5 Anatolian

Alwin Kloekhorst

5.1 Introduction
The Anatolian branch consists of a group of languages once spoken in ancient
Anatolia (modern-day Turkey) and northern Syria, with textual remains dating
from the beginning of the second millennium BCE to the second century CE.1 It
is commonly assumed that in the course of the first millennium CE, the entire
Anatolian branch became extinct. The attested Anatolian languages are (in
chronological order) as follows.2
Kanišite Hittite:3 a dialect of Hittite proper, which is known from hundreds
of personal names and a handful of loanwords attested in Old Assyrian texts
(clay tablets, written in the Old Assyrian version of the cuneiform script, dating
to c. 1935–1710 BCE) mostly stemming from Kaniš/Nēša (modern-day
Kültepe), Central Anatolia.
Hittite (“Ḫattuša Hittite”):4 the main language of the administration of the
Hittite kingdom, written in its own version of the cuneiform script, attested in
some 30,000 fragments of clay tablets (dating to c. 1650–1180 BCE),5 espe-
cially found in the Hittite capital Ḫattuša (modern-day Boğazkale), but also

1
The research for this chapter was conducted as part of the NWO-funded research project Splitting
the mother tongue: The position of Anatolian in the dispersal of the Indo-European language
family (NWO-project number 276-70-026) and the project Multilingualism and minority lan-
guages in ancient Europe, funded by the HERA Joint Research Program “Uses of the past”
(Horizon 2020). I would like to thank Xander Vertegaal and Stefan Norbruis for their useful
comments on earlier drafts of this chapter.
2
Kroonen, Barjamovic & Peyrot (2018) have recently claimed that a number of personal names
that are recorded in texts from Ebla, dated to the twenty-fifth–twenty-fourth centuries BCE,
belong to one or more languages “that clearly fall within the Anatolian Indo-European family”
(2018: 6). However, no detailed analysis of this material is offered, and at present I therefore
regard the linguistic status of these names as too uncertain to make any broad claims.
3
See Kloekhorst 2019 for a full account of this language and its attestations.
4
The authoritative synchronic grammar of Hittite is Hoffner & Melchert 2008. Synchronic
dictionaries are HW² and CHD; etymological dictionaries are HEG, HED, EDHIL. For historical
linguistic treatments, see e.g. Melchert 1994; Kimball 1999; EDHIL.
5
But see Kloekhorst & Waal 2019, who argue that a few Hittite tablets may stem from the latter
half of the eighteenth century BCE.

63

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 64 Alwin Kloekhorst

several other places in Central Anatolia. It is the best attested Anatolian

language by far, and therefore the most important witness of this branch.
Palaic:6 known from several passages embedded in Old and Middle Hittite
texts (sixteenth–fifteenth century BCE), primarily dealing with the cult of the
god Zaparu̯ a. It was the language of the land of Palā, situated in the north-west
of Central Anatolia. The Palaic corpus is small, and therefore many basic
matters regarding grammar and lexicon are unclear.
Cuneiform Luwian (also called Kizzuwatna Luwian):7 only known from
cultic passages cited in Hittite texts (dating to the sixteenth–fifteenth century
BCE). It was certainly spoken in Kizzuwatna (south-east of Central Anatolia)
and possibly also in the western part of Anatolia. In Hittite texts from the New
Hittite period (fourteenth–thirteenth century BCE), we find many Luwian
loanwords, which traditionally were regarded to be Cuneiform Luwian as
well but which may be more appropriately regarded as linguistically belonging
to Hieroglyphic Luwian.
Hieroglyphic Luwian (also called Empire Luwian / Iron Age Luwian):8
closely related to Cuneiform Luwian, written in an indigenous hieroglyphic
script (Marazzi 1998) that seems to have been especially designed for this
language. Seals containing these hieroglyphs can be dated as far back as the Old
Hittite period (c. 1600 BCE), but real texts (mostly inscriptions on rocks and
stone steles) date from the thirteenth to the end of the eighth century BCE. The
c. thirty texts that date from the last phase of the Hittite Kingdom (so-called
Empire period, and therefore “Empire Luwian”) are found all over Anatolia
and northern Syria, whereas the c. 230 post-Empire period inscriptions (Iron
Age, and therefore “Iron Age Luwian”) are restricted to south-eastern Anatolia
and northern Syria, the region of the so-called Neo-Hittite city states. Thanks to
a boost in studies of the language since the publication of Hawkins 2000,
Hieroglyphic Luwian has become one of the better-known Anatolian
languages.
Lydian:9 the language of the land of Lydia (central western Anatolia),
written in its own version of the Greek alphabet, attested in some 120 texts
(the bulk of which are inscriptions on stone steles), dating from the eighth to the
third century BCE (with a peak in the fifth–fourth century BCE). Our

6
For texts, grammar, vocabulary and historical phonology, see e.g. Carruba 1970; Kammenhuber
1969; Melchert 1994: 190–228.
7
Texts are collected in Starke 1985. For grammatical treatments, see Starke 1990; Melchert 2003.
For the lexicon, see Melchert 1993. See Yakubovich 2010 for the term “Kizzuwatna Luwian”.
8
Texts can be found in Hawkins 2000, see also ACLT. For grammatical treatments, see Melchert
2003; Payne 2010; Yakubovich 2015. A good Hieroglyphic Luwian dictionary is a desideratum:
Meriggi 1962 is largely outdated, and the lexical part of ACLT can only be used with caution.
9
For texts, grammar and vocabulary, see Gusmani 1964. Historical linguistic treatments can be
found in Melchert 1994: 329–83; Gérard 2005. A more general introduction to the Lydians and
their language is Payne & Wintjes 2016.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 65

knowledge of Lydian is limited since there are only a few bilingual texts and
since its vocabulary is difficult to compare to the lexicon of the other Anatolian
languages (see also below, Section 5.3.3).
Carian:10 the language of the land of Caria (south-central western Anatolia),
written in its own version of the Greek alphabet, attested in some 200 inscrip-
tions from the seventh–fifth century BCE from Egypt (tomb inscriptions from
Carian mercenaries living there) and from the fourth–third century BCE from
Caria itself. Our knowledge of Carian is very rudimentary: the Carian alphabet
was not successfully deciphered until the 1990s, and many inscriptions contain
personal names only.
Lycian (also called Lycian A):11 the language of Lycia (south-western
Anatolia), written in its own version of the Greek alphabet, in some 150
coin legends and 170 inscriptions on stone, dating to the fifth–fourth
century BCE. Our knowledge of Lycian is relatively advanced, partly
because of some bilingual texts (including the large trilingual inscription
of Letôon) and partly because of its linguistic similarities with the Luwian
languages. Nevertheless, many details regarding grammar and lexicon are
still unclear.
Milyan (also called Lycian B):12 attested in two inscriptions from Lycia
(fifth century BCE) that are written in the Lycian alphabet. Although the name
“Milyan” refers to the region Milyas, situated in the north-east of Lycia, it is
unclear where it originates. The two Milyan inscriptions, which both seem to be
in verse, are difficult to understand, and our knowledge of Milyan is therefore
rudimentary.
Sidetic:13 the language of the city of Side (south coast of Anatolia) and its
surroundings, written in its own version of the Greek alphabet, attested in some
ten inscriptions on coins and stone, dating to the fifth–second century BCE. The
number of textual remains is very low, so we only know a few facts about
Sidetic grammar and lexicon.
Pisidian:14 a language attested in a few dozen tomb inscriptions in the Greek
alphabet that were found in the eastern part of classical Pisidia (south-west of
Central Anatolia), dating to the first–second century CE. The inscriptions
contain only personal names, some of which point to an Anatolian character
to this language.

10
See Adiego 2007 for a full discussion of all Carian texts, and the grammar, lexicon and historical
linguistic interpretation of the language.
11
For text editions see Kalinka 1901; Neumann 1979; Laroche 1979. The vocabulary is compiled
in Melchert 2004 and Neumann 2007. Grammatical treatments and historical linguistic analyses
can be found in e.g. Hajnal 1995; Melchert 1994: 282–328; Melchert 2004; Kloekhorst 2013.
12
Shevoroshkin 2013. The Milyan vocabulary is included in Melchert 2004 and Neumann 2007.
13
Pérez Orozco 2007. 14 Brixhe 1988.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 66 Alwin Kloekhorst

5.2 Evidence for the Anatolian Branch

There is ample evidence to view the Anatolian languages as forming a single
branch: they share enough linguistic features to set them apart as a single
group vs. the rest of the Indo-European language family, cf. e.g. Rieken 2017:
299. A complicating factor, however, is the Indo-Anatolian hypothesis, which
states that Anatolian was the first branch to split off from the Indo-European
mother language, after which the remaining language, which was to become
the ancestor language of all the non-Anatolian Indo-European languages
(“Core Proto-Indo-European”) underwent a set of innovations (see
Section 5.5). Whenever the Anatolian languages show shared features that
are different from the other Indo-European languages, we should therefore
investigate to what extent these differences are caused by innovations that
took place in the prehistory of the Anatolian branch, or by innovations in the
prehistory of Core Proto-Indo-European. In the latter case, the Anatolian
features may in fact be shared retentions, and therefore cannot, strictly
speaking, be used in arguing that the Anatolian languages form a single
branch. In practice, however, it is not always easy to distinguish between
the two.
Another complicating factor is that some of the Anatolian languages have
a very limited attestation (especially Sidetic and Pisidian) or are in general
poorly understood (Carian, Milyan and, to a lesser extent, Lydian and Palaic).
This means that not all features listed below are found in all languages.
The following specific features of the Anatolian languages can be regarded
as examples of common innovations that prove the unity of the Anatolian
branch and allow for the postulation of an ancestor language, Proto-
Anatolian, from which they all derive:
Phonology
• the merger of PIE mediae and aspiratae into a single series that is called lenis
(PIE *d, *dʰ > PAnat. */t/),15 which is distinct from the so-called fortis series,
which is the outcome of PIE tenues (PIE *t > PAnat. */tː/)
• the operation of Eichner’s lenition rules: (1) pre-PAnat. *V̄ ́ C:V > PAnat.
*V̄ ́ CV and (2) pre-PAnat. *V́ . . . VC:V > PAnat. *V́ . . . VCV16

15
As argued in e.g. Kloekhorst 2016: 226–8, within the glottalic theory this merger may be seen as
the result of the development of PIE mediae, which can be interpreted as pre-glottalized lenis
stops, e.g. PIE *d = *[ˀt], into a biphonemic pair of glottal stop + lenis stop: PIE *d = *[ˀt] > pre-
PAnat. *[ʔt] = */ʔ/ + */t/. In this way, the oral part of the PIE mediae was detached from its
glottal part and merged with the PIE aspiratae, which in fact were lenis stops (e.g. PIE *dʰ =
*[t]), whereas the glottal stop merged with the outcome of PIE *h1.
16
Eichner 1973: 79, 100 n. 86. The two lenition rules can in fact be regarded as a single
development, which may be represented as pre-PAnat. *V́ (. . .)VCːV > PAnat. *V́ (. . .)VCV, cf.
Adiego 2001; Kloekhorst 2014: 547–87.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 67

• PIE accented short *ó was lengthened to PAnat. long */ṓ/ (and subsequently
caused lenition according to Eichner’s first lenition rule)17
• the PIE cluster *h2u̯ yields PAnat. monophonemic */qʷː/,18 e.g. PIE *trh2u̯ (e)nt-
> PAnat. */tːrqʷː(ə)nt-/ > Hitt. tarḫuu̯ ant- /tərχʷːənt-/, CLuw. tarḫu(u̯ a)nt-
/tərχʷː(ə)nt-/, Lyc. trqqãt- / trqqñt- /trkʷ(a)ⁿt-/, Car. trqδ- /trkʷⁿt-/
• the development of a lateral in the word for ‘name’: PIE *h3néh3mn- > PAnat.
*/ʔlṓmn-/ > Hitt. lāman, HLuw. álaman-, Lyc. alãma-
Morphology
• the creation of an acc.-dat. form */ʔmːu(-)/ ‘me’ (vs. PIE *h1mmé-)
• the creation of a demonstrative pronoun */ʔopṓ-/ (from virtual PIE *h1o-bʰó-)19
• the loss of the distinction between present and aorist (the “tezzi-principle”)20
• the creation of the ḫi-conjugation (cognate to the PIE perfect)21
• the 1pl. ending */-uén(i)/ (cognate to the PIE dual ending *-ué)22
• the replacement of the post-consonantal pret.act.3sg. ending *-t by the
middle ending *-to (> Hitt. -tta, CLuw. -tta, HLuw. -ta, Lyc. -te)23
• the loss of the subjunctive and optative moods.

For other specifically Anatolian features, see Section 5.5, where a list of shared
retentions of Anatolian will be presented (as arguments in favour of the Indo-
Anatolian hypothesis).

5.3 The Internal Structure of Anatolian

There is some debate on the exact internal subgrouping of the Anatolian
branch, although on some aspects there is broad consensus.

5.3.1 The Luwic Branch

There can be no doubt that Cuneiform Luwian, Hieroglyphic Luwian and
Lycian form a separate branch, which is commonly called “Luwic”. This
17
Kloekhorst 2014: 439–59.
18
Kloekhorst 2006: 102; Melchert 2011: 128–9. Cf. Kloekhorst 2018a for the postulation of
a labio-uvular stop /qʷː/ for the PAnat. stage.
19
EDHIL: 192. 20 Malzahn 2010: 267–8.
21
E.g. Eichner 1975; Kloekhorst 2018b, contra Jasanoff 2003.
22
Jasanoff 2003: 3; EDHIL: 1001.
23
The idea that CLuw. -tta, HLuw. -ta and Lyc. -te reflect the middle ending *-to is generally
accepted (e.g. Yoshida 1993), but the origin of Hitt. -tta is debated. Some scholars assume that
the spelling °C-ta can represent /°Ct/ < PIE *°C-t (e.g. Yoshida 1991: 28); others assume that the
a-vowel is real and developed as a prop-vowel, i.e. /°C-tə/ < PIE *°C-t (e.g. Melchert 1994:
175–6, with references); and still others have argued that the a-vowel is real but cannot be
explained as a prop-vowel, and that Hitt. -tta therefore must reflect earlier *-to (EDHIL: 800–1).
If the latter view is correct, the spread of *-to at the cost of post-consonantal *-t must have been
a common innovation of all Anatolian languages.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 68 Alwin Kloekhorst

means that these languages derive from a “Proto-Luwic” mother language. It is

generally assumed that Milyan and Carian belong to this branch too, and also
Sidetic and Pisidian are often regarded as possibly Luwic languages (e.g.
Melchert 2003: 170–7; Yakubovich 2010: 6; Rieken 2017: 301–3).

5.3.1.1 Shared Innovations of the Luwic Languages The Luwic sub-branch

of Anatolian can be defined through the following innovations (although they
are not always attested in all languages):
Phonological
• the assibilation of PAnat. */ḱː/ > PLuwic */ts/ > CLuw. z /ts/, HLuw. z /ts/,
Lyc. s, Mil. s, Car. s, Sid. ś (vs. Hitt. /kː/, Pal. /kː/, Lyd. k)
• the weakening of PAnat. lenis */ḱ/ > PLuwic *i̯ > Ø: e.g. PAnat. */ḱésːr-/
‘hand’ > CLuw. ī̆ š(ša)ri-, HLuw. istri-, Lyc. izri- (vs. Hitt. /k/ and Pal. /k/)
• the weakening of PAnat. lenis */kʷ/ > PLuwic *u̯ : e.g. PAnat. */kʷṓu-/ ‘cow’
> HLuw. wawa/i-, Lyc. wawa-, uwa- (vs. Hitt. /kʷ/, Pal. /kʷ/, Lyd. k)
• the merger of PAnat. */e/ and */ō̆ / into PLuwic */ə̄̆ /24 (vs. their retention as
separate phonemes in Hittite, Palaic and probably Lydian25)
• Čop’s Law: PAnat. *V̆ ́ CV > PLuwic *V̆ ́ C:V26
Morphological
• the large-scale spread of the proterodynamic i-stem inflection replacing ori-
ginal consonant stem and o-stem inflection (formerly called “i-mutation”)27
• the reshaping of the PAnat. nom.pl.c. ending *-es to PLuwic *-Vns-i (based
on the acc.pl.c. ending *-Vns < PIE *-V-ms + the original pronominal nom.pl.
c. ending *-i < PIE *-oi?) > CLuw. -Vnzi /-Vntsi/, HLuw. -V-zi /-Vntsi/, Lyc.
-i (< *-insi), -ẽi (< *-onsi), -ãi (< *-ānsi), Car. -š (?)
• the grammaticalization of the genitival adjective in *-osːo/i- > CLuw. -ašša/i-,
HLuw. -asa/i-, Lyc. -ehe/i-, Mil. -ehe/i-, Car. -š (?), Sid. -asV, Pis. -s (?)28

24
The development of the Carian vowel system is too poorly understood for us to be certain that
Carian was part of this development. If it was not, this isogloss should be removed from the
inventory.
25
Note that the prehistory of the Lydian vowel system is still relatively unclear.
26
Cf. Kloekhorst 2014: 567–85 for the fact that this law is not only valid in CLuwian (for which it
was originally formulated, cf. Čop 1970), but also in HLuwian and Lycian.
27
See Norbruis 2021: 9–50 (adapting Rieken 2005) for a full treatment of the phenomenon that in
the prehistory of the Luwic branch the proterodynamic i-stem inflection (synchronically
characterized by the presence of an -i- in the nom.sg./pl.c. and acc.sg./pl.c. cases vs. the absence
of -i- in all other cases, therefore termed “i-mutated”), which it had inherited from PIE, spread
widely within the nominal system, first to consonant-stems, and later to *o-stems (but not to *ā-
and *u-stems). See below for the fact that Palaic and Lydian also show some cases of this spread.
28
This suffix is attested in Hittite, too, but it has not been grammaticalized as an inflectional
morpheme, cf. EDHIL: 216.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 69

• the spread of the pret.act.3sg. ending *-to to verbal stems ending in a vowel
(at the cost of the original ending *-t)29
Lexical
• PLuwic *māsːVn- ‘god’ > CLuw. māššani-, HLuw. DEUS-ni-, Lyc. mahana-,
Mil. masa-, Car. mso-, Sid. masara- (vs. PAnat. *tieu- (< PIE *dieu-) in Hitt.
šiu-, Lyd. ciw-)30
• PAnat. *tːrqʷː(ə)nt- ‘(one who has / has been) conquered’ develops into the
generic name for ‘Storm-god’ in PLuwic, yielding CLuw. tarḫu(a)nt-,
HLuw. tarhunt-, Lyc. trqqñt-, Mil. trqqñt-, Car. trqδ-, all ‘Storm-god’ (vs.
Hitt. tarḫuu̯ ant- ‘conquered’ and dIŠKUR-unn(a)- ‘Storm-god’)31

5.3.1.2 Internal Subgrouping of the Luwic Branch The relationships

between the three better-known Luwic languages – Cuneiform Luwian,
Hieroglyphic Luwian and Lycian – are quite clear. It is generally accepted
that Cuneiform Luwian and Hieroglyphic Luwian are closely related, yet
distinct, dialects. The relationship between the two cannot have been a matter
of one of them deriving from the other (cf. Melchert 2003: 171–2), which
means that both must go back to a common ancestor, which may be termed
Proto-Luwian.
Lycian is generally recognized as being closely related to the two Luwian
languages. Yet, although it was attested almost a millennium after the latter’s
first attestations, it was clearly not a direct daughter language of either of them:
the Luwian languages show innovations that are not shared by Lycian (e.g.
merger of PLuwic */ə/ and */a/ into PLuwian */a/; replacement of the dat.-loc.
pl. ending */-əs/ (< PAnat. */-os/) by */-əns/32 > PLuwian */-ants/ (CLuw.
-anza /-ants/, HLuw. °a-za /-ants/); fricativization of */qː/ to PLuwian */χː/
(Kloekhorst 2018a: 73–6)). This means that Lycian stems from a sister
language to Proto-Luwian and that both can be regarded as distinct daughters
of Proto-Luwic.
Although our knowledge of Milyan is limited, it is usually seen as being
closely related to Lycian. This is based on the fact that these two languages have
several linguistic features in common, which may be seen as shared innovations
that set them apart from Proto-Luwian: PLuwic *-Vs > Mil., Lyc. -V (as in nom.
29
Whereas Hittite still shows the old opposition between post-consonantal *-to and post-vocalic
*-t. Palaic has also retained post-vocalic *-t (e.g. pret.act.3sg. lu-ki-i-it). The origin of Lyd. pret.
act.3sg. -l is not fully clear, unfortunately.
30
Eichner (1974: 64) proposed that Luw. *māsːVn- ‘god’ derives from a pre-form *meh1/3-(o)s-h3on-
“freien Willen habend, nach eigenem Belieben handelnd” (~ Lat. mōs ‘custom, usage’, which
Eichner translates as “Wille”). An alternative may be to derive *māsːVn- from *meh1ns-en-,
a derivative of PIE *meh1-ns- ‘moon’.
31
Although it cannot be excluded that already in PAnat. *tːrqʷː(ə)nt- was the name of the Storm-
god, and that Hitt. dIŠKUR-unn- is an innovation, cf. Kloekhorst 2019: 192.
32
Having taken over the nasal of acc.pl.c. */-ns/ < PIE *-ms.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 70 Alwin Kloekhorst

sg.c.) (vs. PLuwian *-Vs); PLuwic *-Vn > Mil., Lyc. -Ṽ (as in acc.sg.c.) (vs.
PLuwian *-Vn); a-umlaut (e.g. Mil. nom.-acc.pl.n. uwadra vs. uwedr(i)-, or
nom.-acc.pl.n. χuzruwãta vs. acc.pl.c. χuzruwẽtiz); syncope in the ethnicon
suffix Mil. -wñni- and Lyc. -ñni- < *-wnːi- (vs. PLuwian *-wanːi-); fronting
of PLuwic */kʷː/ before a front vowel in rel.pron. */kʷːi-/ > Lyc. ti-, Mil. ki- /ci-/
(vs. PLuwian *kʷi-). A shared lexical innovation may be Mil. kibe ~ Lyc.
tibe ‘or’.
The position of Carian, Sidetic and Pisidian is less clear, since the number of
possible isoglosses is very low. In the case of Carian, Adiego (2007: 347) states
that “a meaningful isogloss shared by Carian and Milyan is the copulative
conjunction Car. sb, Mil. sebe ‘and’”, which contrasts with Lyc. se ‘and’. One
may add Car. mso- ~ Mil. masa- vs. Lyc. mahana- ‘god’. In the case of Sidetic,
the dat.pl. ending -a (in masara ‘to the gods’), which must reflect PAnat. */-os/,
shows that this language does not belong to the Luwian subgroup (which rather
shows the dat.pl. ending */-ants/). Furthermore, this ending shows that Sidetic,
just like Lycian and Milyan, has undergone the development *-Vs > -V, which
may be seen as a shared innovation. On the basis of the Sidetic conjuction śa
‘and’, we may assume a closer affinity with Lycian, which has se ‘and’ (vs. Mil.
sebe and Car. sb). In the case of Pisidian, a closer affinity with the Lyco-Carian
subgroup may be seen from the nom.sg.c. ending -V, which then corresponds to
Lyc. -V, Mil. -V, Car. -Ø < PLuwic *-Vs (vs. CLuw. -Vš and HLuw. -Vs). The
exact position of Pisidian within this group must remain undetermined,
however.
All in all, the tree of the Luwic sub-branch may be envisaged as in Figure 5.1.

5.3.1.3 Dating Proto-Luwic The Luwic branch seems to have been rela-
tively shallow. As mentioned above, the linguistic difference between

Pisidian

? Sidetic
?
? Lycian

Milyan

Lyco-Carian Carian

Hieroglyphic Luwian

Luwic Luwian Cuneiform Luwian

Figure 5.1 The Luwic sub-branch of Anatolian

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 71

Cuneiform Luwian and Hieroglyphic Luwian is minimal, and we may therefore

date their pre-stage, Proto-Luwian, to not much more than a handful of gener-
ations before the oldest attested Cuneiform Luwian texts (sixteenth century
BCE), i.e. to c. the eighteenth century BCE. In the same vein, the difference
between the Lyco-Carian branch and Proto-Luwian seems to have been rela-
tively small, so we may assume that Proto-Luwic preceded Proto-Luwian by no
more than two or three centuries. We can thus approximately date this stage to
the twenty-first–twentieth century BCE.

5.3.2 The Position of Palaic

Since our knowledge of Palaic is limited, it is not easy to determine its position
within the Anatolian language family with certainty. Moreover, as Carruba
(1970: 4) and Melchert (2003: 269) show, Palaic shares linguistic features both
with Hittite and with the Luwic languages, adding to the difficulty.
Nevertheless, Oettinger (1978) gives several arguments that would indicate
that Palaic is more closely related to the Luwic languages than to Hittite and
Lydian. According to Rieken (2017: 303), however, “none of the isoglosses
suggested so far [i.e., by Oettinger and others – AK] involve newly created
morphology. In each case, the change consists of a choice among several
inherited morphemes or a shift of a category’s function, mostly extending it”,
and she therefore remains agnostic about the position of Palaic. To my mind,
this is too negative a view: there certainly are some features that in fact can be
used for judging its place in the Anatolian tree.
• The Palaic dative of the 3sg. enclitic pronoun, =tu ‘to him/her’, is identical to
CLuw. =tu and HLuw. =du / =ru, but distinct from Hitt. =šše (later =šši) and
Lyd. =mλ. Oettinger (1978: 78–9) convincingly argues that this =tu origin-
ally was the dative form of the 2sg. enclitic pronoun, which was extended to
the 3rd person. This non-trivial development was thus a shared innovation of
Palaic and the Luwian languages.33
• In Palaic, Proto-Anatolian lenis */kʷ/ (< PIE *gʷ ⁽ʰ⁾) is weakened to /xʷ/ or /χʷ/
in aḫuu̯ ā̆ nti ‘they drink’ < *h1gʷʰénti. This fricativization may be seen as
a first step towards the full weakening that is found in Luwic, where PAnat.
*/kʷ/ > *u̯ .34
• According to Starke (1990: 71–5), Palaic shows some instances of “i-muta-
tion”, indicating a connection with the Luwic branch. Since it has now
become clear that the “i-mutation” inflection in fact goes back to a normal

33
Note that the corresponding Lycian morpheme is =i(je), which then must be a later innovation
through analogy after the nominal dat.sg. ending -i(je).
34
If Watkins’ suggestion (apud Melchert 1990: 207) that Pal. kuu̯ ani- means ‘womanly’ (i.e. from
the PIE stem *gʷen-h2-) is correct, it would show that this weakening did not take place in word-
initial position.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 72 Alwin Kloekhorst

PIE proterodynamic i-stem inflection (cf. footnote 27), the mere existence of
this type in Palaic is not remarkable per se. However, as noticed by Starke, in
Palaic the “i-mutated” inflection also seems to be found in original conson-
ant-stems (e.g. dilaliant(i)-). This implies a secondary spread that is compar-
able to the one found in the Luwic branch, and which may then be viewed as
a shared innovation. Nevertheless, the fact that our evidence for “i-mutated”
stems in Palaic is scanty shows that this spread certainly had not yet taken
place on such a large scale as in the Luwic languages.
• In Palaic, the pret.act.3pl. ending is -(a)nta, which matches Luwic *-Vntə
(CLuw. -anta, HLuw. -anta, Lyc. -Ṽte),35 but contrasts with Hitt. -er and Lyd.
-rs / -riš. Since *-Vntə is generally regarded as deriving from the PIE 3pl.
middle ending *-ento, it may be possible to see the transfer of this ending to
the pret.3pl. of the active as a common innovation of Palaic and Luwic.36
• In Palaic, the only attested pret.act.1sg. ending is -(ḫ)ḫa, which reflects
PAnat. */-qːa/ < PIE *-h2e, and thus originally belonged to the
ḫi-conjugation. Since it is also found in the form aniēḫḫa ‘I did’ (thus
Carruba 1970: 50), which was probably originally mi-conjugating, it seems
that in Palaic the pret.act.1sg. ḫi-ending -(ḫ)ḫa has fully ousted the corres-
ponding mi-ending *-m (attested in Hitt. -un, -nun and Lyd. -ν). The same
development took place in Luwic, where pret.act.1sg. *-q(ː)a (CLuw. -(ḫ)ḫa,
HLuw. -ha, Lyc. -χa, -ga) has fully ousted *-m as well. We may thus assume
that Palaic and Luwic shared this innovation.37
Although the material is scanty and the number of arguments low, it does seem
safe to conclude that Palaic shares some innovations with the Luwic branch.
Nevertheless, it is clear that Palaic cannot be regarded as a proper Luwic
language: for instance, it does not show assibilation of PAnat. */ḱː/ (which
rather yielded Pal. k; Melchert 1994: 210), and it does not show a nom.pl.c.
ending *-Vnsi (but rather -aš and -eš). We should therefore assume that Luwic
and Palaic are related on a higher node, which may be termed Luwo-Palaic.38

35
Yoshida’s scenario, by which the Palaic ending -(a)nta has a different origin from the Luwic
ending *-Vntə < PIE *-Vnto (Yoshida 1991: 370–1), seems too complicated to me.
36
We may also assume that this transfer took place as early as in pre-Proto-Anatolian times, and in
fact consisted of the replacement of the original pret.act.3pl. ending *-Vnt < PIE *-(e)nt by its
middle variant *-Vnto < PIE *-(e)nto in a reaction to the loss of word-final *-t, just as pret.
act.3sg. *-t was for the same reason replaced by middle *-to (cf. Section 5.2). This would fit the
fact that a pret.act.3pl. ending *-an cannot be reconstructed for Proto-Anatolian (contra Yoshida
1991). If correct, we have to assume that Proto-Anatolian, next to ḫi-conjugated pret.act.3pl.
*-ēr / *-rs (with *-rs being the original zero-grade variant of *-ēr?), possessed the mi-
conjugated ending *-(V)nto, and that in all branches only one of these endings survived.
Hittite generalized the ending *-ēr, Lydian the ending *-rs, and Palaic and Luwic the ending
*-(V)nto. This spread of the mi-conjugated ending *-(V)nto at the cost of ḫi-conjugated *-ēr /
*-rs may then be seen as a shared innovation of Palaic and Luwic.
37
Yakubovich (2010: 6) cites this isogloss as the defining feature of the Luwo-Palaic subgroup.
38
Thus also Oettinger 1978: 92; Yakubovich 2010: 6. See now also Giusfredi 2020: 18–19.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 73

5.3.3 The Position of Lydian

The exact position of Lydian is widely debated, which is due to the fact that this
language is poorly understood: only a few Lydian words can be securely
translated, making it difficult to establish etymologies and thus sound corres-
pondences with the other Anatolian languages. Nevertheless, there seems to be
more and more consensus that Lydian, too, was related to the Luwic sub-
branch, since the two share some isoglosses:
• “i-mutation” (cf. Sasseville 2017): since the “i-mutation” inflection reflects
the normal PIE proterodynamic i-stem inflection (Norbruis 2021: 9–50; see
also footnote 27), its presence in Lydian is not remarkable per se. However,
its presence in nouns like sfardẽt(i)- ‘Sardian’ (nom.sg.c. sfardẽtiš vs. dat.pl.
sfardẽtaν), which originally was probably an *-nt-stem, implies the spread of
the “i-mutation” inflection at the cost of the consonant-stem inflection, which
would be an innovation shared with Luwic (and Palaic).
• Lydian pres.act.1sg. -u/-w is identical to PLuwic *-ū̆ (CLuw. -u̯ i, HLuw. -wi,
Lyc. -u),39 which contrasts with Hitt. pres.act.1sg. -mi and -ḫḫi (unfortu-
nately, in Palaic no pres.act.1sg. forms are attested). However, if *-ū̆ indeed
goes back to the PIA thematic pres.1sg. ending *-oH (Kloekhorst 2013: 146),
the ending is not the result of an innovation. Nevertheless, the fact that both in
Luwic and in Lydian (as far as we can tell) *-ū̆ < *-oH ousted the athematic
mi-conjugation ending *-mi and the ḫi-conjugation ending *-h2e-i (which
were retained in Hittite, where a newly created *-o-mi ousted original *-oH)
can be seen as a common innovation.40
• Lydian -cuwe- ‘to erect(?)’ is regarded by Oettinger (1978: 89) and Melchert
(2003: 269) as cognate to CLuw. tūu̯ a- ‘to place’, HLuw. tuwa-i ‘to place’ and
Lyc. tuwe- ‘to place (upright)’, which all reflect a stem *tuu̯ V-. Although
there are different views on the exact origin of this formation, it is mostly
seen as an innovation, which then must have been shared by Lydian and the
Luwic languages.41

39
As Stefan Norbruis and Oscar Billing (pers. comm.) have pointed out to me, since Lycian does
not show a general loss of word-final *-i, Lyc. -u is better derived from PLuw. *-ū̆ than from *-ū̆ i
or *-u̯ i. This means that we have to assume that in the Luwian languages the original ending *-ū̆
was secondarily extended with the present marker *-i.
40
Yakubovich (2010: 6) cites this isogloss as the defining feature of the “Non-Hittite” subgroup.
41
According to Oettinger (1978: 89), the stem tuu̯ V- is based on a false segmentation of the
pres.1pl. form *tuu̯ an(i) < *(dʰe-)dʰh1-u̯ éne(-i) of the verbal root *dʰeh1- ‘to put’. Frotscher
(2012) argues that *tuu̯ V- derives from earlier *dʰh1-oi-, the stem that is found in Hitt. dai-i / ti-
‘to put’, also derived from PIE *dʰeh1-. And Melchert (2004: 74) rather derives *tuu̯ V- from
a stem *(s)teh2w-, ultimately belonging to PIE *steh2- ‘to stand’. Since in all languages *tuu̯ V-
means something like ‘to erect’, a connection with PIE *steh2- may indeed be more attractive
than a connection with *dʰeh1-. Nevertheless, in all analyses the stem *tuu̯ V- is to be viewed as
an innovation.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 74 Alwin Kloekhorst

Note that we are not necessarily dealing with a shared innovation in all cases in
which Lydian coincides with Luwic:
• Lyd. taada- ‘father’ < *tóto- is cognate with PLuwic. *tóti- (CLuw. tāti-,
HLuw. tati-, Lyc. tedi-, Car. ted), which differs from Hitt. atta- and Pal. pāpa-
‘father’. However, since it cannot be excluded that *tóto- is the Proto-
Anatolian form, whereas Hitt. atta- and Pal. pāpa- are innovations, this
isogloss between Lydian and Luwic (see below for the difference in “i-muta-
tion”) could in principle represent a shared retention and is therefore non-
probative.
• Lydian has a 1sg. reflexive particle =m, which is identical to PLuwic *=mi
(CLuw. =mi(?), HLuw. =mi), but contrasts with Hittite, which uses =z(a) <
*=ti in this function (no attestations known for Palaic). Since it cannot be
excluded that Lydian and Luwic reflect the Proto-Anatolian situation,
whereas Hittite may have undergone an innovation, this isogloss may repre-
sent a shared retention and therefore is non-probative.
Moreover, there are also some Luwic isoglosses in which Lydian clearly does
not participate:
• PAnat. lenis /kʷ/ > Lyd. k in kãna- ‘woman’ < *gʷoneh2- (whereas in PLuwic,
PAnat. */kʷ/ is weakened to *u̯ , e.g. *gʷoneh2- > CLuw. u̯ āna-)
• Lyd. ciw- ‘god’ < PAnat. */tieu-/ < PIE *dieu- (vs. PLuwic *māsːVn- ‘god’)
• Lyd. a-stem noun taada- ‘father’ (vs. PLuwic “i-mutated” *tóti-, see the
forms cited above)
I am therefore reluctant to view Lydian as a proper Luwic language; rather,
I assume that both Lydian and Proto-Luwic derive from an earlier node. In
order to establish the position of this node vis-à-vis the Luwo-Palaic node as
assumed above, the following arguments can be used:
• The Lydian dat.sg. form of the 3rd person enclitic pronoun, =mλ ‘to him/her’,
can be derived from *=smei̯ / *=smoi̯ (Kloekhorst 2012: 169), which to my
mind is an archaic morpheme (cognate with the PIE element *-sm- as found
in, e.g., the Skt. pronominal stem tasm-). Lydian thus did not participate in
the Luwo-Palaic innovation by which the original dat.sg. of the 2nd person
enclitic pronoun, *=tu, was extended to the 3rd person.42
• The Lydian pret.act.1sg. ending is -ν, which reflects the PAnat. mi-
conjugation ending *-m. Since in Lydian no trace of the corresponding
PAnat. ḫi-conjugation ending *-q(ː)a (< PIE *-h2e) is found, we may assume
that *-m > Lyd. -ν had been generalized at the cost of *-q(ː)a. This would then
be a reverse development to the generalization of the ḫi-conjugation ending

42
It cannot be excluded, however, that Lyd. =mλ < *=smVi̯ received its -m- from the correspond-
ing dat.pl. form *=smos (Hitt. =šmaš, CLuw. =mmaš, Lyd. =ms) and originally was *=soi, thus
being directly cognate with Hitt. =šše. If this is the case, Lydian would still show an archaic
morpheme vis-à-vis the innovated *=tu of Luwo-Palaic, which would indicate that Lydian
should derive from a higher node.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 75

*-q(ː)a at the cost of *-m that took place in Palaic and Luwic, and which was
mentioned above as a possible shared innovation between these latter two
branches.
It is for these reasons that I assume that the “Luwo-Lydian” node must be
placed higher up the family tree than Luwo-Palaic (thus also Oettinger 1978:
92; Yakubovich 2010: 6).

5.3.4 The Position of Hittite

Hittite proper (“Ḫattuša Hittite”) knew a sister dialect, Kanišite Hittite, that is
very similar to it but in some points does deviate (Kloekhorst 2019). This calls
for the postulation of a Proto-Hittite ancestor language that may have been
spoken only a few generations before the oldest attestations of Kanišite Hittite
(twentieth century BCE), i.e. around 2100 BCE.
As has become clear in the sections above, there are no clear linguistic
innovations that Hittite shares with any of the other Anatolian
languages.43 As Rieken puts it, Hittite is “notorious for [its] conservatism”
(2017: 303). However, this does not mean that Hittite can be directly
equated with Proto-Anatolian: Hittite, too, has undergone its specific
innovations (e.g. the assibilation of dental stop + *i̯ ; the almost complete
elimination of paradigmatic alternations between fortis and lenis stops; the
reshaping of some verbal endings; the transfer of many mi-verbs to the
ḫi-conjugation (Norbruis 2021: 131–207); the spread of the n-suffix in
the word for ‘earth’; etc.).

5.3.5 Dating Proto-Anatolian

Although it is difficult to say anything certain about the absolute dating of
reconstructed ancestor languages, in the case of Proto-Anatolian we have seen
that its two best-known branches, Luwic and Hittite, have proto-languages that
are roughly contemporaneous: Proto-Luwic can be approximately dated to the
twenty-first–twentieth century BCE, and Proto-Hittite to c. 2100 BCE. The
difference between the two is quite sizable, and elsewhere (Kloekhorst in press)
I have therefore argued that they may have been a millennium apart from each
other, which would mean that Proto-Anatolian started to diverge sometime
around the thirty-first century BCE.

43
At first sight, the fact that both Lyd. ciw- and Hitt. šiu- < PAnat. */tieu-/ ‘god’ show assibilation /
palatalization of the word-initial */t/ may be seen as a shared innovation between these two
languages. However, since Lydian shows other features that it shares with Luwo-Palaic, we
have to assume that the assibilation in the word for ‘god’ is a parallel, not shared, innovation in
these languages.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 76 Alwin Kloekhorst

3000 BCE 2000 BCE 1000 BCE 1 CE

Pis.
Sidetic
Lycian
Mil.
PLyco-Car. Carian
HLuwian

PLuwic – PLuwian CLuw.

PLuw.-Pal. Pal.
PLuw.-Lyd. Lydian
Hittite
PAnat. PHitt. KHitt.

Figure 5.2 Tree model of the Anatolian languages

5.3.6 The Dialectal Make-Up of Anatolian

Taking all these points into account, we arrive at a tree model of the Anatolian
branch as shown in Figure 5.2.44

5.4 The Relationship of Anatolian to the Other Branches

In 1994, Puhvel argued that Anatolian shares many linguistic features with
Celtic, Germanic, Italic, Tocharian and, to a lesser extent, Greek, which would
point to a genetic relationship between Anatolian and these “western”
branches. For instance, Hitt. išpant-i ‘to libate’ matches Lat. spondeō and Gr.
σπένδω but has no cognates anywhere else. However, as Melchert (2016: 300)
rightly states, all such cases “can be interpreted as common retentions that just
happen to be preserved in Anatolian and the western dialects” and therefore
“simply are not probative” for determining the position of Anatolian in the
Indo-European family tree: only secured common innovations can be used to
this end. Melchert himself thinks that such common innovations between
Anatolian and “western” languages may indeed exist, but, to his mind, they
would rather prove “post-divergence contact between Anatolian and the west-
ern dialects” (2016: 300), and thus have no bearing on the genealogical
position of Anatolian. Although space limitations do not allow me to examine

44
This tree largely coincides with the trees given by Oettinger (1978: 92) and Yakubovich
(2010: 6).

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 77

the four examples treated by Melchert (2016), it is quite clear that none of them
can withstand scrutiny. There is thus no reason to assume that Anatolian shares
any innovations, either contact-induced or caused by a genetic relationship,
with “western” Indo-European languages or, for that matter, with any of the
other Indo-European languages.

5.5 The Position of Anatolian

A hotly debated issue with regard to Anatolian is the so-called Indo-Anatolian
hypothesis (also “Indo-Hittite hypothesis”), which states that Anatolian was
the first branch to split off from the Indo-European mother language (which
then may be called “Proto-Indo-Anatolian” or “Proto-Indo-Hittite”), after
which the remaining language, which was to become the mother language
of all the non-Anatolian Indo-European languages (and which may be called
“Core Proto-Indo-European”, “Nuclear Proto-Indo-European”, “Classical
Proto-Indo-European” or similar) underwent a set of innovations. There is
some debate on whether this hypothesis is valid at all, and if so, how large the
gap is between the moment Anatolian split off and the time that the first split
within Core Proto-Indo-European (CPIE) took place (which is usually
thought to have been the split-off of Tocharian, see Chapter 6). Some scholars
do not think that there is enough evidence for assuming an early split-off of
Anatolian at all (Rieken 2009; Adiego 2016); others think that there may have
been an early split, but that the gap between Anatolian and the next split is
relatively modest (Eichner 2015; Melchert in press), whereas still others think
that the gap is sufficiently large for the Proto-Indo-Anatolian ancestor lan-
guage to be substantially different from Core Proto-Indo-European
(Kloekhorst 2008: 7–11; Oettinger 2014; Kloekhorst & Pronk 2019).
The validity of the Indo-Anatolian hypothesis can only be proven if enough
secured shared innovations of the non-Anatolian languages can be found. In
Kloekhorst & Pronk 2019: 3–6, a total of thirty-four linguistic features are
listed in which Anatolian deviates from the other Indo-European languages,
and which are presented as possible cases in which Anatolian has retained the
original state of affairs, whereas the other Indo-European languages have
undergone a common innovation (with twenty-three examples classified as
“good candidates”, and eleven as “less forceful” but “promising” ones). These
include:
Semantic Innovations
• The Hittite participle suffix -ant- forms both active and passive participles,
whereas in CPIE the suffix *-e/ont- only forms active participles: narrowing
of the function of *-e/ont- in CPIE (Oettinger 2014: 156–7).

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 78 Alwin Kloekhorst

• The Hitt. active verb ēš-zi < *h1es-ti means ‘to sit’ next to its middle
counterpart eš-a(ri) < *h1e-h1s-°, which means ‘to sit down’, whereas in
CPIE the middle verb *h1e-h1s-to means ‘to sit’ next to the verbal root
*sed- ‘to sit down’: expansion of the meaning of *h1e-h1s- from ‘to sit
down’ to ‘to sit’, with replacement of *h1e-h1s- ‘to sit down’ by *sed-
(Norbruis 2021: 235–41).
• Hitt. ḫarra-i < *h2erh3- means ‘to grind, crush’, whereas CPIE *h2erh3-
means ‘to plough’: semantic specialisation in CPIE (Kloekhorst 2008: 9).
• Hitt. mer- < *mer- means ‘to disappear’, whereas CPIE *mer- means ‘to die’:
semantic shift, through euphemism, in CPIE (Kloekhorst 2008: 8).
Morphological Innovations
• Anatolian has two genders (common/neuter), whereas CPIE has three genders
(m./f./n.): creation of the feminine gender in CPIE (e.g. Melchert in press).
• Anatolian has nom. *ti(H), obl. *tu- ‘ you (sg.)’, whereas CPIE has nom.
*tuH, obl. *tu-: spread of obl. stem *tu- to the nominative in CPIE
(Kloekhorst 2008: 8–9).
• Anat. *h1eḱu- vs. CPIE *h1eḱu-o- ‘horse’: thematization in CPIE
(Kloekhorst 2008: 10).
• Hitt. ḫuu̯ ant- < *h2uh1-ent- vs. CPIE *h2ueh1nt-o- ‘wind’: thematization in
CPIE (Eichner 2015: 17–18).
Sound Changes
• Anat. *h2 = *[qː] and *h3 = *[qːʷ] vs. CPIE *h2 = *[ħ] or *[ʕ] and *h3 = *[ħʷ]
or *[ʕʷ]: fricativization of uvular stops in CPIE (Kloekhorst 2018a).
• Hitt. amm- < *h1mm- (< pre-PIE *h1mn-) vs. CPIE *h1m- ‘me’: degemina-
tion of *mm to *m in CPIE (Kloekhorst 2008: 111 n. 234).
Although it is certainly possible that not all of the arguments listed in
Kloekhorst & Pronk 2019 will eventually become generally accepted, it
seems very unlikely that they will all be refuted, and the Indo-Anatolian
hypothesis can thus be regarded as virtually proven. Moreover, since the
number of arguments listed is relatively large and some of them concern
significant structural innovations (especially the rise of the feminine gender
in CPIE, including the creation of the accompanying morphology), it has been
argued that the temporal gap between the Anatolian split and the subsequent
Tocharian split (cf. Chapter 6) may have been in the range of 800–1000 years.
With the Tocharian split commencing around 3400–3300 BCE, the Anatolian
split may be dated to the period between 4400–4100 BCE. If Proto-Anatolian
indeed first broke up into its daughter languages around the thirty-first century
BCE (see Section 5.3.5), it would mean that it had some 1,300–1000 years to
undergo the specific innovations that define Anatolian as a separate branch (see
Section 5.2). Since these innovations include some large restructurings of
especially the verbal system (loss of the subjunctive and optative mood, merger

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 79

of the present and aorist aspects, creation of the ḫi-conjugation on the basis of
the PIE perfect), such a time span would certainly be fitting.

References
ACLT = Annotated Corpus of Luwian Texts. http://web-corpora.net/LuwianCorpus.
Adiego, Ignasi-Xavier. 2001. Lenición y acento en protoanatólico. In Onofrio Carruba
& Wolfgang Meid (eds.), Anatolisch und Indogermanisch, 11–18, Innsbruck: Institut
für Sprachen und Literaturen der Universität Innsbruck.
Adiego, Ignasi-Xavier. 2007. The Carian language. Leiden: Brill.
Adiego, Ignasi-Xavier. 2016. Anatolian languages and Proto-Indo-European, Veleia 33.
49–64.
Brixhe, Claude. 1988. La langue des inscriptions épichoriques de Pisidie. In Yoël
L. Arbeitman (ed.), A linguistic happening in memory of Ben Schwartz, 131–55.
Louvain-la-Neuve: Peeters.
Carruba, Onofrio. 1970. Das Palaische: Texte, Grammatik, Lexikon. Wiesbaden:
Harrassowitz.
CHD = Hans G. Güterbock, Harry A. Hoffner & Theo P. J. van den Hout (eds.). 1983–.
The Hittite dictionary of the Oriental Institute of the University of Chicago. Chicago,
IL: The Oriental Institute of the University of Chicago.
Čop, Bojan. 1970. Eine luwische orthographische-phonetische Regel. Indogermanische
Forschungen 75. 85–96.
EDHIL = Alwin Kloekhorst. 2008. Etymological dictionary of the Hittite inherited
lexicon. Leiden: Brill.
Eichner, Heiner. 1973. Die Etymologie von heth. mehur. Münchener Studien zur Sprachwis-
senschaft 31. 53–107.
Eichner, Heiner. 1974. Untersuchungen zur hethitischen Deklination. Inaugural-Dissertation
(Teildruck), Erlangen.
Eichner, Heiner. 1975. Die Vorgeschichte des hethitischen Verbalsystems. In Helmut
Rix (ed.), Flexion und Wortbildung, 71–103. Wiesbaden: Reichert.
Eichner, Heiner. 2015. Das Anatolische in seinem Verhältnis zu anderen Gliedern der
indoeuropäischen Sprachfamilie aus aktueller Sicht. In Thomas Krisch &
Stefan Niederreiter (eds.), Diachronie und Sprachvergleich, 13–26. Innsbruck:
Institut für Sprachen und Literaturen der Universität Innsbruck.
Frotscher, Michael. 2012. Die luwischen Entsprechungen der hethitischen Verben des
Typs dāi/tii̯ anzi und ein neues Lautgesetz urindogermanisch *oi̯ > urluwisch *u̯ e.
International Journal of Diachronic Linguistics and Linguistic Reconstruction 9.
137–94.
Gérard, Raphaël. 2005. Phonétique et morphologie de la langue lydienne. Louvain-
la-Neuve: Peeters.
Giusfredi, Federico. 2020. A study in the syntax of the Luwian language. Heidelberg:
Winter.
Gusmani, Roberto. 1964. Lydisches Wörterbuch: Mit grammatischer Skizze und
Inschriftensammlung. With Ergänzungsband Lfg. 1 (1980), Lfg. 2 (1982), Lfg. 3
(1986). Heidelberg: Winter.
Hajnal, Ivo. 1995. Der lykische Vokalismus. Graz: Leykam.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 80 Alwin Kloekhorst

Hawkins, John David. 2000 Corpus of Hieroglyphic Luwian inscriptions. Vol. 1.

Inscriptions of the Iron Age. Berlin: de Gruyter.
HED = Jaan Puhvel. 1984–. Hittite etymological dictionary. Berlin: Mouton.
HEG = Johann Tischler. 1983–2016. Hethitisches etymologisches Glossar. Innsbruck:
Institut für Sprachwissenschaft der Universität Innsbruck.
Hoffner, Harry A., Jr. & H. Craig Melchert. 2008. A grammar of the Hittite language.
Winona Lake, IN: Eisenbrauns.
HW² = Johannes Friedrich & Annelies Kammenhuber. 1975–. Hethitisches Wörterbuch.
2nd ed. Heidelberg: Winter.
Jasanoff, Jay H. 2003. Hittite and the Indo-European verb. Oxford: Oxford University
Press.
Kalinka, Ernestus. 1901. Tituli Lyciae lingua Lycia conscripti. Vindobonae: Hölder.
Kammenhuber, Annelies. 1969. Hethitisch, Palaisch, Luwisch und Hieroglyphenluwisch.
In Johannes Friedrich (ed.), Altkleinasiatische Sprachen, 119–357. Leiden: Brill.
Kimball, Sara E. 1999. Hittite historical phonology. Innsbruck: Institut für Sprachwis-
senschaft der Universität Innsbruck.
Kloekhorst, Alwin. 2006. Initial laryngeals in Anatolian. Historische Sprachforschung
119. 77–108.
Kloekhorst, Alwin. 2008. Etymological dictionary of the Hittite inherited lexicon.
Leiden: Brill.
Kloekhorst, Alwin. 2012. The origin of the Lydian dat.sg. ending -λ. Kadmos 51. 165–73.
Kloekhorst, Alwin. 2013. Likijskij jazyk [The Lycian language]. In Y. B. Koryakov &
A. A. Kibrik (eds.), Jazyki mira: Reliktovye indoevropejskie jazyki Perednej
i Central’noj Azii [Languages of the world: Relict Indo-European languages of
Western and Central Asia], 131–54. Moscow: Academia.
Kloekhorst, Alwin. 2014. Accent in Hittite: A study in plene spelling, consonant
gradation, clitics, and metrics. Wiesbaden: Harrassowitz.
Kloekhorst, Alwin. 2016. The Anatolian stop system and the Indo-Hittite hypothesis.
Indogermanische Forschungen 121. 213–47.
Kloekhorst, Alwin. 2018a. Anatolian evidence suggests that the Indo-European laryn-
geals *h2 and *h3 were uvular stops. Indo-European Linguistics 6. 69–94.
Kloekhorst, Alwin. 2018b. The origin of the Hittite ḫi-conjugation. In Lucien van Beek
et al. (eds.), Farnah: Indo-Iranian and Indo-European studies in honor of Sasha
Lubotsky, 89–106. Ann Arbor, MI: Beech Stave.
Kloekhorst, Alwin. 2019. Kanišite Hittite: The earliest attested record of Indo-
European. Leiden: Brill.
Kloekhorst, Alwin. In press. Proto-Indo-Anatolian, the “Anatolian split” and the
“Anatolian trek”: A historical linguistic perspective. In Kristian Kristiansen &
Guus Kroonen (eds.), The Indo-European puzzle revisited. Cambridge: Cambridge
University Press.
Kloekhorst, Alwin & Tijmen Pronk. 2019. Introduction: Reconstructing Proto-Indo-
Anatolian and Proto-Indo-Uralic. In Alwin Kloekhorst & Tijmen Pronk (eds.), The
precursors of Proto-Indo-European: The Indo-Anatolian and Indo-Uralic hypotheses,
1–14. Leiden: Brill.
Kloekhorst, Alwin & Willemijn J. I. Waal. 2019. A Hittite scribal tradition predating the
tablet collections of Ḫattuša? The origin of the “cushion-shaped” tablets KBo 3.22,
KBo 17.21+, KBo 22.1, and KBo 22.2. Zeitschrift für Assyriologie 109. 189–203.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 5 Anatolian 81

Kroonen, Guus, Gojko Barjamovic & Michaël Peyrot. 2018. Linguistic supplement to
Damgaard et al. 2018: Early Indo-European languages, Anatolian, Tocharian and
Indo-Iranian. doi:10.5281/zenodo.1240524.
Laroche Emmanuel. 1979. L’inscription lycienne. In Henri Metzger (ed.), Fouilles de
Xanthos. Vol. 6. La stèle trilingue de Létôon, 49–127. Paris: Klincksieck.
Malzahn, Melanie. 2010. The Tocharian verbal system. Leiden: Brill.
Marazzi, Massimiliano (ed.). 1998. Il geroglifico anatolico: Sviluppi della ricerca
a venti anni dalla sua “ridecifrazione”. Napoli: Dipartimento di studi glottoantro-
pologici, Università “La sapienza”.
Melchert, H. Craig. 1990. Adjective stems in *-iyo- in Anatolian. Historische
Sprachforschung 103. 198–207.
Melchert, H. Craig. 1993. Cuneiform Luvian lexicon. Chapel Hill, NC: n.p.
Melchert, H. Craig. 1994. Anatolian historical phonology. Amsterdam: Rodopi.
Melchert, H. Craig. 2003. Language. In H. Craig Melchert (ed.), The Luwians, 170–210.
Leiden: Brill.
Melchert, H. Craig. 2004. A dictionary of the Lycian language. Ann Arbor, MI: Beech
Stave.
Melchert, H. Craig. 2011. The PIE verb for ‘to pour’ and medial *h3 in Anatolian. In
Stephanie Jamison, H. Craig Melchert & Brent Vine (eds.), Proceedings of the 22nd
UCLA Indo-European Conference, 127–32. Bremen: Hempen.
Melchert, H. Craig. 2016. “Western affinities” of Anatolian. In Bjarne S. S. Hansen et al.
(eds.), Etymology and the European lexicon, 297–305. Wiesbaden: Reichert.
Melchert, H. Craig. In press. The position of Anatolian. In Michael Weiss & Andrew Garrett
(eds.), Handbook of Indo-European studies. Oxford: Oxford University Press.
Meriggi, Piero. 1962. Hieroglyphisch-hethitisches Glossar. Wiesbaden: Harrassowitz.
Neumann, Günter. 1979. Neufunde lykischer Inschriften seit 1901. Vienna: Österreichische
Akademie der Wissenschaften.
Neumann, Günter. 2007. Glossar des Lykischen: Überarbeitet und zum Druck gebracht
von Johann Tischler. Wiesbaden: Harrassowitz.
Norbruis, Stefan. 2021. Indo-European origins of Anatolian morphology and seman-
tics: Innovations and archaisms in Hittite, Luwian and Lycian. Amsterdam: LOT.
Oettinger, Norbert. 1978. Die Gliederung des anatolischen Sprachgebietes. Zeitschrift
für vergleichende Sprachwissenschaft 92. 74–92.
Oettinger, Norbert. 2014. Die Indo-Hittite-Hypothese aus heutiger Sicht. Münchener
Studien zur Sprachwissenschaft 67. 149–76.
Payne, Annick. 2010. Hieroglyphic Luwian: An introduction with original texts. 2nd
rev. ed. Wiesbaden: Harrassowitz.
Payne, Annick & Jorit Wintjes. 2016. Lords of Asia Minor: An introduction to the
Lydians. Wiesbaden: Harrassowitz.
Pérez Orozco, Santiago. 2007. La lengua sidética, Ensayo de síntesis. Kadmos 46.
125–42.
Puhvel, Jaan. 1994. West-Indo-European affinities of Anatolian. In George E. Dunkel
et al. (eds.), Früh-, Mittel-, Spätindogermanisch, 315–24. Wiesbaden: Reichert.
Rieken, Elisabeth. 2005. Neues zum Ursprung der anatolischen i-Mutation. Historische
Sprachforschung 118. 48–74.
Rieken, Elisabeth. 2009. Der Archaismus des Hethitischen: Eine Bestandsaufnahme.
Incontri Linguistici 32. 37–52.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press

 82 Alwin Kloekhorst

Rieken, Elisabeth. 2017. The dialectology of Anatolian. In Jared S. Klein, Brian

D. Joseph & Matthias Fritz (eds.), Handbook of comparative and historical Indo-
European linguistics. Vol. 1, 298–308. Berlin: De Gruyter Mouton.
Sasseville, David. 2017. The Lydian nominal paradigm of i-mutation. Indo-European
Linguistics 5. 130–46.
Shevoroshkin, V. V. 2013. Milijskij jazyk [The Milyan language]. In Y. B. Koryakov &
A. A. Kibrik (eds.), Jazyki mira: Reliktovye indoevropejskie jazyki Perednej
i Central’noj Azii [Languages of the world: Relict Indo-European languages of
Western and Central Asia], 154–66. Moscow: Academia.
Starke, Frank. 1985. Die keilschrift-luwischen Texte in Umschrift. Wiesbaden:
Harrassowitz.
Starke, Frank. 1990. Untersuchungen zur Stammbildung des keilschrift-luwischen
Nomens. Wiesbaden: Harrassowitz.
Yakubovich, Ilya. 2010. Sociolinguistics of the Luvian language. Leiden: Brill.
Yakubovich, Ilya. 2015. The Luwian language. In Oxford handbooks online.
doi:10.1093/oxfordhb/9780199935345.013.18.
Yoshida, Kazuhiko. 1991. Anatolian verbal endings: The third person plural preterite.
Journal of Indo-European Studies 19. 359–74.
Yoshida, Kazuhiko. 1993. Notes on the prehistory of the preterite verbal endings in
Anatolian. Historische Sprachforschung 106. 26–35.

https://doi.org/10.1017/9781108758666.005 Published online by Cambridge University Press