CC BY
4
DOI: 10.30842/ielcp2306901528072
Alexander Nikolaev
University of Cyprus, [email protected]
WISH FOR HOPE: GREEK ЕЛПОМА1, (E)EAAOMAI, LATIN VOLUP, ARMENIAN GELf
The root of Gk. еХлодоа has traditionally been analyzed as a by-form of PIE *uel(-)h1-, *uel(-)d- and *uel(-)gh- with a labial extension. This paper discusses the problems that arise under the root-extension approach (such as the putative anaptyxis in Lat. volup < *uo/el-p-i-), examines the relevant etymological dossiers, and proposes a new solution for Arm. geij. The paper seeks to rehabilitate E. Hamp's analysis of Lat. volup as reflecting a compound *uelh1-h1p-; it is argued that Gk. еХлодаг can be derived from the same compound with the loss of the laryngeal by Schmidt-Hackstein's Law in athematic aorist 3 sg. *(h1)e-u elhp-t > *(h1)e-u elp-t.
Keywords: Armenian historical phonology, Homeric language, Indo-European etymology, Latin historical phonology, Proto-Indo-European reconstruction, root extensions.
А. С. Николаев
Кипрский университет, [email protected]
Др.-гр. slno^ai, (¿)£X6o^ai, лат. volup, арм. gelj
Праиндоевропейский корень др.-гр. еХпоцаг нередко трактуется как вариант и.-е. *иel(-)h1-, *uel(-)d- и *иel(-)gh- с губным расширителем корня. В статье указывается на проблемы, возникающие при этом подходе (такие, как предполагаемая но маловероятная анаптикса в лат. volup < *uo/el-p-i-), рассматриваются этимологии отдельных слов, обычно привлекаемых к анализу в качестве дериватов и.-е. *иel-, и предлагается новое этимологическое решение для арм. geij. Приводятся аргументы в пользу идеи Э. Хэмпа, который возвел лат. volup к сложению двух и.-е. корней: к этому же композиту *иelh1-h1p-предлагается возводить и др.-гр. еХпоцаг с выпадением ларингала в формах атематического аориста по закону Шмидта - Хакштайна.
Ключевые слова: индоевропейская этимология, историческая фонология армянского языка, историческая фонология латинского языка, праиндоевропейская реконструкция, расширители корня, язык гомеровских поэм.
* Warm thanks are due to A. Alfonso Déniz, D. Buyaner, J. L. García Ramón, S. Hófler, G. Imberciadori, P. Kocharov, S. Neri, M. Peters, B. Vine, R. Viredaz, and M. Weiss. All mistakes are my own.
1. The Proto-Greek root of slno^ai 'expect, think likely, hope, conjecture, imagine'1, slnro 'make hope', slnsov ^im^ov (Hsch. s 2208), perf. solna (with presential meaning), sly (f.) 'hope'2, slmg 'hope', slnrop'n 'hope', aslrc^g (Od. 5.408) 'unexpected', aslrnog 'unexpected, hopeless' can be mechanically back-reconstructed as *uelp-. The initial *u is confirmed both by the Homeric data3 and by the external comparanda long identified in Lat. adv. volup (Naev.+) 'agreeably, to one's satisfaction' and voluptas 'pleasure, delight'4.
As is well known, the zero-grade allomorph *ulp- is reflected in Greek in several nominal forms: (1) snalnvog (Pi. Pyth. 8.84) used as an attribute of voaxog and paraphrased by the scholiast as ^5ug, npoo^v^g 'pleasant, gratifying'. (2) alnviaxog, likewise a Pindaric hapax, conjectured by Callierges on the basis of a scholium for unmetrical avslmaxov transmitted at Pi. Isth. 5.12/3; the adjective is used to modify aroxov 'the finest thing' and was probably understood by Pindar to mean 'joy-giving' vel sim. Wackernagel 1910 corrected the form into morphologically expected superlative almaxog (cf. alyiaxog next to apyalsog), actually attested at Aesch. Pers. 982 where some editors have printed 'Almaxov as a Persian proper name (which it is unlikely to be, see Schmitt 1978: 62-63)
1 For the semantics of elrco^ai and its derivatives see the detailed disquisition by J. N. O'Sullivan (LfgrE 2.559-561) as well as the forthcoming treatment by L. Huitink whose oral presentation in 2023 gave the impetus for the present study. Non vidi Lachnit 1965.
2 CEG 1.51 l.2 = IG I3 1219 l.5 (Attica, 510 BCE); acc. eln' is wrongly taken as "extreme Form der elision" of elrc(i5a) by Ecker 1990: 166 n. 533.
3 E.g. Il. 9.40, 15.288 ^ala (F)elrcsai; 9.371 exi (F)eXrcsxai; Od. 2.275 ensixa (p)eolna; 16.101, 19.84 Kai (p)elni5og odea; plqpf. ernlrcsi is traditionally analyzed as (e)(F)s(F)6Xrcsi. See Chantraine 1942 (2013): 132, 464. Bin- is well represented in personal names (see Bechtel 1917: 152; Minon 2023: 355-356); while there are no Greek epichoric attestations with an initial digamma, Doric *Fslndvrap was borrowed in Etruscan as velparun (4x), see Cassio 1999: 72-75.
4 See DELG 326-327, EDLIL 689, EDG 1.415. Benveniste 1935: 155 took Lat. lepos, -oris 'charm, grace', lepidus 'charming' from "State II" *ulep-, but this analysis has not been widely accepted (the traditional derivation from PIE *lep- 'to peel' ( > *'refine') has been advocated for by Stuber 2002: 67; similarly Garnier 2015-2016: 80-81 (= 2017: 327-328)). Hirt 1927: 310 (followed by Olsen 2007: 104-105) argued that Lat. lupa in the meaning 'prostitute' continues * lp-, but this outcome of initial * l- in Latin is not assured, see the discussion by Pike 2011: 87-88.
but West in his Teubner edition plausibly chose the appellative a^ntGxov5. (3) Finally, there is apna^sog 'pleasant, charming' (Od. 8.164+) from (*)a^na^sog by dissimilation of liquids, for which apya^sog < again provides a parallel6; the original form
is in fact attested in Hsch. a 3267 a^na^sov ayan^xov7. The meaning 'consuming' that apna^sog has in later texts8 is due to a cross-contamination with apna^ra which is also responsible for the non-etymological initial aspiration in this adjective9. We can thus posit a system *(p)a^nv6g / 'desirable' along with a
superlative *(F)s^moTog remade as *(F)aA,moTog and later as *(F)a^nvtGiog after the positive10. Finally, the allomorph a^n- < *ulp- is also attested onomastically11.
On the basis of Gk. s^no^ai and Lat. volup the root *uelp-'Hoffnung schöpfen' is reconstructed in LIV2 68013. The Homeric forms ss^nsxai (Il. 10.105, 13.813), ss^rcoi^v (Il. 8.196, 17.488) may appear to suggest the reconstruction *huelp- but as Beekes (1969: 64) correctly observed14, the initial e- in these forms can be
5 See West 1990: 93 as well as Garvie 2009: 355-356.
6 This explanation is due to J. Wackernagel apud Debrunner 1910: 14.
7 Cod. alnalaiov, with a typical substitution of <ai> for <s> reflecting the monophthongization of [ai]. The correction is due to G. Curtius; I. Vossius's alnvaiov is far less compelling.
8 Cf. CIRB 123.3 äpnalen vouco^ es Kaxe^öicsv (3rd-1st cent. BCE).
9 For the phonological source of the initial aspiration in aprca^, apna^ra see Nikolaev 2020: 37.
10 See van Beek 2022: 456-457.
11 For the Ionian fem. PN AXnalfl < *AXnalen (or AAno&n? Ephesus, 5th cent. BCE, LPGN 5b 22+) see Curbera 2013: 117-118; for the masc. PN AXnraviSn? (Karthaia 3rd cent. BCE) see Bechtel 1917: 488 who posits a substantive as the derivational basis. A. Alonso Deniz kindly draws my attention to the PN AXnalsio^ in the recently published Chian decree (Malouchou 2022, n° 3, col A, l. 11; 400-350 BCE).
12 Viz. a telic version of 'to hope'. This reconstruction is based on the fact that Greek has a perfect eoXrca, and in the Erlangen / Freiburg doctrine, reflected in LIV2, PIE roots with a stative or a durative meaning did not form primary perfects, see e.g. Kümmel 2020: 16. But this doctrine seems too restrictive, and eoXrca can be interpreted as a so-called "intensive" perfect (see van Beek-Migliori 2019: 74-75) made from an atelic root. (I thank M. Peters for a very helpful discussion of this point).
13 Under this analysis, the medial -u- in volup is explained by anaptyxis (see EDLIL 68) which is in fact an unlikely solution, see further below.
14 Beekes' explanation has been widely accepted, see e.g. Ruijgh 1971: 165
explained by analogy to sslSo^ai (< *hueld-15) with which slno^ai is similar in meaning16 and interchangeable in some formulae17. The forms in ssln- can also be explained without taking recourse to lexical analogy: (9u^og) sslnsxai (Il. 13.813) may be analyzed as an instance of "formulaic conjugation"18 of impf. (9u^og) sslnsxo (Il. 12.407), where the initial s- was ambiguous between the augment (which it historically is) and the root-initial vowel. The present stem sslns/o- thus created could then be used by the (later?) poet of the Doloneia (vuv sslnsxai 10.105), as well as to form the optative sslnoi-19. Finally, the variant ssln- could simply be a product of diectasis: as Peters (1980: 316) and Nussbaum (1998: 142-143) have shown, the rule of the poetic language according to which distended ss- could be substituted for s- and vice versa before a double consonant originated in those cases where original *eueRC-lost the intervocalic - -, underwent vowel contraction and then shortened the resulting *eRC- to *eRC- by a secondary Osthoff s Law in Ionic, e.g. *euerge/o- 'enclose' (¿spy©) > *erge/o- (sipyro) > *erge/o- (spyo^evn) or *euers<z- 'dew' (sspan) > *ers<%- > *ersa>-(spaai). The forms sslnsxai, on the one hand, and slSsxai, on the other hand, can thus be analyzed as metrical variants of much more frequent slnsxai (< *uelp-) and sslSsxai (< *eueld-) produced in accordance with this poetic rule. Whichever explanation should be preferred, the allomorph ssln- does not constitute evidence for a root-initial laryngeal in the PIE root *u elp-20.
(= 1991: 594), but note the hesitant notation "*(h;)uelp-" in Willi 2018: 196 vis-à-vis *uelp- on p. 214.
15 LIV2 254. The comparative evidence for this reconstruction of the PIE root is discussed in n. 47 below.
16 Cf. DELG 318: "eXSo^ai "désirer" envisageant le terme du procès, eXrco^ai exprimant plus généralement l'espoir et l'attente".
17 E.g. èéXSo^ai / eXSeai %axa rcavxa (1x Il., 3x Od.) vs. Snobai %axa rcavxa (1x Il.).
18 In the sense of Hoekstra 1965: 112.
19 Somewhat similarly Beekes 1969: 287 who, however, speaks of a "verschlepptes Augment" (of the type Mod. Greek aveßa^) which is a different matter. For a formula-based approach to (9u^oç) èéXnexai / èéXnexo / èéXSexai see Chiattelli 2022: 63-65.
20 The temporal augment in rçXrciÇov, rçXrcica as well as in impf. rçXrcexo, transmitted in the Homeric textual tradition (see West 1998: xxvii) and perhaps at Pi. Py. 4.243 as a variant of eXrcexo, is surely no indication of erstwhile initial laryngeal, contra Wyatt 1972: 75.
2. The older works of reference routinely (if noncommittally) connect this root with the well-established root of Latin velle, English will, OLith. velmi, OCS velitb, e.g. GEW 1.503: "können auf uel- in lat. vel-le, dt. wollen zurückgehen", DELL 752: "on pense au groupe de uolo". This analysis is in principle possible, since the root-extension *-p is known, cf. *ten- 'stretch' next to *temp- 'id.' (LIV2 626) or *(s)ker- 'shear' next to *skerp- 'cut' (LIV2 556, 559), and the semantic distance is not too large21. However, the problem is that the PIE root in question certainly contained a root-final laryngeal: as Hoffmann 1968: 6 (= 1976: 248) was the first to point out, the effects of the laryngeal are seen in the vocalization of the syllabic *r in Vedic trisyllabic 1 sg. aor. avuri 'I have chosen (for myself)', 3 sg. aor. opt. vurita where the allomorph *uur- must go back to *u\H-V, since otherwise *avri, *vrita would have been expected22. Similarly, OAv. 3 sg. inj. varata 'chooses' goes back to Proto-Indo-Iranian *urH-ta23; its Vedic counterpart vrta (which looks like a form made from an anit root) is usually explained by contamination with v - 'to envelop'24. The Vedic 9th class present stem vrnite 'chooses' likewise points to a root-final laryngeal.25 The
21 Possible scenarios for a semantic development include 'choose' (as in Ved. vr-, Av. var-) > 'choose (for oneself)' > 'wish' > 'wish for, expect, hope for' (eXrco^ai) and then 'wish for' > 'desire' > 'pleasure' (volup).
22 The initial v- in the opt. vurita was analogically restored after the root allomorph in other finite forms. Strunk's misgivings (1985: 510 n. 25 = 2005: 686) appear unfounded: there is no reason to suspect original full-grade /avari/ and /va ita/.
3 With the same treatment of accented RHC as in YAv. tauruua- 'overcome'
' - ti ° - ' - -< *trh2-uo-, Ved. türva-, or YAv. varana- 'wool' < *h2ulh1neh2, Ved. ürna-.
24 See Hoffmann 1968: 6 (= 1976: 248). Kümmel 2000: 459 observed that
the form vrta (RV 9.101.13b) can be scanned as a trochee and tentatively
posits *vürta (< *urHta). The pada märto nä vrta tad väcah scans as — w
w w - w - (Anustubh) and Kümmel's proposal restores the iambic form of
the opening (^ - w -) but metrical variations in the opening of an eight-
syllable verse are more common than in the cadence, so the proposal
remains uncertain. (According to van Nooten & Holland 1994: xvii, the
structure ^ - w w is found 179 times in the Rgveda as an Anustubh opening:
it is rare but not non-existing). See Kümmel (ibid.) for a discussion of the
perfect middle stem vavr- which he argues to be secondary.
5 If Vedic urväri- 'wife', Pali ubbari goes back to *uelh1- as 'lady of choice' (Burrow 1984; EWAia 1.229: "nicht sicher einzuordnen"), the preform *ulh1-uer-ih2 may have undergone the Indo-Iranian rule of laryngeal loss in an unaccented syllable (Lubotsky 1997); a laryngeal is
preservation of -l- in Umbrian 3 sg. ipv. veltu, ehueltu 'order'26 suggests a syncope in the middle syllable27: as Meiser 1986: 181182 has observed, the Umbrian form can be explained from an athematic preform *uelatod (< *uelH-) matching the (semi-) athematic inflection of Latin velle28. The acute in Lith. (pa)velti 'allow, prefer' may be due either to an erstwhile laryngeal or to a lengthened grade in a Narten formation. Further support for a root-final laryngeal in this root has been sought in the Doric verb l^v 'to want' (Ar. Lys. 981 l&, Epich. 49.1 lfig, Cretan subj. l^i, lEiOvxi, opt. ls(i)oi, Elean inf. lEv29) < Proto-Greek *u leie/o - where the root vocalism can be explained either from full-grade *u leh1-30 or from
still necessary since otherwise *vrvari would have been expected. For a possible semantic parallel note Lyc. AB lada 'lady, wife' that could go back to Proto-Anatolian *uladar < TVulhreh2-tf 'choosing' > 'the chosen one': for this preform albeit made from a different root (PIE *uel- 'to envelop') see Carruba 1991: 174; however, the traditional connection with Hitt. lazzi- 'good, right', Luwian /lara-(i)/ 'to prosper' (Proto-Anatolian *laHt-) and, with a different dental, Toch. B lare 'dear' (but see Imberciadori 2023a: 653-654) and Russ. lad(n)yj 'fit, good', ORuss. lada 'spouse' deserves continued attention.
26 South Picene ehueli (TE 1) is unrelated, see Weiss 2002: 358-262.
27 Contrast Umbr. muta < *ml-teh2 (Osc. molto, Lat. multa 'penalty').
28 But other solutions are possible, including a secondary thematization (*ueletöd) or a *-ie/o- stem (*uelitöd). Meiser himself in a later publication (1993: 258 n. 21) argued for a derivation from a causative *u ol-ie/o - 'eine Entscheidung treffen lassen' but this preform is morphologically suspect (Goth. waljan, Slavic *uoliti go back to iterative *uol-eie/o-). The relationship between the Umbrian forms and Lat. v lö is not universally accepted, see e.g. Cowgill 1978: 25 ( = 2006: 251) n. 3 for a position of skepticism. See the richly referenced discussion by Garcia Castillero 2000: 377-378.
The laryngeal reflex is notoriously lacking in Latin forms 3 sg. v lt volt < *uelti, 2 sg. vis (< *uel-si with Cowgill 1978) and 2 pl. vultis which are customarily explained as resulting from an analogical transfer of the allomorph *uel- from 1 sg. *uel-ö, participle *uel-ont- by analogy to edö : est (see Haröarson 1993: 85 n. 111; Kümmel 1998: 198; Meiser 2003: 230). It is, however, unclear whether Proto-Italic *uelati > *uelat would undergo syncope in Latin to become vult or not: an extremely frequent form may have been subject to an "Allegro-rule".
29 For a dossier of these forms see Meier-Brügger 1981.
30 The root allomorph *uleh¡- may be found in Venetic instr. sg. leno 'wish, will, voluntary act'? (3x, always in the phrase voltio leno) if this word goes back to *uleno- < *uleh}-no-, parallel to dono.m 'gift' < *dehrno- (leno
zero- grade *«/hi-31. While some of the material discussed in this paragraph remains uncertain, the cumulative evidence unequivocally supports the set character of the root.32
If the PIE root for 'choose; want' is reconstructed as *uelh1-(LIV2 677)33, its extended form is expected to have the shape *uelh1p- which obviously would not give Greek (р)в^люца1. Since Benveniste 1935: 155 this difficulty has been dealt with by analyzing both *-h1 and *-p as extensions used to make two different roots, *uel-h1- (vf-, X^v) and *uel-p- (sXno^ai), from one primordial root *uel-. "[s]uffixation par -э- de la racine *wel-'vouloir'". This influential approach has been widely followed and even modified to include (as was customarily done in pre-laryngeal theory scholarship) a third extended root allomorph * el-d-(sXSo^ai), to be discussed in section 3 below34. A version of this
was compared to Greek Xrçv by Lejeune 1974: 87). The descriptive Schwebeablaut in the verb is not troubling as long as PGk. *u leie/o - is analyzed as a thematized h2e-conjugation /'-present, where "State II" is regular (see Jasanoff 2003: 103; 2021: 169-170). However, zero-grade *ulhrié/ô- will give the same result.
31 See Harôarson 1993: 83. For an alternative etymology of Doric Xrçv see Peters 1987: 289: *leseie/o - (or *lesaie/o - with later substitution of -ee/o-for -ae/o-) from the PIE root *les- 'pick' ( > 'prefer'), cf. Hittite less-m/ 'pick, gather', Gothic l/san 'id.', Lith. lèsti 'peck; pick up' (of birds).
3 Contra the position of skepticism in Seldeslachts-Swiggers 2002.
33 Welsh gwell 'better, preferable, superior' < *uelh-so- 'of choice' (*-do-and *-no- also possible) does not show a laryngeal reflex, but a loss of the laryngeal in this context in Celtic is supported by several examples, see Zair 2012: 181-201. Note that Arm. gel 'beauty' < *uelno- goes back to the root *uel- 'to see' (LIV2 675), not *uelhj- 'to choose' (contra IEW 1137), so no laryngeal reflex is expected here either if one were to expect one at all (cf. gelmn 'wool' < *h2uelhj-mn).
34 Meillet 1908: 148-149 ("dans la racine *wel-p- on distinguera donc une racine plus simple *wel- et un élargissement *-p-; la même racine simple apparait avec un autre élargissement *-d- dans gr. peXSo^oa"); Schwyzer 1939: 701 ("parallele Erweiterungen mit d bzw. p"); Lejeune 1972: 175 ("hom. ééXSexai à côté de eXSexai, ééXrcexai à côté de ëXrcexai (dérivés *wel-d-, *wel-p- de *wel-: lat. wolo)"); Adrados, Bernabé & Mendoza 1996: 8 ("asi en gr. eXrco^ai, 'esperar' y eXSo^ai 'desear' [...] la original es *wel (en lat. wolo)"); Szemerényi 1990: 100 ("It is obvious that *welp- and *weld- must belong together, in fact that they both derive from a simpler root *wel-"); Seldeslachts-Swiggers 2002: 323 ("racines élargies *wel-d- et *wel-p-"); Kocharov and Shatskov 2018: 211 n. 3 ("the root is reconstruc-
explanation has been championed by M. Pike in his very stimulating discussion of volup and related forms (2011: 85-94)35. In support of this analysis Pike cites the root *ten- 'stretch' (LIV2 626) next to *tenh2- (Ved. uttana-, OAv. ustana- 'stretched out' < *-tnh2-no-36; *to/enh2-u- 'extension' ^ *tenh2-(e)uo- 'stretched' > Gk. xavaog, Corn. tanow 'thin', *tnh2-uh2- > Ved. tanu- 'body, self37) and *ten-d-'id.' (LIV2 627: Lat. tendo 'stretch', Ved. tandra- (n.) 'series, row'38, Lith. tandus 'stiff, tight')39. It is important to note that the anit root
ted as *uelhr (LIV: 677f.), which makes it an extended root itself (*uel-hr) when compared to other root forms").
35 "[Alternation of a set root beside an anit root with a consonantal extension. [...] If the root *uel-h1- had three alternating enlargements *-h1-, *-d-, and *-p-, volup must then show both h1- andp-enlargements together" (Pike 2011: 91-92). As Pike's formulation suggests, he is arguing on the basis of *uelh1- next to *uel-p- and *uel-d- without explicitly positing an original *uel-. In doing so, he is following Vine 1982: 228 (and ultimately J. Schindler) in treating at least the d-extended allomorph as exemplifying the pattern of "d-extended CERH- roots, or more precisely, d-extended forms based on roots that are either exclusively, primarily, or partially set". Vine (1982: 229) continues by tentatively suggesting that "at some stage, alternations CERH- : CER-d- (or rather CER-C) were governed by some purely phonological process [...] the precise nature of the rule is not immediately obvious". This may suggest to some readers the operation of the so-called Kortlandt Effect (*d > *h1), see Lubotsky 2013: 162-163 and especially Garnier 2014 who, in particular, uses this sound change to account for one of Vine's examples, *terh1- 'bore' ~ *terd- 'id.', Ved. found in Ved. trnatti, perf. tatarda (2014: 152 = 2017: 365). This may be a possibility in this particular case, but the larger pattern discussed by Vine includes *h2 and *h3, too. For the purposes of the present discussion note that the Kortlandt Effect is an unlikely explanation for the alleged coexistence of *uelh1- and *ueld- since the latter in fact appears to go back to *h1weld-, as will be discussed in section 3. To sum up, the alleged coexistence of *uelh1- and *uelp- is a part of a more general pattern of CERH- ~ CER-C- alternations a general explanation for which could be either morphological (root-extensions) or phonological, but while the latter would be a tempting option, too, no such purely phonological explanation has yet been found. The third option is to look for individual solutions for individual cases: it is pursued in the present paper.
36 See Beekes 1985.
37 See Widmer 2004: 98-99.
38 For Vedic tand- 'be weary' see Kümmel 2006: 323-324.
39 Pike's second example is *kelh2- 'beat' (LIV2 350: Gk. кХй- 'break', Lith kalti 'forge', Lat. calamitas 'misfortune' < *klh2-mo-, etc.) next to p.p.p.
*ten- is very well established, by contrast with the presumed anit *uel-. While root-extensions are certainly an Indo-European phenomenon40, in the case of *uel-hi— *uel-p- this approach is, in my opinion, ultimately not very satisfying since the evidence for the primordial root from which these extended forms are supposed to be derived is effectively absent41, and the status of *hi as a root-extension is quite dubious42.
3. At this point it will be appropriate to discuss another alleged extended allomorph of the same root, namely, * uel-d-, evidence for which has been sought in Gk. sXSo^at 'desire'43. The advent of the laryngeal theory undermined this traditional approach just as it did
perculsus 'beaten down' taken directly from *-kld-to- and extended *Mh2rd-in clades 'ruin'. Here one may doubt the isolated testimony of perculsus which, as Meiser notes (2003: 187) may have been created by analogy to semantically related pellere 'beat, push' : pulsum or vellere : vulsum 'pull out'. The allomorph *kol- < *kl- may also have been extracted from the present stem *kollo < *M-n-h2-e/o- prior to the analogical introduction of the e-grade (LIV2; the present (-)cello may also go back to *kelh2d-e/o- with assimilation postdating syncope as in noXuSeuKn^ > *Poldoukes > Polluces, Weiss 2020: 209). In addition, for clades there is an alternative analysis in terms of a compound *kelh2-dhhi-e/o- (Garnier 2010: 285).
40 For example, see Kazansky 2015 as well as Kazansky 2018 and other
materials of the workshop "Proto-Indo-European Root Extensions" published in Historische Sprachforschung 131 (2018) 98-330. 1 This evidence is limited to Lat. vult, vultis for which other explanations are available (see n. 28 above) and to Proto-Gmc. pres. *utill-, Goth. 3. sg. wili 'wants', Old English wile (< opt. aor. *uelhi-ihi-) where the absence of resonant gemination may give pause, but since this Proto-Germanic sound change only applied after an accented short vowel with another vowel following the cluster (CVRHV > CVRRV, see Neri 2009: 6), its effects would not be expected in most forms of the athematic (Narten) root present or aorist; the allomorph *utel- could thus have been leveled to the optative from indicative. Of course, forms like Goth. waljan or OCS voliti (< *uol(hi)-eie/o-) or Arm. gol 'thief' (< *uolhl-o-) are not informative in regard to the laryngeal.
42 No examples in LIV2; three cases may (or may not) contain *hi: *bhes- / *bhseH- 'chew' (LIV2 82, 98); *pneuf- 'breathe' / *pneuiH- 'be(come) conscious, sense, perceive' (LIV2 489); *(s)ker- 'shear' / *(s)kerH- 'divide, cut' (LIV2 556, 558). Note also the complex of *deuihi- and *deuih2- 'fit together' that invites further challenging questions (see Neri 2017: 165 n. 290 and Nikolaev 2021: 24-25 n. 13).
43 See the references in n. 34 above.
in the case of PIE *uel- (now: *uelhr) ~ *uel-p-: the initial ee- in sslSo^ai was explained as reflecting a root-initial laryngeal (*h1ueld-), as was already mentioned above44. Since PIE *uelh1- clearly did not begin with a laryngeal (cf. Doric lflv < *ulhr), Pike (2011: 90 n. 70) tentatively posits reduplicated *ueuel(h1)de/o- as the preform for sslSo^ai. However, the noun sslSrop 'desire' with its obviously archaic morphology speaks against this analysis, as it is unlikely that this noun was formed from the present stem and not directly from the root. One might argue that the initial sequence ss- in both sslSo^ai and sslSrop represents an artificial poetic distension (see section 1 above), but the distribution of these forms in Homer strongly suggests that the initial ss- is primary: slSs/o- 3x vs. sslSs/o- 17x, sslSrop 9x (with slSrop only appearing in Herodian and Hesychius). An additional argument in favor of the linguistic reality of initial ss- in sslSrop may come from Ibycus fr. 37a PMG owi Kaxa a^sxspav sslSrop: since Herodian cites the verse precisely for the remarkable feminine gender of the noun, the traditional emendation to sslSro (going back to Schneidewin) makes very little sense. However this morphological oddity should be explained45, an argument can be made that Ibycus's use of sslSrop is not owed to the epic tradition (where the word is neuter) and sslSrop is therefore unlikely to display artificial epic diectasis. Following Beekes 1969: 63-64 and LIV2 25446, I prefer to analyze *h1ueld- as a separate root47, no matter how similar to * uelhr, formally or semantically, it
48
may appear .
44 The Chian PN 'ElSiog (Bechtel 1917: 151; probably hypocoristic for 'ElSi^evng vel sim.) does not support original *ueld- as it can equally represent a phonologically regular Ionic development of *eueldi- > eldi- > *eldi- (for which see section 1 above).
45 See Leukart 1987: 355-356.
46 For a similarly cautious position see Dede 2013: 62-63.
47 Cognates of ¿elSo^ai have been long identified in OIr. fled, gen. flede, Welsh gwledd (f.), Breton gloe 'feast, banquet' (Proto-Celtic *ulida- < *hjuld-eh2); Gaulish PN Vlidorix ('rich in feasts'? 'offering rich feasts'? 'king of feasts'?) was added by R. Thurneysen apud Weisgerber 1930: 213. The semantic side of this etymological connection is not immediately obvious: we can either suppose that a meal could be conceptualized as 'the desired thing' or theorize that the root originally meant 'to choose (for oneself)' - making it a synonym of *uelh1— hence gglSo^ai 'desire', while the Celtic words for 'banquet' go back to a stem meaning 'choice', 'portion' with the same semantic development as in Lat. cena 'dinner' <
4. There is one more form for which an analysis with a root-extension has been proposed. Meillet (1900: 393) took Arm. gelj3 'longing, desire' from *uel-gh-49. Assuming that the word-final -j is not simply due to contamination with synonymous ilj or balj, this etymology appears to point to another enlargement of PIE * uel- (the putative source of *uel-hi- and *uel-p-)50. Nevertheless, the meaning
*kertsna 'portion' or Gk. 5aiç 'feast' vis-à-vis Saio^ai 'divide'.
48 In theory, the relationship between *uelhr and *h1uel- (^ *h1uel-d-) can be described as a metathesis, and just such change has in fact been proposed by Rix 1979: 737 and Klingenschmitt 1981: 124 n. 14 in the form *uRHlV- > *H/uRV-, e.g. *urh1-u- (Ved. uru-) > *h1ur-u- > eùpûç 'wide', or *unh1-eh2 (cf. Alb. vê 'to lay down' < PAlb. *uene/o-) > *h1un-eh2 > eùvrç 'bed' (semantically parallel to Gk. Xéxoç 'bed' ~ Xéxexai, PDE l/e). One problem with applying this metathesis to the root of èéXSo^ai is that the rule is formulated as prevocalic, and most forms of PIE *uelh1- were athematic. In general, the metathesis hypothesis has not aged well: for eùpûç a different inner-Greek solution was proposed by Peters 1980: 53-54 and other solutions are available (see e.g. NIL 251 n. 3); the aor. eûpé/o-has been argued to go back to reduplicated *ueur(h1)-e/o- (see Kostopoulos 2014-2015: 158-180); and eùvrç may go back to *h1un-eh2 from root *huen- (see Ziegler 2004 and Schumacher 2007: 268 who compared Brittonic *uenta 'place, neighborhood', MWelsh (cad-)went 'battle field'). Balles 2007 offers a balanced treatment of the problem and discusses the important counterexample *urh1-en > àprçv 'lamb' (but see Neri 2021: 212 n. 29). As far as Gk. èéXSo^ai is concerned, it is important to point out that the evidence for the alleged metathesis is limited to Greek, and it is not clear whether at a late ("pre-Proto-Greek") stage it was still possible to form an extended form *h1wel-d- from *h1wel- (< *uelh1-).
49 For *gh > Arm. j cf. *ghesr-m > jer-n 'hand', *gh(i)i5m > j/wn 'snow', or *h2(e)ngh-u- > anjow-k 'narrow'. Arm. gelj3 'desire' is unrelated to gelj1 'convolvulus' (a plant), cf. OEnglish wel/g, PDE w/llow (see Martirosyan 2010: 203-204), and gelj2, cf. Russ. zeleza 'gland'. See for all three words HAB 1.535.
50 Accepted by Vine 1982: 123, Meillet's etymology appears superior to Bugge's (1893: 6) and Hubschmann's (1899: 45) comparison with Gk. (é)0éXœ 'want', Slavic *zelati, *zeléti 'id.', since palatalization of the initial aspirated labiovelar would be expected in Armenian (*gwhel-gh- > *Jelj, cf. *gwher-mo- > jerm 'warm, warmth'). Even though this etymology might perhaps be salvaged by taking recourse to Meillet's later idea that two palatalized stops could have been dissimilated (1905-1906: 243-245), another problem with Bugge's comparison is that one would expect a prothetic vowel in Armenian if the initial e- of é0éXœ goes back to *h1, as seems to be the case. (For é0éXœ as the original form see Debrunner 1954: 87-101, 105-108; against the old idea of a "deictic element" *(h1)e in
'desire' is easier to align with that of Gk. sslSo^ai, sslSrop and the possibility remains that gelj goes back to *h1uel-gh-, a by-form of unrelated (as argued in section 3) *h1uel-d-51. But there is a different etymological possibility for Arm. gelj: Belardi 2009: 70 mentions Ved. gardh- 'is desirous' (grdhya-h, agrdhat, jagrdh-) and OCS gladb 'hunger', RuCS zbldeti 'crave for' (PIE *gheldh-52), which is semantically attractive and formally possible, since no palatalization of PIE * gh before a front vowel is expected in Armenian, cf. the homophonous Armenian word geij-k' 'glands ' < *ghelgh-, Slav. *zelza 'gland'53. The problem, not discussed by Belardi, is that the expected outcome of the root-final consonant of *gheldh- in Armenian would have been -d-, not -j-. But another origin of Arm. j (beside PIE * h) was discovered by Pedersen 1906: 424 (= 1982: 202) who showed that [dz] (<j>) also emerged from the combination of Arm. d (< *dh) with *s after l und r54, citing aor. (e)helj 'suffocated'55 (from which present stems heljanem, intr. heljnum 'choke' have been derived) for which a preform with root-final Arm. *-d < *dh followed by the sigmatic suffix is made likely by the noun
£98!® see Szemerenyi 1966: 43 (= 1987: 1234), supported by Dunkel 2014: 2. 760 n. 66; for reflexes of word-initial laryngeals in Armenian see Kocharov 2018).
51 The semantic proximity with Gk. gglSrop was signaled by Solta (1960: 34), and Peters (1980: 41) reconstructed *h1uel-gh- under the theory that initial *Hu- and *u- have the same reflex in Armenian. However, Kolligan (2019: 80-82) questioned the latter assumption, arguing that a prothetic vowel should be expected from PIE *H and offering explanations for apparent counterexamples (such as gelmn 'wool' < *h2uelh1mn). Arm. gol 'thief (< *uolh1-o-), mentioned above, does not help resolve the issue, since the initial laryngeal may have been lost by the Saussure Effect.
52 IEW 434; LIV2 185; a different reconstruction *gweldh- is offered in EWAia 1.474 which would not be compatible with an Armenian form with an initial g-, but in fact initial *gh accounts for all of the cognates just as well. (The reconstruction *gweld - is due to Szemerenyi's (1966: 44) very dubious adduction of Gk. poulo^ai to the dossier for this root).
53 See e.g. Olsen 2011: 27; Scala 2017: 108-110 (I owe the latter reference to P. Kocharov).
54 In other environments the cluster was simplified to z as in aor. -eloyz 'let go out' < *h1leudh-s-, esoyz 'plunged' < *keudh-s-, etc., see Viredaz 2018: 202-203. (I am grateful to R. Viredaz and to P. Kocharov for a stimulating discussion of the Armenian material).
55 For a discussion of this verb see Kocharov 2019: 50-52.
xeld (i-stem 'noose; strangulation'), xeldem 'strangle'56. Pedersen originally assumed this development only for secondary clusters (either arising in sigmatized forms or in composition), but there is a growing consensus that inherited *dhs may have had the same outcome57, see the discussion by Kocharov 2019: 37-3858. There do
not seem to be any counterexamples to *dhs > j / L_. With this in
mind, one can posit an s-stem *gheldh-es- 'desire'59 which would be expected in a Caland system next to *ghjd-ro- ( > Ved. grdhra-'desirous' (RV), with a surprising accent60) and *ghjdh-o- ( > YAv. gsrsSa- 'eager, avid, greedy', Yt. 15.47)61. This s-stem would make a possessive derivative *ghldh-s-o- of the type *leuk-es- 'light' (Av. raocah-) ^ *luk-s-o- 'having light' > Ved. ruksa- 'radiant'62. This formation is directly attested in Indo-Aryan *grdzha- > Ved. grtsa-'clever, dexterous; rogue'63, Pali gijjha- 'eagerly desiring; vulture',
56 The analysis goes back to Bugge 1892: 449 and was accepted by Kortlandt 1987 (= 2003: 80); 1994 (= 2003: 105). The variation in the initial consonant is admittedly troubling but is not without parallel. Bugge remarks that "das anlautende h sonst mit x wechselt" without going into further detail; for this variation Djahukian 1967: 310 lists sixteen examples, from different stages of the language, some of them loanwords, e.g. xer ~ her 'beestings', xer 'stubborn, disobedient'- her 'strife, enmity', xul ~ hiul 'shed' (< Gk. uln?), t-xelc ~ t-helc 'deformed, ugly', xalb ~ halb 'snare, net' (Syrian xawla), etc. (I owe the reference to D. Buyaner). Djahukian (2010: 326, 456) maintains the connection between xeld and helj, while for Olsen (1999: 98) xeld is isolated.
57 "Inherited" here does not necessarily mean "original": there is evidence that the phonologically regular development of *-dhs- in PIE may have been to *-ts- with devoicing and deaspiration, but most forms were remade to maintain morphological uniformity (see Nikolaev 2023: 869-872, n. 43).
58 Note Kortlandt's retraction of his earlier view that PIE *ds, *dhs, *gs, *ghs merged as Arm. c' (apudKocharov 2019: 37 n. 24).
See Rau 2010: 107.
60 Possibly analogical to the substantivization grdhra- 'vulture'. B. Vine suggests the influence of the present stem grdhya-" (the only attestation of which in the Rg-Veda is the participle grdhyantam RV 4.38.3c).
61 See EWAia 1.493-494 and Hofler 2017: 460 (who operates with a root *g(w>eldh-, following the EWAia).
6 "russus-type" in the terminology of Hofler 2017; see also Hofler 2018: 133-134.
63 This meaning is traditionally explained as the result of influence of grdhnu- 'hasty, quick'. One meaning of Proto-Indo-Iranian *g(hardh-as-may have been 'greediness', see Rau 2010: 107 n. 88.
gijjhati 'covets'64. The stem *ghldh-s-o- 'desiring' substantivized with vrddhi as full-grade *gheldh-s-o- 'desire'65 will give Arm. gelj 'longing'66.
5. It is time that we return to the root of Gk. sXno^at, Lat. volup and its relationship with *uelhi- 'to choose; to want', as there is another hypothesis advanced by Pike in his important treatment of Lat. volup and related forms that deserves careful consideration. The adverb volup is traditionally taken from apocopated nom.-acc. sg. neut. *uolupe < *uolupi made from an adjective *uolupis61. Under the theory that *-hi- and *-p- (and *-d-) are alternating enlargements that could be combined in the same root-allomorph 68 , Pike reconstructs an acrostatic i-stem *uelhip-i-, offering further parallels for this nominal type. This reconstruction appears morphologically justified in view of the "Narten" behavior of the verbal root * uelhi-(opt. *uelhi-ihi- > Lat. velim). The weak stem *uelhip-i- would regularly give Proto-Italic *uelapi- > *uolapi- (*e > o before l pinguis, prior to vowel weakening69) that may conceivably have served as the source of Lat. volup10.
64 See Wackernagel 1907: 313 (= 1955: 502).
65 Alternatively, *gheldh-s-o- 'desire' may be viewed as a substantivization of the possessive derivative *gheldh-s-o- derived on the model of *leuk-es-'light' ^ *leuk-s-o- 'having light' > Old Norse ljoss 'bright', substantivized ljos n. 'light' ("pexus-type" in the terminology of Höfler 2017). As Höfler 2017: 32 remarks, "ein vermeintlich durch Vrddhi substantiviertes russus-Adjektiv formal gar nicht von einem bloß mit Kontrastakzent substantiviertenpexus-Adjektiv zu unterscheiden ist".
66 Under the theory that the development *-Ts- > j after liquids in Armenian was limited to secondary clusters, one may still argue that geij represents a later derivative made from the reflex of PIE *gheldh-es-, *gheldh-s- or that the erstwhile presence of this base stem (> Arm. *geid) in Armenian allowed the secondary treatment (*ds > dz).
67 See e.g. Weiss 2020: 159 n. 84. Pike (2011: 87) has plausibly argued that the medial vowel in volup(e) escaped syncope thanks to its being accented in a fixed phrase volup est 'it is a pleasure' (Pl.) < *volupest(i). The noun voluptas (Pl.+) is probably built directly to the stem volup since pre-syncope *uolupitas should be expected to give *uolpitas (especially after a liquid, as Pike 2017: 86 notes), cf. *deksiteros > dexter 'right', *oinodekem > undecim '11' or *uirotüts > virtüs 'virtue'.
68 See n. 35 above.
69 For this relative chronology see Nussbaum 1999: 386-387.
70 See below for phonological details.
With considerable ingenuity, Pike derives Gk. slnig 'hope' from the same word, arguing that in the strong stem * uelhp-i- the laryngeal was lost in the coda of a super-heavy syllable, giving Proto-Greek *uelp-i- which by Osthoff s Law gave *uelp-i-. This laryngeal loss rule (*CeRH.C > *CeR.C), comparable to the well-known Saussure Effect (*CoRH.C > *CoR.C), was discussed in the late 2000's but was never formally published, and doubts remain71. In addition, it is not particularly likely that Gk. slnig continues an i-stem: while a few original (barytone) i-stems have indeed developed by-forms in -iS-72, e.g. omv 'sight' next to omSog or ©sxiv next to ©sxiSog73, there is no evidence that slnig (Od. 16.101, 19.84 slniSog, Hes. Op. 498 slmSa) ever had i-stem declension74. The Greek word can be analyzed as a productive deverbative formation
71 The example Pike cites, following M. Weiss, is Kpn^vog '(steep) river bank, (overhanging) cliff' which appears to go back to *krem(h2)-no-derived from a Narten present stem made to PIE root *hremh2- (Gk. Kpg^a^ai 'to hang, to be hung up', Ved. sram'-, sramyati 'is weary'). It is true that Kpn^vog (Hom. +) cannot be derived within Greek from the late and analogical present stem Kp^vnm first securely attested in Appian (for which the influence of the noun is traditionally assumed). However, another, admittedly more complicated way of getting Kpn^vog from the root of Kpg^a^ai would be by assuming a deverbal noun *hremh2-mn ^ *kremh2-mn-o- > *hremmn-o- by Schmidt-Hackstein's Law (see Neri 2021 for an example of this law in a similar context) > * hremn-o- with degemination. As S. Hofler reminds me, we could also start with a zero-grade form * mh2-no- the lautgesetzlich outcome of which would have been PGk. *krmano-: this phonotactically impossible form would be subject to further irregular changes and a metathesis is at least thinkable. In any event, the etymological connection between Kpn^vog and Kpg^a^ai may be supported by Slav. *strbmb 'steep' which may be taken from the same root (some of the Slavic reflexes mean 'hard': the word appears to have been contaminated with the reflexes of *strh1 -mo- 'stiff'; see Nikolaev 2014: 712-713). Differently Sturm 2021: 59 n. 60 who analyzes Kpn^vog as a completely unrelated derivative of Kp a- / Kpn- 'head' in the meaning 'edge', comparing PDE headland.
72 Thanks to Scharnierformen such as nom.sg. -ig, voc. sg. -i, and dat. pl. -ioi.
73 See in general Meier 1975. A Mycenaean example can be seen in nom. du. te-mi-dwe-te /termidwente/ (KN So 4433+) 'provided with a border/flange?' next to Myc. te-mi-we-te (KN Sg 1811) if the latter form is not a spelling mistake but a genuine form derived from an older i-stem (Hsch. t 542 T£pm<^ noug?, cf. Hom. xepmoeig).
74 This evidence should certainly not be sought in acc. euelniv, first found in Plato.
of the type Sai© 'burn' ^ SaiS- 'torch', Xsn© 'peel' ^ XsmS- 'skin, shell', etc.75.
Despite the position of skepticism taken in the previous paragraphs, Pike's discussion of volup opens the door to interesting questions about the origin of -u- in this word76. The standard works of reference are completely silent on this issue. M. de Vaan hesitantly supposes anaptyxis (EDLIL 68) which, however, remains without parallels77. We have Lat. culpa 'guilt' which appears to go back to *kuolpa- from PIE *kuelp- 'bend' (LIV2 375), scalpo 'scratch' from PIE *skelp- (Gmc. *skelf-), sulpur 'sulfur' if from PIE *solp- (Toch. B salype 'oil, fat', Ved. sarpis- 'lard')78, and importantly, vulpes19 'fox' < *uo/elpi- so it is not obvious that Proto-Italic *uo/elpi- would undergo epenthesis. Differently from these words, however, neut. *uo/elpi > *uolpe would have undergone apocope of the final vowel80, and it cannot be excluded that a late epenthesis became necessary to eliminate the unique (and thus illicit?) word-final cluster *-lp81.
75 Another possibility is a derivation from the root noun eXrc- (cf. ai'^ ^ aiyiS-), as M. Peters kindly points out to me.
76 A form volop is attributed to Naevius by Charisius (GL 1.239.12) who, however, quotes the fragment (com. 12) for the exclamation eho, and the status of the form volop (not volup) is not assured, as M. Weiss points out to me. (I thank Weiss for his help in affirming the reading of the Naples manuscript). Weiss also kindly informs me of the form [v]oluptates in an inscription from Etruria (Blera) dated to the 4th-3rd century BCE (see Ambrosini 2012-2013 (2014)): since the form predates the earliest examples of the change of o to u in an unaccented closed syllable by 50-100 years, it casts doubt on Charisius' testimony.
77 Even though consonant clusters ending in *-l- (*-dhlV-, *-tl-, *-kl-) regularly undergo epenthesis, with the vowel quality determined by the quality of the l (± velar), as in stabulum / stabilis, this is of course a very different environment.
78 Cf. further palpus 'palm of the hand' / palpo 'to caress', pulpa 'meat', and talpa 'mole' which have no etymology. It is important to note that syncope remains a possibility (to my knowledge, there is no evidence that primary and secondary -lp- clusters were treated differently in Latin).
9 volipes (Quint. Inst. 1.6.33, Isid. Orig. 12.2.29) is a grammarian's coinage, based on the presumed etymology (vo lvo / volo + pes).
80 For this late apocope cf. simul 'at the same time' < semol < *semli (next to similis); see Weiss 2020: 159.
81 The result, of course, was still a unique Latin word ending in -p. There is no reason to believe that voluptas could be made directly from a verbal root
In view of this uncertainty, a search for an alternative solution is warranted, and it may indeed be advantageous to derive volup from either Proto-Italic *uolVpi-8 or *uelVpi-: in the latter case V would have to be a back vowel83, causing velarization of the preceding lateral as l-pinguis (= [I])84. The requisite phonological changes are generally well understood though not entirely straightforward: as is well known, a short vowel in a medial open syllable before a labial consonant becomes a sound written with <u> in Old Latin but eventually spelled with <i>: optumus ~ optimus85. While the spelling *volip is not attested, it is important to bear in mind that a back vowel in the initial syllable of the word has been known to favor as the resulting vowel, cf. auceps, aucup- 'bird-catcher' vs. princeps, princip- 'chief (< *-kap-) or monumentum 'monument' vs. alimentum 'nourishment' (< *-e-mentom). Nussbaum (2016: 221-222 n. 36) has recently drawn attention to an additional conditioning factor: was the outcome of a short vowel in an open medial syllable immediately preceded by Pol- (where P = labial consonant), cf. volucer 'flying' < *uolV-tli- or molucrum 'part of a mill' (Festus 141 Marx = 124.2 Lindsay) < *moletlo-. It is easy to see that *uolVp(i) fulfills both of these conditions, whether its -o- is an inherited o-grade or a product of *e before lpinguis; the outcome *uolupi > *uolupe > volup is thus assured.
- which would make it a unique stem in -tat- - rather than from the i-stem adjective (cf. *fakli-tat- > facultas) but even in a theoretically unlikely preform *uelp-tat- no epenthesis would be expected, cf. sculptus 'carved'. 2 Technically, in an o-grade form *uolHpi- the laryngeal should have been lost in accordance with the Saussure Effect, making the derivation of the vowel in the medial syllable from Proto-Italic *a < *H problematic, but the laryngeal could have been analogically restored. Likewise possible is PIE *uol(H)upi- vel sim. Zero-grade *ulHpi- would have an entirely different outcome, while the reflex of * lpi- would not undergo an epenthesis, as argued above.
83 Since long e caused velarization (e.g. Hercules < Her(a)cles), it has been assumed that short e had the same effect but see Imberciadori 2023b who argues for Latin celeber 'busy, crowded, frequent' < *kelh1-es-ro-.
84 A different approach is taken by Olsen 2007: 104-105 who analyzed volup- as a compromise between the full grade *uolp- < *uelp- and *lup-which she, following Hirt, argued to be a regular development of *ulp-.
85 See Leumann 1977: 87-90; Sommer-Pfister 1977: 82; Meiser 2010: 68; Kazansky 2017: 49-50; Weiss 2020: 128.
Now, one possible source of a short velarizing vowel in * uelVpi-is Proto-Italic * a from PIE * H. It is important to emphasize that Pike's *uel(-)hi-p- will get us to this protoform without any problems. But a different approach deserves to be explored, if only for the sake of those who are less comfortable with the uncontrollable use of root-extensions. I would like to return to an idea by E. Hamp86, who, in a characteristically succinct 1980 article, theorized that Lat. volup continues a compound of the root noun *uelhi- with the root *hiep-of Hitt. epp-z 'grabs, seizes', Lat. apiscor 'I seize', coepi 'I began', Ved. apa 'has reached', Old Albanian ep 'gives ' (LIV2 2 3 7) 87. Indeed, a compound *u elhi-hip-88 would be expected to undergo a simplification of double *hihi and the resulting *uelhip- (no longer transparent to the speakers) would start a life of its own as a neo-root; a nominal i-stem made from it would be directly reflected in Latin volup, as Hamp had argued. Hamp glossed this verbal governing compound (originally a bahuvnhi, of the same type as Gk. x8PyiV 'lustral water' < *'hand-washing') as "'desire-reaching' i.e. 'attaining one's wish'", hence 'pleasure', which is possible, given that the root *uelhi- appears to have developed the meaning 'to want' next to the arguably original meaning 'to choose' already in the protolanguage89.
6. So where does this leave us regarding Gk. sXno^ai? Without Lat. volup, the root * elp- appears completely isolated, but as mentioned above, *uelhi-p- with a root extension would not give Proto-Greek *uelp-, nor would Hamp's compound *uelhi-hip-. However, this does not mean that we must fall back on Benvenistian * uel-p- next to *uel-hi-, *uel-d- and *uel-gh-, the last two of which have been
86 Reported but rejected by de Vaan (EDLIL 689).
87 As M. Weiss reminds me, there is another possibility: volup could go back to *uolu-Vp-(i) which would lose the second *u before a rounded vowel, cf. *suesor > *suosor > soror 'sister' (Weiss 2020: 165). Under this analysis, the compound would consist of a Latin reflex of *uelhi-u- 'desire' and the Latin root ap-.
88 = Hamp's "*m el-Hep-": he did not take into consideration the evidence for the root-final laryngeal in PIE *u elh i-.
89 To judge by the meaning of Lat. volo, Goth. wiljan, Gk. (Doric) X^v, etc. Pike 2011: 89 n. 69 aptly adduces Lat. opto as a parallel for 'choose' > 'wish'.
argued above to be at least unnecessary90, especially since the whole picture is made less than plausible by the lack of evidence for the primordial *uel-. I propose instead to equate slno^ai with Lat. volup under Hamp's analysis.
The neo-root *uelh1p- 'take a wish (to heart), wish for, hope' (based on a nominal compound *uelh1-h1p-) would be expected to be used to make finite forms91. In athematic aorist forms such as 3 sg. *(h1)e-uelhp-t92 the internal laryngeal would be lost in a heavy CHCC cluster by Schmidt-Hackstein's law93, giving *(h1)e-uelp-t94. As a parallel, compare the compound *uelh1-bheh2- 'speak in riddles' made with a different root *uelh1- as the first member and the root *bheh2- 'speak' (Gk. as the second member95: used as a verbal root, *uelh1-bhh2- gives the neo-root *uelbh- (Ved. valh- 'to riddle', Lith. vilbinti 'to lure') without an internal laryngeal which was presumably lost in aor. * (h1)e-uelh1bh-t > * (h1)e-uelbh-t but not in the noun *uelh1bh-r ( ^ Gk. sls^aipo^ai)96. Similarly, the resulting root allomorph *uelp- (< *uelh1p-) is the basis for the nominal forms in Greek: slnig, slnrop^ 'hope', *(p)alnv6g,
*(F)alnalsog, *(F)slmaTog 'hoped for'; the aorist stem *uelp- / *ulp- forms a subjunctive *uelp-e/o- giving Greek slno^ai.
90 *uel-d- has been revised as (unrelated) *h1ueld- (cf. gglSrop 'desire') and * el- h- has been eliminated since an entirely different source for Arm. gelj 'longing, desire' has been proposed (*gheldh-s-o- 'desire', cf. Ved. grdh-'to covet', grtsa- 'desirous').
91 Just as in Indo-Aryan the neo-root gop- / gup- (from compound go-pa-'cowherd' > 'protector', see Scarlata 2001: 303-304) developed a full Averbo with perf. jugup-, p.p.p. gupta-, etc. (see EWAia 1.499-500).
92 If, contrary to the LIV2 reconstruction discussed in n. 12 above, this PIE (neo)root was atelic and did not make a root aorist, the laryngeal would still be lost in sigmatic *(h1)e-uelh1p-s- or its precursor (replaced in attested Greek by ^lrcica) by the same sound law, as M. Peters reminds me.
93 See Schmidt 1973; Hackstein 2002.
94 This laryngeal loss would of course not affect the nominal compound *uelh1-h1p- > *uelhp-(i-), the putative source of Lat. volup.
95 For the first member compare Lith. vilti, Latv. vilt 'trick, beguile'.
96 See AleW 1434 for the compound analysis and Goto 1995 for Gk. gle^aipo^ai. For another parallel, compare the structurally similar analysis I proposed for Gk. ^an- < *mmp- / *memp- < *(h1)e-menh1p-t 'seized' < *men- + hp- 'take by hand' without, unfortunately, knowing Hamp's article at the time (Nikolaev 2022).
97 A verbal abstract (cf. ^p- 'help', K^p 'doom', liy 'libation'), see n. 2.
References and abbreviations
Adrados, F. R., Bernabé, A., & Mendoza, J. 1996: Manual de lingüística indoeuropea. Vol. 2: Morfología nominal y verbal. Madrid: Ediciones Clásicas.
AleW = W. Hock et al. (eds.). Altlitauisches etymologisches Wörterbuch, Version 2.0. URL: https://alew.hu-berlin.de.
Ambrosini, L. 2012-2013 (2014): Le divinitá dei pocola deorum: un nuovo pocolom di voluptas del Volcani Group. RPAA 85, 337-363.
Balles, I. 2007: A Greek Laryngeal Metathesis That Needn't Be Either... Verba Docenti: Studies in Historical and Indo-European Linguistics Presented to Jay H. Jasanoff by Students, Colleagues, and Friends, ed. by A. J. Nussbaum, 15-24. Ann Arbor, New York: Beech Stave Press.
Bechtel, F. 1917: Die historischen Personennamen des Griechischen bis zur Kaiserzeit. Halle: Niemeyer.
Beek, L. van. 2022: The Reflexes of Syllabic Liquids in Ancient Greek: Linguistic Prehistory of the Greek Dialects and Homeric Kunstsprache. Leiden, Boston: Brill.
Beek, L. van, Migliori, L. 2019: Active versus Middle Perfect in Homeric Greek: Synchrony and Diachrony. The Paths of Greek: Literature, Linguistics and Epigraphy. Studies in Honour of Albio Cesare Cassio, ed. by E. Passa and O. Tribulato, 71-106. Berlin; Boston: de Gruyter.
Beekes R. S. P. 1969: The Development of the Proto-Indo-European Laryngeals in Greek. The Hague, Paris: Mouton.
Beekes, R. S. P. 1985: Skt. uttñná-, ZVS 98, 47-48.
Belardi, W. 2009: Elementi di armeno aureo. Vol. 3: Repertorio delle voci armene di origine indoeuropea. Formazione lessicale. Composi-zione. Elementi di morfologiapronominale. Roma: Calamo.
Benveniste, É. 1935: Origines de la formation des noms en indo-européen. Paris: Adrien-Maisonneuve.
Bugge, S. 1892: Beiträge zur etymologischen Erläuterung der armenischen Sprache. IF 1, 437-459.
Bugge, S. 1893: Beiträge zur etymologischen Erläuterung der armenischen Sprache. ZVS 32, 1-87.
Burrow, T. 1984: Vedic urvári; Lady of Choice, Wife. JRAS 1984/2, 209216.
Carruba, O. 1991: Searching for Woman in Anatolian and Indo-European.
Perspectives on Indo-European Language, Culture and Religion. Studies in Honor of Edgar C. Polomé, vol. 1, 155-181. McLean VA: Institute for the Study of Man.
Cassio, A. C. 1999: Epica greca e scrittura tra VIII e VII secolo a.C.: Madrepatria e colonie d'Occidente. Scritture mediterranee tra il IX e il VII sec. a.C. Atti del Seminario (Milano, 23-24 febbraio 1998),
ed. by G. Bagnasco and Gianni/F. Cordano, 67-84. Milano: Universita degli Studi di Milano.
CEG = P. A. Hansen (ed.) Carmina epigraphica graeca saeculorum VIII—V a. Chr. n. Berlin; New York: de Gruyter, 1983.
Chantraine, P. 1942: Grammaire homerique. Vol. 1: Phonetique et morphologie. Paris: Klincksieck (cited from the 2013 edition).
Chiatelli, E. 2022: The Long Augment in Homer: A Formula-Based Approach. University of Cambridge Ph.D. Dissertation.
CIRB = Struve, V. V., Tikhomirov, M. N, Gaidukevic, V. F., Dovatur, A. I., Kallistov, D. P., Knipovic, T. N. (eds.) Corpus inscriptionum Regni Bosporani. Moscow; Leningrad, 1965 Струве, В. В., Тихомиров, М. Н., Гайдукевич, В. Ф., Доватур, А. И., Каллистов, Д. П., Книпович, Т. Н. Корпус боспорских надписей. Москва; Ленинград, 1965.
Cowgill, W. 1978: The Source of Latin vis 'thou wilt'. Die Sprache 24, 2544 (= The Collected Writings of Warren Cowgill, ed. by J. S. Klein, 251-266. Ann Arbor; New York: Beech Stave Press, 2006).
Curbera, J. 2013: Simple Names in Ionia. Personal Names in Ancient Anatolia, ed. by R. Parker, 107-144. Oxford: Oxford University Press.
Debrunner, A. 1910: Review of E. Boisacq, Dictionnaire etymo-logique de la langue grecque, fasc. 1-3, Heidelberg, Winter, 1907-1909, Göttingische gelehrte Anzeigen 1910/1, 1-18.
Debrunner, A. 1954: "Das Augment n-," Festschrift für Friedrich Zucker zum 70. Geburtstag, 85-110. Berlin: Akademie Verlag.
Dede, F. 2013: Inomigreci in -ap e -mp. Roma: Il Calamo.
DELG = Chantraine, P., Dictionnaire etymologique de la langue grec. Histoire des mots. Paris: Klincksieck, 1968-1980 (cited from the 2009 edition).
DELL = Ernout, A., and A. Meillet, Dicti nnai e etym l giq e de la langue latine. 4th ed. Paris: Klincksieck, 1959.
Djahukian, G. B. 1967: Ocerki po istorii dopis'mennogo perioda armjanskogo jazyka [Studies in the History of Armenian in the Pre-Literal Period]. Yerevan: Armenian Academy of Sciences. Джаукян, Г. Б. 1967: Очерки по истории дописьменного периода армянского языка. Ереван: Издательство Академии Наук Армянской ССР.
Djahukian, G. B. 2010: Armenian Etymological Dictionary, ed. by V. Sarkisian. Yerevan: Asoghik.
Dunkel, G. E. 2014: Lexikon der indogermanischen Partikeln und Pronominalstämme. Heidelberg: Winter.
Ecker, U. 1990: Grabmal und Epigramm. Studien zur frühgriechi-schen Sepulkraldichtung. Stuttgart: Steiner.
EDLIL = de Vaan, M., Etymological Dictionary of Latin and the Other Italic Languages. Leiden; Boston, Brill, 2008.
EWAia = Mayrhofer, M., Etymologisches Wörterbuch des Altindo-arischen. Heidelberg: Winter, 1986-2001.
García Castillero, C. 2000: La formación del tema de presente primario osco-umbro. Vitoria: Universidad del País Vasco.
Garnier, R. 2010: Sur le vocalisme du verbe latin: étude synchro-nique et diachronique. Innsbruck: Institut für Sprachwissen-schaft der Universität Innsbruck.
Garnier, R. 2014: Nouvelles considérations sur l'effet Kortlandt. Glotta 90, 139-159 (= Scripta selecta: Études d'étymologie indo-européenne, 353-374. Paris: Les Cent Chemins, 2017).
Garnier, R. 2015-2016: Nouvelles réflexions sur les adjectifs en -idus du latin. Wékwos 2: 73-86 (= Scripta selecta: Études d' étymologie indo-européenne, 319-334. Paris: Les Cent Chemins, 2017).
Garvie, A.F. 2009: Aeschylus. Persae. Oxford; New York: Oxford University Press.
GEW = Frisk, H., Griechisches etymologisches Wörterbuch, Heidelberg: Winter, 1960-1972.
Goto, T. 1995: Griechisch èXe^aipo^ai. Kuryiowicz Memorial Volume, Part 1, ed. by W. Smoczynski, 365-370. Cracow: Universitas.
HAB = H. Acarean, Hayerën armatakan bararan [Armenian Etymo-logical Dictionary]. Erevan: University Press, 1926-1935.
Hackstein, O. 2002: Uridg. CH.CC > C.CC. HS 115, 1-22.
Hamp, E. 1980: Latin uolup(e). ZVS 94, 158.
Haröarson, J. A. 1993: Studien zum urindogermanischen Wurzel-aorist und dessen Vertretung im Indoiranischen und Grie-chischen. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
Hirt, H. 1927: Indogermanische Grammatik. Bd. 1. Heidelberg: Winter.
Hoekstra, A. 1965: Homeric Modifications of Formulaic Prototypes: Studies in the Development of Greek Epic Diction. Amsterdam: N. V. Noord-Hollandsche Uitgevers Maatschappij.
Höfler, S. 2017: Der Stier, der Stärke hat. Possessive Adjektive und ihre Substantivierung im Indogermanischen. University of Vienna Ph.D. dissertation.
Höfler, S. 2018: A Look over Lat. umerus 'shoulder'. Proceedings of the 29th Annual UCLA Indo-European Conference, ed. by D. M. Goldstein, S. W. Jamison and B. Vine, 129-146. Bremen: Hempen.
Hübschmann, H. 1899: Overview of H. Hübschmann, Armenische Grammatik. 1. Theil: Armenische Etymologie, Leipzig: Breitkopf & Härtel, 1897, Anzeiger zu IF 10, 41-50.
IEW = Pokorny, J., Indogermanisches etymologisches Wörterbuch. Bern: Francke, 1959.
Imberciadori, G. 2023a: Etymologische Untersuchungen zum System der tocharischen Adjektive. University of München Ph.D. dissertation.
Imberciadori, G. 2023b: "Zur Etymologie von lat. celeber 'verkehrsreich, belebt; berühmt', lat. celer 'schnell', und der Artikulation von frühlat. *-l- / _ e. Glotta 99, 66-92.
Jasanoff, J. H. 2003: Hittite and Indo-European Verb. Oxford; New York: Oxford University Press.
Jasanoff, J. H. 2021: Vedic dhaya-, citaya- and an Indo-Iranian Sound Law. HS 134, 166-185.
Kazansky, N. N. 2015: [Lack of Semantic Differentiation Among Roots and the Problem of Root Extensions in Proto-Indo-European]. Indo-European Linguistics and Classical Philology 19, 328-340. Казанский, Н. Н. 2015: Семантическая недифференцирован-ность корней и проблема расширителей в праиндоевропейском языке (*terHj-, *triHx-b- etc.). Индоевропейское языкознание и классическая филология 19, 328-340].
Kazansky, N. N. 2017: Ocerk fonologii latinskogo iazyka [An Outline of Latin Phonology]. Moscow: Russian Academy of Sciences Казанский, Н. Н. 2017: Очерк фонологии латинского языка. М.: Российская академия наук].
Kazansky, N. N. 2018: Root Extension as a Linguistic Problem. HS 131, 201-209.
Klingenschmitt, G. 1981: Albanisch und Indogermanisch. MSS 40, 93-131 (= Aufsätze zur Indogermanistik, ed. by M. Janda, R. Lühr, J. Matzinger and S. Schaffner, 217-242. Hamburg: Kovac, 2005).
Kocharov, P. 2018: A Comment on the Vocalization of Word-initial and Medial Laryngeals in Armenian. Farnah: Indo-Iranian and Indo-European Studies in Honor of Sasha Lubotsky, ed. by L. van Beek, A. Kloekhorst, G. Kroonen, M. Peyrot, T. Pronk and M. de Vaan, 123-136. Ann Arbor; New York: Beech Stave Press.
Kocharov, P. 2019: Old Armenian Nasal Verbs: Archaisms and Innovations. University of Leiden Ph.D. dissertation.
Kocharov, P., Shatskov, A. 2018: Towards a Classification of PIE root extensions. HS 131, 210-226.
Kölligan, D. 2019: Erkink' ew erkir: Studien zur historischen Grammatik des Klassisch-Armenischen. Hamburg: Baar.
Kortlandt, F. 1987: Sigmatic or Root Aorist. AAL 8, 49-52 (= Armeniaca: Comparative Notes, 79-82. Ann Arbor: Caravan Books, 2003).
Kortlandt, F. 1994: Palatalization of Dentals in Armenian. AAL 15, 27-31 (= Armeniaca: Comparative Notes, 104-106. Ann Arbor: Caravan Books, 2003).
Kostopoulos, G. 2014-2015: On Two Problems of Greek eüpov 'I found'. Die Sprache 51/2, 158-236.
Kümmel, M. J. 1998: Wurzelpräsens neben Wurzelaorist im Indogermanischen. HS 111, 191-208.
Kümmel, M. J. 2000: Das Perfekt im Indoiranischen: eine Unter-suchung der Form und Funktion einer ererbten Kategorie des Verbums und ihrer Weiterentwicklung in den altindo-iranischen Sprachen. Wiesbaden: Reichert.
Kümmel, M. J. 2006: Vedisch tand- und ein neues indoiranisches Lautgesetz. Indogermanica: Festschrift für Gert Klingenschmitt: Indische, iranische und indogermanische Studien dem verehrten Jubilar dargebracht zu seinem fünfundsechzigsten Geburtstag, ed. by G. Schweiger, 321-332. Taimering: VWT Verlag.
Kümmel, M. J. 2020: The Development of the Perfect within IE Verbal Systems: An Overview. Perfects in Indo-European Languages and Beyond, ed. by R. Crellin and Th. Jügel, 15-48. Amsterdam; Philadelphia: John Benjamins.
Lachnit, O. 1965: Elpis: eine Begriffsuntersuchung. University of Tübingen Ph.D. Dissertation.
Lejeune, M. 1972: Phonétique hist orique du mycénien et du grec ancien. Paris: Klincksieck.
Lejeune, M. 1974: Manuel de la langue Vénète. Heidelberg: Winter.
Leukart, A. 1987: Po-ro-qa-ta-jo, to-sa-pe-mo, a-mo-ra-ma and Others: Further Evidence for Proto-Greek Collective Forma-tions in Mycenaean and Early Alphabetic Greek. Minos 22-24 ( = Studies in Mycenaean and Classical Greek Presented to John Chadwick, ed. by J. T. Killen, J. L. Melena, J.-P. Olivier), 343-366.
Leumann, M. 1977: Lateinische Grammatik. Vol. 1: Lateinische Laut- und Formenlehre. München: Beck.
LfgrE = Lexikon des frühgriechischen Epos, ed. by Snell, B., Mette, H. J., and M. Meier-Brügger. Vol. 1-4. Göttingen: Vandenhoeck und Ruprecht, 1955-2010.
LIV2 = Lexikon der indogermanischen Verben. Die Wurzeln und ihre Primärstammbildungen, ed. by H. Rix, M. Kümmel, et al. 2nd ed. Wiesbaden: Reichert, 2001.
LPGN = A Lexicon of Greek Personal Names, ed. by P. M. Fraser, E. Matthews, and R. Parker. Oxford: Clarendon Press, 1987-.
Lubotsky, A. 1997: The Indo-Iranian Reflexes of PIE *CRHUV. Sound Law and Analogy. Papers in Honor of Robert S.P. Beekes on the Occasion of his 60th Birthday, ed. by A. Lubotsky, 139-154. Amsterdam; Atlanta: Rodopi.
Lubotsky, A. 2013: The Vedic Paradigm for 'Water'. Multi Nominis Grammaticus: Studies in Classical and Indo-European Linguistics in Honor of Alan J. Nussbaum on the Occasion of his Sixty-fifth Birthday, ed. by A. I. Cooper, J. Rau, and M. Weiss, 159-164. Ann Arbor; New York: Beech Stave Press.
Malouchou (MaXoûxou), G. E. 2022: XiaKà èmypaqiKa. Athens: Hellënikë Epigraphikë Hetaireia.
Martirosyan, H. 2010: Etymological Dictionary of the Armenian Inherited Lexicon. Leiden; Boston: Brill.
Meier, M. 1975: -iô-: Zur Geschichte eines griechischen Nominal-suffixes. Göttingen: Vanderhoeck und Ruprecht.
Meier-Brügger, M. 1981: Greek Xœ 'I wish'. Indo-European Studies 4, ed. by C. Watkins, 196-206. Cambridge, Mass.: Department of Linguistics, Harvard University.
Meillet, A. 1900: Étymologies arméniennes. MSL 11, 390-401.
Meillet, A. 1905-1906: Varia. MSL 13, 237-253.
Meillet, A. 1908: Introduction à l'étude compar ative des langues indo-européennes. 2nd ed. Paris: Librairie Hachette.
Meiser, G. 1986: Lautgeschichte der umbrischen Sprache. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
Meiser, G. 1993: Das Gerundivum im Spiegel der italischen Ono-mastik. Sprachen und Schriften des antiken Mittelmeerraums: Festschrift für Jürgen Untermann zum 65. Geburtstag, ed. by F. Heidermanns, H. Rix and E. Seebold, 255-268. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
Meiser, G. 2003: Veni, vidi, vici: die Vorgeschichte des lateinischen Perfektsystems. München: Beck.
Meiser, G. 2010: Historische Laut- und Formenlehre der lateinischen Sprache. 3rd ed. Darmstadt: Wissenschaftliche Buchgesellschaft.
Minon, S. 2023 (ed.): Lexonyme. Dictionnaire étymologique et sémantique des anthroponymes grecs antiques. Vol. 1 (A-E). Genève: Librairie Droz.
Neri, S. 2009: Review of D. Ringe, A Linguistic History of English. Vol. 1: From Proto-Indo-European to Proto-Germanic. Oxford; New York, Oxford University Press, 2006, Kratylos 54, 1-13.
Neri, S. 2017: Wetter. Etymologie und Lautgesetz. Perugia: Università degli Studi di Perugia.
Neri, S. 2021: Zur Wortbildung von lat. germänus 'leiblich; echt'. Studies in General and Historical Linguistics Offered to Jon Axel Hardarson on the Occasion of his 65th Birthday, ed. by M. Tarsi, 365-386. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
Nikolaev, A. 2014: An Etymological Miscellany. Indo-European Linguistics and Classical Philology 18, 708-717.
Nikolaev, A. 2020: Greek aprca^ 'robber; robbery'. IF 125, 33-40.
Nikolaev, A. 2021: Review of Neri 2017. Kratylos 66, 19-33.
Nikolaev, A. 2022: Greek fian- and its Congeners. Acta Linguistica Petropolitana 18: 267-281.
Nikolaev, A. 2023 : New Phrygian <€>ôiksç, Greek Qiyyavm (with Remarks on Miller's Law and the Treatment of *dhs in PIE). Indo-European Linguistics and Classical Philology 27, 858-879.
NIL = D. S. Wodtko et al., Nomina im indogermanischen Lexikon. Heidelberg: Winter, 2008.
Nussbaum, A. J. 1998. Two Studies in Greek and Homeric Linguistics. Göttingen: Vandenhoeck und Ruprecht.
Nussbaum, A. J. 1999: *Jocidus: An Account of the Latin Adjectives in -idus. Compositiones indogermanicae in memoriam Jochem Schindler, ed. by H. Eichner, H.-Chr. Luschützky and V. Sadovski, 377-420. Praha: enigma corporation.
Nussbaum, A. J. 2016: A Note on Latin Syllables and Anaptyxis. Tavet Tat Satyam: Studies in Honor of Jared S. Klein on the Occasion of His Seventieth Birthday, ed. by A. M. Byrd, J. DeLisi and M. Wenthe, 214-229. Ann Arbor; New York: Beech Stave Press.
Olsen, B. A. 1999: The Noun in Biblical Armenian: Origin and WordFormation, with Special Emphasis on the Indo-European Heritage. Berlin: de Gruyter.
Olsen, B. A. 2007: Three Latin Phonological Details. Proceedings of the 18th Annual UCLA Indo-European Conference, ed. by K. Jones-Bley, M. E. Huld, A. della Volpe, and M. Robbins Dexter, 101-115. Washington DC: Institute for the Study of Man.
Olsen, B. A. 2011: Review of Martirosyan 2010. Kratylos 56, 14-30.
Pedersen, H. 1906: Armenisch und die Nachbarsprachen. ZVS 39, 334-485 (= Kleine Schriften zum Armenischen, ed. by R. Schmitt, 112-263, Hildesheim: Olms, 1982).
Peters, M. 1980: Untersuchungen zur Vertretung der indogermanischen Laryngale im Griechischen. Wien: Verlag der Österreichischen Akademie der Wissenschaften.
Peters, M. 1987[1988]: Review of G. de Boel, Goal Accusative and Object Accusative in Homer. Brussel: AWLSK, 1988. Die Sprache 33, 285-292.
Pike, M. 2011: Latin -täs and Related Forms. University of California, Los Angeles Ph.D. Dissertation.
Rau, J. 2010: Indo-European Nominal Morphology: The Decads and the Caland System. Innsbruck.
Rix, H. 1976: Abstrakte Komplemente im Urindogermanischen. Studies in Diachronic, Synchronic, and Typological Linguistics: Festschrift for Oswald Szemer enyi on the Occasion of his 65th Birthday, ed by. B. Brogyanyi, 725-747. Amsterdam: John Benjamins.
Ruijgh, C. J. 1971: Review of DELG vol. 2, Lingua 28: 162-173 (= Scripta minora ad linguam graecam pertinentia, ed. by J. M. Bremer, A. Rijksbaron, and F. M. J. Waanders, vol. 1, 591-602. Amsterdam: Gieben).
Scala, A. 2017: Vittore Pisani e la continuazione delle dorsali indeuropee in armeno. Alessandria 11, 103-118.
Scarlata, S. 1999: Die Wurzelkomposita im Rg-Veda. Wiesbaden: Reichert.
Schmidt, G. 1973: Die iranischen Wörter für "Tochter" und "Vater" und die Reflexe des interkonsonantischen H (a) in den idg. Sprachen. ZVS 87, 36-83.
Schmitt, R. 1978: Die Iranier-Namen bei Aischylos. Wien: Verlag der Österreichischen Akademie der Wissenschaften.
Schumacher, S. 2007: Kontinuanten urindogermanischer Wurzel-aoriste im Albanischen. Teil 1: Wurzelaoriste mit frühur-albanischem Stamm auf Vokal oder auf *s. IJDL 4, 207-280.
Schwyzer, E. 1939: Griechische Grammatik, Vol. 1: Allgemeiner Teil. Lautlehre. Wortbildung. Flexion. München: C.H. Beck.
Seldeslachts, H., Swiggers, P. 2002: A propos de la flexion de latin velle. Donum grammaticum: Studies in Latin and Celtic Linguistics in Honour of Hannah Rosen, ed. by L. Sawicki and D. Shalev, 317328. Leuven: Peeters.
Solta, G. R. 1960: Die Stellung des Armenischen im Kreise der indogermanischen Sprachen. Wien: Mechitaristen-Buchdruckerei.
Sommer, F., Pfister, R. 1977: Handbuch der lateinischen Laut- und Formenlehre: eine Einführung in das sprachwissenschaftliche Studium des Lateins. Heidelberg: Winter.
Strunk, K. 1985: Flexionskategorien mit akrostatischem Akzent und die sigmatischen Aoriste. Grammatische Kategorien: Funktion und Geschichte. Akten der VII. Fachtagung der Indogermanischen Gesellschaft (Berlin, 20.-25. Februar 1983), ed. by B. Schlerath and V. Rittner, 490-514. Wiesbaden: Reichert (= Kleine Schriften, ed. by H. Hettrich, W. Hock, P.-A. Mumm and N. Oettinger, vol. 2, 666-690. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck, 2005).
Stüber, K. 2002: Die primären s-Stämme des Indogermanischen. Wiesbaden: Reichert.
Sturm, J. 2021: Nasal Presents from Homer to Attic Greek: Analogy and Reanalysis in the Greek Verb. Harvard University Ph.D. dissertation.
Szemerenyi, O. J. L. 1966: The Labiovelars in Mycenaean and Historical Greek. SMEA 1, 29-52 (= Scripta Minora: Selected Essays in Indo-European, Greek, and Latin, ed. by P. Consi-dine and J. T. Hooker, vol. 3, 1220-1243. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck, 1987).
Szemerenyi, O. J. L. 1990: Introduction to Indo-European Linguistics. Oxford: Clarendon Press.
Vine, B. 1982: Indo-European Verbal Formations in *-d-. Harvard University Ph.D. Dissertation.
Viredaz, R. 2018: Armenian Aorist and Imperfect Endings. Acta Linguistica Petropolitana 14, 163-268.
Wackernagel, J. 1907: Indisches und Italisches. ZVS 41, 305-319 ( = Kleine Schriften, Göttingen: Vandenhoeck und Ruprecht, 1955, vol. 1, 494-508).
Wackernagel, J. 1910: ölnvicxog. ZVS 43, 377-378 (= Kleine Schriften, Göttingen: Vandenhoeck und Ruprecht, 1955, vol. 1, 831-832).
Weisgerber, L. 1930: Die Sprache von Festlandkelten. Frankfurt.
Weiss, M. 2002: Observations on the South Picene Inscription TE 1 (S. Omero). Indo-European Perspectives, ed. by M. Southern, 251266. Washington, D.C.: Institute for the Study of Man.
Weiss, M. 2020: Outline of the Historical and Comparative Grammar of Latin. 2nd ed. Ann Arbor: Beech Stave Press.
West, M. L. 1990: Studies in Aeschylus. Stuttgart: Teubner.
West, M. L. 1998: Homeri Ilias. Stuttgart; Leipzig: Teubner.
Widmer, P. 2004: Das Korn des weiten Feldes. Interne Derivation, Derivationskette und Flexionsklassenhierarchie: Aspekte der nominalen Wortbildung im Urindogermanischen. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
Willi, A. 2018: Origins of the Greek Verb. Cambridge: Cambridge University Press.
Wyatt, W. F., Jr. 1972: The Greek Prothetic Vowel. Cleveland: American Philological Association.
Zair, N. 2012: The Reflexes of the Proto-Indo-European Laryngeals in Celtic. Leiden: Brill.
Ziegler, S. 2004: Uridg. *wenh1- '(hin)schütten, ausbreiten'. HS 117: 1-12.
