Abstract
The Luwian nominal class of common gender a-stems, which has been argued to reflect Proto-Indo-European stems in *-eh2-, has not yet been explored to a full extent. In Cuneiform Luwian, a number of action nouns in -a- c. derived from verbs in -i(ya)- (< *-i̯e/o-) have been noticed by scholars, but a larger analysis of all examples including Hieroglyphic Luwian is still lacking. It is the goal of the present paper to fill this gap. After a synchronic analysis of all examples, their stem formation will be compared to a similar derivational process found in Sanskrit. Consequently, the comparative evidence will shed light on the function of the suffix *-eh2- in Proto-Indo-European, including Anatolian.
1 Introduction
Recent papers on the nominal suffix *-eh2- in Anatolian have demonstrated that it is used productively in combination with thematic suffixes to build agent nouns or “individualizations”, e.g. *-ti̯o- + *-eh2- c. (Hajnal 2003: 193), *-lo- + *-eh2- c. (Sasseville 2014–15), *-si̯o- + *-eh2- (Sasseville 2018). In her extensive analysis of Lycian a-stems (< *-eh2-), Martínez Rodriguez (2018: 279–83) reemphasizes the “abstract” function of *-eh2-, previously known almost solely from Eichner’s (1973: 59) morphological analysis of Hitt. miyaḫḫuntešš-(mi) ‘to become old’ (< *mii̯-eh2-u̯ent-eh1-s-), in which the reconstructed noun in *-eh2-, i.e. *mii̯eh2- ‘growth’, represents a deverbal abstract noun derived from the Hittite verb māi-/mi(ya)-(ḫi) ‘to grow’. Martínez Rodriguez (2018: 280) establishes the existence of this derivational process in Lycian by adducing e.g. tama- c. ‘building’ (< PIE *doméh2- ← *dem- ‘to build’) and xupa- c. ‘grave (← heap)’, equivalent to Hitt. ḫūppa- ‘heap, pile’ (< PIE *h2upéh2-) derived from the non-attested Lycian cognate of the Hittite root verb ḫuwāpp-/ḫu(wa)pp-(ḫi) ‘to cast, to hurl down’.1
While preliminary investigations of the verbal abstract function of *-eh2- have been undertaken, there is yet no broader study of this derivational process. Based on examples cited above, Anatolian root verbs could generate a common gender action or result noun in *-eh2-. In his monograph on Luwian nominal morphology, Starke (1990: 63 & 584) points out a number of verbal abstract nouns in -a- c. that are derived from verbs in -i- (*-i̯e/o-), whose origins have never been addressed, e.g. tūmmanti-(ti) ‘to hear, listen’ → tūmmantiy-a- c. ‘obedience’. Therefore, it is the goal of this paper to explore first whether the Luwian suffix -a- c. that derives abstract nouns from verbs in *-i̯e/o- can go back to the Indo-European suffix *-eh2- and, second, whether there is a comparable derivational process outside of the Anatolian branch that could support its reconstruction for Proto-Indo-European.
2 Luwian verbal abstract nouns in -iy-a- c.
In this section, Luwian verbal abstract nouns in -iy-a- c. and their derivational bases will be extensively analyzed in both the Cuneiform and Hieroglyphic Luwian corpora. Sasseville (2014–15: 106–07) argues that the Luwian nominal class of non-mutated a-stems should be equated with the Lycian class of a-stems going back to common gender stems in *-eh2- (likewise Norbruis forthcoming). The lack of i-mutation in this class is the prime argument for the equation. The second argument is the Luwian dative singular in -a, which is equivalent to the Lycian locative singular in -a, both being confined to non-mutated common gender a-stems.2 Therefore, while investigating the Luwian suffix -iya- c., it will be important to pay attention, first, to whether it ever takes i-mutation and, second, with which dative singular ending it is inflected. In the remainder of this section, action nouns in -iya- c. with an attested derivational base will be discussed first, followed by those without an attested base. We begin with the clearest examples.
2.1 Tūmmantiya- c.
Luwian tūmmantiya- c. has been semantically compared to Hitt. ištamaššuwar (Laroche 1959: 99, followed by Melchert 1993: 233) and translated accordingly as ‘obedience’. In his morphological analysis, Starke (1990: 134) describes the Cuneiform Luwian substantive tūmmantiya- c. as an action noun derived from the verb tūmmanti-(ti) ‘to hear, listen’, which is in turn a denominative verb in -i-(ti) (< *-i̯é/ó-) derived from the noun tūmmant- n. ‘ear’.3 Regarding its occurrences, it is found solely in Hittite contexts, though sometimes marked with a Glossenkeil: nom. sg. tu-u-ma-an-ti-ia-aš (KUB 17.10+ iv 33), 
du-um-ma-an-te-ia-aš (KUB 17.20 ii 10, acc. sg. 
tu-um-ma-an-ti-ia-an (KBo 2.9+ i 49), tu-um-ma-an-ti-ia-an (KUB 24.1 iii 10′), dat./loc. sg. [t]u-um-ma-an-ti-ia (IBoT 3.83, 4′). These five attestations demonstrate that the stem-final -iya- undergoes neither i-mutation nor contraction and that the dative/locative singular ending is not -i, but -a. Therefore, tūmmantiya- c. can go back to a common gender abstract noun in *-eh2-.
2.2 Ārraḫḫaniya- c.
The following example, which is not mentioned by Starke (1990: 584), is arraḫḫaniya- c. This lexeme is found at least once in a Luwian context (acc. sg. a-ar-ra-aḫ-ḫa-ni-a[n] (KUB 35.109 iii 15)) and several times in Hittite contexts (nom. sg. 
ar-ra-aḫ-ḫa-ni-ia-aš (KBo 41.210, 12′), acc. sg. ar-ra-aḫ-ḫa-ni-ia-an (KUB 55.38 i 3), dat./loc. sg. ar-ra-aḫ-ḫa-ni-ia (KUB 50.26+ iii 34)). The common gender stem of arraḫḫaniya- shows no i-mutation in the nominative or accusative singular and has a dative/locative singular in -a, which points towards an original stem in *-eh2-. In a philological and linguistic analysis of the verb ārraḫḫani-(ti) and its derivatives, Rieken (2017) argues convincingly that ārraḫḫani-(ti) means ‘to curse, to grumble’ and its action noun ārraḫḫaniya- c. ‘cursing, crumbling’.4 Thus, the abstract noun ārraḫḫaniya- c. is a perfect example of a formation in -iy-a- derived from a verbal stem in -i-(ti) (< *-i̯é/ó-).
2.3 Tiwataniya- c.
The substantive tiwataniya- c. is attested only in Hittite contexts (gen. sg. ti-wa-ta!-ni-ia-aš (KBo 41.210 obv. 12′), dat. sg. [ti-wa-t]a-ni-ia (KBo 54.99+ iii 35)). It was not mentioned in Starke’s summary of action nouns in -iya- c. (1990: 584), although its derivational base is the well-known verb tiwatani-(ti) (Starke 1990: 147467).5 On contextual and etymological grounds, tiwatani-(ti) ‘to swear by the Sun-god, to utter a curse’ has been analyzed as a denominative of tiwat- c. ‘Sun-god’ (Watkins 1993: 470, Melchert 1993: 230, Rieken 2017: 242–43). Consequently, the action noun tiwataniya- c. should refer to the action of ‘swearing by the Sun-god, uttering a curse’. As can be seen from the two attested forms, no i-mutation occurs in the stem-final -iya- of tiwataniya- c. and the dative singular ending is -a, which allow us to posit an original nominal stem in *-eh2-.
2.4 Tarāwiya- c.
The noun tarāwiya- c. /trawiya-/ is attested exclusively in a Hittite context: nom. sg. 
ta-ra-a-ú-i-i[a-aš] (KBo 11.1 obv. 15), ta-ra-a-wi5-ia-aš (KBo 11.1 obv. 44), acc. sg. 
ta-ra-a-wi5-an (KBo 11.1 obv. 28). Goetze (1963: 60–61) connects tarāwiya- c. with the Luwian verb tarāwi-(ti), which he interpreted as ‘to enrich’ and, thus, suggests the meaning ‘wealth’ (followed by Starke 1990: 63). Laroche (1963: 246) likewise connects it with the Luwian verb tarāwi-(ti), which he translates as ‘to put down’ instead, and consequently suggests for the derivative tarāwiya- c. the meaning ‘calm’. However, Melchert (1992: 53) has convincingly argued that Luwian tarāwi-(ti) means ‘to hand over’, connecting it further with Lydian tro-(d) ‘to hand over’. Thus, he translates the action noun tarāwiya- c. as ‘control’; cf. the meanings of Hitt. maniyaḫḫ-(i) ‘to hand over, to manage, to control’. Considering the context in which tarāwiya- c. occurs, it definitely refers to something good that a land needs to prosper:
nu D10 EN-IAA-NA DINGIRMEŠ [m]e-˹ma˺-ú nu-kán DINGIRMEŠ KUR-TAM an-da ˹ták˺-šu-li-it IGIḪI.A-it ú-wa-an-du nu-kán ŠÀ-BI KUR-TI a-aš-šu
KBo 11.1 obv. 27–28ta-ra-a-wi5-an ták-šu-ul aš-šu-ul mi-i[a-tar]-ra i-ia-an-du
The Storm-god, my Lord, must speak to the (other) gods. The gods must look at the land with peaceful eyes. They must generate in the land goods, tarāwiyan, peace, goodness and growth.
From a Luwian perspective, tarāwiya- c. is clearly a derivative in -a- c. of the Luwian verb tarāwi-(ti) ‘to hand over, to give over (to a god)’. The verb can be used in the negative sense of ‘handing an enemy over’ (KUB 35.45 + KBo 29.3 ii 25–26) or in a positive sense of ‘handing (offerings) over to the gods’ (KUB 35.54 ii 35′–36′). Therefore, since it can be interpreted formally as an action noun with positive semantics, a meaning such as ‘the action of handing offerings over to the gods’ or in other words ‘(religious) commitment’ is suggested here. On the other hand, the meaning ‘control’ suggested by Melchert in the sense of ‘management’ or ‘administrative control (over the land)’ cannot be excluded. Until further attestations in different contexts are discovered, the passage cited above may be tentatively translated as:
They must generate in the land goods, commitment (or control?), peace, goodness and growth.
Based on the few attested forms, one can notice that neither i-mutation nor contraction occurs in the derivational sequence -iya- (< *-i̯-eh2-).
2.5 Lappiya- c.
The next example lappiya- c., adduced by Starke (1990: 584), occurs only in a Luwian context: acc. sg. la-ap-pí-ia-an (KUB 35.109 ii 7′), la-ap-pí-an (KUB 35.109 iii 14). It refers to something negative alongside ‘woe’, ‘blood’ and ‘wailing’. Its derivational base is surely the verb lappi-(ti), whose semantic assignment depends on the etymological connection with Hitt. lāpp- (mi) ‘to glow (?)’ (CHD L–N: 39–40, Kloekhorst 2008: 519). The same semantics are suitable for the Luwian verb:
nu-kán GIŠa-la-an-za-na-aš KI-LI-LIḪI.A […] nu kiš-an SÌRRU GIŠa-aš-ta-ra-la-an[(-) …] la-ap-pí-in-ta
KUB 35.142 iv 8′–10′
[They …] the crowns of alanza(n)-wood. (The crowd) sings in the following manner: “They made […] the throne glow.”
Regarding the action noun lappiya- c., a sense ‘glowing’ would be expected from a morphological perspective. In comparison with Hitt.? lappiya- c. ‘fever, embers’, a meaning such as ‘heat, glow (?)’ has been suggested (Starke 1990: 63, Melchert 1993: 126, Kloekhorst 2008: 519, Soysal 2017: 82 ‘glowing fire, embers’). This is appropriate for the Luwian context, which requires something negative that applies to a human being. As for its morphology, Luw. lappiya- can be reconstructed as an action noun in *-eh2- derived from a verb in *-i̯e/o-.
2.6 Walliya- / BONUS-li-ia- c.
The substantive walliya- c. is found once in a Hittite context and translated as ‘glory’ (Kloekhorst 2008: 948, HEG W–Z: 260): gen. sg. wa-al-li-ia-aš (KUB 19.13 i 48). It also occurs once in a Hieroglyphic Luwian inscription, in which servants incised someone’s name with glory or pride: abl./instr. BONUS-li-ia-ti (ANCOZ 8, § 7). The derivational base is surely the verb walli-(ti) ‘to raise’ attested in Cuneiform Luwian. Thus, walliya- c. can be interpreted as an action noun in -a- c. derived from the verb walli-(ti) ‘to raise’ with the meaning ‘raising, exalting’ → ‘glory’. However, the two attestations available do not tell us anything regarding the presence or absence of i-mutation in this noun. Furthermore, no dative/locative singular form is attested so far. Therefore, the morphological interpretation of this lexeme as an action noun in *-eh2- is dependent on the other examples presented here.
2.7 Wa/i+ra/i-ia- c.
In the Hieroglyphic Luwian corpus, we find the substantive wa/i+ra/i-ia- /warriya-/ c. ‘help, assistance’, which is an a-stem showing no contraction of the sequence /-iya-/: acc. pl. wa/i+ra/i-ia-zi-a (SULTANHAN, § 13), abl./instr. wa/i+ra/i-ia-ri+i (SULTANHAN, § 45). Its derivational base is the verb /warriya-(i)/ ‘to help’ (BOHÇA, § 7, 8, 11, 12), whose semantics were argued convincingly by Hawkins (1975: 139). Therefore, the noun /warriya-/ would refer to ‘the act(ion) of helping’ and, since the first attestation listed above shows absence of i-mutation, it would represent another example of a common gender action noun in /-iy-a-/ (< *-i̯-eh2-).
2.8 LEPUS+ra/i-ia- c.
In Hieroglyphic Luwian, the noun /tabariya-/ c. ‘authority’ is attested several times: nom. sg. [LE]PUS+ra/i-ia-sa (KARKAMIŠ A23, § 2), gen. sg. LEPUS+ra/i-ia-sa (BOROWSKI 2), abl./instr. ta-LEPUS+ra/i-ia-ti (BOROWSKI 3, § 5), LEPUS+ra/i-ia-ti (KÖRKÜN, § 3; MARAŞ 1, § 5). It is also found in full phonetic spelling in Hittite context as tapariya- c. ‘authority’: acc. sg. ta-pár-ri-ia-an (KUB 40.1 rev. 19, 21, 27), dat./loc. sg. ta-pa-ri-ia (KUB 14.7+ i 7, 15), unclear [t]a-pár-ia-aš-š[a(-) (KBo 64.256, 5′). Its derivational base is most likely a verb */tabari-(ti)/, which does not seem to be directly attested. The stem tapariye/a-(mi), which one encounters in the literature, can actually be taken as a verb of the ḫatrāi-class instead, i.e. tapariyā(i)-(mi) ‘(intr.) to command, decide’ (KUB 13.4 iv 9), ‘(tr.) to designate, appoint (a person or a house)’ (KUB 21.1+ i 65, Bo 86/299 ii 36. ii 94 (Bronze Tablet), KUB 2.2+ ii 48) and thus can be morphologically interpreted as a denominative derived from the noun tapariya- c. ‘command’. But one may assume a variant tapar-(ti) beside *tapari-(ti) and a derivation *tapari-(ti) → tapariya-. The same situation is found with kup-(ti)/*kupi-(ti) ‘to plot’ → kupiyat(i)- c. ‘plot’. Not only the dative/locative in -a, but also the lack of i-mutation in the nominative singular speak for /tabariya-/ c. as a common gender stem in -a- c. (< *-eh2-).
2.9 Lalḫiya- c.
The next example adduced by Starke (1990: 584) is lalḫiya- c. ‘campaign’, attested once in a Luwian context: acc. sg. la-al-ḫi-ia-an (KUB 25.39 iv 12). As its derivational base, Laroche (1959: 62) reconstructs a verb *lalḫi(ya)- ‘to march’, comparing Hittite cognates such as laḫḫiye/a-(mi) ‘to go on an expedition’ (followed by Melchert 1993: 123). Furthermore, a mediopassive stem in -i-(ti) is attested in a fragmentary Luwian context, i.e. laḫḫintari (KUB 35.79 iv? 12′). The base *lalḫi(ya)-, on the other hand, is reduplicated and must belong to a different stem class; cf. ḫišḫiya-(i)? ‘to bind’, ḫanḫaniya-(i) ‘to be perturbed’. Thus, there is enough comparative evidence to make the claim that Luw. lalḫiya- c. ‘campaign’ is an action noun in -a- c. (< *-eh2-) derived from a verb in -iya-(i); cf. /warriya-(i)/ above.
2.10 Zaršiya- c.
The last potential action noun in -iy-a- c., which was also adduced by Starke (1990: 37), is zaršiya- c. ‘guarantee, safe conduct’ found solely in Hittite contexts: nom. sg. 
za-ar-ši-ia-aš (KUB 14.3 ii 63), 
za-ar-ši-ia-aš (KUB 8.79 rev. 14), acc. sg. 
za-ar-ši-ia-an (KUB 14.3 ii 62), dat. sg. 
za-ar-ši-ia (KUB 14.3 ii 1), 
za-ar-ši-ia (KUB 14.3 ii 65), za-ar-ši-ia (KUB 8.79 rev. 4). This substantive shows a dative singular in -a and a lack of contraction or i-mutation in the sequence -iya-. Morphologically, it can be interpreted as an action noun derived from an unattested verb *zarši-(ti) ‘?’.6
Based on the preceding analysis, verbal abstract nouns in -a- c. are derived from verbs in -i-(ti) (< *-i̯é/ó-) or in -iya-(i) (cf. Hitt. -āi-/-i(ya)-(ḫi)). The large number of examples demonstrates that this process is highly productive in Luwian. As to the prehistory of the suffix -a-, the lack of contraction and of i-mutation combined with the dative/locative singular ending -a speaks for an origin in PIE *-eh2-. Consequently, there is now evidence for a derivational chain creating verbal abstract nouns in *-eh2- based on verbal stems in *-i̯e/o- in Anatolian. In the next section, a parallel derivational chain in another branch of the Indo-European family will be addressed.
3 The Sanskrit suffix -yā́- f.
The Sanskrit language of the Indo-Iranian branch has a suffix -ā́- f. going back to *-eh2-, which is productive in forming derivatives from both primary and secondary verbal stems. This includes deverbal derivatives such as desideratives in -ṣ-, e.g. śuśrūṣ- ‘to want to listen’ → śuśrūṣā- f. ‘obedience’, jigīṣa-te ‘to be greedy’ → jigīṣā́- f. ‘the wish of achieving, greed’, bhíkṣa-te ‘to desire to have a share for oneself’ → bhikṣā́- ‘the wish of begging’ (AiGr. II, 2: 242–43, Tucker 1988: 101), and root verbs, e.g. nind- ‘to rebuke’ → nind-ā́- f. ‘reproach’, rādh- ‘to succeed’ → -rādh-ā́- ‘success’, dhvṛ- ‘to cause to fall’ → dhvar-ā́- ‘causing to fall’ (AiGr. II, 2: 245–47). Furthermore, the same nominal suffix is productive with verbal stems in -ya- (< *-i̯e/o-), e.g. tṛ́ṣ-ya- ‘to be thirsty’ → tṛṣy-ā́- f. ‘thirst’ (cf. tṛṣyā́-vant- with Hitt. miyaḫḫ-unt-ešš-(mi)),7 dá-ya- ‘to take part in, to sympathize’ → day-ā́- f. ‘sympathy’, gṛ́dh-ya- ‘to be greedy’ → gṛdhy-ā́- f. ‘greed’, apasyá- ‘to be active’ → apasyā́- f. ‘activity’, mṛgáya- ‘to hunt’ → mṛgayā́- f. ‘hunt’, vacasyá- ‘to be audible’ → vacasyā́- f. ‘eloquence’ (AiGr. II, 2: 243–44), namasyá- ‘to do reverence’ → namasyā́- f. ‘the act of doing reverence’. All these derivatives designate the actions of their verbal bases and are thus genuine verbal abstract nouns. The derivational chain -yā́- f. in Sanskrit further belongs to a productive system including participles in -y-ánt- and agent nouns in -y-ú-, e.g. avasya- ‘to seek help’ → avasyánt- ‘seeking help’, avasyú- ‘seeking help, helping’.8
Because of the unclear morphemic analysis of certain examples, the morphological origins of these suffixes have long been obscure. In cases where roots end in -ay, it has been questioned whether the suffix was really -yā́- or -yú- or simply -ā́- or -ú-, e.g. tāyú- ‘thief’ ← Av. stay-át ‘stolen’, māyā́- ‘magical power’ and dur-māyú- ‘using evil craft’ ← mī- ‘to deceive’ (AiGr. II, 2: 469–70).9 However, on comparative grounds, the first example is actually a root ending in a laryngeal, whereas the glide is part of the verbal suffix; cf. OIr. táid, OCS tatь < *táh2-ti- m. ‘thief’, Gk. τήτη f. ‘lack’, Hitt. tāye/a-(mi) ‘to steal’ < *téh2-i̯e/o-. Even though the Indo-Iranian branch may have created a neo-root *teh2(i̯)- including the glide, the comparative evidence of Gk. τηΰσιος ‘in vain’ requires the reconstruction of the derivative *teh2-i̯-ú- already in the common prehistory of Greek and Sanskrit (LIV2: 616). The second example māyā́- has a highly debated etymology, and the analysis of the derivation depends strongly on the shape of the root. An etymological connection with the root *meh1- ‘to measure’ has been defended.10 In this case, one could assume that the derivational chain was *meh1-i̯éh2-, unless one posits a verbal formation in *-i̯e/o-, i.e. *meh1-i̯e/o-, as the derivational base of the action noun in -ā́- (and the agent noun in -ú-). However, independent evidence for a PIE present in *-i̯e/o- derived from this root would be welcome (cf. LIV2: 424). There is a different etymological connection with the root *mei̯h1- ‘to grow’,11 in which case the glide would indeed be part of the root, while the derivative in -ā́- (and -ú-) would be directly derived from the root verb. In any case, as long as the etymology of an example is not assured, it cannot be used to determine the origin of the underlying morphological process.
There is an existing view that the derivational chain described above owes its existence to a primary suffix -yā́- or -yú-, which was derived directly from verbal roots (AiGr. II, 2: 243). For example, Stüber (2012: 127–28, 2015: 113) takes as prototype Skt. sarvavidyā́- f. ‘all-knowing’ ← ved- (against older-looking vidā́- f. AiGr. II, 2: 247) and śayyā́- f. ‘warehouse’ ← śay- ‘to lie’ (against older śayā́- in śayā́su AiGr. II, 2: 247), because she posits that the origin of the abstract suffix *-i̯eh2- lies in the collective form of the neuter suffix *-i̯o-. However, sarvavid-yā́- is much more likely to represent a late formation instead of an inherited process, which would indicate that the suffix was metanalyzed from -y-ā́- to -yā́- within the history of Indo-Iranian. In the next section, we will turn to the origin of the process and address the Celtic data cited by Stüber.
4 The origin of the derivational process
On the basis of the Anatolian data, which was until now unknown to the wider Indo-European literature, the original derivational pattern can be more precisely reconstructed. First, it is known that the suffix *-eh2- was productive in deriving verbal abstract nouns from root verbs, e.g. *bheu̯gh- ‘to flee’ → *bhugh-éh2- ‘flight, escape’ (Gk. φυγή, Lat. fuga), *gu̯hen- ‘to strike’ → *gu̯hon-éh2- ‘the action of striking’ (Gk. φοναί ‘bloodbath’, Cz. ú-hona ‘wound’, both having developed to result nouns).12 Further examples of this type in Hittite, which would make a welcome addition, have never been properly sought. Kimball (2015) looks for the Indo-European τομός-type of noun formations in Hittite and ends up tracing all oxytone result nouns in -ā- to this type, e.g. parš- ‘to break’ → paršā- c. ‘crumb, morsel’, ḫarp- ‘to combine’ → ḫarpā- c. ‘heap’, but a consideration of the type in *-eh2- is completely missing in her analysis. Since Proto-Indo-European stems in *-o- and *-eh2- merged paradigmatically in Hittite, a distinction between the τομός- and the τομή-types will not be demonstrable without a Luwian or Lycian cognate.
Considering that root verbs co-existed with verbs in *-i̯e/o-, the suffix could have easily been extended first to deradical stems in *-i̯e/o-, e.g. *u̯l̥H-i̯é/ó- (Luw. walli-(ti) ‘to raise’) → *u̯l̥H-i̯éh2- (Luw. walliya- c. ‘raising, glory’) and gu̯l̥dh-i̯é/ó- (Skt. gṛ́dhya- ‘to be greedy’) → gu̯l̥dh-i̯éh2- (Skt. gṛdhyā́- f. ‘greed’),13 and second to denominatives in *-i̯e/o-, whose productivity is assured for both Luwian and Sanskrit. Furthermore, the extension of the abstract suffix *-eh2- to the verbal suffix *-i̯e/o- is supported by the parallel extension to Sanskrit desideratives in -ṣ- (< *-se/o-), e.g. bhikṣā́- ‘the wish of begging’ ← bhíkṣa-te ‘to desire to have a share for oneself’. Thus, abstract nouns in *-eh2- could be derived from various kind of verbal suffixes. The question then becomes which derivational process was present already in Proto-Indo-European and which must be considered an independent innovation of the two separate branches. The productivity of *-eh2- with denominative stems in *-i̯e/o- is not likely to be inherited despite its existence in Anatolian (though so far only in Luwian) and Indo-Iranian (only in Sanskrit), especially if one considers how difficult it is to reconstruct denominative verbs based on word equations for Proto-Indo-European. Therefore, this derivational process is more likely to be an independent innovation of Luwian and Sanskrit. On the other hand, the extension of the derivational process of *-eh2- suffixation from verbal roots to deradical verbs in *-i̯e/o- can be safely reconstructed for the protolanguage.
Before concluding, a similar suffix that has recently been discussed in Celtic will be addressed. Stüber (2015: 113–14) addresses the morpheme *-ii̯ā- within the Celtic branch and argues for the existence of an inherited Indo-European suffix *-i̯eh2- derived from verbal roots, e.g. OIr. rige ‘stretching’ (< *rig̑i̯eh2-) beside OIr. rigid ‘to stretch, rule’ (< *réi̯g̑-/rig̑-´). However, some of her examples clearly co-existed with a deradical verb in *-i̯e/o-, e.g. OIr. guide f. ‘plea, prayer’ < PIE *gu̯hedhi̯eh2- beside OIr. guidid ‘asks for, prays’, Gk. θέσσεσθαι ‘to ask, to be a suppliant’ and Av. jaiδiiemi ‘to beg’ < PIE *gu̯hedhi̯e/o-,14 OIr. fige f. ‘weaving’ < PIE *u̯egi̯eh2- beside OIr. -figther ‘is woven’ < PIE *u̯égi̯e/o-, OIr. slaide f. ‘strike’ beside OIr. slaidid ‘strikes’ < Proto-Celtic *sladi̯e-. These examples would speak for the existence of *-eh2- derived from deradical verbs in *-i̯e/o- in Celtic. Stüber nevertheless takes the Celtic derivatives in *-i̯eh2- that are found beside root verbs as prototypes of this derivational process. However, such an incomplete synchronic picture is also found in Anatolian, where a prehistoric verb in *-i̯e/o- must be posited beside the attested root verb (see Section 2.8). Consequently, instead of taking Proto-Celtic *CVC- : *CVC-i̯eh2- as the prototype of the derivational process, it would be preferable on morphological grounds either to restore the missing formation in *-i̯e/o- or to assume a metanalysis of the suffix from *CVC-i̯-eh2- → *CVC-i̯eh2-.
In sum, having established the process *CR̥C-i̯-eh2- for Luwian, Sanskrit and additionally Old Irish, its genesis can be posited for the Proto-Indo-European stage. The development of the morphological process can be postulated as follows:
| PIE | 1. *CoC-éh2- / *CR̥C-éh2- |
| PIE | 2. (deradical *-i̯e/o-) *CR̥C-i̯-eh2- (Luwian, Sanskrit, Old Irish) |
| Post-PIE | 3. (denominal *-i̯e/o-) *CERC-VC-i̯-eh2- (Luwian, Sanskrit) |
5 Conclusion
The derivational chain of deradical *-i̯e/o- + *-eh2- building verbal abstract nouns occurs in at least three branches of the Indo-European family, i.e. Anatolian, Indo-Iranian and Celtic. On the basis of the comparative data adduced above, the creation of this derivational chain is argued to have begun already in the Proto-Indo-European period. Two further developments of this derivational type can be observed in these languages. In Luwian and Sanskrit, the suffix *-eh2- was extended to denominal verbs in *-i̯e/o-, and then, in Sanskrit and Old Irish, it seems to have been metanalyzed to *-i̯eh2-. These two innovations must however be considered late and independent.
Acknowledgment
This paper was written within the frame of the eDiAna-project financed by the DFG (Deutsche Forschungsgemeinschaft).
See Eichner (2017: 283). On abstract nouns in *-eh2- in Anatolian and their existence in the prehistory of Hittite, see also the brief overview of Melchert (2014: 259).
On the paradigmatic appurtenance of the Luwian dative ending -a to the non-mutated a-stems, see Yakubovich (2015: 6.2).
Contra Starke (ibid.), the participial form tūmmātaīmmiš does not belong to the same verbal paradigm as that of tūmmanti-(ti) (see Melchert 1993: 232).
There are also a few “Hittitized” forms of the Luwian verb in Hittite contexts, i.e. ārraḫḫaniye/a-(mi); cf. [a-ar-r]a-aḫ-ḫa-ni-ia-ad-du (VSNF 12.124, 3′) and [a-ar-ra-]˹aḫ˺-ḫa-ni-ia-at (KBo 43.14, 5′).
The other forms in which the sequence -iya- is preserved, e.g. ti-wa-ta-ni-ia-at-ta (KUB 35.39 i 23, KBo 22.254 rev. 9), ti-wa-ta-ni-ia-am-ma-ti (KUB 35.58 ii 3′), reflect a factitive derived from the action noun in -iya- c. For a new classification of Luwian verbal stems in /-i-(ti)/, /-iya-(i)/ and /-iy-a-(ti)/, see Sasseville forthcoming.
HEG W–Z: 673–74.
Whether one takes tṛ́ṣ-ya-as a deradical formation or a denominal formation derived from the co-existing root noun is a more general problem within the reconstruction of the Proto-Indo-European language. However, it is likely that this ambiguity was also perceived by native speakers of Proto-Indo-European, which may have been the starting point for the extension of the derivational process: deradical → denominal.
Gotō 2013: 53 & 130–31. These formations also acquired a desiderative sense, whose origin has not always been clear. Tucker (1988) argues that the desiderative value comes from formations in -ṣánt-, -ṣú- and -sā́-. Furthermore, she claims that since the full system of parallel derivatives in -yáti, -yú-, -yā́- is productive only in the Saṃhitās, that this might indicate that the extension of -yú-, -yánt- and -yā́- to form new derivatives with double desiderative and non-desiderative value is a poetic feature (Tucker 1988: 11323).
Albino (2012) raises the possibility that dur-māyú- is a haplological form of *dur-māya-yú-.
KEWA II: 624 with further literature.
KEWA II: 624 with further literature. Cf. also Yakubovich (2010) for a different reconstruction of the root.
Rix (1979: 736). For their remarkable productivity in Ancient Greek, see Chantraine (1933: 18–26).
On the Proto-Indo-European etymology, see EWA: 474.
LIV2: 217. For the Old Irish verb, see Cowgill (1980: 49–59).
References
AiGr. II, 2: Wackernagel, Jacob & Albert Debrunner. 1954. Altindische Grammatik. Band II, 2. Die Nominalsuffixe. Göttingen: Vandenhoeck & Ruprecht.
Albino, Marcos. 2012. Vedisch durmāyú-. Münchener Beiträge zur Sprachwissenschaft 66/1: 7–10.
Chantraine, Pierre. 1933. La formation des noms en grec ancien. Paris: Champion.
CHD: Güterbock, Hans G. & Harry A. Hoffner, Jr. 1980–. Chicago Hittite dictionary. Chicago: The Oriental Institute of the University of Chicago.
Cowgill, Warren. 1980. The etymology of Irish guidid and the outcome of *gu̯h in Celtic. In Lautgeschichte und Etymologie. Akten der sechsten Fachtagung der Indogermanischen Gesellschaft Wien, 24.–29. September 1978, ed. Manfred Mayrhofer, Martin Peters, & Oscar E. Pfeiffer, 49–78. Wiesbaden: Reichert.
Eichner, Heiner. 1973. Die Etymologie von heth. mehur. Münchener Studien zur Sprachwissenschaft 31: 53–107.
Eichner, Heiner. 2017. Ein philologisch-sprachwissenschaftlicher Blick auf den Fortgang der lykischen Studien seit Emmanuel Laroche. In Hittitology today. Studies on Hittite and Neo-Hittite Anatolia in honor of Emmanuel Laroche’s 100th birthday. 5èmes rencontres d’ archéologie de l’ IFÉA, Istanbul, 21–22 novembre, ed. Alice Mouton, 277–99. Istanbul: Institut Français d’ Études Anatoliennes Georges Dumézil.
EWA: Mayrhofer, Manfred. 1992. Etymologisches Wörterbuch des Altindoarischen. Heidelberg: Winter.
Goetze, Albrecht. 1963. Review of H.G. Güterbock and H. Otten, Keilschrifttexte aus Boghazköi XI. Journal of Cuneiform Studies 17: 60–64.
Gotō, Toshifumi. 2013. Old Indo-Aryan morphology and its Indo-Iranian background. In co-operation with Jared S. Klein and Velizar Sadovski. Wien: Verlag der Österreichischen Akademie der Wissenschaften.
Hajnal, Ivo. 2003. “Jungluwisch”. Eine Bestandsaufnahme. In Licia e Lidia prima dell’ ellenizzazione. Atti del convegno internazionale, Roma, 11–12 ottobre 1999, eds. Mauro Giorgieri et al., 187–205. Roma: Consiglio Nazionale delle Ricerche.
Hawkins, J. David. 1975. The negatives in Hieroglyphic Luwian. Anatolian studies 25: 119–56.
HEG: Tischler, Johann. 1977–2016. Hethitisches etymologisches Glossar. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
KEWA II: Mayrhofer, Manfred. 1963. Kurzgefaßtes etymologisches Wörterbuch des Altindischen. A concise etymological Sanskrit dictionary. Band II: D–M. Heidelberg: Winter.
Kloekhorst, Alwin. 2008. Etymological dictionary of the Hittite inherited lexicon. Leiden/Boston: Brill.
Laroche, Emmanuel. 1959. Dictionnaire de la langue louvite. Paris: Maisonneuve.
Laroche, Emmanuel. 1963. Review of H.G. Güterbock and H. Otten, Keilschrifttexte aus Boghazköi XI. Orientalistische Literaturzeitung 58. 245–48.
Martínez Rodrigez, Elena. 2018. Revisiting gender and morphology in Lycian a-stem nouns. In 100 Jahre Entzifferung des Hethitischen. Morphosyntaktische Kategorien in Sprachgeschichte und Forschung. Akten der Arbeitstagung der Indogermanischen Gesellschaft vom 21. bis 23. September 2015 in Marburg, ed. Elisabeth Rieken, 275–87. Wiesbaden: Reichert.
Melchert, H. Craig. 1992. The third person present in Lydian. Indogermanische Forschungen 97: 31–54.
Melchert, H. Craig. 1993. Cuneiform Luvian lexicon. Chapel Hill: self-published.
Melchert, H. Craig. 2014. PIE *-eh2 as an “individualizing” suffix and the feminine gender. In Studies on the Collective and Feminine in Indo-European from a Diachronic and Typological Perspective, eds. Sergio Neri & Roland Schuhmann, 257–71. Leiden: Brill.
Norbruis, Stefan. forthcoming. The origin and spread of the ‘i-mutation’ paradigm and the prehistory of the Luwic nominal stem classes. Historische Sprachforschung.
Rieken, Elisabeth. 2017. Schimpfen und Fluchen im Luwischen. Zur Etymologie von kluw. arraḫḫani(ya)-. Indogermanische Forschungen 122: 241–51.
Rix, Helmut. 1979. Abstrakte Komplemente im Urindogermanischen. In Studies in Diachronic, Synchronic and Typological Linguistics, Festschrift für Oswald Szemerényi on the Occasion of his 65th Birthday, Part 2, ed. Bela Brogyanyi, 725–47. Amsterdam: John Benjamins B.V.
Sasseville, David. 2014–15. Luwian and Lycian agent nouns in *-é-leh2. Die Sprache 51: 105–24.
Sasseville, David. 2018. New evidence for the PIE common gender suffix *-eh2 in Anatolian: Luwian -ašša- (c.) and Lycian B -asa- (c.). In 100 Jahre Entzifferung des Hethitischen. Morphosyntaktische Kategorien in Sprachgeschichte und Forschung. Akten der Arbeitstagung der Indogermanischen Gesellschaft vom 21. bis 23. September 2015 in Marburg, ed. Elisabeth Rieken, 303–18. Wiesbaden: Reichert.
Sasseville, David. forthcoming. Anatolian verbal stem formation. Luwian, Lycian and Lydian. Leiden: Brill.
Soysal, Oğuz. 2017. A new join KUB 17.8 + Bo 6172: Mythological description of a natural disaster in Ancient Anatolia? NABU (Nouvelles Assyriologiques Brèves et Utilitaires) 2017/2: 80–83.
Starke, Frank. 1990. Untersuchung zur Stammbildung des keilschrift-luwischen Nomens. Wiesbaden: Harrassowitz.
Stüber, Karin. 2012. Kollektive Verbalabstrakta im Indogermanischen. Münchener Studien zur Sprachwissenschaft 66/1: 113–45.
Stüber, Karin. 2015. Die Verbalabstrakta des Altirischen. Bremen: Hampen.
Tucker, Elisabeth. 1988. Some innovations in the system of denominative verbs in early Indic. Transactions of the Philological Society 86: 93–114.
Watkins, Calvert. 1993. Some Anatolian words and forms. In Indogermanica et Italica. Festschrift für Helmut Rix zum 65. Geburtstag, ed. Gerhard Meiser, 469–78. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck.
Yakubovich, Ilya. 2010. Indo-European *mā ‘to grow’. In Indoevropejskoe jazykoznanie i klassicheskaja filologija. XIV. chtenija pamjati I.M. Tronskogo, ed. N.N. Kazanskij, 478–92. N.N. St. Petersburg: Nauka.
Yakubovich, Ilya. 2015. The Luwian language. Oxford Handbooks Online. https://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780199935345.001.0001/oxfordhb-9780199935345-e-18.