Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Dipartimento
di Studi
Letterari
ANNALI
Linguistici
e
Comparati
del Dipartimento di Studi
Letterari, Linguistici e Comparati
Sezione linguistica
ANNALI
sezione
linguistica
AIΩN
N.S. 5
2016
ESTRATTO
UNIVERSITÀ D E G L I S T U D I D I N A P O L I
“L ’O R I E N TA L E ”
A I ΩN
N.S. 5
I S S N 2281-6585 2016
ANNALI
del Dipartimento di Studi
Letterari, Linguistici e Comparati
Sezione linguistica
AI ΩN
N.S. 5
2016
4 Indice
ISSN 2281-6585
Registrazione presso il Tribunale di Napoli n. 2901 del 9-1-1980
web: www.aionlinguistica.com
e-mail: redazioneaion@unior.it
Per la redazione delle proposte i collaboratori sono invitati ad attenersi con cura alle ”norme”
disponibili nel sito della rivista.
Le proposte di pubblicazione inviate alla rivista vengono valutate da revisori anonimi. A tal fine
una loro copia dev’essere priva di qualunque riferimento all’autore.
ANNALI
del Dipartimento di Studi
Letterari, Linguistici e Comparati
Sezione linguistica
AI ΩN
N.S. 5
2016
PROPRIETÀ RISERVATA
INDICE
Abstract
This paper argues that the traditional derivations of Luwian gurta- ‘city,
town’ (*gh3dhó- / *ghórto-; *gw3Hto-) are phonologically doubtful and the
loanword theories are similarly flawed. Instead, it proposes to return to
Sturtevant’s quickly abandoned theory (PIE *kw3tó-) fitting both
phonologically and semantically. As a by-product the etymology of the name
of Gordion will be also discussed.
1. The problem
2. *gh3dhó- / *ghórto-?
The most widespread view assumes that gurta- is the Luwian reflex of
PIE * gh3dhó- (cf. Skt. g3há- ‘house’, YAv. gərəδa- ‘dwelling place of the
daevas’ and also Goth. gards ‘house’, Lith. ga9das ‘fence, enclosure, stall’,
OCS gradъ ‘wall, town, garden’, Alb. gardh ‘hedge, fence’, etc.)1.
Nevertheless, as most recently Kloekhorst (2008: 495) rightly pointed out,
the Luwian reflex of *C3C is CarC-, thus this etymology cannot be upheld
(Kloekhorst classifies it only as ‘problematic’)2.
Melchert (1994: 260, 2012: 210) believes that next to the above
quoted regular reflex there is some evidence for PA *3 > Luwian ur
as well (Szemerényi, 1955: 72 even considered it as regular),
however, this evidence is extremely meagre constituting of only two
words3: gurta- itself and Hier.-Luwian zú+ra/i-ni ‘horn (pl. acc.)’
(ASSUR letter f+g §36) from *%3n°. However, considering that the
same root appears as zarwani(ya)- ‘of a horn’ in Cuneiform Luwian
(KUB 35.42, 2; KUB 35.43, 13), one is forced to assume that the <zú>
sign must have had a value <za x> as well (proposed in Simon, 2008,
see there for a detailed discussion; not quoted in Melchert, 2012) 4.
1 Szemerényi, 1955: 72; Georgiev, 1958: 17; GEW 1114 (only as a possibility);
Melchert, 1994: 260, 2012: 213-214; EIEC 199; Mallory & Adams, 2006: 221; Mottausch,
2011: 80. Before the identification of its Luwian origin it was of course considered as the
Hittite reflex of this PIE root.
2 See already Heubeck, 1961: 60-61; Durante, 1968: 57 n. 17; Tischler, 1983: 659;
1999: 359-360, who assumes that this is a loanword from Luwian and explains it from
*%orh2-%3h2 (cf. PIE *%erh2- ‘mischen’). However, as Kloekhorst (2008: 1032) pointed out,
it cannot be proved that this is indeed a loanword from Luwian and Rieken based her
etymology on the irregular “reflex” of PIE *3, which is methodologically doubtful.
4 Melchert’s recent explanation from *%3ng-id-, cf. Skt. ś4ṅga- <%3n-g (w)-o- (2012:
212, 213) is attractive, but as Steer (2014) points out, this is not a regular PIE
derivation. Melchert (2012: 210) adduces Hitt. karkidant- ‘horned’, which, put aside its
unclear meaning (cf. Puhvel, 1987: 89), cannot be connected due to obvious
phonological reasons. A simple derivation *%3n-ó- (cf. PGerm. *hurna- > Goth. haúrn,
etc.) → *%3n-id-, however, is entirely regular. Lipp’s suggestion (2009: 299-302), an
Indo-Aryan word through Hurrian borrowing is phonologically entirely arbitrary in
all its steps (ś4ṅga- → Hurr. *š/surŋg-i- → Anatolian substitution with *-ŋŋ- simplified
to -n-; note that the substitution of *3 with /ur/ in Hurrian is contradicted by Artatāma
from Çtá-dhāman-, EWAia/I: 255 with refs.), and in view of the proposal above,
unnecessary. For further criticism see also Mottausch, 2011: 80-81 and Melchert, 2012:
210, who points to the sibilant in Lipp’s scenario instead of the attested affricate:
Lipps’s etymology is based on the misreading of the Hieroglyphic Luwian sign *448
(<sù> instead of <zú>, see already Melchert, 1987: 201-202; cf. most recently Simon,
2008: 21-26).
5 Čop’s suggested solution, “*gh°rdho-”, “mit Schwa secundum gerundeter Art”
(1979: 15) is phonologically flawed, since schwa secundum appears in PIE only between
two obstruents (Schindler, 1977: 31; Mayrhofer, 1986: 175-177; Byrd forthcoming). This
restriction is not that strict in the daughter languages, but even in these cases schwa
secundum does not appear before a syllabic resonant, since by definition it cannot
appear there.
6 Followed by Braun, 1935-1936: 388; Walde & Hoffmann, 1938-1954/I: 660; Feist,
1939: 198; Szemerényi, 1942: 15; Kretschmer, 1951: 11 n. 1; Devoto, 1962: 492.
7 Kloekhorst (2008: 495) reconstructs these roots with *“h (so did Melchert, 1994: 260,
who has, however, retracted it in 2012: 210 n. 11). Nevertheless, this assumption is
contradicted by the Balto-Slavic, Sanskrit and Albanian data (Beekes’s suggestion [2010:
1645] of depalatalization is entirely ad hoc) and required only by the gratuitous and
semantically weak assumption that this word is connected to Skt. hárati ‘to bring, carry’
(a root attested only in Indo-Iranian, cf. LIV2: 177), so it can be easily abandoned (cf. also
GEW 1114; Kluge – Seebold, 2002: 332).
3. *gw3Hto-?
The other popular hypothesis (Bomhard, 1976: 220; Bader, 1991: 127;
Kimball, 1999: 250) suggested to derive gurta- from *gw3to- from *gwer-
‘mountain, height’ (cf. Skt. girí-, YAv. gairi-, OCS gora ‘mountain’, Lit.
girià ‘woods’, Alb. gur ‘rock, stone’). Puhvel (1997: 276) again did not
bother to argue, instead labelled it as a “shot in the dark”. However,
Kloekhorst (2008: 495) rightly pointed out that the correct root sounds
*gwerH- (cf. also Lipp, 2009: 300, who derives gurta- from *gw3h1tó- ‘mit
Berg / Anhöhe versehener Ort’), but the consequences of this have not
been pursued further (for which see below).
Melchert (1994: 260) first rejected this etymology claiming that it is
falsified by the preserved <g->, however later (2012: 214) said that this is
not a real problem, since “the labiovelar could have been unrounded by
the round anaptyctic vowel before being deoccluded: *gw3(H)to- > *gworto-
> gurta- (read /gorta-/)”. While the delabialization in such an environment
is plausible (though gratuitous), Melchert again operated with the false
reflex of *3, thus his remark does not provide an explanation for the
preservation of the initial consonant. However, since the loss of the
guttural part of the labiovelar in initial position is based exclusively on
prevocalic evidence (*gwon° > wāna- ‘women’, *gwow° > wawa/i- ‘cow’,
*gwhen° ‘to strike’ > wi/enal- ‘stick’, Melchert, 1994: 254, 2012: 212-213 with
refs.), one may surmise that the reflex was different before syllabic
resonant (assumed also by Lipp, 2009: 300-301: *gw3- > *gwur- > gur-),
though one may expect a similar behaviour as well.
Nevertheless, no example with *3 is known, but exactly this was
proposed by Melchert (2012: 212-213) in case of *‚. His two examples
are, however, problematic. One of them has nothing to do with words
with initial *gw8, and the other one, HLuwian REL-za- /kwanza-/ ‘to
incise, carve’, has now a competing etymology:
Kloekhorst’s analysis (2006: 97-101): *h2w- is not one of the sources of Lyc. q, but the
only one and thus qãn- has nothing to do with words with *gw-. Based on this example
Yakubovich (2013-2014: 294 n. 1) even assumes a Proto-Luwic change *gwh- > kw- that
would solve our problem as well if it were correct.
Melchert (2012: 213) derived it from *gwh‚s%e/o- ‘to strike’ (cf. also
Tischler, 1983: 604-605 with refs.) and as he rightly points out, the Luwian
form cannot reflect the e-grade since in that case the guttural would have
disappeared. It must therefore reflect the zero grade, and accordingly,
one must assume the preservation of the labiovelar at least in front of *‚.
The second – and in my view, more convincing – proposal was put
forward by Yakubovich (2013-2014: 285-287), who derives it from PIE
*kwels- ‘to draw’ (via a change [lts] > [nts] in the Hattusha & Iron Age
Luwian dialect), which has the advantage that it provides an
explanation for the plausible cognate Hittite gulš- ‘to carve, inscribe’
as well. In this case there is no evidence for the Luwian reflex of *gw in
front of the syllabic resonants.
If one prefers to follow the former explanation, it provides an
opportunity to extend this sound law to every syllabic resonant and
thus explain the preservation of the guttural, but unfortunately there is
no evidence or counter-evidence for that. Nevertheless, the Luwian
reflexes of the verb *kwer- ~ *kw3- show /kwar-/ ~ kūr- ‘to cut’ which
means that *gw3- could also lead to *gur-, if it was not preceded by the
degutturalization. Fortunately, the date of the change *CwÇC- > *CuRC-
is known and it is a Proto-Anatolian change (Melchert, 1994: 55-56) and
thus *gw3C° would have regularly lead to *gurC° already in PA.
However, there is a phonological circumstance that practically
excludes this derivation. As Kloekhorst, 2007 demonstrated, the
reflexes of *CwÇCV- and *CwÇCC- are different: while the former yields
*CuRCV-, the latter one results in *CwaRCC-. Since the root in question
contained a laryngeal, as Kloekhorst emphasized (see above), the
assumed form *gw3Htó- should have lead to †<kuwarta->. The only
possibility to escape this conclusion is to assume either that unlike
*CwÇCV- > *CuRCV- the *CwÇCC- > *CwaRCC- change was not PA but
has different reflexes in different languages (which is implausible)9, or
9 Would Luwian walwa/i- mean ‘lion’ and originate in PA *wDgwo- <*wDgwo- < PIE *wCkwo-
‘wolf’ as it is generally assumed since Steinherr 1968 and Lehrman, 1978: 228-230, 1987 (e.g.
Melchert, 1993: 254, 1994: 127, 360; EIEC 646; Mallory & Adams, 2006: 138; Yakubovich, 2010:
97), it would provide a direct confirmation of the PA date of *CwÇCC- > *CwaRCC-. However,
the meaning is not secure and the derivation is problematic, see the detailed discussion in
that the laryngeal in *gw3Htó- was lost before the *CwÇCC- > *CwaRCC-
change, which, however, cannot be proven now. All in all, the
phonological circumstances are not favourable for this etymology.
Finally, Melchert (2012: 214) believes that the “insurmountable”
obstacle is semantics: the root means ‘mountain’, while Melchert duly
argues that gurta- means only ‘city, town’ and not ‘acropolis, citadel’,
otherwise it would not have appeared in the collocation šarāzzi gurti
‘into the upper citadel/walled city’ (KBo 5.6 iii 33). This semantic shift,
however, is not “insurmountable”. A nice parallel can be provided by
Middle Persian kalāγ <kl’k> ‘city’ (cf. also Arm. k‘alak‘) <*kalXk+- (if the
derivation from PIE *kolH- / *kCH-, that served as a root for various
derivations meaning ‘hill’ by Klingenschmitt, 1980: 23-24 with n. 10 and
refs., is correct10). A more precise definition of the meaning of gurta- as
‘city, town’ is, however, an important result of Melchert’s article.
4. A loanword?
Kloekhorst, 2008: 951. Note furthermore the Lydian PN Walwe-, which was considered to be
a Luwic loanword from the above mentioned form by Yakubovich (2010: 97) due to the
specifically Luwian change PIE *kw > *gw > w, which is, however, not possible, see
Kloekhorst’s quoted discussion. Therefore, if the Lydian name is indeed connected with the
Luwian word, they both can represent a PA word *w(a/o/e)lwo- vel sim. of unknown meaning.
10 Cf. Lith. kalvà ‘small hill’, Latv. kalva ‘hill, island in a river’ (<*kolH-wo-, De Vaan,
2008: 124), Lat. collis (*kolH-wi- with Nussbaum, 1997: 195 and not*kolH-ni-); Lith. kálnas
(<*kolH-no-), OE hyll (<*kCH-ni-), Gr. κολώνη (*kolH-ōn-o-) ‘hill’ (De Vaan, 2008: 124); Lat.
columen ‘top, summit’ (<*kelH-men-), culmen ‘summit (of a building), peak’ (<*kolH-men-),
cf. Kroonen, 2013: 254.
11 Beekes (2010: 1645) also maintains that gurta- “can hardly be Anatolian”, but his
12 Tischler (1983: 661) cautiously suggested a connection with Skt. kuṭa- ‘fort,
stronghold’ deriving it from *kurta-. While this is theoretically possible, the etymology
of this word is unclear, see KEWA/I: 221 with ref. calling attention to similar Dravidic
and Austroasiatic words.
13 See e.g. Lidén, 1916: 22 (very cautiously); Georgiev, 1938: 199; Walde &
Hoffmann, 1938-1954/I: 660; Feist, 1939: 198; GEW 1114; Neumann, 1972: 824, apud
Neumann & Strobel, 2003-2005: 544; Brixhe & Lejeune, 1984: 80; Roller, 1984: 260 n. 17
(contra her, Gordion did not mean ‘city’ in Phrygian); Innocente, 1997: 40; EIEC 199;
Mallory & Adams, 2006: 199; Berndt-Ersöz, 2012: 34 (“possible”); Adams, 2013: 211.
2) just as the indigenous Greek phrase meaning ‘into the city’ has been
reinterpreted by the Turkish newcomers as the name of the settlement in
the case of Istanbul, the local Hittite-Luwian labelling gurta- ‘city’ could
also have been reinterpreted by the Phrygian newcomers as the name of
the settlement, though, again, this is less probable due to the voiceless
initial consonant of the Hittite-Luwian word;
3) just as Ázatiwatiya- ‘the Luwian name of Karatepe’ was named
after its founder, Ázatiwata-, Gordion may reflect and probably does
reflect the name of its founder, a Gordios / Gordiēs (suggested by Bryce,
2012: 4114), which was a widespread name in Central Anatolia, notably
also the name of the father of Midas, the most famous king in Gordion,
and the founder of the city according to Stephen of Byzantium (s.v.
Gordieion, for Gordios’s role in the legends on the origins of the kingdom
and dynasty of Midas see Strobel, 2010: 33-34 with refs.)15.
It can be concluded that there is no convincing proposal for a
loanword.
14 His statement, however, that this is the general assumption is plainly wrong (he
and Roman literature and that it is an artificial name created from the city name
Gordion. This is clearly refuted by the individuals called Kurtis and Gurdî in the
Hieroglyphic Luwian and Neo-Assyrian texts (for the former see the inscriptions
BOHÇA, HİSARCIK 1 §5 (and perhaps 2 §2), KULULU lead strip 1 §4.17, Hawkins
2000; for the latter cf. Aro-Valjus 1999; Fuchs 2000 with refs.).
6. Conclusions
Acknowledgements
16 This derivation is regular also in that case if someone maintains the Hittite origin
of gurta-.
References