCHOLARSHIP
4. There have been a variety of attempts to elucidate the unexpected shape of LVGVEI.The first, with which I am in sympathy in principle, posits that u-stem dat.sg.-VEI directly continues inherited *-ou¾ ei¾ .Though he is not explicit, Lejeune 1955, pp. 9 & 9 8 apparently suggests that *-ou¾ ei¾ > *-uu¾ ei¾ .He does not, however, discuss the realisation of the latter as orthographic -‹VEI›.One might presume that he would have agreed with Wodtko's 2000, p. 235 report that there are not any geminates attested in the inscription. 10 Schmidt 1960, p. 183  An analogy with the i-stem desinence would have yielded dat.sg.-ou¾ , as may have been the case in Translapine Celtic (see n. 5).
6. Wodtko 2000, p. 235, in addition to the view that -VEI continues inherited *-ou¾ ei¾ , also suggests that u-stems in Hispano-Celtic were remodelled after consonant stems, yielding lugu¾ -as the stem for the theonym.The existence of forms such as Latinised LVGOVIBVS and LVCOVBV render such a view extremely unlikely, however, for inherited */ou/ was monophthongised in Latin (cf.§3); the diphthong must, then, be Celtic.7. Stifter 1997, pp.218-219 proposes that not only LVGVEI, but also the dative singular verbal abstracts (which function as infinitives) in (3), which likewise bore inherited proterokinetic flexion 13 , reflect a remaking of the proterokinetic flexion in the dative singular of Hispano-Celtic only, such that the inherited full-grade vocalism of the weak stem suffix was replaced by nil-grade -u-.
(3) a.The motivation for this change, according to Stifter, would be that the proterokinetic flexional paradigm possessed a descriptively hysterokinetic or amphikinetic (see Meier-Brügger 2002, pp. 218-220) instrumental singular (Schindler apud Hollifield 1980: 45) 14 , in both of which the suffix bore nil-grade vocalism, and that syncretism between the dative and instrumental singular commonly occurs in later Celtic.

EMERITA (EM) LXXIV 1, enero-junio 2006 pp. 77-88 ISSN 0013-6662
But syncretism between cases normally involves not the suffix of a noun, but the desinence, e.g., cf.inherited n-stem dat.sg.-on-ei, e.g., Piuonei (CI 26), with innovatory -on-i, e.g., Kuas ´oni (CI 20), ‹ locative singular in early Cisalpine Celtic (Eska & Wallace 2001).We do not have hard evidence for syncretism elsewhere in Hispano-Celtic, but we can look to Old Irish.To be sure, proto-IE instr.pl.*-b s is replaced dat.pl.*-b s os in all stem classes in Old Irish as evidenced by -(a)ib.But it is the instrumental desinence that took over the function of the dative case.It is also far from clear that the dative and instrumental cases syncretised in any stem class in Old Irish, as demonstrated in (4).
(4) a. Thematic dative singulars, e.g., fiur «man», presuppose a desinence with u-quality, which could continue proto-Celt.instr.Cf. the fact that in Old Latin, all unstressed vowels were rounded to proximate phonetic [Y], spelt both as ‹u› and ‹i›, before labial consonants; see Leumann 1977, pp. 87-90, Sommer & Pfister 1977, pp. 82-85, and Meiser 1998, p. 68. 16 Though let us note TVRROS (MLH K.3.8)beside TVROS (MLH K.3.9) in the graffiti from Peñalba de Villastar.17 For which, to the best of my knowledge, there is no evidence either for or against.In only the thematic and the u-stems, then, is it possible that the dative singular continues the instrumental singular form.But even were one to grant that such a merger is what occurred, we should note that the proto-Indo-European suffix -*-o-or *-u-, respectively -and desinence *-h 1 had long since merged in the distant prehistory of Celtic and that, in synchronic terms, it is the desinence, not the suffix, that syncretised in Old Irish.
IV.A NEW APPROACH.CELTIC TEXT AND LATIN ORTHOGRAPHIC PRACTICE 8.The view of Lejeune, Schmidt, and Wodtko that inherited *lugou¾ ei¾ > *lu-guu¾ ei¾ appears to me to have substantial merit.It is well known that there is a tendency in Hispano-Celtic for /o/ to be labialised to /u/ when adjacent to a non-final labial (Eska 1995: 9-13).Thus proto-IE thematic dat.pl.*-ob s os is usually attested as -uPos ´, e.g., louCaiTeiTuPos ´ (MLH K.0.7), but may appear with unaltered stem vowel in OILOBOS (MLH K.3.11) 15 .Cf. also COMEIMV (MLH K.3.3) and r ´usimus (MLH K.1.1A11), which, in my view, continue 1. pl. *-mos.9. Should we adopt the view that *lugou¾ ei¾ > *luguu¾ ei¾ , I would suggest that the engraving of ‹v› for /uw/ may not be either merely a random failure to note an orthographic geminate (as hinted at by Wodtko) 16 or a regular orthographic practice of Hispano-Celtic when engraved in Roman characters (as per Schmidt, who sees an identity with Latin orthographic practice) 17 .The fact that the inscription is engraved in Roman characters may be pivotal to our analysis, however.As an alternative explanation, then, I would propose that the orthographic representation of -/uw/-by -‹v›-is a token of «econo-It is possible that the forms LVCVBVS and LVCVBO, discussed in §3, are susceptible to a similar analysis.Under such an analysis, Schmidt's view is substantially correct, though the economic orthography would not be native to Hispano-Celtic practice, but an artifact of the inscription being engraved in Roman characters.
For a typological parallel, compare the situation in the Sabellic languages, in which /i/ and /j/ in combination and /u/ and /w/ in combination are normally both written in the native script, while economic orthography is employed to an overwhelming degree in inscriptions engraved in Roman characters (Buck 1904, pp. 28-29), e.gg.: That such a literate Roman practice should be employed in a Hispano-Celtic inscription should not be surprising.There were a number of Latin inscriptions engraved at Peñalba de Villastar; one, located amongst the Celtic inscriptions and engraved in the same type of hand, even is so erudite as to record a sentence from the second book of Vergil's Aeneid (II 268-269) (see further de Hoz 1979, pp. 245-246): (7) tempus erat quo prima quies mortalibus aegris inc [ipit] "It was the time when, for weary mortals, their first rest begins".10.The orthography of LVGVEI, then, is likely not to be essentially aberrant in any way.While it is possible that it represents a common sound change and a random failure to engrave an orthographic geminate, it is at least as likely to represent a token of a Latin orthographic practice solely because the inscription is engraved in Roman characters 18 .As so often in the Conti-19 The sixth character of this form is clearly engraved as ‹M›, but it is mostly likely to be analysed as a n-stem idionym.LXXIV 1, enero-junio 2006 pp. 77-88 ISSN 0013-6662 nental Celtic linguistic record, whereas LVGVEI has often been thought to be the product of engraver error or, at least, linguistically problematic, in the end, it is not so.

V. CODA. THE FORMANS -UN-IN VERBAL ABSTRACTS
11.The fact that LVGVEI and the dative singular verbal abstracts in -unlisted in (3) both appear to bear nil-grade vocalism in their suffixes when full-grade is expected does not mean that both must be explained in the same way.The verbal abstracts, for which we expect proterokinetic *-u¾ enei¾ or *-menei¾ , are not susceptible to the phonological and orthographic explanation that I have advanced for LVGVEI.Their unexpectedly vocalism can only be explained as analogical.