Professional Documents
Culture Documents
Counting in Hieroglyphic Luwian
Counting in Hieroglyphic Luwian
ABSTRACT
This study investigates Hieroglyphic Luwian nominal number marking in connection
with numerals in a detailed corpus analysis, bringing together philology and linguistic
analysis. The result is that the numerals two, three and four mostly take plural
nouns, while numerals from ve onwards never do so. This leaves the language with a
two-tier system of grammatical number marking, which has many typological parallels.
With lower numerals, plural marking in the noun is triggered by the inection of the
numerals themselves, and data from other languages in the region show that the (non-)
application of plural marking may well be an areal feature.
1. INTRODUCTION
Numerals and what they quantify are a staple of any language, yet often they receive little
attention in grammatical description. This lack of information may be attributed to the
relatively low degree of attention numerals have attracted in general. In particular, older
grammars frequently neglect to mention the agreement numerals call for, or whether some
numerals are treated dierently from others. Such is also the case for Hieroglyphic Luwian
(HLuwian).
1.1. Grammatical description
Several grammatical sketches are available for HLuwian, the most detailed one being Plochls
2003 Introduction to Hieroglyphic Luwian. On the topic of numerals, he states that even with
values greater than one, the accompanying noun can still show singular inection, to which
he adds a small collection of examples (Plochl 2003: 99). Melchert (2003: 201) comes to the
same conclusion in his short chapter on grammar: Cardinal numbers greater than one may
take either singular or plural. A similar statement is made by Payne in the 2nd edition of her
introduction to HLuwian (2010: 33). Nothing more detailed has been said on agreement in
quantied phrases so far, and it is the aim of the present paper to ll this particular gap in the
grammatical description of HLuwian.
1.2. Linguistic aliation
Luwian is an Indo-European language and belongs to the Anatolian branch. It is a sister
language to Hittite, which has a much larger corpus of texts. The languages more closely
related to Luwian are Lydian, Carian and Lycian, and they are considerably less well attested.
1
I am grateful to Professors Elisabeth Rieken and Michael Job for their valuable suggestions and constant support,
and I would also like to thank the participants of the Cambridge Symposium of the Philological Society in March
2010 for their thought-provoking questions and comments on my talk, which the present paper is based on.
The author 2011. Transactions of the Philological Society The Philological Society 2011. Published by Blackwell Publishing, 9600
Garsington Road, Oxford OX4 2DQ and 350 Main Street, Malden, MA 02148, USA.
BAUER
221
Within the Anatolian branch, the exact relationship of the languages to each other is still
being debated (cf. Melchert 2003: 1757).2
Luwian is attested in two dierent strata. Cuneiform Luwian (CLuwian) originates from
the Hittite period in the second millennium BC, while HLuwian attestation mostly followed
in its wake. There are numerous hieroglyphic seals and some texts predating the collapse of
the Hittite empire; however, the bulk of the HLuwian material belongs to the period
between c.1200 and c.700 BC (cf. e.g. Hawkins 2000: 13). The HLuwian texts are, for the
most part, inscriptions on stone, scattered over a vast area stretching from southeastern
Anatolia to northern Syria, with some locations oering a large quantity of textual
evidence, others very little. There are also several lead strips, whose origins are more
dicult to ascertain, but which may be dated to the eighth century BC (cf. Hawkins 2000:
505, 534).
The writing system of HLuwian is a combination of a syllabary and numerous logograms,
from which at least some of the syllabic readings are derived. The syllabic signs complicate the
spelling of consonant clusters considerably, particularly as the HLuwian system does not
include any VC or CVC signs. Thus the phonetic shape of the words and especially the
endings often remain somewhat obscure due to the large number of dead vowels (Hawkins
2003: 1569). Additionally, a syllable-nal n is usually omitted in writing (and very possibly
in pronunciation) if followed by another consonant, occasionally resulting in ambiguity
(Melchert 2003: 182).
2. CORPUS
STUDY
2.1. Data
The data listed in Table 1 were collected through searches of an electronic version of the
HLuwian corpus, which includes all published texts. The data are split into three large
columns and two smaller ones according to the form of the quantied noun, which can be
either singular, plural or a bare logogram, i.e. without any visible inection. Ablative Instrumental case forms are listed separately in a smaller column, since that particular case does not
distinguish number. The rightmost column contains passages which are too poorly
understood to permit classifying the noun that might be modied by the numeral; the
column also includes partly destroyed and incomplete phrases.
2.2. Analysis
As the corpus of HLuwian is comparatively small, only a limited number of numerals is
attested.3 Thus the numerals ranging from two to twelve are all present in the data, while
higher values are less often available. Most older Indo-European languages feature a decimal
system, to which HLuwian is no exception, and the attestation of nearly all decades up to one
hundred and in part even beyond comes as no surprise.
Nearly all numerals are written as logograms, not syllabically. In a mere handful of cases,
the words for two and three are actually spelt out (a fully syllabic spelling for four is only
attested for CLuwian):
For the purpose of the present study, the data concerning the numeral one is excluded from Table 1.
222
109, 2011
1
1
4
1
10
1
8
11
12
14
15
2
7
21
22
25
27
28
30
32
40
42
50
55
60
68
20
Uncertain status
Abstract
Measure
AI
Inanimate
Concrete
Unclear meaning
Animate
Non-hum
Measure
Inanimate
Concrete
Non-hum
4
5
Animate
Human
Human
Abstract
Measure
Non-hum
Inanimate
Concrete
No.
Human
Animate
Bare
Unclear meaning
Sg
1
1
BAUER
223
Table 1. Continued
Pl
69
100
120
131
133
140
150
151
170
200
220
308
400
500
1
1
1
3,000
4,000
112
1,000
Uncertain status
97
600
Abstract
80
141
Measure
75
111
Inanimate
Concrete
AI
Non-hum
Animate
Human
Measure
Concrete
Non-hum
Inanimate
Unclear meaning
Bare
Animate
Human
Abstract
Measure
Inanimate
Concrete
Non-hum
No.
Human
Animate
Unclear meaning
Sg
1
1
4,100
4,400
224
109, 2011
(1)
MARAS
(2)
TOPADA
There are no such attestations for any numerals beyond that, and their phonetic shape can at
most be inferred from other, related languages.
Table 1 also shows that the nouns that are quantied cover the entire range of basic
semantics from [HUMAN] to [ABSTRACT]. Abstract entities are overall rare due to the nature of
the texts. Numerals in HLuwian are mainly used to name the exact size of commodities in
business transactions or rituals, and they seldom occur in other contexts.
2.3. The distribution of plural marking
The most striking fact in the distribution is that the attestation of plural nouns cuts o after
the numeral four, as can be seen from Table 1. It thus appears that plural marking depends
on the value of the numeral. Up to four, the numerals combine with mostly plural nouns, but
thereafter they do not. This suggests the presence of a grammatical rule of the following
content: the numerals ve and higher permit only singular nouns, while the numerals two
to four permit plurals.
In light of the comparatively small size of the corpus and the fact that HLuwian attestations
are scattered over a rather large geographical area, it may be tempting to pass these forms o
as accidental. However, a closer look at the data reveals that the distribution holds true within
one and the same text, and even with the same word:
(3)
Here, the recipients are three statues, marked as plural. The following two examples are
taken from the preceding paragraphs on the same lead strip:
4
Evidence from this passage must be taken with caution, as it is slightly damaged, and the word boundaries are not
altogether clear; cf. also Hawkins remarks (2000: 257). However, the numeral is the only attestation of two so far
that is fully spelt out.
5
Double quotation marks are conventionally used to transliterate logogram markers, which are used frequently, but
by no means reliably. The sign | indicates the presence of a so-called word divider and || a line break in the
HLuwian original. A single apostrophe at the end of a word marks a case of initial-a-nal (cf. Hawkins 2003: 159
61). The combination of asterisk and gure marks logograms without an assured interpretation.
6
HLuwian, like most Anatolian languages, distinguishes two grammatical genders: common (C) and neuter (N). The
basic word order is SOV, and the rst word of a sentence will usually be followed by at least one particle (PTC) that
attaches to it (Melchert 2003: 185, 200). Some words are accompanied by so-called determinatives (DET) that act as
graphemic classiers (Hawkins 2003: 158). Other abbreviations include: GA (genitival adjective), GER (gerundive), IPV
(imperative), PRN (pronoun), PRS (present tense), PRT (preterite).
BAUER
(4)
(5)
225
The recipients are ve statues and seven statues respectively, and the nouns are marked as
singulars both times. Similarly, the sheep that are being allocated in these examples are also
consistently marked as singulars (or appear as bare logograms). Yet the HLuwian word for
sheep can show plural marking:
(6)
KULULU lead strip 3 provides additional evidence for the split in number marking between
four and ve with <-zi> marking the plural and <-sa> the singular:
(7)
The pattern is by no means limited to the KULULU lead strips. Further evidence can be
adduced, for instance, from the ASSUR letters:
(8)
(9)
ASSUR
ASSUR
While the meaning of the noun aruta i- is not wholly clear, its inection poses no diculties.
In example (8) it is clearly marked as a plural, following the numeral four directly, and both
numerals are also inected as plurals. In example (9), however, the noun follows the numeral
ve and shows singular inection as expected. The numeral ve also appears to be inected
226
109, 2011
in this case, but the form is a hapax; since nothing else is known about the actual phonetic
shape of ve, the issue of its inection must remain speculation.
For the distribution of number marking, the grammatical gender of the quantied nouns
appears to be of no importance. The same rules apply to common gender nouns as to neuter
nouns, as can be seen in the previous examples: the noun taru(t)- statue (examples (3)(5)) is
a neuter, while OVIS sheep (example (6)), VIR(-ti-) man (example (7)) and aruta i- basket?
(examples (8) and (9)) are all common gender. Likewise, it has become clear from the
examples that the numerals ve and higher do not take nouns in the genitive. Instead, the
noun always appears in the case required in the sentence.
There is one case of a plural noun that could possibly be cited as a counterexample, because
it is combined with the numeral fty:
(10)
ASSUR
Yet a closer look at the context shows that in this passage, the numeral occurs after the
noun it quanties, which is very unusual and happens in only a handful of (mostly
doubtful) cases out of more than 150. It is therefore suggested that post-nominal numerals
are appositions and should be translated accordingly instead of treating them in the same
way as the other numerals. As apposition, the numeral has no direct inuence on the
inection of the noun; instead it is a quantication added to a formerly unspecied group
of more than one.
There are two other possible counterexamples. One is 6? VIR-pzi?q (KULULU lead strip 3 ii
reg. 1). However, the sign tentatively restored zi is located in the fold of the strip and thus so
badly damaged that it is impossible to be certain about the reading. The other possible
counterexample occurs in the following sentence:
(11)
In this case, one can be reasonably certain about the plural ending of the noun DEUS-[ni]zi, yet much of that part of the stele is so badly abraded that the numeral is anything but
legible.
What seemed solid counterexamples at rst sight have, under closer scrutiny, turned out to
be nothing of the kind. These three passages therefore highlight the importance of a sound
philological investigation of the material when conducting linguistic analyses, because it
ideally forestalls phantom problems. On the other hand, the examples also emphasise the
importance of a linguistic basis for philological work. The extreme rarity of a construction in
a certain language (e.g. a numeral following a noun in HLuwian) becomes wholly clear only in
a statistical analysis, which puts the phenomenon into the right perspective. Without such
empirical work, the grammatical description of a language must remain impressionistic in
parts.
While there are thus no combinations of numerals ve and higher with plural nouns,
Table 1 shows that there are a handful of examples in which the numerals two, three and
BAUER
227
four are followed by a singular noun. This is the case for four7 out of 24 unambiguously
inected nouns, and it is unlikely that these singulars are simply mistakes. Yet it is dicult to
nd a conditioning factor.
For instance, deniteness can probably be ruled out: (7) 5 includes: 3 VIR-ti-zi 2
FEMINA-ti-sa, i.e. a correct plural noun followed by an unexpected singular noun. The
context is in no way dierent from any of the preceding ones, and the NP cannot possibly be
argued to be indenite. It may be that the animacy hierarchy (cf. e.g. Corbett 2000: 69, 217
18) plays a small role in the distribution. Table 1 reveals that out of six [HUMAN] nouns with
number marking, only one is a singular, whereas two out of three [NON-HUMAN, ANIMATE]
nouns are singulars. However, all [INANIMATE] nouns show plural marking, which runs counter
to all expectations based on animacy.
Finally, the most promising explanation may be language change, as three of the four
singular nouns8 are late attestations, dating to the mid- to late eighth century (cf. Hawkins
2000: 465, 505, 534). It would not be far-fetched to assume that the consistent singular
marking with numerals from ve onwards eventually aected the nouns quantied by lower
numerals.
2.4. Morphological cueing
The data thus conrm the hypothesis presented earlier: the presence or absence of plural
marking depends on the numeral quantifying the noun. This pattern raises a number of
questions, the rst of which is why plural marking is cut o after the numeral four.
HLuwian is by no means the only language to treat nouns following ve and higher
dierently. Old Church Slavonic, for instance, requires the nouns to be genitives in those
cases, even though it does not with lower numerals. Similar patterns are still in use in some
of the modern Slavic languages (Comrie 1991: 7489). The evidence from Old Church
Slavonic and some of the other Slavic languages draws attention to a pattern common to
most older Indo-European languages, viz. the indeclinability of numerals from ve
onwards contrasted with the numerals two, three and four, which inect fully for
number, gender and case, usually according to a pronominal pattern (cf. e.g. Fortson 2004:
130). This system is maintained by Vedic Sanskrit (Wackernagel 1930: 339), Greek
(Schwyzer 1939: 587) and Old Irish (Thurneysen 1946: 242), while most of the other IndoEuropean languages have either restricted full inection to fewer numerals or abandoned it
altogether.
HLuwian undoubtedly inected its numerals two, three and four, as has become clear
from several examples in which a phonetic complement is added to the number used to write
the numeral (cf. (8)). Apart from one unclear case (cf. (9)), there is no phonetic
complementation with any of the higher numbers. This suggests that the presence or absence
of inection works as a morphological cue: a plural ending in the numeral can thus trigger a
plural ending in the noun that follows it. Without such a trigger, however, the noun is not
turned into a plural. Singular or plural marking of a noun does not therefore depend on the
inherent value of the numeral as such (indicating a plurality of entities), but on its form, viz.
the presence of the plural ending in the numeral itself. Such a pattern brings the
7
See Table 1; the relevant passages are: 2 FEMINA-ti-sa two women (KULULU lead strip 3 1.5), 2 OVIS-sa two
sheep (SULTANHAN 29), 3 ANNUS-si-si-na OVIS(ANIMAL)-wa i-na three yearling sheep (MARAS 11 8) and 4-zi
(*78)a-ru-ti-na four baskets (ASSUR letter c 9, cf. (9)), however, in that particular case, the noun is directly
preceded by ve, and not four.
8
The dating of
MARAS
228
109, 2011
Suxaufnahme phenomenon to mind, which is marginally present in Hittite9 and a little more
common in Luwian, very probably stemming from prolonged contact with Hurrian (cf.
Yakubovich 2006; 2008).
2.5. The system in typological terms
The results so far lead to the question of what kind of number system HLuwian employs.
Number systems have received renewed attention over the past decade after a rst foray into
the eld by Hurford (1975). Corbett (2000) explores the area systematically and uncovers an
astonishing amount of variation across languages. Thus it is quite common for languages to
mark some nouns for number, but not others,10 or to treat number marking as optional, and
it is also possible to employ more than one system for phrases quantied by numerals
(Corbett 2000: 11, 211; Rijkho 2002: 38). These subsystems, or tiers, are arranged under a
top tier, which in HLuwian is a singular plural dichotomy.
Like many other languages, HLuwian usually marks number redundantly in a sentence by
requiring number agreement between subject and verb. Often this kind of redundancy extends
to quantied expressions as well, resulting in numerals greater than one being combined with
plural11 nouns. In a combination of a numeral and a noun, the numeral is the controller and
the noun the target, whose grammatical number is controlled by the numeral. It therefore
stands to reason that in HLuwian, the controllers two, three and four have a feature value
[+PLURAL], which the controllers ve and higher apparently lack. Any noun following a
numeral that lacks the feature [+PLURAL] therefore defaults to the only other form there is, i.e.
the singular (cf. Corbett 2000: 178). It is clear that a singular form does not necessarily signify
a single entity, instead it constitutes an instance of general number (also called a
transnumeral) when combined with a numeral.12 Cross-linguistically, general number forms
are often identical with singulars (Corbett 2000: 14).
Unfortunately, no further agreement data can be adduced in the case of HLuwian
quantied expressions. One reason is that quantied expressions occupy subject position only
extremely rarely. Yet when they do constitute subjects, it is usually in existential sentences,
which lack a verb form altogether. There are also several cases with ambiguous verb forms,
whose orthography does not permit any conclusive interpretation; cf.
(12)
8
REX-ti-sa
sa-tax
PTC
PRN.1.SG abbr. NUM king.N.SG abbr. be.PRT.3.SG PL
Against me in the city Parzuta eight kings, lesser and more
important, were hostile(?).
TOPADA
Similarly, there are no relevant anaphoric data that could be used.13 Thus it is impossible on
the basis of the available data to determine whether HLuwian requires syntactic or semantic
number agreement beyond the very limited connes of the NP.14
9
Cf. also the occurrence of both singulars and plurals in Old Script texts in Hittite (Honer & Melchert 2008: 159).
10
Cf. e.g. English cow cows vs. sheep sheep and sh sh.
11
12
For a somewhat dierent account see Rijkho, in whose classication the HLuwian nouns would be termed set
nouns (2002: 289).
13
ASSUR
14
letter f+g 41 could be adduced as an example if the reference of the pronoun were clear.
Cf. English for semantic agreement: The committee have decided (Corbett 2000: 18791).
BAUER
229
a-i-ya-am-mi-zi
make.PPL.N.PL.C
GIS
GA.ZUM-za
comb.N A.SG.N
GIS
GA.ZU[(M-za
comb.N A.SG.N
The numeral four in (13a) is combined with a plural noun, whereas both instances of the
numeral nine are followed by singular nouns.16 While these ndings correspond exactly to
the hypothesis postulated here for HLuwian, three tokens are too few on which to base a
conclusion.
The Lycian data, albeit more plentiful, is even less informative, since the analysis of
numerous forms is still under debate. Even the exact meaning of many numerals is far from
clear for instance, nuntata could mean nine, nineteen or even ninety (cf. Melchert 2004:
44; Neumann & Tischler 2007: 245).
Hittite, the most well-attested Anatolian language, oers a much larger amount of data and
also a far more complex picture. Most numerals can take both singulars and plurals already in
Old Script texts (cf. Honer & Melchert 2008: 1589, 1612), sometimes even resulting in a
fairly even distribution between the two. Additionally, plural marking does not cut o at any
point, and the Hittite system is apparently one in which numerals greater than one can be
accompanied by singular and plural nouns alike i.e. plural marking after numerals is
optional in all cases. However, further complications arise from the frequent use of logograms
in Hittite, some of which are followed by logographic plural markers. While this would
usually be taken as a sign that the underlying Hittite form is a plural, there are cases in which
such a logographic plural is then followed by a Hittite phonetic complement that clearly
marks it as singular:
(14) Hittite, KBo 32.13 ii, lines 15 and 17
a. SIG7-an GU4H I:A -un
NUM
cow.PL-A.SG.C
ten thousand cattle
UDU
b. SIG7
GUKKAL?+KUNH I:A -n=a
NUM
fat-tailed.sheep.PL-A.SG.C=PTC
ten thousand fat-tailed sheep
It is therefore impossible to be certain about the underlying grammatical number of
logographic forms if there is no phonetic complement.
15
16
Both numerals are also suxed with the particle -za, which is commonly assumed to attach only to neuters in the
nominative accusative singular. Its function, however, is entirely unclear and thus requires further investigation. I
would like to thank Anna Morpurgo Davies for raising this question.
230
109, 2011
KBo 32.19 i 6
three kings
KBo 32.19 i 4
nine kings
KBo 32.13 i 15
ten thousand (heads of) cattle
KBo 32.13 i 16-17
thirty thousand sheep
If the analysis of the forms is correct, then there is plural marking in nouns that follow the
numerals three and nine, while the numerals ten thousand and thirty thousand
apparently do not eect such marking. It is also possible that the dierent animacy status of
the nouns plays a role in the distribution.
The evidence is certainly too sketchy to permit a denite conclusion regarding the origin of
the HLuwian system. However, it has become clear that plural marking with higher numerals
is by no means mandatory in any of the languages in Anatolia that provide the necessary
data. What appeared to be an idiosyncratic quirk of Luwian at rst sight can, therefore,
actually be regarded as an areal feature.
3. CONCLUSION
This paper has shed light on an unexpected distribution in nominal number marking in
HLuwian optional plurals with numerals two to four, but strictly singular nouns from
ve onwards and the results have been explained from dierent points of view. Languageinternally, the distribution is maintained through the morphological triggering caused by the
plural endings of the numerals two, three and four. Without such a trigger, the nouns
default to general number, which is covered by the forms of the singular. This distribution
points to a system of grammatical number that has two tiers, of which the top tier consists of a
singular plural dichotomy. The second tier includes a general number besides the other two,
and is invoked whenever a quantier position is lled by a numeral, yet it can be overruled by
morphological cueing.
For lack of data, outside inuence in the development of the system could neither be proved
nor disproved. Nevertheless, the other languages in Anatolia also possess optional plural
marking, and we may at least speak of an areal feature. The study has also highlighted the fact
that linguistic analysis and philology must go hand in hand. A purely linguistic approach
would not have been able to deal with the apparent counterexamples, and thus would not
have uncovered the distribution. At the same time, the linguistic ndings send out a word of
caution to the purely philological approach, as reading badly preserved passages can result in
incorrect restoration and emendation that, rather than having the desired eect, actually
obscure the picture.
BAUER
231
Sprachwissenschaftliches Seminar
Georg-August-Universitat Gottingen
Kate-Hamburger-Weg 3
37073 Gottingen, Germany
Email: [email protected]
References
COMRIE, BERNARD, 1991. Balto-Slavonic, in Jadranka Gvozdanovic (ed.), Indo-European Numerals, Berlin: Mouton
de Gruyter, 717833.
CORBETT, GREVILLE G., 2000. Number, Cambridge: Cambridge University Press.
FORTSON, BENJAMIN W., 2004. Indo-European Language and Culture, Malden, Mass.: Blackwell.
HAWKINS, J. DAVID, 2000. Corpus of Hieroglyphic Luwian Inscriptions, 3 vols, Berlin: de Gruyter.
HAWKINS, J. DAVID, 2003. Scripts and texts, in H. Craig Melchert (ed.), The Luwians, Leiden: Brill, 12869.
HAZENBOS, JOOST, 2007. Hurritisch und Urartaisch, in Michael P. Streck (ed.), Sprachen des Alten Orients,
Darmstadt: Wiss. Buchgesellschaft, 13558.
HOFFNER, HARRY A. & MELCHERT, H. CRAIG, 2008. A Grammar of the Hittite Language, Winona Lake, Ind.:
Eisenbrauns.
HURFORD, JAMES R., 1975. The Linguistic Theory of Numerals, Cambridge: Cambridge University Press.
MELCHERT, H. CRAIG, 2003. Language, in H. Craig Melchert (ed.), The Luwians, Leiden: Brill, 170210.
MELCHERT, H. CRAIG, 2004. A Dictionary of the Lycian Language, Ann Arbor, Mich.: Beech Stave Press.
NEUMANN, GUNTER & TISCHLER, JOHANN, 2007. Glossar des Lykischen, Wiesbaden: Harrassowitz.
PAYNE, ANNICK, 2010. Hieroglyphic Luwian, 2nd rev. edn, Wiesbaden: Harrassowitz.
PLOCHL, REINHOLD, 2003. Einfuhrung ins Hieroglyphen-Luwische, Dresden: Verl. der TU Dresden.
RIJKHOFF, JAN, 2002. The Noun Phrase, Oxford: Oxford University Press.
SCHWYZER, EDUARD, 1939. Griechische Grammatik, vol. 1, Munich: C.H. Beck.
STARKE, FRANK, 1985. Die keilschrift-luwischen Texte in Umschrift, Wiesbaden: Harrassowitz.
THURNEYSEN, RUDOLF, 1946. A Grammar of Old Irish, Dublin: DIAS.
WACKERNAGEL, JACOB, 1930. Altindische Grammatik III: Nominalexion, Zahlwort, Pronomen, Gottingen: Vandenhoeck & Ruprecht.
YAKUBOVICH, ILYA, 2006. The free-standing genitive and hypostasis in Hittite, Journal of Near-Eastern Studies 65(1),
3949.
YAKUBOVICH, ILYA, 2008. The origin of Luwian possessive adjectives, in Karlene Jones-Bley et al. (eds.), Proceedings
of the 19th Annual UCLA Indo-European Conference, November 34, 2007, Washington, DC: Institute for the Study
of Man, 193217.
YAKUBOVICH, ILYA, 2010. Sociolinguistics of the Luvian Language, Leiden: Brill.