Sketch of Yakkha Grammar

Aspects of Yakkha Grammar
Diana Saow
(Working paper
Department of Linguistics, University of Leipzig)
May 21, 2010
Contents
1
.
.
.
.
7
7
8
8
9
Phonology
2.1 Orthography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Phoneme inventory and allophonic rules . . . . . . . . . . . . . . . . . . . . . .
11
11
11
Parts-of-spee
17
Nominal Morphology
4.1 Number . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Pronouns . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.1 Personal pronouns . . . . . . . . . . . . . . . . .
4.2.2 Possessive pronouns . . . . . . . . . . . . . . . .
4.2.3 Demonstrative pronouns . . . . . . . . . . . . .
4.2.4 Interrogative pronouns and other interrogatives .
4.3 Case system . . . . . . . . . . . . . . . . . . . . . . . .
Introduction
1.1 e Yakkha language and its speakers . . . .
1.2 e language name . . . . . . . . . . . . . . .
1.3 Earlier work on Yakkha language and culture
1.4 Genetic Aliation . . . . . . . . . . . . . . .
Verbal Morphology
5.1 Stem formation . . . . . . . . .
5.2 Person and number agreement
5.3 Flexible agreement . . . . . . .
5.4 Verbal negation . . . . . . . . .
5.5 Tense . . . . . . . . . . . . . .
5.5.1 Nonpast . . . . . . . .
5.5.2 Past . . . . . . . . . .
5.6 Mood . . . . . . . . . . . . . .
5.6.1 e subjunctive . . . .
5.6.2 e imperative . . . . .
5.6.3 e optative . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
19
19
19
19
20
21
22
23
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
29
29
32
36
36
37
38
40
44
44
44
46
Verbal compounding
6.1 Bipartite verbs . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1.1 Verb-Verb-compounds . . . . . . . . . . . . . . . . . . .
6.1.2 Noun-verb-compounds . . . . . . . . . . . . . . . . . .
6.2 Grammatical compounding: valency anges, aspect, aktionsart
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
49
49
49
50
50
Adverbs and adjectives

7.1 Adverb-building strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.2 Spatial deixis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3 Adjectives and comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
53
53
53
55
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Contents
7.4
Reduplication, triplication and ideophonic paerns in adverbs and adjectives . .
56
e structure of simple clauses

8.1 e noun phrase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2 Copulas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
59
59
60
Argument structure
9.1 Valency ange . . . . . .
9.1.1 Detransitivisation
9.1.2 Causatives . . . .
9.1.3 Applicative . . .
9.2 Experiencer constructions
.
.
.
.
.
61
61
61
61
62
62
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10 Relative clauses and nominalisation

10.1 Nominalisation . . . . . . . . . . . . . . . . .
10.1.1 e nominaliser =na ~ =ha . . . . . .
10.1.2 Nominalisation with -khuba ~ -khuwa
10.2 Corelative clauses . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
65
65
65
67
68
11 Adverbial clauses and converbs

11.1 e supine converb . . . . .
11.2 e simultaneous converb .
11.3 e negation converb . . .
11.4 e comitative . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
71
71
72
72
73
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
12 Conditional and counterfactual clauses
75
13 Complement clauses
13.1 Nonnite complement clauses . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.2 Finite complement clauses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
77
77
78
14 More on nite clause linkage constructions

14.1 Sequential linkage . . . . . . . . . . . .
14.2 Simultaneous linkage . . . . . . . . . .
14.3 Causal conjunctions . . . . . . . . . . .
14.4 e reportatives and text structure . . .
.
.
.
.
81
81
82
82
83
15 Grammatical relations
15.1 Coding properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
85
85
16 Particles used for information structure
87
17 Bibliography
93
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
A note to the reader

is manuscript is the rst outcome of my PhD resear on the Yakkha language, and as su it
represents work in progress. e current second round of eld resear will lead to new insights
and possibly also to dierent analyses of the data gathered in spring 2009. Any comments and
suggestions for improvement are welcome.
Although this is just a preliminary result, I already feel indebted to many people that helped
and still help me in my endeavour to understand and describe Yakkha.
First of all, I want to thank Prof. Novel Kishor Rai for suggesting this almost undocumented
language to me and for establishing the contact to the Yakkha community. e Kirant Yakkha
Chumma (Indigenous Peoples Yakkha Organisation) welcomed me in Kathmandu and supported
my resear in the Yakkha villages.
My supervisor Prof. Balthasar Biel read the manuscript and his comments helped mu to
improve it.
I would also like to thank my main hosts Kamala Linkha and Dhan Kumari Jimi, my assistant
Kamala Koyongwa and many others from the villages of Madi Mulkharka, Tamaphok, Dadagaun,
Mude Saniscare and Mamling for their hospitality and their invaluable eorts in helping me to
understand their language. I am also sincerely indebted to all Yakkha speakers who allowed me
to record their voices in the form of songs, conversations, narrations and ritual language.
e PhD resear is funded by Landesstipendium des Landes Sasen. I would also like to thank
the Max Plan Institute for Evolutionary Anthropology, Department of Linguistics for nancing
both of my eld trips.
Contents
Abbreviations
1,2,3
s/d/p/ns
A
ABL
ADD
ADVBL
ALL
AFF
BEN
COM
COMPL
COND
COP
CTFT
CVB
DEM
DIST
DOWN
ERG
e
FOC
G
GEN
HORT
IMP
INF
INSTR
LEVEL
LOC
NEG
NMLZ
NOM
NPST
OPT
P
PERF
POSS
PROX
PST
Q
REP
S
SBJV
SIM
SUP
T
TOP
TEL
UP
person (1>3: rst acting on third person, etc.)

numerus: singular, dual, plural, nonsingular (when dual and plural are not distinguished)
most agent-like argument of a transitive verb
ablative
additive focus
adverbialiser
allative
armative
benefactive
comitative
complementizer
conditional
copula
counterfactual
converb
demonstrative pronoun
distal (spatial deixis)
down (spatial deixis)
ergative
exclusive
focus particle
most goal-like argument of a transitive verb
genitive
hortative
imperative
innitive
istrumental
same level (spatial deixis)
locative
negation
nominaliser
nominative
non-past
optative
most patient-like argument of a transitive verb
perfective aspect
possessive
proximal (spatial deixis)
past
question
reportative
most subject-like argument of a transitive verb
subjunctive
simultaneous
supine
most theme-like argument of a ditransitive verb
topic particle
telic
up (spatial deixis)
1 Introduction
1.1 e Yakkha language and its speakers
is section deals with the historical, geographical and sociolinguistic baground of the Yakkha
language.
Yakkha is a Kiranti language spoken in eastern Nepal. e Yakkha-speaking villages are located
in the southern part of the Sankhuwasawa district and in the northern part of the Dhankuta district.
Within the area of eastern Nepal known as Kirant, the Yakkha area belongs to the Pallo Kirant
Far Kirant area.
e Yakkha area is mainly surrounded by other Kiranti languages. Going clowise, starting
in the east, these are Tamarkhole Limbu, Phedappe Limbu and Chathare Limbu. Athpare, Belhare, Chlng and Chintang follow in the south, Dungmali and Bantawa in the west, Mewahang,
Lohorung and Yamphu in the north. Further north, there is the Shingsaba language, whi is
also Tibeto-Burman, but not Kiranti. is geographical classication has to be understood in an
idealised sense. Most of the villages are ethnically and linguistically diverse. According to the
UNESCO Working Paper No. 7 (Toba et al. 2005) and the Nepali census of 2001 there are 14.648
native speakers out of about 17.000 ethnic Yakkha, whi makes up 0.07 per cent of the Nepalese
population (Central Bureau of Statistics 2001). One reason why the language is not spoken by all
Yakkha people is migration outward, but there are also whole Yakkha villages inside the original
homeland that have swited to Nepali completely.
e main selements are the villages of the Tin um (e ree Regions) known as Das
Majhiya, Pan Majhiya and Pan Khapan to the east of the Arun river (Kongren 2007a:86).
ese designations originate in the land system introduced in the 18th century by the King Prithivi
Narayan Shah (1723-1775), whi was part of his strategy to create a uniform Nepali nation. He
enforced Hindu law and caste system to the indigenous peoples of the area that is known as
Nepal today. Surnames like Dewan and Jimi are also inherited from these times. Titles su as
Dewan and Jimindar were given to individuals and village headmen in the Yakkha area in order
to implement the Gorkha tax system, and they were later adopted as surnames because of the
power and high social status associated with them. Among the Limbu, this is what happened
with the Nepali/Mughal title Subba, and among the Khambu, the title Rai became a common
name (Whelpton 2005:51). e name Dewan is especially common to those Yakkha who gave
up their homeland and migrated outward. Apart from these non-indigenous surnames however,
ancestral clan names play a vital role in social life and in the ritual sphere. A list of the clans and
subclans (Sameo) can be found in Kongren (2007b:168). e villages with the highest density
of speakers are, according to Kongren (2007a), Syabun, Jaljala, Siddhapokhari, Siddhakali, Wana
in the Pan Khapan area, Haisudhe, Kingring, Kharang and Baneshwor in the Pan Majhiya
area, and Madi Rambeni, Bagha, Waleng (Nep. Madi Mulkharka), Tumok (Nep. Tamaphok),
Tellok, Salle, Mamling, Ankhinbhuin, Yaiten (Nep. Dandagaun) in the Das Majhiya area.
e language use is in decline, for the well-known reasons of intercaste marriage, low prestige
compared to Nepali, and the availability of education beyond the primary-sool-level exclusively
in Nepali or English. Especially the young generation replaces Yakkha with the Himalayan Lingua
Franca Nepali (own observations). According to an ethnological study of T.B. Subba, Yakkha is
replaced with Nepali even at home (Subba 1999). Im am not aware of statistical evaluations about
1 Introduction
how many speakers are still fully competent and how many ildren still acquire the language. It
appeared to me that the general knowledge and awareness of the origins, myths and folk stories of
the Yakkha is comparatively low in the speaker community, with an increase in knowledge among
the elderly speakers. What is still vividly known and practised is a ri tradition of folk songs.
e traditional occupation of the Yakkha is farming and cale breeding. Another means of
subsistence is shing and hunting. e main crops are maize, rice and millet. A typical household
has pigs, bualos, ien and goats. e pig is also a major feature in the ritual design, as a
sacrice to the ancestors. e Yakkha follow their own religion, in whi the worship of the
ancestors plays a major role. Based on the clan division, ritual specialists called Mahawa,
Chamwa or Bijuwa have to undertake the ritual activities for ea family, on occasions like birth,
marriage and death, and annual festivities based on the crop cycle. e main festivals are Casowa
(Nep.: Udhauli) in autumn and Yuya (Nep. ubhauli) in spring (Kongren 2007a:102.).
In the rituals, a special spee register is used, that is very dierent from colloquial spee and
has its own vocabulary and a rigid structure and formalised style. is register is usually referred
to as Munthum in the Kiranti languages, and it has the function to invoke supernatural powers
and the spirits of the ancestors.
1.2 e language name

e language name is tracable to the Proto-Kiranti root *rok, whi is the Kiranti autonym. It
is not found in non-Kiranti languages of the Tibeto-Burman sto. e second syllable might be
traced ba to the Proto-Tibeto-Burman root for language, talk. e root *rok is also the source
for other Kiranti ethnonyms, su as for the Puma Rai (Southern Kiranti), who have the autonym
rokho. e sound ange of /r/ to /y/ is typical for Eastern Kiranti, to whi Yakkha belongs. e
neighbouring groups Lohorung, Yamphe and Yamphu also call their languages Yakkhaba1 , as it is
pointed out in van Driem (1994).
1.3 Earlier work on Yakkha language and culture

e material on Yakkha so far available is scarce. e oldest source on Yakkha that I could nd
was a apter in the Linguistic Survey of India, with a brief introduction and some Yakkha texts
that were collected in Darjeeling (Grierson 1909). More recent works on the language are the
Yakkha-Nepali-English dictionary (Kongren 2007b), a synoptic glossary (Winter et al. 1996) and
an article about the inectional morphology (van Driem 1994), whi is unfortunately based on
one paradigm only. Resear on cultural and political aspects was undertaken by Subba (1999)
und Russell (2007, 2004, 1997).
As the language use is in decline, and as there are already whole Yakkha villages that have no
single Yakkha speaker, Yakkha must be considered endangered. In the recent past, with ethnic consciousness rising in post-monary Nepal, several steps were taken to preserve the language and
the cultural knowledge that it contains, both by the Kirant Yakkha Chumma (Indigenous Peoples
Yakkha Organization) and by individuals of the broader Yakkha community in Nepal and India.
Among these we can nd the implementation of Yakkha language textbooks (Jimi et al. 2009) and
a curriculum for rst and second class, an ethnographic introduction avaible in English and Nepali
(Kongren 2007a), a dictionary (Kongren 2007b), a collection of poems (Dewan et al. 2002 (VS 2059)
and a collection of thematically ordered wordlists and articles on the Yakkha traditions (Linkha
and Dewan 2005 (VS 2064). For a more detailed bibliography of the works on Yakkha that were
published in Nepali the interested reader is referred to Rapaa et al. (2008). All these publications
and recordings were very helpful sources for my own work. Furthermore, the Global Recordings
1
-ba is a nominaliser.
1.4 Genetic Aliation

Network oers several recordings of biblical content in Yakkha (hp://globalrecordings.net/). Although the nature of a translated text diers highly from natural spee and must be treated with
caution, the recordings provided useful baground for my own data.
1.4 Genetic Aliation

Yakkha belongs to the eastern bran of the Kiranti languages, together with Limbu and several
Rai languages. One feature of Eastern (and Central) Kiranti is the merger of voiced and voiceless
obstruents. Concerning voice, Eastern and Central Kiranti build one group, concerning other
features, they are dierent. A typical feature of Eastern Kiranti is the sound ange of proto */r/
and */R/ to /y/ (hence yak in Eastern and rak ~ rok in others). Sometimes, the resulting /y/ also
elided. e tree in gure 1.1 represents just one possible scenario of the genetic aliation of the
Kiranti languages, based on the regular sound anges in initials (Biel 2008a).
Kiranti
Western
(*C C)
Chaurasiya
(* s):
Jero
Wambule
Central-Eastern
(*voiced voiceless;
k,*c kh, )
Midwestern
(*p,*t b,d):
ulung
Koyu
Northwestern: Upper Dhko:

Bahing/Bayung
Khaling
Hayu
Dumi
Sunuwar/Koc
Central
(*p,*t b,d)
Khambu
(* g, hr):
Kulung
Nairing
Sampang
Sam
Greater Eastern
(*p,*t ph, th)
Southern:
Camling
Upper Aru
Eastern
Bantawa (PE *ph,*th ):
(Yakkha-Limbu)
?Dungmali
Lohorung
Puma
Yamphu
Limbu
Mewahang Greater Yakkha :
( s;
Yakkha
,r y)
?Mugali
Chlng
Chintang
Athpare
Belhare (PE *th )
Figure 1.1: Yakkha within the Kiranti Language Family, citing Biel (2008a)
e status of Yakkha within Eastern Kiranti is controversial. e proto-pregloalized consonants became either aspirated consonants or zero: */c/ > //, */p/ > /ph/, */t/ > //, while in
Central Kiranti, they became voiced stops. e nontrivial question of the phonemic status of
rhotics and laterals in Yakkha is treated in apter 2 on the phonology. e table in gure 1.2 is
based on data from van Driem (1993, 1987), Biel et al. (in prep.), Kongren (2007b) and own data.
Supposedly there are several dialects, even one spoken in Ilam, but no detailed dialectal study
was undertaken yet for Yakkha.
1 Introduction
Proto-Kiranti
*/d/
*/j/
*/b/
*/r/
*/r/
*/R/
*//
*/p/
*/t/
*/t/
Dumi (west)
den
ju
bhii
rep
rodi
rm
pu
t
Puma (central)
ten
ca
poo
rep
rodu
rum
apdbu
du
dok
Yakkha (east)
ten
ca
pik
ep
yakthu
yum
ep
phu
u
ak
Limbu (east)
tn
ca
pit
yep
yak
yum
sap
phu
thu
thak
Gloss
village
eat
cow
stand
Kiranti Autonym
salt
write
ower
drink
loom
Figure 1.2: Comparison of sound anges in Eastern, Central and Western Kiranti
10
2 Phonology
2.1 Orthography
is apter deals with the consonant and vowel inventory and allophonic rules. At rst I will
briey lay out the orthography used here. I found it most practical for my purposes to follow the
common orthography used in other grammars on Kiranti languages. It neither represents the phonetic level, because it is unpracticable to note down ea and every phonetic dierence, especially
as the analysis of the sound system is not the major goal of this work. Nor does it represent the
phonemic level, because the reader would have to bear in mind the whole section on allophonic
variance to gure out how the words are pronounced. A second reason not to represent the phonemic level is that there are too many dialectal and possibly individual deviations and exceptions to
the allophonic rules, so that this information would be lost in a phonemic representation. What I
employ here is what could be called the allophonic level, including allophones that are the result
of the voicing rule for instance.
e orthography used here looks as follows: e symbol [y] is used for the palatal approximant
(IPA: j), [w] is used for the bilabial approximant, [c] is used for the alveolar fricative (IPA: ts),
and [] stands for its aspirated counterpart (IPA: ts). In the same way, [ph], [th], [kh] and [wh]
represent aspirated consonants. e signs [] and [] are used for retroex (or postalveolar) stops,
and [] is used for the velar nasal.
2.2 Phoneme inventory and allophonic rules

Charts with the vowel and consonant inventories are provided in Figures 2.1 and 2.3 below. Phones
in braets occur only in loan words, or they are questionable, as it is the case for the gloal stop.
Length dierences are not contrastive, they are found only on the allophonic level, as for instance
in kama fall vs. pagyam old man/husband. ere are ve vowels, and there are no central
vowels like // or // as in other Kiranti languages. Minimal pairs for the vowel distinctions are
provided in gure 2.2.
i
u
e
o
a
Figure 2.1: Yakkha vowel phonemes
ree diphtongs /ai/, /ui/ and /o/ were found, occuring marginally, as in waikapu (a tree,
bot. sauraja nepaulensis), waitna exists, uimala steep descent and osipma feel shy. e
nasalised diphtong in the last example is a stem variation of the original stem /ont/. e sequences
/ai/ and /ya/ are sometimes reduced to [e], as in [wetna] /waitna/ it exists, [khe] /khya/ go or
the nonpast morpheme [me] /mya/. Furthermore, the vowels of open stems or of stems ending in
a gloal stop can be reduced to glides or even be elided if suxes are following. us, /ca-wa/
11
2 Phonology
Phonemes
/e/ vs. /i/
/e/ vs. /a/
/o/ vs. /u/
/o/ vs. /a/
/u/ vs. /i/
nema
tema
okma
thokma
ukma
Examples
sow seed
nima
lay on one side tama
shriek
ukma
spit
thakma
bring down
ikma
know, see
come
bring down
weigh
collect
Figure 2.2: Minimal pairs for vowels
and /cu-wa/ can be pronounced [cwa], /khe-a/ can be reduced to [khya], /pi-a/ can be reduced
to [pya], likewise /si-a/ can be reduced to [sya]. is may result in homophonous verb forms, for
instance the bare stem khya and its 3s.PST/IMP form khya, whi is underlying /khya-a/.
Stops
asp.
Stops voiced
asp.
Aricates
Ar. asp.
Aricates
Ar. asp.
Fricatives
Nasals
Vibrants
Laterals
Glides
Glides asp.
bilabial
p
ph
(b)
(bh)
dental
t
th
(d)
(dh)
alveolar
()
(h)
()
(h)
palatal
velar
k
kh
(g)
(gh)
gloal
()
(j)
(jh)
m
s
n
r
l
w
wh
Figure 2.3: Yakkha consonant phonemes

A typical feature of Eastern Kiranti is the merger of voiced and voiceless stops and aricates.
Voiced stops and aricates only occur allophonically. Due to phonological rules of voicing and
reduction, sequences like /-ka-ha/ may become [-gha], but voiced aspirated stops are not found
in phonemic status. Aspirated nasals, as they are found in other Kiranti languages, are absent in
Yakkha. e glide /w/ however has an aspirated counterpart /wh/, standing in phonemic opposition (cf. Table 2.4).
As for rhotics in eastern Kiranti, according to van Driem (1990), [l] and [r] have a complementary distribution: [l] occurs word-initially and syllable-initially aer stops, and [r] occurs between
vowels and in Cr-clusters. At the time of this claim however, only Limbu data was available. For
Yakkha, the situation is a bit dierent. e rhotic /r/ does not appear word-initially, as there are
words like lok anger, but no words starting with /r/, except for the ancient ritual word ren that
was translated as homeland, village, place of walking, playing, seeing, and Nepli loans. A place
where only /r/ occurs is in syllable-initial consonant clusters, for instance in ikhrum-ikhrum on
all sides, phuphruwa milk and makhruna bla, and suthrukpa a special kind of knot. On the
other hand, only /l/ appears syllable-initially aer closed syllables, as in khala in this way, and
pulicakli being many. Both /r/ and /l/ may occur in intervocalic position, as in ulippa old, sala
language and tarokma start or mora big. e two sounds /l/ and /r/ are distributed over complementary environments, but on the other hand, they also occur in the same environment. e
problem with the explanation of this distribution is that the dierent syllabications (e.g. .kr vs.
k.l) were established just on acoustic impressions and need independent evidence. It is not clear
12

Phonemes
/k/ vs. /kh/
/p/ vs. /ph/
/t/ vs. /th/
/c/ vs. //
/y/ vs. /w/
/y/ vs. /l/
/w/ vs. /wh/
/s/ vs. /h/
/k/ vs. //
// vs. /m/
// vs. /n/
/m/ vs. /n/
kepma
pakna
tumma
tapma
cikma
cimma
yapma
yamma
yapma
wapma
wama
sima
somma
pekma
okma
toma
tuma
-a
muma
Examples
come up
khepma
young guy
phak
understand
thumma
receive
thapma
age, ripen
ikma
tea
imma
become rough wapma
disturb
wamma
become rough lapma
put on clothes whapma
curve, bend
whama
die
hima
stroke gently
homma
break
pema
shriek
oma
agree
tomma
pour
tumma
(ERG)
na
uproot
numa
go
pig
tie
winnow
measure
ask
put on clothes
trap
accuse, blame
wash clothes
boil
spread
t into
peel
aa
stand up
understand
(NMLZ.s)
heal, get well
Figure 2.4: Minimal pairs for consonants
yet whether two dierent sounds /r/ and /l/ trigger the dierent syllabications or whether it is
the other way round, and dierent syllabication results in dierent realisations of one phoneme.
Despite these unsolved questions, the synronic occurence of /r/ and /l/ in intervocalic environment justies calling them two phonemes. With rhotics, metathesis can be found in allophonic
variation, as in tepruki ~ tepurki ea. is might be due to Nepali inuence (B. Biel, p.c.).
e stops /p/, /t/ and /k/ can all be unreleased or neutralised to a gloal stop syllable-nally.
e gloal stop is also prothesised before every vowel-initial syllable, to have a minimal onset.
However, it is not clear yet whether it also stands in phonemic contrast to the other stops. Morphophonological processes can be found in abundance in Yakkha, and more data is needed to
clearly distinguish all the underlying forms.
Unaspirated stops and the aricate, but not the sibilant, may undergo postvocalic and postnasal
voicing. e voiceless counterparts can be heard in this context as well, but far less than the voiced
stops. is rule applies to both lexical stems and inectional morphemes, with the exception of
the morpheme -ci (dual or nonsingular, depending on morphological context). e only instance
of voiced -ci is in the second person dual pronoun ncita ~ njida. In example (1), we can see
how the locative case morpheme -pe gets voiced, but -ci does not. e domain of the voicing rule
is apparently the word.
(1)
homa-ci-be
river-ns-LOC
in the rivers
Yakkha has several nasal prexes that are realised as homorganic prenasalisation, i.e. they do
not constitute syllables of their own. e nasal prexes also trigger voicing stem-initially (see
example (2)). e rule only applies at stem boundaries, so that kipa tiger or kucuma dog will
not be voiced. Complex words, su as compound verbs and nouns are also exempt from this
rule, for instance koncama go for a walk. e voicing is however active in the grammatical
13
2 Phonology
compounding of verbs (see example (3), cf. apter 6 on the structure and basic semantics of the
compound verbs). Several morphemes are realised by the above mentioned nasal prexes. In the
verbal domain, they code agreement with third person plural (A and S arguments) and negation.
In the nominal domain, there is the possessive prex for second person singular. In spatial and
temporal adverbs, the prex codes distal meaning.
(2)
a.
m-bi-a-na
3pA-give-PST-NEG
b.
ey did not give it to us. (underlying: /pi/)

n-ja-wa--an=na
NEG-eat-ASP-1sA-NEG=NMLZ.s
c.
I do not eat it. (underlying: /ca/)

m-ba
2sPOSS-house
your house (underlying: /pa/)

One more exception to the voicing rule has to be admied, that is shown in (3a). Stem-nal /t/
stays voiceless even before a vowel, so that for instance the imperatives et-u Hunt sh! and phatu Help! do not undergo voicing. If the stem ends in a nasal plus /t/ however, stem-nal voicing
will apply, as in (3b). To sum up, we can say that there are two exceptions to the voicing rule, one
being the morpheme -ci, the other being stem-nal /t/. e reason for the second exception might
be that the stem-nal /t/ in these verbs has its origin in an augment and was reanalysed as part
of the stem (cf. section 5.1 on stem formation). In addition to the exceptions, the rule is subject to
considerable dialectal and even individual variation. us, the orthography adapted here reects
the voicing as it was recorded, and not the underlying voiceless forms. e bilabial stop can
also ange to a bilabial glide /w/, via the voiced stop /b/, as for instance awa is an alternative
pronounciation for aba (underlying /ap-a/) Come!.
(3)
a.
eko ceya lut-u-bi-ci

one story tell-3P-BEN-ns
b.
She told them a story. (underlying /pi/)

em end-u-ga=na?
song apply-3P-2=NMLZ.s
Did you put on music? (underlying: /ent/)

Another morphophonological process is connected to the nasal prexes. As they are unspecied
for the place of articulation, they assimilate in place to the subsequent consonant, thus yielding
prenasalisation. e examples above in (2) also illustrate this rule, and three more examples are
provided in (4). If the verb starts in a vowel, the default option is a velar nasal, as exemplied in
(4d). e nasal prex, if aaed to a nasal-initial stem, yields an initial nasal geminate, shown
in (4e).
(4)
a.
-khya-n=na
NEG-go-NEG=NMLZ.s
b.
He did not go.

n-ya-ci
wai=ya-ci?
2sPOSS-ild-ns exist=NMLZ.ns-ns
c.
Do you have ildren?

n-da-ya-ci
3pS-come-PST-ns
d.
ey came.
mamha wa-ci-a
big
14
mimiyah wa-ci
ien-ns-ERG small
-og-wa-ci=ha
ien-ns 3pA-pe-NPST-ns=NMLZ.ns
e.
e big ien pe the small ien.

m-ma
2sPOSS-mother
your mother
Plosives also undergo several assimilation processes. A stem-nal plosive, if followed by a morpheme starting in a sibilant, will also become a sibilant. In other words: p,t,k s/_s. Examples
are provided in (5). ere are stems that have a coronal augment aaed, whi is visible only
when the stem is followed by vowels, due to a restriction on complex consonant clusters. As for
these stems, they ban the assimilation to a sibilant, as shown in example (5d) 1 .
(5)
a.
/nip-sa/
[nissa]
count-SIM.CVB
b.
counting
/sop-sa/
[sossa]
wat-SIM.CVB
c.
wating
/et-se/
[esse]
hunt.sh-PURP.CVB
d.
to hunt sh
/ept-se/
[epse]
write-PURP.CVB
to write
Stem-nal /t/ will assimilate to place of articulation if it is followed by morphemes starting in
/m/, i.e. t p/_m. e innitive of et hunt sh, thus, is epma, for phat help it is phapma. To
nd out the stem forms, the inected forms are always needed, for instance the imperative etu for
epma or phatu for phapma.
Some sounds and clusters also assimilate in nasal quality to a subsequent nasal. e verb understand for instance, with the underlying stem /tunt/ (realized as [tund]), has the innitive tumma.
e inected form tum-me-nen=na (stem-NPST-1>2=NMLZ.s) could either mean I understand
you or I meet you, with the underlying stems /tunt/ and /tups/ respectively. e innitive is
tumma for both. Not all stems undergo the nasal assimilation, and the inectional morphology
seems to behave dierent from the innitive sux -ma. Some verbs keep a stem-nal stop in the
innitive, but assimilate to a nasal when being inected. As a rule of thumb, verb stems without
augments and those ending in a stop-t-cluster (e.g. -pt, -kt), retain the stop in the innitive, but
stems ending in nasal-t or stop-s (e.g. -nt, -mt, -ps, -ks) ange their coda to a corresponding nasal.
e variation of stem-nal stop-s cluster and nasal (e.g. -ks/-, -ps/-m) however is also a robust
paern of stem variation in Kiranti and therefore is not neccessarily to assimilation. e dialectal
and individual variation is also true for the application of the assimilation rules discussed here, so
that the orthography reects the forms that were recorded, not the underlying structure.
See also Chapter 5.1 on stem formation
15
2 Phonology
16
3 Parts-of-spee
Yakkha distinguishes nouns, verbs, adjectives and adverbs in the open class of parts-of-spee. In
the closed class, there are dierent kinds of pronouns, adpositions, conjunctions, and one classier
for human reference. Furthermore, there are many particles that are used for information structure.
Yakkha nouns inect for number, case and possession. As is usual for Kiranti and generally
Tibeto-Burman, there is no gender system. e markers -ma for female reference and -pa for male
reference do exist. is is particularly pronounced in the kinship system, as for instance isapa
means nephew, while isama refers to niece. Sometimes, this corelation is also found in the
beginning of words refering to kinship relations, su as in pagyam husband, magyam wife,
(a-)pum grandfather and (a-)mum grandmother1 . As for the etymology, probably there used
to be some stem that hosted the markers and further morphology, and got lost by the time. In
some nouns, one of the two markers got lexicalised and the derived nouns simply have neutral
reference, as for instance kipa means tiger and kucuma denotes dog, regardless of their sex.
Many nouns exhibit the marker -wa. Etymologically, there are competing stories. It could be
related to the Tibeto-Burman nominaliser -pa, as in pelewa lightning, phawa pig fodder
(from phak pig), phetawa shaman, healer (pheta means turban, the typical clothing for healers). But in words su as miwa tear (from mik eye) it could be related to the root for water
wa. Also terms for birds, insects and plants, su as phamiyuwa (an insect) are built with -wa,
whi could be related to the Kiranti root for bird, whi is also wa (in Yakkha now: nwak). Furthermore, there are also neologisms like phaniwa constitution.
Yakkha verbs, as Kiranti verbs in general, are aracterised by a complex interplay of person and
number markers that cannot always be identied straightforwardly. e verbs show agreement
with S, A and P, and in ditransitive scenarios, the agreement is aligned with the primary object,
that is, the goal or recipient. e verbs furthermore inect for tense, aspect, mood and polarity.
Most verbs have two stem forms that are triggered by the phonological context. Generally, verb
stems are monosyllabic. Polysyllabic verb stems, su as suncama it, incama sell, yuncama
smile, koncama walk and osipma feel shy are compounds.
Some aspectual categories, Aktionsarten and valency anging derivations su as benefactive
and causative are constructed with a compound verb, a typical feature of South Asia. It consists of
two verb stems of whi the second stem has grammaticalised to a marker of the above mentioned
categories. e morphology and semantics of the compound verbs will be discussed in detail in
Chapter 6.
Many properties that are expressed via adjectives in other languages, are covered by verbs in
Yakkha. ey can be used as predicates or aributively. A nominaliser2 that agrees in number
with the head noun is aaed to the verb stem. Examples are hana/haha spicy, huna/huya
burnt or cina/ciha cold, with =na as the singular form of the nominaliser, and =ha/=ya as
the plural form. Other adjectives are also tracable to verbs, but without a nominaliser aaed.
Some of them look like frozen forms of inected verbs, su as attu fat, from apma spread.
Others look like bare verb roots, su as em clear from emma be clear. Rare examples
1
e prex a- in braets is a rst person singular possessive prex. Kin terms are used with possessive prexes
obligatorily, a grammatical reex of the concept of inalienable possession. See also apter 4.2.2 on possessive
pronouns and prexes.
2
Alternative names in other descriptions of Tibeto-Burman languages are aributiviser or article, but the multiple
functions of this particle are best covered by the term nominaliser.
17
3 Parts-of-spee
of adjectives not tracable to a verb are ikhrum round and lukluk short. Comparison is not
a dening aracteristic of adjectives in Yakkha, as the same construction also applies to verbs,
yielding expressions like I see/walk beer/faster than you, where no degree adverb as in English
is needed (cf. apter 7.3 on comparison).
We can nd a range of adverbs in Yakkha, thinking of adverbs as grounding or modifying a verbal event in time, space, manner or degree. ere are temporal adverbs asen yesterday, locational
and directional adverbs, su as tutunne far up above, those that are exclusively directional like
yokhala towards over there. Some adverbs are derived by adding the comitative case -nu to a
nominal, verbal or adjectival root, as in (1). e domain of ideophonic adjectives and adverbs is
also very ri in Yakkha (cf. section 7.4).
(1)
a.
hasu-nu
monary-COM
b.
royal
ucun-nu
nice-COM
c.
well
khik-nu
be.bier-COM
bier
Compounds exist in nouns, verbs, adjectives and adverbs. Naturally, the etymology is not
always transparent any more. In all domains, Nepali borrowings are frequent.
Yakkha has one identicational copula for armative contexts om, and two negative copulas
manna for existential, and menna for identicational use (cf. apter 8).
In the eld of pronouns, personal pronouns, demonstrative pronouns and interrogative pronouns
can be found. Reexive and reciprocal pronouns do not exist in Yakkha, because these valency
modications are expressed by verbal morphology. As relativisation is done via nominalisation,
Yakkha also las relative pronouns. ere is however the corelative construction, where question
pronouns and demonstrative pronouns may function as relative pronouns (cf. section 10).
Yakkha has one native classier pa for human reference. For example, uci hippa these two
people. Other classiers, su as wora piece are loans from Nepali, as in bis wora khibak twenty
ropes. Native numerals are found upto number ve: i, hic, sum, thum and a, but the Nepali
numerals are also used. Yakkha uses relational nouns in a possessive construction to express
spatial relations (cf. apter 4.3 on the genitive and possessive constructions). ere are some
adpositions that were taken over from Nepali, like dekhi since, samma until, anusar according
to, lagi for, the last example with a loan translation of the genitive case that it requires.
ere are some interjections, su as hoi Enough!, om Yes, menna/manna No.
Furthermore, there are several conjunctions and particles for information structure, that will be
dealt with in their respective apters and sections.
18
4 Nominal Morphology
4.1 Number
Yakkha distinguishes singular, dual and plural in the verbal domain and in pronouns, and just
singular and nonsingular in nouns and third person pronouns. e nonsingular marker on nouns
is -ci. e sux applies to whole NP, aaing always to the nal component. erefore, modifying material does not agree in number. e number marking on nouns is not obligatory, and,
impressionistically, it is more frequent in nouns denoting animates. Nonsingular marking can be
interpreted associatively, as it is done in other Kiranti languages and in Nepali also (cf. example
(1)).1 e associative semantics are a feature that has reexes in other parts of the morphosyntax
as well, for instance in the exible agreement (cf. section 5.3).
(1)
a.
a-koma-ci-nu
1sPOSS-aunt-ns-COM
b.
with my auntie and her people

Lila didi-ci
L.
elder.sister-ns
sister Lila and her folks
4.2 Pronouns
4.2.1 Personal pronouns
e distinctions found in the personal pronouns are not as ne-grained as in the verbal inection.
An overview of the personal pronouns is provided together with the possessive pronouns in gure
4.1 below. e rst and second person pronouns distinguish singular, dual and plural number. e
morpheme -ci conveys a dual meaning in the rst and second person pronouns, as opposed to -ni
for plural. In the third person, -ci simply conveys a nonsingular meaning, but the verbal inection
distinguishes dual number also in third person (cf. apter 5.2). e rst person pronouns do not
distinguish inclusive/exclusive, but this distinction occurs in the possessive pronouns and in the
verbal inection. Hence, both inected verb forms in (2) appear with the same pronoun. Probably
the diaronic base of the second person pronoun is the stem-initial /n/, to whi the number
morphemes -ci and -ni were aaed. Finally, -da presumably has its origin in a topic particle
that was aaed to the pronouns and got lexicalised.2
(2)
a.
kani khe-i-a
we.pl go-1/2p-excl
b.
We went. (adressee excluded)

kani khe-i=ha
we.pl go-1/2p=NMLZ.ns
We went. (adressee included)

1
In Yakkha, kinship terms are used regularly to adress people, rather than their proper names. ey need not correspond to actual kinship relations. Strangers will also be adressed as sister, auntie, brother etc., depending on
their age and status.
2
Parallel etymologies of second person pronouns and topic particles can at least be found in the Puma (also Kiranti)
second person pronoun khan-na
19
1s
1di
1de
1pi
1pe
2s
2d
2p
3s
3ns
personal pronoun
ka
kanci
kanci
kani
kani
nda
njita
nnita
u
uci
possessive pronoun
akka pa
enciga pa
anciga pa
ega pa
aniga pa
ga pa
njiga pa
nniga pa
ukka pa
uciga pa
prex
apa
encipa
ancipa
emba ()
anipa
mba
njipa
nnipa
upa
ucipa
pronoun plus prex

akka apa
nga mpa
ukka upa
Figure 4.1: Personal and possessive pronouns (illustrated by the head noun pa house)
4.2.2 Possessive pronouns

e possessive pronouns consist of the personal pronouns to whi a genitive sux is added, with
slight irregularities. ey distinguish inclusive and exclusive, a category that is absent in the
personal pronouns. e rst person possessive pronouns are aracterized by the loss of the initial
/k/. e inclusive forms have no parallel in the personal pronouns. e third singular pronoun
shows an unusual assimilation of the velar nasal to a stop. e expected form would be uga,
but instead we have ukka here, perhaps to distinguish the possessive pronoun beer from the
ergative-marked third person pronoun u-a.
e possessive prexes mark the possessum, i.e. the head, indexing person and number of the
possessor. eir form is similar to the possessive pronouns, whi suggests that both have the
same etymology. e nasal in the inclusive prex eN- and the nasal in the second person prex
N- assimilate in place to the subsequent consonant. ere are dierent structural options for a
possessive phrase. One could either have the possessive pronoun plus the head noun, or simply
the head noun with a possessive prex. In the singular, and only there, the possessive pronoun
can occur together with the possessive prex. is option was rejected for the nonsingular numbers, with the explanation of sounding awkward and redundant. In the singular however, this is
not perceived ungrammatical. e heterogenous behaviour of the pronouns can be interpreted as
an ongoing grammaticalization process of the pronouns to prexes. Expectedly, as the singular
pronouns are shorter compared to the nonsingular pronouns, it is easier for them to become grammaticalised. Currently, the prexes are in any number or person just an alternative option to the
pronouns. is might ange over the time, and they might become obligatory, a development not
unknown from other Kiranti languages, as for instance Puma (Sharma (Gautam) et al. 2005). Table
4.1 provides an overview of the pronouns and prexes, exemplied by possessive phrases with the
head noun pa house. Another irregularity that we can see here is the only partial application
of the voicing rule that was introduced in apter 2.2.
Another remark on the possessive prexes is in order. e category of inherent/obligatory possession exists, where certain concepts cannot be expressed without belonging to another entity.
is is true for kinship terminology and, beyond kinship, for body parts and parts of plants. As for
kinship terms, the rst person singular possessive prex is the default option. Terms like a-mum
grandmother, a-pum grandfather, a-na elder sister and many more are never used without
possessive prexes. Literally, they mean my grandmother and so on.
It has to be noted that the terms for family relations acquired by marriage do not fall within
the domain of obligatory possession, reecting their non-consanguineous status (cf. example (3)).
While the default option (e.g. for vocatives) for kin terms is the rst person prex, for other terms
it is the third person singular, as for instance u-ti thorn.
20
4.2 Pronouns
(3)
taba
hetne tas-wa-ga=na
male.in-law where arrive-NPST-2=NMLZ.s
Where will (your) husband arrive?
4.2.3 Demonstrative pronouns

e demonstratives pose some questions that could not be answered yet. e following section
shall only present the data and rst ideas on a possible analysis. e form and semantics of
the demonstrative pronouns correspond to the nominalisers (cf. apter 10.1.1). For proximal
(and probably also neutral) reference, there is na for singular and kha-ci for nonsingular and
uncountables. Both pronouns can appear aributively (example (4a)), in predicative function and
independently as head of an NP, (see example (4b)). In non-modifying position, the nonsingular
pronoun always appears with the nonsingular sux, yielding khaci, the rst naturally may not.
Independent kha without plural sux, as in kha-ga pa (intended: their house) was rejected as
ungrammatical. However, due to the plural being a phrasal sux, phrases like *khaci yapmici
are ungrammatical. Instead, kha yapmici is used (cf example (4d)). e following examples shall
illustrate the dierent positions in the NP. In (4a) we have the demonstrative in modifying position,
in (4b) and (4c) we have the pronouns as heads of NPs.
(4)
a.
na
toba imin et-u-ga=na?
DEM beer
b.
how like-3P-2=NMLZ.s
How do you like this beer?

kha-ci imin et-u-g=ha?
DEM-ns how like-3P-2=NMLZ.ns
c.
How do you like these?

na yond-a-bo
yenda sund-i-mya=bu
DEM come.out-PST-COND beer
d.
become.sour-3P-NPST=REP
When this one (the lizard) comes out, the beer gets sour, they say.
kha toba-ci khumdu=ha-ci
DEM beer-ns tasty=NMLZ.ns-ns
ese beers are tasty./e beers are tasty.

e morpheme kha has another function that is not connected to the demonstrative, but it also
carries plural semantics. It may be used to express a plurality of items, without specifying the exact
number, as for instance khadalci heap of lentils or khaphuci bun of owers. e oice of the
right interpretation seems to depend on the context, from the data available so far, but this issue
cannot be explained satisfactorily yet. e second use also points towards a general unmarked
use of na and kha. e addition of further morphology to yield a distal meaning supports this
assumption. Whether these bases code proximal or neutral deixis with regard to distance, the
addition of a homorganic nasal prex turns them into distal demonstratives, translatable with
that and those. e examples in (5) shall illustrate this. e proximal-distal-distinction is also
present in some adverbs of manner, for instance the manner adverbs khala in this way, and
khala in that way, and in the time adverbs khatni this time, now, and khatni that time,
then.
(5)
a.
na-ga
pa
this-GEN house
b.
his (this ones) house (Nep.: yasko ghar)

nna-ga pa
that-GEN house
c.
his (that ones) house(Nep.: usko ghar)

kha-ci-ga
pa
these-ns-GEN house
21
d.
their (these) house (Nep.: yiniharuko ghar)

kha-ci-ga pa
those-ns-GEN house
their (those) house (Nep.: uniharuko ghar)

e etymology of kha can most probably be traced ba to Proto-Kiranti *khol all/ everybody,
that has developed to dierent markers with plural semantics throughout the Kiranti languages:
quantiers (Yakkha khak ~ ghak , Khulung khl, ulung khole) all; antipassive markers
(Puma), rst person patient markers (Puma, Chintang, Camling) (Biel and Gaenszle 2007).
ere is another set of demonstratives that refer to dierent directions and locations with respect
to the speaker. ese are khe for proximal, yo for distal reference on the same level of altitude, to
for up and mo for down. ese roots may form demonstratives or spatial adverbs, depending
on the derivational morphology that is added to them (see also apter 7). e nominaliser (=na
for singular, and =ha ~ =ya for nonsingular reference) turns them into demonstratives. Examples
are provided below. ey may also occur in predicative function (see example (6c)).
(6)
a.
khe=na
toba
PROX-NMLZ.s. beer
b.
this beer
to=ya
pik-ci
UP=NMLZ.ns cow-ns
c.
the cows up there

-ga
toba yo=na,
2sPOSS-GEN beer
ak-ka
toba khe=na.
DIST-NMLZ.s, 1sPOSS-GEN beer
PROX-NMLZ.s
Your beer is that one, my beer is this one.
4.2.4 Interrogative pronouns and other interrogatives

e interrogative pronouns and adverbs are mainly built by the combination of the unspecied
interrogative root i with case suxes, postpositions and the nominaliser. ere is a second interrogative root, namely het ~ he, whi may basically combine with the same morphological
material as the rst-mentioned root, though sometimes with dierent semantic results, as i=na
means what, but het=na yields whi. All interrogatives found so far are listed in Table 4.2
below.
i
i/ ina/ iya what
ibe where
ibeni when
ipha where from
imin how
ikhi how mu/how many
isa who
ija why
het
hetna whi
hetne where
hetni when
hetna where from
Figure 4.2: e interrogative pronouns and other interrogatives

e root i usually occurs with further morphology. Depending on the number of the items in
question, it will either take =na, the nominaliser for singular reference, or =ya ~ =ha, the nominaliser for plural and uncountables. For example, food is expected to consist of several dierent
22
4.3 Case system

items, and will be asked for with the plural form. If however not an item but en event is the focus
of the question, the bare interrogative root, without further morphology, is used (see (7)).
(7)
a.
i=ya
ca-ma
what=NMLZ.ns eat-INF
b.
What to eat?
i
leks-a?
what happen-PST
What happened?
e root can also combine with the genitive -ga and the loan postposition lgi for from Nepali,
to ask for purpose, as shown in example (8).
(8)
i-ga-lagi
ta-ya-ga=na?
what-GEN-for come-PST-2=NMLZ.s
What did you come for?

As for the pronouns with the root het, their etymology is quite transparent. Hetna whi consists of the root het plus the singular nominaliser =na. Hetne where and hetna where from
are built via the addition of the locative and the ablative case suxes respectively. e bilabials
in the locative -be and in the ablative -pa have anged to a nasal, adapting the place of articulation of the preceding /t/ in het. e ange in quality to a nasal however cannot be explained
straightforwardly. e particle ni, found in hetni and ibeni means while, and is also used
in clause linkage (cf. apter 14.2). e origin of the other particles -min, -khi, -sa and -ja is
not transparent. ey do not exist independently.
Some of the interrogative pronouns can be duplicated to yield unspecic reference, like hetni
hetni sometimes, or iya iya some things. Some more examples for the interrogative pronouns
are provided in (9). In (9f) we can see that isa may also take the plural sux if the identity of
more than one person is in question.
(9)
a.
hetna
ta-ya-ka=na,
mamu?
where.from come-PST-2=NMLZ.s girl
b.
Where do you come from, daughter?

hetni ta-ya-ga=na?
when come-PST-2=NMLZ.s
c.
When did you arrive?

hetne wa-me-ka=na,
mamu?
where live-NPST-2=NMLZ.s girl
d.
Where do you live, daughter?

na ten-be
ikhi
wa-me-ka=na?
e.
How long will you stay in this village?

ani-ga
ten imin et-u-ga=na?
this village-LOC how.mu live-NPST-2=NMLZ.s
1pe.POSS-GEN village how have.opinion.about-3P-2=NMLZ.s
f.
How do you like our village?

kha-yapmi-ci
isa-ci?
these/unspec.pl-person-ns who-ns
Who are these people/the group of people?
4.3 Case system

Yakkha has seven cases, namely the nominative, the ergative/instrumental, the genitive, the locative, the ablative, the allative and the comitative (cf. the table in Figure 4.3).
23
case
nominative
ergative/instrumental
genitive
locative
ablative
allative
comitative
marker
/
-a
-ga
-pe/ -ge
-pa/ -ga
-khala
-nu
Figure 4.3: Case markers

e unmarked nominative is used for subjects of intransitive verbs (S), objects of transitive verbs
(P) (for both theme (T) and goal (G) grammatical roles), and also for subjects of transitive verbs (A),
as long as they are represented by rst and second person pronouns. It is also used for predicate
nominals. Arguments in the A-role that are not represented by rst or second person pronouns
are marked by the ergative case -a. is also includes nouns with second person reference (for an
example cf. section 5.3 on exible agreement). Examples of nominative and ergative are provided
in (10) below.
(10)
a.
ka
khe-me-=na
1s[NOM] go-NPST-1s=NMLZ.s
b.
I go.
massina=ca
-und-wa-ci
small[NOM]=ADD 3pA-pull.out[3P]-NPST-ns
c.
(ey) also pull out the small (sh)

ka
nta
caklet
pi-me-nen=na
1s[NOM] 2s[NOM] sweet[NOM] give-NPST-1>2=NMLZ.s
d.
I give you the sweet.

u-a ka
mokt-wa-=na
3s-ERG 1s[NOM] beat[3sA]-NPST-1s=NMLZ.s
He beats me.
e ergative is also used to mark instruments or eectors, as shown in (11). Also the medium for
communication, i.e. the language will be marked with an instrumental (cf. (11b)). Some temporal
adverbs also take an ergative/instrumental marker, as in example (11c).
(11)
a.
lukhak-a la
stone-INS
b.
lukt-i=na
foot[NOM] bump.into[3s]-PST=NMLZ.s
Did you hit (your) foot at the stone? (in.class/exception what is -i here?)
e-ga ceya-
sarab pi-ci=a
1pe-GEN language-INS curse give[3pA,PST]-3nsP=NMLZ.ns
c.
(e sun) cursed them in our language.

khibela-a
thesedays-ERG
thesedays
e genitive case is realised by the sux -ka. It is used for possessive constructions, as in (12).
As mentioned in Chapter 4.2.2 on possessive pronouns, the possessum may carry the possessive
prex, as in (12a) and (12b).
(12)
a.
-ga
m-ba-ci
m-ma-ci
2sPOSS-GEN 2sPOSS.pr-father-ns 2sPOSS.pr-mother-ns
your parents
24
4.3 Case system

b.
isa-ga
u-cya?
who-GEN 3sPOSS-ild
c.
whose ild
ani-ga
likha-ga utpati
1pePOSS-GEN a_clan-GEN origin
the origin of our Linkha clan

Some local postpositions also require the genitive, as shown in (13a). Some of them are relational
nouns, used in a possessive construction, to whi a locative must be added, as in (13b) and (13c).
Example (13a) has no locative aaed to the head noun, whi is the reason for interpreting it as
a postposition.
(13)
a.
tebul-ga mobarik
table-GEN under
b.
under the table

si-ga u-sam-be
tree-GEN 3sPOSS-boom-LOC
c.
under the tree

tayar-ga u-yum-be
tyre-GEN 3sPOSS-side-LOC
next to the tyre

e teknonymic genitive -gu is the respectful way to adress or refer to elder people, instead of using
their names. ey are referred to as father or mother of their eldest ild. e more frequent option
is apparently the eldest son, but exceptions in favour of the eldest daughters name are possible.
Etymologically, the teknonymic genitive is the result of merging the genitive sux -ga with the
third person singular possessive prex u-. In case the ilds name does not end in a vowel, an
epenthetic element -e is added. In this construction, the initial consonant of the head noun is
geminated. Examples are provided in (14).
(14)
a.
Ram-e-gu-ppa
R.-EPEN-TEK.GEN-father
b.
Rams father
Bal-e-gu-ppa
B.-EPEN-TEK.GEN-father
c.
Bals father
Sita-gu-ppa
S.-TEK.GEN-father
Sitas father
Yakkha has only one general locative case -pe, that is used for location as well as for direction. is is exceptional for Kiranti languages, that usually exhibit four dierent locative cases,
distributed on a vertical axis. e neighbouring closely related Athpare language however also
exhibits only one locative -Ni (Ebert 1997). Examples are provided in (15), with (15d) exemplifying
directional use. According to the voicing rule, there is an allomorph -be, whi can become -we
in fast spee. It might be that the ritual language still exhibits the deictic locative distinction, but
this was not eed yet.
(15)
a.
khorek-pe cuwa
b.
ere is beer in the bowl.

u-thok-pe
to-me=na
bowl-LOC beer
3sPOSS-body-LOC t-NPST=NMLZ.s
It suits her.
25
c.
nwak-ka ohop-pe
bird-GEN nest-LOC
d.
in the nest of the bird

salle-we khem-me--ci-a=i
Salle-LOC go-NPST-e-d[1]-e=EMPH
We go to Salle.
Propriation also has to be expressed by constructing the possessor as a locative, illustrated in
(16).
(16)
e-ka-be
paisa manna
1diPOSS-GEN-LOC money NEG.EXIST
We do not have money.

ere is a second locative, used to express the notion at Xs place, whi is a loan translation
from Nepali. e morpheme -ge is a combination of the genitive -ga and the locative -pe. e
same combination of morphemes is found in Nepali (e.g. tapi-ko-m at your place, mero-m at
my place).
(17)
a.
isa-ge?
who-at
b.
at whose place?
dhanu-ge ta-ya-=na
D.-at
come-PST-1s=NMLZ.s
I came to Dhanus place.

e locative is not obligatory. In colloquial language, examples like (18) can be heard. It has to
be eed yet, whether this is generally allowed, or only with inherent locations or motion verbs.
(18)
tumok
wai-a=na
Tamaphok stay-e=NMLZ.s
I am in Tamaphok.
Reduplication in combination with the locative is used to convey several locations, in a continous movement with iterative stops, as shown in (19).
(19)
pa-pa-be
house-REDUP-LOC
from house to house

e ablative -pa denotes some movement away from a source. It has the following allomorphs,
some of whi are phonologically conditioned: -ba, -na, -nda. e voiced version may simply
be accounted to the voicing rule, as shown in (20). e last allomorph is found in connection with
spatial adverbs only, and might have a dierent etymology.
(20)
a.
mamli-ba
Mamling-ABL
b.
from Mamling
tumok-pa
Tamaphok-ABL
from Tamaphok
e allomorph -na seems to be an assimilation to the place of the preceding stop, but the ange
to a nasal cannot be explained straightforwardly. In parallel to the ablative, the locative -pe is also
realised as [-ne] aer the interrogative root het, to yield hetne where.
26
4.3 Case system

(21)
a.
het-na tai-ka=na,
mamu?
what-ABL come-2=NMLZ.s girl
b.
Where did you come from, girl?

jarman-ba ta-ya-=na
Germany-ABL come-PST-1s=NMLZ.s
I came from Germany.

e ablative does not distinguish dierent levels of altitude, at least according to the data that is
available so far. e form -pa is used also when the direction of the movement is upwards or
downwards.
(22)
a.
tumli-ba
Tumlingtar-ABL
b.
from Tumlingtar (lower altitude than deictic centre)

otemma-ba
plains-ABL
c.
from the Tarai (lower altitude)

himal-ba
Himalaya-ABL
from the Himalayas (higher altitude)

e alternative -nda was found in the following example.
(23)
a.
mo-nda kya-=na
below-ABL come.up[PST]-1s=NMLZ.s
I came up from below.

e older version of the ablative can be found thanks to the data provided in Grierson (1909).
In most cases it is transcribed as -bohu, as in (24).
(24)
nakhok barsa-bohu
so.many year-ABL
since so many years

In parallel to the second locative, there is also a second ablative -ga, used to express from Xs
place, as illustrated in (25). It has the same composite structure as the second locative.
(25)
dhanu-ga ta-ya-=na
D.-from
come-PST-1s=NMLZ.s
I came from Dhanus place.

ere is an allative case -khala, but it is rarely used, as the locative may also express movement
towards a goal. ere is a homophonous manner adverb khala, that means in this way/ like this.
(26)
wale-ba tumok-khala
Waleng-ABL Tamaphok-ALL
from Waleng to Tamaphok

Another case is the comitative -nu.
(27)
a.
nda-nu ka=ca lo-me-=na
b.
I will also come with you.

toba-nu n-do-me-n=na
2s-COM 1s=ADD come.along-NPST-1s=NMLZ.s
beer-COM NEG-agree-NPST-NEG=NMLZ.s
It (the ocolate) does not go with the beer.
27
e comitative also plays a role in the derivation of some adverbs out of adjectives, as shown
in (28) (for more examples cf. apter 7.1). Furthermore, the comitative is also found in clause
linkage (cf. apter 11.4).
(28)
a.
sua-nu et-u-=ha
b.
It tasted sour to me.

khumdu-nu na-ma
sour-COM feel-3P-e=NMLZ.ns
tasty-COM
smell-INF
to smell tasty
28
5 Verbal Morphology
5.1 Stem formation
Kiranti verb stems usually alternate between two stem forms. ere are various stem-nal alternations and augmentations, leading to consonant clusters that will be pronounced only before
vowels in the same phonological word, i.e. before vowel-initial sux-strings. Otherwise, the
unaugmented stem form will be used. Augmented stems end in the coronal obstruents -s or -t.
Some also end in -r whi might historically have come from -t as well.
e augments can be traced ba to Proto-Sino-Tibetan derivational suxes. ey got lexicalised and largely lost their morphological content. In Kiranti, some reexes of this old system
can still be found in valency-anging correspondences like in gure 5.1, but this is by no means
a regular synronic paern, and there are also many intransitive verbs with an augment. Some
examples of these correspondences in Yakkha are listed in Figure 5.1.
uks ~ u
hons ~ hom
poks ~ po
phops ~ phom
si
come down
be open
explode
be covered
die
ukt ~ uk
hont ~ hom
pont ~ pom
phopt ~ phop
sis ~ si
bring down
open
spread out grain
cover, hide
kill
Figure 5.1: Valency increase via stem augments
Due to assimilation processes many innitives can have the same form, but they can be distinguished via the inected forms. Many verbs are also homophonous, but they have dierent
valency and thus the inectional morphology is dierent. In the following, the dierent stem
types will be presented.
a) e rst group are non-alternating stems. Open stems as well as closed stems belong to this
group. e alternations found here are just due to general phonological rules su as the voicing
rule or the bilabial assimilation of /t/ to [p], as discussed in apter 2.2 on the various assimilation
processes. e homophonous verbs like shoot and come (both apma) or cough and pri
(both hopma) are distinguished by their dierent valency. e verb ekma in the list below shows
variation between plosive and nasal. It might be triggered by subsequent nasals in the inectional
suxes, but it is not an obligatory phonological rule, as the innitive shows. us, I tend to
perceive it as a simple exception.
stem
ap
ap
yep
hap
up
hot
hot
at
innitive
apma
apma
yepma
hapma
upma
hopma
hopma
apma
gloss
shoot
come
stand
cry
earn
cough
pri
spread
29
5 Verbal Morphology
phat
cok
yok
ek ~ e
yak
to
hi
u
to
ya/e
um
ca
khi
phapma
cokma
yokma
ekma
yakma
toma
hima
uma
toma
yama/ema
umma
cama
khima
help
do
sear
break
sele down
t, meet, agree
survive
drink
get, t, agree
emanate sound
su
eat
quarrel
b) e second stem group shows alternations of open stems and stem-nal plosives. ese stops
are elided before vowels, su as in khya.na (/khek-a-=na/) I went, so.na (/so-a-=na/) I
wated, in contrast to so.nen.na I wated you. In the innitive, the stops always assimilate
to [p], but the inection reveals the underlying plosives /k/, /t/ and //. is group is atypical
for Kiranti stem variation, because the stem-nal stops behave contrary to the usual augments:
ey appear before consonants and they are elided before vowels. is paern might result from
reanalysing augments as belonging to the stem.
stem forms
khek ~ khe
so
~ so
ha
~ ha
tu
~ tu
pi
~ pi
lut
~ lu
innitive
khepma
sopma
hapma
tupma
pipma
lupma
gloss
go
wat
bite
tread with feet
give
tell sth.
c) e stems in this group alternate between open and closed stems ending in augment -s. e
augment will never appear before a consonant, whi is why the innitive, in contrast to the group
above, always has open syllables. is kind of alternation is only found between open stems and
stems with an augmented -s, never with -t.
stem forms
nis
~ ni
yas
~ ya
cis
~ ci
khes ~ khe
us
~u
es
~e
us ~ u
innitive
nima
yama
cima
khema
uma
ema
uma
gloss
see, know
be able to
cool down
create diculty
boil, be cooked
defecate
shrink (food)
d) e next group has closed stems ending in a plosive and augmented -t. It appears only before
vowels, su as in ka gagri ipt-u-=na I lled the pot. e last verb in this list, hokma, is special
insofar as the shorter stem also shows variation between plosive and nasal. Although the nasal is
triggered by other nasals in the inectional morphology, it is no overall morphophonological rule,
as the innitive form shows.
30
5.1 Stem formation

stem forms
ipt
~ ip
opt
~ op
ukt
~ uk
tupt ~ tup
yokt ~ yok
ukt ~ uk
cikt ~ cik
khupt ~ khup
hokt ~ hok/ho
innitive
ipma
opma
ukma
tupma
yokma
ukma
cikma
khupma
hokma
gloss
ll sth.
correct, save
bring down
light up
pierce
be sold
ripen
carry
bark
e) e augment -t is also found on stems ending in the nasals /m/, /n/ and //. In innitives,
stems ending in -nt assimilate to subsequent [m], while those ending in-t stay as they are.
stem forms
et ~ e
hat ~ ha
tunt ~ tum
ent ~ em
unt
~ um
hont ~ hom
umt ~ um
imt ~ im
homt ~ hom
innitive
ema
hama
tumma
emma
umma
homma
umma
imma
homma
gloss
staple, raise
sweat
understand
become aware
pull
t into
shrink (clothes)
ask
swell
f) e next group of stems shows variation between a nal nasal and a corresponding stops-paern. Examples can be found for the pairs -m/-ps and -/-ks, the augmented form again
surfacing only before vowels. Some irregularities are found also here, e.g. ipma can also be found
for sleep, pointing towards an underlying /p/. More inected forms of the verbs listed here are
needed to distinguish between genuine stem variation of the kind described here and just another
instance of stop assimilation to a nasal. If the second possibility is true, we needed to establish
one more stem group, namely the variation between stem-nal stop and stop plus augmented -s.
is stem type is not uncommon in other Kiranti languages.
stem forms
ips
~ im
tups ~ tum
ceps ~ cem
sops ~ som
uks
~ u
paks ~ pa
hiks ~ hi
kaks ~ ka
keks ~ ke
hiks ~ hi
innitive
imma
tumma
cemma
somma
uma
pama
hima
kama
kema
hima
gloss
sleep
meet, nd, get
recover
stroke gently
come down
send
return
accept, fall down
bear fruit
return
31
5 Verbal Morphology
h) One more alternation, namely that between an open or closed stem and nal -r was found.
is augment is not typical, and might have developed from an original -t, via -d.1 So far, there
are only three representatives of this stem type. An example of an inected form is her-a=na it
dried up.
stem forms
her
~ he
por
~ po
pher ~ phep
innitive
hema
poma
phepma
gloss
dry up
fall down
open the eyes wide

e Yakkha verb shows a complex system of agreement markers for person and number, whi is
typical for Kiranti languages. In transitive scenarios, both actants trigger agreement on the verb.
In ditransitive scenarios we have primary object alignment, i.e. it is the most goal-like argument
(henceforth G-argument) that triggers object agreement. e single morphemes may code just
one category, like -ka for 2 or - for e. Some are portmanteu morphemes combining person
and number or even a specied transitive scenario, like -nen for 1>2. Other morphemes are
ambiguous, like -ci for dual or nonsingular, but their meaning is specied via their morphological
context. Due to elision of vowels to avoid a hiatus of two vowels, some morphemes are rarely
overtly realised.
e agreement axes may have dierent forms depending on whi argument type they represent (indicated by capital leers in the glosses). e dierent aligment paerns that are exhibited
by the paradigms will be treated in detail in apter 15.1 on grammatical relations. Polarity is also
marked by verbal axes, mostly nasals that can be copied several times throughout the inected
verb form, basically to fulll the function of closing open syllables. To give an example, the verb
form pimecuna they give it to her/him has the negated form mbimencunna. Other nasal morphemes, su as - e, 1s and -m 1/2pA may also be copied for the same purpose. Tense, aspect
and mood are expressed by verbal axes as well, leading to a huge number of possible forms for
ea verb.
e agreement system will be explained in the following. e table in gure 5.2 below gives
an overview of the dierent axes in intransitive and transitive (indicative) inection, extracted
from the verb forms to the exclusion of negation markers, mood and tense/aspect markers that
will be included and discussed later. e cliticised nominalisers =na and =ha/=a are not obligatory, but as the table and the paradigms show, it is very common to aa them to the verb. e
morphemes are listed in their underlying forms, in contrast to the examples that are presented as
they were recorded.
In general, the person and number marking has more distinctions in the S and A roles. For
instance, the dual number for the agreement with P arguments is only there in the second person.
Especially where the rst and second person are involved, many forms have collapsed into one,
due to the rare occurence of the respective scenarios in real life. If we look at the boxes for rst
person P, we see that a distinct form is only used if both actants have singular number. Remarkably,
the form for rst person nonsingular patient has no person marking at all anymore, only the verb
stem and negation, tense and aspect remained. In the box for rst person acting on second person,
we see a nicely layered paern: As soon as one actant, no maer whi one, has dual number,
1
is phonological ange is not unknown in Yakkha, as some correspondences between the more araic ritual
language and profane spee show, e.g. ren and ten for village, place to live.
32

intransitive
1s
1s
1di
1de
1pi
1pe
2s
2d
2p
3s
3d
3p
-=na
-ci=ha
--ci-=ha
-i
-i-=ha
-ka=na
-ci-ka
-i-ka
=na
-ci=ha
N- =ha-ci
1ns
2s
-nen=na
reexive
-nen-ci=na
transitive
2d
2p
-nen-ci=na -nen-i=na
reexive
-nen-i=na
reexive
-nen-i=na
--ka=na
-ka
reexive
-=na
-ka=na
-ci-ka
-i-ka
N- -ka=na
3s
-a
-c-u=na
--c-u-=na
-m=na
-m-a=na
-ka=na
-c-u-ka=na
-m-ka=na
=na
-c-u=na
N- =na
3ns
--ci-=ha
-c-u-ci=ha
--c-u--ci-a
-m-ci-m=ha
-m-ci-m-a
-ci-ka
-c-u-ci-ka
-m-ci-m-ka
-ci=ha
-c-u-ci=ha
N- -ci=ha
Figure 5.2: Table of person and number agreement

the dual form will be used. As soon as one actant has plural number, the plural form will be
used. ere was one exception though, namely for 1pe>2d, where the sux string -nen-ci=na
was also possible for some speakers. Generally, as these nonsingular forms are not so frequent,
there was considerable insecurity and discussion about the verb forms among the speakers. Not
only agreement with the patient shows syncretisms. If we look at the box for third person acting
on second, we see that even singular and dual number of the agent have collapsed, and in the last
column of the box, all numbers have collapsed into one form.
e verbal morphology is templatic, with clearly dened slots for ea ax. e order is shown
in Figure 5.3, including the sux copying.
1
-nen
1>2
2
-i
1/2pS/P
3
-~ -a
e
4
-ci
d
5
-u/-i
3P
6
-m
1/2pA
-
e[copy]
7
-ci
3nsP
8
-m
1/2pA[copy]
9
-ka~ -ga
2
-~ -a
e[copy]
10
=na ~ =ha
NMLZ
11
-ci
3pS
Figure 5.3: Sux slots for agreement

In the following, the morphemes and their properties will be discussed, with examples. e
prex slot can be occupied by an unspecied nasal, whi either codes agreement with third
person plural (in S and A roles) or a negation marker (cf. section 5.4). As it is unspecied with
regard to the place of articulation, it assimilates to the place of the initial consonant of the verb
stem. Before vowels and the glide /w/, it is realised as a velar nasal. e nasal prex coding
3pS/A is missing in the paradigms in combination with rst person patient and second person
plural patient. In general, the labels for the morphemes stand for a maximal extension. It is oen
the case that a morpheme is not found in all the expected slots. Examples for intransitive and
transitive inection can be found in (1).
(1)
a.
-khy-a-ma-ci
3pS-go-PST-PERF-ns
b.
ey have gone.
m-bi-a-ga=na
3pA-give-PST-2=NMLZ.s
ey gave it to you.
e rst sux slot is occupied by the portmanteau-morpheme -nen for rst person acting on
second. e second slot is occupied by -i, whi codes rst and second person plural S and second
person P arguments. It appears in agreement with the S argument in intransitive verbs, and with
33
5 Verbal Morphology
P arguments in transitive verbs. Examples can be found in (2). is analysis cannot cope with one
ambiguity exemplied in (2b), where -i can also stand for agreement with plural rst person A
arguments. But as mentioned above these forms are exactly those that caused discussion among
the speakers, so that I assume this neutralisation of agreement forms to be a new development.
(2)
a.
pi-me-nen=na
give-NPST-1>2=NMLZ.s
b.
I give it to you.
pi-me-nen-i=na
give-NPST-1>2-1/2p=NMLZ.s
c.
I give it to you (p). OR We (d) give it to you (p). OR We (p) give it to you (s/d/p).
pi-i-ga
give-2p-2
d.
He gave it to you (p).

khe-i-wa
go-1p-NPST
e.
We (incl) go.
khe-i-wa-ga
go-2p-NPST-2
You (p) go.

e exclusive morpheme - ~ a, strictly speaking, codes the non-inclusive, because also the
1s-agreement is covered by this sux. us, though it is the morphologically marked form, it is
semantically the unmarked form, dened by the exclusion of the adressee.2 e allomorph -a
results from the epenthesis of /a/ to avoid complex consonant clusters. e inclusive/exclusive
distinction that we have in the inection is not found in the personal pronouns, but it remains in
the possessive pronouns (cf. apter 4.2.2). e exclusive sux can also be copied to build a coda
of otherwise open syllables. To account for all the occurrences of -a, we have to dene two slots:
slot 3 (cf. (3a) and (3b)) and slot 9 (cf. (3c)).
(3)
a.
khe-me-=na
go-NPST-1s=NMLZ.s
b.
I go.
khya-me--ci-a
go-NPST-[1]e-d-[1]e[copy]
c.
We (dual, excl) go.

tund-wa-m-ci-m-a
understand-NPST-1pA-3nsP-1pA[copy]-e
We (p, excl) understand them.

e fourth sux slot is for -ci d and the h slot is for -u/-i for third person patient (3P).3
Due to the hiatus of the two vowels, both suxes melt to -cu. In the indicative paradigm, the
distribution of -u is only d>3P, but as we will see later in the mood paradigms, it has a wider
range of application. Additionally, it is plausible that -u is underlying in the forms where it is
not audible, and is not pronounced due to the phonologically impossible sequence */-wa-u-m/.
In slot 6, sux -m, coding 1/2pA, has the same distribution with respect to person and number
as -i 1/2pS/P, but it appears in scenarios where rst or second person plural are in the A role.
Example (4c) and (4d) provide two examples. Furthermore, the slot can also be lled by a copy of
the exclusive sux -a.
2
3
Depending on the actual verb form, the morpheme is glossed 1s in the singular or e.
e alternative sux -i occurs only in a handful of verbs, whi rather seem to be marginal exceptions than indicators
for any inectional classes (cf. (ie) and (if)).
34

(4)
a.
saptha-me-c-u=na
like[3A]-NPST-d-3P=NMLZ.s
b.
ey (dual) like it/her/him.

tum-me--c-u-=na
understand-NPST-[1]e-d-3P-[1]e=NMLZ.s
c.
We (dual, excl) understand it/her/him.

pi-wa-m=na
give-NPST-[3P]-1pA=NMLZ.s
d.
We (incl) give it to him.

pi-wa-m-ga=na
give-NPST-[3P]-2pA-2=NMLZ.s
e.
You (pl) give it to him.

mend-i
nish[IMP]-3P
f.
Finish it.
ka sabun mend-i-=na
1s soap
nish[PST]-3P-1s=NMLZ.s
I nished the soap.

Slot number seven is for the second -ci, whi marks nonsingular number agreement with third
person patient (3nsP). e second -ci can also be followed by nasal copy suxes, namely slot 8
for -m and and slot 9 for - (cf. (5b)). Slot number nine also contains the second person morpheme
-ka ~ -ga, whi is aligned neutrally. It makes no distinction for the argument type it agrees with.
Both -ci and -ka are illustrated by example (5).
(5)
a.
saptha-me-c-u-ci-ga
like-NPST-d-3P-3nsP-2
b.
You (dual) like them.

n-dund-wa-m-ci-m-a-na
NEG-understand-NPST-1pA-3nsP-1pA-e-NEG
We (pl, excl) do not understand them.

e third person S and A singular is zero, in parallel to other Kiranti languages. e nal clitics
=na (NMLZ.s) and =ha ~ -a (NMLZ.ns) are actually nominalisers that function as focus markers
here. ey are not obligatorily, but frequently aaed to the verbs. As they always correspond
to the number of the S and P arguments, they are aligned ergatively. With respect to ditransitive
verbs, they correspond to the number of the most theme-like argument (henceforth T-argument).
For =na there are exceptions to its usual appearance in singular forms: In the paradigms and
the ax table above, we can see that for the scenario 1>2, =na is always aaed, never =ha.
Also in colloquial spee, =na seems to be the unmarked oice, because it extends its scope to
nonsingular forms (cf. example (6)).
ere is one peculiar thing about the inection in the third person plural in intransitive verbs.
Here, the dual/plural distinction partly seems to be expressed by the order of the morphemes -ci
and =ha. In fact, the plural verb form rather looks like a nominalization with the homophonous
nominal nonsingular marker -ci. But this paern is consistent and regular throughout all tense and
mood forms. It is the only case so far that any inectional ax comes aer the nal nominalisers.
is is in contrast to other Kiranti languages su as Chintang and Bantawa, where the order of
axes is freely permutable without any consequences for semantics or scope (Biel et al. 2007).
Example paradigms of intransitive and transitive verbs are provided on page 38 and 39, with the
negation and the nonpast markers included.
Talking about the agreement morphology, a further feature of the paradigm has to be mentioned,
namely the absence of any marking for rst person patient (except for the forms where agent
35
5 Verbal Morphology
AND patient have singular number). is is probably a politeness strategy of negative face, i.e.
downplaying the role of oneself in patient scenarios.
5.3 Flexible agreement

As for the syntax of the agreement in Yakkha and Tibeto-Burman in general, it is mu more
exible than in Indo-European languages. As noted earlier by Biel (2000), the purely identicational agreement that is known from Indo-Aryan is accompanied by associative (appositional
and partitive) agreement types. In appositional agreement, the NP that corresponds to the agreement marker is semantically an apposition to the referent of the marker, but syntactically, it is
the argument. For instance, in (6a), the identity of the NP as second person is only revealed by
the agreement marker. e corresponding NP mamu gives additional information about the referent, without being an appositional phrase syntactically. is kind of agreement is not found in
Indo-Aryan languages.
(6)
a.
mamu
hetne khe-i-ga=na
girl[NOM] where go[PST]-2p-2=NMLZ.s
b.
Where did you girls go?

kamniwak sori
yu-i-yo
u-u-m
friend[NOM] together sit[PST]-1p-SEQ drink[HORT]-3P-1pA
c.
Having sat down together, let us friends drink.

a-koma-a=le
ta-ga=na
raea
1sPOSS-aunt-ERG=PTCL bring[PST,3P]-2=NMLZ.s MIR
You, auntie, really brought her!

e second type is partitive agreement, shown in example (7). Here, the number reference of
the NP and the number denoted by the agreement marker cannot agree, because the agreement
marker denotes a group of possible referents, while the NP denotes the subset of actual referents.
e verb here agrees with plural, although the referent of the A argument has singular number.
In (7b) this referential mismat even extends over the embedded and the main clause.
(7)
a.
nita sum-pha-be isa=a

2p
b.
ya
khus-uks-u-m-ka
three-CLF-LOC who=ERG money[NOM] steal-TEL-3P-2pA-2
Who of you three guys stole the money?

khatni=go isa=a=ca
khus-u-m-a=ha
-ka-ya-ma-nin
while.that=TOP who=ERG=ADD steal-3P-1pA-e=NMLZ.ns 3pA-say-PST-PERF-3pNEG
But none of them said:I am the one who stole the money.
5.4 Verbal negation

e negation is marked by a nasal prex and several nasal suxes that can get copied throughout
the inected verb form, to provide open syllables with a coda. e unspecied nasal prex assimilates in place to the rst sound of the verb stem. Before vowels and the glide /w/, the default
option is the velar nasal. For some forms, especially in the blo of rst person acting on second
person (cf. paradigm table 5.5 on page 39), it is the only negation marking device (cf. example (8)).
(8)
a.
n-dum-me-nen=na
NEG-understand-NPST-1>2=NMLZ.s
b.
I do not understand you.

tum-me-nen=na
understand-NPST-1>2=NMLZ.s
I understand you.
36
5.5 Tense
e negation sux is -n, sometimes also -na. It may appear in dierent slots. e rst negation
slot must be preceding the dual marker -ci, and follow the NPST markers me and wa, and thus
it is the same as the slot for - e. We can see in example (9a) that in the armative, - builds
the coda of the second syllable, and that in (9b), the negation marker fullls the same function.
e oice in favour of the negation marker may be conditioned by the proximity of /n/ to the
subsequent /c/. e examples in (9c) and (9d) illustrate the same point.
(9)
a.
tum-me--c-u-=na
understand-NPST-[1]e-d-3P-[1]e=NMLZ.s
b.
We (d,e) understand him.

n-dum-me-n-c-u-a-n=na
NEG-understand-NPST-NEG-d-3P-[1]e-NEG=NMLZ.s
c.
We (d,e) do not understand him.

tum-me--c-u--ci-a
understand-NPST-[1]e-d-3P-[1]e-3nsP-e
d.
We (d,e) understand them.

n-dum-me-n-c-u-n-ci-a-na
NEG-understand-NPST-NEG-d-3P-NEG-3nsP-[1]e-NEG
We (d,e) do not understand them.

Two more slots for negation are aer -u 3P and aer -ka 2, both exemplied in example (10):
(10)
n-dum-me-n-c-u-n-ci-ga-na
NEG-understand-NPST-NEG-d-3P-NEG-3nsP-2-NEG
You (d) do not understand them.

As mentioned before, the negation prex is homophonous with the prex coding 3pS/A. ere is
another sux, -nin to indicate negation in the third person plural. It is found in most, but not all
verb forms that agree with 3pA, and also in the two forms of the scenario 1pi>3. e occurrence
in the rst person verb forms cannot be explained with the homophonous prex. It is glossed as
p.NEG, but this has to be understood as a maximal extension, as it is the only common feature
of all the occurences of this sux (cf. (11a) and (11b)). As we will see later, it also occurs in the
mood paradigms, with an allomorph -nni. It is also hard to tear apart the nal negation marker
-na and the nominaliser =na in the singular forms, as shown in (11c). e sequence -nna could
either be geminated -na, or a string of negation marker -n and nominaliser =na, but it is dicult
to tell. In the forms with plural patient, the nal -na can only be a negation marker.
(11)
a.
n-dund-wa-m-ci-m-nin=ha
NEG-understand-NPST-1pA-3nsP-1pA-p.NEG=NMLZ.ns
b.
We (i) do not understand them.

n-dund-wa-n-ci-nin=ha
3pA-understand-NPST-NEG-3nsP-p.NEG=NMLZ.ns
c.
ey do not understand them.

n-dund-wa-n=na
NEG-understand[3s>ps]-NPST-NEG=NMLZ.ns
He does not understand him.
5.5 Tense
Yakkha basically distinguishes past and nonpast. As many verbs have inceptive semantics, we
nd past inection in many situations with nonpast time reference.
37
5 Verbal Morphology
5.5.1 Nonpast
e nonpast marker has two allomorphs -me (~ -me ~ -mya) and -wa. ey originated most
probably in two vector verbs (cf. Section 6 on verbal compounding) that got grammaticalised
as aspect markers and were now reanalysed as tense markers, to disambiguate the otherwise
homophonous past and nonpast forms (van Driem (1994) made the same observation). e lexical
verbs wama sit, stay, live, and metma make, apply do exist in Yakkha as well, though metma
was only found as causative marker so far. In other Kiranti languages though, metma is still found
as a regular lexical verb.
e distribution of these two allomorphs is not random, but grammatically conditioned. Both
forms may appear in the singular forms, though -me is more common. e oice might also be
due to dialectal variation, because the speaker using -wa in 1s came from Madi Mulkharka, while
the other paradigms were collected in Tamaphok village. e dual forms always take -me, and
plural usually occurs with -wa, but some verbs also have -me in plural, especially in the third
person.
e picture is slightly more complex in the transitive paradigms. e usual oice is -me, but
-wa occurs in the forms of third person acting on second plural (3>2p), and in the forms with third
person patient, -wa is the default oice except for those with dual agent (cf. nonpast paradigms4
on page 38 and 39, with armative and negative polarity). e two allomorphs occupy dierent
slots in the Yakkha verbal template. While -me comes immediately aer the stem and before the
agreement, -wa follows the sux -i 1/2pS/P.
1s ka
1di kanci
1de kanci
1pi kani
1pe kani
2s nta
2d ncita
2p nnita
3s u
3d uci (hippa)
3p uci
NPST-a
khemena
khemeciha
khemecia
kheiwa
kheiwaa
khemekana
khemeciga
kheiwaga
khemena
khemecia
khemeaci
NPST-neg
khemeanna
khemencina
khemenciana
kheiwana
kheiwaana
khemekanna
khemencigana
kheiwagana
khemenna
khenmencina
khemenaci
Figure 5.4: Paradigm of khepma go (NPST, armative and negative)
e non-obligatory nominalising clitics are included in the paradigm on the one hand because they are so frequent
that certain forms are never heard without them, on the other hand because sometimes one cannot tell whether the
sequence /na/ is the nominaliser or the negation sux.
38
3p
3d
3s
2p
2d
2s
1pe
1pi
1de
1di
1s
tummena
ndummeanna
reexive
1ns
tummya
ndummenna
tummeka
ndummekana
tummegana
ndummeganna
1s
tummeciga
ndummencigana
reexive
tummenenina
ndummeneninna
tundiwaga
ndundiwagana
tummenennina
ndummenenina
2p
tummenenina
ndummenenina
3s
tundwaa
ndundwaanna
tummecuna
ndummencuna
tummecuna
ndummencuanna
tundwamna
ndundwamninna
tundwamana
ndundwamanna
tundwagana
ndundwaganna
tummecugana
ndummencuganna
tundwamgana
ndundwamganna
tundwana
ndundwanna
tummecuna
ndummencunna
ndundwana
ndundwaninna
Figure 5.5: Paradigm of tumma understand (NPST, armative and negative)
ndummekana
ndummekaninna
tummekana
ndummekanna
2d
tummenencina
ndummenencina
reexive
tummenencina
ndummenencina
reexive
2s
tummenenna
ndummenenna
3ns
tundwaciha
ndundwaciana
tummecuciha
ndummencuncina
tummecucia
ndummencuncianna
tundwamcimha
ndundwamcimnina
tundwamcima
ndundwamcimana
tundwaciga
ndundwancigana
tummecuciga
ndummencunciganna
tundwamcimga
ndundwamcimgana
tundwaciya
ndundwancina
tummecucia
ndummencuncina
ndundwacia
ndundwancinina
5.5 Tense
39
5 Verbal Morphology
5.5.2 Past
e past tense is marked by -a ~ -ya, as opposed to nonpast me ~ wa, exemplied in (12).
(12)
a.
pi-a-ga=na
give[3sA]-PST-2=NMLZ.s
b.
he gave it to you
pi-me-ka=na
(3s>2sNPST)
give[3sA]-NPST-2=NMLZ.s
he gives it to you
e glide /y/ is inserted in a sequence of /-a-a/ (thus, [-a-ya]). When the past morpheme collides
with other vowels, it can be elided, as we will see below in this section. In many verb forms the
only dierence between nonpast and past is the nonpast morpheme. is is the case with all the
forms for 1>2, 3>2p and before the 3P sux -u, to avoid a sequence of /-a-u/ (cf. paradigm
tables on page 42 and 43)5 . A phonological explanation for the absence of the past morpheme
however cannot hold exclusively. As we have seen in the sequence -ci-u, the vowel /i/ is elided
to avoid a hiatus, but as for the past paradigm, there are many examples, where the clash of two
vowels cannot be the reason (for instance example (13)). At least for open stems and the suxes
-a and -i however, it is reasonable to suppose a phonotactic reason for the elision, because the
sequence /pi-a-i-ga/ (3>2p) violates the constraints on well-formed syllables in Yakkha. In forms
like pi-ga=na (/pi-a-u-ga-na/, 2s>3s), an elision of both -a and -u took place, seemingly because
of the two competing constraints to delete -a before -u, and not to allow /i-u/ sequences either. In
Puma, a Kiranti language of the Southern Central bran, a similar elision of these vowel-suxes
results in vowel lengthening and low tone. In Yakka however, this cannot be reported, and the
only instance of long // in the pipma-paradigm are the forms where /-i-i/ is underlying (3>2p).
(13)
a.
pi-nen=na
(1s>2sPST)
give-1>2=NMLZ.s
b.
I gave it to you.
pi-me-nen=na
(1s>2sNPST)
give-NPST-1>2=NMLZ.s
I give it to you.
Based on the past paradigms, two more tense forms can be constructed, namely perfect and past
perfect. ese labels are not the result of an in-depth analysis yet. ey just rely on the Nepali
translations of the inected verbs. e morphology is as follows : e past form serves as base to
whi the sux -ma ~ -mi for the perfect is aaed. is, in turn, serves as base to whi -ssa
~ -ssi is aaed for the past perfect. e sux -ssa might correspond to the somewhat irregular
verb stem sa be, was. e slot for these suxes is the same as the nonpast slot for -wa. e vowel
/a/ of the two suxes assimilates to /i/ when the verb form contains the sux -i, as in 1/2pS/P
(cf. table 5.7). Some examples are provided in (14).
(14)
a.
lop sak-a
n-sy-a-ma-a-n=na
now hunger-INST NEG-die-PST-PERF-1s-NEG=NMLZ.s
I am not hungry now. (lit. I did not get hungry.)

5
e oice of the nominalisers in this ditransitive paradigm depends on the number of the T-argument. In some forms,
the grammaticality of the plural nominalisers was doubted or rejected by the speakers. I ascribe this insecurity or
irregularity on the one hand to the low frequency of these forms in natural discourse, and on the other hand to the
apparently unmarked status of the singular nominaliser.
40
5.5 Tense
b.
eno
su-ca-ya-ma-ssa,
khatnigo hensen
day.before.yesterday it-STEM-PST-PERF-PST.PRF, but
these.days
n-su-ca-me-n=na
NEG-it-STEM-NPST-NEG=NMLZ.s
e day before yesterday it had ited, but thesedays it does not it any more.
41
42
3p
3d
3s
2p
2d
2s
1pe
1pi
1de
1di
1s
reexive
piya
mbiana
piana/-ha
mbiaanna/-ha
piaga
mbiagana
piagana/-ha
mbiagana
1s
1ns
piaciga
mbiancigana
reexive
pinenina
mbinenina
pinencina
mbinencina
reexive
2d
pinencina
mbinencina
reexive
piiga
mbiigana
pinenina
mbinenina
2p
pinenina
mbinenina
3s
pina/-ha
mbianna
piacuna/-ha
mbiancunnaha
piancuna/-ha
mbiancuanna
pimna/-ha
mbimninna/-ha
pimana/-ha
mbimanna
pigana/-ha
mbiganna
piacugana/-ha
mbiancuganna/-ha
pimgana/-ha
mbimganna/-ha
pina/-ha
mbinna/-ha
piacuna
mbiancunna/-ha
mbina
mbininna
Figure 5.6: Paradigm of pipma give (PST, armative and negative)
mbiagana/-ha
mbiaganinna
piagana/-ha
mbiaganna/-ha
2s
pinenna/-ha
mbinenna/-ha
3ns
picia
mbicianna
piacuciha
mbiancuncina
piancucia
mbiancuciana
pimcimha
mbimcimninha
pimcima
mbimcimana
piciga
mbinciganna
piacuciga
mbiancuncigana
pimcimga
mbimcimgana
piciya
mbincina
piacuciya
mbiancuncina
mbiciya
mbincinina
5 Verbal Morphology
1s
1di
1de
1pi
1pe
2s
2d
2p
3s
3d
3p
PST
khyana
khyaiha
khyacia
kheiha
kheia
khyagana
khyaciga
kheiga
khyana
khyacia
khyaci
PRF
khyamana
khyamaciha
khyamacia
kheimiha
kheimia
khyamagana
khyamaciga
kheimiga
khyamana
khyamaciya
khyamaci
PRF.NEG
khyamaanna
khyamancina
khyamanciana
kheimina
kheimiana
khyamaganna
khyamancigana
kheimigana
khyamanna
khyamancina
khyamanaci
PST.PRF
khyamassana
khyamassaciha
khyamassacia
kheimissiha
kheimissia
khyamassagana
khyamassaciga
kheimissiga
khyamassana
khyamassaciya
khyamassaci
Figure 5.7: Past, perfect and past perfect paradigm of khepma go
PST.NEG
khyaanna
khyancina
khyanciana
kheina
kheiana
khyaganna
khyanigana
kheigana
khyanna
khyancina
khyanaci
PST.PRF.NEG
khyamassaanna
khyamassancina
khyamassanciana
kheimissina
kheimissiana
khyamassaganna
khyamassancigana
kheimissigana
khyamassanna
khyamassancina
khyamassanaci
5.5 Tense
43
5 Verbal Morphology
5.6 Mood
Yakkha distinguishes subjunctive, optative and imperative mood. e subjunctive is morphologically unmarked (the forms consist just of the stem plus the agreement markers), the imperative is
marked by -a, and the optative is marked by -ni. Note that the nominalising focus particle =na/
=ha is notorically absent in these paradigms.
5.6.1 e subjunctive
A subjunctive paradigm for rst person is provided in Table 5.8, exemplied by the two intransitive verbs khepma go and apma come.6 It is used for hortative contexts, i.e. the rst person
seeking permission to do something or encouraging others to do something together, and also
for warnings. Examples are provided in (15). In the corresponding negative forms there is the
same negation paern as in the indicative inection, except for the gemination, that cannot be
explained morphologically, because two underlying nasals are not reasonable here, at least not in
the nonsingular forms.
(15)
a.
hetne khe-i?
where go[SBJV]-1p
b.
Where should we go?

ciya
hops-u-m?
tea[NOM] sip[SBJV]-3P-1pA
c.
Shall we have tea?

sori
khe-ci?
together go[SBJV]-[1]d
Shall we go together?
1s
1di
1de
1pi
1pe
Subjunctive Negation
khepma
khea
kheanna
kheci
khecinna
kheci
khecianna
khei
kheinna
khei
kheianna
Subjunctive Negation
apma
apa
apanna
apci
apcinna
apci
apcianna
abi
abinna
abi
abianna
Figure 5.8: Subjunctive, intransitive verbs
5.6.2 e imperative
e imperative is coded by the morpheme -a, homophonous with the past tense. e correspondence of past tense and deontic morphology is also known from other Kiranti languages (Biel
2003, Ebert 2003). e imperative forms are almost identical to those in the past paradigm (cf.
Chapter 5.5.2), except for the missing agreement and the new plural morpheme -ni 2/3p, whi
is also found in the optative forms. A particle -eba can be added to the imperative forms to make
them more polite (same as the Nepali n), as exemplied in (16). Table 5.9 shows the imperative
paradigms for the intransitive verbs khepma go and apma come.
(16)
a.
ab-a-eba
come-IMP-POL.IMP
I am aware of the fact that the mood paradigms are incomplete regarding person. ey will be completed as soon as
more data are available.
44
5.6 Mood
b.
Please come.
-ab-a-n-eba
NEG-come-IMP-NEG-POL.IMP
c.
Please do not come.

-khy-a-n-ci-n-eba
NEG-go-IMP-NEG-d-NEG-POL.IMP
Please do not go.
2s
2d
2p
Imperative Prohibitive
khepma
khya
khyan
khyaci
khyancin
khyani
khyanin
Imperative Prohibitive
apma
aba
aban
abaci
abancin
abani
abanin
Figure 5.9: Imperative, intransitive verbs

e picture for the imperative in transitive verbs is expectably more complex, as the verb agrees
with both actants. e agreement suxes however are the ones that we already know, with
some slight dierences. e imperative morpheme -a is elided in the forms coding the 2s>3scenarios, due to the subsequent 3P-marker -u. In the paradigm of pipma, the absence of the
suxes -a and -u in the upper row is due to the open stem pi, because a sequence of /pi-a-u/ is
not licensed by the phonotactic rules of Yakkha (cf. section 5.5.2). For the rst person patient,
we have the same paern as in the indicative paradigms. e only scenario that is distinguished
from the remaining ones is the one with actants having singular number. e object marker -u,
that had the distribution of dual acting on third person in the indicative, covers all scenarios with
a third person undergoer involved in the imperative (hence the gloss 3P). e sux -m is also
known from the indicative paradigm, it stands for (rst and) second person plural agent (1/2pA).
Figure 5.10 shows the imperative of the transitive verbs imma ask and pipma give, with the
prohibitive forms in the lower rows respectively.
1s
2s
1ns
pya
mbyaan
2d
2p
2s
pya
mbian
imda
nimdaa
2d
2p
imda
nimdan
3s
3ns
pipma give
pi
mbin
piacu
mbiancun
pianum
mbianumnin
pici
mbincin
piacuci
mbiancuncin
pianumcim
mbianumcimnin
imma ask
imdu
nimdun
imdacu
nimdancun
imdanum
nimdanumnin
imduci
nimduncin
imdacuci
nimdancuncin
imdanumcim
nimdanumcimnin
Figure 5.10: Imperative and prohibitive, transitive verbs
45
5 Verbal Morphology
5.6.3 e optative
e optative is marked by the sux -ni, following the respective number agreement, whi is
again -ci for dual and -ni for plural. It also belongs to the irrealis mood category, expressing
a general statement of interest towards the realisation of an event, while the fulllment of the
proposition is beyond the S or A arguments rea, as in the examples in (17). e optative is
also used to express requests, but it is less direct and hence more polite than the imperative. e
optative of lema be/become is also used in permissive contexts (cf. (17d) and (17e)). It is used
in a complement-like construction in these examples (with embedded innitives), but ellipsis is
frequent, leaving the bare leni O.K./Fine. In the negative, the plural and the singular forms
are syncretistic, due to the homophony of the nasal prexes for negation and third person plural,
but I suppose a dierent underlying structure. e gemination in the plural is due to a sux
string -ni-n-ni 3p-NEG-OPT, while for the singular, the plural -ni cannot apply, and therefore
the gemination must be a phonological eect (record paradigms and e again whether it was
really there!).
(17)
a.
oom-be
tas-u-ni
summit-LOC rea[3sA]-3P-OPT
b.
May she rea the top./May she be successful.

yakkha bhasa cekt-u-m-ni
Y.
c.
language speak-3P-1p-OPT
May we speak Yakkha.

ucun le-ni
good become[3sS]-OPT
d.
May it (your work) turn out nicely.

lup-ma
le-ni
tell.to.someone-INF become[3sS]-OPT
e.
You may tell them./ It is O.K. to tell them./ Please tell them.
ka-ma
le-ni
tell.something-INF become[3sS]-OPT
You may tell it./ It is O.K. to tell it./ Please tell it.
3s
3d
3p
khepma
Optative Negative
kheni
kheninni
khecini khecinni
kheni
kheninni
apma
Optative Negative
abni
abninni
apcini
apcinni
abni
abninni
Figure 5.11: Optative, intransitive verbs

e new, full list of inectional axes is presented in Figure 5.12.
46
Tense
Deontic
Negation
Agreement
-a
IMP
-me
~ -mya
NPST
-a
~ -ya
PST
1>2
2
-nen
1/2pS/P
3
-i
NPST
-ma
~ -mi
PRF
-wa
-ssa
~ -ssi
PST.PRF
2/3p
d
-ni
7
-ci
3P
8
-u/-i
NEG
e
-n
1/2pA
-
9
-m
3nsP
10
-ci
1/2pA
11
-m
Figure 5.12: Sux slots for the Yakkha verb
NEG
-n
6
-
~ -a
e
-ni
OPT
12
-ka
~ -ga
2
-
~ -a
e
-n
~ -(n)na
NEG
-nin
~ -(n)ni
p.NEG
13
14
=na
~ =ha
NMLZ
3pS
15
-ci
5.6 Mood
47
5 Verbal Morphology
48
6 Verbal compounding
Yakkha exhibits two dierent paerns of verbal compounding. One type is lexical compounding,
when a noun is incorporated into a verb, or when two verbs combine to yield a new meaning that
cannot be derived from its single components. Representatives of this group are henceforth called
bipartite verbs. e second type is the derivational and sometimes inectional compounding of one
lexical and one grammaticalised verb stem. is will be referred to as compound verb, as this term
is commonly used in the literature (Pokharel 2001, Ebert 1997, Doornenbal 2009). Both bipartite
verbs and grammatical compound verbs have the same formal structure (except for the noun-verb
type, whi can only be lexical). e inection always applies to both verbal limbs, following
certain prosodic rules. If there are any prexes, they aa to the rst verb, and the sux ain
is aaed to the vector verb. Inbetween the two verb stems, sux material (if available) has to
be copied from the nal sux string, because the vector verb subcategorises for a (maximally)
disyllabic unit as host, and the bare verb stems are always monosyllabic. Some verbs do not allow
inectional morphology between the two stems, but the reason for the dierence is not clear
yet. e Kiranti paern diers from the Indo-Aryan paern of compound verbs, where the whole
inection applies only to the vector verb, as in Hindi (Montaut 2004) and in Nepli (Pokharel 2001).
Both compound types are treated as dierent categories here only because of their lexical structure.
While the bipartites form a new meaning that cannot be retrieved from the single units, the second
verb in the grammatical compounds productively and transparently anges the semantics of the
rst verb, as will be shown in the following two sections.
6.1 Bipartite verbs

e lexical compounds may consist of two verbs or an incorporated noun and a verb. If a verb
is the rst part, both verbal limbs will host inectional morphology. If a noun is the rst part,
the inection only applies to the verb, so that prexed material will stand between the noun
and the verb. One of the incorporated nouns even takes the nominal plural marker and triggers
plural agreement in the verbal part (cf. example (2) on page 50). Morphologically, the parts of the
compounds are independent, although lexically they form a unit.
6.1.1 Verb-Verb-compounds
Examples for the verb-verb-type are suntipma get sour (cf. example (1a)), osipma feel shy
(cf. example (1b)), himsipma oke, swallow wrongly (cf. example (1c)), comtipma dry up
(cf. example (1d)), consipma be happy (cf. example (1e)), suncama it (cf. example (1f)). e
verb sipma ~ si is frequent in lexical compounding, as is cama eat, but the origin of sipma is
not claried yet. As for tipma, the inected forms reveal that the root must be pi ~ pi, give (cf.
examples (1a) and (1d)). e innitive forms again show the preference for syllables to be closed
by a nasal, whi is aieved either via epenthesis or via the mutation of stops (e.g. [suntipma]
has the underlying stems /sus ~ su/ and /pi ~ pi/).
(1)
a.
cuwa sus-a-by-a-ma=ha
beer get.sour-PST-STEM-PST-PRF=NMLZ.ns
e beer got sour.
49
b.
ond-a-si-a-=na
feel.shy-PST-STEM-PST-1s=NMLZ.s
c.
I felt shy.
cama-a hipt-a-si-a-ga=na
food-INS oke-PST-STEM-PST-2=NMLZ.s
d.
You oked because of the food.

copt-a-by-a=na
dry-PST-STEM-PST=NMLZ.s
e.
It is dried up
cond-a-sy-a
be.happy-PST-STEM-PST
f.
She/he was happy.

sus-u-ca-ya-ma-ssa
it[PST]-3P-STEM-PST-PRF-PST.PRF
It had ited.
6.1.2 Noun-verb-compounds
e noun-verb-compounds do not show uniform morphological behaviour. e noun may be a
monosyllabic stem or a disyllabic unit, because of the occurence of the nominaliser (or innitive)
-ma. Some of the compounds belong to the possessive experiencer verbs, su as lumatukma
love, sapthakma like, yupma(ci) cipma have enough sleep, yupma(ci) yuma be tired. ey are
dierent from the usual bipartites because a possessive construction is involved. eir morphology
and syntax is treated in detail in section 9.2, so that two examples shall suce here (cf. (2)). As we
can see, both parts of the compound are morphologically independent, as they can host inectional
material. e literal translation of lumatukma would be liver-hurt. In (2b), the noun yupma
shows plural marking and also triggers plural agreement in the verb. In contrast to lumatukma,
yupma(ci) yuma always shows default agreement with third person.
(2)
a.
a-luma
n-du-me-nen=na
1sPOSS-love NEG-STEM-NPST-1>2=NMLZ.s
b.
I do not love you.

a-yupma-ci yus-a-ci
1sPOSS-be.tired STEM-PST-3pS
I am tired.

e compound verbs of this kind consist of a lexical base and a vector verb (V2). e vector verb
can express derivational notions su as applicative, causative, reexive or reciprocal, but also
inectional categories like aspect and mood in the Kiranti languages. All vector verbs originate
in lexical verb stems and retain the typical augmented stem structure. e dierence between
the grammatical and the lexical compounding becomes obvious from examples where both are
combined, as in (3). In this example, we even nd two verbs added to the lexical stem /inca/ (itself
a compound), namely causative/applicative met and uks, that was tentatively interpreted as telic
marker.
(3)
in-ca-met-uks-u--ni
play-STEM-APPL-TEL-3P-1sA-while
while I played (sh the sh)
50

e following section will introduce some of the vector verbs that are found in Yakkha. ere
is the benefactive applicative, derived from pipma ~ pi ~ pi give. Another verb hama ~ haks
send is used in similar contexts, and it may also express malefactive notions (cf. (4d), at least not
aested for pipma yet). One argument position is added to the verb, so that intransitive becomes
transitive and transitive becomes ditransitive. e benefactee will trigger object agreement in
the verb, following the usual paern of agreement with G arguments for ditransitive verbs (cf.
example (4c)). is example also exhibits an aspiration of the voiced bilabial, that cannot be
explained yet. As a vector verb, pipma may also have a completive interpretation (cf. example
(4e)).
(4)
a.
end-u-bi-=ha
pour-3P-BEN-1s=NMLZ.ns
b.
I poured it (the ketup) for her.

-end-u-bi--an=ha
NEG-pour-3P-BEN-1s-NEG=NMLZ.ns
c.
I did not pour it for him.

ka katha lend-a-bhy-a-
1s story exange-IMP-BEN-IMP-1s
d.
Tell me a story.
mokt-haks-wa-=na
beat-BEN-NPST[3P]-1s=NMLZ.s
e.
I will give him a beating.

ca-ya-bhy-a-=na
eat-PST-TEL-PST-1s=NMLZ.s
I have nished eating (i.e. the whole procedure is done, including washing ones
hands).
e vector verb khepma ~ khe ~ khya go conveys a telic meaning. Some event or action has
reaed its ultimate end, as the clauses in example (5) shall illustrate. e respective innitives
are imkhepma fall asleep, kakhepma fall down, phomkhepma spill over, somkhepma slide,
slip and hokhepma crumble down. e stems in the innitives preferably end in a nasal, also
if another coda is underlying (cf. example (5d) and (5e)).
(5)
a.
ips-a-khy-a=na
sleep[3s]-PST-TEL-PST=NMLZ.s
b.
She fell asleep.

kaks-a-khy-a=na
fall[3s]-PST-TEL-PST=NMLZ.s
c.
She fell down.

phom-khem-mya
spill[3s]-TEL-NPST
d.
It spills over.
latho uimala-be
sos-a-khy-a=na
elephant steep.slope-LOC slip[3s]-PST-TEL-PST=NMLZ.s
e.
e elephant slipped at the steep slope.

parkhal hor-a-khy-a=na
wall
crumble[3s]-PST-TEL-PST=NMLZ.s
e wall crumbled down.

A telic meaning is also conveyed by the verb uma ~ uks come down. e exact dierences
between uma and khepma cannot be established without a deeper analysis and more data. e
usual Nepali translation of uma as a vector verb is saknu nish. Impressionistically, in all examples collected so far, except for those with forget, the lexical verbs denote events that involve
51
a movement towards the S/A argument. We have examples with soma see/know, tumma understand, wapma put on clothes, uma drink. e vector verb is sometimes hard to distinguish
from the remaining morphology, because its stem vowel can get lost, as in (6a) and (6b). is is
optional, as (6c) shows, where the sequence /di-u/ anges into [dyu]. In example (6e) a gloal
stop is prothesized to uma, because the elision of the /a/ that belongs to the stem does not seem
to be a valid option.
(6)
a.
so-ks-u-ga=na
see-TEL[PST]-3P-2=NMLZ.s
b.
Have you seen him?

mun-di-ks-u-m-a
forget-STEM-TEL[PST]-3P-1pA-e
c.
We have forgoen it.

mun-dy-uks-u-
forget-STEM-TEL[PST]-3P-1s
d.
I have forgoen it.

nis-uks-u=na
know-TEL[PST]-3P=NMLZ.s
e.
He knows it.
te
wa-uks-u-=ha
clothes put.on-TEL-3P-1s=NMLZ.ns
I have put on the clothes.

e verb hema ~ heks is used when the event denoted by the main verb is about to begin (Nep.:
lgnu). Its literal meaning is cut/saw in Yakkha. Note that because of the inceptive semantics of
many verbs, the past form is used. If the lexical verb ends in an obstruent, it will get aspirated.
(7)
a.
lop wasik ta-heks-u=na

now rain
b.
half
c.
bring-be.about.to-3P=NMLZ.s
It is about to rain now.

theklup leks-heks-a=na
become-be.about.to-PST=NMLZ.s
Half (of the book) is about to be nished.

mend-heks-a
nish-be.about.to-PST
It (the soap) is about to be nished.

e verb emma ~ end has the lexical meaning apply, insert. In compounds, the ad-hoc interpretation from the few examples that are available so far, is the expression of an immediate and
qui accomplishment of an event (Nepali: hlnu). In example (8c), I assume that the underlying
stems of [amd] are /ap-end/. e sux -i however is puzzling here.
(8)
a.
n-lept-end-wa=na
3pA-throw-TEL-NPST-NPST
b.
ey throw it. (Nep.: phlidinan)

khe-nd-i-me-ci-ga
go-TEL-2p-NPST-ns-2
c.
You go away suddenly/ immediately. (Nep.: gihlau)

am-d-i-me-=na
come-TEL-?-NPST-1s=NMLZ.s
I will come immediately/ without delay. (ihalu)
52
7 Adverbs and adjectives

7.1 Adverb-building strategies
Some adverbs are derived by aaing a particle -lo ~ -lo and the sux -wa, that is found as a
nominaliser and as an adverbialiser. e adverb peklowa usually for instance is derived from
pek ~ pyak many/mu with the two morphemes aaed to it. In other contexts, loa ~ lowa
is used like a postposition meaning like, as in khobek loa like a crab. e particle -lo ~ lok ~
lo is also known from Belhare as a comitative (Biel 1993) and as manner sux from Bantawa
(Doornenbal 2009). In Yakkha, it is found in derivational nominal morphology, as in iplok unity,
lolok dress, uniform and numerals like oplok ten thousand.
Another strategy that derives adverbs is the aament of the comitative case -nu to adjectival roots, su as khumdunu tasty and ucunnu nice, beautiful. ese roots occur with the
nominaliser =na/=ha to yield adjectives, and with the comitative to yield adverbs (cf. (1)).
(1)
a.
ucun-nu nam-ma
nice-COM smell-INF
b.
to smell tasty
khumdu=na cama
tasty=NMLZ.s food
c.
the tasty food

su-a-nu
et-u-=ha
be.sour=NMLZ.ns-COM taste/feel.like-3P-1s=NMLZ.ns
d.
It tasted sour (to me).

tu-tu=nna
cokcoki
above-REDUP=NMLZ.s star
the star far up above

e particle -ni, translatable with while and otherwise used in clause linkage, may also be
employed to derive temporal adverbs, su as asenni long ago, or imni the year before lost
year, khatni this time, now.
Another particle that is found frequently in adverbs, especially in ideophonic adverbs, is -lik,
as in kakkulik-kakkulik fall down slowly and at, cicagalik fall down overturned, but also in
adverbs that are derived from lexical bases, su as cumagalik lost, hiklik turning around,
iplik-iplik twisted, siklik hurt, dead, wakulik bent, crooked. is particle is also found in
nominal compounding, especially for words denoting small animals, and it might be a lexicalised
diminutive. Some of these nouns are yacukulik ant, siplik bed bug, opokalik bug, mukalik
garden lizard, hokalik wall lizard. Athpare and Puma have a diminutive let, whi supports
this assumption (cf. Ebert (1997), Biel et al. (2006)).
7.2 Spatial deixis

Elaborate deictic systems in case, adverbs, demonstratives, and even in interjections (Biel 1999a)
are typical for Kiranti. Yakkha is no exception to this, apart from the gap in the locative case (cf.
Chapter 4.3 on the case system).
53

e set of demonstratives that make a distinction along the proximal-distal continuum was
already introduced in Chapter 4.2.3. Table 7.1 repeats the paradigm.
PROX
DIST
singular
na
nna
nonsingular
kha
kha
Figure 7.1: e vertical dimension in motion verbs

e proximal-distal distinction is cross-cut by another deictic system that includes the vertical
axis. e forms of the deictic roots are as follows: khe here, yo there, across (both for the same
level), to up and mo down. e roots can occur without further morphology. ey may also
host the locative, the allative and the ablative case, as shown in the examples in (2). e locative
-pe anged to -ne, sometimes it is geminated. e ablative -pa appears as -nda.
(2)
a.
khe-khala
PROX-ALL
b.
(towards) here
khe-nda
PROX-ABL
c.
from here
yo-khala
across-ALL
d.
towards there
mu-nne
down-LOC
down - need loc. example from o-set!

ere is a second paradigm with dierent stem vowels for all except the proximal root, namely tu
up, mu down and yu across. is might indicate a transposed system, where the deictic centre
is dierent from the place where the speaker is, transposed to somewhere further away. is is
used to express notions su as that mountain up there, beyond the other mountain/beyond that
river. is assumption is not yet supportable by any further data. It needs a closer look and a
careful analysis. It is supported however by the fact that for this set no proximal root is available.
e second set of adverbs also involves the gemination of the nasal (for an overview cf. the table
in gure 7.2).
e same deictic roots may also serve as bases for demonstratives, via the aament of the
nominalisers =na and =ha ~ -a ~ -ya. e demonstratives may appear independently as pronouns,
but also in modifying position.
(3)
a.
khe=na
toba
PROX=NMLZ.s beer.in.small.barrel
b.
this beer
khe=ya
toba-ci
PROX=NMLZ.ns beer.in.small.barrel
c.
these beers
mo=ha
pik-ci
DOWN=NMLZ.ns pig-ns
the pigs below

Via duplication of the initial CV-cluster, the adverbial roots can intensify their original meaning,
su as in example (4). e reduplicated forms may function as adverbs (e.g. mumunne) or as
54
7.3 Adjectives and comparison

demonstratives in modifying position (e.g. mumunna). Table 7.2 summarises the spatial adverbs
and demonstratives. Not all forms are aested yet.
(4)
a.
tu-tu=nna
cokcoki
above-REDUP=NMLZ.s star
b.
the star far up above

to-to
khy-a!
above-REDUP go-IMP
Go up there!
To complicate the picture further, combinations are also possible, su as mumo down there,
tuto up there and yuyo over there (same level), that are also used while pointing and showing
something. us, together with the single and reduplicated forms, we arrive at ve distinctions
for ea word above, below and beyond/across (same level).
UP
PROX
DIST
DOWN
UP
DIST
DOWN
UP
DIST
DOWN
root
to
khe
yo
mo
tu
yu
mu
tuto
yuyo
mumo
+ LOC
(tone above)
(khene this side, near)
(yone far, across)
(mone below)
munne
yunne
tunne
+ ABL
tonda
khenda
yonda
monda
+ ALL
tokhala
khekhala
yokhala
mokhala
Demonstrative
tona
khena
yona
mona
tunna
yunna
munna
REDUP(=NMLZ/-LOC)
toto(-nna/-nne)
yoyo(-nna/-nne)
momo(-nna/-nne)
tutu(-nna/-nne)
yuyu(-nna/-nne)
mumu(-nna/-nne)
Figure 7.2: Summary of spatial deixis in adverbs and demonstratives

e motion verb roots have also lexicalised dierent directions on the vertical axis, as shown
in table 7.3.
NEUTRAL
UP
LEVEL
DOWN
come
tama come
kepma ~ kya come up
apma come (same level)
uma ~ uks come down
bring
tapma ~ ta bring
apma ~ apt bring
ukma ~ ukt bring down
Figure 7.3: e vertical dimension in motion verbs
7.3 Adjectives and comparison

Adjectives in Yakkha are oen tracable to verbs denoting some property. Many properties are
expressed by verbs translatable with be sour, be bla, be cold, and the adjectives are usually
built by the verb stem and the nominalisers =na and =ha. Tenically, the adjectives are participles,
but they seem to be perceived as lexical words by the speakers. Examples are provided in (5) (some
have been repeated from Chapter 3).
(5)
a.
ha=na/
ha=ha
be.spicy=NMLZ.s/ be.spicy=NMLZ.ns
hot, spicy
55

b.
hu=na/
hu=ya
burn=NMLZ.s/ burn=NMLZ.ns
c.
burnt
ci=na/
ci=ha
be.cold=NMLZ.s/ be.cold=NMLZ.ns
d.
cold
su=ha
be.sour=NMLZ.ns
e.
sour
makhru=na
be.bla=NMLZ.s
f.
bla
o=ya
fry=NMLZ.ns
fried
Other adjectives are also derived from verbs, but without the nominaliser, and sometimes without any further morhology at all. Not all adjectives have a transparent etymology of course.
(6)
a.
att-u
spread[3sA]-3P
b.
fat
em
be.clear
clear
Comparison is covered by a construction involving the particle -hatni, that is aaed to the
standard of comparison, whi is typically a noun or pronoun. It can also be an adverb, as in (7a).
Not only properties denoted by adjectives can be compared using this construction. As we see
in (7), verbs frequently denote the quality that is compared, so that we cannot use comparison as
dening criterion for adjectives.
(7)
a.
u-la-ci
eno-hatni
n-sas-a-ma.
3sPOSS-leg-ns some.time.ago-COMPAR 3pS-be-PST-PRF
b.
Her legs got stronger than last time.

ka nda-hatni ke-gek-a=na
1s 2s-COMPAR grow-STEM-1s=NMLZ.s
c.
I am taller than you.

ka nda-hatni lam-ma yam-me-=na
1s 2s-COMPAR walk-INF be.able-NPST-1s=NMLZ.s
d.
I can walk (beer/more) than you.

nda-hani pak=na?
2s-COMPAR be.young=NMLZ.s
e.
Is he younger than you?

ka u-hani tum-a=na
1s 3s-COMPAR be.ripe/be.old-1s=NMLZ.s
I am older than him.
7.4 Reduplication, triplication and ideophonic patterns in adverbs

and adjectives
Reduplication is a frequent means of emphasis in adjectives and adverbs. Examples are provided
in (8).
56
7.4 Reduplication, triplication and ideophonic paerns in adverbs and adjectives

(8)
a.
lop-lop
now-RED
b.
just now
mi-mi=ya
small-RED=NMLZ.ns
c.
very small
i-i
tight-RED
d.
very tight
u-u
sad-RED
e.
worried
tum-tum-ma
senior-RED=NMLZ
f.
adult
ip-lik-ip-lik
twist-PTCL-RED-PTCL
g.
twisted
elok-elok
from.far.away-REDUP
from very far away

Many adjectives and adverbs are ideophonic, too. In these examples, the base for the reduplication does not occur independently. e reduplication is obligatory. Reduplicated and ideophonic
adverbs are provided in (9). Sometimes the initial consonant is anged in the reduplicated part,
as in (9b).
(9)
a.
kakkulik-kakkulik
IDEOPH-RED
b.
falling slowly and at

kakulin-wakuli
IDEOPH-RED
c.
tumbling down
ga-ga
IDEOPH-RED
feeling full aer consuming too mu food

Triplication paerns as they are found in Bantawa and Chintang (cf. Rai (1984), Rai et al. (2005))
were also found, so far only with ideophonic bases (cf. example (10)). e ange in the initial
consonant usually involves the rhotic /r/ or the labial /l/.
(10)
a.
hu-ru-ru
IDEOPH-TRIPL-TRIPL
b.
wind blowing
tu-ru-ru
IDEOPH-TRIPL-TRIPL
c.
tears falling
phi-li-li
IDEOPH-TRIPL-TRIPL
d.
buery jiering
pe-le-le
IDEOPH-TRIPL-TRIPL
something heavy being bloed while pulling
57

e.
tho-lok-lok
IDEOPH-TRIPL-TRIPL
boiling water
58
8 e structure of simple clauses

is apter deals with phrase structure and constituent order. e unmarked constituent order is
head-nal, but it can be modied for purposes of information structure. In clauses, the unmarked
order is SOV. Noun phrases are optional. In ditransitive constructions, the recipient usually precedes the theme. is order can be anged, due to a tendency to put thematic elements rst,
with rhematic increase towards the end of the clause (cf. the examples in (1)). e order is not
anged in interrogative clauses. Oen, the interrogative words are found in the focus position
right before the verb (cf. examples (1c) and (1d)).
(1)
a.
ilen
patapa-ga u-nip-ma
mas-a-py-a-ma-sa
one.day landlord-GEN 3sPOSS-count-NMLZ lose-PST-TEL-PST-PERF-PST.PERF
b.
One day, the landlords money got lost.

ani-ga
likha-ga
uhile utpatti mamli-be
leks-a=na=bu
1pPOSS.e-GEN a_subclan-GEN long.ago origin a_village-LOC happen-PST=NMLZ.s=REP
c.
Our Linkha clans, origin, long ago, was in Mamling, they say.
taba
hetne tas-wa-ga=na
male.in-law where arrive-NPST-2=NMLZ.s
d.
Where will (your) husband arrive?

na-ga suru
imin leks-a=na
this-GEN beginning how be-PST=NMLZ.s
How did she originate?

ere is a clause-nal aerthought position, as shown in (2).
(2)
likha-ci
nam-nu bagari n-jog-a,
a_subclan-ns sun-COM bet
bagari
3pS-do-PST, bet
e Linkas had a bet with the sun, a bet.
8.1 e noun phrase

e NP has the same head-nal structure as the clause has. e head can also be omied, as the
possessive phrase in example (3a) shows. Modiers within the NP can be topicalised with =go (cf.
(3b)). As for the order of modiers within the NP, there is not sucient data to make any claims.
NPs are rare anyway, and there is rarely more than one modier.
(3)
a.
ego ani-ga
likha-ga
one 1pePOSS-GEN a_subclan-GEN
b.
one (story) of our Linkha clan

uk-ka=go
pik-ci
3sPOSS-GEN=TOP cow-ns
his cows
Any element in the NP can move behind the head noun. In example (4), it is the numeral (with
classier) that has moved to the end. e position behind the noun is restricted to one element.
(4)
na ak-ka
kamniwak-ci hip-pa
this my-GEN friend-ns
two-CLF
59
8 e structure of simple clauses

these two friends of mine
Noun phrases and their constituents can be coordinated just by juxtaposition. In example (5a),
we have two coordinated adjectives with the same head noun. In the coordination of nouns, a
possible case marker will only stand at the end of the whole conjoined phrase, because it is a
phrasal ax (cf. example (5b)).
(5)
a.
mam=ha
mimi=ya
nasa-ci
big=NMLZ.ns small=NMLZ.ns sh-ns
b.
big and small sh

m-ba
m-ma-a
2sPOSS-father 2sPOSS-mother-ERG
your father and mother
8.2 Copulas
Yakkha has three copulas: the armative copula om, and two negative copulas manna for existential, and menna for identicational use (cf. example (6)). ey do not show person agreement
markers, only a number marker can be found in some instances.
(6)
a.
ceya
-khala
om
language DIST-like.this COP
b.
e language is like that.

na
kamala-ga pa om?
DEM.PROX K.-GEN
house COP
c.
Is this Kamalas house?

e-ka-be
paisa manna
d.
We do not have money.

na
paisa ak-ka
1diPOSS-GEN-LOC money NEG.EXIST
menna
DEM.PROX money 1sPOSS-GEN NEG.IDENT
is is not my money.
e forms of the copulas are frozen. Except for nonsingular number, they cannot host any
inectional axes. Only the negative copula can be inected for nonsingular number, as in (7):
(7)
na-khala
manna-ci?
DIST-like.that NEG.EXIST-ns
ose kinds of (grindstones) are not there (in Germany)?
60
9 Argument structure
9.1 Valency ange
9.1.1 Detransitivisation
Any transitive verb can basically be inected intransitively, whi is a typical Kiranti feature.
e object is omied in the intransitive agreement morphology. e verb cokma in example (1a)
is inected transitively, while in (1b) it is inected intransitively. e oice has to do with the
referential status of the object. If it is denite or specic, the transitive form will be used, if it is
unspecic or generic, or if the maer is rather about the structure and manner of the event, the
intransitive form will be used.
(1)
a.
cog-uks-u=na
do[3sA,PST]-TEL-3P=NMLZ.s
b.
He did it.
ekdam cog-a-nu
very
cog-a-nu
do[3sS]-PST-while do[3sS]-PST-while
while he worked hard/while he did a lot

An option in Yakkha to be unspecic about the person and number of the agent of an event is the
use of intransitive inection with 3s default person marking, comparable to using the indenite
pronoun one in English or man in German. e oice of the nonpast allomorph men in (2a) shows
that the inection is intransitive, because the transitive inection would have the allomorph wa
in the 3s>3s scenario.
(2)
a.
i=ya=ca
cok-ma n-ya-me-n=na
what=NMLZ.ns=ADD do-INF NEG-be.able[3sS]-NPST-NEG=NMLZ.s
b.
One cannot do anything (about it).

i=ya=ca
cok-ma n-yas-wa-m-nin=na
what=NMLZ.ns=ADD do-INF NEG-be.able-NPST[3P]-1pA-NEG=NMLZ.s
We cannot do anything (about it).
9.1.2 Causatives
Causatives are constructed via the compound verb construction that was introduced in section
6.2. e vector verb involved here is metma make, homophonous with the nonpast marker, and
probably from the same source etymologically. Some of the causatives got lexicalised, as in yokmetma tell about something, make someone curious. As single verb, yokma means sear. e
valency of the verbs is increased, as yokmetma is ditransitive, with the person told as G argument
and a noun phrase or an embedded clause as T argument (with the conditional conjunction bo
as complementiser), as shown in example (3).
(3)
ak-ka
kamniwak-ci, ka
isa
om, isa
om-bo
1sPOSS-GEN friend[NOM]-ns 1s[NOM] who[NOM] COP who[NOM] COP-COMPL
u-ci-a=ca
n-yok-me-me-ka-nin
3-ns-ERG=ADD 3pA-sear-CAUS-NPST-2-NEG
My friends, they also will not tell you who I am.
61
Other examples for causatives are tamema bring, guide, and incametma play with something.
e second example is not a canonical causative semantically, but it also increases the valency.
While incama just means play, in the causative version an argument is added. is is again in
parallel to the Nepali word khelnu play and the causative khelaunu play with something.
9.1.3 Applicative
ere is a marker -na that I call theme-applicative, but currently the data is restricted to one
example (cf. (4)). e T argument of give is promoted at the cost of the demotion of the G
argument to an oblique-marked (locative) adjunct, so that the verb agrees with A and T now.
e reason here seems to be that the spee-act participant, i.e. the second person, needs to be
promoted to a status where it can trigger agreement. Note that the order of G and T argument has
anged as well.
(4)
a.
ka
nta
caklet
pi-me-nen=na
1s[NOM] 2s[NOM] sweets[NOM] give-NPST-1>2=NMLZ.s
b.
I give you the sweets. (agreement with A and G arguments)

ka
nta
a-phu-be
pi-na-me-nen=na
1s[NOM] 2s[NOM] 1sPOSS-e.B.-LOC give-DETR-NPST-1>2=NMLZ.s
I give you to my elder brother. (agreement with A and T arguments)

Experiencer constructions are particularly interesting, because their syntactic properties oen differ considerably from usual argument structure (cf. Biel (2004b)). Experiencer constructions are
aracterised by some core participant of the verb being emotionally or sensationally aected by
the event denoted by the verb. Some examples are provided in this section, to shortly introduce
the range of possible syntactic options for the expression of experiencer events. Morphologically,
the experiencer verbs are compounds of an incorporated noun and a verb. e degree of incorporation varies, with respect to the possibility of intervening material and phonological integration.
But in any case, the meaning is compounded by the noun and the verb, and it is dierent from
the meaning of the components taken by themselves. Morphologically, both limbs are still independent, as ea takes the inection according to its word class. e prenasalisation that codes
negation will always aa to the verbal root, never to the noun-verb-complex as a whole (see
(5a)).
One possibility is a construction with the verb sima (die or kill - distinguished by dierent
stem behaviour). e instrumental case or an ergative case can be aaed to the stimulus.
(5)
a.
sak-a
n-sy-a-ma--an=na
hunger-INS NEG-die-PST-PERF-1s-NEG=NMLZ.s
b.
I am not hungry.
raksi-a sis-a-ga=na-i?
spirit-ERG kill[3sA]-PST-2=NMLZ.s-PTCL
Are you drunk?

Another construction is called possessive experiencer-construction. Here, the experiencer is
conceptualised and syntactically constructed as the possessor of a feeling or of an aected body
part, whi is represented by a noun that is incorporated into the verb. ere are dierent possessive experiencer constructions. Either the verb agrees with the feeling or aected body part1 , or
1
Another interpretation is to say that the verb shows default 3s-agreement. But as we also nd nonsingular-agreement
with some of the (nonsingular) incorporated nouns, the rst option seems more plausible to me.
62

the verb agrees with the experiencer and the stimulus (you and them in (6a), he and you in
(6b)). e verb in example (6d) appears to be inected intransitively.
(6)
a.
nda u-ci n-luma

2s
b.
tukt-wa-ci-ga
3-ns 2sPOSS-liver hurt-NPST-[3]ns-2
You love them.

u-sap
tha-me-ka=na
3sPOSS-soul(?) raise[3sA]-NPST-2=NMLZ.s
c.
He likes you.
ka dotpen a-luma
kipt-wa-=na
1s dotpen 1sPOSS-liver be.greedy-NPST[3P]-1s=NMLZ.s
d.
I got greedy for the dotpen.

tawa-ga
u-lok
-kho-me-n=na?
male.in.law-GEN 3sPOSS-anger NEG-scrat(?)[3sS]-NPST-NEG=NMLZ.s
Isnt your husband angry?

e other group of possessive experiencer verbs mentioned above shows agreement not with the
experiencer, but with the possessee, hence, the feeling or aected body part. Verbs like for instance
pomma kepma feel lazy, yupma(ci) cipma have enough sleep and yupma(ci) yuma be tired
belong to this group. Example (7) illustrates this paern. Note that the noun yupma sleep has
plural number. e plural ending at the noun is not obligatory, in contrast to the obligatory plural
agreement on the verb. Note that while yupma(ci) yuma be tired and pomma kepma feel lazy
have intransitive morphology, yupma(ci) cipma have enough sleep has transitive morphology.
In example (7d), yupma is the S-argument of the verb (at least if we take agreement as the dening criterion), while in example (7e), it is the P-argument, with the experiencer as A-argument.
Regarding (7a) (and (7b)), note the voicing of the initial in the verb. is was not observed for the
other experiencer verbs (e.g. example (7e)), where the voicing rule does not apply.
(7)
a.
a-pomma
-gy-a
1sPOSS-lazyness 3pS-come.up-PST
b.
I feel lazy.
m-pomma
-gy-a
2sPOSS-lazyness 3pS-come.up-PST
c.
You feel lazy.

-ga
yupma n-yus-a-ci?
2sPOSS-GEN sleep[N] 3pS-STEM[3]-PST-ns
d.
Are you tired?

a-yupma
n-yus-a-ci
1sPOSS-sleep[N] 3pS-STEM[3]-PST-ns
e.
I am tired.
a-yupma
cips-u--ci-=ha
1sPOSS-sleep have.enough.sleep-3P-1s-ns-1s=NMLZ.ns
I had enough sleep.

Other verbs have intransitive morphology, and they only agree with the experiencer. Again, all
of them are compounds. Examples are osipma feel shy and yasipma feel exhausted (not to
be confused with the verbs containing sima).
(8)
a.
o-si-me-=na
feel.shy-STEM-NPST-1s=NMLZ.s
b.
I feel shy.
n-yag-a-sy-a-ma-a-n=na
NEG-feel.exhausted-PST-STEM-PST-PRF-1s-NEG=NMLZ.s
63
I am not exhausted.
Sometimes the stimulus hosts an ablative case. is was found for instance in kisipma be afraid
in (9).
(9)
ka
nda-bha kisit-a=na
1s[NOM] 2s-ABL
be.afraid-1s=NMLZ.s
I was afraid of you.
64

10.1 Nominalisation
10.1.1 e nominaliser =na ~ =ha
Yakkha has a nominaliser that also appears in aributivising function. It exhibits a grammatically
conditioned allomorphy, triggered by the number of the head noun. Its form is =na for singular
reference, and =ha ~ =ya for plural and uncountables, i.e. for everything nonsingular. e nominaliser connects any modifying material, i.e. adjectives, demonstratives, question pronouns and
relative clauses, to a nominal head (cf. also section 7.3 (on adjectives), section 4.2.3 and section 7.2
(on demonstratives), section 4.2 (on question pronouns)). e clausal nominalisation is also found
in independent clauses in the function of a focus particle.
Adjectives are usually based on a verbal root plus the nominaliser. Depending on the number of
the head noun, the singular or the plural nominaliser is osen. For instance, lakna/lakha means
salty, derived from the verb (lakma be salty), siana means dead, derived from (sima die).
Demonstrative pronouns consist of an adverbial root and the nominaliser, as in the following
examples.
(1)
a.
khe=na
toba
DEM.PROX=NMLZs. beer
b.
this beer
to=ya
pik-ci
DEM.UP=NMLZ.ns cow-ns
the cows up there

Also interrogative words host the nominaliser. e interrogative pronoun ina what (s) is employed if the number of the item in question is singular, while iya what (ns) is used if it is obvious
that the answer will contain more than one item. Nominalised like this, the NPs may also host
the nonsingular marker, as in example (2b).
(2)
ca-ma
a.
i=ya
b.
What to eat? (consists typically of more than one item)

imin=ha=ci?
what=NMLZ.ns eat-INF
how=NMLZ.ns-ns
c.
How are they?

imin=na
kokomek?
how=NMLZ.s buery
What kind of buery?

Case-marked phrases may also be nominalised (cf. example (3a)). Clauses become relative clauses
via the nominalisation, as far as one can tell to date, without restrictions on the argument types.
While the example in (3b) is a nominalised innitive, the participants in (3c) are not specied. As
the examples in (3d), (3e) and (3f) show, the head nouns need not be nominative arguments of the
nominalised verbs that modify them, but can be locative or other adjuncts.
65

(3)
a.
u-ti-nu=ha-ci
3sPOSS-thorn-COM=NMLZ.ns-ns
b.
(the sh) with the thorns

nasa ep-ma=na
tawasi
c.
the shing rod to cat sh

ca=ya
cama
asen
sh cat-INF=NMLZ.s shing.rod
yesterday eat=NMLZ.ns food
d.
the food that was eaten yesterday

asen
khe-i=ya
aphis
e.
the oces that we went to yesterday (plural marking in nouns denoting inanimates
is not obligatory)
na tihar
din-be lak-ma=na
ceya n-jog-u
f.
On this Tihar day, they decided the maer about the dance.
yakkha-ba yakkha-ma-ci-ga -wa-ya=na
ten-be
yesterday go-1p=NMLZ.ns oce[p]
this a_hindu.festival day-LOC dance-INF=NMLZ.s maer 3pS/A-do[PST]-3P
Y.-m
Y.-f-ns-GEN
3pS-live-PST=NMLZ.s village-LOC
in the village where the Yakkha men and women lived

e number allomorphy of these nominalisers parallels the two demonstrative pronouns na
this and kha these that were discussed in Chapter 4.2.3. An etymological connection is likely.
(4)
a.
na toba imin et-u-ga=na?

this beer
b.
how like-3P-2=NMLZ.s
How do you like this beer?

kha-ci imin et-u-g=ha?
these-ns how like-3P-2=NMLZ.ns
How do you like these?

e nominalizers are also quite frequent in inected verbs at the end of a main clause, to signal
focus on some element within the clause. e examples in (5) illustrate the focus on a negated
verb and in a question. In Athpare, the focus is even obligatory in questions (Ebert 1997). is is
not yet eed for Yakkha.
(5)
a.
eno
sun-ca-ya-ma-sa,
khatnigo hensen
day.before.yesterday it-eat-PST-PRF-PST.PRF but
these.days
n-sun-ca-me-n=na
NEG-it-eat-NPST-NEG=NMLZ.s
b.
Some days ago it had ited, but thesedays it does not it any more.
cu-a
sy-a-ga=na?
cold-INSTR die-PST-2=NMLZ.s
Are you feeling cold?

As the nominaliser agrees in number, the question must be asked whi argument of the clause it
selects to agree with. e following examples will show that the nominaliser is aligned ergatively.
In intransitive verbs, it agrees with the S-argument (most subject-like argument), shown in (6a).
In transitive verbs, the nominaliser agrees with the P-argument (the most patient-like argument),
illustrated by the examples (6b) and (6c). In ditransitive verbs, the verb agrees with the T-argument
(the most theme-like argument). is is illustrated by (6d), where the most theme-like argument is
beer, triggering the nonsingular allomorph, because the T-argument is perceived as uncountable.
us, in ditransitive verbs, agreement with all three core participants is possible, although the
66
10.1 Nominalisation
agreement with the T-argument via the nominaliser tells us only about the number, not the person
of the participant.
(6)
a.
khe-me-=na
go-NPST-1s=NMLZ.s
b.
I go. (S)
makhu ses-wa-=na
garlic
peel-NPST-1s=NMLZ.s
c.
I peel the garlic (one clove). (P)

mankhu ses-wa-=ha
d.
I peel the garlic (many). (P)

ka nta cuwa
pi-me-nen=ha
garlic
peel-NPST-e=NMLZ.ns
1s 2s beer.in.bowl give-NPST-1>2=NMLZ.ns
e.
I give you the beer. (T)1

ka nta caklet pi-me-nen=na
1s 2s sweet give-NPST-1>2=NMLZ.s
I give you the sweet.(T)

e nominalised clause can also be the complement of postpositions, as example (7) shows. e
postposition is borrowed from Nepali.
(7)
ka-ya=na
anusar
say[3sA]-PST=NMLZ.s according.to
according to what he promised
10.1.2 Nominalisation with -khuba ~ -khuwa

is nominalisation can apply to any verb, and the result is a participle or a noun that has the
semantic role of S/A with respect to the verb. e nominaliser aaes to the verb stem, without
inectional markers. e second syllable of the nominaliser corresponds to the common TibetoBurman nominaliser -pa ~ -ba, that also has the function of a male/neutral nominaliser in Yakkha.
It is not an agent nominalisation, because the semantic role of the S-argument need not be an
animate or human controller of the event, as the following examples in (8) show.
(8)
a.
cin-khuwa mamu
learn-NMLZ girl
b.
the learning girl

im-khuwa
sleep-NMLZ
c.
the sleeping one

si-khuwa
die-NMLZ
d.
the dying one

cek-khuwa
speak-NMLZ
e.
speaker
ep-khuwa
write-NMLZ
writer (also instruments, like pencils)

1
cuwa is perceived as uncountable mass noun, hence the marker at the verb is =ha; in contrast to the beer in a small
barrel with a pipe called toba, whi is perceived as countable and whi gets the marker =na.
67

f.
muni-khuwa yapmi
forget-NMLZ person
g.
forgetful person
khi-khuwa
be.bier-NMLZ
the bier one (e.g. fruit)

Example (9) illustrates that the nominalisation really applies to all kinds of S/A arguments, regardless of their case marking. Here, the semantic relation is one of propriation, expressed by a locative.
e only grammatical relation that was not found yet with this nominalisation is anything P-like.
(9)
kai-ga-be
wa-ya.
wa-khuba-a
me-wa-khuba
someone-GEN-LOC exist[3s]-PST exist-NMLZ-ERG NEG-exist-NMLZ-[NOM]
m-bi-n-ci-nin
3pA-give-NEG-3nsP-NEG
Some had (food). ose who had (food) did not give it to those who did not have (food).
To express negation, the prex men- ~ mi- is aaed to the verbal root. e resulting participle
or noun may also denote general, time-stable properties, illustrated by and (10).
(10)
a.
men-ni-khuwa
NEG-see=NMLZ
b.
blind person
khep-ma mi-ya-khuba
go-INF
NEG-be.able=NMLZ
lame, not able to walk

Non-nite relative clauses with object and modifying adverb are also possible with this kind of
nominalisation (see (11)). As the verb appears without any inection, the reference of the arguments and the time reference is retrieved from the lexical context.
(11)
asen
limna pi-khuwa kamniwak
yesterday sweet give=NMLZ friend
the friend who gave me the sweet yesterday
10.2 Corelative clauses

e corelative clause construction involves a headless relative clause (as in English What I did
was wrong.), whi denes a topic about whi the second part will provide a statement. e
construction requires a pronoun in the relative clause, and another pronoun in the corresponding
main clause. e pronouns can be question pronouns or demonstrative pronouns. In the main
clause, the pronoun can also be a question pronoun, or a demonstrative pronoun. e corelative
construction is relatively common in South Asian languages, and it has parallels to what Biel
(1999d) calls the Informational Dipty. is is a structurally similar strategy of information
management in discourse, where the rst part is used to announce the kind of information one
would supply, before conveying the information in the second part of the dipty. is textstructuring device is also found in Yakkha (cf. Chapter 16).
(12)
a.
ka
ikhi
nis-u-ks-u-,
ikhi
ka-me--na
1s[NOM] how.mu[NOM] know-3P-TEL-3P-1s how.mu[NOM] tell-NPST-1s=NMLZ.s
b.
I will tell as mu as I know.

na hetne camyoba wa=ya
DEM where food
68
nakha camyoba nak-se
exist=NMLZ.ns DEM food
khe-i
ask.SUP.CVB go[HORT]-1pS
10.2 Corelative clauses

Let us go where the food is, to ask for food, they said.
69
70

is apter deals with converbs and the comitative clause linkage. e comitative linkage diers
from the converbs, because it applies to a fully inected verb. But it cannot be called coordination,
because certain clause-nal markers of independent clauses, su as the mirative raea and the
verb-nal nominalisation are not found in the comitative clause linkage. e clause linkage types
discussed in this apter show strong dependencies with the main clause. Despite this close clause
linkage type, they exhibit some properties that are unexpected here, su as illocutionary marking
independent of the main clause.
11.1 e supine converb

e supine converb -se is used to express the purpose of some movement and possibly also caused
movement (in analogy to other Kiranti languages one can expect that, although examples are
needed). e main verb will always be a motion verb, and the S/A-argument of the supine clause
is identical in reference to the S/A-argument of the main verb. e marker is always aaed to
the subordinate verb that does not take any agreement morphology.
(1)
a.
yakkha ceya
cin-se
ta-ya-=na
Yakkha language learn-SUP.CVB come-PST-1s=NMLZ.s
b.
I came to learn the yakkha language.

kon-ca-se
khe-i
walk-STEM-SUP.CVB go[HORT]-1p[i]
c.
Let us go for a walk.

cama ca-se
aw-a-eba
food eat-SUP.CVB come.LEVEL-IMP-HON
d.
Please come to eat the food!

ciya-ga oom sem-se
khe-i-a
tea-GEN leaf
e.
plu-SUP.CVB go[PST]-1p-e
We went to plu tea.

paip cok-se
khe-me-=na=i
pipe do-SUP.CVB go-NPST-1s=NMLZ.s=EMPH
I will go to x the water pipe.

As mentioned in Chapter 2.2 on the phonology, verb stems ending in stops assimilate to the sibilant,
to yield a geminate. is happens only when the stem does not have a coronal augment. For
instance, whapma in example (2) has the stem form whapt, thus bloing the assimilation. Stems
like /nip-se/ however become [nisse], /sop-se/ becomes [sosse].
Illocutionary force may have scope over both clauses or only over one of them. It is oen the case
that the entity in question is only the converbal clause, as in example (2). e context is two people
meeting at a water tap, so that the main verb come must be part of the presupposition. What
cannot be answered straightforwardly however is the non-trivial question whether in clauses like
in example (2) really only the converbal clause is in the focus of the question or the converbal
clause in relation to the main clause.
(2)
tek
whap-se
ta-ya-ga=na?
clothes wash-SUP.CVB come-PST-2=NMLZ.s
71

Did you come to wash clothes?
e converbal clause and the main clause may have intervening material between them. In (3),
it is even a question word, conrming that the focus of the question may as well be on the main
clause, and the content of the converbal clause is presupposed.
(3)
tukhi ca-se
khep-ma lai?
pain eat-SUP.CVB what go-INF
PTCL
Why should I go (and marry someone) to earn pains?
11.2 e simultaneous converb

e simultaneous converb -sa connects two events that happen at the same time or during the
same period. e reference is not neccessarily the same between the arguments of the subordinate
and the main clause, as example (4b) shows. In example (4a) the converb is duplicated. e sibilant
assimilation mentioned above in section 11.1 applies here too. ere is a structural dierence
between (4a) and (4b), as in the rst example the whole converbal clause is repeated, while in the
second example one converbal marker scopes over two juxtaposed verb phrases.
(4)
a.
ten
kos-sa
kos-sa
ta-i-ci-a
village walk-SIM.CVB walk-SIM.CVB come-1p-ns-e
b.
We came, walking from village to village.

em
ca kusa
ca-sa
din khe-me=na
c.
Joking around, the day passes by.

yapmi pa-pa-be
nak-sa
song/joke eat happiness eat-SIM.CVB day go[3s]-NPST=NMLZ.s
khep-ma
people house-house-LOC ask-SIM.CVB go-INF[OBLG]
e people have to go from house to house, asking (for food).

e simultaneous converb is also used in a periphrastic aspect construction that has structural
parallels to a periphrastic continuous aspect in Nepali. e main verb khet take along is partly
grammaticalised to a marker of continuative events, although on the surface the construction looks
like the converbal clause linkage.
(5)
nha
pheri lep-sa
-khet-wa-ci-a
and.then again throw-SIM.CVB 3pA-take.along-NPST[3P]-ns=NMLZ.ns
And then again keep throwing it (the shing net).
11.3 e negation converb

e negation converb consists of the prex men- and the sux -le, aaed to the bare verb stem,
without any inection. It is used to express that one event will not take place without another
event, as shown in example (6).
(6)
a.
ka cama men-ja-le
-im-me-a-n-na
1s food NEG.CVB-eat-NEG.CVB NEG-sleep-NPST-1s-NEG=NMLZ.s
b.
I will not go to sleep without eating.

ka eptap me-khe-le
1s book
c.
iskul -khem-me-a-n=na
NEG.CVB-take.along-NEG.CVB sool NEG-go-NPST-1s-NEG=NMLZ.s
I will not go to sool without taking my books.

ke men-nan-le
-im-me-a-n=na
tooth NEG.CVB-brush-NEG.CVB NEG-sleep-NPST-1s-NEG=NMLZ.s
I will not go to sleep without brushing my teeth.
72
11.4 e comitative
e focus and scope properties are still open to investigation when more data is available, but
example (7) suggests that, in line with other Kiranti languages su as Puma (cf. Saow (2008))
the scope of a question may also rea over the converbal clause, or even only over the converbal
clause. e event stated in the main clause was already active in discourse when the following
sentence was uered.
(7)
lambu men-sop-le
way
lam-bu-ga=na?
NEG.CVB-look-NEG.CVB walk-STEM-3P-2=NMLZ.s
Did you walk without looking at the way (that you fell into a mud hole)?
e same negation prex as in the negation converb can also be aaed to innitives, as in
example (8), where the innitive has obligative reading. e obligative scopes over the negation,
yielding a prohibitive reading.
(8)
i=ya
i=ya
men-jok-ma
bani, ...
what=NMLZ.ns what=NMLZ.ns NEG-do-INF[OBLG] TOP,
As for what kinds of things he was not allowed to do,
11.4 e comitative
e comitative covers a wide semantic range, as long as one event is happening alongside the
other. It is dierent from the simultaneous and the supine converb, as it aaes to fully inected
verbs, with the exception of the nal nominaliser. e marker is homophonous with the nominal
comitative, in line with a common Tibeto-Burman paern of utilising case markers as clause linkage markers. At rst glance, it might appear to be an allomorph of the conjunction -ni while
that relates simultaneous events, as it also aaes to fully inected verbs. But, in analogy to what
we know from neighbouring languages, I assume that there are subtle dierences in syntactic and
scope behaviour, that need a closer look. ere are no constraints on coreference. In (9a), the
referents of the S/A arguments are dierent, while in (9b) it is the same referent.
(9)
a.
ne n-nis-u--nu
pauba sa=na.
here NEG-see-3P-1s-COM old.man be[PST]=NMLZ.s
b.
As I have not seen this place (for a long time), he (my uncle) got very old.
khatni=go likha ekdam cog-a-nu
cog-a-nu
bis
wora
while.that=TOP a_subclan very.mu do[3s]-PST-COM do[3s]-PST-COM twenty CLF
khibak=ca ipt-i-ci.
rope=ADD weave[3sA]-3P-ns
But the Linkha, working hard (to win the bet with the sun), wove twenty ropes.
ere is also one example from an older stage of the language, taken from Grierson (1909). e
variety documented here is about one hundred years old, and the data were collected in Darjeeling.
In the two texts that can be found in this source, the comitative is found transcribed as nuhu,
nu and nu (cf. (10)). is example also shows that the illocutionary force marker (the imperative)
does not scope over the comitative clause, so that it needs imperative marking itself.1
(10)
ani yemnu=na pik u-ya
apt-a-nu
sis-u
and fat=NMLZ.s cow 3sPOSS-ild bring[3sP]-IMP-COM kill[3sP]-IMP
Bring a fat cow and kill it.
An interesting detail of this example is the detransitivised verb in the comitative clause, while the main verb is not
detransitivised. is might however also be due to mistranscription. e missing genitive on the possessor in the
possessive construction is also unusual.
73
74

e conditional conjunction is bo ~ bho, combining full-edged independent clauses. On the
other hand, innitives are also exceptionally frequent in conditional constructions, but there is no
explanation for them yet. Examples can be found in (1). e proposition in the condition provides
a baground for the proposition in the main clause.
(1)
a.
ep-ma-bho
mu-ni-me-n=na.
men-ep-ma-bho
write-INF-COND forget-STEM[3s]-NPST=NMLZ.s NEG-write-INF-COND
mu-ni-mya
forget-STEM[3s]-NPST
b.
If one writes down, one will not forget it. If one does not write it down, one will
forget it.
emma u-ma-bo
kam cok-ma n-yas-a-an=na
raksi
c.
drink-INF-COND work do-INF NEG-be.able-1s-NEG=NMLZ.s
If I have to drink raksi, I cannot do my work. (why PST?)

batti
n-da-me-n=na.
wani halo.
electricity NEG-come-NPST-NEG=NMLZ.s O.K. by.the.way
n-da-me-n-na-bo,
n-da-nin-ni
halo
NEG-come-NPST-NEG=NMLZ.s-COND NEG-come-NEG-OPT by.the.way
e electricity does not come. Its O.K., by the way. If it does not come, may it not
come
e particle bo could be related to another topic marker and reportative bani (literally while
saying), in parallel to the Nepali markers bhne, bhner, bhndi. e morpheme bo ~ bho
~ ba must have its origin in a verbal root. In other Kiranti languages (e.g. Puma), we also nd
a verb bma with the meaning talk, say, whi was not found yet in Yakkha (where it would
have to be pma or phma, according to the sound anges). Another parallel to the Nepali
conjunction bhne is the usage of bo as complementiser in embedded clauses (cf. example (2),
repeated from section 9.1.2).
(2)
ak-ka
kamniwakci, ka isa om, isa om-bo u-ci-a=ca
1sPOSS-GEN friend-ns
1s who COP who COP-REP 3-ns-ERG=ADD
n-yokme-me-ka-ni-n
NEG-tell-NPST-2-p-NEG

In irrealis contexts, when unfullled conditions are involved, the counterfactual clitic =pi is
aaed to the verb of the main clause, and optionally also to the clause part that contains the
condition, where it is oen preceded by other clause linkers su as simultaneous -ni and sequential -ho, and by the topic particle =go. e verbs in these clauses either take subjunctive or
past inection. is is not claried yet. ey may also appear in the innitive, like in conditional
clauses.
(3)
a.
ka nis-u-=ni=bi
ikhi
lu=bi
1s know[SBJV/PST]-3P-1sA-while=CTFT how.mu tell=CTFT
If I knew it, how mu would I tell
75

b.
khibela-a=go, nasa tok-ma-ci=ho=go=(bi),
diana=ca pip-ma
this.time-ERG=TOP, sh get-INF-ns=SEQ=TOP=(CTFT), D.=ADD give-INF[OBLG]
sa=bi
be.PST=CTFT
c.
If there was sh available these days, we had to give it to Diana also.

diana=ca pip-ma=ho=go=bi
cond-a-sy-a=bi=ba
D.=ADD give-INF[OBLG]=SEQ=TOP=CTFT be.happy-PST-STEM-PST=CTFT=PTCL
Aer giving them to Diana too, she would have been happy!
76
13.1 Nonnite complement clauses
Various verbs in Yakkha can embed verbal complements, and in all that are found so far it is
possible for a lower (i.e. an embedded) argument to trigger agreement in the matrix verb, hence,
to be part of the argument frame of the matrix-verb. For some verbs this is obligatory, for others,
I have reason to assume raising constructions. is phenomenon is also known as long-distanceagreement (henceforth LDA). It is found for instance in Romance languages, in Hindi/Urdu (Bu
2008), Tsez (Polinsky and Comrie 1999), Godoberi (Haspelmath 1999), Belhare (Biel 2004a) and
Puma (Saow 2008). Within the Kiranti languages, this type of agreement is not uncommon.
e complement verbs that are found so far in Yakkha are yama be able, sima IPVF, tarokma
start, mitma like doing, pipma allow, kama agree, tokma get the ance, memma ~ ment
nish, cokma try and lema as honoric marker, in parallel to the Nepali huna-forms. e
behaviour of the dierent complement verbs is far from homogenous, as was shown for Puma in
Saow (2008). In Puma, some of them assimilate in valency to their embedded verb, some have
LDA only with certain embedded verbs, some have LDA only with certain argument types, e.g.
with P, but not with G. e precise description and analysis of the behaviour of complement verbs
is crucial for understanding the syntax of Kiranti languages. ese agreement paerns may show
ergative alignment, as will be shown later.
In (1) below we nd some examples. In (1a), agent (1s) and patient (2) both trigger agreement
in the matrix verb, while the embedded verb is in the innitive form. With an embedded ditransitive verb, as in (1d), however, the expected agreement with the recipient (2s you) does not take
place. e object agreement slot in the matrix verb takes the default option, third person singular
(-u). is could either mean that the matrix verb does not agree with any embedded argument,
or that it agrees with the theme, money, whi is however unexpected, given the usual Kiranti
agreement paern of secundative alignment (i.e. with the goal argument in ditransitives). In all
the other examples, the matrix verb agrees with the embedded object.
(1)
a.
nda ka im-ma ya-me--ga=na
b.
You can ask me.

sop-ma si-me-nen=na
c.
I am looking at you.
nasa-ci um-ma ta-rokt-wa-ci=ya
2s
1s ask-INF be.able-NPST-1s-2=NMLZ.s
look-INF IPFV-NPST-1>2=NMLZ.s
sh-ns pull-INF start[3sA]-STEM-NPST-ns=NMLZ.ns
d.
He starts pulling out the sh.

ka nda ya pip-ma ta-rokt-u-=na
1s 2s
e.
money give-INF start[PST]-STEM-3P-e=NMLZ.s
I started giving you the money.

sop-ma m-mim-me-nen=na
look-INF NEG-like-NPST-1>2=NMLZ.s
f.
I do not like looking at you.

emd-u-ci,
ab-a-ho
cigik-a ap-ma
cog-u-ci
tease[PST]-3P-ns, come-PST-SEQ bow-INS hunt.with.bow-INF do[PST]-3P-ns
77
She teased them, she came and tried to shoot them with an arrow.
As was said before, the long distance agreement is even obligatory in some verbs. In the example
(2), with tokma as matrix verb, default 3s agreement is not grammatical, i.e. this is not a raising
construction.
(2)
a.
ka mu-ci
im-ma tokt-u--ci-
1s mushroom-ns buy-INF get.ance[PST]-3P-1s-ns-1s
I got the ance to buy mushrooms.

b. *ka mu-ci
im-ma
tokt-u-
1s mushroom-ns buy[PST]-INF get.ance-3P-1s
Intended: I got the ance to buy mushrooms.

Among the complement verbs, there is one special case. Syntactically, the construction behaves
like a complement construction, but the matrix verb is zero. e obligative have to is expressed
only by an innitive, without any matrix verb. e intive takes nonsingular number agreement
with P-arguments, being aligned to the usual paern of primary object (hence, agreeing with the
number of P and G). Examples are provided in (3).
(3)
a.
yapmi pa-pa-be
nak-sa
khep-ma
people house-house-LOC ask-SIM.CVB go-INF[OBLG]
b.
e people have to go from house to house, asking (for food).

ca-ma-ci
eat-INF-ns
c.
ey have to be eaten.
wama-ci pip-ma-ci
ien-ns give-INF[OBLG]-ci
(e fodder) has to be given to the ien.

e complement verb lema (lit.: be, be alright) is interesting insofar as the whole construction
is apparently a loan translation from Nepali. In Nepali, any verb can be inected for honoric
status by adding huna, the third person form of the verb be to an innitive. For instance, garnu
huna is the honoric way of saying he does. e same is possible with lema in Yakkha, as
shown in example (4) below.
(4)
a.
hetne khep-ma leks-a-ga=na

where go-INF
b.
be-PST-2=NMLZ.ns
Where did you go?

hetna
ta-ma leks-a-ga=na
where.from come-INF be-PST-2=NMLZ.ns
Where did you come from?
13.2 Finite complement clauses

Finite complement clauses usually occur with a complementiser. So far, only the conditional
conjunction bo was found in this position. e same paern is known from Nepali, where the
conditional bhane is also used as complementiser in the same kind of constructions, for instance
with verbs of saying and thinking (cf. example (5)).
(5)
ak-ka
kamniwakci, ka isa om, isa om-bo u-ci-a=ca
1sPOSS-GEN friend-ns
n-yokme-me-ka-ni-n
NEG-tell-NPST-2-p-NEG
78
1s who COP who COP-REP 3-ns-ERG=ADD
13.2 Finite complement clauses

79
80

14.1 Sequential linkage
e sequential clause linkage combines clauses that refer to subsequent events. ese are -yo,
-ho and -ma. e rst two forms can very well be allomorphs, as the [y] ~ [h] allophony is
found elsewhere in the language, for instance in the plural nominaliser. e last form however
might be dierent, but the data are not sucient to answer this question. e sequential linker
is aaed to the inected verb. Diaronically, the sequential linker might be connected to the
ablative marker -pa. Generally, for Tibeto-Burman we oen nd the same morphemes applying
to noun phrases and clauses, i.e. the same morphemes for case and for clause linkage markers.
e sequential clauses found so far exhibit referential identity of the S and A arguments of both
clauses. e possessor in (1b) is regarded as A argument semantically.
(1)
a.
kamniwak sori
friend
b.
yu-i-yo u-u-m
together sit-1pS-SEQ drink-3P-1/2pA
Having sat down together, let us friends drink.

cama pyak ca-i-ho a-phok
kei loa leks-a-by-a=na
food/rice mu eat-TEL-SEQ 1sPOSS-belly drum like become-PST-TEL-PST=NMLZ.s
c.
Aer eating mu rice, my belly became like a drum.

pyak cuwa u-ma mund-i-wa-m=ha
mu beer drink-SEQ forget-3P-NPST-1pA=NMLZ.ns
d.
Aer drinking mu beer, we will forget it.

yo -khem-mya-ho -und-wa-ci=ha
there 3pS-go-NPST-SEQ 3pS-pull.towards.self-NPST-ns=NMLZ.ns
e.
ey go a bit further and then they pull them (the sh) out.
uncas
sal-be ta-ya-=ho
ne we-a=na
thirty-nine year-LOC come-PST-1s=SEQ here stay[PST]-1s=NMLZ.s
Having come here in the year thirty-nine, I stayed here.

e domain of the sequential is not restricted to verbs. It may aa to verbs, demonstratives and
also adverbs, as the example (2) illustrates. Here, it aaes to the distal demonstrative kha (cf.
section 4.2.3), yielding kho. But it is never used to conjoin two nouns, whi is always done
with the comitative (cf. example (2b)) or with juxtaposition.
(2)
a.
pheri nam=a lond-a-khy-a,
kh-o
likha
baji-pe
har-a
again sun=ADD come.out[3s]-PST-TEL-PST, DIST-SEQ a_subclan race-LOC lose-NATIV
cog-a-khy-a
do[3s]-PST-TEL-PST
b.
e sun rose again, and so the Lingkha lost the bet.

minuma nu mima
cat
COM mouse
the cat and the mouse

e sequential linker is frequent in narrations, where events are recalled in a linear order. e
passage in example (3) illustrates this nicely, with four occurrences of the sequential marker.
81

(3)
nha
hattabatta lukt-ab-a-ho
muok-ho
pa-be
and.then very.quily run[3s]-come.LEVEL-PST-SEQ carrying.in.front-SEQ house-LOC
ta-ya,
nhaha pa-be
ho pheri lukt-ab-a-ho
sidak
come-PST, and.then house-LOC SEQ again run[3s]-come.LEVEL-PST-SEQ medicine
end-a-bhy-a-ba
apply[3sA]-PST-BEN-PST-1s-PTCL
en, he came running quily, he lied me up and went into the house. And then, as (we
were) in the house, he came running again and applied medicine on my wounds.
14.2 Simultaneous linkage

Two clauses with simultaneous time reference can be combined with the verbal sux -ni ~ nia. I am not aware yet whether the dierent forms are related to any dierences in semantics
or structure. So far, they are treated as allomorphs. e sux can be aaed to the verb, following
the subjunctive (or past) inection (cf. example (5)), or to innitives of the verb, that maximally
agree with the number of the P argument of the clause (cf. (4b)). e related events may also
proceed continuously over a period of time, as in (4a). e examples collected so far suggest that
there is no constraint on coreference. As for the inected example in (5), it is not clear yet whether
the subjunctive or the past inection are used obligatorily, or if others, like the indicative nonpast
for instance, are also possible.
(4)
a.
uthamla uimala lam-ma-nia la-ci n-sa-mya-ci
b.
While walking up and down, the legs get stronger.

khamboma sem-ma-ci-ni, u-ti-a
hot-ni-ma
c.
While pluing the berries, their thorns pried me.

lagura-ci lam-ma-nia n-du-me-a=na-ci,
steep.up
berry
knee-ns
steep.down walk-INF-while leg-ns 3pS-become-NPST-ns
plu-INF-ns-while 3sPOSS-thorn-INS pri-STEM[3s,PST]-PERF
yu-ma-nia
walk-INF-while NEG-ae-NPST=NMLZ.ns-NEG-ns, sit-INF-while
n-du-me-a-ci
3pS-ae-NPST=NMLZ.ns-ns
While walking, (my) knees do not ae, while siing, they ae.1
e linker may also host the topic particle =go, whi is not surprising as the content of the linked
clause is grounding the main assertion in time (cf. (5)).
(5)
plestik-be, plestik-ka jhola-be in-ca-met-uks-u--ni=go, inh cog-a-ma-ssa- wa=ba

plastic-LOC plastik-GEN bag-LOC play-STEM-CAUS-TEL-3P-1s-while=TOP interj do-PSTPERF-PST.PERF-1s maybe=PTCL
While playing (with the sh) in a plastic, a plastic bag, I shrieked Eek!, probably.
14.3 Causal conjunctions

e conjunction ijabani because is used for causal clause linkage. It is constructed in parallel
to the Nepali kina bhne, out of the interrogative word why and a topic marker. An example is
provided in (6).
1
e plural form is strange here, and could be due to an ongoing neutralisation in the verbal paradigms.
82
14.4 e reportatives and text structure

(6)
ka lop caklet n-ja-wa-a-n=na
ija-bani toba-nu
1s now sweets NEG-eat-NPST-1s-NEG=NMLZ.s why-saying beer-COM
n-do-me-n=na
NEG-agree-NPST-NEG=NMLZ.s
I do not eat the ocolate now, because it does not go with the beer.
14.4 e reportatives and text structure

ere are several reportative particles, employed for marking spee, but also for marking topics,
and structuring the text. e rst to be discussed here is bani, whi literally means while
saying, and it is oen used to introduce a topic of a longer passage. In (7), all examples stand at
the beginning of a longer description or enumeration. e particle is also lexicalised in the word
for because (cf. the previous section).
(7)
a.
lakt-a-ni
imin lak-ma
bani
...
dance-(?)-while how dance-INF while.saying
b.
While dancing, as for how to dance,

na-ga suru
imin leks-a=na
bani,
...
this-GEN beginning how be-ASP=NMLZ.s while.saying,
c.
As for how she (this one) originated,

i=ya
i=ya
men-jok-ma
bani,
...
what=NMLZ.ns what=NMLZ.ns NEG-do-INF[OBLG] while.saying,
d.
As for what kinds of things he was not allowed to do,

haku
ina bani,
...
from.now.on what while.saying
What comes next,

e genuine reportative particle is =bu, with the function to disburden the speaker of the responsibility for the truth value of the uerance, as is done in English by the adverb allegedly. It
was frequently found in a narrative that was transmied orally. e particle always aaes to
the inected (and optionally nominalised) verb.
(8)
a.
ani-ga
likha-ga
uhile
utpatti mamli-be
1pePOSS-GEN a_subclan-GEN long.time.ago origin a_village-LOC
leks-a=na=bu
happen-PST=NMLZ.s-REP
b.
Our Lingkha clan originated long ago in Mamling, they say.

eko cahi lukt-a-khy-a=na=bu
one TOP run-PST-TEL-PST=NMLZ.s-REP
c.
One (cow), it ran away.

na
yo-be=ca
me-im-ma-ci=bu,
likha-ci-ga
and.then bamboo-LOC=ADD NEG-sleep-INF[OBLG]-ns-REP, a_subclan-ns-GEN
And they are not allowed to sleep in a bamboo basket, the Linkhas (ildren).
83
84
15.1 Coding properties
In this section, the alignment paerns in morphology will be summarised. For Yakkha, this includes case, number marking and agreement paerns. Both ergative and accusative paerns can
be found in the morphology, and with regard to object agreement, both indirective and secundative alignment paerns can be found. eir distribution is as follows: e basic cases for core
arguments are the nominative and the ergative, while the ergative is not found on A arguments
that are coded by rst or second person pronouns (cf. (1)). e nominative marks the S arguments
of intransitive verbs, the rst and second person A arguments (if pronouns) of transitive verbs
and all types of P arguments (cf. (1c)). Hence, the case alignment is split according to person and
referential status, in line with the expectations of the referential hierary that was introduced by
Silverstein (1976). ere is no particular object case in Yakkha, but the comitative and the locative
may also serve to mark oblique arguments.
(1)
a.
uci-a u
mo-me-c-u=na
3ns-ERG 3s[NOM] beat[3pA]-NPST-ns-3P=NMLZ.s
b.
ey beat him.
ka
u
mo-mya-=na
1s[NOM] 3s[NOM] beat-NPST-1sA[>3P]=NMLZ.s
c.
I beat him.
ka
nda
coklet
pi-nen=na
1s[NOM] 2s[NOM] sweet[NOM] give-1>2=NMLZ.s
I gave you a sweet.

e alignment of the ergative case only excludes rst and second person pronouns, not rst
or second person reference in general. A second person A argument not represented by a pronoun is possible in Kiranti because the agreement here works along dierent parameters than in
Indo-European languages. It does not always have to be identicational (cf. example (2)). More
examples on exible agreement are provided in section 5.3 on page 36.
(2)
a-koma-a=le
ta-ga=na
raea
1sPOSS-aunt-ERG=PTCL bring[PST]-2[>3P]=NMLZ.s MIR

If we look at person and number agreement, we can nd dierent alignment types that are distributed according to person. e Kiranti family exhibits a very robust alignment paern for
agreement, that is similar throughout the single languages. It was described and discussed in
Biel (2008b) and can be aracterised basically as follows: e rst person is aligned ergatively,
the second person is neutrally aligned and the third person exhibits accusative alignment. Not all
agreement morphemes are part of this paern in Yakkha. Naturally, the portmanteau-morpheme
coding -nen 1>2 cannot be part of it. e exclusive morpheme - ~ -a and the dual -ci are also
neutrally aligned (for an exhaustive list of agreement axes cf. apter 5.2). Table (2) summarises
the relevant morphemes. e Yakkha language is deviant from the usual Kiranti paern in that
there is no morpheme only for rst person that is ergatively aligned. e ergative paern applies
to two suxes that code plural agreement with rst and with second person. is alignment split
85
stands in opposition to the expectations of the referential hierary as rst introduced by Silverstein (1976), if its range of application is interpreted as including not only case, but also agreement
marking.
1/2
2
3
A
S
P
-m 1/2pA
-i 1/2pS/P
-ka 2
N- 3pS/A
-u 3P
-ci 3nsP
Figure 15.1: Alignment of agreement
As was shown earlier, the alignment of agreement in Yakkha (and Kiranti in general) is secundative (agreement with the primary object, cf. apter 5.2), at least if both objects of the ditransitive
verb are marked with the nominative. I expect other valency classes, but so far, I have no data
to support this claim. e nominalisers =na (for singular) and =ha ~ =ya (for nonsingular) are an
exception to the secundative paern. ey aa to the verbs frequently and have the following
distribution. ey agree with the S and P arguments, i.e. when the S or P argument is singular, =na
will be used, if nonsingular (or uncountable mass noun), =ha ~ =ya will be used. e examples in
(3) shall illustrate this. e paern can also be observed in the armative forms in the paradigm
tables in apter 5.2. e negative forms have a sux -na, apparently in the same slot, as the
nominaliser never aaes there. An exception yet unexplained are the scenarios of rst person
acting on second in the transitive paradigm, where exclusively =na is found, also with nonsingular P arguments. It is a general observation that the singular marker is also used in nonsingular
contexts in natural spee. It seems to be the unmarked option.
(3)
a.
ka khe-me-=na
1s go-NPST-1s-NMLZ.s
b.
I go.
kanci khe-me-ci=ha
1d
c.
go-NPST-ns[i]-NMLZ.ns
We go.
n-dund-wa=na
3pA-understand[3P]-NPST=NMLZ.s
d.
ey understand him.
n-dund-wa-ci=ha
3pA-understand-NPST-[3]ns-NMLZ.ns
ey understand them.
In ditransitive scenarios, the oice of the nominalisers corresponds to the number of the T
argument, and not the G argument, i.e. they follow indirective alignment, as shown in example
(4). e sequence -ga=ha is contracted to a voiced aspirated velar stop [gh].
(4)
a.
eko dotpen pi-a-ga=na

one dotpen give[3sA]-PST-2=NMLZ.s
b.
He gave you one dotpen.

cama pi-a-g=ha
food give-PST-2=NMLZ.ns
He gave you the food.
86

is apter shall introduce the dierent devices to structure the information in the uerances
and texts. It is not an analysis of these devices, but rather a description. e exact establishment
of the pragmatics and scope of all the particles treated here deserves a study in its own right.
e topic particle =go cliticises to phrases but also to any word in a phrase, whether noun, verb
or adverb, as we can see in (1a) where it aaes to an entity within a noun phrase.
(1)
a.
uk-ka=go
pik-ci wa-i-sa=bu
3sPOSS-GEN=TOP cow-ns exist-STEM-PERF=REP
b.
As for him, he had cows.

ka=go nala paip cok-se
khe-me-=na=i
1s=TOP pipe pipe do-SUP.CVB go-NPST-1s=NMLZ.s=EMPH
c.
As for me, I will go to x the water pipe.

macwa a-me=na=go,
yama
water
d.
come-NPST=NMLZ.s=TOP auntie
It is raining though, auntie.

ako=go bura leks-a-khy-a=na
hola
now=TOP old.man become-PST-TEL-PST=NMLZ.s probably
Now he probably became an old man.

e topic particle is also part of the lexicalised adversative expression in (2).
(2)
khat-ni=go
that-while=TOP
but
Restrictive focus is expressed by the particle =se. It is aaed to whatever it has scope over.
(3)
a.
pa=se
sop-ma=na
house=only wat-INF[OBLG]=NMLZ.s
b.
Only the house has to be wated.

ka=go na mamu hen=se
nis-wa-=na
1s=TOP this girl
c.
today=only know-NPST-1s=NMLZ.s
But I know this girl only since today!

hokhala-o, hokhi=se
like.that-SEQ
that.mu=only
Like that, that mu only.

e additive focus is marked by the clitic =ca. In negation, it is oen aaed to interrogative
words to express negation to the most possible extent, as in English any (cf. example (4a)).
(4)
a.
i=ya=ca
cok-ma n-ya-me-n=na
what=NMLZ.ns=ADD do-INF NEG-be.able-NPST-NEG=NMLZ.ns
b.
One cannot do anything about it.

nda-nu ka=ca lo-me-=na
2s-COM 1s=ADD come.along-NPST-1s=NMLZ.s
I will also accompany you.
87

c.
mo-khala=ca uks-a-ni
down-ALL=ADD come.down-IMP-p
Come down too.

In clause linkage, the additive focus can be added to the sequential linker to express a concessive
relation of two propositions (cf. example (5)).
(5)
marej end-a-ho=ca
khumdu sa
mustard apply[3sA]-PST-SEQ=ADD tasty
be[PST]
Even though one applied mustard, it was tasty.

e two particles to be discussed next oen occur together, when imperatives are involved.
ey are =i ~ [e] and =pa ~ [ba] ~ [wa] (in fast speea). In imperatives, adding -pa makes the
imperative more polite, while the aament of -i adds emphasis to it, similar to the Nepli particle
-n:
(6)
aw-a=i=wa
come.LEVEL-IMP=EMPH=PTCL
Please come.
Both particles are also found in single usage. Particle -i is oen found aaed to questions,
negated sentences and contrastive focus, as in (7) below:
(7)
a.
menna=i, paip cok-se
khe-me-=na=i
no=EMPH, pipe do-PURP.CVB go-NPST-1s=NMLZ=EMPH
b.
No, I will go to x the pipe.

raksi-a sis-a-ga=na=i?
spirit-ERG kill[3sA]-PST-2=NMLZ.s=EMPH
Are you drunk?

Particle -pa, if used without -i, can also be understood emphatically. Please note the intervocalic
gemination.
(8)
hen khi=se=ppa
today this.mu=only=PTCL
Today this mu only.

Both particles are also translatable with the Nepali emphatic particle -i ~ -ni ~ -nai. e question
whether -i is a Nepali loan can be negated, as it has cognates in the Limbu and Belhare question
markers (B. Biel, p.c.). Particle -pa is also found on questions, but not on negated sentences. In
example (9), it is aaed to a clause that contains an armative answer. Probably, both particles
express dierent nuances. Particle -pa carries an armative notion, and in questions it expresses
the expectation of the speaker to get a positive answer. It aaes to inected verbs, but also to
noun phrases (cf. (9c)).
(9)
a.
na-be
yu-ma le-me=pa?
le-me=pa/
this-LOC sit-INF be.alright-NPST=PTCL be.alright-NPST=PTCL
le-me-n=na
be.alright-NPST-NEG=NMLZ.s
b.
Is it O.K. to sit here? It is O.K./ It is not O.K.

om=ba!
COP=PTCL
c.
Yes!
a-pa-a
et-u-ci=ba,
samundra-be=pa
1sPOSS-father-ERG hunt.sh[PST]-3P-ns=PTCL ocean-LOC=PTCL
88
My father went shing, in the ocean.

e particle =lai is also an emphatic marker, sometimes carrying a contrastive meaning, but not
necessarily.
(10)
a.
ikhi
luma-kipt-u-a=lai
how.mu liver-dig[3sA]-3P-PST=PTCL
b.
How greedy he is!

om=lai!
COP=PTCL
c.
Yes of course! - ho ni!

hetne wei-ka=na? pa-be=lai.
tukhi ca=se
khep-ma=lai?
where live-2=NMLZ.s house-LOC=PTCL pains eat-PURP.CVB what go-INF=PTCL
d.
Where do you live? - At home, of course. Why should I go to suer (in a marriage)?
n-da-ci=bo=go
im-m-ha-ci=lai
3pA-bring-ns=COND=TOP buy-INF[OBLG]=NMLZ.ns-ns=PTCL
ca-m-ha-ci
eat-INF[OBLG]=NMLZ.ns-ns
e.
If they bring some (sh), we will have to buy them and eat them.
khala in-ca-met-uks-u-=ni=go=lai
...
f.
Just playing with the sh, like this,

koi khala=lai
like.this play-STEM-CAUS-TEL-3P-1s=while=TOP=PTCL
some like.this=PTCL
Some are just like this.

Another particle is =le, translatable with Nepli po, whi always has a contrastive notion. As
we see in (11b), the particle also occurs together with the Nepali mirative raea, expressing the
unexpectedness of the proposition.
(11)
a.
hako
im-me-=na=le
from.now.on sleep-NPST-1s=NMLZ=PTCL
b.
Now I will sleep.

a-koma-a=le
ta-ga=na
raea
1sPOSS-aunt-ERG=PTCL bring[PST]-2[>3P]=NMLZ.s MIR

e particle =le is also found in a xed expression, a kind of ller:
(12)
i=na=le=ba
what=NMLZ.s=PTCL=PTCL
what to say/ watamacallit

e phonological boundedness is also evidenced by a gemination that can be found in some instances (cf. (13)).
(13)
a.
yo=lle
khe-ks-a-ga=na?
DIST=PTCL go-TEL-PST-2=NMLZ.s
b.
So you went there?- para po jandaiau

khe=lle
PROX=PTCL
here
e Nepli mirative rhe was borrowed into Yakkha, but the mirativity does not seem to be
very strong. In example (14b), the proposition cannot be an entirely new discovery for the speaker.
89

(14)
a.
macwa mi=na
water
b.
rahea
small=NMLZ.s MIR
e water/rain got less.

pena-be wa-i-m-i-a
raia, kani
a_place-LOC live[PST]-1p-PERF-1p-e MIR
1p
We lived in Singapur, we. (refering to times when the speaker was too young to
remember)
e particle halo is used frequently in colloquial spee, meaning something like by the way.
It is oen reduced to =lo, when it aaes to a host with open syllable structure (cf. (15)). Example
(15d) shows that it is also used in questions.
(15)
a.
ka=go jokor=le=lo
1s=TOP self.deciding.person=PTCL=PTCL
b.
I can decide for myself, by the way.

wani=halo. n-da-me-nna=bo,
n-da-ni-ni=halo
O.K.=by.the.way NEG-come-NPST-NEG=COND NEG-come-OPT-NEG=by.the.way
c.
It is O.K., by the way. If it (the electricity) does not come, may it not come, by the
way!
kha=go=le=lo
this=TOP=PTCL=PTCL
d.
It is this one!
hetna
ta-i-ka=na=lai=halo?
e.
Where did you come from, by the way?

pi-a-ks-a=ba=lo
where.from come-STEM-2=NMLZ.s=PTCL=by.the.way
give-PST-TEL-PST=PTCL=by.the.way
ey gave me away (in marriage), just like that.

ere is a dialectal variety =ko of the above mentioned particle, found for instance in the dialect
spoken in Ankhinbhuin village. e question in (15d) sounds like in example (16) in that variety.
(16)
hetna
ta-i-ka=na=lai=ko?
where.from come-STEM-2=NMLZ.s=PTCL=by.the.way
Where did you come from, by the way?

e particle =ma is probably a sequential linker used for emphasis. In example (17), the emphasis lies on the fact that the speaker gives in into an undesired circumstance. e grammaticalised
expression of this state of mind is also refered to as frustrative. As one allomorph of the Yakkha sequential linker was also [ma], and the usage of sequential linkers in this sense is found in Nepli
as well as other Kiranti languages, this interpretation is very plausible, though more examples are
needed.
(17)
cok-ma? yakkha ten-be
what do-INF
khalo=ma
Yakkha village-LOC like.this=FRUST
What to do? Aer all, it is like this in a Yakkha village! (referring to low infrastructure)
e expression of two alternatives is coded by the particle em. It is aaed to both of the
conjoined phrases, as exemplied in (18):
(18)
a.
laka=em lima=em
salty=or sweet=or
Is it (the tea) salty or sweet?
90
b.
om=em men=em?
COP=or COP.NEG=or
Yes or no?
91
92
17 Bibliography
Biel, Balthasar (1993), Belhare subordination and the theory of topic, in K. H.Ebert, ed., Studies
in clause linkage, ASAS, Zri, pp. 23 55.
Biel, Balthasar (1996), Aspect, mood, and time in Belhare. Studies in the semantics-pragmatics
interface of a Himalayan language, Vol. 15 of ASAS, Seminar fr Allgemeine Sprawissensa.
Biel, Balthasar (1998), Rhythm and feet in Belhare morphology, Working Paper No. 287, Rutgers Optimality Arive; http://roa.rutgers.edu .
Biel, Balthasar (1999a), Cultural formalism and spatial language in Belhara, in B.Biel and
M.Gaenszle, eds, Himalayan space: cultural horizons and practices, Zri, pp. 75 104.
Biel, Balthasar (1999b), Grammatical relations, agreement, and genetic stability. Ms., University
of California, Berkeley; hp://www.uni-leipzig.de/ biel/resear/papers.
Biel, Balthasar (1999c), Nominalization and focus constructions in some Kiranti languages, in
Y. P.Yadava and W. W.Glover, eds, Topics in Nepalese linguistics, Royal Nepal Academy, Kathmandu, pp. 271 296.
Biel, Balthasar (1999d), Principles of event framing: genetic stability in grammar and discourse.
Ms., University of California, Berkeley; hp://www.uni-leipzig.de/ biel/resear/papers.
Biel, Balthasar (2000), On the syntax of agreement in Tibeto-Burman, Studies in Language
24, 583 609.
Biel, Balthasar (2003), Belhare, in G.urgood and R. J.LaPolla, eds, e Sino-Tibetan languages, Routledge, London, pp. 546 70.
Biel, Balthasar (2004a), Hidden syntax in Belhare, in G.urgood and R.LaPolla, eds, Himalayan
languages, past and present, Mouton de Gruyter, Berlin.
Biel, Balthasar (2004b), e syntax of experiencers in the Himalayas, in P.Bhaskararao and
K. V.Subbarao, eds, Non-nominative subjects, Benjamins, Amsterdam, pp. 77 112.
Biel, Balthasar (2008a), Kiranti: an introduction, Seminar Paper, Universitt Leipzig.
Biel, Balthasar (2008b), On the scope of the referential hierary in the typology of grammatical
relations., in G. G.Corbe and M.Noonan, eds, Case and grammatical relations: papers in honor
of Bernard Comrie, Benjamins, Amsterdam.
Biel, Balthasar, Goma Banjade, Martin Gaenszle, Elena Lieven, Netra Paudyal, Arjun Rai,
Iha P. Rai, Manoj Rai, Novel K. Rai, Shree Kumar Rai, Vishnu S. Rai, Narayan P. Gautam (Sharma), Sabine Stoll and Mark Turin (2006), e Chintang and Puma Corpus. DOBES
Multi-media Corpus, www.mpi.nl/DOBES.
Biel, Balthasar, Goma Banjade, Martin Gaenszle, Elena Lieven, Netra Paudyal, Ica Rai, Manoj
Rai, Novel K. Rai and Sabine Stoll (2007), Free prex ordering in Chintang, Language 83, 43
73.
93
17 Bibliography
Biel, Balthasar and Martin Gaenszle (2007), Generics as First Person Undergoers and the Political
History of the Southern Kirant. Paper presented at the 7th Biannual Meeting of the Association
for Linguistic Typology, Paris, September 26, 2007.
Biel, Balthasar, Martin Gaenszle, Arjun Rai, Shree Kumar Rai, Vishnu S. Rai, Diana Saow,
Sabine Gnther and Narayan P. Gautam (Sharma) (in prep.), Puma-Nepali-English Dictionary,
Chintang and Puma Documentation Project, Kathmandu.
Bu, Miriam (2008), Revisiting Long-Distance Agreement in Urdu. Talk given at the 30th Annual
Convention of the German Society of Linguistics, Bamberg.
Central Bureau of Statistics, Nepal (2001), Population census report.
Dewan, Gopal, Anupa Dewan, Santa Dewan and Jimi Radha (2002 (VS 2059)), Opyongme
(Palam Rang), Kirat Yakkha Chumma, Dharan, Sunsari.
Dixon, R. M. W. (1994), Ergativity, Cambridge University Press, Cambridge.
Doornenbal, Marius Albert (2009), A grammar of Bantawa, LOT Publications, Utret, e
Netherlands.
Ebert, Karen (2003), Camling, in G.urgood and R.LaPolla, eds, e Sino-Tibetan languages,
Routledge, London, pp. 533 545.
Ebert, Karen H. (1993), Kiranti subordination in the South Asian areal context, in K. H.Ebert, ed.,
Studies in clause linkage, Zri, pp. 83 110.
Ebert, Karen H. (1994), e structure of Kiranti languages, ASAS Press, Zri.
Ebert, Karen H. (1997), A grammar of Athpare, LINCOM Europa, Mnen.
Ebert, Karen H. (1999), Nonnite verbs in Kiranti languages an areal perspective, in Y. P.Yadava
and W. W.Glover, eds, Topics in Nepalese Linguistics, Royal Nepal Academy, Kathmandu.
Ebert, Karen H. (2003), Equivalents of conjunctive participles in Kiranti languages, in
T. R.Kansakar and M.Turin, eds, emes in Himalayan languages and linguistics, South Asia
Institute and Tribhuvan University, Heidelberg and Kathmandu, pp. 27 48.
Grierson, George A. (1909), Tibeto-Burman family, Part I, General Introduction, specimens of the
Tibetan dialects, the Himalayan dialects and the North Assam group., Vol. Vol. III of Linguistic
Survey of India, Superintendent of Government Printing, India, Calcua.
Haspelmath, Martin (1999), Long-distance agreement in Godoberi (Daghestanian), Folia Linguistica 33, 131 151.
Hildebrandt, Kristine (2007), Prosodic and Grammatical Domains in Limbu, Himalayan Linguistics Journal 8, 1 34.
Jimi, Indira, Visvakaji Kongren and Manita JImi (2009), Engka Yakkha Cheptap, Siksa tatha
Khelkud Mantralaya, Sanothimi, Bhaktapur.
Kongren, Ramji (2007a), Yakkha Jatiko Samskar ra Samskriti (Yakkha Indigenous Peoples Tradition and Culture), Kirat Yakkha Chumma, Kathmandu.
Kongren, Ramji (2007b), Yakkha-Nepali-English Dictionary, Kirant Yakkha Chumma, Indigenous
Peoples Yakkha Organization, Kathmandu.
94
17 Bibliography
LaPolla, Randy J. (1992), On the dating and nature of verb agreement in Tibeto-Burman, Bulletin
of the Sool of Oriental and African Studies 55, 298 315.
Linkha, Magman and Bam Dewan (2005 (VS 2064)), Yakkha Ceya Sikla - Prarambhik Yakkha
Sabda Samgraha, yakkha edn, Kirant Yakkha Chumma, Indigenous Peoples Yakkha Organization, Sunsari.
Montaut, Annie (2004), A Grammar of Hindi, Lincom Europa, Muenen.
Pokharel, Madhav P. (2001), Compound verbs in Nepali, Royal Nepal Academy, pp. 185 208.
Polinsky, Maria and Bernard Comrie (1999), Agreement in Tsez, Folia Linguistica 33, 109 130.
Rai, Novel K., Balthasar Biel, Martin Gaenszle, Elena Lieven, Netra Paudyal, Iha P. Rai,
Manoj Rai and Sabine Stoll (2005), Triplication and ideophones in Chintang, in Y. P.Yadava, ed.,
Current issues in Nepalese linguistics, Linguistic Society of Nepal, Kirtipur.
Rai, Novel Kishore (1984), A descriptive study of Bantawa. Ph.D. dissertation, Deccan College
Post-Graduate Resear Institute, Pue (India).
Rapaa, Syankarelu Lal, Bag-Ayagyami Yalunga and Amar Tumyahang (2008), Indo-Nepal
Kiranti Bhasaharu, Kirantivigyan Adhyayan Samsthan, Kathmandu.
Russell, Andrew (1997), Identity Management and Cultural Change: Religion and Politics amongst
the Yakkha, in D. N.Gellner, J.Pfa-Czarnea and J.Whelpton, eds, Nationalism and Ethnicity in a Hindu Kingdom: the Politics of Culture in Contemporary Nepal, Harwood Academic
Publishers.
Russell, Andrew (2004), Traditions in Transition: Sanskritization and Yakkhacation in East
Nepal., History and Anthropology 15(3), 25161.
Russell, Andrew (2007), Writing Traveling Cultures: Travel and Ethnography amongst the Yakkha
of East Nepal, Ethnos 72(3), 361 382.
Saow, Diana (2008), Clause linkage in Puma (Kiranti). MA. thesis, Department of Linguistics,
U. Leipzig.
Sharma (Gautam), Narayan P., Balthasar Biel, Martin Gaenszle, Arjun Rai and Vishnu S.
Rai (2005), Personal and possessive pronouns in Puma (Southern Kiranti), in Y. P.Yadava,
G.Bhaarai, R. R.Lohani, B.Prasain and K.Parajuli, eds, Contemporary issues in Nepalese linguistics, Linguistic Society of Nepal, Kathmandu, pp. 225 233.
Silverstein, Miael (1976), Hierary of features and ergativity, in R.Dixon, ed., Grammatical
Categories in Australian Languages, Humanities Press, New Jersey, pp. 112 171.
Subba, T B (1999), Politics of Culture, Orient Longman, Chennai.
Toba, Sueyoshi, Ingrid Toba and Novel Kishore Rai (2005), Diversity and Endangerment of Languages in Nepal, UNESCO Kathmandu Series of Monographs and Working Papers: No 7, Kathmandu: United Nations Educational, Scientic and Cultural Organization, Kathmandu Oce.
van Driem, George (1987), A grammar of Limbu, Mouton de Gruyter, Berlin.
van Driem, George (1990), e fall and rise of the phoneme /r/ in Eastern Kiranti: sound ange
in Tibeto-Burman, Bulletin of the Sool of Oriental and African Studies 53, 83 86.
van Driem, George (1993), A grammar of Dumi, Mouton de Gruyter, Berlin.
95
17 Bibliography
van Driem, George (1994), e Yakkha verb: interpretation and analysis of the Omruwa material
(a Kiranti language of Eastern Nepal), Bulletin of the Sool of Oriental and African Studies
57, 347 355.
van Driem, George (2001), Languages of the Himalayas: an ethnolinguistic handbook of the
Greater Himalayan Region, containing an introduction to the Symbiotic eory of Language,
Brill, Leiden.
Whelpton, John (2005), A History of Nepal, Cambridge University Press, Cambridge, UK.
Winter, Werner, Gerd Hansson, Alfons Weidert and Biram Ingwaba Subba (1996), A Synoptic
Glossary of Athpare, Belhare and Yakkha, Lincom Europa, Untersleissheim, Mnen.
96

Sketch of Yakkha Grammar

Uploaded by

Copyright:

Available Formats

Sketch of Yakkha Grammar

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sketch of Yakkha Grammar

Uploaded by

Copyright:

Available Formats

Aspects of Yakkha Grammar

Adverbs and adjectives

Reduplication, triplication and ideophonic paerns in adverbs and adjectives . .

e structure of simple clauses

10 Relative clauses and nominalisation

11 Adverbial clauses and converbs

12 Conditional and counterfactual clauses

14 More on nite clause linkage constructions

16 Particles used for information structure

A note to the reader

person (1>3: rst acting on third person, etc.)

1.2 e language name

1.3 Earlier work on Yakkha language and culture

1.4 Genetic Aliation

1.4 Genetic Aliation

Northwestern: Upper Dhko:

2.2 Phoneme inventory and allophonic rules

Figure 2.2: Minimal pairs for vowels

Figure 2.3: Yakkha consonant phonemes

2.2 Phoneme inventory and allophonic rules

Figure 2.4: Minimal pairs for consonants

ey did not give it to us. (underlying: /pi/)

I do not eat it. (underlying: /ca/)

your house (underlying: /pa/)

eko ceya lut-u-bi-ci

She told them a story. (underlying /pi/)

Did you put on music? (underlying: /ent/)

He did not go.

Do you have ildren?

2.2 Phoneme inventory and allophonic rules

e big ien pe the small ien.

See also Chapter 5.1 on stem formation

with my auntie and her people

sister Lila and her folks

We went. (adressee excluded)

We went. (adressee included)

pronoun plus prex

4.2.2 Possessive pronouns

male.in-law where arrive-NPST-2=NMLZ.s

Where will (your) husband arrive?

4.2.3 Demonstrative pronouns

toba imin et-u-ga=na?

How do you like this beer?

How do you like these?

ese beers are tasty./e beers are tasty.

his (this ones) house (Nep.: yasko ghar)

his (that ones) house(Nep.: usko ghar)

their (these) house (Nep.: yiniharuko ghar)

their (those) house (Nep.: uniharuko ghar)

the cows up there

DIST-NMLZ.s, 1sPOSS-GEN beer

Your beer is that one, my beer is this one.

4.2.4 Interrogative pronouns and other interrogatives

Figure 4.2: e interrogative pronouns and other interrogatives

4.3 Case system

What did you come for?

where.from come-PST-2=NMLZ.s girl

Where do you come from, daughter?

When did you arrive?