Content uploaded by Zhang Shuya
Author content
All content in this area was uploaded by Zhang Shuya on Dec 10, 2022
Content may be subject to copyright.
The history of the polyfunctional 𗗙jij
in Tangut
How did the accusative/genitive syncretism come
about?
Shuya Zhang
Tokyo University of Foreign Studies
This paper focuses on the history of a polyfunctional case marker 𗗙jijin
Tangut, an extinct Rgyalrongic language (Sino-Tibetan). This versatile case
morpheme is a typological rarity of maximum syncretism among several
abstract case functions, including dierential object marking, the genitive,
and the oblique (which overlaps with the dative). For one thing, accusatives
originating from datives or spatial sources are rarely found with additional
genitive functions; for another, reported instances of accusative/genitive
syncretism seldom include other functions. The principal hypothesis of this
paper is that the Tangut 𗗙jijmay be the result of multiple
grammaticalization processes stemming from a proto-locative source. These
processes can be subsumed under two pathways, one leading from an
allative to an accusative, with an intermediate oblique stage, and the other
from a locative to a genitive. Although both of these development pathways
are frequently attested, the Tangut 𗗙jijremains a typological rarity due to
their superposition.
Keywords: case syncretism, dierential object marking, locative, Tangut,
Rgyalrongic languages
. Introduction
This paper focuses on the diachronic development of the polyfunctional case
marker 𗗙jijin Tangut (Glottolog: tang, in Tangut 𗼇𗟲 mjiŋwuu, in
https://doi.org/10.1075/sl.21085.zha |Published online: 8 December 2022
Studies in Language ISSN 0378-4177 |E‑ISSN 1569-9978 © John Benjamins Publishing Company
Chinese 西夏语 xīxià y), an extinct language with its own scripts, once spoken
in the Tangut Empire (–) in present day northwestern China.1
Though once classied into the Qiangic branch (Tibeto-Burman, Sino-
Tibetan) (Wang ; Sun ; etc.), recent studies on Sino-Tibetan phylogeny
and historical linguistics (Lai et al. ; Sagart et al. ) show that Tangut is a
close relative of the modern Rgyalrongic languages, a group of unwritten Tibeto-
Burman languages, spoken in southwestern China (Aba Tibetan and Qiang
Autonomous Prefecture and Ganzi Tibetan Autonomous Prefecture, Sichuan
Province cf. Figure ). These languages are known for their complex phonology
and rich polysynthetic morphology, which several authors consider to be conser-
vative features (Jacques ; Hill ; Zhang et al. ; etc.). The subgrouping
of the Rgyalrongic group is illustrated in Figure .
Figure . Geographic distribution of Tangut and modern Rgyalrongic languages
. Since this paper has been written with typologists/diachronicians, Sino-Tibetanists, and
Tangutologists in mind, Tangut characters are provided alongside their number as listed in the
Tangut-Chinese Dictionar y (Li ). Transcriptions follow the reconstruction of Gong ().
However, knowledge on Tangut characters is not necessary for understanding the content of
the paper. The present analysis is mainly based on two materials translated from Chinese texts,
𗴮𘊳 (Leilin, ‘The Forest of Categories’, eds. Shi et al. ) and 𗆧𗰖𗕿𘓓𘐆𘚔𘐳 (Xinji Cixi-
aozhuan, ‘Newly Collected Biographies of Aection and Filial Piety’, ed. Jacques ).
[] Shuya Zhang
Figure . Stammbaum of Rgyalrongic languages (Sun ; Lai et al. )
The most important functions of the postposition 𗗙jijinclude dierential
object marking (i.e. accusative), and the genitive. It reects the Proto-Rgyalrongic
yod postposition =*j(V), the functions of which vary across dierent sub-groups.
No modern Rgyalrongic varieties exhibit the same accusative/genitive syncretism
as the Tangut 𗗙jij. Before discussing the source of 𗗙jijand the processes
leading to its accusative/genitive syncretism, I will rst briey introduce the phe-
nomenon of dierential object marking and the sources of its markers.
. Dierential object marking
‘Dierential object marking’ (henceforth DOM) refers to the non-uniform mark-
ing of objects within the same language, depending on their semantico-pragmatic
properties (Aissen ; Bossong ; Dalrymple & Nikolaeva ; Næss ;
etc.), as illustrated by the pair of examples from Hindi given in ().
(-)
() a. Ilaa-ne
Ila-
ek
one
bacee-ko
child-
uaayaa
li-
‘Ila lied a child.’
b. Ilaa-ne
Ila-
ek
one
haar
necklace.
uaayaa
li-
(Mohanan :)‘Ila lied a necklace.’
DOM is oen used as an umbrella term,2most frequently including systems
exhibiting optional object marking, in which an object marker can optionally
be present or absent without aecting its grammatical function, as is the case in
Hindi, and those showing alternating marking, where two morphological case
. Distinction is also made between dierential case marking of object, which is dependent
marking by morphological case or adpositions, and dierential object indexation, occurring on
the predicate (Iemmolo ; Witzlack-Makarevich & Seržant ). In the present paper, the
term DOM is used only to designate case marking on dependent.
The history of the polyfunctional 𗗙jijin Tangut []
markers alternatively mark the object (for a detailed discussion on this issue see
Chappell & Verstraete ).
Recurrent parameters of DOM involve animacy (humanness), deniteness,
specicity, and information structure of the object (Aissen ; Dalrymple &
Nikolaeva ; Sinnemäki ; etc.). In many languages DOM operates under
a synergy of multiple parameters. In Hindi, accusative marking is related to ani-
macy and deniteness, while animate and denite inanimate objects receive overt
marking (Mohanan :). In Dolakha Newar (Tibeto-Burman), DOM is co-
regulated by the animacy and topicality of the object. Its distribution can be
described by two probabilistic trends: (i) topic animate objects oen receive case
marking; (ii) inanimate objects can occasionally be case-marked if they refer to a
salient entity (Genetti , , cited in Dalrymple & Nikolaeva : –).
. Sources and grammaticalization of DOM markers
Bossong (, ) points to datives and other spatial related markers such as
locatives, allatives as common sources for DOM markers, as found in Romance
and Semitic languages. This observation is also relevant to the Tibeto-Burman
languages in which ‘locative-allative-dative-accusative’ syncretism prevails
(LaPolla , ).
The spatial and dative sources of DOM markers seem to be further compat-
ible with a widely attested grammaticalization chain which accounts for the evo-
lution of the accusative, with the allative as a starting point, and the dative as an
intermediate stage (Heine :–; Heine & Kuteva :–; Lehmann
:; among others), as illustrated in ().
() Common grammaticalization chain of case functions
(Heine :)allative > dative > accusative (O)
. Genitive syncretism
It is typologically rare for the dative/accusative morpheme to be isomorphic with
the genitive. Yet in the Gunnän Gurage (Ethio-Semitic) languages, a highly poly-
functional case marker jä- is used as the (optional) accusative, dative, and genitive
(Meyer :), as shown by the examples from Ezha given in ().
(-)
() a. abza
Abza
j-mss-we
-man-
k’t’t’r--n-m
kill.-3..-..-
‘Abza killed the man’
[] Shuya Zhang
b. abza
Abza
j-dsta
-Desta
xwett
two
brr
Birr
ab--n-m
give.-3..-3..-
‘Abza gave two Birr to Desta.’
c. j-dsta
-Desta
gred
girl.
(Endalew :, , )‘Desta’s daughters’
In Tshangla (Bodish), the ending -ga falls into two functional domains, the
locative-dative-(limited) accusative, and the genitive (Andvik : –). Both
Gurage and Tshangla dier from languages like modern Mongolic (Janhunen
), in which the accusative/genitive morpheme does not involve other abstract
case functions (Baerman’s Type , i.e case syncretism of a core case and a
non-core case).
The question of whether dative-accusative-genitive syncretism resulted from
a single source or represents an accidental isomorphy remains controversial in
both Gurage and Tshangla. For Gurage, Hetzron (:) assumed an accidental
syncretism due to the phonetic fusion of two distinct case morphemes – *lä- ‘to,
for’ for the accusative/dative, and *zä- ‘for’ for the genitive. However, later schol-
ars (Endalew ; Meyer ) prefer a common source for all of these func-
tions, though a clear grammaticalization pathway leading to this syncretism has
yet to be proposed. Concerning Tshangla, Andvik (:; :) expresses
uncertainty about a single source on the basis of dialectal evidence suggesting a
proto-locative form *-gu probably distinct from the genitive *-ga.
This paper focuses on the case of Tangut (Rgyalrongic), in which a poly-
functional case marker 𗗙jijexhibits an accusative/genitive syncretism similar
to that observed in Gurage and Tshangla. I will argue that such syncretism in
Tangut has resulted from the multi-path grammaticalization of a single locative
source. Tangut is also unique in that the accusative/genitive morpheme 𗗙jij
(i) exhibits functional overlap with the dative 𘋩do, and (ii) its locative source
is obscure and can only be reconstructed through comparative data from modern
Rgyalrongic languages. The aim of the paper is to illustrate that a typological
rarity of extreme syncretism among abstract case functions can arise due to the
superimposing of divergent grammaticalization pathways stemming from a spa-
tial case, even if each individual pathway is typologically frequent.
The rest of the paper is organized as follows. In Section , I describe relevant
morpho-syntactic features of Tangut. In Section , I analyze the clausal functions
of the Tangut 𗗙jij, paying special attention to comparison with the dative
𘋩do. Section provides comparative data from West Rgyalrongic and core
Rgyalrong, laying a framework for Section , which attempts to explain
accusative/genitive syncretism as resulting from multiple grammaticalizations of
The history of the polyfunctional 𗗙jijin Tangut []
the Proto-Rgyalrongic general locative *=j(V). Section oers concluding
remarks.
. Morpho-syntactic prole of Tangut
Tangut is a verb nal language, exhibiting templatic verbal morphology. The ver-
bal template of Tangut is much simpler than those of modern Rgyalrongic lan-
guages (Jacques ), and can be divided into three domains, as summarized in
Table : (i) the verb stem (including noun-verb and verb-verb compounds), (ii)
verbal prexes, including directional prexes, encoding TAM values as well as
certain orientation meanings (cf. Arakawa ; Beaudouin Accepted), negation,
and other modal prexes, and (iii) argument indexation suxes (cf. Jacques ;
Kepping ).
. Transitivity
Since transitivity is crucial to understanding the clausal functions of 𗗙jij,
which will be discussed in this paper later on Section , this section briey intro-
duces transitivity in Tangut.
As in other Rgyalrongic languages, transitivity in Tangut can be determined
through indexation patterns, with intransitive and transitive verbs in Tangut fol-
lowing dierent indexation patterns (Beaudouin ; Gong ; Jacques
:–, etc.). Intransitive subjects (S) are pronominally represented by
three suxes: 𗧓-ŋa, 𘉞-nja, and 𗐱-nji/. The third per-
son is unmarked.3
Transitive verbs have a maximum of two indexation slots, one for the subject
(A) and one for the object (O). Alignment distinguishes three scenarios: (i) in
a non-local scenario (), the verb is always unmarked; (ii) in a local scenario
(, ), the verb agrees with the patient; (iii) in a mixed domain (/, /
), the SAP argument is always pronominally represented on the verb, indepen-
dent of its syntactic role. Table illustrates the Tangut transitive paradigm with
the causative auxiliary verb 𗟻phji‘to send, cause to do’.
A small number of transitive verbs have an additional stem [B] form, occur-
ring in / forms (Gong ; Jacques , ; Gong ). As shown
in Table , the verb 𗟻phjihas a distinct stem [B] 𗠔phjo, showing vowel
alternation with the basic stem [A]. Jacques () hypothesizes that vowel alter-
. Tangut also has an infrequent dual sux 𘙌-kj(Arakawa ).
[] Shuya Zhang
nations in stem [B] originate from a third person patient sux *-w(well attested
in modern Situ, see Prins :), which merged with the basic stem [A].
() Reconstructed stems of ‘to send, cause to do’
(Jacques :)
Stem [A]
Stem [B]
Tangut
𗟻0749phji1
𗠔4568phjo2
Pre-Tangut
*phja
*phja-w
Table . Simplied verbal template of Tangut (based on Jacques , :)
Directional
prexes Negation Modal
Noun
stem
Verb
stem Auxiliary Person
:
𗈪
a-
𗱢
nja
𗅋
mji-
𘖑
mj
𗉘
thj-
…
𗧓
-ŋa
… …
𗭪
-Sji …
: … …
𗭊
jij-
…
𘀆
njij-
… …
Table . Person indexation of 𗟻 phji(stem [B] 𗠔 phjo) ‘to send, cause to do’
(Gong :)
𗟻 𘉞
phji-nja
𗟻 𗐱
phji-nji
𗠔 𗧓
phjo-ŋa
𗟻 𗐱
phji-nji
𗟻 𗧓
phji-ŋa
𗟻 𗐱
phji-nji
𗠔 𘉞
phjo-nja
𗟻 𗐱
phji-nji
𗟻 𘉞
phji-nja
𗟻 𗐱
phji-nji
𗟻 phji
Tangut also diers from modern Rgyalrongic languages in the emergence of
non-nite forms in chained clauses (Jacques ; Beaudouin ). As in (), ver-
bal inections are only present in the nal clause, and are neutralized in the pre-
ceding clauses.
The history of the polyfunctional 𗗙jijin Tangut []
() 𘋨 𘝦 𘃡 𗴢 𘇒 𘉁 𗁦 𘓺 𘜹 𗞞 𗨻
[dzjw1=dj.wji1
emperor=
bjuu1-wier1]
to.respect-to.cherish
[mjii1
rank
bjj1]
to.raise
[ŋwr1.dzow1
empress
dja2-we2]
1:-to.become
‘…The Emperor respected and loved her, and elevated her rank, she became
(Cixiaozhuan, Jacques :)the empress.’
The neutralization of verbal inections leads to ambiguities in transitivity mark-
ing in Tangut texts. Where there is an absence of morphological evidence, transi-
tivity can be indicated on the basis of syntactic criteria, such ergative marking by
𘝦𘃡dj.wji,4as in ().
. Status of 𗗙jijin Tangut case system
In contrast to its simplied verbal inections, Tangut is rich in dependent case
marking (cf. Table ), which distinguishes itself from languages with their domi-
nant head marking.
Table . Tangut case system (adapted from Beaudouin )
Abstract cases Spatial cases Non-spatial cases
𘝦 𘃡 dj .wjiErgative 𘋩 doAllative 𗳒 ŋwuInstrumental
𗗙 jijGenitive 𘕿 aLocative 𗑠 rjirComitative
𗗙 jijAccusative, oblique 𗅁 uIn-essive 𗖵 bjuInstrumental
𘋩 doDative 𘂤 khaMid-essive 𗸒 suComparative
𗀔 thjaaSuper-essive
𗯴 khjuSub-essive
𘇂 guMed-esssive
… …
The 𗗙jijis versatile and is available for two functional domains (Kepping
:–; Kotaka ; etc.). First, it functions as a genitive marker, express-
. The ergative 𘝦𘃡dj.wjiis optional aer transitive subjects (Kepping :),
and its distribution still awaits further investigation. Nishida ( []) analyzes 𘝦𘃡
as an ‘emphasizing’ agentive (強調主格), suggesting that its distribution could be related to
semantico-pragmatic factors, as is widely observed in Tibeto-Burman (DeLancey ).
[] Shuya Zhang
ing adnominal possession linking two noun phrases, or a noun phrase and a
pronominal phrase (e.g. ()).
() Genitive 𗗙jij
𗌵𘏸 𗗙 𗪯
[kjow1.i1]=jij1
:Jiang.Shi=
[gji2]
wife
(Cixiaozhuan, Jacques :)‘Jiang Shi’s wife’
Second, it occurs following a wide range of clausal functions, which can be sub-
sumed under two labels, the accusative, marking indexable objects, or the oblique,
occurring aer various non-indexable functions.5
In terms of clausal functions, the distribution of 𗗙jijpartially overlaps
with that of the dative 𘋩do, in particular aer the R of ditransitive verbs.6
Beaudouin () argues that the dative use of 𘋩dodeveloped from its alla-
tive source, i.e. marking the endpoint of a motion.7
The clausal functions of 𗗙jijwill be detailed in comparison with 𘋩do
in Section .
. Clausal functions of 𗗙jij
This section examines the clausal functions of the case morpheme 𗗙jijacross
three syntactic constructions, (i) object (O) marking with monotransitive verbs
. 𗗙jijwas previously described as an ‘anti-ergative’ marker (Jacques , etc.), a term
introduced in the eld of Sino-Tibetan linguistics by LaPolla (, ) as a kind of
semantico-pragmatically-based marking of grammatical relations, opposed to the ergative. This
label is not used in this article, as one of the objectives of the paper is to specify the precise syn-
tactic distribution of a particular case marker in order to trace its historical development. Given
its exible syntactic distribution, 𗗙jijwill be glossed here as , or , according to
the syntactic function it marks.
. The understanding of ditransitive verbs here follows Malchukov et al. (: ) and refers to
verbs having three arguments, prototypically associated with the semantic roles of Agent (A),
recipient (R), and Theme (T).
. There is no consensus on the case morpheme 𘋩do. Nishida ( []:) and
Arakawa (:) analyzed it as a pure locative case (場所格), expressing stative location as
well as the endpoint of a motion. At the same time, they recognized 𘕿aas an accusative/
dative (對格/與格) case, co-existing with its locative source. Such dierences could be relevant
to corpus types. In my own corpus, since 𘋩dois attested with a dative use while 𘕿a
does not have signicant dative/accusative functions, I follow the analysis of Beaudouin ().
Systematic analyses of these two case morphemes awaits future investigation.
The history of the polyfunctional 𗗙jijin Tangut []
(Section .), (ii) recipient (R) marking with (non-derived) ditransitive verbs
(Section .), and (iii) non-subject argument marking in auxiliary causative con-
structions, involving argument promotion (Section .).
. Monotransitive verbs: Optional O marking
One of the most common uses of 𗗙jijin Tangut is as a DOM marker. The
presence or absence of 𗗙jijdoes not aect object indexation, with the latter
following syntactic rules.
Optional O marking with 𗗙jijoccurs on the basis of semantico-pragmatic
properties of objects, and can thus be described as a prominence-based8rule co-
regulated by humanness, as noted in previous studies (Arakawa :; Kotaka
:–), deniteness, and the discourse topicality of the object.
Examples () and () illustrate the eect of deniteness on the distribution of
𗗙jij. With the same monotransitive verb 𗜍sja‘to kill’, overt case marking
is present for a denite human object in (), but is absent for an indenite human
object in ().
() 𗕾 𗍁 𘓐 𘝦𘃡 𗉋𗺉 𗗙 𗞞 𗜍
[kju1.we2
Ju.city
dzjwo2]A=dj.wji1
people=
[tiow1.thji2]O=jij1
:Zhuochi=
dja2-sja1
1:-to.kill
(Shi et al., : )莒城人殺卓齒。
(Leilin .A.)‘The people of Ju city killed Zhuochi.’
() 𘝨 𘋨 𘊶 𘓺𘁤 𗩍𗷬 𘝦𘃡 𗾞 𗇸 𘓐
𗖌 𗞞 𗜍
zj2
time
[dzjw1
emperor
nj1
aunt
ŋwr1.thjiw2
royal.woman
w1.thjwor2]A=dj.wji1
servant=
nj2
day
zji
1
time
[dzjwo2
people
gj2]O
one
dja2-sja1
1:-to.kill
(Shi et al. :)時帝姑公主奴白日殺一人。
‘At that time, the servant of the emperor’s aunt killed a person in the daytime.’
(Leilin .A.–)
Human and denite objects are not marked if they do not have discourse promi-
nence. In (), the human and denite object the Minister Pi is unmarked. In this
. The term ‘prominence’ in this paper is used in a similar way as in Aissen () and de
Swart () and refers to features triggering DOM. The only dierence is that here promi-
nence also includes topicality in addition to humanness/animacy and deniteness/specicity.
[] Shuya Zhang
instance, it is a non-topical object mentioned in an account as incidental informa-
tion at the ending of a story.9
() 𘜶 𗂸 𗁡 𘂆 𗜍
[tha2.bji2
Minister
phji1]O
Pi
tsj1
also
sja1
kill
(Shi et al. :)又殺大臣嚭
‘(Finally, the King of Yue destroyed the country of Wu, exiled the King Fuchai
(Leilin, .A.)of Wu) also killed the Minister Pi.’
The prominence-based rule of Tangut optional O marking shows three proba-
bilistic tendencies: (i) human/denite objects are most likely to take overt case
marking; (ii) animate and denite objects are occasionally found bearing case
marking; (iii) inanimate objects are not excluded from case marking (Arakawa
:) (e.g. ()), but are the least frequently marked type.
() 𗪘 𗥓 𗗙 𗇐
[ji1
Previous
ŋo2]O=jij1
illness=
dj2
to.heal[A] (心経a-–, r Arakawa :)‘Cure the previous illness (先の病治す)’
. Ditransitive verbs: Alternating R marking
Typical ditransitive verbs in Tangut include ‘Give’- and ‘Say’-verbs, which are
morphologically transitive, with two indexation slots for A and O. Most of them
are morphologically indirective,10 with the T indexed on the verb as the object
(T=O).11 The R, though non-indexable, receives obligatory case marking with the
dative 𘋩door the oblique 𗗙jij.
The distribution of 𗗙jijand 𘋩doin R marking tends to be comple-
mentary in Tangut. The postposition 𗗙jijshows idiosyncratic distribution
. Topicality can be measured by referential distance (Givón :), as no antecedent is
found in the three preceding sentences in the text. A corpus-based study measuring the referen-
tial distance and topical persistence of unmarked objects in Tangut is beyond the scope of this
paper, but is an anticipated line of inquiry for future studies.
. This paper follows Malchukov et al.’s () distinction between secundativity and indirec-
tivity. In Rgyalrongic languages, secundative and indirective verbs are distinguished through
verbal morphology, see Table , and Lai (:–; ) for a detailed discussion.
. Due to the neutralization of verbal inections in Tangut texts (cf. Section .), morpholog-
ical indirectivity is not always obvious.
The history of the polyfunctional 𗗙jijin Tangut []
with a few Give-verbs, such as 𘓯khjow([B] 𘎾khjj)12 ‘to give, grant’ and
𘈈mji‘to grant’. The dative 𘋩dois found with many other ‘Give’-verbs,
among which are 𗼒khu‘to tribute, consecrate’, or 𗨳tj‘to return’ (for a list
of verbs selecting 𘋩do, see Beaudouin ).
However, there are verbs of which the R can be marker with either the
oblique 𗗙jijor the dative 𘋩do, as exemplied by the verb 𘎪tshiij([B]
𘀽tshjii) ‘to say, speak’. In Example (), the addressee (R), Chen Zheng, is
marked with 𘋩do.
()
𘄴 𘟙 𗋕 𗳷 𗂧 𘒏 𘆝 𘅣 𘔜 𗑃 𘋩 𘙇
𘎪
[tshji1
Qi
njij2]A
king
thja1
[thj1
Chu
lhjj2
country
gia1
army
rjijr1
horse
d2]T
aaire
[thjĩ1.tjĩ1]R=do2
:Chen.Zheng=
rjr2-tshjiij1
1:-to.say[A]
‘The king of Qi spoke about the aair of the army of Chu to Chen Zheng.’
(Twelve Kingdoms, Solonin :, cited by Beaudouin )
However, in (), the addressee (R), your majesty, is marked by 𗗙jij. Note that
morphological indirectivity is clearly indicated in (): the verb occurs in its stem
[B] form 𘀽tshjii(< Pre-Tangut *tshjeej-w, Jacques :–), indicating
that the theme, a counsel, is indexed as the direct object.
()
𗂸 𗅉 𘟙 𘜶 𗗙 𗤱 𗖌 𘀽 𗧓
[bji2]A
your.servant
nioow1
[njij2
king
tha2]R=jij1
grand=
[rjj2
stratagem
gj2]T
one
tshjii2-ŋa2
to.say[b]-1
(Shi et al. :)臣更與大王說一計。
‘I’ll tell your majesty a counsel (to save yourself from Wu’s invasion).’
(Leilin, .B.)
Note that in (), the argument marked by 𗗙jijis the addressee (R), as well
as the beneciary of the event. In the example above, extracted from the story of
Goujian Miewu, King Goujian of Yue is faced with the threat of Wu’s invasion and
asks his minister Fan Li for counsel. The addressee, the King Goujian, is also the
one who would benet from the speaker’s counsel. The alternating R marking in
. Alternation between khjowand khjjis not due to the merger of the third person patient
sux *-w (Jacques :–).
[] Shuya Zhang
() and () is a potential indicator of prototypical semantic values of 𗗙jijand
𘋩do, a hypothesis which will be developed in Section ...
. Auxiliary causative
In Tangut, productive causativization is built with the auxiliary verb 𗟻phji
(Stem [B] 𗠔phjo, cf. Table ), originally a motion verb meaning ‘to send’
(Arakawa ; Shi :). The derived causative compound shows bipartite
morphology, with the base verb and the causative auxiliary sharing TAM marking
and argument indexation.
The case postposition 𗗙jijis particularly frequent in Tangut auxiliary
causatives. While previous studies (Arakawa : –; Kotaka :; Shi
:) mention the optional presence of 𗗙jijaer the causee, case mark-
ing for non-subject arguments in causative compounds is more complicated. I dis-
tinguish three such situations below.
.. Optional causee marking with 𗗙jij
The syntax of causative compounds in Tangut distinguishes two situations
according to the transitivity of the base verb.
Causative compounds derived from an intransitive base are bivalent, with the
causer, the transitive subject (A), and the original S as the causee. In this case, the
causee is also the syntactic object ([Acauser, Ocausee]). Both the causer and causee
can be indexed on the verb. The causee optionally receives overt case marking
with 𗗙jij, in accordance with the same prominence-based rules observed for
non-derived monotransitive verbs (cf. Section .).
Example () shows a causative compound derived from the intransitive verb
𗈦lh‘be confused’. In addition being indexed (sux 𘉞-njamarks the
causer, and the stem [B] form of the causative auxiliary reects the causee), the
causee, common people is also case marked with 𗗙jij.
()
𗍳 𗩱 𘝦 𘕣 𗲉 𗏹 𘓐 𗗙 𗈦 𗠔 𘉞
nji2
2
njwi2.dj
skill
wa2
what
djo2
to.have[B]
[ju2
common
dzjwo2]C AUSEE=jij1
people=
lh2-phjo2-nja2
be.confused-[B]-2
(Shi et al. :)汝有何術, 而誣惑常人?
‘What technique do you have to cause confusion in common people?’
(Leilin, .B.–)
The history of the polyfunctional 𗗙jijin Tangut []
With a transitive base, causativization derives a three-place predicate ([Acauser,
Ocausee, oblique]). It introduces a new agent as the causer (Acauser), and the original
A becomes the causee (Ocausee). Both the causer and the causee are indexable. The
original O becomes an oblique of the derived causative compound, which is non-
indexable. The phenomenon of optional causee marking is also observed with
trivalent causative compounds. Example () has a trivalent causative compound
𘟀𗟻 ‘to cause to see, to show’, based on the transitive verb 𘟀ljij‘to
see’. The plural rst person causee is pronominally represented on the verb by the
plural sux -nji, and is also case marked with 𗗙jij.
()
𗧓 𘆄 𗗙 𘙌 𘟀 𗟻 𗐱
[ŋa2.nj2]CAUSEE=jij1
1=
kj1-ljij2-phji1-nji2
1:-to.see-[]-1
(Kepping :, citing Nevskij :)‘Show it to us.’
.. Optional marking on the oblique argument with 𗗙jij
The postposition 𗗙jijis not dedicated only to optional causee marking.
Though infrequent, it is also possible for the oblique argument to be overtly
marked with 𗗙jij. As evidenced by (), with the causative verb 𗯹-𗟻
‘to cause to catch’, case marking occurs aer the oblique argument Shijing.
() 𗩇 𗺹 𗈪 𗕣 𘓐 𘒫 𘏸 𘊐 𗗙 𗯹 𗟻
w1.xew1
:Wenhou
a-tshj1
1:-be.angry
dzjwo2
people
zeew2
to.send
i1.kjij1=jij1
:Shijing=
lju2-phji1
to.catch-[]
(Shi et al., : )文侯怒, 使人捕師經。
(Leilin, .A.)‘Wenhou became angry, and sent people to catch Shijing.’
Although the presence of 𗗙jijcan be explained by the discourse prominence
of the argument Shijing as the protagonist of the story, we also notice that Shijing
is the maleciary,13 aected by the causative event.
. One of the reviewers points out that Examples () and () do not have a maleciary in a
typical sense, but a highly ‘aected patient’. The suggestion is not adopted, since there is no clear
morphological evidence indicating that the arguments marked by 𗗙jijin the two examples
are objects. Moreover, the term ‘aectee’ is avoided in the paper, as it is reminiscent of the the-
ory of aectedness in DOM (Næss ), which might be relevant to DOM in Tangut but is
beyond the scope of the current paper. Terminological clarication is expected for future works,
as the term ‘aectee’ is used as an umbrella term subsuming both beneciary and maleciary in
Geshiza Horpa (Honkasalo :–).
[] Shuya Zhang
.. Alternative causee marking with 𘋩do
There are very few examples in my corpus in which the causee is marked by the
postposition 𘋩do. One such example is given in ().
() 𘑲 𗦵 𗢄 𗘼 𗶃 𘋩 𗋚 𗄾 𗟻
tjow1.p1
:Zhang.ben
jwã2.giuu1.tji2=do2
:Chunyu.zhi=
wj2-sej1-phji1
1:-to.calculate-[]
‘(Zhang Ben’s mother was seriously ill) Zhang Ben sought divination from
(Leilin .B.–)Chuyu Zhi.’14
The occurrence of 𘋩doinstead of 𗗙jijaer the causee in () could be
motivated by semantic factors. In () Chunyu Zhi is not the argument aected
by the causative event; it is the causer, Zhang Ben, asking aer the fortune of his
sick mother, who would benet from the causative event.
. Interim summary
.. Prototypical semantic values of 𗗙jijand 𘋩do
The functional overlap between 𗗙jijand 𘋩doin three-place predicates,
including non-derived ditranstive verbs (Section .) and causative constructions
(Section .) indicates potential semantic overlap in that they are goal-oriented.
However, alternations aer the R and the causee also suggest that they dier from
each other semantically. The arguments marked by the yod postposition 𗗙jij
are also associated with the semantic role of a bene-/maleciary, which is aected
by the event denoted by the predicate. Therefore, I propose Figure to illustrate
the prototypical semantic values of 𗗙jijand 𘋩doin Tangut.
This hypothesis is tested in Example (), which contains a verb-noun incor-
porated predicate 𘞝-𗢯dzjw-lhjwa‘to speak ill of ’. The predicate involves
two non-subject arguments: King Fuchai, the addressee, marked by the dative
𘋩do, and Wu Zixu, marked by 𗗙jij. Although the syntactic status of the
argumet marked with 𗗙jijis ambiguous and could be either an object or an
oblique, it is clear that Wu Zixu is the maleciary, aected by the action of Minis-
ter Pi’s slandering.
. Shi et al. (: ) provided a nearly word-to-word translation, 張本就淳于智處卜, in
which 𘋩dois translated as 處chù ‘place’, reminiscent of its original allative function. From
the context we know that Zhang Ben seeks divination from someone skilled, Chuyu Zhi. The
ambiguity here is in line with the functional continuum of 𘋩doaer an NP human, rang-
ing from the endpoint of a motion to an abstract goal. For a detailed account of the grammati-
calization of 𘋩do, see Beaudouin ().
The history of the polyfunctional 𗗙jijin Tangut []
Figure . Prototypical semantic values of 𗗙 jijand 𘋩do
() 𘜶 𗂸 𗁡 𘕘 𗍏 𗗙 𗥑 𗉪 𘋩 𘙌 𘞝 𗢯
[tha2.bji2.phji1]
Minister.Pi
[ts1.sju2]=jij1
:Wu.Zixu=/
[xu1.thij2]=do2
:Fuchai=
kj1-dzjw1-lhjwa1
1:-to.slander-tongue
(Shi et al., : )大臣嚭谗子胥於夫差。
‘The Minister Pi slandered Wu Zixu to the King Fuchai.’
(Leilin, .A.–B.)
.. Syntactic distribution of 𗗙jijand 𘋩do
Table summarizes the syntactic distribution15 of 𗗙jijand 𘋩doin mark-
ing clausal functions.
Table . Syntactic distribution of 𗗙 jijand 𘋩 do
Object (indexable) Non-indexable
Patient Causee Oblique ( , etc.) recipient; addressee Goal of motion
𗗙 jij rare
𘋩 dorare
. Kepping (:–) also mentioned the occurrence of 𗗙jijaer intransitive sub-
jects. Due to a scarcity of examples, the phenomenon is still unclear and should await further
studies. Optional intransitive subject marking also exists in other Tibeto-Burman languages. In
Lhasa Tibetan, the dative can optionally occur aer intransitive subjects, which could be related
to both event semantics of controllability and information structure (Simon :–). On
the other hand, the use of a subject marker for objects is also observed sporadically, for exam-
ple in Burmese, for which Chappell & Verstraete () suggest a possible reanalysis of a rele-
vant morpheme as a general marker of information structure. In addition, I also found optional
T(heme) marking with very few ditransitive verbs in my corpus. These may be described as
examples of Kittila’s () ‘extended dierential object marking’. Both phenomena remain to
be investigated with a larger corpus.
[] Shuya Zhang
The most active range of 𘋩dolies in the non-indexable interval. As
Beaudouin () remarks, within the source function of allative (i.e. goal of
motion verbs), 𘋩doalready shows a preference to combine with human par-
ticipants, which suggests a clear grammaticalization from allative to dative.
In contrast, 𗗙jijshows a decreasing distribution from indexable to non-
indexable functions. There is no internal evidence from Tangut suggesting a spa-
tial related source for 𗗙jij, except for the fact that it is goal-oriented (cf.
Figure ). Furthermore, it is also dicult to suggest any directionality between the
accusative and the genitive, although there are cases where 𗗙jijexhibits a gen-
itive/oblique ambiguity (see Section .. for the same phenomenon in modern
Rgyalrongic). In general, reanalysis of possessors as accusatives is typologically
rare, and is considered counter-directional grammaticalization (Narrog :).
The following sections examine comparative data from modern Rgyalrongic
languages, in hopes of unveiling the origins of the postposition 𗗙jijand
explaining genitive/accusative syncretism.
. Cross-Rgyalrongic comparison
The Tangut 𗗙jijreects the Proto-Rgyalrongic yod postposition *=j( V)
(Jacques :). Reexes of this postposition are still productive in modern
West Rgyalrongic languages and Situ Rgyalrong (i.e. the eastern branch).16 How-
ever, the yod postpositions in these modern relatives of Tangut function very dif-
ferently. In modern West Rgyalrongic languages, the yod postpositions show a
genitive/oblique syncretism (Section .), and those in Situ basically function as
general locatives (Section .).
Despite their functional dierences, the productive uses of the yod postpo-
sitions in modern Rgyalrongic languages can oer clues to possible sources of
the Tangut 𗗙jijand the associated genitive/accusative syncretism discussed in
Section .
. Modern West Rgyalrongic: Genitive/oblique =jV
The yod postpositions in modern West Rgyalrongic languages present genitive/
oblique syncretism. For the genitive function, =jV serves to express adnominal
. Japhug (Northern Rgyalrong) preserves traces of the yod locative. The word qhuj ‘tonight’
is potentially a fossilized locative form of the relator noun -qhu ‘aer’ (Jacques : –).
The history of the polyfunctional 𗗙jijin Tangut []
possession, as exemplied in (). This clearly corresponds to the Tangut genitive
𗗙jij(e.g. Example ).
( )
() genitive =ji
[tî]PR=ji
:Bkra.shis=
[pîr]PE
pen
(Lai :)‘Bkrashis’ pen’
The oblique function of the yod postposition should be understood with respect
to verb classication in modern Rgyalrongic languages (cf. Table ), which is
strictly based on morphological transitivity (i.e. indexation patterns) and case
frames. In general, three non-subject categories can be distinguished, (i) the
accusative (i.e. the indexable O), (ii) the dative, which is in most cases non-
indexable, but is indexable with secundative verbs, and (iii) the oblique, which is
non-indexable.
In the non-indexable domain, we also notice competition between the dative
and the oblique in marking extended arguments in two constructions: (i)
extended intransitive verbs, and (ii) indirective verbs.
Table . Verb classication, indexation types and case frames in modern West
Rgyalrongic languages (based on Lai : )*
Verb type Indexation Case frame
Intransitive
Semi-transitive
Extended intransitive
V[S] S=ø
S=ø
S=ø
E=ø
E=
E=
Transitive Indirect transitive V[A,O] A(=)
A(=)
A(=)
O=ø
O=
O=: (Khros)
Secundative V[A,R=O] A(=) T=ø R=
Indirective V[A,T=O] A(=)
A(=)
T=ø
T=ø
R=
R=
Multitransitive A(=) T=ø R(E)= E=
* E stands for the extended argument (cf. Dixon & Aikhenvald ).
.. Extended intransitive verbs
The dative that marks extended arguments shows idiosyncratic distribution, and
is found with verbs such as ‘to help’ (
rin Khroskyabs, or in Geshiza), ‘to fear’
(nsc
rin Khroskyabs, stær in Geshiza) (Honkasalo :; Lai :–).
[] Shuya Zhang
The distribution of the oblique =jV is rarely idiosyncratic, and is associated
with a predictable semantic role of experiencer (Honkasalo :). It is typ-
ically found with stative verbs, alternating between basic intransitive [S=ø] and
extended intransitive [S=ø, E=].17 This is exemplied by the modal verb ró ‘be
necessary, should’ in Khroskyabs, which is intransitive and basically involves an
impersonal subject, as in (a). It can be ‘supplemented’ with an additional non-
subject argument marked by the oblique =ji, expressing the meaning ‘to want,
need; be necessary to’, as in (b).
( )
() a. ró [S=ø] ‘be necessary, should’
ægæ
jd
=
water=
tsâtsâ
well
r
to.washI
ró
should
‘Then, one should wash them well with water. (Lit: washing them well
(Lai :)with water is necessary)’
b. ró [S=ø, E=] ‘want, be necessary to’
nû=ji
2=
tǽ=æ
whatever=many
r-ró
-be.necessaryI
‘You want too much (Lit: Too much things are necessary to you)’
(Lai :)
.. Indirective verbs
The dative is the most important means for R marking. It occurs aer the R of
all secundative (e.g., ‘to feed’: Khroskyabs b
, Geshiza m) and most indirective
(e.g., ‘to ask’: Geshiza rjæ, Khroskyabs rǽ) verbs in modern West Rgyalrongic
languages. Only a small proportion of indirective verbs with their R marked by
the oblique, are typically associated with the semantic role of a bene-/maleciary
(Honkasalo :–).
First, manipulation verbs (e.g. Khroskyabs vǽ ‘to bring’, tǽ ‘to take away
(to)’; Geshiza mbe ‘to carry away’, næ ‘to bring’ are subsumed under a larger
semantic class of orientable verbs, further details of which are discussed in
Section .) consist a particular subtype of indirective verbs in Rgyalrongic lan-
guages. They denote an induced motion, the R of which can be understood as an
abstract goal intrinsically required by the verb’s semantics. As in (), the oblique
=je, merged with plural marker i, serves to mark the R of the verb næ ‘to bring’.
. Productive alternations between intransitive and extended intransitive could ultimately
give rise to idiosyncrasies. The Geshiza verb æn ‘to miss (someone)’, requiring an experiencer
marked by the oblique =je (Honkasalo :), could be lexicalized from the alternating
structure [S, E=] of a basic intransitive verb *æn ‘to lack’ [S].
The history of the polyfunctional 𗗙jijin Tangut []
( )
() næ ‘to bring’ [S=, T, R=]
go
evening
ŋæ=i
1=.
vdæ=wo
friend=
<tiænxua>
phone.call
dæ-næ-si
-bring.3-
(Honkasalo :)‘In the evening, our friend called us.’
Second, the oblique can occur with a group of monotransitive verbs which do
not denote a motion event, and optionally ‘supplement’ a third argument acting
as the bene-/maleciary of the denoted action (Honkasalo :). In (), the
yod postposition =je supplements a beneciary, the Balang villagers, of the action
denoted by the verb v-ql ‘to divide’.
( )
() v-ql ‘to divide’ [S=, T, R=]
lmo
3.
wo
bear
æ-l
one-..
dæ-v-sæ
--kill..3
zda
..
dæ-v-sæ
--kill..3
ta=ræ
when=
næ-v-tæ=ræ
.--bring..3=
bra
stpa=je
villagers=
dæ-v-ql-si
--divide..3-
ŋu
.3
‘He has killed a bear. When killing the bear he brought it down (to Balang),
(Honkasalo :)and divided it for Balang villagers.’
Some transitive verbs can be extended to become quadrivalent, associating with
three non-subject arguments: a non-casemarked Theme, a Recipient marked
with the oblique, and a Source argument marked with the dative (Lai :;
Honkasalo :–). In (), with the manipulation verb (r)-vǽd (Stem
II z
m) ‘to bring’ in Khroskyabs, the dative =ke yields an ablative reading and
marks the source of the transfer event. The ablative reading, which seems to
be semantically contradictory with the dative one, originates from its original
adessive function ‘near, at the place of…’ (Lai : ).
( )
() z
m‘to bring’
tî=
Bkra.shis=
nû=ji
2=/
kp
book
ŋ
=ke
1=
k-u-z
m
--to.bringII (Lai :)‘Bkrashis brought you a book from me.’
Although in () the yod postposition =ji is ambiguous due to the contiguity of
the R and the T and could be either an oblique (encoding the beneciary) or a
genitive (encoding the possessor), its goal-oriented property is obvious, and the
inverse case alignment is ungrammatical.
[] Shuya Zhang
. Situ Rgyalrong: General locative =j
The yod postposition occurs in a non-syllabic form =j in most Situ dialects (Lin
:–; Lin : –; Prins :–; Zhang : –; etc.).
Like its syllabic cognates =jV in the West branch, it also operates at the phrasal
level.18
As a general locative, =jin Situ has two identiable interpretations: as a loca-
tive expressing stative location (e.g. (a)) and as an allative indicating the end
point of a motion (e.g. (b)). The allative reading is triggered with orientable19
verbs, including (i) motion verbs such as ‘to go’ and ‘to come’, (ii) manipula-
tion verbs denoting an induced motion, such as ‘to take’, and (iii) orienting verbs
denoting a ctive motion. These verbs morpho-syntactically require a goal argu-
ment obligatorily marked with the locative.
-
() a. Locative: stative location
ŋ-wûj
1.-grandson
kn
s
two
t
[t-
m=j]
.-house=
ŋ
s-nt
beI-3
‘My two grandsons are at home.’
b. Allative: endpoint of a motion
[a-ti
=j]
-up=
r-bê-n
:-comeI-2
(Lit: come to the upper part)‘Come up. ’
Locative phrases in Situ oen occur in one of the two syntactic positions: as
an oblique argument (e.g. (a); (b)), or as an adjunct. Two locative phrases
can occur successively, with an ‘outer’ locative adjunct providing background
information and preceding an ‘inner’ locative, as in (). Note that though not
obligatory, the ‘outer’ locative is followed by the prominence marker k, which
developed from an ergative source (Prins :–; Zhang :–).
. In some varieties, the yod postpositions develop new functions as clause subordinators.
The Khroskyabs =ji has developed the function of a relativizer from a genitive (Lai: ). In
Brag-bar (Situ), the locative =jcan occur aer nominalized adverbial clauses.
. Orientable verbs in Rgyalrongic languages constitute a semantic class exhibiting distinct
morphological behaviors. While most verbs have one or two directional prexes which have
lexicalized for the perfective and imperative as pure markers, orientable verbs are free
to select directional prexes, encoding both spatial orientations and values. See Zhang
(:–) for details on directional prexes and orientability in Situ, and Jacques
(:–) for the Northern branch.
The history of the polyfunctional 𗗙jijin Tangut []
-
() [xán
downstream..
comcó=j
:Kyom.kyo=
k]OUTER
[<nongchang>=j]INNER
farm=
na-thê-n
:-to.I-2
‘(You) go down to the farm at Kyom-kyo.’
The general locative =jin Situ has various non-spatial uses, all of which are related
to the two syntactic positions of adjunct and oblique.
.. Delimitative adjuncts
The yod postposition =jin Situ can be used to encode sentence topics, roughly
equivalent to English ‘as for’ topics. This function originates from the clause-
external function of the locative =j, which is typically associated with the sentence
initial position, as exemplied in ().
-
() r-trpi
=j
one-:catty=
r-pthár
one-banknote
kn
s-pthar=ksn
two-banknote=like.this
cám
about
na-ŋôs.
.-beII
‘As for each catty (of Iris), it sells for about one or two yuan.’
.. Oblique
The oblique function of the Situ locative =jis similar yet not identical in terms of
its distribution to the oblique =jV of modern West Rgyalrongic languages.
As shown in Table , the Situ locative =jis mainly found in constructions with
(i) extended intransitive and (ii) indirective verbs. The locative and the dative also
compete in the marking of extended arguments in these two constructions. How-
ever, the Situ dative wo-phá=j is only a semi-grammaticalized locative phrase,
comprising (i) a possessive prex, pronominally representing the possessor, (ii) a
nominal base *pha ‘vicinity’,20 and (iii) the yod locative =j.21
... Extended intransitive verbs
With extended intransitive verbs, the Situ =jfunctions in the same way as does the
oblique =jV in modern West Rgyalrongic languages (Sections ..). The yod loca-
tive =jin Situ is also found with stative verbs, to which it optionally introduces an
experience (e.g. k-râ V[S=ø] ‘be necessary’; V[S=ø, E=] ‘be necessary to’).
. Reconstruction of this root awaits further revision. Dative relator nouns within Situ show
irregular correspondences, with an aspirated labial ph- in Brag-bar wo-phá=j, and a soened
labial approximant w- in Cogtse w-wâ=j (Lin : ).
. In Brag-bar the yod locative can be replaced or followed by the locative t in Brag-bar
(Zhang :). In Cogtse it can be replaced by the sibilant locative =s(Lin ).
[] Shuya Zhang
Table . Verb classication, case marking and indexation types in Brag-bar Situ
Verb type Indexation Argument structure
Intransitive
Semi-transitive
Extended intransitive
V[S] S=ø
S=ø
S=ø
S=ø
E=
E=
E=ø
Transitive V[A,O] A(=) O=ø
Secundative ditransitive V[A,R=O] A(=) T=ø R=ø
Indirective ditransitive V[A,T=O] A(=)
A(=)
O=ø
T=ø
R=
R=
... Indirective verbs
A major distributional dierence between the locative =jin Situ and the oblique
=jV in modern West Rgyalrongic languages exists in ditransitive constructions.
With indirective verbs, the Situ locative =jis reserved for manipulation verbs
intrinsically involving a goal argument. The argument marked by =jcan either be
the concrete goal of motion, or the abstract goal of a bene-/maleciary (typically
a human participant). As exemplied by (), =jmarks the maleciary of manip-
ulation verb ka-s-bê ‘to induce (a disaster)’, derived from the cislocative verb ka-
bê ‘to come’ with the causative prex s-.
-
() wo-ŋkhú=j
3.-aer=
t
=j
oneself=
máksn
like.that
ma-k-bd
-:/-be.good
o
ma-ka-n-s
-be
----come
wo-rô
3.-words
n
-ŋs.
-beI
‘(This story is about) not bringing bad things to oneself in the future.’
Unlike modern West Rgyalrongic languages (Section ..), the yod postposition
in Situ is not available for non-orientable monotransitive verbs, such as ‘to buy’, or
‘to divide’, since they do not intrinsically involve a goal argument. For these verbs,
bene-/maleciary encoding relies on pronominal possessive prexes. As in (),
the beneciary of the action of buying is encoded by pronominal possessive pre-
xes, with ŋ- occurring on the theme, which is indexed on the verb as the syn-
tactic O.
-
() ka-kî ‘to buy’ ()
The history of the polyfunctional 𗗙jijin Tangut []
ŋá
1
ŋ-wô
1.-head
na-m
ŋam,
-be.painfulI
ŋ-sm
n
1.-medicine
r
one
r-kî-n
-to.buyI-2
(Lit: buy my medicine)‘My head aches, buy me some medicine. ’
The restriction of locative encoding of bene-/maleciary to orientable verbs in
Situ reects a very common extension of the allative; from a concrete goal to an
abstract goal (cf. Heine ).
.. Locative =j and possession
In core Rgyalrong languages, possession is predominantly expressed through pre-
xing morphology. Most nouns are inalienably possessed, lexically taking an
indenite possessive prex t-orta-. Denite possession is expressed by replacing
the indenite possessive with a set of pronominal possessive prexes (cf. Jacques
:; Zhang : ; etc.). The possessive prexes occur on the possessee
NP, and agree with the person and number of the possessor, which can oen be
omitted, for instance:
-
() t-mí (ŋá) ŋ-mí
.-daughter .-daughter
‘daughter (default form)’ ‘my daughter’
In addition to the prevailing prexing possessive, the locative =jin Situ can also
serve to express possession in three constructions, as discussed below.
... Locational possessive
One important function of the locative-existential clauses in Situ is to express the
‘locational possessive’ (cf. Stassen :–).22
The locational possessive in Situ comprises three elements: (i) the existential
predicative ndó ‘exist’, occurring in an invariant form; (ii) the possessor (),
occurring as an oblique argument, marked by the locative =j; and (iii) the pos-
sessee (), which is the intransitive subject of the existential copula, as in ().
Note that the possessee NP does not agree with the possessor NP in this example,
and occurs in its default form with the indenite possessive prex t-.
-
() locational possessive
. The phenomenon prevails in Tibeto-Burman languages, for further discussions see Stassen
(:–) and Chappell & Lü ().
[] Shuya Zhang
-
ŋá=j
1=
t-
m
.-house
ks
m
three
ndó
’.
(Zhang :)‘I have three houses (Lit: to me, there are three houses)’
The distribution of the locational possessive varies across Situ dialects. For
instance, it is used for kinship relations in Cogtse Situ (Lin :–), but such
uses are not attested in Brag-bar Situ.
... Locative pronouns
In Situ, the locative =jcan be directly attached to personal and interrogative pro-
nouns, causing them to function like possessive pronouns (Nagano : –;
Prins :; etc.). Situ locative pronouns oen occur in equative copula
clauses, and serve to emphasize a possessive relationship (Lin :).
For instance, in (), the singular rst person locative pronoun ŋá=j appears
with the equative copula ŋ
s‘be’. The construction expresses a meaning similar to
the dative possessive in French (c’est à moi).
-
() t
ta-rkâ
.-mule
t
ŋá=j
1=
ŋ
s
beI:
‘The mule is mine.’
The possessive reading of locative pronouns could have originated from the struc-
tures where the gure is omitted, and the locative phrase has a contextual referen-
tial status, as in ().
-
() ŋá
1
ŋa-zgû=j
1.-back=
z
also
n
=j=ksn
2==like.that
nda∼ndâ
be.identicalI.
ostó
really
k-dî
:/-be.heavy
ndó
existI’ (Zhang : )‘That (burden) on my back is as heavy as (the one) on you.’
... ‘Double marking’
Situ also has a ‘double marking’ pattern, in which the possessor has an encliticized
locative =j, and is indexed on the possessee NP with a possessive prex. As in
(), the possessor NP, t-rmî o ‘people’, is both case marked by the yod locative
=j, and pronominally represented on the possessee NP by the plural third person
possessive prex i-.
The history of the polyfunctional 𗗙jijin Tangut []
-
() mâ
.
wo-ŋkhú
3.-aer
tn,
t-rmî
.-people
o=j
=
i-ka-ndzi
.3-:-to.eat
wo-spâ
3.-material
t-rg
k
.-grains
t
máksn
like.that
tni
n-k
-va-u
--to.doII-3
ŋ
s
beI.
ka-ts
-to.say
n-ŋs.
-beI
‘Aer that, he made human food like that.’
As Heine (:) remarks, the phenomenon of ‘double marking’ of a possessor
could be related to a ‘Topic Schema’ ‘(As for X), X’s Y’. As for Situ, the double
encoding of possessors might have originated in constructions in which a posses-
sor is also encoded as a sentence topic. In (), the possessor NP marked by the
locative occurs as the sentence topic. The possessive prex i- occurring on the
possessee NP reects the person and number of the topic possessor, and serves
as morphological evidence that distinguishes the Topic possessive () from the
Locational possessive (cf. Section ...).
-
() smó=j
:So.mang=
t
i-ralp
3.-king
t
n-ndo
-existI’
‘As for Somang (people), they had their (own) king.’
In (), the sentence topic and the possessee NP are separated by the optional
determiner t, which is used as a contrastive marker. The absence of this element
could easily lead to contiguity of topic possessor and possessee, resulting in the
‘double marking’ pattern.
. Multi-directional development of the yod postposition
The unidirectionality hypothesis of grammaticalization predicts increasing gram-
maticalization from spatial concepts to more abstract case functions (Heine et al.
:). Among the three Rgyalrongic branches discussed in this paper, it is
appropriate to postulate that the Situ locative represents one of the more archaic
uses of the Proto-Rgyalrongic yod postposition. The yod postposition’s devel-
opment towards genitive/oblique syncretism in modern West Rgyalrongic, and
towards genitive/accusative syncretism in Tangut can be explained through two
main grammaticalization pathways. They are summarized in ().
() a. From allative to accusative (Section .)
b. From locative to genitive (Section .)
[] Shuya Zhang
. From allative to accusative
The promotion of the Proto-Rgyalrongic yod locative postposition to an
accusative could have occurred in two steps. First, the allative was reanalyzed as
an oblique case in Proto-West Rgyalrongic (Section ..), then the oblique was
further promoted to an accusative in Tangut (Section ..).
.. From allative to oblique in West Rgyalrongic
One of the major functions of the oblique case =jV in modern West Rgyalrongic
languages is to mark bene-/maleciaries in indirective constructions (cf.
Section ..). This function is also attested with the Situ locative =jas an exten-
sion of its allative function, albeit with a slightly dierent distribution (cf.
Section ...). The transition from the Proto-Rgyalrongic allative function to
the West Rgyalrongic oblique function (with bene-/maleciary related meanings)
may have involved: (i) the loss of the original spatial meaning, and (ii) reanalysis
of the allative as a marker of bene-/maleciaries.
This hypothesis nds support in the morpho-syntactic properties of motion
verbs in West Rgyalrongic languages, which alternate between semitransitive, tak-
ing an unmarked goal argument ([S=ø, E=ø]), and extended intransitive, with
a postpositional goal argument ([S=ø, E=]) (cf. Table ). At the synchronic
level, dierential goal marking is regulated by semantico-pragmatic factors:
default goals are unmarked (e.g. (a)), and overt marking occurs when there is a
need for topological specication (e.g. (b), (c)).
( )
() rbjǽ ‘to arrive’
a. æ
mo
mother
j
m
house
r-rbjǽ=mbæ
-arriveI=until
(Lai :)‘Before mom arrives home…’
b. rdzngá
:Rdzong.’gag
z
m=ke
bridge=
læ-rbj-
ŋ
-arriveII-1
(Lai :)‘I arrived near the Rdzong.’gag bridge’
c. <gn>
:Gun
vú=l
people=:inside
næ-rbjî
-arriveII
k
(Lai :)‘Gun went into the crowd.’
Unmarked goals in West Rgyalrongic languages (e.g. (a)) correspond to strict
locative marking in Situ (Section .), and can be considered to represent a gap
le by the loss of the original allative (spatial) function of the yod postposi-
The history of the polyfunctional 𗗙jijin Tangut []
tion.23 However, the yod postposition =jV did not completely disappear follow-
ing the goal arguments of motion verbs in West Rgyalrongic languages. Rather,
it is restricted to human goals. In (), the yod postposition =ji occurs aer a
human goal, which can be understood as a metaphorical goal of the ctive motion
denoted by the verb v
a‘to go’, which in this case can be translated to ‘to go well
with, to t’.
( )
() c
χtulû =t
hat=
ŋæ
=ji
1=
dvâ
exactly
næ-a=t
-to.goI=
r-ŋǽ
-beI(Lai :)‘This hat ts me very well (Lit: this hat goes well with me).’
This yod postposition’s semantic shi to goal [+human], as demonstrated in ()
also helps to explain its occurrence in ditransitive constructions with manipula-
tion verbs (cf. Section ..). On the one hand, manipulation verbs are orientable
and intrinsically include a goal argument; on the other hand, the R of a ditran-
sitive construction is oen a human participant. Under these circumstances, in
extending to non-orientable verbs, such as ‘to buy’, it is likely that =jV develops
into a genuine bene-/maleciary marker.
()
Therefore data from modern West Rgyalrongic languages support an increasing
grammaticalization from allative to oblique, with bene-/maleciary related func-
tions (as summarized in ); a typologically frequent grammaticalization pathway
(Heine et al. :; Heine ; Narrog ; etc.).
.. Promotion of oblique to accusative in Tangut
Although it is widely accepted that the dative is a transit stage for the second
grammaticalization of a spatial case to the accusative (cf. Section .), the promo-
tion of the Proto-West-Rgyalrongic oblique to the Tangut accusative seems to be
an exception.
. In Situ, topological specications are realized by locative phrases based on spatial relator
nouns such as wo-li
=j ‘in the middle of ’. Despite the semantic discrepancy, in West Rgyalrongic
languages, some spatial case postpositions with topological specication have transparent
sources cognate with spatial relator nouns in Situ. For example, Khroskyabs =l ‘locative:inside’
is cognate with Brag-bar wo-li
‘middle’, and =g ‘loctaive:inside’ with Brag-bar wo-ŋgú ‘inside’.
[] Shuya Zhang
The yod postposition did not develop into a dative in any modern Rgyal-
rongic languages; on the contrary we observe that datives developed indepen-
dently in dierent branches: (i) Khroskyabs and most Horpa languages have velar
datives (Honkasalo ; Gates ; Lai ; Sun & Tian ); (ii) the Horpa
of Stodste and Rtsangkhog have a dental dative =do (Sun ; Lai ) cog-
nate with the Tangut allative/dative 𘋩do; (iii) Situ has less grammaticalized
dative relator nouns (cf. Section ..). This formal diversity suggests that dative
as a dedicated case category was formed very late in Rgyalrongic languages. In
Tangut, it is unclear whether or not 𗗙jijunderwent a dative stage. Its limited
occurrences aer the R of non-derived ditransitive verbs still fall within the func-
tional domain of the oblique.
An alternative hypothesis is that the Tangut accusative developed from
oblique. This development is likely to have been realized via bridging syntactic
constructions involving argument promotion, such as the auxiliary causative (cf.
Section .). For one thing, the causative auxiliary in Tangut 𗟻 (stem [B]
𗠔) is originally an orientable verb which means ‘to send’ and syntactically
involves a goal argument. Therefore, the optional causee marking with 𗗙jij
can be considered to have been initially driven by semantics properties, and in
particular by humanness, as observed with =jV in modern West Rgyalrongic lan-
guages (e.g. Example ). For another, since the auxiliary causative in Tangut
is an argument promotion mechanism, it can promote the causee, which can
be marked by the yod postposition, to a syntactically indexable O. In this case,
𗗙jij, which was originally a marker for a human goal, would have been rean-
alyzed as an accusative marker.
Since an auxiliary causative based on a motion verb is Tangut innovation and
is not found in any other Rgyalrongic languages, this also explains why the yod
postposition did not develop an accusative in other West Rgyarongic languages.
.. Information structure and accusative marking
Unlike the Romance languages, where DOM rst emerges as a topic marking
device (Iemmolo ), information structure is likely to have been a contributor
in the grammaticalization of the yod postposition in Rgyalrongic languages.
The existing data only allow us to speculate that information structure may
have promoted the spreading of DOM in Tangut. In modern West Rgyalrongic,
the two semantic roles that are associated with the oblique =jV, the experiencer
and the bene-/maleciary, are typically human and denite. Although Tangut
DOM shows a strong tendency to appear with human and denite objects, it is
not limited to them. Tangut DOM can potentially be grouped into the Type of
Dalrymple & Nikolaeva’s (:) typology, wherein DOM rst applies to topic
The history of the polyfunctional 𗗙jijin Tangut []
and non-topic arguments with the semantic features of human and denite, and
then spreads to other topic-worthy arguments. Needless to say, the relationship
between information structure and DOM in Tangut is worthy of further study.
. From locative to genitive
Since the genitive function of the yod postposition is shared by all West Rgyal-
rongic languages, reanalysis from the locative as a genitive should predate the
West-Rgyalrongic split.
Reanalysis of spatial and directional markers as genitives is a typologically fre-
quent occurrence (Heine :; Stassen :). For the Rgyalrongic lan-
guages, the reanalysis of a locative as a genitive could be the result of multiple
dierent pathways. Among the ve source schemes that Heine (:) pos-
tulates for the expression of attributive possession, at least two (i) the Location
Schema (Y at X), and (ii) the Topical Schema (as for X, X’s Y) that are relevant
to the present discussion. Both schema involve the reanalysis of constituency
(Stassen :–): two noun phrases occurring in adjacency in an original
predicative possessive construction are reanalyzed as a single noun phrase, with
the preceding noun phrase becoming the adnominal modier of the subsequent
noun phrase.
.. Location schema
The locational possessive, as observed in Situ (cf. Section ...), is the bridging
construction which has allowed the reanalysis of a locative possessor as a genitive
in West Rgyalrongic.
As illustrated in (), due to the contiguity, the locative possessor, originally
an oblique argument, is reanalyzed as an adnominal modier of the subsequent
possessee, which is the original subject of the copula predicate.
() Genitive possessive reanalyzed from locational possessive
a. Locational possessive [=] [] [.]
b. Adnominal possessive [= ] ([.])
A locational possessive with the yod postposition can be traced back to Proto-
Rgyalrongic. Tangut has a dozen existential verbs (Ikeda ; Shi :–),
among which 𘟣dju, a common Rgyalrongic etymon (cognate with
Khroskyabs d
, Situ ndó), demonstrates idiosyncratic combination with the yod
[] Shuya Zhang
postposition when expressing possession (Kotaka : ; Ikeda :), as
evidenced in ().24
() TANGUT (WEST RGYALRONGIC)
𗡸 𗗙 𗥃 𗣼 𘟣
sji2=jij1
woman=
ljr1
four
thja2
integrity
dju1
(Shi et al. :, cited in Ikeda : )婦有四德
(Leilin, .B.)‘Women have four integrity.’
A second pathway, through semantic extension from the Location Schema, could
also be assumed. This would have happened with locative phrases with referential
status (cf. Section ...), and in particular with pronouns, as illustrated in ().
() Metaphorical extension from the Location Schema
= ‘what is at X’ > = ‘X’s’
.. Topic Schema
Section ... illustrated that the ‘double marking’ pattern in Situ might have
developed from the topic possessive, which suggests that a reanalysis from topic
possessor to genitive possessor was a possibility in Situ, though it remained a mar-
ginal phenomenon. (from (a) to (b)). Additionally, the ‘double marking’ pat-
tern itself (b) can be reconstructed as an intermediate stage in the reanalysis
from topic possessive to genitive in West Rgyalrongic.
() Genitive possessive reanalyzed from the Topic Schema
a. Topic possessive [=] [.-]
b. Double marking [= .-]
c. Adominal possessive [= ]
The transition from stage (b) to (c) was achieved through the loss of pos-
sessive prexes, which is a shared morphological innovation of the West branch.
Although there are no clear traces of personal possessive prexes le in any West
Rgyalrongic languages, indenite possessive prexes *tV- can be found as fos-
silized preinitials in a few lexical forms, such as the preinitial t- in Khroskyabs tv
‘st’ (Proto-Khroskyabs *t[]-C.kut, Lai : ), cognate with Brag-bar Situ t-
rkút.
. Tangut has another locational possessive predicated by the existential verb 𘂬o, but
the locative possessor is encoded by the general locative 𘕿a.
The history of the polyfunctional 𗗙jijin Tangut []
. Beneciary possessor
Heine (:) also suggests a Goal Schema, namely a unidirectional grammat-
icalization from markers of allative, dative, or benefactive to genitive. Yet in West
Rgyalrongic languages, it is not clear whether the bene-/maleciary meaning his-
torically precedes the genitive meaning.
Bene-/maleciary encoding for non-orientable verbs with the yod postposi-
tion may have been a simple analogy from orientable verbs, and a side eect of the
increasing grammaticalization of the allative in West Rgyalrongic (Section ..).
However, we cannot preclude an alternative hypothesis that the newly developed
genitive =jV occurs in compensation for the loss of possessive prexes in West
Rgyalrongic. Note that Core Rgyalrong languages use the possessive prexes
to encode the bene-/maleciary of non-orientable verbs (cf. Section ..., see
also Jacques :–, for Northern Rgyalrong languages). The ambiguity
between the bene-/maleciary and genitive always exists.
Figure summarizes all the assumed historical developments of the yod post-
position in Rgyalrongic languages presented in research to date.
Figure . Diachronic development of the yod postposition in Rgyalrongic Languages
. Conclusion
This paper explained the historical development of a polyfunctional case marker
𗗙jijin Tangut, and attributed its genitive/accusative syncretism to multiple
grammaticalization from a proto general locative case. This syncretism appears
confusing in Tangut owing to the complete loss of the original locative function,
which can only be reconstructed through comparison with more conservative sis-
ter languages (Section ).
The two main grammaticalization pathways assumed in this paper, from alla-
tive to accusative, and from locative to genitive with the help of constructional
change (Section ), conrm the hypothesis of unidirectional grammaticalization
[] Shuya Zhang
(Heine et al. ). Grammaticalization from a spatial concept to an accusative,
and to a genitive are both recurrently attested. However, the case of Tangut shows
that typological rarity can also arise through the layering of various well-attested
grammaticalization pathways.
The Tangut case can contribute to typological studies of case functions in at
least two ways. First, it seems that Tangut, together with Gurage languages and
Tshangla, can be classied as demonstrating maximum syncretism of abstract
case functions (oblique/dative, accusative, genitive). They can also be classied
as a new subtype of Type case syncretism (between a core case and a non-core
case) in Baerman’s () typology.
Second, the development of postpositional accusative marking in Tangut, as
well as in modern West Rgyalrongic languages, challenges the ‘function conti-
nuity’ hypothesis, that states ‘If an adposition occurs as both Object marker and
Allative marker, it also occurs as Dative marker’ (Blansitt :). Dedicated
dative markers were formed aer the branching-o of West Rgyalrongic, since in
Core Rgyalrong the dative is expressed by semi-grammaticalized relator nouns;
even inside the West branch etymological diversity of dative postpositions can be
observed. In Tangut, it is unclear whether a dative stage preceded the accusative
in the increasing grammaticalization of 𗗙jij. Postpositional accusative mark-
ing also emerges in modern West Rgyalrongic languages. In Khroskyabs, objects
of transitive verbs denoting actions involving surface contact are marked by the
superessive postposition =t (Lai ). In Dgebshes Horpa (Sun & Tian )
and Mazur Stau (Gates ), the locative/allative = is developing towards an
object marker. In both cases, the relevant morphemes have not passed through a
dative stage.
One nal remark should be made concerning the diachronic development
of morpho-syntax in Rgyalrongic languages. In West Rgyalrongic languages, we
observe a correlation between the simplication of verbal inections and the
emergence of dependent marking. Prevailing optionality is observed with case
marking in Tangut, which has the most innovative and simplied verbal system (a
few other cases of optional object marking are reported in Horpa, which has also
undergone a partial collapse of its verbal system, cf. Gates ; Honkasalo ).
Lexicalized case frames developed in Khroskyabs, a language which well pre-
serves the verbal inections. If we assume that adposition case marking emerges
in compensation for the disintegration of old inectional systems, as Bossong
() hypothesized for the Romance and Semitic languages, then the factors lead-
ing to optionality, as found in Tangut, and split marking, as found in modern West
Rgyalrongic, are still in need of investigation.
The history of the polyfunctional 𗗙jijin Tangut []
Funding
This research was funded by the Grant-in-Aid for JSPS Research Fellow (F).
Acknowledgements
I would like to thank Guillaume Jacques, Lai Yunfan, Mathieu Beaudouin, Arakawa Shintaro,
Shirai Satoko, Sawada Hideo, Hayashi Norihiko, Agnes Conrad, Nikita Kuzmin, and the
anonymous reviewers for their valuable comments and suggestions. All errors are my own.
[] Shuya Zhang
Abbreviations
rst person
second person
third person
Stem I
Stem II
’ Stem I’
’ Stem II’
[] Stem A
[] Stem B
transitive subject
ablative (before/above ABS)
absolutive
accusative
adessive
allative
autobenefactive
auxiliary
causative
classier
clause marker
copula
dative
denite
demonstrative
determiner
directional prex
distal
dierential object marking
downward direction
dual
extended argument
ergative
existential predicate
experiential
factual
genitive
inferential
imperative
indenite
innitive
instrumental
inverse
imperfective
linker
locative
masculine
negative
nominalizer
non-past
() object
oblique
optative
possessor
perfect
perfective
plural
prominence marker
person name
potential
possessive
possessor
proximate
prex
pronoun
past
participle
recipient
reduplication
intransitive subject
singular
sensory
subject
theme
tense-aspect-modality-evidentiality
toponym
upward direction
The history of the polyfunctional 𗗙jijin Tangut []
References
Aissen, Judith. . Dierential Object Marking: Iconicity vs. economy. Natural Language
and Linguistic Theory (): –. https://doi.org/10.1023/A:1024109008573
Andvik, Erik. . A grammar of Tshangla. Leiden: Brill.
https://doi.org/10.1163/ej.9789004178274.i-490
Andvik, Erik. . Tshangla. In Graham Thurgood & Randy J. LaPolla (eds.), The Sino-
Tibetan languages, –. nd edn, London: Routledge.
Arakawa, Shintaro. . Seika-go no kakuhyo shiki ni tsuite [on the Tangut case markers]. In
Hideo Sawada (ed.), Chibetto=Biruma-kei gengo no bunpo gensho : Kaku to sono shu
hen
[Grammatical phenomena of Tibeto-Burman languages : Case-marking and related
matters], –. Tokyo: Research Institute for Languages and Cultures of Asia and
Africa.
Arakawa, Shintaro. . On the Tangut verb prexes in ‘Tiansheng Code’. In Irina Popova
(ed.),
:
- .
.. [Tanguts in Central Asia: a collection of articles marking the th
anniversary of Prof. E. I. Kychanov], –. Moscow: Oriental Literature.
Arakawa, Shintaro. . Seikago no so su
setsuoji ni tsuite [On the “dual” sux of Tangut]. In
Tooru Hayasi, Tomoyuki Kubo, Setsu Fujishiro, Noriko Ohsaki, Yasuhiro Kishida &
Mutsumi Sugahara (eds.), Yu
rashia sho gengo no tayo sei to do tai go kinengo [Diversity
and dynamics of Eurasian languages: The th commemorative volume], –. Tokyo:
Consortium of Studies of Eurasian Languages.
Arakawa, Shintaro. . Seika-go no shieki ni tsuite [On the causative construction in Tangut]
In Takumi Ikeda (ed.), Shina=Chibetto-kei sho gengo no bunpo gensho : Shieki no shoso
[Aspects of causative construction], –. Kyoto: Institute for Research in Humanities,
Kyoto University.
Baerman, Matthew. . Case syncretism. In Andrej L. Malchukov & Andrew Spencer (eds.),
The Oxford handbook of case, –. Oxford: Oxford University Press.
Beaudouin, Mathieu. . Localiser en tangoute: entre continuums et bipartitions. Bulletin de
la Socie
te
de Linguistique de Paris (): –.
Beaudouin, Mathieu. . Tangut verb agreement: optional or not? Linguistics of the Tibeto-
Burman Area (): –. https://doi.org/10.1075/ltba.21008.bea
Beaudouin, Mathieu. Accepted. Tangut and Horpa languages: some shared morphosyntactic
features. Language and Linguistics.
Blansitt, Edward. . Datives and allatives. In Michael Hammond, Edith A. Moravcsik &
Jessica Wirth (eds.), Studies in syntactic typology, –. Amsterdam: John Benjamins.
https://doi.org/10.1075/tsl.17.14bla
Bossong, Georg. . Empirische universalienforschung: Dierentielle objektmarkierung in den
neuiranischen Sprachen. Tu
bingen: Narr.
Bossong, Georg. . Dierential object marking Romance and beyond. In Dieter Wanner &
Douglas A. Kibbee (eds.), New analyses in Romance linguistics, –. Amsterdam: John
Benjamins. https://doi.org/10.1075/cilt.69.14bos
Chappell, Hilary & Shanshan Lu
. . A semantic typology of location, existence, possession
and copular verbs: Areal patterns of polysemy in Mainland East and Southeast Asia.
Linguistics (): –. https://doi.org/10.1515/ling-2021-0219
[] Shuya Zhang
Chappell, Hilary & Jean-Christophe Verstraete. . Optional and alternating case marking:
Typology and diachrony. Language and Linguistics Compass (). Available at:
https://doi.org/10.1111/lnc3.12311 (last access May ).
Dalrymple, Mary & Irina Nikolaeva. . Object and information structure. Cambridge:
Cambridge University Press. https://doi.org/10.1017/CBO9780511993473
DeLancey, Scott. . “Optional” “ergativity” in Tibeto-Burman languages. Linguistics of the
Tibeto-Burman Area (): –.
Dixon, Robert M.W. & Alexandra Y. Aikhenvald (eds.). . Changing valency. Cambridge:
Cambridge University Press. https://doi.org/10.1017/CBO9780511627750
Endalew, Assefa. . Multifunctional Morpheme j- and -m in Ezha: An Ethio-Semitic
Language. Macrolinguistics (): –.
Gates, Jesse P. . A grammar of Mazur Stau. Paris: EHESS PhD dissertation.
Genetti, Carol. . Object relations and dative case in Dolakha Newari. Studies in Language
(): –. https://doi.org/10.1075/sl.21.1.03gen
Genetti, Carol. . A grammar of Dolakha Newar. Berlin: De Gruyter Mouton.
https://doi.org/10.1515/9783110198812
Givo
n, Talmy. . The pragmatics of de-transitive voice: functional and typological aspects
of inversion. In Talmy Givo
n (ed.), Voice and Inversion, –. Amsterdam: John
Benjamins. https://doi.org/10.1075/tsl.28.03giv
Gong, Hwang-cherng. . Xı xia
yu
do
ngcı de re nche
ng hu
yıng yu
yı nyu
n zhua
nhua
n
[Personal Agreements and Phonological Alternations in the Tangut Verb]. Language and
Linguistics (): –.
Gong, Hwang-cherng. . Tangut. In Randy J. LaPolla & Graham Thurgood (eds.), The
Sino-Tibetan languages, –. London: Routledge.
Gong, Xun. . Verbs stems in Tangut and their orthography. SCRIPTA . –.
Heine, Bernd. . Possession: Cognitive sources, forces and grammaticalisation. Cambridge:
Cambridge University Press. https://doi.org/10.1017/CBO9780511581908
Heine, Bernd. . Grammaticalization of cases. In Andrej L. Malchukov & Andrew Spencer
(eds.), The Oxford handbook of case, –. Oxford: Oxford University Press.
Heine, Bernd, Ulrike Claudi & Friederike Hu
nnemeyer. . Grammaticalization: A
conceptual framework. Chicago: The University of Chicago Press.
Heine, Bernd & Tania Kuteva. . World lexicon of grammaticalization. Cambridge:
Cambridge University Press. https://doi.org/10.1017/CBO9780511613463
Hetzron, Robert. . The Gunna
n-Gurage languages. Napoli: Istituto orientale di Napoli.
Hill, Nathan W. . The historical phonology of Tibetan, Burmese, and Chinese. Cambridge:
Cambridge University Press. https://doi.org/10.1017/9781316550939
Honkasalo, Sami. . A grammar of Eastern Geshiza: A culturally anchored description.
Helsinki: University of Helsinki PhD dissertation.
Iemmolo, Giorgio. . Towards a typological study of dierential object marking and
dierential object indexation. Pavia: University of Pavia PhD dissertation.
Ikeda, Takumi. . Verbs of Existence in Tangut and Mu-nya. In Nathan W. Hill (ed.),
Medieval Tibeto-Burman languages IV, –. Leiden: Brill.
Jacques, Guillaume. . Phonologie et morphologie du japhug (rGyalrong). Paris:
Universite Paris VII – Denis Diderot PhD dissertation.
Jacques, Guillaume. . Textes tangoutes I, nouveau recueil sur l’amour parental et la pie
te
liale (Languages of the World/Text Collections ). Mu
nchen: Lincom Europa.
The history of the polyfunctional 𗗙jijin Tangut []
Jacques, Guillaume. . The origin of vowel alternations in the Tangut verb. Language and
Linguistics (): –.
Jacques, Guillaume. . The Structure of the Tangut verb. Journal of Chinese Linguistics
(): –.
Jacques, Guillaume. . Esquisse de phonologie et de morphologie historique du tangoute.
Leiden: Brill. https://doi.org/10.1163/9789004264854
Jacques, Guillaume. . Tangut, Gyalrongic, Kiranti and the nature of person indexation in
Sino-Tibetan/Trans-Himalayan. Linguistics Vanguard. Available at
https://doi.org/10.1515/lingvan-2015-0033 (last access March ).
Jacques, Guillaume. . A grammar of Japhug. Berlin: Language Science Press.
Janhunen, Juha. . On the taxonomy of nominal cases in Mongolic. Altai Hakpo : –.
Kepping, Ksenija Borisovna. . Subject and Object agreement in the Tangut Verb.
Linguistics of the Tibeto-Burman Area (): –.
Kepping, Ksenija Borisovna. . : [The Tangut Language:
Morphology]. Moscow: Nauka.
Kittila
, Seppo. . Object-, animacy- and role-based strategies: A typology of object
marking. Studies in Language (): –. https://doi.org/10.1075/sl.30.1.02kit
Kotaka, Yuji. . Seikago no kakujoshi ni kansuru to goronteki kenkyu
[A synthetic study of
Tangut case particles]. Hiroshima: Hiroshima University PhD dissertation.
Lai, Yunfan. . Grammaire du khroskyabs de Wobzi. Paris: Universite Sorbonne Nouvelle –
Paris PhD dissertation.
Lai, Yunfan. . Relativisation in Wobzi Khroskyabs and how genitivisation enters it.
Linguistics of the Tibeto-Burman Area (): –. https://doi.org/10.1075/ltba.17015.lai
Lai, Yunfan. . Jia ro
ngyu
zu
yu
ya
n do
ngcı de fe
nle i – yı Chuo
sıjia
(La wu
ro
ng) yu
we ilı[The
classication of verbs in Rgyalrongic languages, from a Khroskyabs perspective].
Yu
ya
nxue
lu
nco ng [Essays on linguistics] : –.
Lai, Yunfan, Guillaume Jacques, Xun Gong & Jesse Gates. . Tangut as a West Rgyalrongic
language. Folia Linguistica Historica (): –. https://doi.org/10.1515/flih-2020-0006
LaPolla, Randy J. . ‘Anti-ergative’ Marking in Tibeto-Burman. Linguistics of the Tibeto-
Burman Area (): –.
LaPolla, Randy J. . On Nominal Relational Morphology in Tibeto-Burman. In
Ying-chin Lin, Fang-min Hsu, Chun-chih Lee, Jackson T.-S. Sun, Hsiu-fang Yang &
Dah-an Ho (eds.), Studies on Sino-Tibetan languages: Papers in honour of Professor
Hwang-cherng Gong on his seventieth birthday, –. Taipei: Institute of Linguistics,
Academia Sinica.
Lehmann, Christian. . Thoughts on grammaticalization (Classics in Linguistics ). Berlin:
Language Science Press, https://doi.org/10.26530/OAPEN_603353
Li, Fanwen. . Xia
-Ha
n Zı
dia
n[Tangut-Chinese Dictionary]. Beijing: Zhongguo Shehui
Kexue Chubanshe.
Lin, Xiangrong. . Jia
ro ngyu
ya
njiu
[A study of the Rgyalrong language]. Chengdu: Sichuan
minzu chubanshe.
Lin, You-jing. . Jia
ro ngyu
Zhuo ke
jı
hua
yu
fa
bia
ozhu
we
nbe
n[Cogtse Rgyalrong texts:
Fully analyzed spontaneous narratives with an updated sketch grammar of the language].
Beijing: Shehui Kexue Wenxian Chubanshe.
Lin, You-jing. . How grammar encodes space in Cogtse Rgyalrong. Himalayan Linguistics
(): –. https://doi.org/10.5070/H916130394
[] Shuya Zhang
Malchukov, Andrej L., Martin Haspelmath & Bernard Comrie. . Ditransitive
constructions: A typological overview. In Andrej L. Malchukov, Martin Haspelmath &
Bernard Comrie (eds.), Studies in ditransitive constructions: A comparative handbook,
–. Berlin: De Gruyter Mouton. https://doi.org/10.1515/9783110220377.1
Meyer, Ronny. . The morpheme ya
-in Muher. Journal of African Languages and
Linguistics (): –.
Meyer, Ronny. . Gurage. In Stefan Weninger (ed.), The Semitic languages: An international
handbook, –. Berlin: De Gruyter Mouton. https://doi.org/10.1515/9783110251586.1220
Mohanan, Tara. . Argument structure in Hindi. California: CSLI Publications.
Næss, A
shild. . What markedness marks: The markedness problem with direct objects.
Lingua : –. https://doi.org/10.1016/j.lingua.2003.07.005
Nagano, Yasuhiko. . Gyarongo bunpo kenkyu
[A reference grammar of the rGyalrong
language – Bhola dialect]. Tokyo: Kyu
ko Shoin.
Narrog, Heiko. . The grammaticalization chain of case functions: Extension and
reanalysis of case marking vs. universal of grammaticalization. In Silvia Luraghi &
Heiko Narrog (eds.), Perspectives on semantic roles, –. Amsterdam: John Benjamins.
https://doi.org/10.1075/tsl.106.03nar
Nevskij, Nikolaj Aleksandrovich. . T ,
[Tangut Philology, Research and Dictionary]. Moscow: Vostochnaya Literatura.
Nishida, Tatsuo. []. Seikago [Tangut]. In Gengogaku Daijiten, vol. , –. Tokyo:
Sanseido.
Prins, Marielle. . A grammar of rGyalrong Jia
omu
zu
(Kyom-kyo) dialects. Leiden:
Universiteit Leiden. https://doi.org/10.1163/9789004325630
Sagart, Laurent, Guillaume Jacques, Yunfan Lai, Robin J. Ryder, Valentin Thouzeau,
Simon J. Greenhill & Johann-Mattis List. . Dated language phylogenies shed light on
the ancestry of Sino-Tibetan. PNAS (): –.
https://doi.org/10.1073/pnas.1817972116
Shi, Jinbo. . Tangut language and manuscripts: An introduction. Leiden: Brill.
https://doi.org/10.1163/9789004414549
Shi, Jinbo, Zhenhua Huang & Hongyin Nie. . Le
ilı
n ya
njiu
[A study on ‘the Forest of
Categories’]. Ningxia Renmin Chubanshe.
Simon, Camille. . Morphosyntaxe et se mantique grammaticale du salar et du tibe tain de
l’Amdo: analyse d’un contact de langues. Paris: Universite Sorbonne Nouvelle – Paris
PhD dissertation.
Sinnemäki, Kaius. . A typological perspective on Dierential Object Marking. Linguistics
(): –. https://doi.org/10.1515/ling-2013-0063
Solonin, Kirill. . Dvendtsat’ Tsarstv. Neizvstnij Pamyatnik Tangutskoi Literatury [Twelve
kingdoms: An unknown monument of Tangut literature]. St. Petersburg: Peterburskoe
Vostokovedenie.
Stassen, Leon. . Predicative possession. Oxford: Oxford University Press.
Sun, Hongkai. . Co
ng cıhuı bıjia
o ka
n xı xia
yu
yu
za
mia
n yu
zu
qia ngyu
zhı de gua nxı[The
relationship between Tangut and the Qiangic branch of Tibeto-Burman from the
perspective of vocabulary comparison]. Mı
nzu
yu
we
n[Minority Languages of China]
(): –.
Sun, Jackson T.-S. . Parallelism in verb morphology of Sidaba Rgyalrong and Lavrung in
Rgyarongic. Language and Linguistics (): –.
The history of the polyfunctional 𗗙jijin Tangut []
Sun, Jackson T.-S. . Morphological causative formation in Shangzhai Horpa. Bulletin of
Chinese Linguistics (): –. https://doi.org/10.1163/2405478X-90000031
Sun, Jackson T.-S. & Qianzi Tian. . Huo
’e
r yu
ge xı hua
do
ngcı duıxie [Verb agreement in
Gexi Horpa]. Bulletin of Chinese Linguistics (): –.
https://doi.org/10.1163/2405478X-90000120
de Swart, Peter. . Cross-linguistic variation in object marking. Nijmegen: Radboud
University PhD dissertation.
Wang, Jingru. . Lu
n Sıchua n Qia ngyu
jı Mıya
oyu
yu
Xı xia
yu
[The Shishiah and the
Ch’iang and Minia Languages in Szu
ch’uan Province]. In Xı
xia
Ya
njiu
, vol. , –.
Taipei: Academia Sinica.
Witzlack-Makarevich, Alena & Ilja A. Serz
ant. . Dierential argument marking: Patterns
of variation. In Ilja A. Serz
ant & Alena Witzlack-Makarevich (eds.), Diachrony of
dierential argument marking, –. Berlin: Language Science Press.
Zhang, Shuya. . Le rgyalrong situ de Brag-bar et sa contribution a
la typologie de
l’expression des relations spatiales: L’orientation et le mouvement associe . Paris: Institut
national des langues et des civilisations orientales PhD dissertation
Zhang, Shuya, Guillaume Jacques & Yunfan Lai. . A study of cognates between Gyalrong
languages and Old Chinese. Journal of Language Relationship (): –.
https://doi.org/10.31826/jlr-2019-171-210
Address for correspondence
Shuya Zhang
Japan Society for the Promotion of Science
Research Institute for Languages and Cultures of Asia and Africa
Tokyo University of Foreign Studies
Asahicho -
Fuchu, - Tokyo
Japan
bragbarskad@gmail.com
Publication history
Date received: December
Date accepted: June
Published online: December
https://orcid.org/---
[] Shuya Zhang
