Towards a typological classification of Old Hittite relative clauses

Guglielmo Inglese - [email protected]

Università degli Studi di Pavia\Università di Bergamo
2. Indogermanistiches Forschungskolloquium, Würzburg 31/03/16

1 Old Hittite RCs: state of the art

Based on Held (1957), Hale (1987), Garrett (1994), Hoffner & Melchert (2008: 423 ff.)
Features of Hittite RCs
 Occurrence of the inflected relative pronoun kui- who, which .
 Position: either preposed or postposed. They never interrupt the main clause.
 Linked to their main clause either asyndetically or through sentence connectives.
 Contain the relativized noun, which can be anaphorically referred to in the main clause.
 Indentification of the correlative diptych since Justus (1976)

Types of RCs
A. Preposed RCs:
(1) Determinate
G)ŠSUKURḪI.A=ma Ù G)ŠPAḪI.A kue ḫarkanzi nu=šmaš=at=kan
spear:PL=PTC CONJ scepter:PL REL.ACC.PL.N hold:3PL CONN=3SG.DAT=3SG.ACC.N=PTC
1-aš LÚMEŠEDI arḫa dāi
one:NOM guard away take:3SG
But which spears and scepters they hold, one guard takes these (lit. it) away from them. (KBo
4.9 Rev. v 3-5)1

(2) Indeterminate
nu=šši=ššan kuit šaḫḫan LUGAL-uš dāi nu apāt ēššai
CONN=3SG.DAT=PTC REL.ACC.N šaḫḫan:ACC.N king:NOM give:3SG CONN DEM.ACC.N make:3SG
Whichever šaḫḫan the king imposes on him, he does that. KBo . iv -16)

B. Postposed RCs

(3) Non-restrictive
nu=za dKumarbiš GALGA-tar ZI-ni katta daškizzi UDKAM-an
CONN=REFL Kumarbi:NOM windsom:ACC.N mind:DAT with take:ITER.3SG day:ACC
kuiš LÚ ḪUL-an šallanuškizzi
REL.NOM man evil:ACC big:CAUS.ITER.3SG
Kumarbi takes wisdom into his mind, he who raises the day as evil. (KUB 33.98 i 4-5)

(4) Indefinite
nu 8 DUMUMEŠ-uš uwadanzi MUNUS-ni=ššan kuiēš nāui pānzi
CONN 8 son:ACC.PL bring:3PL woman:DAT=PTC REL.NOM.PL not-yet go:3PL
They bring eight boys who have not yet gone to a woman. KUB . ii -10)

 Problems: description not based on large scale corpus analysis, obsolete theoretical
background, the determinate vs. indeterminate opposition has proven untenable
(Goedegebuure 2009, Huggard 2015, Melchert ftc.)

1 Examples (1) to (4) are taken from Garrett (1994).

1
 2 The typology of RCs

How to identify RCs cross-linguistically (Hendery 2012: 15)

i. they contain some form of a verb (vs. adjectives)
ii. they are linked in some way to another clause (vs. nominalized verbs)
iii. they delimit the reference of a noun phrase by specifying the role of the referent of that noun
phrase in the situation described by the relative clause (vs. adverbial clauses)

 i. and ii. refer to coding properties, iii. provides a functional definition

Classification of RCs (de Vries 2002, 2005, Cristofaro 2003, Andrews 2007, Hendery 2012)

A. Semantics:
1) Restrictive
The man [who is sitting in that office] is a psychologist
2) Non-restrictive, appositive
They went to a number of Bach concerts, [for which they had booked tickets several
months in advance]
3) Maximalizing (Grosu & Landman 1998)
John looked at the mice [that there were in the cage]

Maximalizing RCs
 Denote the maximal amount of entities that satisfies the properties given by the RC
 Do not allow for stacking, i.e. embedding of further relatives
 Incompatible with determiners such as few, some, often paired with there-constructions or
maximalizing quantifiers such as all.

B. Accessibility Hierarchy (Keenan & Comrie 1977):

Subject > Direct Object > Indirect Object > Oblique > Genitive > Object of comparison
C. Structural variation:
a. Position of the head with respect to the MC: head-internal vs. head-external, absence of
the head > free RC
b. Position of the RC with respect to the MC: prenominal vs. postnominal, left-adjoined vs.
right-adjoined
c. Degree of syntactic integration: embedded vs. adjoined

Figure 1: RELATIVE CLAUSE HIERARCHY

2
Correlative RCs:
a subtype of adjoined relative clauses: specifically, those that are left-adjoined and have a nominal
phrase in the adjoined clauses that is co-referent with a nominal phrase in the main clause. The
former NP is marked with (or sometimes substituted by) either a relative marker or an interrogative,
and the correlated NP in the main clause is marked by an anaphoric marker of sorts, commonly a
demonstrative. (endery :

The correlative diptych in ancient IE languages (Justus 1976, Lehmann 1986)

(5) Latin
quei ager ex private in publicum commutatus est de
REL.NOM land:NOM from private:ABL in public:ACC convert:PTCP.PST.NOM be:3SG about
eo agro siremps lex esto
DEM.ABL land:ABL in-the-same-way law:NOM be:3SG.FUT.IMPV
Any land that has been converted from private into public, to such land the law shall apply in the
same way. C)L )2, 585)
(6) Old Indic
agne yaṃ yajñam adhvaraṃ viśvataḥ paribh”r asi sa
Agni:VOC REL.ACC sacrifice:ACC offering:ACC all-around surrounding be:2SG DEM.NOM
id deveṣu gacchati
truly god:DAT.PL go:3SG
Agni, the sacrifice and the offering that you completely encompass, that sacrifice truly goes to
the gods. RV , ,
(7) Celtiberian
iomui liś taś TiTaś sisonti ś omui […] ś aum TeCameTinaś
REL.DAT land:ACC.PL assigned:ACC.PL sow:3PL DEM.DAT DEM.GEN.PL tithe:ACC.PL
TaTus
give:IMP.3SG
To whom they give assigned lands, to this one (applies): he shall give the tithes of those.
(8) Ancient Greek
ὅς ε ε ῖς ἐ εί α ά α ἔ αὐτοῦ
REL.NOM PTC god:DAT.PL obey:3SG.SUBJ.MID indeed PTC hear:3PL.AOR DEM.GEN
Whoever obeys the gods, they indeed listen to him. (om. Il. 1.218)
(9) Armenian
or oḳ i noc̣ anē vał aṙ is hasṣ ē, zna ṭagawor
REL.NOM ever from 3PL.GEN quickly to 1SG reach:3sG.SUBJ DEM.ACC king:ACC
araric̣
make:1SG.SUBJ.FUT
Whichever of them reaches me the sooner, him shall ) make king.

 Omission of the resumptive item is licensed by its position on the Accessibility Hierarchy
(Keenan & Comrie 1977) and by language-specific syntactic rules, i.e. pro-drop (cf. Pompei 2011:
494).

Free relatives: RCs lacking a lexical head, a syntactic configuration compatible with all structural types
(de Vries 2005: 11, Latin examples from Pompei 2011)
(10) Free correlative
ita fit, ut, quod bonum sit, id etiam honestum
so happen:3SG COMP REL.NOM.N good:NOM.N be:3SG.SUBJ DEM.NOM.N also honest:ACC.N
So it turns out that what is good is also honest. Cic. Fin. 3,28)

3
 (11) Free embedded
quem di diligent adulescens moritur
REL.ACC god:NOM.PL love:3PL.FUT young:NOM die:3SG.PASS
(e whom the gods love dies young. Plaut. Bacch. 816-817)

Table 1.: MAPPING BETWEEN SYNTACTIC AND SEMANTIC TYPES OF RCS

Appositive Restrictive Maximalizing
emb. postnominal + + +
emb. prenominal - + +
adj. correlative - - +
free - - +

 Restrictive semantics constitute a good diagnostic for the embedded status of RCs

3 Corpus analysis: RCs in Old Hittite

The corpus (only OH/OS): Zalpa’s text (Otten 1976 [StBot 17]), Anitta’s text (Neu 1974 [StBot 18]),
Rituals (Neu 1980, 1983 [StBot 25/26]), Palace Chronicle (Dardano 1997), Laws (Hoffner 1997),
Instruction (Miller 2013).

Table 2.: TOKEN FREQUENCY OF RCS IN OH

Text PREPOSED PREPOSED PLUS CONNECTIVE POSTPOSED
StBot 18 3 3 -
StBot 17 - 2 -
Chronicle 1 2 2
StBot 25/26 3 3 -
Laws 9 8 -
Instruction - - -
Total 16 18 2

1 Preposed plus connective:

(12) Correlative
dUTU-az ut-ne-e [ku-it k]u-it=pát a-ra-iš n=u-uš
Sun-God:ABL land:NOM.N REL.NOM.N REL.NOM.N=EMPH rise:3SG.PST CONN=3PL.ACC
ḫu-u-ma-an-du-uš=p[át ḫ]u-u[l-la-nu-un]
all:ACC.PL=EMPH destroy:1SG.PST
Which land rose against the Sun God, ) have destroyed them all. (Anitta Obv.11-12 [StBot 18])
(13) Free correlative
ku-iš pa-ap-re-ez-zi nu a-pa-a-aš=pát 3 G[ÍN KÙ.BA]BBAR pa-a-i
REL.NOM impure:3SG CONN DEM.NOM=EMPH 3 shekel silver give:3SG
Whoever is impure, that one gives three shekels of silver. KBo . i [Laws § 25])
(14) Left-adjoined
[k]u-iš ša-ga-i-[i]š ki-i-ša-ri ta Ø LUGAL-i MUNUS.LUGAL=ya ta-ru-e-ni
REL.NOM sign:NOM become:3SG.MID CONN king:DAT queen=CONJ speak:1PL
Whatever sign occur[s], we report it to the king and the queen. KBo . + KBo . iv

4
(15) Free left-adjoined
ú-e-š=a ku-wa-pí-it a-ru-me-en nu Ø MUNUS-z[a …] DUMU x [ḫa]-a-ši
1PL.NOM=PTC wherever arrive:1PL.PST CONN woman:NOM son give-birth:3SG
But wherever we arrived, a woman gives [bir]th to just one child. Zalpa. A Obv. 11 [StBot 17])
 Given their syntactic structure, RCs from (12) to (15) must be semantically maximalizing!
o Indefinite relative pronoun kuit kuit whichever , kuwapit wherever
o Universal quantifier ḫ”manduš all
o Focus particle = pát
o Stacking of more preposed RCs is unattested

Table 3.: UNAMBIGUOUS LEFT-ADJOINED RCS IN OH

Sentence type Correlative Free correlative Adjoined Free adjoined Total
Frequency 2 12 1 3 18

2 Postposed
(16) Restrictive
na-at-ta a-pu-u-un GESTIN-an pí-i-e-er LUGAL-uš ku-in a-uš-ta
NEG DEM.ACC wine:ACC give:3PL.PST king:NOM REL.ACC see:2SG.PST
They did not give me that wine that you, the King, saw. (KBo 3.34 ii 5-6 [Chronicle A])

 The fact that kuin is not adjacent to GESTIN-an does not prevent the clause to be embedded,
but it depends on the basic OV word order, cf. (16):

(17) Hos deterrere […] ne frumentum conferant quod debeant

DEM.ACC.PL deter.INF COMP corn.ACC.N supply.SUBJ.3PL REL.ACC.N ought.SUBJ.3PL
They are deterring […] from contributing the corn which they ought to supply. Caes. Gal. 1.17)

(18) Appositive:
A-HI LUGAL A-NA P[A-NI A-BI LUGAL] ku-i-e-eš e-eš-kan-ta
brother king to front father king REL.NOM.PL sit:ITER.3PL.MID
(Behold) the brothers of the king, who sit in front of the father of the king KBo . iii
[Chronicle A])

3 How should the remaining 16 preposed RCs be classified?

(19) Center-embedded postnominal restrictive RC

ma-a-an URUA-ri-in-na 11 ITU-aš ti-ez-zi nu a-pé-e-[el É=SÚ] ku-e-la
when Arinna:ALL 11 month:NOM arrive:3SG CONN DEM.GEN house=1SG.POSS REL.GEN
G)Še-ya-an a-aš-ki-i=š-ši ša-ku-wa-a-an a-[pé-ni-iš-ša-an]
eyan:NOM.N gate:DAT=3SG.POSS see:PTCP.NOM.N likewise
)n Arinna, when the eleven month arrives, the house of him at whose gate an eyan is standing is
likewise exempt . KBo . ii 61-62 [Laws § 50])

Probert s (2006) identifies two main syntactic types:

EMBEDDED RCs (no connective and no resumption) vs. ADJOINED RCS (connective and resumption)

5
Two exceptions:
 No resumption in MCs introduced by ta (tendency of ta to occur in contexts of DO omission,
cf. Rieken 1999)
 No connective and possessive resumption (only 1 occurrence)

(20) Internally headed embedded (x 6)

[LÚK]AŠ.E [(tar-ru-uḫ-zi ku-i)]š 2 NINDAwa-ga-ta-aš 1 MA[.NA KUBABBAR LUGAL-wa-aš
runner win:3SG REL.NOM 2 wagada:ACC.N 1 mine silver king:GEN
[(ki-iš-šar-ra-az=š)]e-et da-a-[i]
hand:ABL=3SG.POSS.ABL take:3SG
Which [runner wins], he take[s] two wagada breads and one mi[ne of silver] [from the hand] of
the King. (KBo 19.9 + 20 + KBo 20.5 + KBo 25.12 + ABoT 5 ii 10-12)
(21) Free embedded (x9)
ar-ḫa-an-n=a ku-iš pár-ši-ya 1 UD[U 10 NINDAḪI.A 1 DUG KA.GAG
boundary:ACC.N=CONJ REL.NOM break:3SG.MID 1 sheep 10 bread 1 jug KAGAG-beer
pa-a-i]
give:3SG
And who ever violates the boundary [shall give] one sheep, [10 loaves, and one jug of KAGAG-
beer]. KUB . -11[Laws § 168])

Probert s account is problematic in many respects:

a) Lack of connective does not entail per se lack of sentence boundary, asyndeton is a lively clause
linkage strategy in OH (cf. Hoffner & Melchert 2008)
b) Role of Garret s (1996) rule: unaccusative intransitive verbs require a clitic subject, unergative
verbs do not. One only finds adjoined RCs with intransitive verbs, as clitic resumption in the MC
is compulsory in this case:

(22) [ku-i] š=za LÚḫi-ip-pa-ri ḫa-a-ap-pár i-ez-zi n=a-aš=kán

REL.NOM=REFL ḫipparaš-man:DAT business:ACC.N make:3SG CONN=3SG=PTC
ḫa-a-ap-pa-ra-az [še-me-e]n-zi
business:ABL forfeit:3SG
[Whoever] transacts business with a ḫipparaš-man, he [shall forfeit] his purchase price.
(KBo 6.2 ii 51-52 [Laws § 48])

 If RCs such as (20) – (21) were truly embedded, thus serving as argument for the verb of
the MC, one expects them to occur also with unaccusative verbs without clitic resumption,
as the subject slot would be filled by the RC itself. The fact that this pattern never occurs
suggests that embedding is at best a syntactic configuration available only if the MC has a
transitive or unergative verb.

c) Connectives and resumption do not provide a reliable diagnostic, patterns opposed to Probert s
rules are attested

(23) Connective and no resumption

ku-it ú-e-mi-ez-zì a-pa-aš=a [(pá)r]-na-a=š-ša pí-it-ta-iz-zi
REL.ACC.N bring:3SG DEM.NOM=PTC house:ALL=3SG.POSS.ALL bring:3SG
What he brings back, this one brings it to his [hou]se. (KBo 3.34 i 12 [Chronicle A])

(24) No connective and resumption

KAKSAL-za ku-it a-aš-šu ú-taḫ-ḫ[u-un (a-pé-e-da-an-da ḫa-liš-ši-ya-nu-un)]
way:ABL REL.ACC.N good:ACC.N bring:1SG.PST DEM.INST furnish:1SG.PST
6
 Which good ) have brought back from the military campaign, [with this I furnished (the temple of
the Storm God ]. (Anitta Rev. 58-59 [StBot 18])

 Restrictive semantics is the only viable diagnostic for the embedded status of these RCs, but this
cannot apply to free relatives (9/15). Of the other 6 internally headed RCs, only (25) seems to
support a contextual restrictive reading:

(25) [ták-ku LÚ-aš MUNUS-na-an ḫar-zi ta LÚ-aš a-k]i DAM=S[Ú

if man:NOM woman:ACC have:3SG CONN man:NOM die:3SG wife=3SG.POSS
ŠEŠ=ŠU da]-a-i [ta=an at-ta-aš=ši-iš da-a-i ma-a-an
brother=3SG.POSS take:3SG CONN=3SG.ACC father:NOM=3SG.POSS.NOM take:3SG when
ta-a-a]n-n=a at-ta-aš=ši-š=[a a-ki MUNUS]-na-an-n=a [ku-in
after=ADV father:NOM=3SG.POSS.NOM=PTC die:3SG woman:ACC=PTC REL.ACC
ḫar-ta DUMU ŠEŠ-ŠU da]-a-i
have:3SG.PST son brother=3SG.POSS take:3SG
[)f a man has a wife and this man dies], [his brother shall ta]kes the widow as wife. (If the brother
dies), [his father shall take her. But when afterwards] his father [dies as well], [the son of his brother
shall] take the [woman whom he i.e. the father had.] KUB . + . iv -4 [Laws § 193])

 Problem: the unusual occurrence of =(m)a points to the fact that MUNUS-an belongs to the main
clause. In this case, (25) is a center-embedded postnominal RC (cf. 19).

4 Conclusions

 A careful scrutiny of corpus data reveals that OH attests to a variety of different RC types.
Structurally one can identify PREPOSED CORRELATIVE and LEFT-ADJOINED RCs, both configuration
displaying FREE variants, POSTNOMINAL CENTER-EMBEDDING RCs, and POSTPOSED RCs. This mirrors
the variability attested in other Anatolian languages, as well as in post-OH times (Melchert ftc.).
 Semantic and syntactic differences: PREPOSED RCs are syntactically less integrated and
semantically maximalizing, POSTPOSED and CENTER-EMBEDDING are structurally more integrated
and semantically either restrictive or appositive.
 Among PREPOSED RCs, there is no evidence for a syntactic adjoined vs. embedded distinction
(contra Probert 2006). This is consistent with the fact that in post-OH almost all preposed RCs
can be only interpreted as adjoined, since connectives become increasingly obligatory after all
kind of dependent clause.

7
 References
Andrews, Avery. 2007. Relative clauses. In Language typology and syntactic description. Vol. 2, Complex
constructions (2nd ed.), T. Shopen (ed.), 206–236. Cambridge: CUP.
Cristofaro, Sonia. 2003. Subordination [Oxford Studies in Typology and Linguistic Theory]. Oxford: OUP.
Dardano, Paola. 1997. L’aneddoto e il racconto in età antico-hittita: La cosidetta cronaca di palazzo.
[Biblioteca di ricerche linguistiche e filologiche 43]. Roma: Il Calamo.
de Vries, Mark. 2002. The syntax of relativization. PhD dissertation, University of Amsterdam
de Vries, Mark. 2005. The fall and rise of universals on relativization. Journal of Universal Language 6:
125-157.
Garret, Andrew J. 1994. Relative Clause Syntax in Lycian and Hittite. Die Sprache 36: 26-69.
Garret, Andrew J. . Wackernagel s Law and Unaccusativity in (ittite. )n Approaching Second: Second
Position Clitics and Related Phenomena, A. L. Halpern & A. M. Zwicky (eds.), 83-133. Stanford: CSLI
Publications
Goedgebuure, Petra. 2009. Focus Structure and Q-word Questions in Hittite. Linguistics 47/4: 954-969.
Grosu, Alexander & Landman, Fred. 1998. Strange relatives of the third kind. Natural Language Semantics
6: 125-170.
Hale, Mark. 1987. Notes on Wackernagel s Law in the Language of the Rigveda. )n Studies in Memory of
Warren Cowgill, C. Watkins (ed.), 38-50. Berlin: de Gruyter.
Held, Warren H., Jr. 1957. The Hittite Relative Sentence. Language Dissertations 55. Baltimore: Linguistic
Society of America.
Hendery, Rachel. 2012. Relative Clauses in Time and Space: A Case Study in the Methods of Diachronic
Typology. Amsterdam/Philadelphia: John Benjamins.
Hoffner, Harry A. & Melchert, Craig H. 2008. A Grammar of the Hittite Language. Part I: reference grammar.
Winona Lake, Indiana: Eisenbrauns.
Hoffner, Harry A. 1997. The Laws of the Hittites. A Critical Edition. Leiden: Brill.
Huggard, Mattyas. 2015. Wh-words in Hittite: A Study in Syntax-Semantics and Syntax-Phonology
Interfaces. PhD dissertation, University of California, Los Angeles.
Justus, Carol. 1976. Relativization and Topicalization in Hittite. In Subject and Topic, C. N. Li (ed.), 213-
246. New York: Academic Press.
Keenan, Edward & Comrie, Bernard. 1977. Noun phrase accessibility and universal grammar. Linguistic
Inquiry 8:63-99.
Lehmann, Christian. 1986. On the typology of relative clauses. Linguistics 24: 663-680. Linguistics 49 (2):
365-390.
Melchert, Craig. Forthcoming. Relative Clauses in Anatolian. To appear in a Festschrift for Rosemarie Lühr.
Miller, Jared L. 2013. Royal Hittite Instructions and Related Administrative texts [Writings from the Ancient
World 31]. Society of Biblical Literature: Atlanta.
Neu, Erich. 1974. Der Anitta-Text [Studien zu den Bogazköy-Texten 18]. Wiesbaden: Harrassowitz.
Neu, Erich. 1980. Althethitische Ritualtexte im Umschrift [Studien zu den Bogazköy-Texten 25].
Wiesbaden: Harrassowitz.
Neu, Erich. 1983. Glossar zu den althethitischen Ritualtexten [Studien zu den Bogazköy-Texten 26].
Wiesbaden: Harrassowitz.
Otten, Heinrich. 1973. Eine althethitische Erzählung um die Stadt Zalpa [Studien zu den Bogazköy-Texten
17]. Wiesbaden: Harrassowitz.
Pompei, Anna. 2010. Les propoposition relatives entre restrictivité et maximalisation. Studi Italiani di
Linguistica Teorica e Applicata 39: 439-456.
Probert, Philomen. 2006. Clause Boundaries in Old Hittite Relative Sentences. Transaction of the
Philological Society 104: 17-83.
Rieken, Elisabeth. 1999. Zur Verwendung der Konjunction ta in den hethitischen Texten. Münchener
Studien zur Sprachwissenschaft 59: 63-88.