1
Yakubovich
The origin of the Luvian possessive adjectives.
Ilya Yakubovich
University of Chicago
Luvian, Lycian, Milyan and Carian use a set of related suffixes for the formation of
possessive adjectives (Luv. -assa/i-, Mil. -ese/i-, Lyc. -ehe/i-, Car. -s/ś-), which are further cognate
with the rare Hittite suffix -(a)ssa- used for the formation of derived nouns.
The etymology of
this morpheme is debatable. Melchert (1994: 77) and Hajnal (2000: 163-64) suggest that Luv.
-assa/i- and Lat. -ārius are ultimately derived from the Proto-Indo-European possessive adjective,
the suffix of which can be reconstructed as *-eh2so-. The Proto-Anatolian assimilation rule *-h2s> -ss-, which is introduced based on this single phonetically imprecise comparison, comes at a
very high price. One has to assume that Hittite lexemes such as pahs- ‘to protect’, nahsaratt‘fear’, tuhs- ‘to cut off’, and palahsa- ‘(a garment)’ owe their existence to the analogical restitution
of the “laryngeal”.
I believe that the Luvian possessive adjectives in -assa/i- and their Luvic relatives have no
direct cognates in Indo-European. They rather represent straightforward outcomes of the Anatolian
genitive endings *-osso and *-osyo modified by case attraction. The Indo-European comparanda of
these two morphemes are the rare genitive marker *-eso/-oso (Fortson 2004: 114, 129) and the
thematic genitive ending *-osyo. In terms of internal reconstruction, one can analyze both ProtoAnatolian endings as containing the common Indo-Hittite genitive ending *-os extended by
respectively the deictic particle *so and the relative particle *yo. The reflex of the first particle is
independently attested in Luvian as the extension -sa/-za of nom./acc. sg. n. nominal forms (cf.
Melchert 2003: 186-87).
The clearest evidence for the existence of Anat. *-osso comes from the Luvian genitives in
-assa attested in cuneiform transmission, which are usually dismissed as scribal errors or nonce
formations (1-2). Their closest external comparanda are the patronymics in -ehe and -ese attested
in Lycian and Milyan respectively. The Luvian genitives rendered as -(C)a-sa in the hieroglyphic
orthography can represent both /-as/ and /-assa/.
The reflexes of *-osyo are primarily known from Luvian hieroglyphic inscriptions, where
they are usually recorded as -(C)a-si. Their attestations are nearly restricted to the subject and
direct object noun phrases (3-4). Luvian cuneiform texts contain a number of forms in -assi,
although their morphological analysis is debatable. Luv. is(sa)ra/i- < Anat. *g’esr- ‘hand’ suggests
that the gemination of Hittite /s/ at the syllabic boundary can be reconstructed not only for Hittite,
but also for Luvian. If so, the etymological sequence *-osyo may have yielded *-assya and then,
after the apocope, *-assi.
One of the features of the Anatolian Linguistic Area is case agreement in the possessive
construction (Luraghi 1993). “Suffixaufnahme”, or double marking of the case of the possessum
and the case of the possessor on the dependent noun, constitutes a well-known peculiarity of the
Hurro-Urartian languages (5). Hittite case attraction (“partitive apposition”) represents an attempt
2
Yakubovich
to copy the areal construction with “Suffixaufnahme” with the means available in an inflectional
language. As a result of case attraction, the possessor appears not in the expected genitive case, but
rather assumes the case marking of its head-noun (6). This morphosyntactic phenomenon is never
attested in the subject NPs.
The difference between contact-driven case attraction in Hittite vs. Luvian is that the
Luvian case agreement marker does not replace the original genitive case ending, but is rather
appended on top of it. This results in the variation between genitive case nouns and secondary
possessive adjectives. When such alternations can be observed in one and the same text, the
subject NPs contain the genitive case noun, while the construction with possessive adjectives is
reserved for the oblique NPs (7). Nevertheless, the Luvian possessive adjectives begin to spread
across the board in hieroglyphic texts, frequently replacing genitive case nouns even in the subject
NPs. In Kizzuwatna Luvian, where the impact of Hurrian must have been stronger, one can see
only isolated vestiges of the original construction with the possessor in the genitive.
Thus the original Luvian construction apassa parnanza (or apassi parnanza) ‘his house’
underwent case attraction in the oblique cases to become (instr.) apassati parnati ‘from his house’,
while the paradigm of the newly-formed possessive adjectives was eventually completed with the
nominative, hence apassanza parnanza ‘his house’. The rare instances of Hittite nominal
derivatives in -assa- are to be explained through case attraction followed by head noun ellipsis.
Thus the original *hapassa DUG ‘vessel for river(-water)’ could undergo case attraction in oblique
cases (e.g. abl. *hapassaz DUG-az), and then the disappearance of the word for ‘vessel’ would
lead to the grammaticalization of the new nominal stem
(DUG)
hapassa-. This must have happened
before the generalization of the unextended genitive ending *-os > -as in Hittite. The mechanism
of hypostasis triggered by case attraction has been described in detail in Yakubovich 2006: 44-45
in connection with the genesis of Éarzana- ‘inn’.
The vocalism of Lyc. -ahe/i- and Mil. -ase/i-, the secondary allomorphs of Lyc. -ehe/i- and
Mil. -ese/i-, cannot be used as an argument against the suggested reconstruction. Hajnal (2000:
170-171) has shown that the possessive adjectives in -ahe/i- and -ehe/i- are equally well attested in
Lycian, and furthermore that there is a good correlation between the stem type and the quality of
the initial vowel in the possessive suffix. The thematic stems usually take the variant -ehe/i-, while
the allomorph -ahe/i- is particularly frequent with lexemes that end in -a in the nominative.
Therefore, the distribution between the two allomorphs of the Lycian and Milyan possessive
adjectives can be trivially derived from the stem vowel generalization.
3
Yakubovich
EXAMPLES
(1)
KBo 4.14 ii 57-59, CTH 123 (NH/LNS), Stefanini 1965: 42
nasma=kan LUGAL-us tūwali KASKAL-si arha paizzi nasm=at GIM-an asān
imam ¶mar-ša-aš-ša mēhur
‘… Or (when) the king goes to a long campaign, or whatever time of calamity?
there may be…’.
(2)
KUB 35.54 iii 6-8, 758.1.C (MS), Starke 1985: 68
a-ta a-ap-pa za-aš-ta-an-za DINGIR.MEŠ-an-za pár-ra-an ni-iš <<???>>
im-ma-ra-aš-ša dIŠKUR-aš-ša-‰an™-za ha-li-iš-ša
‘Let the halis- of the Storm-God of an Open Country not … before these gods’.
(3)
KARATEPE 1, § 51 (Hu.), Hawkins 2000: 55
pi-ia-tu-há-wa/i-tu-u (DEUS)TONITRUS-hu-za-sá ARHA u-sa-nú-wa/i-mi-sá
za-si-há-wa/i |(“CASTRUM”)há<+ra/i>-na-sá-si DEUS-ní-zi
(LITUUS)á-za-ti-wa/i-tà-ia || “LONGUS”-ta5-ia (DIES)há-li-ia mi-ia-ti-zi-ha ||
(ANNUS)u-si-zi sa-na-wa/i-sá-ha-wa/i || tá-mi-hi-sá
‘And may Tarhunt the highly blessed and this fortress’s gods give to him,
to Azatiwada, long days and many years and good abundance’.
(4)
KARKAMIŞ A 2+3, § 16, Hawkins 2000: 109
|za-ti-pa-wa/i |kar-ka-mi-si-za(URBS) (DEUS)TONITRUS-ti-i Ika-tu-wa/i-sa
|REGIO-ni-ia-si |DOMINUS-ia-sa REL-i-zi |(“*273”)wa/i+ra/i-pa-si
|DOMINUS-ia-zi-i pi-ia-tá
‘The “lords of lore” whom Katuwa the country-lord gave to this Karkamisean
Tarhunt…’
(5)
KBo 20.129 ii 56-57, CTH 777.5 (MH/MS), Wilhelm 1995: 118
hi-pa-ap-te-ni-ta
d
hibapte=ne=da
Šauška=ve=ne=da
hibapte=RltSg=Dir
Sawoska=GEN=RltSg=Dir
GAŠAN-ka-bi-ne-da
‘…to the hibapte of the goddess Sawoska’.
4
Yakubovich
(6)
KBo 6.2 i 24, CTH 1.a.A (OH/OS), Hoffner 1997: 26
takku LÚ. U19.LU-as (gen.) ELLAM-as (gen.) KIR14=set (acc.) kuiski wāki
KBo 6.3. i 33, CTH 1.b.A (OH/NS?), Hoffner 1997: 26
takku LÚ.U19.LU-an (acc.) ELLAM (acc.) KIR14=set (acc.) kuiski wāki
‘If someone bites off the nose of a free person…’.
(7)
KARABURUN § 9-10, Hawkins 2000: 481
si-pi-sa-pa-wa/i ni-ia-sa (gen.) REL-ti si-pi-ia REX-ti MALUS-za CUM-ni ||
za<+ra/i>-ti-ti ni-mu-wa/i-zi ni-pa-wa/i ha-ma-si si-pi-ia-pa-wa/i-ta ni-ia-sá-na
(poss. adj.) hara/i-na-wa/i-ni-sa(URBS) (DEUS)ku+AVIS-ia ku-ma-pi ta-wa/i
INFRA-ta á-za-tu
‘But if Sipis, (son) of Nis, wishes evil for Sipis the king, for (his) son or grandson,
for Sipis, (son) of Nis, may the Haranean (god) together with? Kubaba swallow
down the eyes’.
BIBLIOGRAPHY
Fortson, Benjamin W IV. 2004. Indo-European Language and Culture: an Introduction. Oxford:
Blackwell.
Hajnal, Ivo. 2000. “Der adjektivische Genitivausdruck der luwischen Sprachen (im Lichte neuerer
Erkenntnis)”. 125 Yahre Indogermanistik in Graz. Ed. M. Ofitsch and Ch. Zinko. Graz:
Leykam. Pp. 160-84.
Hawkins, J. David. 2000. Corpus of Hieroglyphic Luvian Inscriptions. Volume I. Part I, II: Texts;
Part III: Plates. Berlin-New York: W. de Gruyter.
Hoffner, Harry A. 1997. The Laws of the Hittites: a Critical Edition. DMOA 23. Leiden: Brill.
Luraghi, Silvia 1993. “La modificazione nominale nelle lingue anatoliche”. Archivio Glottologico
Italiano 68 (2): 145-166.
Melchert, H. Craig. 1994. Anatolian Historical Phonology. Amsterdam-Atlanta: Rodopi.
________ 2003.
Starke, Frank. 1985. Die keilschrift-luwischen Texte in Umschrift. StBoT 30. Wiesbaden:
Harrassowitz
Stefanini, Ruggero. 1965. “KBo IV.14 = VAT 13049”. Atti della Accademia Nazionale dei Lincei.
Series 8, Volume 20, pp. 39-79.
Wilhelm, Gernot. 1995. “Suffixaufnahme in Hurrian and Urartian”. Pp. 113-35 in Plank 1995.
Yakubovich, Ilya. 2006. “The Free-standing Genitive and Hypostasis in Hittite”. JNES 65/1: 3949.
Yakubovich
5