A Grammar of Karbi
A Grammar of Karbi
A Grammar of Karbi
by
A DISSERTATION
March 2014
DISSERTATION APPROVAL PAGE
This dissertation has been accepted and approved in partial fulfillment of the
requirements for the Doctor of Philosophy degree in the Department of Linguistics by:
and
Original approval signatures are on file with the University of Oregon Graduate School.
ii
1. © 2014 Linda Anna Konnerth
2.
iii
DISSERTATION ABSTRACT
Doctor of Philosophy
Department of Linguistics
March 2014
iv
Noun phrases may be marked for their clausal role via - ‘non-subject’
or - ‘locative’ but frequently remain unmarked for role. Their pragmatic status can be
indicated with information structure markers for topic, focus, and additivity.
Commonly used discourse constructions include elaborate expressions and
parallelism more generally, general extenders, copy verb constructions, as well as a
number of final particles.
Audio files are available of the texts given in the appendices, particular examples
illustrating phonological issues, and phonetic recordings of tone minimal sets.
v
CURRICULUM VITAE
DEGREES AWARDED:
PROFESSIONAL EXPERIENCE:
Small Professional Grant, Center for Asian and Pacific Studies, University of
Oregon, 2012, 2010, 2008
vi
Gary E. Smith Summer Professional Development Award, funding attendance of
the Summer Linguistic Institute ‘Language in the World’ of the Linguistic
Society of America at the University of Colorado at Boulder, 2011
Graduate Student Research Grant for the project ‘The Status of Women in Karbi
Society,’ Center for the Study of Women in Society, University of Oregon,
2011
PUBLICATIONS:
vii
Konnerth, Linda. 2011. “Functions of nominalization in Karbi.” In North East
Indian Linguistics Volume 3, edited by Gwendolyn Hyslop, Stephen Morey,
and Mark W. Post, 120-34. New Delhi: Foundation/Cambridge University
Press India.
Konnerth, Linda, and Dan Wood. 2008. “Report on NEILS 3.” Linguistics of the
Tibeto-Burman Area 30.2, 173-6.
viii
ACKNOWLEDGMENTS
My deepest gratitude goes to my advisor Scott DeLancey. Scott has been my true
mentor ever since I came to Oregon in 2006. In addition to sharing his knowledge, Scott
has always supported me, inspired me, and led me back to see the big picture when I was
lost in details. The Karbi descriptive grammar project was born in 2007 when members
of the Karbi Lammet Amei (Karbi Literary Association) contacted Scott. This was
serendipitous, as Scott had just weeks earlier given me Karl-Heinz Grüßner’s 1978
grammar of Karbi phonology and morphology to look at; I happily took on this exciting
project. Before my first trip to India, I was able to get in touch with Karl-Heinz and
visited him in Tübingen, Germany, where he let me into the magical realm of his attic. I
spent several hours perusing his notes, transcriptions, and other Karbi materials he had
collected. I cannot possibly thank Karl-Heinz enough for his generosity, encouragement,
and the friendly welcome I received from him and his family, both in Tübingen and in
Shillong. During my first trip to India I met Sikari Tisso, who was to become my main
language consultant and collaborator, and who made everything possible. His dedication
and tireless efforts make him a true hero for all who value the wonderful language and
culture of the Karbi people. While this project was ongoing, his son passed away too
young; this dissertation is also dedicated to the memory of Sarpo Rongkhelan Tisso.
I am deeply grateful to UV Jose. His enormous knowledge of the languages in the
area and of the people speaking them as well as his generosity and advice and logistical
support inspired and helped me during all this time. Many thanks also to Brother
Benjamin Ingti Kathar, who shared his language and warm friendship with me.
This project would not have existed without the effort of the Karbi Lammet Amei
to preserve their beautiful and rich language and inherent culture and knowledge for
future generations. Khorsing Teron as well as Dharamsing Teron (not actually a KLA
member) volunteered much time and effort to the project. Thanks also to Hokursing
Rongpi, and to Budheswar Timung (Nowgong KLA). Due to space limitations, other
Karbi speakers that have made this project possible are acknowledged in §2.3.1.
I would like to thank Amphu Rongpipi, Klirdap Teronpi, Serdihun Beypi,
Larshika Tissopi, and Sarlomet Tisso, who all greatly contributed to the project by
preparing transcriptions, struggling through translations, and helping with the analysis.
ix
None of my research would have been possible without a place for me to stay. In
Diphu, Kro hem always had their doors open for me: Sashikola Hansepi, Member Kro,
Ruplin, Sintu, Rasinza, and Sampri. Jirsong Asong in Diphu also offered me a warm and
friendly place to stay - many thanks to Fr. Vinod! In Umswai, I stayed with the wonderful
Hanse family: Albina Teronpi, Holiwel Hanse, Hunmily, Basapi, John, and Platinum.
Closer to home, my dissertation committee has offered great feedback that has
given me plenty of food for thought: thank you to Spike Gildea, Doris Payne, and Zhuo
Jing-Schmidt! I would also like to thank everybody else in the wonderful Linguistics
Department at the University of Oregon, where I always received support, friendship, and
encouragement from all sides. I feel more fortunate than I can put in words to have been
able to learn from the great faculty at UO along with my fellow grad students, as well as
from these same grad students, in particular Shahar Shirtz, Amos Teo, Danielle Barth,
Anna Pucilowski, and Dan Wood. I am also particularly thankful to Gwen Hyslop, who
was always willing to help, give advice, and, of course, talk Tibeto-Burman. In addition, I
wish to express my gratitude to Wolfgang Schulze at the Ludwig-Maximilians-
Universität München, who made it possible for me to come to the UO in the first place.
I am grateful to a number of other linguists who have given me feedback on my
work and have inspired me, in particular, Northeast Indianists Mark Post, Stephen Morey,
and Jyotiprakash Tamuli; Tibeto-Burmanists David Bradley, Christina Willis, Dave
Peterson, Stephen Watters, and Robert Schikowski; as well as Antoine Guillaume.
This research was funded in part by National Science Foundation Grant # BCS-
0951749. At the UO, further support was provided by the Center for the Study of Women
in Society; the Center for Asian and Pacific Studies; and the Oregon Humanities Center.
Finally, I have always had incredible support from home in Germany. If I am any
good at explaining linguistic issues to non-linguists, it is because I got to practice many
many many times with my mom, who miraculously always wanted to know more. I have
received love, encouragement, and advice from Roswitha, Arthur, Arne, Sascha, Irmi,
Juli, Franzi, and everybody else in my extended family and of my wonderful friends; they
all have sustained me throughout all these years and especially in the last months of
writing this dissertation. Lastly, thank you, Eric, for everything - your help, your humor,
your understanding, your support, your encouragement, your love, every step of the way.
x
To the Karbi people
xi
TABLE OF CONTENTS
Chapter Page
I. INTRODUCTION .................................................................................................... 1
1.1. Karbi People, Language, and Culture ............................................................. 2
1.1.1. Names and ISO Codes for the Karbi Language ................................... 2
1.1.2. Number of Speakers and Geographical Spread of Karbi ..................... 2
1.1.3. Aspects of Karbi Traditional Culture and Social Organization ........... 4
1.1.4. The Karbi Lammet Amei (KLA) .......................................................... 6
1.1.5. The Karbi Anglong District ................................................................. 6
1.2. Karbi in Tibeto-Burman.................................................................................. 7
1.2.1. Tibeto-Burman Languages of Northeast India .................................... 7
1.2.2. Karbi in Tibeto-Burman Classification Proposals ............................... 8
1.3. The Role of Contact in the Development of Karbi ......................................... 10
1.4. Varieties of the Karbi Language ..................................................................... 12
1.4.1. Plains Karbi (“Amri Karbi”) and Hills Karbi ...................................... 13
1.4.2. Relationships between the Hills Karbi ‘Dialects’ ................................ 15
1.4.3. Hills Karbi: Differences between Rongkhang (Hills Karbi; Diphu)
and Amri (Hills Karbi; Western Karbi Anglong) Dialects................... 16
1.4.4. Variation in Lexemes ........................................................................... 17
1.5. Sociolinguistic Profile of Language Endangerment ....................................... 17
1.5.1. Setting Factors ..................................................................................... 18
1.5.1.1. Education ................................................................................. 18
1.5.1.2. Virtual and Physical Infrastructure .......................................... 18
1.5.1.3. Government.............................................................................. 19
1.5.1.4. Media ....................................................................................... 19
1.5.2. Contact and Change ............................................................................. 19
1.5.3. Domains/Vitality/Attitudes .................................................................. 20
1.5.3.1. Domains ................................................................................... 20
1.5.3.2. Vitality ..................................................................................... 21
1.5.3.3. Attitudes ................................................................................... 21
xii
Chapter Page
xiii
Chapter Page
xiv
Chapter Page
xv
Chapter Page
xvi
Chapter Page
xvii
Chapter Page
xviii
Chapter Page
xix
Chapter Page
xx
Chapter Page
xxi
Chapter Page
xxii
Chapter Page
xxiii
Chapter Page
xxiv
Chapter Page
xxv
Chapter Page
xxvi
Chapter Page
xxvii
Chapter Page
xxviii
Chapter Page
xxix
Chapter Page
xxx
Chapter Page
xxxi
Chapter Page
xxxii
Chapter Page
xxxiii
Chapter Page
xxxiv
Chapter Page
xxxv
Chapter Page
xxxvi
LIST OF FIGURES
Figure Page
11. Female speaker vowel formant values for ke- and ne- .......................................... 109
12. Male speaker vowel formant values for ke- and ne- .............................................. 109
13. Semantic map of basic clausal functions of Karbi parts of speech ........................ 133
19. Grammaticalization scenario for =si ‘focus’ (starting as copula) ......................... 423
20. Possible grammaticalization pathway for the imperfective construction .............. 427
xxxvii
Figure Page
23. Waveform and spectrogram of enutvetlo ‘he was alone’ (HK, TR 005-6) ............ 525
25. Prosodic emphasis in full predicate reduplication of baithekthe (KK, CC 012) ... 594
xxxviii
LIST OF TABLES
Table Page
5. Working and presentation software used for different data types ......................... 40
xxxix
Table Page
29. Stopped syllable tone minimal pairs for M-H, L-H, and L-M ............................... 74
33. Tone sandhi effects on low tone suffix after mid tone stem .................................. 84
xl
Table Page
44. Monosyllabic stem tone changes after che- ‘RR’ and cho- ‘AUTO.BEN/MAL’ ........ 100
45. Disyllabic stem tone changes following possessive prefixation ............................ 101
47. List of derivational suffixes that participate in tone change .................................. 102
52. Initial vowel deletion in ing- stems after prefixes ke-, pV-, cho- ........................... 107
53. Vowel deletion between ar- stems and prefixes ke-, pV-, che-, a- ........................ 107
54. Vowel formant values of ke- ‘NMLZ’ and ne- ‘1EXCL.POSS’ .................................. 108
55. Morphosyntactic criteria for the attempt to identify a class of ‘adjectives’ .......... 120
57. Sample complex stems expressing concepts of HUMAN PROPENSITY .................... 122
61. List of sortal classifiers that occur in typical classifier construction ..................... 137
65. Nouns counted with numeral only (without classifier) .......................................... 147
xli
Table Page
71. Morphological structure of independent pronouns and possessive prefixes ......... 168
xlii
Table Page
90. Sample PDs that mean ‘quite’ (productive and non-productive PDs) ................... 249
92. Some predicate derivations and related lexical items ............................................ 250
95. Sample PDs that indicate a considerable degree (‘quite’) ..................................... 252
97. PDs indicating argument quantification and degree or extent ............................... 253
101. Sample predicate derivations that function as argument classifiers .................... 262
xliii
Table Page
117. Irrealis contexts for -ra ‘NF:IRR’ and =le ‘FOC:IRR’ ............................................. 553
121. Metadata of texts that have formed the main corpus for this dissertation ........... 729
xliv
CHAPTER I
1. INTRODUCTION
This is a grammar of Karbi as spoken in the hills of the Karbi Anglong district in
Assam, Northeast India. It expands on research findings reported in Grüßner's (1978)
grammar of the phonology and morphology of the language, but also offers a more
comprehensive treatment of issues in Karbi syntax, semantics, and pragmatics.
The remainder of this chapter is organized as follows. In §1.1, the Karbi people
and their language and culture are introduced. Next, an overview of the linguistic context
of Karbi is offered, i.e., the relationships between Karbi and surrounding languages, as
that context helps understand why modern Karbi grammar is the way it is. On the one
hand, that involves the relationship to other Tibeto-Burman languages, as discussed in
§1.2. On the other hand, and very importantly, it also involves the contact relationship
specifically with the Austroasiatic Khasi languages, and possibly other Tibeto-Burman
languages, such as Meithei, as discussed in §1.3.
In §1.4, an overview of what little is known on different varieties of Karbi is
provided, including the major dialectal divide between Hills and Plains (or, Amri) Karbi,
and some notes on variation within Hills Karbi, the major dialect group that this grammar
is based on.
A sociolinguistic profile of language endangerment (mostly based on Hills Karbi)
is included in §1.5.
Elements of Karbi oral literature are discussed in §1.6, including notes on specific
genres of oral literature in the context of Northeast India, as well as including remarks on
the Karbi song language.
In §1.7, an overview of the linguistic literature and linguistic resources on Karbi
is provided.
Finally, §1.8 discusses Karbi orthography and ongoing issues in standardization
efforts as well as an outline of conventions followed in this dissertation. The organization
of this grammar is outlined in §1.9.
1
1.1. Karbi People, Language, and Culture
In the last few decades, there has been a movement among the Karbis to push for
the autonym Karbi or the elaborate form Karbi Karbak (see §12.2.2 on elaborate
expressions). While this name has long been in use, it is a recent development that Karbi
is favored over the logonym Arleng (i.e., ‘man, person’).1 This might be due to the
existence of as a simple noun root for ‘man, person’.2
Mikir is a formerly commonly used exonym, which has become pejorative within
the last few decades in particular. Now most Karbis have strong objections against it,
which has to do with a number of offensive hypotheses for the etymology of this name.
Bhoi Mynri is mentioned by Grüßner (1978:6) as an exonym used by the neighboring
Khasis of Meghalaya, to the immediate west of Karbi Anglong (see §1.3). According to
my language consultants, this term Bhoi Mynri may also specifically refer to the variety
of Plains Karbi spoken across the western border of Assam in Meghalaya (§1.4.1).
The Hills Karbi variety has the ISO 639-3 code 'mjw', whereas the Plains Karbi
variety has the ISO code 'ajz' (for dialect differences, see §1.4.1).
The Census of India from 2001 reports a total of 419,534 native speakers of Karbi,
which is also the figure cited in the Ethnologue (Lewis, Simons, and Fennig 2013). The
Karbi Lammet Amei (§1.1.4) estimates a higher number of speakers, at over half a million.
Karbi is spoken in Assam and adjacent areas in neighboring states in Northeast
India. There are also reported to be Karbi villages in Bangladesh and Burma. For a map
of Northeast India, that shows the borders with Bangladesh, Bhutan, China (to the
northeast of Northeast India), and Burma, see Figure 1.
1
In one of the recorded texts collected for the corpus of this dissertation, the storyteller finds himself
saying Arlengpi for ‘Karbi woman’ (using the female - suffix) and corrects himself and says ‘Karbipi’.
2
Note, however, that there also is another general noun t 'person, man’, which is a borrowing from
Asamese.
2
Figure 1. Northeast India (taken from maps.google.com)
3
river valley area and the hill range that extends to the south and the southeast as hinted at
in Figure 1.3
3
This is part of the hill range that extends all the way into Southeast Asia and represents something of a
cultural area, with similar histories of the people inhabiting them, see Scott (2009).
4
crucial times such as sickness or before going on a long trip, priests perform rituals that
typically involve sacrifice of animals (such as chickens or ducks) in conjunction with
chants that are orally transmitted from generation to generation, typically using the Karbi
song language (§1.6.2), see Figure 3.
An important cultural symbol is the Jambili athon (Figure 4). The bird on top
represents values such as wisdom, intellectuality, and leadership. The lower four birds in
the four directions are the followers.
There are five major clans in Karbi society: Terang, Teron, Inghi (also spelled
Enghi or Enghee), Ingti (also spelled Engti), and Timung.4 These five major clans are
4
These are the clan names in the Hills Karbi variety; in Plains Karbi, some names are slightly different, e.g.,
Timung is Tumung (see §1.4.1 on dialect differences between what I refer to as Plains and Hills Karbi).
5
further divided into subclans. This division into clans and subclans has important societal
consequences such as marriage restrictions.
An excellent resource on Karbi cultural studies are the two volumes ‘Karbi
Studies’. The first volume is edited by Dharamsing Teron, with contributions both from
Karbi and international scholars (D. Teron 2012), while the second volume is entirely
authored by Teron (D. Teron 2011). Further information can also be obtained from the
official website of the Karbi Anglong district (http://www.karbianglong.nic.in/) and a
blog maintained by Morningkeey Phangcho (http://karbi.wordpress.com/).
This dissertation is the result of close collaboration with members of the Karbi
Lammet Amei, who in fact initiated the project in 2007: most notably Mr. Sikari Tisso, as
well as Mr. Khor Sing Teron. The Karbi Lammet Amei (KLA; from - t -
‘Karbi word-artful POSS-assembly’) is a language and literature organization based in the
district capital Diphu, but with branches in larger villages and towns. The KLA was
founded on March 27, 1966, with the goal of preserving and promoting the Karbi
language so it could be taught in schools and other institutions of higher education, while
also engaging in the promotion of Karbi literature (Khor Sing Teron, p.c.).
The Karbis have had their own autonomous Karbi Anglong district (lit., -
‘Karbi POSS-hill’ > ‘Karbi hills’) for a number of decades. The district was first
formed in 1951, although at that time, the North Cachar Hills to the south of the present-
day western and eastern part still belonged to the district (then called ‘United Mikir and
North Cachar Hills District’). In 1970, the two parts were separated, and the ‘Mikir Hills’
district was renamed as Karbi Anglong in 1976, with Diphu in the eastern part as the
capital. Karbi Anglong is the largest district of Assam.5
5
This information comes from http://www.karbianglong.nic.in/, which is the official website of the Karbi
Anglong District Administration, accessed on February 3, 2014.
6
1.2. Karbi in Tibeto-Burman
While there has never been any doubt that Karbi is a Tibeto-Burman language, the
exact phylogenetic status inside Tibeto-Burman has not been possible to determine. This
is despite the fact that information on Karbi has been available early on (§1.7.1), and that
it has been considered in the early large-scale Tibeto-Burman classification proposals as
well as in the modern, detailed comparative work. The difficulties of working out the
exact phylogenetic status of Karbi are likely in large part due to a history of language
contact and grammatical reorganization as a result of it (§1.3).
Bradley’s Sal branch is a more inclusive version of this branch whose name was
coined by Burling (1983), but renamed later more transparently as the ‘Bodo-Konyak-
Jinghpaw’ branch (Burling 2003). Besides this Bodo-Konyak-Jinghpaw subbranch,
7
Burling’s (2003) attempt at classifying the Tibeto-Burman languages of Northeast India
includes a substantial number of other low-level branches, whose higher-level groupings
remain far from clear. As seen in the classification of the languages of the ‘Eastern
Border’ in Figure 6, Karbi has in this context always been one of two languages (the
other being Meithei, the state language of Manipur) that have been particularly difficult
to associate with one of the other low-level branches.
Karbi (then referred to as ‘Mikir’) was included in the Linguistic Survey of India
(LSI) by Grierson and Konow in the early 20th century (Grierson 1903), which represents
8
the first attempt at classifying Tibeto-Burman languages. Already at that time, there was
a fair amount of information available on the language. In the LSI, it is noted that Karbi
“has received some attention from the missionaries who work among them”, and “we
have a vocabulary and some short pamphlets written in it and an admirable grammar with
selected texts from the pen of the late Sir Charles Lyall” (Grierson 1903:69). A
classificatory problem Grierson and Konow encountered is noted in the next two
sentences:
In Volume III, Part ii of the Survey I have classed Mikir as falling within the
N g -Bodo Sub-Group. The language has affinities with Bodo, but subsequent
investigation has shown that it is much more closely connected with Kuki, and
that it should be classed […] as belonging to the N g -Kuki Sub-Group, in which
it occupies a somewhat independent position.
While the absence of a closer link between Karbi and Bodo-Garo6 has not been
controversial since, there are three other groups in particular that have been linked to
Karbi: Meithei, Naga, and Kuki-Chin (the latter two of which were put into one group,
going back to the LSI, see above).7
However, the evidence that underlies the grouping of Karbi with Meithei might
better be analyzed as borrowings (§1.3). The putative grouping with Naga is complicated
due to the fact that it is not currently clear at all what ‘Naga’ actually is, as there is a
long-standing confusion of ethnic and linguistic labels surrounding the term ‘Naga’
(Burling 2003) (i.e., using ethnic labels as linguistic labels, see also §1.4.2 for a similar
problem within Karbi ‘dialects’). A possible link to Kuki-Chin currently appears
promising. As pointed out in various places throughout this dissertation, particular links
to Kuki-Chin exist, for example with respect to: the negative equational copula (§4.6.2.2);
the cislocative as well as speech act participant non-subject marking (§6.3.1.4); the
reflexive/reciprocal prefix (§6.4.3); and the focus marker =si (§9.7.3.1.5), among other
constructions. However, working out the exact details as well as implications of these
6
Bodo-Garo languages form a “compact, low-level branch of Tibeto-Burman” (DeLancey 2012). The few
similarities that exist between Karbi and Bodo-Garo, such as the Karbi ke- nominalizer that is a cognate of
a Bodo-Garo adjectival prefix (Konnerth 2009, 2012), stem from a very high node, possibly Proto-Tibeto-
Burman.
7
There also was a proposal by Bauman (1976) to consider Karbi the missing link in a connection between
Kuki-Chin and Lepcha.
9
similarities and apparent cognates is a matter of future research. This will likely include
the difficult work of carefully disentangling areally from genetically shared features.
In sum, it has remained difficult to come up with a classification proposal that
places Karbi in a closer relationship with one of the (geographically) neighboring
branches. This is despite early availability of information on Karbi grammar and lexicon
(§1.7.1), but certainly has to be seen in the context of the remaining lack of information
on some of the Kuki-Chin and so-called ‘Naga’ languages. What appears quite obvious,
however, is that a major factor in obscuring the relationships between Karbi and other
Tibeto-Burman languages has been language contact and contact-induced changes in
Karbi grammar and lexicon. In particular, it has been known since the Linguistic Survey
of India that Karbi has been in close contact (and, in fact, the closest contact of all TB
languages) with the Austroasiatic Khasi languages to the west in Meghalaya.
DeLancey (2012) offers a creolization account for the origins of the Bodo-Garo
branch of Tibeto-Burman, which likely applies in the historical development of Karbi as
well, as in fact also suggested by DeLancey. At a smaller scale than Bodo-Garo, Karbi
shares the ethnolinguistic profile of a language a) spoken by a relatively large number of
people (in the local linguistic context), b) spread across a relatively large area, and c)
extending into the plains (as opposed to being only spoken in the hills) (see §1.1.2).8
While Bodo-Garo is argued to have an origin in a lingua franca in DeLancey’s
proposal, such an extreme case of contact influence does not need to be assumed in the
case of Karbi. However, a considerable impact from contact has to be part of the history
of Karbi. Specifically, there is robust evidence of changes in Karbi grammar and lexicon
8
This was also noted as a striking difference between the Karbis and other ethnic groups by Walker (1925)
in the preface to his dictionary. He says, “[…] the Mikirs are among the more numerous of the Assam
frontier races, and […] they are scattered over a wide area, from Golaghat to Kamrup and the Khasi Hills
beyond Gauhati, and from the Cachar plains near Silchar to the forests north of Bishnath in Darrang […].”
A similar remark about the relatively wide geographic spread of the Karbis in the local context stems from
the Linguistic Survey of India, where it is noted that “it cannot be doubted that in former times the Mikirs
occupied a comparatively large tract of country in the lower hills and adjoining lowlands of the central
portion of the range stretching from the Garo Hills to the Patkoi” (Grierson 1903:69).
10
(as well as, in fact, culture and social organization9) due to contact with the Khasi
languages (Grierson 1903; Grüßner 1978; Joseph 2009). This branch of Austroasiatic is
centered in a region to the west of the Karbi-speaking area, across the border in the state
of Meghalaya. More research on the Khasi branch is needed in order to be able to connect
a particular language more closely to extensive contact with Karbi. At this point,
specifically the language (and ethnic group) referred to as Jaintia or Pnar appears to be
the one closest linked to contact with Karbi.
In addition to contact with the Khasi languages, there is some evidence hinting at
a possible contact scenario with Meithei. The one strong piece of evidence is the
peculiarity of numerals ‘eight’ and ‘nine’ being morphologically complex forms that
translate as ‘ten minus two’ and ‘ten minus one’ (§4.8.2). This subtractive construction
for ‘eight’ and ‘nine’ is not attested so far in any other language in the region. The
corresponding constructions in Karbi and Meithei look calqued, since the individual
morphemes do not correspond.
Another strong piece of evidence is that Meithei has a female -pi suffix
corresponding to Karbi - ‘female’ (§5.4.1.1). This is noteworthy because it is not a
typical Tibeto-Burman form and to my knowledge not attested anywhere else in the
family.10
In addition to the evidence from numerals and the female suffix, there are several
other correspondences that could potentially contribute to a model of Karbi-Meithei
contact (without being strong evidence). For example, the Karbi word ‘meat’ is
peculiar because the more common Tibeto-Burman root for a word ‘flesh’ is something
like Matisoff’s (2003) reconstructed *sya. Now the Meithei word for ‘pig’ is (Chelliah
1997), and considering that pork is the major and favorite type of meat eaten by the
Karbis, might be a borrowing from this Meithei word for ‘pig’. At the same time, the
Meithei is similar to roots for the word ‘pig’ in other Tibeto-Burman languages.
9
As pointed out by Grüßner (1978), a number of lexical items referring to social organization as well as the
kingdom system are borrowings from Khasi, obviously suggesting that the concept was borrowed along
with the word. Examples include - ‘king’ (with the male suffix - ) from Khasi lyngdoh ‘priest’,
and the word k for ‘clan’.
10
The typical Tibeto-Burman female suffix has a bilabial nasal. For example, in Tibetan, the male and
female suffixes are -pa and -ma, respectively.
11
Matisoff (2003) here reconstructs *pwak (the Karbi word is . Therefore, this does not
represent strong evidence for a borrowing from Meithei into Karbi.11
While the details of the Karbi dialect situation are outside the scope of this
dissertation, it appears that there is a high degree of homogeneity - perhaps surprisingly
so, given the large geographic spread of the language. This was also noted by Walker
(1925) as he writes in the preface to his dictionary that “in spite of the fact […] that [the
Karbis] are scattered over a wide area, […], the language is practically one and the same
throughout.”
The highly simplified ‘big picture’ of the dialect situation, is that there is a major
dividing line (political as much as, or even more than, linguistic in nature) between the
Hills Karbis (Karbis from Karbi Anglong (KA)) and the Plains Karbis (Karbis mostly
living in the plains of Assam largely north of KA), as discussed in §1.4.1.
Within each of these major two varieties, there is relatively little dialectal
variation. However, investigating the nature of dialectal variation is complicated due to
the application of dialect labels by Karbi native speakers, which are grounded in
historical ethnic/familial and/or geographical affiliation, as outlined in §1.4.2.
Following this discussion, §1.4.3 further discusses two of these dialect labels from
the Hills Karbi variety: the Rongkhang or Ronghang dialect, which (with apparently
wide-spread acceptance) is being used as the basis for standardization; and the Hills (not
Plains) Amri Karbi dialect, which is spoken in the western part of Karbi Anglong, where
the traditional-cultural center of the Karbis lies.
Finally, §1.4.4 offers a list of some lexemes that have been found to exhibit
(mostly, vowel) alternations in the speech of different native speakers, without, however,
actually appearing to represent dialect isoglosses.
Note that besides these geographical and historical/ethnic dialect groups, there
appears to be some evidence for a Christian sociolect (possibly specifically in the Tika
region), with some slight differences in lexicon and grammar from the non-Christian
11
In addition, there is also a demonstrative si in Meithei (Chelliah 1997) that represents an alternative (or
possibly ultimately the same) cognate for the focus marker =si in Karbi (which is suggested to be
connected to an equational copula si(i) in Central Kuki-Chin in §9.7.3.1.5).
12
sociolect (see §1.7.2 on Grüßner’s work, which was based on a variety with some such
features). All of these issues pertaining to linguistic varieties of Karbi require further
research.
In the discussion of varieties of the Karbi language, in the following subsections
as well as in the entire dissertation, I want to emphasize that not a single word is written
with a political motive behind it. My goal has always been to describe the linguistic
landscape in a scientific way and to be as neutral as possible when it comes to the politics
that are, of course, tied to it in real life. I truly hope that no part of the discussion of the
different varieties of the Karbi language is offensive to anybody.
The Plains Karbi variety spoken in the Kamrup and Marigaon districts of Assam
as well as partly in the Ri-Bhoi district of Meghalaya is commonly referred to as ‘Amri
Karbi’ in the linguistic literature and in the Ethnologue (Lewis, Simons, and Fennig
2013). However, it should be noted right away that Karbis identifying with the western
subvariety of Hills Karbi use the same name for themselves, possibly because of an
ultimately shared geographical origin and/or common ancestors (§1.4.3).
Members of the Karbi Lammet Amei (§1.1.4) have expressed their concern to me
over the use of the name Amri for the Plains Karbi variety, while this name is embraced
by many speakers of this variety. The term ‘Dumra’ or ‘Dumrali’ is also used to refer to
this variety of the Karbi language and the people that speak it.
In the following discussion, I will use the geographically based terms ‘Plains
Karbi’ and ‘Hills Karbi’ to refer to the two major linguistic varieties of the Karbi
language, which, again, include further ‘sub’-varieties based on linguistic features, which
are, however, not as different from one another as are the two major varieties. Although
these geographic terms are not ideal either since there are Karbis living in the plains who
do not speak the ‘Plains Karbi’ variety, I follow M. Teron and Tumung (2007) in using
‘Plains Karbi’ and ‘Hills Karbi’ in these ways, as the terminological debate is currently
still ongoing and there simply is no ideal set of terms to use at this point. Note that the
name Amri refers to a historical administrative unit in the Karbi kingdom, and, as
mentioned above, in addition to Plains Karbis, the group of Hills Karbis living in the
13
western part of the Karbi Anglong district also identify with this name, see §1.4.2 and
§1.4.3 below.
There is a strong political movement on part of the Plains Karbis to consider their
variety of Karbi a different language rather than just a different dialect from the variety of
Karbi that is spoken in the Hills. This likely has to do with the unequal power relations
between the two groups. While there are close to half a million native speakers of Hills
Karbi that have autonomy in the Karbi Anglong district, the Plains Karbi speaking
population is scattered across a number of districts, and a 2003 figure reported by the
Ethnologue estimates the number of speakers at a total of 125,000 (Lewis, Simons, and
Fennig 2013).12
According to my Hills Karbi language consultants, there is a high degree of
mutual intelligibility between the Hills and Plains Karbi. This is especially true for Hills
Karbi speakers that are fluent in Assamese (which most people living in the urban areas
are), as the Plains Karbi variety has a large number of Assamese loans due to closer
contact with Assamese in the plains. It is also noted in the Ethnologue that “some Amri
Karbi villages shifted completely to Assamese due to intermarriage and the perception
that Assamese is preferred for children to do well in school.”
In addition to the larger number of Assamese loans, some of the more noticeable
ways in which Plains Karbi is different from Hills Karbi are the following. First, there are
differences between the song language (§1.6.2) and the ordinary language. Specifically,
Plains Karbi uses lexemes in ordinary, colloquial speech, which are only used in the song
language of Hills Karbi, and vice versa.
A phonological difference is that Plains Karbi has preserved coda /l/, which in
Hills Karbi has changed into diphthongs or glide codas, depending on the analysis (see
§3.1.2 and §3.4).
In the domain of morphology, there are two salient differences. First, the unusual
onset-reduplicative negative suffix - in Hills Karbi (§3.8.6.3) corresponds to just -e
without the onset reduplication in Plains Karbi. Second, the Hills Karbi negative
existential copula is , while Plains Karbi instead uses the form ingjong (§4.6.2.1.1).
12
It is not clear what the basis is for this figure provided by the Ethnologue. The Census data of close to a
half a million native speakers of Karbi do not specify whether a distinction was made between Hills and
Plains Karbi (and hence it likely was not).
14
With respect to syntax, there are two frequently occurring Plains Karbi
constructions that are not used in Hills Karbi. On the one hand, the positive existential
copula is often used following a bare stem. It is not clear to my Hills Karbi language
consultants what the function of this construction is, but it is very striking to them. On the
other hand, habitual aspect is marked by the suffix -man in Plains Karbi, where in Hills
Karbi simply the bare stem is used.
Within the Plains Karbi variety, there is also some amount of dialectal variation.
For example, in the Ri-Bhoi district of Meghalaya, the Bhoi Mynri variant is spoken.
A good resource on the Plains Karbi variety is the trilingual dictionary edited by
M. Teron and Tumung (2007), which includes both Plains and Hills Karbi forms of each
lexical item, as well as translations into Assamese and English.
There are four ‘dialect’ labels that are used by Hills Karbi native speakers to
identify their own and other people’s speech: Amri, Rongkhang (or Ronghang),
Chinthong, and Killing. Originally, however, these labels are connected to historical
administrative units of the Karbi kingdom and the people that lived in these
administrative units. Therefore, while there certainly is a historical connection between
an individual’s affiliation to one of these groups and the variety s/he speaks, this is not
always the case anymore in the year 2014.
Since this discussion of these different groups is only for the purpose of sketching
out the linguistic landscape, everything said here comes through the lense of linguistic
variation and is considered in its relevance to linguistic varieties. I would like to
acknowledge that it is shorthand to speak of ‘Amri Karbi’ or ‘Rongkhang Karbi’ and that
the more accurate way of referring to the people that identify with these names is to say
Amri aso ‘children / inhabitants of Amri’ and Rongkhang aso ‘children / inhabitants of
Rongkhang.’
According to Dharamsing Teron (p.c.), the following can be said about these
‘dialect’ groups:
The three major groups are Amri, Rongkhang, and Chinthong. The Killing group
appears to be a part of the Rongkhang group. The Amri group was the first to migrate into
15
the present-day Karbi Anglong area. They split up and some of them went to present-day
western Karbi Anglong (i.e., the group discussed in §1.4.3), while others moved into the
plains (i.e., the Plains Karbis, discussed above in §1.4.1).
The Rongkhang group is mostly located in the southern portion of the eastern part
of Karbi Anglong, where the district capital Diphu is located. This group became the
most dominant and influential group, which is why the Rongkhang ‘dialect’ is currently
considered the standard dialect. Note, however, that linguistically, it is not possible for
my language consultants to pinpoint defining differences between Rongkhang and the
‘dialects’ of eastern Karbi Anglong and the area to the south, i.e., Chinthong, and Killing.
(The Killing group is geographically centered in places such as Kheroni, Jyrikyndeng,
and further into the North Cachar Hills area.)
1.4.3. Hills Karbi: Differences between Rongkhang (Hills Karbi; Diphu) and Amri
(Hills Karbi; Western Karbi Anglong) Dialects
While any particular differences between the Rongkhang dialect and other dialects
to the north and the south are not easy to discern for my language consultants (although
among themselves, they may identify with different dialect groups, see §1.4.2 above),
there are a number of differences between, on the one hand, Rongkhang and the other
putative dialects, and, on the other hand, the Amri dialect in western Karbi Anglong.
These differences are not only lexical in nature, but also include two systematic
phonological differences. First, the Amri dialect has a sixth phonemic vowel, which is a
high to mid-high, front, centralized /ɪ/ (§3.2.1). Second, the Amri dialect has preserved
/ei/ and the /ai/ diphthongs (or vowels /e/ and /a/ with palatal glide coda, see §3.4), while
the Rongkhang dialect has merged them to /ai/ (§3.2).
In addition, there are number of differences in other grammatical domains as well.
For example, the pe-~pa- ‘causative’ prefix is seemingly only ever produced as pa- in the
Amri dialect, i.e. without any allomorphy (see §3.9.2.1). Another difference is that that
‘afterthought’ particle =he (§12.3.9) is more frequently used in Amri speech, and that the
particle is prominently used as a marker of narrative style specifically in this dialect
(§12.3.12).
16
1.4.4. Variation in Lexemes
Table 1 provides lists of sample lexemes, of which there are two (or more)
variants based on vowel alternations. No study has been conducted yet to see whether
these alternation patterns somehow align in the speech of individual native speakers.
While Table 1 shows that many different vowel alternation patterns exist, it
should be noted that in almost all cases, the vowel alternation occurs in the first syllable
of a disyllabic word. The only two exceptions are the two words listed last:
‘girl’ and ‘carry on back’.
In addition to the simple vowel alternations in lexemes listed in Table 1, there
also exists lexical variation in other words, such as mensopi or nemsopi for ‘papaya’.
The following setting factors that have an impact on the degree of language
endangerment are discussed here: number of speakers, which, however, is discussed
above in §1.1.2; languages represented in education (§1.5.1.1); virtual and physical
infrastructure in the community (§1.5.1.2); the relationship between government and
language (policy) (§1.5.1.3); and, finally, how Karbi is represented in mass media
(§1.5.1.4).
1.5.1.1. Education
Even within the Karbi Anglong District, most schools have either Assamese or
English as the medium of instruction, depending on whether they are government or
Christian schools. Within the last fifteen to twenty years or so, a few Christian primary
schools have started to teach in Karbi (and other local minority languages), spearheaded
by Frs. UV Joseph and Joseph Teron, and textbooks have been developed (along with Br.
Benjamin Kathar). There also are a few private schools that teach in Karbi. Moreover,
there has been an effort to translate existing textbooks from Assamese into Karbi. By and
large, however, children go to Assamese or English medium schools.
A very sudden increase in virtual infrastructure in recent years (which was very
noticeable even just between 2008-2012), i.e., availability of electricity, TVs, cell phones,
13
This profile was put together as part of a seminar on ‘The Sociolinguistics of Language Endangerment’
offered by David and Maya Bradley at the 2011 LSA Institute in Boulder, CO. I would like to thank David
and Maya for the feedback they provided me and for engaging me in thinking about these issues.
18
and the internet, will likely affect the community. Increase in physical infrastructure has
recently improved mobility, and will almost certainly affect the community as well. A lot
is currently changing, and the endangerment situation five years ago was likely
substantially different from what the situation will be like in five to ten years from now.
Language shift to Assamese is a lot more common among Amri Karbis, who live
in the plains, compared to Karbi Anglong Karbis, who live in the hills. This has likely
been the case historically (in the last several centuries or so) as well, since there are a lot
more Assamese loans in the Plains Karbi variety (§1.4.1).
1.5.1.3. Government
The Karbis live in their autonomous Karbi Anglong district with a Karbi Anglong
Autonomous Council (and that is also its official name, i.e. it is in English rather than
Karbi) (see §1.1.5). This political autonomy strengthens the status of the language.
1.5.1.4. Media
There are a few newspapers in Karbi. The Arleng Daily and the Thekar14 are
written in Roman script, while there are also newspapers that use the Assamese script.
The KLA publishes quite a lot of books in Karbi. There is a local film industry that
produces movies and comedy shows (and some documentaries) in Karbi. There also is a
Karbi language TV program.
There has been a recent increase in availability of Karbi media on the internet. In
particular, there are Karbi blogs and there are a number of songs sung in Karbi (but
following the ‘mainland’ Indian Bollywood style) on Youtube.
The major dominant language in the area that many Karbi speakers shift to is
Assamese. Shifting to Assamese occurs among all of Assam's minority languages, since
it is the main lingua franca for speakers of different minority languages. While English
14
The Thekar now also has an online edition: http://thekararnivang.com/.
19
functions as a lingua franca to some degree as well (especially in the Christianized areas),
Assamese is more widespread.
1.5.3. Domains/Vitality/Attitudes
1.5.3.1. Domains
There are two situations of language use within the Karbi community that I have
experienced first hand and can comment on. One is life in the district capital Diphu, the
other is life in a Christian village in rural western Karbi Anglong. As would be expected,
the differences are very noticeable.
In Diphu, especially among the middle class, it is typical for Karbis under the age
of 35 to 40 years or so, to be quadrilingual. While Karbi is the native language that is
spoken in the home, it is also common for them to use Assamese, English, and Hindi
(probably in that order of frequency) on a regular basis. It is therefore no problem for
them to switch to whatever language is shared with their interlocutor(s). If several
languages are shared, it is typically with first priority Karbi and then Assamese that are
used, but that is not always the case, and people enjoy switching between languages. For
example, I have heard Karbi native speakers in their late twenties talk to each other using
Karbi, but suddently switch to Assamese or English words or entire sentences.
In the villages that I have visited, this multilingual situation does not exist to that
extent. While it is common for Karbis except for the older generation to speak Assamese
to varying degrees, they may only ever use Assamese when they go to the weekly or
biweekly market where many sellers only speak Assamese and no Karbi. There is a trend
for the younger generation, especially with increases in physical and virtual infrastructure,
to be exposed to English and Hindi and to know how to say a few things in these
languages. In addition, in the villages in western Karbi Anglong with Tiwa-speaking15
villages nearby, it is also common for Karbis to know Tiwa to varying degrees (and vice
versa).
15
Tiwa is a Bodo-Garo language spoken mostly inside the Karbi Anglong district.
20
1.5.3.2. Vitality
The overall vitality is currently good for Hills Karbi. Most commonly, Karbi is
transmitted to the younger generation and used in the homes.
1.5.3.3. Attitudes
The attitudes of the Karbi community towards their language are generally
positive. I have not met Karbis who did not consider their language important to them.16
The fact that the KLA (§1.1.4) exists also speaks to that: The KLA is a non-governmental
organization that survives on private donations; its members volunteer, with no monetary
compensation for their work.
According to the Indian constitution, every Indian citizen has the right to go to
school in his/her mother language, but this right is not strongly financially supported and
therefore not well implemented.
The Karbi language currently looks healthy. However, especially the increase in
infrastructure is changing so dramatically that consequences will likely become more and
more noticeable in the near future. Although the facts are that a) there are a large number
of speakers; b) the language is almost always transmitted to the children; and c) the
community is politically protected by having their own autonomous district, the dominant
Assamese language creates a lot of pressure that many in the community feel. The KLA's
largest current concern is the standardization of the writing system. This will serve both
the purpose of doing language maintenance as well as result in added prestige.
16
Although it is the case that since I was closely collaborating with the KLA the whole time, the group of
Karbis I came in contact with were people who particularly cared about their language.
21
1.5.6. Current and Suggested Future Remedial Action
22
used in a discussion of elaborate expressions in Khumi by Peterson (2010:96-7), and
DeLancey (p.c.) has come across stories with the same basic structure in other languages
of Northeast India as well.
Another narrative that is characteristic of the region (specifically the hill region
stretching from Northeast India across Southeast Asia) concerns the loss of an allegedly
previously existing script. This narrative is analyzed by Scott (2009) as a literary-
mythological account of an intentional decision by these peoples for an oral literary
tradition and against a written tradition. He makes this argument in the context of his
larger hypothesis that the hills peoples of Southeast Asia17 have a history of intentionally
fleeing the developing civilizations in the valleys (which were heavily built on slavery in
their early beginnings) in order to maintain (cultural and political) independence and
societal equality.
In Karbi, there also exists a narrative of a lost script that was used to write Karbi.
But, and this account is similar to stories in other languages of the region (Scott 2009),
the only record of the script was on a deer hide, which in times of starvation had to be
eaten in order to survive, and was therefore lost.
The Karbi song language is used for oral literature that is sung or chanted (hence
the name) rather than narrated. It is also referred to as the poetical language. According to
my language consultants, between the two major dialects of Hills and Plains Karbi, there
is an interesting relationship between song language and ordinary language such that
Hills Karbi song language words are ordinary language words in Plains Karbi and vice
versa.18
Many lexical items that occur in the Hills Karbi song language represent
borrowings from Khasi languages (with which there exists a history of contact, §1.3). For
example, the word um is used for ‘water’ in the song language, which is a common
17
That is, the peoples inhabiting the hill range that stretches across Southeast Asia, which Scott refers to as
Zomia, with the claim that that is not only a geographic label but also needs to be understood as an area of a
shared cultural-political history.
18
Note that it is not common for members of the Karbi language community to understand all the song
language words. They typically know a few individual words, but no more than that.
23
component of toponyms in western Karbi Anglong in names such as Umswai, Umlapher,
Umkachi (or Amkachi), etc. The song language is thus an important object for further
study in order to trace Khasi borrowings in Karbi. A first move in this direction is the
book Karbi lamlir achili (lit., ‘the seeds of the Karbi poetical language’), a collection of
Hills Karbi song language words (some of which with context in songs and chants in
which they are used) edited by eminent Karbi language and literature scholar Longkam
Teron (L. Teron 2008). It is furthermore a topic for future study to investigate the
grammatical structure of song language texts.
Note that an interesting aspect in the transition from traditional to modern culture
is that the song language is also used in modern (Indic, Bollywood-style) Karbi pop
songs. However, Christian songs do not make use of the song language but of the
ordinary language.19
There are a small handful of important names in Karbi linguistics, which should
be mentioned at the outset of this section. The chronologically first mention should be
made to George D. Walker, who published a fairly comprehensive dictionary in 1925. In
the mid 1960s, the late French missionary Father Balawan put together a word list with
parts of a grammatical description (published as Balawan (1978)). Not much later, Karl-
Heinz Grüßner worked on Karbi grammar. Around the same time, Karbi scholar
Professor Rongbong Terang published a Karbi dictionary (Terang 1974). Within the last
decade or so, substantial contributions have also been made by another important Karbi
scholar, Longkam Teron.
Early resources on Karbi include word lists by Robinson (1849), Stewart (1855),
and Kay (1904), as well as information provided in sections on Karbi in the Linguistic
Survey of India (Grierson 1903). A historical ethnographic description with some
19
There might be several reasons for this. For once, it might be because most Karbi speakers do not
understand song language words. Another reason could be that the song language is closely linked to the
traditional religious belief and rituals.
24
references to Karbi grammar as well as several texts with translation was “edited,
arranged and supplemented” by Sir Charles Lyall based on notes by the Indian Civil
Service officer Edward Stack (Stack and Lyall 1908). Several other early resources on
Karbi linguistics are listed by Grüßner (1978:218-21).
20
Grüßner has since digitized his old audio recordings.
25
(with the first element in the compound being t ‘wood’), Grüßner reports the word to
be me-dang (with the first element being ‘fire’). These subtle differences are
suspected to be part of an (again, subtly) distinct sociolect typical of the Christian
villages in the area. While the issue of Karbi varieties cannot be treated here in any
comprehensive way (see also §1.4), it should be noted that certain discrepancies in
grammatical description between Grüßner’s work and the present work might be due to
dia-/socio-lectal differences.
The lasting contribution of Grüßner’s work cannot be underestimated. The level
of detail and carefully researched description, especially given the historical context
when much less was known about the Tibeto-Burman language family, is truly
remarkable.21
In 1966, besides Father Michael Balawan, Father John Mariae also produced
work on Karbi grammar and lexicon, which was only later published and is now available
as a booklet called ‘Karbi Self-taught’ (Mariae 2007). An overview of the history of
research on Karbi grammar is offered by D. Teron (2011:148-57).
21
I also want to take this opportunity to again thank Karl-Heinz for all his support, his enthusiasm, and for
freely and fully sharing everything and anything he had that could possibly help me in my endeavours to
work on Karbi grammar.
26
In addition to the resources on Karbi grammar mentioned above, important
contributions are also a number of dictionaries. The first comprehensive dictionary (Karbi
to English and English to Karbi) was published by Walker (1925).
The Karbi scholar Bidorsing Kro produced a Karbi to Karbi dictionary with
explanations in Assamese and English that was first published in 2002, with a second
edition that has been published since Kro (2009). Another dictionary emerging from
scholarship from within the Karbi community is Taro (2010). A comparative dictionary
of Plains and Hills Karbi by M. Teron and Tumung (2007) represents an important
resource on the major dialectal divide within the language.
The Karbi Roman script orthography was developed by missionaries in the 19th
century. Currently there also exists a Karbi orthography using the devanagari-based
Assamese script. However, the Karbi Lammet Amei advocates for the use of the Roman
script. Although there is a standardized set of letters in the Roman script that adequately
represent Karbi phonemes, the orthography is to date not standardized, which is a matter
of concern for many in the community, specifically, of course, for the members of the
Karbi Lammet Amei.
The controversial issues arise with respect to a) word boundaries; b) capitalization
after adding prefixes; c) syllable boundaries; and, most importantly, d) the representation
of tone.
Regarding a) word boundaries, the status of clitics is controversial. For example,
it is not clear whether the very frequently occurring topic marker =ke (§10.7.1) should be
attached to the last word of the noun phrase that it is phonologically bound to, or not.
With respect to b) capitalization after adding prefixes, the most frequent issue is
the occurrence of the possessive prefix a- (§5.3.1) on proper names. For example, in the
simple noun phrase ‘this Karbi woman’, the word ‘Karbi woman’, i.e., Karbipi, needs to
have the a- prefix on it. That is, one could write this noun phrase as laso aKarbipi, with
a- in lower case and Karbipi capitalized, but that looks odd to some members of the
community.
27
The issue of c) representing syllable boundaries is often (but not exclusively)22
encountered when a vowel-initial syllable occurs in the middle of a word. This represents
a problem because vowel onsets are accompanied by glottalization, which is not
represented in the writing system, but is noticeable in this context since it prevents
resyllabification (see §3.3). For example, adding the predicate derivation suffix - ‘much’
(§6.5.5.1) to the existential copula results in a pronunciation of [d ʔ ]. As a result,
many members in the community dislike a representation as doo, resorting to either using
a dash (i.e., do-o) or an apostrophe (i.e., ’ ).
Finally, d) the representation of tone has typically been avoided. Neither the
Roman script nor the Assamese devanagari-based script has been successfully modified
to represent tone. Different proposals have been put forth to do so, but none has been
systematic; either only one tone (usually the most salient mid tone that features
glottalization) is represented and/or only certain rhymes are marked for tones (and not
actually consistently). For example, tone in open syllables that lack a coda consonant has
been represented with an <h> coda by some Karbis - however, it is sometimes the low
tone that is represented by this orthographic <h> coda (e.g., neh ‘1EXCL’), and sometimes
the mid tone (e.g., meh ‘fire’). Another proposal specifically aiming at the representation
of the mid tone in nasal coda syllables was to write an orthographic homorganic stop, e.g.,
(a-)tump for the plural word with a bilabial nasal coda, (a-)phant for the non-subject
marker with an alveolar nasal coda, or langk for ‘water’ for a velar nasal coda.
Except for where the representation is not systematic or phonological (i.e., the
attempts so far at representing tone), all of these issues are mostly political in nature and
there is no right or wrong in linguistic terms. In this dissertation, I follow the orthography
Grüßner (1978) used. This has the following implications for the four orthography
problems outlined above:
For a) word boundaries, clitics are written as one word with the element that they
are phonologically bound to. For b) capitalization after adding prefixes, the above
representation is used, i.e., lower case prefix with capitalized proper noun stem (i.e.,
aKarbipi). With respect to c) syllable boundaries, the apostrophe is used, i.e., ’ for the
22
The other situation where this issue arises is when a multisyllabic word contains a consonant
combination of /pl/, /pr/, /kl/, or /kr/ between two vowels, where the two consonants could be split up as
coda plus initial or an open syllable followed by a consonant cluster.
28
example word from above. Finally, for d), the representation of tone, Grüßner’s approach
with diacritical marks is used: the grave for low tone (e.g., low tone t ‘die’); the acute
for high tone (e.g., t ‘snatch, grab’); and the macron23 for mid tone (e.g., t ‘be short’).
In this dissertation, examples are offered with both a word line and a morpheme
line, where tone is only indicated in the morpheme line but not in the word line. The
details of the representation of tone is further discussed in §3.5.9; the details of the
representation of data in general in this dissertation is further discussed in §2.4.3.
23
Note that in his dictionary manuscript, Grüßner also sometimes used the circumflex for the mid tone, e.g.,
representing ‘be short’ as thî.
29
A discussion of clause types and clause combining, including non-declarative
speech acts, is offered in Chapter XI.
Finally, Chapter XII provides an overview of the major constructions that have
functions on the level of the larger discourse.
30
CHAPTER II
This chapter deals with the methodology and general approach to grammar
writing that underlies this dissertation, as well as the various data-theoretical aspects and
practical workflows involved in the collection and organization of the data that this
grammar is based on.
Most of the existing literature on linguistic data management has been published
within the fields of language documentation (Himmelmann 1998; Gippert, Himmelmann,
and Mosel 2006; Himmelmann 2006a; Woodbury 2011) and the description of
(especially endangered) languages (Austin and Sallabank 2011). While this grammar of
Karbi is primarily aimed to be a descriptive resource, attempts were made to incorporate
the insights from the recently emerged (or, as some would argue, revived) field of
‘language documentation’ (or ‘documentary linguistics’).
This chapter begins with a discussion of the general approach and theoretical
framework underlying this grammar (§2.1). In §2.2, an overview of the corpus, including
data types and data formats is offered. Aspects of primary data collection are discussed in
§2.3, while §2.4 deals with the processing of primary data to derive annotation files
representin the main corpus for this dissertation.
The approach taken in this dissertation contains three major components. First, it
is firmly rooted in a functional-typological framework. Second, it embraces collaboration
with the language community as the best approach for data collection and analysis. Third,
the analysis presented in this descriptive grammar gains explanatory force through a
historical-comparative perspective on the grammatical constructions that are discussed.
31
2.1.1. Functional-Typological Framework
32
cross-linguistically, which is due to the shared aspects of human life everywhere in the
world. But it also motivates why other aspects of language are shaped differently cross-
linguistically, which is due the aspects of human life that are different in different parts of
the world, just as there exist different cultures.
This dissertation project was initiated by the Karbi community through the Karbi
Lammet Amei (KLA; §1.1.4), and was therefore collaborative in nature from the
beginning. Due to this close collaboration with the KLA, numerous aspects of data
collection and processing were enhanced.
For example, the KLA (and specifically my closest collaborator Sikari Tisso)
identified speakers that were able to tell particular stories and provide particular
information. Due to the KLA’s own interest in the success of the project, they (and again,
most importantly Sikari Tisso) also greatly helped with all aspects of the practical
realization and facilitation of the project, which always represents a major and time-
consuming task in fieldwork (which at times is even impossible for an outsider linguist
without collaborators from within the community).
33
2.2. Corpus
The data that were collected and generated for this grammar are organized into
three subcorpora, as laid out in Table 2. They are 1) data based on individual recording
sessions; 2) context-free elicitation data; and 3) the lexical database. All of these data are
- or are based on - spoken data. Additionally, three other types of data were available: 4)
data from Grüßner's (1978) work and specifically his transcriptions and translations of
audio data he had collected, which, therefore, are also spoken data, 5) an extensive
dictionary manuscript that Grüßner provided me, containing 240 pages of morphemes
with indication of tone, and 6) data extracted from written texts, mostly school textbooks.
The most important data type are the data from recording sessions, as they
represent the most natural use of the language (§2.1.1). Most of the time and effort spent
on this dissertation research were devoted to the transcription, translation and analysis of
the speech recordings. Within this data type, I distinguish between primary or media data
and derived data or annotations. Within the derived data type, I distinguish between the
two types of representation/translation and grammatical annotation as shown in Table 3.
34
Table 3. Recording session based data types
Primary data / Audio recordings
media data Video recordings
Derived data / Representation Transcription
annotations & Translation Translation
Analysis Grammatical Parsed Karbi words
annotation English glosses for Karbi
morphemes
Part-of-speech labels for
Karbi morphemes
Grammatical / cultural notes
Text-based elicitation
As detailed in Appendix H, which shows the metadata for those texts that were
fully transcribed, translated, analyzed and further annotated, the primary data that these
texts are based on consist of a total of 01:21:36 (hh:mm:ss) of media files that include
video, and a total of 00:40:02 of audio-only recordings, for a total of about two hours. As
for the data derived from these media data, the two hours of spoken language are
represented by a total of approximately 12,500 transcribed Karbi words.
35
the media files can be said to mutually inform one another. For example, the analysis
(very broadly speaking, i.e. consisting of parsing, glossing, assigning part of speech
labels, etc.) is derived from the transcription, translation, and media files, while also in
return informing transcription and translation.
Basic
Annotation Transcription Translation
Level
A distinction between primary data and data from how primary data is further
processed (i.e., derived data), as done in Table 3 above, is typically made in the language
documentation literature, although the dividing line may be drawn in different ways.
For example, Himmelmann seemingly distinguishes between primary data and
their transcription and translation on the one hand and the analysis on the other hand in an
earlier publication (1998:161–2), but in a later publication (Himmelmann 2006a:14)
distinguishes between primary data consisting of "recordings/records of observable
linguistic behavior and metalinguistic knowledge (possible basic formats: session and
lexical database)" and the "apparatus", which consists of metadata and annotations
(including transcription, translation, and "further linguistic and ethnographic glossing and
36
commentary"). That is, Himmelmann also distinguishes between essentially three types
of data (primary - transcription and translation - analysis), but puts the transcription and
translation in earlier work with the primary data and in later work with the analysis. As
represented in Table 3, I consider transcription and translation derived data or annotations
just like the different types of grammatical annotations (see also Schultze-Berndt 2006).
Primary data in my conceptualization are media data, i.e. audio and video
recordings as well as images such as photographs of the speakers and of the recording
session. The idea is that primary data are the most objective, 'raw' data.
Derived data are derived from primary data and include transcription and
translation on the one hand, and analysis of the texts on the other hand. We can
distinguish here the former two from the latter in that transcription and translation are still
more objective and can be fairly easily carried out by trained research assistants, while
the analysis requires a fully trained linguist. The analysis may include grammatical
annotations such as the parsed Karbi text, morpheme-by-morpheme glosses and part of
speech tags as well as grammatical and cultural notes for different layers of syntactic
analysis of the given parts of a text and ethnolinguistic commentary. The analysis may
also include separate files with elicitation based on the given parts of a text.
Although there is a sense that we can distinguish between primary and derived
data, where primary data are objective and 'raw', while further data that are derived from
the primary data consist of more (inter)subjective analysis and interpretation that may
turn out to be incorrect or flawed, this distinction is actually blurred. In fact, even with
the primary, 'raw' media data, no true objectivity is guaranteed, since even here,
subjective decisions about where to set up the camera and microphones and what types of
microphones to choose have an impact on the resulting recordings (Good 2011).
37
Phonological elicitation was mostly aimed at the Karbi tone system, e.g., eliciting
roots that undergo morphophonemic changes due to adding certain prefixes, or elicitation
of words with similar tone patterns.
The lexical database has been built up parallel to the text database of recording
session based data due to the way the software Toolbox, which was used for grammatical
annotations, works. Therefore, the lexical database includes all morphemes that occur in
the recorded texts with some additional opportunistic data entries. It is a basic lexical
database, which will require further work to be usable as a dictionary. As of now, entries
generally only consist of the Karbi morpheme, an English gloss, a part of speech label,
and in some cases additional grammatical, lexical, or cultural notes.
Data formats ideally depend on their purpose: a) for the purpose of working with
the data; b) for the purpose of presenting the data within the academic research
community or the Karbi community; or, c) for archiving and longevity of the data
(Johnson 2004, 146; Austin 2006, 96–7; Good 2011, 227–8).
Table 4 summarizes what formats are used for what data type depending on the
respective goal. For the master copies of all types of files, archival formats were chosen
following best practices, which recommend the use of uncompressed, widely used, non-
proprietary formats in order to ensure long-term preservation and avoid loss of quality
(Bird and Simons 2003; Johnson 2004; Austin 2006).
The choice of working format depends on the software used, and in the case of
these Karbi data involve, for example, the .eaf format, which is the format the ELAN
software generates.24 Still the working format should also be uncompressed, i.e. be of the
highest possible quality, for example the working format for audio recordings is the
uncompressed .wav format, which is also the archiving format. Video files, on the other
hand, involve three different working formats. The .mts format is the high definition
format that is generated by the video camera used for this project (the Canon Vixia HF S-
24
Note, however, that the .eaf format is in fact a type of XML format (see MPI LAT).
38
10). This format is neither a good working, nor presentation, nor archiving format and
therefore has to be converted first. As a working format, a smaller format is advantageous
since it requires less storage space and runs faster. For the current project, high resolution
video data were not required and so we typically used smaller .wmv or .avi formats as the
working format as they are supported by ELAN.
The presentation format is largely intended to serve Karbi community members.
Therefore, the important consideration is to try and make the files small and portable and
widely accessible. That means that unlike the working and archiving formats, the
presentation format does not have to be of the highest possible quality but it is more
important that data transfer is easy and that the data do not get modified. Therefore, .mp3
for audio or .pdf files for text data are preferred presentation formats.
Table 5 provides a list of all software used in the course of the project. For basic
text annotation, ELAN26 was used. ELAN (the ‘EUDICO Linguistic Annotator’) is
software developed by the Language Archiving Technology (LAT) group of the Max
Planck Institute (MPI) for Psycholinguistics in Nijmegen, The Netherlands. The other
main software, which was used for linguistic analysis and the creation of a lexical
database, is Toolbox.27 Toolbox was developed by the Summer Institute of Linguistics
25
Text data includes data types such as elicitation data, tabular data, coding schemes, concordance /
examples lists, and questionnaire responses.
26
ELAN, developed by the Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands:
http://www.lat-mpi.eu/tools/elan/
27
Toolbox, developed by SIL International: http://www.sil.org/computing/toolbox/
39
(SIL). For phonetic analysis, Praat28 was used. In order to edit audio files, the freely
available Audacity29 software was used. In order to extract audio from video files, the
VLC media player30 software was used. Lastly, to write texts and create tables, Microsoft
Word and Microsoft Excel were also used.
Table 5. Working and presentation software used for different data types
Type Working software Presentation software
Audio recordings Audacity, ELAN, Praat, Audacity
Emu
Video recordings VLC media player, ELAN VLC media player
Images N/A
Transcriptions ELAN ELAN
Translations Toolbox, ELAN
Grammatical Toolbox
annotations
Lexical database Toolbox, MS Word MS Word, Adobe Reader,
(Lexique Pro)
Text data MS Word, Toolbox, MS MS Word, MS Excel, Adobe
Excel Reader
Data collection was carried out over a total of 15 months consisting of five phases:
January - March 2009 (phase 1); February - May 2010 (phase 2); January - March 2011
(phase 3); September - December 2011 (phase 4); September - December 2012 (phase 5).
Specifically, phase 1 mostly consisted of word elicitation aimed at Karbi
phonology and simple sentence elicitation without recording much. During phase 2 a
festival to celebrate Karbi culture took place in the local capital Diphu. This festival
brought a number of village elders to Diphu, about ten of who agreed to be recorded,
performing a variety of genres. We were able to record them in a recording studio in
Diphu and collected a lot of primary data in the form of recording sessions involving both
audio and video data. These data are mostly folk stories as well as some procedural texts.
28
Praat, developed by Paul Boersma and David Weenink: http://www.fon.hum.uva.nl/praat/
29
Audacity, developed by Dominic Mazzoni and Roger Dannenberg at Carnegie Mellon University:
http://audacity.sourceforge.net/
30
VLC media player, developed by the VideoLAN Project: http://www.videolan.org/vlc/
40
A lot of song data were also recorded during phase 2, which, however, have not been
analyzed yet since Karbi song language is entirely different from the ordinary language
and requires further research (§1.6.2).
During phase 3 some additional phonological data were collected. As part of
phase 4, a one-week recording trip was carried out in November 2011, which resulted in
a number of recorded interviews, conversations, narrations of local histories and folk
stories, most of them pertaining to a research project dedicated to investigating the status
of women in Karbi society. During the final stage of phase 5, a few other texts were
collected, including an on-line narration of the Pear Story (Chafe 1980).
I would like to acknowledge the participation and support from the following
people: Maloti Rongpharpi, Kache Rongpharpi, the Assistant Teachers at Lorulangso LP
School: Punyalata Ingtipi and Rani Teronpi, Maggie Katharpi, Chandra Kanta Terang,
Puspa Engtipi, Kahan Terangpi, and Dr. Janta Pator (Diphu); Burnesh Milikpi, Keshop
Terang, Jugal Timung, Rajen Kro, Dim Teron, and Hemari Rongpi (Jyrikyndeng); Aren
Ingti, Pretty Ingtipi, and Kamal Chandra Kro (Dongkamokam); Chandra Sing Tisso
(Podumsarpo); Harsing Ronghang, Ronghang Lindok and his assistants (Ronghang
Rongbong); Longsing Tokbi (Amtereng); Bidyaram Rongpi (Hamren); Dhansing Terang,
Kare Rongpipi, and Harsing Kro (Boythalangso); Anjan Teron and Dhaniram Ingleng
(Bhoksong); Modon Kro (Rambangla); Sarthe Phangcho (Putsari Hindu arong); Mahin
Phangcho (Umpanai); Hem’ari Ingjai (Pharkong Abi); Sika Hansepi (Men Terang);
Kache Kropi (Balijuri); Chandra Sing Teron; Kasang Teronpi (Dingso Terang);
Rongbang Teron and Seng Tisso (Ujandongka); Welisbon Ronghipi (Umswai Model);
Longsing Bey (Murap, Umlaper); Joysing Tokbi (Laru aum, Umlaper); the people of
Sohliya (Meghalaya); Owen Terang and the people of Marmein (Meghalaya); Dhiren
Ingti and Dhiren Ronghang (Kamrup); Khayasing Hanse (Bokoli); Sarhon Ronghang,
KLA; Sangvai Teronpi and Hangmiji Hanse (Hongkram); Dr. Mansing Rongpi, MLA,
Dispur; and Father Joseph Teron, Don Bosco. Kardom.
41
2.3.2. Data Collecting Team
The data collecting team consists first and foremost of Mr. Sikari Tisso, who
represents the Karbi Lammet Amei (§1.1.4). As my main collaborator on this project, he
was involved in and/or facilitated almost every recording session. He also interviewed
various Karbi speakers for this project.
Mr. Tisso was born in 1955 and grew up in Western Karbi Anglong but moved to
Diphu in his late twenties. He says of himself that he now speaks an idiolect where he
mostly speaks the main Diphu dialect of Karbi interspersed with several features of the
Western Karbi Anglong Amri dialect that he grew up speaking.
Furthermore partially involved in the primary data collection efforts were
especially Mr. Bhudeswar Timung, as well as Ms. Amphu Rongpipi, and the Kro and
Hanse families in Diphu and Umswai, respectively. Mr. Bhudeswar Timung played an
important role in the data collection trip of November 2011, where he interviewed
speakers and also facilitated data collection otherwise. He currently lives and is originally
from Socheng in eastern Karbi Anglong. He speaks the standard (Rongkhang) dialect of
Karbi. Ms. Amphu Rongpipi facilitated the collection of recordings from her mother,
Puspa Ingtipi, and her grandmother, Kahan Terangpi. She and her family are from Diphu
and speak the standard dialect. The Kro family has been my main host family. I stayed
with them and enjoyed their generous hospitality for large periods of my time in Karbi
Anglong. I was able to record Mrs. Sashikola Hansepi and Ms. Rasinza Kropi of the Kro
family. The Hanse family was my host family in Umswai, West Karbi Anglong for a few
weeks in fall 2011 and fall 2012. They also facilitated the recording of native Karbi
speakers of Umswai.
Metadata information about the fifteen speakers of the fully annotated texts that
represent the main corpus for this dissertation is provided in Appendix H. Before
recording sessions, the data collection team made sure to obtain informed consent from
the speaker(s) that we were going to record following standard practice (see for example
Dwyer 2006:43-5) as well as fulfilling requirements imposed by the University of
Oregon's Institutional Research Board (IRB). To that end, consent forms were created in
42
English and translated into Karbi, which are attached as Appendix I.31 The consent forms
gave speakers the following explicit options:
The speakers were asked to check the appropriate and sign and date. Although
Dwyer (2006:44) reports that in certain cultural settings "written forms may breed
mistrust" and that they "may wisely be viewed with suspicion", my experience was that
the written consent forms were actually viewed as carrying prestige, and one of my host
families asked for another copy to keep and told other people about it in my presence.
31
A good idea would have been to have designated fields on the consent form for speaker metadata.
43
Table 6. Text genres
Spontaneity Category Genre
Less spontaneous Monologue Folk tale
Procedural text
Personal narrative
Pear story
More spontaneous Dialogue Interview/Conversation
The folk tales are the least spontaneously produced genre, as they follow a given
plot line and to some degree probably also particular structural patterns. Despite this (in
some sense, undesirably) low level of spontaneity, this genre was recorded the most: Not
only was it very important for the Karbi Lammet Amei (§1.1.4) to record as many folk
tales as possible, but this genre also has a high priority from a documentary viewpoint as
it represents the major component of the Karbi oral literature (§1.6.1). As such, it is
important for academic disciplines besides linguistics, such as anthropology, ethnography,
or folklore studies, and, it can also provide information on the cultural-linguistic history
of the Karbis, thus relating back to being of immediate concern to linguistics.
Procedural texts may or may not be as spontaneous as folk tales. Some of the
procedural texts in the main corpus narrate traditional cultural practices that are likely to
be retold and transmitted, and therefore not spontaneous. An example is the text about the
alkaline food kangmoi ahan that involves burning jhum fields and picking up the ashes as
an ingredient (‘SiH, KH’). Other procedural texts, however, such as the recipe for pork
with fermented bamboo shoots (‘PI, BPR’), are clearly spontaneously produced.
Personal narratives are intermediate with respect to spontaneity. The two personal
narratives in the main corpus, ‘SH, CSM’ and ‘SiT, HF’, both represent retellings of trips
to festivals the day before. In both cases, the speakers retell the events from their
memories without much planning or outlining, but due to the nature of them being
monologues, less spontaneity is involved.
The Pear Story (Chafe 1980) was recorded as an on-line narration: The speaker
was telling what was happening as he was watching the video clip. Therefore, this was all
spontaneous.
Finally, another recording made from a spontaneously produced genre, which was
fully annotated, were about two thirds of an interview between two speakers. While the
44
interviewer had an outline with questions he wanted to ask, everything produced on part
of the interviewee was fully spontaneous.
To ensure highest quality, audio recordings were created in .wav format with a
sampling rate of 48 kHz with a bit depth of 16 or 24 following best practices that
recommend at least 44.1 kHz and 16 bits (Johnson 2004, 147; Austin 2006, 107), while
anticipating that 48 kHz and 24 bits will be the future archival standard for audio files
(Nathan 2011, 260). During all recording sessions, the actually recorded audio was
monitored by using headphones (Austin 2006:90).
Another aspect involved in the attempt to collect high quality audio data has to do
with the recording setting. A notorious difficulty in collecting high quality audio data in
the fieldwork context is the level of background noise. This was certainly the case for this
project, where it often seemed impossible to escape the constant background noise.32
However, early on in the project, Sikari Tisso was able to identify a sound studio in
Diphu, owned privately by Mr. Chandra Kanta Terang, who made it available for
recordings for this dissertation research. As a consequence, a number of recordings were
made in the sound studio in order to get audio data of better quality by eliminating most
of the background noise.
However, a new issue (of unclear ramifications) that arose as a result of solving
the background noise problem, was that this recording studio would get hot and stuffy
after awhile since it did not have any direct opening to the outside. Due to the stuffiness,
other native Karbi speakers were not very motivated to join the recording sessions,
resulting in several sessions that were recorded with speakers one-on-one, which is not
ideal considering the goal of obtaining data in natural settings. Nonetheless, it is not clear
32
Specifically, background noise such as animal sounds, people chatting or singing, children crying, people
physically working on some project, echo inside rooms, noisy electricity or generators, fans, etc.
45
what type of impact (if any) this procedure had on the speech or grammar used in the
recordings obtained that way. It might be the case that some speakers were actually more
comfortable being in a dark room by themselves without other people gathering around
them and watching them.
The recording equipment that was used includes a high definition video camera,
two different digital audio recorders, and various microphones for use in different
recording contexts (for details, see metadata spread sheet in Appendix H). More often
than not, a session was recorded both with the video camera and with the audio recorder
to ensure high quality audio data.33
A variety of microphones were used, including a cardioid condenser hand-held
microphone, an omni-directional condenser hand-held microphone, 2 lavalier clip-on
omni-directional condenser microphones, one head-mounted unidirectional dynamic
microphone, and one omni-directional dynamic hand-held microphone (see metadata
spread sheet in Appendix H for exact models).
The most versatile and most often used microphone was the hand-held condenser
microphone, which was used in various settings. Depending on the model used on a given
the trip, the hand-held condenser microphone picked up sound either in a cardioid or an
omni-directional pattern, both of which can be used to record several speakers, and
thanks to the above-mentioned recording setting in a recording studio (§2.3.5.2), the
omni-directional microphone did not pick up too much background noise in the
recordings obtained in that setting.
The condenser omni-directional microphone (which has enhanced capabilities of
picking up sound due to additional battery / phantom power) proved a lot more useful
than the dynamic omni-directional microphone (which in fact turned out useless).
The two omni-directional lavalier microphones were used in recording interviews
and other conversations, especially those involving just two speakers.
33
This practice proved useful since the special cable connecting external microphones to the video camera
(female XLR to 3.5mm mini plug) turned out - after several recordings - to be of inconsistent quality during
the 2010 trip.
46
Lastly, the head-mounted unidirectional microphone was used for phonetic
recordings, as well as in the monologue narration of the Pear Story. The head-mounted
type was considered somewhat uncomfortable by speakers, and thus was not used much.
This section describes the methods and workflows followed in processing primary
(i.e., media) data.
47
Transfer media files to laptop
Label files
Back up on
Toolbox: Analysis external
hard drives
After all files were accordingly labeled, some of the audio files were edited to cut
the beginning or end before the speaker/singer started or after s/he had ended. This was
done using the free software Audacity. In addition, video files also had to first be
converted before proceeding further. The video camera used for this project generates
AVCHD or .mts files, which cannot be used for processing the data. Therefore, all
video .mts files were converted into .avi or .wmv format for working/processing purposes.
Furthermore, using the freely available VLC media player software, audio files in .wav
48
format were extracted from the video files. The reason for this is that ELAN (currently)
only provides a waveform (which is helpful for segmenting and transcribing) if an actual
audio file is uploaded. Therefore, audio files were extracted and uploaded into ELAN
together with their video files in order to obtain a representation of the waveform.
Next, an ELAN file was created for segmentation and transcription purposes (see
§2.4.2.1 and §2.4.2.2 below). The ELAN file was then exported into Toolbox to proceed
with the analysis. In order to supply Toolbox with the fields necessary to do the analysis,
i.e., at least the "/mb" field for the parsed Karbi text, the "/ge" field for the English gloss,
the "/ps" field for part of speech labels, and the "/ft" field for the free translation, I added
these fields into new tiers in ELAN and left them blank in ELAN. After export into
Toolbox, these fields were exported as well and were ready to be used in Toolbox for the
analysis (§2.4.2.3), the free translation (§2.4.2.4), and the ethnographic and linguistic
notes (§2.4.2.5).
2.4.2. Annotation
In the subsections to follow, I discuss the various annotation components, both the
more basic components of segmentation, transcription, and translation, as well as the
types of annotation that both represent and are based on the analysis of Karbi grammar.
This notion of 'annotation' including all of transcription, translation, and analysis follows
the terminology of Bird and Liberman (2001) and Schultze-Berndt (2006).
The discussion is roughly ordered according to the typical workflow: what needs
to be done first is the segmentation of the audio or video file into intonation units
(§2.4.2.1), then followed the transcription (§2.4.2.2) (sometimes but rarely along with the
translation), and then followed the analysis including parsing, glossing (§2.4.2.3),
typically the translation (§2.4.2.4), and adding ethnographic and grammatical notes
(§2.4.2.5).
2.4.2.1. Segmentation
Segmentation of audio files of spoken texts was carried out based on auditory
impressions, such that each segment represents an intonation unit. Typically, I myself did
49
the segmentation, mostly due to the fact that this was something I could do relatively
quickly compared to research assistants who did not use computers on a regular basis.34
2.4.2.2. Transcription
The transcription of texts was carried out by native Karbi speaking research
assistants Amphu Rongpipi, Klirdap Langne Teronpi, Serdihun Beypi, and Larshika
Tissopi. All of them read and write English and Karbi.
All texts were transcribed using a Karbi orthography that follows Grüßner’s (1978)
work (§1.8); phonetic transcriptions were not prepared. Within the orthographical
transcription, however, certain very basic non-linguistic and paralinguistic aspects were
transcribed as suggested by Schultze-Berndt (2006:229): whispering, laughing, the
existence of short pauses, and hesitation markers are indicated. Furthermore, false starts
were transcribed, as also advised by Himmelmann (2006b:269), as they "may prove
crucial for various interpretative and analytical tasks."
An issue was the accurate transcription of minor dialect differences such as using
a different vowel. For example, one dialect has richo for 'king', whereas the standard
dialect has recho (see §1.4.4). Research assistants transcribing texts would
understandably often slip into transcribing how they would say a word instead of how the
recorded speaker said it. Overall, text transcription was very close to the text. False starts
and other types of misspoken words were indicated by angular brackets < >.
The way Toolbox works, every project typically involves a text database and a
lexical database. Because of that, morphological parsing and glossing becomes semi-
automated with a growing lexical database.
In choosing glosses for grammatical morphemes, the conventions of the February
2008 version of the Leipzig Glossing Rules were followed (see also §2.4.3).
34
This was generally possible due to the considerable overlap in the phonological marking of intonation
units cross-linguistically (Himmelmann 2006). There still remained a few segmentation issues in some
cases that had to be changed at a later point.
50
2.4.2.4. Free Translation
Free translations from Karbi into English were typically added while analyzing a
text, intonation unit by intonation unit, although a few texts were translated in the process
of being transcribed. The type of free translation provided for most texts was rather close
to the original Karbi structure, so in fact more literal than free. This decision was based
on the wish to highlight Karbi idiomatic expressions or constructions. Where the actual
meaning became obscure, a brief explanation was added in the free translation or in the
ethnographic notes. As advised by Schultze-Berndt (2006:236), anything that had to be
added in the translation in order to render the particular English clause grammatical, most
typically noun phrases or pronouns, was put in brackets to indicate that the corresponding
form was absent in the Karbi original text.
Ethnographic notes were added, for example, for customs that were perhaps
referred or alluded to in a particular clause in a text.
Linguistic notes were added if, for example, a) a language consultant told me that
the use of a particular morpheme was unacceptable in her/his dialect or odd; or b) if there
was something peculiar about a phonological issue, such as a much lengthened vowel; or
c) when doing minor text-based elicitation, such as asking whether a particular
information structure marker could be replaced by a different one with roughly the same
meaning.
51
(a) atomo kethan kangton’iklo; kardom'iklo ho
(b) [a-tom ke-th n ke-ingt n- k-l ] kard m- k-l ho]
(c) POSS-story NMLZ-tell NMLZ-conclude-FRML-RL GREETING-FRML-RL EMPH:INTERACT
(d) ‘[…] this matter I'm telling, just this one thing (i.e., doesn't want to tell any other
stories), I'm finished telling the story, thank you’ [SeT, MTN 052]
In (a), the word line, tone is not marked (see §3.5.9), except in some cases where
prosodic extra high pitch occurs, typically serving as non-final marking, as in (2) below
(see §11.2.1.2). The spelling conventions for Karbi that are used in the word line (and
elsewhere) are outlined in §1.8. Square brackets with three dots […] are included
whenever either the beginning or the end of an intonation unit, or both, are left out
(because they are irrelevant to the point being made). In order to indicate words or parts
of words that the speaker did not intend to say (where s/he misspoke), <angle brackets>
are used. If these are full words, then they are also listed in the morpheme and gloss lines
(and translated in the free translation line).
In the morpheme and gloss lines, (b) and (c), there are three symbols that may
indicate a morpheme boundary: the dash (‘-‘) as the default, which includes the
morpheme boundary between affixes, and between roots in compounds; furthermore, the
equal sign (‘ ’) for clitics; and finally, the tilde (‘~’) for reduplication.
Moreover, in the morpheme line (b), tone is marked following the conventions
discussed in §3.5.9. This line may also include square brackets that indicate constituency.
In the gloss line (c), abbreviations follow the Leipzig Glossing Rules (§2.4.2.3) where
possible. Borrowings are indicated where known, e.g., t ‘person(<Asm)’ is identified
as being a borrowing from Assamese. The period (‘.’) is used when more than one
English word or abbreviation is needed as a gloss for the Karbi morpheme. The colon (‘:’)
is used to indicate a subtype: for example, ‘CLF:round’ stands for the classifier that
classifies round entities. Here, the first component of the gloss, ‘CLF’, represent the
general type or category of element, while the second component, ‘round’, indicates the
particular subtype of classifier, i.e., the classifier for round entities.
Finally, the free translation line (d) corresponds with the word line in indicating
(with square brackets and dots inbetween) when the beginning or end of the intonation
unit are not included. Parentheses are used when an additional word is needed in the
translation to render the English sentence grammatical, or when context is needed that is
52
pragmatically understood in Karbi but not in English, e.g., the fact that in (2) refers to
the wife’s parents’ place.
(2) Tone marking in word line: prosodic extra high pitch on - ‘realis’ as marker of
non-final status
amat la apiso abangke akhalun
am t [l a-pis ab ng ke] a-khal n che-ingh r-t ng-l
and.then this POSS-wife NPDL=TOP POSS-kd.big.basket RR-carry.load-finish-RL
ha kedamlo t
h ke-d m-l am t
over.there NMLZ-go-RL and.then
'and then, the wife put the basket rope around her head and was just leaving to go over
there (to her parents' place)' [SeT, MTN 041]
53
CHAPTER III
3. PHONOLOGY
3.1. Consonants
There are a total of 18 consonant phonemes in Karbi that contrast with each other
in minimal sets. All but one of the 18 phonemes, which is the velar nasal /ŋ/, occur at the
54
beginning of syllables (see §3.1.1 and Table 8), whereas the syllable coda position is
limited to a much more restricted set of consonants (see §3.1.2 and Table 16).
In the class of syllable onset consonants (see Table 8), stops are the only manner
of articulation that exists at all places of articulation except for the glottal stop.
Phonetically, there is a glottal stop in the language, which, however, only surfaces as part
of the mid tone and occurs in conjunction with glottalization across the whole syllable
(see §3.5), as well as with syllable-initial vowels (§3.3). Note that Table 8 shows one
phoneme in two different cells: the palatal /ɟ~j/ has allophonic variation in its manner of
articulation, and is therefore given as both a stop and a glide. Details will be discussed
below.
The bilabial voiceless aspirated stop /ph/ is in free variation with a bilabial
voiceless fricative /ɸ/. It seems as though the younger generation, and particularly
speakers of the Christian sociolect, tend to use /ɸ/ more. It should also be noted that the
name of the capital Diphu as a specific lexical item is most commonly pronounced with
35
Representation in <angle brackets> indicate the orthographic representation of the respective phoneme
followed in this grammar.
55
/ɸ/. Grüßner (1978: 12) also noted the use of the fricative /ɸ/ in the then younger
generation and ascribes it to contact with the dominant language Assamese, where the
fricative production is the standard realization of orthographic, or perhaps ‘original’, <ph>
according to him. Scott DeLancey (p.c.) notes that this allophonic alternation between
/ph~ɸ/ also occurs in Boro (Boro-Garo, Tibeto-Burman; Assam, Northeast India), and
may be an areal feature. Indeed, the alternation between /ph~ɸ/ geographically extends
into Northeastern Bhutan, where it is found in at least Kurtöp (but presumably other
languages as well) (Hyslop 2011: 106).
The alveolar stop series /d/, /t/, /th/ is complete, while the palatal and velar series
each lack one member: the palatal series /ɟ/, /c/ (<ch>) lacks the voiceless aspirated stop,
while the velar series /k/, /kh/ lacks the voiced stop. Lacking specifically the velar voiced
stop is common across phonological systems of languages of the world for a phonetic
reason: it is more difficult to produce a negative voice onset time (VOT) for the velar
place of articulation because there is less room in the vocal tract for voicing to build up
(Ohala 2010: 667).
The palatal voiced stop [ɟ] is in free variation with the palatal glide [j]. The choice
of which allophone should be considered primary largely depends on whether one wants
to focus more on synchrony or more on diachrony. The stop articulation [ɟ] is more
common in the dialects now considered standard. Perhaps related to the orthographic
representation as <j>, there is a prescriptive tendency in favor of the stop [ɟ], while
rejecting the glide pronunciation. However, especially in syllable onset but word-medial
position, most native speakers at least sometimes produce a glide instead of the stop. The
glide [j] pronunciation appears quite frequent in the more conservative varieties in West
Karbi Anglong, and perhaps also more so among older people. It also seems that
sometimes an intermediate, fricative-like version is produced such that it almost seems to
be more of an allophonic continuum of manner of articulation that spans from a more
stop-like across a more fricative-like to a more glide-like production.
56
Grüßner (1978: 12) also reports a voiced “fricative” production of /ɟ~j/,36 which
he says he especially noticed among older people and particularly in the context of
singing and reciting traditional texts. He also cites data from Robinson (1849), which
show transcriptions of this phoneme sometimes as <j> and sometimes as <y>. Grüßner
(1978: 12) suspects that there was and is an ongoing sound change from the “fricative” to
the stop citing Shafer (1966) for the claim that “*/y/” is historically earlier. More recent
research by Matisoff (2003) suggests that the Karbi /ɟ~j/ phoneme does indeed
voiced fricative [v] as suggested by the spelling). By describing this as a bilabial voiced
fricative /β/, we can explain both salient facts, which are, on the one hand, observing
speakers produce this consonant more like a bilabial rather than a labiodental, and, on the
other hand, especially preceding unrounded vowels, it sounds a lot more like a fricative
(and, indeed, closer to [v]) than a glide. The exact production of this phoneme also seems
to be variable between a more fricative-like and a more glide-like articulation when
comparing different dialects. While the subdialects of Hills Karbi tend more towards a
fricative-like articulation, a more glide-like articulation is found in the Ri-Bhoi variety
spoken in the border area between Assam and Meghalaya.
36
Grüßner (1978: 12) writes that it is a fricative using [j] as a phonetic symbol for it; it appears that he has
the glide in mind instead.
37
Evidence includes the following forms: ‘to stand’ from Proto-Tibeto-Burman *r(y)ap; hijap ‘fan’
from *ya:p; ‘rat’ from *b-yəw; ‘to point’ from *y ŋ ‘f ’; ‘night’ from *ya; ‘to sell’
from *ywar; t ‘deer’ from *d-yuk; and ‘to be lightweight’ from *r-y :ŋ.
57
In onset position, nasals are limited to bilabial /m/ and alveolar /n/. Other
sonorants include the lateral /l/ and rhotic /r~ɾ/, the latter of which is usually produced as
the flap /ɾ/ in onset position rather than the trill /r/.38
The onset position allows clusters of two consonants. These are exclusively
combinations of stops with the lateral or the rhotic. Permissible onset clusters are
discussed in §3.3.
A minimal triplet for the bilabial stop series (voiced, voiceless, aspirated) in onset
position with an open, mid tone rhyme is presented in Table 9.
Table 10 offers a minimal triplet for the alveolar stop series (voiced, voiceless,
aspirated) in onset position with an open, low tone rhyme.
Table 11 gives a minimal pair for the two palatal stops: voiced /ɟ~j/ (<j>) (also
listed in a minimal set with liquid onsets in Table 15) and voiceless unaspirated /c/
(<ch>), with a low tone, velar nasal coda rhyme.
38
The rhotic occurs in an onset cluster with the voiceless aspirated alveolar stop in two lexical items, t
‘six’ and t ‘seven’. In this case, the rhotic is produced like the approximant /ɹ/ (see §3.3 and Table
25), which is a production also found in coda position (see §3.1.2).
58
Table 12 offers a minimal pair for the aspirate-nonaspirate contrast among
voiceless velar stops. The voiced velar stop is not a native phoneme of Karbi (but see
§3.1.1.4 for the voiced velar onset in borrowings).
In Table 13, monosyllabic roots with the same, stopped low tone rhyme but
different fricative onsets, voiced /β~w/ (<v>), voiceless /s/, and voiceless /h/, are
contrasted.
Table 14 and Table 15 give minimal pairs for the two nasal onset phonemes /m/
and /n/, and for the liquids /l/ and /r/ and the palatal glide (/stop) /ɟ~j/, respectively.
59
3.1.1.4. Marginal Onset Consonants and Consonant Clusters
39
In particular, those with some fluency in Assamese, Bengali, and/or Hindi.
60
In syllable-final position, there is only one member each of stops and nasals at the
bilabial, alveolar, and velar positions, as well as the rhotic. Stops are unreleased. The
velar nasal is the only syllable-final consonant that does not also occur as a syllable onset.
The rhotic /r~ɾ/ varies between a more flap-like production and a more trill-like
Table 17 shows two sets of three monosyllabic verb roots that are minimally
contrasted by their coda stops.
Table 18 offers two sets of near minimal sets contrasting sonorant codas, though
not consistently with the same tone. The /d/-initial set has the same tone for nasals at all
three places of articulation, however.
3.2. Vowels
Karbi has a five vowel monophthong system with /i/, /e/, /a/, /o/, and /u/, see
Figure 9 (though also note §3.2.1). In open syllables, the vowels are articulated as more
maximally dispersed from each other, while in closed syllables, they are more centralized.
While /i/, /e/, /a/, and /o/ are produced in a manner very similar to the basic phonetic
values of [i], [e], [a], and [o], the /u/ vowel is often produced without much lip rounding.
In syllable-initial position, vowels are always preceded by a glottal stop (also see §3.3).
i u
e o
a
Table 19 offers minimal sets for vowel monophthongs in open syllable roots with
a bilabial voiced /b/ initial. The data are organized into three sets according to tone: the
first set has low tone roots, the second set has mid tone roots, and the third set has high
tone roots.
62
Table 19. Minimal sets for vowel monophthongs in open syllables
/i/ ‘to be small’, ‘goat’ ‘to keep’
/e/ ‘handle’, ‘to ‘CLAN’ ‘to sieve’
chase away’
/a/ ‘paddy disease’ ‘to carry (child) on back’
/o/ ‘to apply fish ‘inner part of fruit’ ‘to sacrifice’
poison’
/u/ ‘to carry (child) on back’, ‘to plait’,
‘small bamboo basket’ ‘bundle’
Karbi rhymes /ei/, /ai/, /oi/, and /ui/ may be analyzed as vowel diphthongs (see
§3.4). Table 20 provides a minimal set of these rhymes following the same initial /b/ as in
Table 19. Diphthongs are never followed by a coda consonant, they only occur in open
syllables. They do occur with all three tones. Note that the /ei/ diphthong only occurs in
the Amri dialect of Hills Karbi spoken primarily in West Karbi Anglong. The Amri
dialect has both the /ei/ and the /ai/ diphthongs, while the standard Rongkhang dialect has
merged them to /ai/ (see §1.4.3 on Hills Karbi dialects).
Although both examples with the /oi/ rhyme in Table 20 are borrowings from
Assamese, there are lexical items with /oi/ in Karbi that are not borrowed from Assamese
as well, such as ‘to agree’ or ‘back’.
As mentioned above, the Hills Amri Karbi variety is interesting in that it has the
/ei/ diphthong that does not exist in the standard Hills Karbi (Rongkhang) variety. The
Amri Karbi variety is also very interesting for its vowel monophthong system. While the
five monophthong vowel system shown in Figure 9 above is the accepted standard vowel
inventory of Karbi, the Amri Karbi variety from West Karbi Anglong has an additional
63
vowel.40 This sixth vowel is a high to mid-high, front, centralized vowel /ɪ/ (see Figure
10).
i u
ɪ
e o
A full study of this vowel still needs to be conducted to examine any co-
occurrence or positional restrictions in detail. It seems clear, however, that wherever /ɪ/
occurs in Amri Karbi, the standard Karbi varieties have the high front vowel /i/. A few
items where /ɪ/ occurs are given in Table 21. Note that the standard Karbi variety has the
same items except for using /i/ instead of /ɪ/. The minimal pairs that contrast /i/ and /ɪ/ in
Amri Karbi are therefore homophonous in the standard Karbi varieties.
40
Note that there may be a difference between the Christian and non-Christian sociolects of the Amri Karbi
variety. While the non-Christian sociolect definitely has the sixth vowel, it appears that the Christian
sociolect does not. This would also explain why Grüßner (1978), who had worked with speakers of the
Christian sociolect of the Amri Karbi variety, does not report this vowel.
41
t is a lexicalized item, where t ‘to die’ is clearly recognizable, but n does not appear to
synchronically be a suffix.
64
The ten examples given in Table 21 have /ɪ/ in different positions in the word and
with different tones. The existence of minimal pairs between /i/ and /ɪ/ suggests that /ɪ/
historically was a phoneme of Pre-Karbi that was later on merged in the standard dialects,
while the more conservative Amri Karbi variety has preserved the /ɪ/.
Basic five vowel systems with a less stable sixth vowel are also found in the
Boro-Garo branch. Burling (2013) gives an interesting comparative description of the
‘sixth’ vowel in these languages, which in Boro-Garo, however, are all back and not front
like in Karbi (though also high to mid-high). In the modern Boro-Garo languages, the
basic five vowels have not changed and can be easily reconstructed, while the ‘sixth’
vowel is different in the different languages, and has merged with either /u/ or /o/ in one
of them, Dimasa.
There are three marginal monophthong vowels that occur in a very limited set of
morphemes. These are the open [ɛ], the open [ɔ], and the [oʷ] off-glide version of /o/.
Table 22 shows all three with examples. Note that all of the examples in Table 22 are
phrase-final or phrase-‘independent’ markers with highly pragmatic functions.42 It seems
that the open [ɛ] and the open [ɔ] are better analyzed as part of prosody (perhaps with the
function of marking the end or boundary of a phrase).43 This will need to be addressed in
future research.
The off-glide [oʷ] is not so much associated with morphemes that occur in regular
discourse. Seemingly the most frequent morpheme where this off-glide occurs is ho
‘here.you.go’, which is an interjection that accompanies an act of offering an object,
usually when the intended recipient of that object has their attention focused elsewhere.
This ho ‘here.you.go’ and a separate discourse particle ho form a very clear minimal pair
42
Their highly pragmatic function in phrase-final position is also the reason why these are unmarked for
tone. It has been impossible to determine the tone given that the pitch on these markers is heavily
influenced by prosodic patterns (§3.5.9).
43
Another phenomenon I have observed is that in Y/N-questions without the question particle ma at the
end, if the verb has the suffix - ‘RL’ (for example in common questions like - t- ? ‘have you
eaten?’), the - is typically produced with a very open /ɔ/ as compared to the parallel answer or statement
- t- ‘(I) have eaten.’
65
for /oʷ/ versus /ɔ/. Otherwise, /oʷ/ also occurs in the expression used to call dogs when
they are fed, in dododo (multiple repetitions of do with /oʷ/, not just exactly three).
Finally, the off-glide [oʷ] also occurs as a variant of [o], seemingly for stylistic reasons,
whose exact effects are unclear. For example, I have heard people sometimes say avelo
‘not exist anymore’ articulated more as [aβeloʷ] than [aβelo].
For a discussion of the functions of the interjections and hesitation words listed in
Table 22, see Chapter XII.
Besides these marginal monophthong vowels, there also is one lexical item
borrowed from Assamese, which has a unique two vowel sequence: ‘forgive(<Asm)’.
There is no glottal stop inbetween the two vowels, but instead a glide transition. Some
speakers produce a more nativized version of this verb root by inserting a palatal voiced
stop (which anyway alternates with the glide), resulting in .
3.3. Syllable
Table 23 gives an overview of all types of open syllables, i.e. those that lack a
coda consonant, and provides a sample monosyllabic root for each. Table 24 does the
same for closed syllables, i.e. those that do have a coda consonant (see Table 16 for an
overview of possible coda consonants).
66
Table 23. Open syllable types – diphthong analysis
Description
Schematic Example Gloss
Onset Nucleus
None Monophthong V ‘to sleep’
Diphthong VV ‘to pick up sth. spread out (like a cloth
that was spread out to dry in the sun)’
Simple Monophthong CV ‘to send’
Diphthong CVV t ‘to wrap by rolling’
Cluster Monophthong CCV ‘to fall’
Diphthong CCVV kr ‘to agree’
44
This can be explained as being (at least partly) due to a diachronic development whereby rhymes with a
historical coda /l/ turned into the modern day diphthongs, see footnote 49 in the next section §3.4.
45
There appears to be some amount of dialectal variation such that /phr/ (as in the western Amri,
Rongkhang, and Chinthong dialects) may be produced without aspiration as [pr] in the eastern dialects. For
example, the word samph ‘sun (poetic); NAME’ may be pronounced as samp , ‘front’ may be
pronounced as , and ‘have sweet smell’ may be pronounced as .
67
Table 25. Onset cluster types
Stop Liquid Cluster Example Gloss
/l/ /pl/ ‘to become’
/p/
/r/ /pr/ ‘to be fast’
/l/ /phl/ ‘to burn’
/ph/
/r/ /phr/ phr ‘front’
/th/ /r/ [/thr/] t ‘six’
/l/ /kl/ ‘to do’
/k/
/r/ /kr/ ‘CLAN’
/kh/ /r/ [/khr/] ng ‘tree sp.’46
As for alveolar stops, only the aspirated stop occurs with the rhotic, and only in
two words, and really just in one morpheme: t ‘six’ and t ‘seven’ (which is
derived from t ‘six’ plus ‘one’). I have heard Karbi speakers47 say that t ‘six’
comes from a longer form therok without the onset cluster, and this longer form is also
recorded by Matisoff (2003: 145). That means then, that the alveolar stops essentially do
not participate in onset clusters with the lateral and the rhotic, which also makes sense
from an articulatory point of view. It presumably is also because of this reason that in
t ‘six’ and t ‘seven’, the rhotic is produced as an approximant, which it is not
otherwise in onset position. The velar voiceless unaspirated stop /k/ productively occurs
in clusters with both the lateral and the rhotic, while the voiceless aspirated stop /kh/ only
occurs with the rhotic and only in a limited number of morphemes, mostly in suffixes.
The example given in Table 25, ‘tree sp.’ might be the only actual root that has
this onset cluster.
Finally, it should be noted that the syllable is a very salient unit in Karbi. It
appears that the majority of roots and certainly the great majority of affixes are
monosyllabic. To my knowledge, resyllabification does not occur across morpheme
boundaries.
46
Grüßner (1978: 13) mentions that the botanical name for is “Amora Rehituka”.
47
I particularly noted that Mr. Sikari Tisso mentioned this longer form therok. Mr. Tisso is originally from
Boksong in West Karbi Anglong and identifies himself as an Amri dialect speaker, specifically of the non-
Christian sociolect.
68
3.4. Palatal Glide Coda versus Diphthongs
For syllables such as bai, bei, boi, or bui, it is not obvious which analysis is
preferable: whether they should be treated as CVC syllables where the coda is a palatal
glide /j/, or whether they should be treated as CVV syllables with a diphthong nucleus.
Both analyses introduce similarly small amounts of complexity to the description of the
phonological system.
The palatal glide analysis requires us to pose a new coda phoneme, i.e. the palatal
glide. Since there are no other palatals or glides in coda position, this is a disadvantage to
this analysis. Furthermore, the same palatal glide is in a sense an entirely new phoneme,
since in syllable onset position, we only have /ɟ~j/. This allophonic alternation is only
(PTB) (see footnote 37 in §3.1.1). But even if the stop /ɟ/ is a ‘new’ development in Karbi,
it is still a synchronic fact that there is allophonic variation between /ɟ~j/ in onset position,
but not in coda position. Although Matisoff (2003) reconstructs what he calls ‘palatal
diphthongs’ for PTB, i.e. the type of rhyme under consideration, it does not appear that
the Karbi rhymes are reflexes of those PTB palatal diphthongs.48 Instead there are a
48
I only found one item where PTB *-ay may be reflected by Karbi -ai, which is i from PTB *dzay
‘cattle’ (Matisoff 2003). In my own corpus I only have ‘cow’, which is clearly a compound with
the second root coming from ‘to cultivate, loosen soil’. In a dictionary manuscript that Grüßner was
working on in the 1970s, he also lists i as an individual root meaning ‘mithun, Gaxaeus Ganrus’.
69
number of instances in which Karbi /ei/, /ai/, /oi/, and /ui/ reconstruct back to a
monophthong plus coda */-l/ in PTB.49 This reconstruction also explains the dialectal
difference, where Plains Karbi has the forms phirul ‘snake’, mol ‘back’, or ingkol
‘twenty’.
Following the current orthographic standard, diphthongs /ei/, /ai/, /oi/, and /ui/
rather than palatal glide codas are written in this dissertation.
3.5. Tone
Karbi has three tones, which I will refer to as low, mid and high, following
Grüßner (1978). However, while low and high would seem to be fitting labels, this is not
necessarily the case for the mid tone, which is the odd one out of the three tone categories.
When contrasting monosyllabic roots, for example, in a minimal triplet of tone, the
auditory characteristics of the three tones are as follows.
The low tone seems to be produced at about the normal pitch level of a given
speaker. It is usually realized with a bit of a falling contour.
The high tone is produced at a significantly higher pitch level, and sometimes has
a bit of a rising contour associated with it when followed by at least one other syllable
within the same word. When the high tone is on a word- or phrase-final syllable, for
example in the case of eliciting a monosyllabic stem as its own word, then the high tone
is produced with a steep falling contour as part of the additional layer of phrase-final
prosody.
Lastly, the mid tone sometimes has an intermediate pitch level between the low
and the high tone, but not always, as will be discussed below in more detail below. It is
different from the low and the high tone in that it has glottalization associated with it.
This glottalization surfaces most clearly in open and sonorant-final syllables when they
occur as the last (or only) syllable of a word.
A significant finding of this dissertation is that stopped or checked syllables, i.e.,
those ending in a stop /p/, /t/, or /k/, may be mid tone. The previous description of the
tone system by Grüßner (1978) had stopped syllables be exclusively low or high, while
49
Evidence includes Hills Karbi words ‘to mix, stir’ from PTB *ŋw ; ‘snake’ from *s-brul;
t ‘to wrap (something large)’ from *r-tul; he ‘hail’ from *ryal; ‘twenty’ from *m-kul; and,
t ‘arrow’ from *tal.
70
reserving the mid tone for non-stopped syllables only, i.e., open or sonorant-final
syllables.
The three tones can be contrasted in minimal sets as shown in §3.5.1 and §3.5.2.
These minimal sets were discovered both in an early stage of the research for this
grammar that was dedicated to phonological issues, as well as later on and throughout the
project when encountering new roots in the process of analyzing texts. In the early stages
of the research project, some time was spent going through the list of phonotactically
possible monosyllables and eliciting which possible syllables were indeed morphemes of
Karbi and which were not. That way, a few minimal pairs were discovered. However,
identifying the tone even just of monosyllabic roots, but even more so of disyllabic or
polysyllabic roots or of suffixes or clitics has remained a challenge throughout the course
of research for this grammar. It is my goal for this section to both justify why what we
have in Karbi should indeed be analyzed as a tone system, and to then lay out why it has
been so challenging to describe this tone system, and to apply the tonal analysis to the
task of accurately representing spoken Karbi in the transcription of texts recorded for the
main corpus of this grammar. Note that this grammar only indicates tone at the
morpheme but not the word level, and only on roots and suffixes, but not on prefixes, the
one proclitic, or enclitics (see §3.5.6 and §3.5.9).
In §3.5.3, I describe how the minimal sets and at least two other kinds of evidence
form the basis of the claim that Karbi does indeed have tone. With this in mind, §3.5.4
lays out why this tone system is weak and carries a low functional load.
The next section §3.5.5 discusses two tone sandhi effects that occur across stems
and suffixes. In §3.5.6, the interactions between tone and different phonological and
morphological levels are discussed. This includes a discussion of the tone-bearing unit
and tone at the word level, which is not marked in this grammar. In §3.5.7, remarks on
some of the tone patterns and tone changes that occur in compounds are offered. A
discussion of practical strategies that can be and have been used to identify the tones of
individual morphemes is given in §3.5.8. This is followed by a section that describes the
conventions for marking tones that are followed in this grammar in §3.5.9. Finally,
§3.5.10 provides some remarks on a few tone minimal pairs across nouns and verbs that
71
may suggest a historically derivational function of tone in Karbi, which is, however,
synchronically not productive.
There are two tables below that offer tone minimal sets for monosyllabic roots
that have either a voiceless onset (Table 26), or a voiced onset (Table 27). Since the
voicing status of the onset is known to give rise to tonogenesis cross-linguistically, it is
important to note that all three tones occur after voiced and voiceless onsets in Karbi.
Both Table 26 and Table 27 also give minimal triplets across coda type. A
‘perfect’ minimal triplet that has all three minimally contrasted roots belong to the same
word class, so either all nominal or verbal (following criteria outlined in §4.1.1), could
not be found. This is telling of the low functional load of this tone system (see §3.5.4.2).
As mentioned above, this study has found stop coda syllables carrying the mid tone,
contrary to Grüßner (1978). However, stopped mid tone syllables are still a bit different
from non-stopped mid tone syllables: stopped mid tone syllables appear to participate in
phonological contrast a lot less frequently than non-stopped mid tone syllables. In fact,
when I first came across stopped mid tone syllables, I thought they were mid tone
because they did not participate in tone contrast. These were roots such as t ‘to dig (a
small hole)’, ‘to boil’, and - t ‘one-CLF:HUM:SG’, which do not have segmentally
identical counterparts with a low or a high tone. Further research has revealed, however,
that there are some minimal pair contrasts that mid tone stopped syllables participate in
(see below). Regarding a three-way tone contrast, I only have found one single fairly
good minimal triplet for stopped monosyllabic roots, the one given in Table 26: t, t,
and t. Even this minimal triplet is a bit problematic, because t may only ever be used
as a relator noun with a dependent noun or a subordinator with a dependent clause, in
both of which cases it occurs with the a- ‘possessive’ prefix. Non-stopped monosyllabic
roots, on the other hand, display the three-way tone contrast clearer with a number of true
minimal triplets (presumably around a dozen to perhaps a couple of dozen). Table 26 and
Table 27 show , , and , and , , and as sample minimal triplets
following a voiceless onset, and , , and , and , , and as sample
minimal triplets following a voiced onset.
72
Table 26. Minimal sets for tones after voiceless onsets
Tone Open syllable Sonorant coda Stop coda
L ‘grandmother’ ‘to spread’ t ‘to dig (a large hole)’
M ‘to roast’ ‘raw rice’ t ‘time’ ( t ‘during’)
H ‘to give birth’ ‘to take rest’ t ‘to question/examine a
wrongdoer’
There is one other, not even as good, minimal triplet for stopped syllables that I
have come across, which is given in Table 28. While the low versus mid tone opposition,
t ‘to know’ and t ‘to move wood in fire’ is solid, the high tone counterpart is a type
of exclamative expression that is usually accompanied with a gesture that indicates the
amount or size of a given entity.
Table 29 offers minimal pairs of stopped syllables displaying all three pairwise
contrasts: the mid tone versus the high tone, the low tone versus the high tone, and the
low tone versus the mid tone. While especially for the low versus high but also the low
versus mid oppositions of stopped syllables, there are a few other minimal pairs besides
the ones listed in Table 29, the two minimal pairs of the mid versus high opposition given
in this table are the only ones I have come across.
Note also that ‘hardship; be poor’ is a borrowing from Assamese. Borrowings
from the non-tonal surrounding languages that have been strong donor languages in the
history of contact with Karbi, i.e., Indic languages (Assamese, Hindi, Bengali), English,
and Khasi, typically receive the high tone (see also Grüßner (1978: 31-32)).
73
Table 29. Stopped syllable tone minimal pairs for M-H, L-H, and L-M
Minimal pair L tone M tone H tone
‘dust’ ‘hardship(<Asm)’,
M vs. H ‘be poor(<Asm)’
‘truth’ ‘to approve’
‘to stick’ ‘to help’
L vs. H
‘to beat’ ‘to be okay’
‘wound’ ‘eye’
L vs. M
‘to get water’ ‘paddy’
Table 30 offers two minimal triplets of disyllabic roots, where only the ultimate,
i.e., prominent, syllable (cf. §3.6) carries contrastive tone. These were the only tone
minimal triplets of disyllabic roots I was able to identify, although there might be
additional ones, especially with one of the frequent first syllables and frozen prefixes ing
or ar (cf. §4.7).50
50
The phonological study of disyllabic roots with initial ing and ar would have been impossible for me
without the thorough compilation of those forms by Grüßner in his dictionary manuscript.
51
See §3.5.9 for explanations of how tone is marked in morphemes with more than one syllable. Note that
in this dissertation, tone is only marked at the morpheme level but not at the word level (§3.5.6).
74
Table 31. Minimal pairs of disyllabic roots
Minimal L tone M tone H tone
pair
‘carrying ‘to carry a
ing
load’ load’
M vs. H
‘be
‘sp.grass’
intoxicated’
t ‘half’ t ‘mosquito’
L vs. H
‘sister’ ‘to dissolve’
‘cotton’ ‘potash’
‘yam’ ‘snake’
L vs. M
t ‘be ing t ‘water leech’
slippery’
There are at least three different pieces of evidence for the phonological basis of
the Karbi tone system that also serve to justify analyzing it as a tone system, rather than,
e.g., a pitch accent system. First and foremost, of course, we need to consider the
evidence of minimal pairs and triplets that contrast monosyllabic and also a couple of
disyllabic roots for tone, presented in the previous sections §3.5.1 and §3.5.2. The fact
that there is a large number of monosyllabic roots, and still a considerable number of
disyllabic roots that are segmentally identical and are only distinguished by pitch height
or a combination of pitch height and glottalization should lead us to believe that tone is as
phonological as are the consonant and vowel phonemes.
Second, there is a certain level of awareness of tones among native speakers. For
example, my consultant Sikari Tisso pointed out a small number of cases where in the
recordings that we collected for this project and then analyzed, the speaker in a particular
recording made a speech error that consisted in using the wrong tone.52 The fact that he
was able to tell me he heard the wrong tone speaks to the phonological basis of tones.
Furthermore, many different people have mentioned to me the tone minimal pairs of
‘yam’ versus ‘snake’, and ‘to see, look’ versus ‘water’, after I
say that I am studying Karbi. What is interesting, however, is that it tended to be exactly
52
Examples are the use of a high tone in *a- instead of low tone a- ‘POSS-grandmother’ in KK, BMS
093, and a low tone pronunciation of high tone n ‘now’ in -pu- n-t ‘up until now’ in SiH, CW 017.
75
these two minimal pairs that people point out to me, as if they were the conventionalized
prototypes of tone minimal pairs. Another example of the awareness of tones among
native speakers is the tongue twister given in (3), which also is something several
different people have pointed out to me, and Grüßner (1978: 26) also recorded it.
The idea that different phonemic contrasts may have different degrees of
functional load in the overall phonemic system goes back to the Prague School (inter alia
Mathesius 1929; Jakobson 1931). There have also been approaches to quantify the
functional load of phonemic contrasts (e.g. Hockett 1967; Surendran and Niyogi 2006).
The goal of this section is, however, to argue in a qualitative way for why tones in Karbi
only carry a low functional load.
76
I present six different types of evidence that the Karbi tone system does indeed
carry a low functional load within Karbi phonology. While each piece of evidence by
itself would not be conclusive, the various types of evidence together form the argument
for the low functional load of tones in Karbi.
The first type of evidence comes from native speakers’ difficulty in identifying
tone categories. The next two types of evidence come from what we can refer to as the
paradigmatic and syntagmatic dimensions of the functional load of tone. The
paradigmatic dimension consists in the contrastiveness of the tone system or the existence
of tone minimal sets. The syntagmatic dimension consists in the context of a tonal
morpheme, both within the same word and at the phrasal or clausal level, which may
greatly contribute to the identification of that morpheme. The fourth type of evidence
presented here is the occurrence of categorical tone changes and cases of
indeterminability of the tones of certain morphemes due to over-layering prosody. The
fifth type of evidence comes from a phonetic study that looked at both acoustic
measurements of tones and at the performance of native speakers as they tried to identify
lexical items in a perception study (Konnerth and Teo in press)
Lastly, evidence for the low functional load of the tone system also comes from
the fact that only roots and suffixes bear tone (see §3.5.6).
77
a suffix, which, I believe, is already suggestive of the low functional load of this tone
system.
78
remarkably frequent, such that they seem a) semantically close to the prototypical
meaning of the stem without the suffix, and b) to some degree lexicalized as a unit of
their own. As a result of that, native speakers have quite frequently used predicate
derivations to differentiate between members of minimal sets in the course of research on
tone for this project. Examples are t - ‘be.short-small’ and t - ‘snatch-quickly’
(§6.5.1.1.3). Thus the existence of predicate derivations with narrow semantics and
especially the cases of frequently collocating stem-suffix combinations are another factor
that reduces the functional load of tone in Karbi.
Another, quite curious piece of evidence that suggests that the context of a root
matters is that for recordings made for the phonetic analysis of tone, where stems were
recorded once in isolation followed by three times in a carrier phrase (“Neli ___ pusi
kepu.”), one particular speaker merged mid and high tone items produced in isolation to
the low tone such that the phonetic recordings of the items t ‘be short’ and t ‘snatch’
would be (and sound to other native speaker) as follows:
“ . Neli t pusi kepu. Neli t pusi kepu. Neli t pusi kepu.” (for t ‘be short’)
“ . Neli t pusi kepu. Neli t pusi kepu. Neli t pusi kepu.” (for t ‘snatch’)
Note that other native speakers thought this speaker was first saying the verb ‘die’
and then switched to ‘be short’ or ‘snatch’, respectively.
It is perhaps not clear whether this should count as evidence that the context
matters so much that an item without a context, i.e. if produced in isolation, does not
receive tonal specification by this speaker. However, it does represent more proof for the
low functional load of tone.
An acoustic study of Karbi tone has shown that there are differences between
speakers in whether they realize the mid versus high tone distinction in pitch (Konnerth
and Teo in press). This study examines two native Karbi speakers, one female and one
male, in their respective realizations of the three tones in the following three contexts: 1)
monosyllabic bare stems, 2) monosyllabic stems with the suffix - ‘IRR2’, and 3)
monosyllabic stems with the suffix - ‘IRR1’. Although both speakers originally come
from different areas of Karbi Anglong, they have both lived in Diphu for a long time.
79
Averaged F0 values of the three tones show that the male speaker consistently
differentiated the mid and the high tone through F0 in all three contexts, while the female
speaker merged the mid and the high tone. For both speakers, the low tone was
consistently lower than the mid or the high tone in a statistically significant way.
It is not quite clear what the reason behind this difference between the two
speakers in the realization of the mid versus the high tone is, but to evaluate the
phonological basis of the mid versus high tone distinction produced by the male speaker,
a follow-up perception study was conducted, discussed in the next section §3.5.4.5.
53
Listeners could listen to the stimuli as much as they wanted. The stimuli were the target item once in
isolation and three times in the carrier phrase Neli ___ pusi kepu. ‘I said ___’.
80
3.5.4.6. Prosodic Tone Changes and Indeterminability
There are at least two types of instances in the corpus of recorded texts where a
morpheme that we know to have a particular tone in a pragmatically unmarked context
occurs with a different tone in a particular pragmatic or prosodic context. This is different
from morphophonemic tone changes, discussed in §3.9.1, which are explained purely by
the morphological environment of a tonal morpheme without reference to prosody.
In (4), ‘how’ in kopuloma is originally low tone, as indicated in the
morpheme line. However, in the emphatic context in this example (i.e., with a sense of
‘how only’, as the speaker is desperate) is actually produced with a high tone, as
indicated in the word line.54
This prosodic tone change is even more striking considering that with the
high tone exists separately and means ‘where’.55
The suffix - t ‘PFV’ also consistently becomes high tone - t when followed
by -pen ‘NF:with’ as in (5), or when followed by a relator noun derived subordinator such
as ‘after’ in (6) (see also §11.2.1.2).
54
This tone change appears to not be restricted to this speaker or text. My language consultants recognize
the change to high tone as something natural in the given context.
55
The two items and are parallel to ‘like this’ and ‘this side, here’, although with the
demonstrative la, this prosodically-driven tone change does not happen to my knowledge.
56
The audio file for KK, BMS 067 is available under the DOI name 10.7264/N3TT4P7M, see Appendix B.
81
puta] kroikredetlo ansi
p t kr i-C -d t-l nsi
QUOT=ADD:although agree-NEG-PFV-RL and.then
‘he didn't let her go, he was embracing her tightly, although she said «leave me (alone)!»,
he didn't agree’ [KK, BMS 080]57
kedam kechenglo
ke-d m ke-ch ng-l
NMLZ-go NMLZ-begin-RL
‘after the car came, we started going from here to Hongkram’ [SH, CSM 008]58
57
The audio file for KK, BMS 080 is available under the DOI name 10.7264/N3Q23XJN, see Appendix B.
58
The audio file for SH, CSM 008 is available under the DOI name 10.7264/N39W0CR4, see Appendix B.
59
See HK, TR 023 and 092.
82
3.5.5. Tone Sandhi Effects
Two consistent tone sandhi effects have been found. The first occurs on the tone
of a stem that follows the prefix cho- ‘auto-benefactive/malefactive’ (‘AUTO.BEN/MAL’).
While both prefixes che- ‘reflexive/reciprocal’ (‘RR’) and cho- ‘AUTO.BEN/MAL’ change
the tone on monosyllabic stems such that a low tone turns into a mid tone and a mid tone
into a high tone (see §3.9.2.2), the cho- ‘AUTO.BEN/MAL’ prefix induces additional tone
sandhi (see also §3.9.2.2 for a discussion of the differences between che- and cho-). The
sandhi surfaces as a higher pitch level compared to either the che- ‘RR’ prefixed
counterpart or a ke- ‘NMLZ’ prefixed stem that matches the stem tone derived by che- and
cho-. Table 32 contrasts the same stems with cho- ‘auto-benefactive/malefactive’ and
che- ‘reflexive/reciprocal’. The tone sandhi is indicated by bold print. The sandhi effect
also persists on high tone stems, although no category tone change occurs. The example
in Table 32 is t ‘to tell’. Here, t with the che- ‘RR’ prefix and t with the
ke- ‘NMLZ’ prefix are identical in pitch contour, whereas t with the cho-
‘AUTO.BEN/MAL’ prefix displays the higher pitch sandhi.
The second consistent tone sandhi effect is found on low tone suffixes after mid
tone stems. Specifically, low tone suffixes following mid tone stems share characteristics
with mid tone suffixes. Table 33 gives examples of a low tone suffix, - ‘much:S/O’ and a
mid tone suffix, -t ‘again’, as they follow the mid tone stem ‘eat’. In the stem plus
suffix forms in the second column, i.e., - and -t , the low tone suffix -
‘much:S/O’ is still clearly different from -t ‘again’: the latter carries the salient word-
final glottalization associated with the mid tone, while the former does not.
The low tone suffix does, however, behave like a mid tone suffix in the third and
fourth columns, after - ‘RL’ or - ‘NEG’ are added. This is diagnosed with the help of
60
See §3.5.6 and §3.5.9 on how tone is marked in this dissertation (specifically, only on roots and suffixes
but not on prefixes, for reasons outlined in those sections).
83
these two suffixes, which can be used to identify the tone of a stem (see §3.5.8.2 and
§3.5.8.3), and particularly attaching - ‘RL’ makes the mid tone very salient.
Table 33. Tone sandhi effects on low tone suffix after mid tone stem
Suffix ‘to eat’ with suffix + - ‘RL’ +- ‘NEG’
- ‘much:S/O’ - - - - -
-t ‘again’ -t -t - -t -t
This sandhi effect, whereby a low tone suffix shares contour characteristics with
mid tone suffixes, only occurs after mid tone stems. After a low tone stem such as
‘go’ (or after a high tone stem), this sandhi effect does not occur, and - - and -
t - have clearly distinct pitch contours.
Due to the low functional load of the tone system argued for in §3.5.4, it has not
been possible to describe tone patterns at the word level. While certain word tone patterns
at a phonetic, non-contrastive level could be observed, they have continually proven to
not be strong, consistent, or categorical enough to allow for a phonological description.
Future research with a more narrow focus on these issues will hopefully shed more light
on them.
For the purposes of this grammar, tone is considered a property of syllables in
tonal morphemes, i.e., in roots and suffixes. Tonal morphemes can be regularly
contrasted with other tonal morphemes and their tone can thus be identified. The one
(verbal) proclitic slot, the half a dozen prefixes, and the roughly same number of enclitics
(see §3.8) – which are all monosyllabic – are considered toneless morphemes. Likewise,
non-final syllables of multisyllabic morphemes are also typically toneless. Most probably,
an analysis of tone at the word level and the difference between tone-bearing syllables
and toneless syllables requires a phonetic in-depth study of stress. In the current work,
stress is only discussed briefly in §3.6.
While the accuracy of tone assignment on monosyllabic tonal morphemes is near-
perfect, some difficulty is encountered in disyllabic and, even more so, in polysyllabic
tonal morphemes, when a syllable other than the last syllable may carry an inherent tone.
84
This is especially the case in borrowings and in compound nouns (for the latter see
§3.5.7). While borrowings generally carry the high tone (Grüßner 1978: 31-32), it is not
always clear whether in disyllabic borrowings, the first, unstressed syllable tends to also
be high tone or toneless. To my knowledge, most instances of disyllabic borrowings do
not have a tonally specified first syllable, such as t ‘date (<Asm)’, t
‘baked.snack (<Eng)’, or ‘vow (<KhJ)’. Grüßner, in his dictionary manuscript,
however, also lists t ‘knife (<Asm)’, t ‘doctor (<Eng)’, and t ‘stress, wear
and tear (<Khs)’, all with a high tone first syllable.
While the present work does not describe tone at the word-level, Grüßner (1978:
19) does offer a description of how tone is realized on toneless syllables. This description
suggests that toneless syllables alternate between low and high surface realizations (see
Table 34) which needs to be read from right to left. If the tone-inherent, stressed syllable
(right-most column in the table) is low, then the preceding toneless syllable (middle
column in the table) is realized as (‘realized as’ is indicated by the arrow in the table)
high. If the tone-inherent syllable is mid or high, then the preceding toneless syllable is
realized as low. If there is another toneless syllable (left-most column) preceding a first
toneless syllable, then this syllable is realized as low tone if followed by a high tone
syllable, and is realized as a high tone if followed by a low tone syllable. According to
Grüßner, sequences of toneless syllable follow this alternating pattern.
While auditory observations confirm Grüßner’s basic idea that the first toneless
syllable of a sesqui- or disyllabic word appears to build up a pitch contrast to the
following tonal syllable, the alternating pattern in multisyllabic words described by
Grüßner (1978: 19) has not been found. My own impression is that instead, the pitch
contrast builds up across the sequence of toneless syllables, followed by the fully realized
tone of the tone-inherent syllable.
85
3.5.7. Tone (Changes) in Compounds
There are certain tendencies for the tone patterns in disyllabic compounds. Table
35 gives an overview of those. The tone patterns low-mid (LM), mid-mid (MM), and
high-low (HL) are underlined, because they appear to be robust patterns, which occur on
many items. The patterns mid-low (ML) and mid-high (MH) are in [square brackets] to
indicate their marginal status: Only t as ML, and t and t as MH
were found in the corpus. The LL and LH patterns do not appear as marginal as ML and
MH, but also not as robust as LM, MM, and HL. Those compounds that have changed
one or both tones compared to the respective tone of each part that they appear to be
derived from are in bold.
The only two tone patterns that are not listed in Table 35 are high-mid (HM) and
high-high (HH). The HH pattern does appear to exist outside of borrowings (for which
there are a number of attested items, see §3.5.6) in compounds recorded by Grüßner
(1978: 36): t - ‘sp.mosquito’ (first part from t ‘mosquito’), or -
‘opening’ (first part from ‘hole’). Grüßner (1978: 36) also offers one item with
86
what he transcribes as a HM pattern: t - ‘wood chip’ (from t ‘splinter’ and
‘thorn’). However, the first part t might actually be mid tone,61 which then would
match the word t ‘bamboo strap’, which occurs in the corpus of this grammar. The
HM pattern might then not actually occur in compounds. Evidence in favor of this
hypothesis is that in Table 35, the first part of - ‘molasses’ is ‘sugarcane’,
and the first part of - ‘cloth to carry baby on back’ is ‘cloth’, so that HM patterns
changed into LM patterns. While the HM pattern perhaps does not occur in compounds, it
does, however, occur in words formed by a high tone stem with the diminutive suffix - ,
such as - ‘a small matter’ or - ‘a little song’. Among them, there is also
‘splinter’, which is (historically) composed of ‘to chip off’ and - but has arguably
undergone lexicalization.
It appears then that there are no clear restrictions on tone patterns in disyllabic
bimorphemic stems, but there are tendencies for more common patterns.
There are several strategies that have greatly facilitated the daunting task of
marking tone in the texts that represent the corpus for this grammar. In what follows, I
describe these strategies. The two strategies described in §3.5.8.2 and §3.5.8.3 use
suffixes to help identify root tones and are largely based on differences in the distribution
of perceived prominence among the target tonal morpheme and the added suffix.
3.5.8.1. Syllable under Investigation Is the Last Syllable of the Word and Non-Stopped
Adding - ‘realis’ (§6.9.1) to stems has proven the best way to perceive tone
categories of roots or derivational suffixes. Compared to determining the tone category of
a word-final tonal morpheme (§3.5.8.1), this strategy eliminates the problem of phrase-
final intonation, because the target syllable is not the last syllable of the word. This helps
distinguish low from high tone on the target syllable. However, in comparison to that
same strategy, we are also left without the salient glottalization associated with the mid
tone. It turns out, however, that even in the absence of glottalization, the mid tone leaves
a very salient auditory trace when - ‘realis’ is added. Specifically, a prominence shift
happens with mid tone stems, whereby mid tone stem plus - ‘RL’ has a perceived
prominence on the suffix, while with low or high tone stems plus - ‘RL’, the perceived
prominence remains on the stem. Phonetically underlying this perceived prominence shift
from mid tone stem to suffix appears to be a delay in the falling pitch contour of low tone
- ‘realis’. While after low and high tone stems, the falling pitch contour over - ‘realis’
is more or less linear, after mid tone stems it is delayed across the /l/ and only falls at the
beginning of the vowel /o/.
Since this strategy helps identify the mid tone without relying on glottalization, it
led to the discovery of the mid tone on stopped syllables, which due to their segmental
nature with unreleased coda stops are already perceived as glottalized.
An additional strategy that facilitates distinguishing between the mid and the high
tone (which often is a difficult task, see §3.5.4.4 and §3.5.4.5) consists in adding the mid
tone negative suffix - . Here again (as with - ‘realis’, see the previous section
§3.5.8.2), the emerging tone patterns between the three different target syllable tones and
the suffix tone are different in the distribution of perceived prominence. While the
88
emerging tone pattern of a low or mid tone target syllable and - ‘NEG’ has the - ‘NEG’
suffix more prominent, while the emerging tone pattern of a high tone target syllable and
- ‘NEG’ has the high tone target syllable more prominent. This strategy then helps
distinguish between a high tone versus a mid tone (or low tone) target syllable, while it
does not help distinguish between a low tone and a mid tone target syllable.
In the text examples in this grammar, tone is only marked on morphemes but not
on words. In particular, it is only marked on roots and suffixes, not on the items in the
verbal proclitic slot or on prefixes. On enclitics, it is generally not marked, although the
additive particle =t is realized with a clear and consistent glottalization that identifies it
as mid tone. The other enclitics are not marked for tone due to their susceptibility to
assimilate to pitch contours of prosody (§3.8.4).
If a tone changes due to morphophonemics (see §3.9.1), the new tone (and not the
underlying tone) is indicated. For example, the prefix che- ‘reflexive/reciprocal’ changes
the low tone of ‘go’ to a mid tone, and so the in-text example will indicate the tone
of the stem in - ‘RR-go’ as mid.
In disyllabic or sesquisyllabic morphemes (or prefix-stem combinations, for that
matter), the default realization of the preceding, unstressed, toneless syllable is to build
up a contrast to the tone of the tone-inherent, stressed, final syllable such that a toneless
syllable preceding a final low tone syllable is realized as high tone, and a toneless
syllable preceding a final mid or high tone syllable is realized as low tone (see §3.5.6). If
the unstressed syllables in a particular multisyllabic morpheme follow that default pattern
of realization, their tone is not marked. If they deviate from this pattern, then their tone is
marked. Table 36 gives examples of how tone is represented on the unstressed syllable of
disyllabic roots in this grammar.
The first three columns show the default patterns H-L, L-M, and L-H, for which
the unstressed first syllable remains unmarked. One example of a disyllabic stem and one
example of a prefix plus monosyllabic stem each are given. The second three columns
show the non-default patterns L-L, M-L, M-M, M-H, and H-H, for which the unstressed
first syllable is marked for tone. An example for each is given; note, however, that the
89
examples of M-L and M-H are the only ones there are in the corpus (see also §3.5.7).
Furthermore, the H-M example ‘splinter’ is (historically) composed of ‘to
chip off’ and the diminutive suffix - but has arguably lexicalized so that it is listed here
as a disyllabic stem.
Note that there are some individual roots and suffixes that are not marked for tone
in this grammar. These represent instances, where I have not yet been able to confirm the
correct tones.
While tone in Karbi is synchronically only lexical (if that), Grüßner (1978: 47; 53)
has compiled a list of corresponding noun and verb stems that form tone minimal pairs. I
repeat some of his examples along with my own examples.
It can be noted that nouns tend to take the mid tone, and may correspond to (and
perhaps be derived from) low or high tone verbs, although there is also one example each
of a high tone noun corresponding to a low tone verb and of a low tone noun
corresponding to a high tone verb. In addition, there are also homophonous verb-noun
pairs such as ’ ‘cheek; to carry in mouth’.
90
Table 37. Verb-Noun tone minimal pairs (G = Grüßner 1978)
Tone pattern
Verb Noun
(Verb-Noun)
‘to sieve’ (G) ‘strainer’ (G)
t ‘to cut’ (G) t ‘piece’ (G)
L-M ‘to compile’ ‘pile’
‘to filter’ ‘filter’
‘to step’ ‘step (classifier)’
‘to chop off’ (G) ‘chip, piece’ (G)
‘be wide’ (G) ‘width’ (G)
H-M
‘to plait’ ‘bamboo container’
‘to carry a load’ ‘carrying load’
L-H ‘to smell’ (G) ‘smell’ (G)
H-L ‘to sing’ (G) ‘song’ (G)
There is another pattern among tone minimal pairs that Grüßner (1978: 91)
mentions. This is between verbs that Grüßner suggests are semantically related. However,
the semantic relationships between the seven minimal pairs he lists (copied into Table 38)
are not always obvious. There is one potentially compelling example, which is c
‘wash’ and ‘be wet’. In absence of further minimal pairs with an action-result type
of relationship, however, this may just be a coincidence.
3.6. Stress
Stress arguably is not part of Karbi phonology proper, because it never creates a
lexical contrast between two morphemes, and a detailed phonetic study of stress is
outside the scope of this grammar. That said, however, a few remarks on stress are in
91
order. Stress does play a major role in the surface realizations of word-level tone, and the
study of stress would also complement the (diachronic and synchronic) understanding of
word and morpheme structure.
There is an interaction between stress and tone. Grüßner (1978: 23; my translation)
remarks that “in syllables with main stress and medium stress, tones are pronounced
clearly. In syllables with weak stress, tones become indistinguishable and approximately
converge towards Tone 1 [i.e., the low tone].” He continues to link the strength of stress
to morpheme type: “the main stress usually resides on the stem syllable, [i.e.] the syllable
with the semantically most important function. It [i.e., that syllable] is usually the syllable
of a free morpheme. If this [morpheme] is multisyllabic, then the main stress resides on
the last syllable.”
My own observations match Grüßner’s. There exists an iambic stress pattern in
Karbi multisyllabic morphemes such that the sequence of syllables is unstressed-stressed.
There are a large number of disyllabic roots that almost exceptionlessly follow this
iambic pattern, and the same stress pattern exists on combinations of prefixes with
monosyllabic stems such as - ‘POSS-language’ or - ‘NMLZ-eat’. In fact, this
iambic pattern is typical in many branches of Tibeto-Burman (though note that Tibetan
and Kiranti, for example, are trochaic), and exists both across the modern languages and
has diachronically shaped cognates all across the family. Matisoff (2003) in his
Handbook of Proto-Tibeto-Burman remarks on this as follows.
“Compounding has been a pervasive morphological process for at least the past
two millennia of the history of the ST [Sino-Tibetan] family, as part of the
languages’ response to the ever-present danger of homophony among their
monosyllabic morphemes. […] The unstressed vowel of the first syllable in such a
compound is typically schwa; the tone loses its original contour and becomes
“neutral”; if there is a final consonant it tends to drop; and eventually its semantic
identity is likely to become obscured. This is the process of “prefixization”,
whereby a fully meaningful morpheme is reduced to a prefix, in such a way that
the original disyllable becomes a sesquisyllabic unit.” (Matisoff 2003: 153-154)
While there is a tendency in Karbi for unstressed syllables (and especially pre-
stem ones and non-ultimate ones inside the stem) to be toneless (see §3.5.6), future
research needs to investigate this in detail. In particular, the unstressed (because more
92
inflection-like, see §6.2.1) modal suffixes - ‘realis’, - ‘IRR1’, and - ‘IRR2’ are
certainly tonal: - and - are low tone, while - is high tone.
The derivational suffixes, on the other hand, are part of the stem, and as such, a
derivational suffix added to a simple stem may receive the main stress. Take, for example,
- ‘reach-NEG’, where the stress is on the suffix and indicated by underlining (though
see §3.5.8.3 for a description of the prominence shift that happens with high tone simple
stems), or - ‘go-CONT’, where it is also the suffix that is more prominent. This
can be contrasted with =t ‘1EXCL=ADD’, where the first person exclusive pronoun is
stressed and not the additive enclitic.
There is a single exception to the iambic stress pattern that I have come across,
which is the distal demonstrative . It has a very prominent first syllable, although the
vowel is actually quite short as if the onset /l/ of the second syllable also closed the first
syllable. The second syllable is unstressed and reduced.
The change described by the rule [/t/ Ø / ___ v, l] is not specific to particular
morphemes but happens generally. Examples of this hypoarticulation pattern are
t= ‘where FOC’, which ends up pronounced as or, showing the same
pattern twice: - t- t- ‘one-CLF:HUM:SG-only-RL’, which ends up pronounced as
enuvelo.
3.8.1. Roots
Roots carry tone, and probably the majority of roots are monosyllabic. There are,
however, also a large number of disyllabic roots and sesquisyllabic roots (i.e., with a
reduced first syllable), while very few roots have more than two syllables. There are two
very prominent, synchronically non-morphemic first syllables that occur in many of the
disyllabic roots: /ing/ and /ar/, which are discussed in §4.7.
3.8.2. Suffixes
Suffixes are minimally syllabic, and mostly monosyllabic, although there are a
few disyllabic ones. Besides roots, suffixes are the only morpheme type that carries tone.
94
However, the tones of subordinating suffixes that have clausal scope and therefore are
highly susceptible to prosodic pitch contours have been virtually impossible to determine,
such as -te ‘conditional’ and -si ‘non-final:realis’, and are therefore not marked for tone.
3.8.3. Prefixes
Prefixes are toneless, and there are only a small number of them, which are
exhaustively listed in Table 39. With the exception of personal possessive prefixes, the
category of prefixes in Karbi is characterized by a highly reduced phonological shape.
When added to monosyllabic stems, they form sesquisyllables (Matisoff 2003: 153 ff.)
such as - ‘POSS-language’ or - ‘NMLZ-eat’ (see also §3.6).
Only the second and third person possessive prefixes have a coda consonant, and
only the personal possessive prefixes as a group can be extended with the honorific -
(§12.4.1).
3.8.4. Clitics
All attested clitics are listed in Table 40. They are generally unmarked for tone,
since they occur at phrasal boundaries, where prosodic pitch contours are so prominent
that the underlying tone labels such as ‘low tone’ or ‘high tone’ do not appear to apply.
There are two exceptions: =t ‘additive’, which consistently occurs with mid tone
95
glottalization and is therefore marked as such, and = ‘this much’, which consistently
occurs with high tone.
Clitics are generally reduced monosyllables without a coda consonant. The
exception is nang= and the longer pronominal forms with the - ‘honorific’ in the
proclitic category, as well as =pen ‘with; from’ and ‘non-final’ among the enclitics, as
well as disyllabic discourse markers that appear synchronically lexicalized, i.e., =mati,
=kema, and =t , the last of which appears to be historically composed of =t ‘additive’
and ‘be good’, so literally ‘also good’.
3.8.5. Interjections
Table 41 lists some of the examples of reduplication found in the corpus. Note
that the tone patterns recorded here should be considered preliminary and require further
research.
62
The reduplicated form is apparently derived from ‘be straight’ (ultimately from
‘foot’).
97
As we can see, reduplication without vowel change occurs with all vowels,
including monophthongs and diphthongs. While it is typically only the last syllable of the
stem that is reduplicated (which is the suffix in the cases of t - and - - ,
we also find disyllabic reduplication, as in .
98
The sample forms in Table 43 show that the rhyme of the suffix is invariably / /,
while the onset of the suffix repeats the simple onset of onset cluster (as in kr - ) of
the last syllable of the stem. If the last syllable of the stem does not have an onset
consonant, i.e., if it is vowel-initial, in which case the syllable starts with a glottal stop,
then the suffix repeats that (as in ar’ - , or, without the morpheme boundary: ar’ ).
Example (7) gives an in-text example of the verbal negation of the complex stem
’ t that has one prefix and two suffixes. It shows that only the last syllable / t/
is relevant to determine the form of the negative suffix, which consequently is simply - .
Note that the /k/ coda of the preceding syllable does not resyllabify to become the
onset of the following syllable / t/.
3.9. Morphophonemics
99
3.9.1. Tone Changes
There are several robust patterns of tone change, whereby the tone of a stem or a
suffix changes from one category to another. These tone category changes represent
evidence for the phonological basis of the Karbi tone system.
3.9.1.1. Stem Tone Change Induced by che- ‘reflexive/reciprocal’ and cho- ‘auto-
benefactive/malefactive’
Grüßner's (1978: 37) excellent work on Karbi phonology was the first to note that
the che- ‘reflexive/reciprocal’ and cho- ‘auto-benefactive/malefactive’ prefixes change
the tone of immediately following monosyllabic stems according to the pattern shown in
Table 44: low tone stems become mid, mid tone stems become high, and high tone stems
do not change (though see §3.5.5 for additional phonetic sandhi effects that occur as a
result of prefixing cho- ‘auto-benefactive/malefactive’).
Table 44. Monosyllabic stem tone changes after che- ‘RR’ and cho- ‘AUTO.BEN/MAL’
Underlying New
Sample With
stem stem With che- ‘RR’
stem cho- ‘AUTO.BEN/MAL’
tone tone
‘find (for
LOW MID r che-r ‘find (one’s own)’ cho-r
oneself)’
k p/ ‘cover (one’s own, ‘dig (for
MID HIGH che-k p cho-t k
t k oneself)’ oneself)’
‘teach (one’s own ‘teach (for a
HIGH HIGH th n che-th n cho-th n
children)’ living)’
100
such as - from ‘take’, subsequently turned into a low one stem if followed by the
negative suffix: * - - ; instead, the stem remained high tone, - - .
As first pointed out by Grüßner (1978: 21; 39), disyllabic nominal stems may
change their tones as well when a prefix is added. Grüßner described this change
primarily as a stress shift and the tone change as epiphenomenal, which is certainly an
interesting idea worth investigating further. The reliable pattern is that mid-mid disyllabic
stems change to high-low after a- ‘possessive’ or ne- ‘1EXCL:POSS’ (or presumably a
different personal possessive prefix, although the other prefixes were not specifically
checked) is attached. This is illustrated in Table 45.
The mid-mid to high-low tone change is almost exceptionless,63 but there also
seems to be a pattern for low-high stems to change to high-low as well, such as t and
t in Table 45. At this point, it is not clear how productive this pattern is and whether
it might be limited to borrowings. Note that no change was found to occur in LL stems
such as ‘kind of shovel’ and - , LM stems such as ko ‘sister-in-law’
and - or t ‘matter’ and - t , HL stems such as t ‘same gender
siblings’ and - t , but also other LH stems such as t ‘literature’, which remains
63
The only exception I have encountered is in compounds with the clan name (although the clan name
y, pronounced as /b /, is regular). Here, if ‘house, family’ is added, then - does not change
to * - - , but remains - - m. With the clan name y, the pattern is regular and y-
changes into - y- . The same is true for the difference between - ‘CLAN-female’, which remains
mid-tone - - , while y- regularly turns into - y- , curiously enough.
101
- t and does not undergo the tone change compared to other LH stems given in
Table 45.
There are a number of derivational suffixes that have low and mid tone
allomorphs, where the mid tone allomorph occurs after low or mid tone stems, while the
low tone allomorph occurs after high tone stems. Table 46 shows this pattern with the
sample suffix - - ‘JOIN’ following the low tone stem t ‘answer’, the mid tone
stem ‘give’, and the high tone stem ‘get’.
Table 47 offers an exhaustive list of all derivational suffixes that have so far been
found to participate in the tone change shown in Table 46.64 It also lists the lexical source
of the suffix where synchronically found in the language. It may not be a coincidence that
all of the lexical source verbs are low tone, i.e., that the original low tone leads to this
allomorphy.
64
Grüßner (1978: 37) mentions some of these as well.
102
In addition to the derivational suffixes in Table 47, the bound forms -t
~ -t of the numeral t ‘three’, which occur suffixed to classifiers, undergo the
same tone allomorphy. Examples of low, mid, and high tone classifiers used with -t
~ -t are: -t ‘CLF:long.cylindrical-three’, -t ‘CLF:animal-three’, and
-t ‘CLF:times-three’.65
There also are cases of more idiosyncratic tone changes. There is one example in
the corpus, where pV- ‘causative’ changes mid tone ‘be good’ into a high tone, ,
see (8). Since pV- ‘causative’ occurs with a very low frequency, it is not clear whether
this is a robust pattern or not.
65
Grüßner (1978: 65) also mentions the alternation in -t -t , and claims that mid tone classifiers
become low tone in this construction, e.g. -t were to become -t . I don’t have any evidence,
however, that this particular construction results in any idiosyncratic tone change of that kind; I rather
assume Grüßner here describes a general pattern of tonal unspecificity on unstressed syllables.
103
Table 48. Idiosyncratic tone allomorphy of - ~- ‘only’
Only - t acceptable Only - t acceptable Both - t and - t acceptable
‘nine’ ‘one’ ‘two’
t ‘three’ ‘four’
‘ten’ ‘five’
‘one hundred’ t ‘six’
t ‘seven’
‘eight’
t ‘three hundred’
3.9.2.1. Prefix Vowels: ke- ‘NMLZ’, pV- ‘CAUS’, che- ‘RR’, cho- ‘AUTO.BEN/MAL’, a-
‘POSS’
Grüßner (1978: 93) describes a pattern of allomorphy between ke- and ka- of the
nominalizer as well as pe- and pa- of the causative, whereby ke- and pe- occur before
monosyllabic stems, while ka- and pa- occur otherwise; “suffixes do not count in
determining monosyllabicity, prefixes, however, make the verb multisyllabic” (93). This
104
regular pattern is not found in the corpus of this grammar. Instead a much messier pattern
emerges as shown in Table 49 and Table 50.
As shown in Table 49, monosyllabic stems always take ke-, and there appears to
only be one exception: - ‘NMLZ-be.different’. As pointed out by UV Jose (personal
communication), however, is likely to be an early borrowing from Assamese belek
(which has now been reborrowed as belek). If that is indeed the history of Karbi ,
then the ka- is a neat piece of evidence because it tells the disyllabic origin of .
In every context other than pure monosyllabic stems, there seems to be free
allomorphic variation between ke- and ka-. Although some tendencies may turn out to
align with dialectal areas, there is some amount of variation within the same speaker,
which shows that it is not just a matter of different dialects.
Table 50 gives examples of the two allomorphs of pV- ‘causative’. Unlike the
allomorphy of ke- ‘nominalizer’, pe- appears to only ever occur before monosyllabic
stems, although pa- is used in that context as well, seemingly especially in the non-
Christian variety of the Hills Amri dialect in western Karbi Anglong.
The reflexive/reciprocal prefix che- alternates with ch-, which only occurs before
disyllabic stems that begin with /ar/ or /ing/, as shown in Table 51. Although the corpus
of this grammar always has ch- before disyllabic stems that start in /ar/, discussions
within our research team suggest that the resulting first syllable can often be pronounced
either as /char/ or as /cher/ (as in t in Table 51, but also, e.g., ~ ‘to
105
learn, study’). Forms with /cher/ are also recorded by Grüßner in his dictionary
manuscript.
106
Table 52. Initial vowel deletion in ing- stems after prefixes ke-, pV-, cho-
Prefix Stem Vowel deletion
ke- -
pV- ‘to sit’ -
cho- -
a- t ‘type of bamboo basket’ - t
Examples of the vowel deletion that occurs between prefixes ke- ‘nominalizer’,
pV- ‘causative’, and, less robustly, che- ‘reflexive/reciprocal’ and a- ‘possessive/modified’
and disyllabic stems that start in ar- is given in Table 53 (see discussion of Table 51
above for the case of che- ‘RR’). Note that in the case of the a- prefix, there is typically no
morphophonemic change if the ar- disyllabic root is a lexical noun (e.g., ‘god’). If
instead the ar- disyllabic root is, however, a noun that almost always occurs with the a-
prefix (like relator noun , or presumably any other noun that requires a possessive
prefix, see §4.4.5), one of the two /a/ vowels typically is deleted.66
Table 53. Vowel deletion between ar- stems and prefixes ke-, pV-, che-, a-
Prefix Stem Vowel deletion
ke- ‘to ask’ ka
pV- t ‘to spin’ t
[che- t - t - t]
a- ‘god’ [ ’ ]
‘inside (relator noun)’
Lastly, one question regarding prefix vowel allomorphy concerns the difference
between ke- and pV- as they occur before monosyllabic stems with either a high or a non-
high vowel. Due to the perception of vowel harmony that assimilates ke- to ki- and pe- to
pi- before high vowels, the Karbi Lammet Amei (see §1.1.4) has issued the spelling rule
to write ki- and pi- (or pa- depending on the dialect, see discussion above) in those
instances. However, the first person exclusive possessive prefix ne- is not perceived to
66
More research is required to understand under exactly which circumstances the a- prefix is not used or
pronounced on nouns that begin with ar-, or, rather, under which circumstances it actually is used, as there
are only two instances in the corpus, where that is the case (a- m ‘POSS-god’ in HK, TR 111 and a-
ng ‘POSS-person’ in SH, CSM 039).
107
change to ni- before high vowels. The research question then is whether ke- and ne-
behave the same before monosyllabic stems with high vowels or not.
A brief phonetic study to evaluate this research question was conducted. The
stimuli listed in Table 54 were recorded from one female speaker and one male speaker,
both residing in Diphu.67 Four tokens of each type were recorded: once in isolation, and
three times in the carrier phrase Neli ___ pusi kipu. ‘I said ___.’ All four tokens were
then measured for F1 and F2 formant values of the vowel of the first syllable, i.e. either
the vowel of the prefix ke- ‘nominalizer’, or the vowel of the possessive prefix ne-
‘1EXCL:POSS’, and in one case, the vowel in the first syllable of ‘mother-in-law’ for
reference of a lexical reduced /i/. The results of these measurements are given in Table 54.
Table 54. Vowel formant values of ke- ‘NMLZ’ and ne- ‘1EXCL.POSS’ preceding
high and non-high vowels
Female Male
Following
Speaker Speaker
stem Prefix Item Gloss
Avg. in Hz Avg. in Hz
vowel
F1 F2 F1 F2
ke- ki ‘be difficult’ 386 2,519 395 1,916
‘NMLZ’ ki ‘to give’ 350 2,727 337 2,036
High ki ‘to measure’ 391 2,691 354 1,938
vowel
/i/ or /u/ ne- ne ‘my biscuit’ 400 2,650 364 1,965
‘1EXCL.POSS’ ne ‘my mother’ 400 2,628 386 1,964
ne ‘my song’ 409 2,582 403 1,927
67
These are the same two speakers that were consulted in the phonetic tone study mentioned in §3.5.4.4
and §3.5.4.5.
108
In order to visualize these results, Figure 11 shows the formant values of the
female speaker’s stimuli, and Figure 12 of the male speaker, respectively, in the spatial
arrangement typical of the discussion of vowels, i.e. as in Figure 9 above.
F2 Female Speaker
2,800 2,700 2,600 2,500 2,400 2,300
300
320
kV- before high vowels
340
360 ne- before high vowels
380 kV- before non-high vowels
400 F1 ne- before non-high vowels
420 /ni/ in /nipi/
440
460
480
500
Figure 11. Female speaker vowel formant values for ke- and ne-
before high and non-high vowels
F2 Male Speaker
2,100 2,000 1,900 1,800 1,700
280
300
kV- before high vowels
320
340 ne- before high vowels
360 kV- before non-high vowels
380 F1 ne- before non-high vowels
400 /ni/ in /nipi/
420
440
460
480
Figure 12. Male speaker vowel formant values for ke- and ne-
before high and non-high vowels
109
The most /i/-like vowel is therefore in the left upper corner, while more
centralized productions are further right and further down. The blue symbols are the data
points for ke- and the red symbols are the data points for ne-, while the diamonds are the
prefixes occurring before high vowels and the triangles are the prefixes occurring before
non-high vowels. The green square is for the vowel in the first syllable of ‘mother-
in-law’.
Keeping in mind that this study only looks at a highly limited amount of data, i.e.,
every data point is the average across just four tokens, the emerging picture is somewhat
different between the female and the male speaker. The formant values of the female
speaker’s stimuli suggest that ke- ‘nominalizer’ and ne- ‘first person exclusive possessive’
before high vowels group together, and the two prefixes before non-high vowels group
together (i.e., all diamond symbols group together and all triangle symbols group
together). At the same time, there is quite a bit of variability among blue diamonds, i.e.
ke- before high vowels, which really means more data are required. At the same time,
because of the grouping just mentioned, the ke- and the ne- vowels appear to largely
participate in the same vowel harmony, although the vowel harmony for ke- appears a bit
stronger.
The male speaker’s data, on the other hand, are a lot messier with regard to ke-
and ne- before non-high vowels, i.e., the triangle symbols. Before high vowels, ke- and
ne- also behave quite similarly, and, again, more data are needed to see whether the
difference bears out statistically.
I would argue that the fact that for both speakers, the data points for ke- before
high vowels, i.e. the blue diamonds, are spread apart quite a bit, whereas the data points
of ne- before high vowels, i.e. the red diamonds, are clustering quite closely, suggests
that there really is difference between ke- and ne-. This difference may of course be
expected due to the much higher frequency of ke- compared to ne-. An interesting
parallel between both speakers is also that specifically the data point of is the one
that has the lowest F1 and the highest F2 average values. Since is likely to have a
higher frequency than the other two stimuli from the same category, and , as
well as having the /i/ in the stem as opposed to /u/, this piece of evidence contributes to a
bigger picture of a host of different factors all playing into these formant differences. I
110
would conclude that this brief phonetic study suggests that there is indeed a difference
between the vowels or vowel variability in ke- compared to ne-, and choosing different
spelling conventions for the two can in fact be argued for with these data. Further data, of
course, would help to evaluate the research question with more accuracy.
This section considers three differences between the similar prefixes cho- ‘auto-
benefactive/malefactive’ and che- ‘reflexive/reciprocal’ with regard to what may be
understood as phonological strength. The similarity between the prefixes is not only their
shared onset palatal stop, but also the tone change both prefixes cause in monosyllabic
stems, see §3.9.1.1.
The first difference is also mentioned in §3.9.1.1 and consists in cho- changing
low tones into mid tones in disyllabic stems in addition to monosyllabic stems such that
low tone ‘choose’ turns into mid tone - due to prefixation of cho-, while
prefixation of che- does not result in a tone change, so - remains low tone. The
tone change from mid tone to high tone in disyllabic stems, however, does not seem to
occur. Instead, mid tone disyllabic stems remain mid tone after prefixation of cho-.
Examples are cho- from ‘break a long object’; - from ‘steal’;
and - from ‘cook with alkaline’.
The second difference can be seen in the same example: cho- keeps its /o/ vowel
when attaching to disyllabic stems that start with /ing/, while che- looses its vowel.
The third difference is described in §3.5.5: While che- and cho- both change the
tone category of low and mid tone monosyllabic stems (as described in §3.9.1.1), cho-
additionally induces a sandhi effect that is absent in the prefixation and resulting tone
change caused by che-. Therefore, while - and - both change low tone
‘to go’ to mid tone, the resulting - has the same pitch contour as -
‘NMLZ-be.old’, a non-derived sesquisyllabic mid tone stem, while - has a higher
pitch contour.
These three differences thus provide converging evidence that che- is
phonologically/phonetically weaker than cho-. Due to the related semantics of che- and
111
cho- in that both function to detransitivize, i.e., che- as a reflexive/reciprocal marker and
cho- as a auto-benefactive/malefactive marker, the following historical scenario can be
hypothesized: The origin of cho- could be a bimorphemic sequence of che- plus another
prefix, which was fused while leaving behind these traces of being more than just a
typical, sesquisyllabically patterning prefix like che-. It is not clear at this point, however,
what the second morpheme in that assumed historical morpheme sequence could have
been.
112
CHAPTER IV
4. WORD CLASSES
This chapter deals with Karbi word classes. It starts with the two major word
classes of nouns and verbs (§4.1), and then tackles the question of what the status of
property-concept terms or ‘adjectives’ is in Karbi in §4.2. Specifically, it is argued that
property-concept terms are a subclass of verbs, which is substantiated with a number of
morphosyntactic tests. The next section §4.3 then offers a summary of the previous two
sections by discussing the clausal functions of nouns, verbs, and property-concept terms.
Section §4.4 is devoted to the discussion of the various subclasses of nouns found
so far, and a separate section §4.5 gives an overview of pro-forms that include personal
pronouns, demonstratives, interrogative pronouns and pro-adverbs, and pronouns and
pro-adverbs of universal quantification. Section §4.6 is about verb subclasses, although
the discussion of the large verbal subclass of property-concept terms occurs in §4.2.
In §4.7, frozen prefixes on Karbi nouns and verbs are surveyed. Some of these
have long been known and used in the reconstruction of Proto-Tibeto-Burman
morphology. Finally, §4.8 goes over several minor word classes, including adverbs,
numerals, as well as subordinators and coordinators.
113
Below I will argue that Karbi has old prefixal morphology that distinguishes
between nominal and verbal roots, but that the (perhaps younger) verbal suffixal
morphology does not help much.
4.1.1. Morphosyntactic Criteria for Nouns and Verbs at the Root Level
The two most frequent prefixes in Karbi are a- ‘possessive (POSS)’ (see §5.3.1)
and ke- ‘nominalizer (NMLZ)’ (see §6.4.1). Using these two prefixes, it is possible to
divide the Karbi lexicon into three categories: those root that only take a- ‘POSS’ but not
ke- ‘NMLZ’, those roots that only take ke- ‘NMLZ’ but not a- ‘POSS’, and finally those roots
that take neither. While this last category of items that take neither is a residual category
that needs to be further investigated with the help of other morphosyntactic tests, the first
two categories are Karbi noun roots and Karbi verb roots. At the root level, all and only
Karbi verbs can take the ke- ‘NMLZ’ prefix, e.g., ‘eat’ in (9), and all and only Karbi
nouns can take the a- ‘POSS’ prefix, e.g., ‘cooked rice’ in (10).
Since ke- is a nominalizer, however, the derived stem that carries ke- becomes a
member of the word class of nouns, and is then eligible to take a- ‘POSS’, as in (11),
where the verb root ‘crow (of a rooster)’ is nominalized via ke- and then takes the
possessive a-.68
68
It should be noted that besides the a- ‘possessive’ prefix, the other personal possessive prefixes (see
§5.3.2) should also work as a morphological test for nounhood.
114
There is thus an asymmetry between these two word-class diagnostic prefixes: the
nominalizer ke- attaches to roots, while the possessive a- attaches to stems.
As for syntactic criteria that distinguish between nouns and verbs, the largely
cross-linguistically valid ones apply. For example, nouns head noun phrases in which
they may be modified by a range of different elements including demonstratives,
numerals (in classifier constructions), or relative clauses (see Chapter VII). Verbs
typically form the predicate and occur clause-finally (although nominal predicates do
occur in Karbi), and have different restrictions as to what types of predicate constructions
they may occur in. There are certain predicate constructions that only verbs occur in, but
never nouns (see Chapter VIII).
4.1.2. Lack of Word Class Distinctions with Suffixal Aspect, Modality, and Negation
115
functions as a noun phrase in this example, and specifically a nominal predicate. Again,
the irrealis suggests a future interpretation, translating as ‘(she) will be alone.’
What holds for irrealis - also holds for irrealis - . In (14), - attaches to
‘truth’, which is clearly recognized as a noun by the possessive prefix a-. Here the irrealis
has an epistemic reading of expressing uncertainty, ‘might (they) be the truth?”.
In (18), the certainty marking suffix - ‘definitely’ occurs on a verb stem, which
is where it usually occurs (although more typically with one of the irrealis suffixes - or
- ). But in (19), we find - ‘definitely’ on the interjection ‘yes’ with the purpose
of showing that the speaker emphatically agrees.
Lastly, (20) and (21) provide another example of a typically verbal suffix used on
a noun stem: the formality marker - (see §12.4.2). In (20), it attaches to the verb stems
t ‘meet’ and t ‘tell’. Note also that the pronoun ilitum carries a separate -
117
‘honorific’ marker, which acts as a nominal honorific/formal counterpart to the verbal -
marker.
chethan’ikronglonang
che-th n- -r ng-lon ng
RR-tell-FRML-instead-lo.HORT
'[…] today we meet, let's tell a story instead (of doing other things)' [CST, HM 002]
The evidence presented in §4.1.1 and §4.1.2 shows that there are inherently
nominal roots and inherently verbal roots. A pair of diagnostic affixes are the a-
‘possessive’ prefix, which only attaches to nominal roots functioning as stems (or
nominal stems derived from verbal roots), and the ke- ‘nominalizer’ prefix, which only
attaches to verbal roots. As is the case cross-linguistically, knowing that a particular root
belongs to the Karbi word class of verbs predicts that it will most frequently function as
the predicate of a clause; knowing that a particular root belongs to the Karbi word class
of nouns predicts that it will relatively rarely function as the predicate of a clause, and
instead regularly heads a noun phrase in order to indicate reference to an entity.
The evidence presented above has further demonstrated that noun roots/stems do
not need to be modified in any way to function as predicates. Nominal predicates may
118
contain a range of prototypically verbal suffixes including the irrealis modal - and -
suffixes, aspectual suffixes such as - t ‘perfective’, - - ‘continuative’, the
certainty expressing suffix - ‘definitely’, the formality marker - , and even the quasi-
reduplicative negative suffix, which perhaps is the most surprising given its tight
morphophonological integration with the stem it attaches to as well as considering that
there is a negative equational copula , which could easily (and does usually) serve to
negate nominal predicates (see §4.6.2.2 and §8.1.2).69 Overall, the ease with which
nominal roots or stems function as predicates in Karbi has likely functioned as a catalyst
for the grammaticalization of nominalization-based constructions such as nominalized
main clause constructions (see §9.7.3), or the development of subordinators from relator
nouns (§4.8.3).
As for the other direction, for verbal roots to function as noun stems,
nominalization via ke- is required (see Chapter IX).
The previous section has shown that we can sort roots into the large categories of
nouns and verbs. A third major category of word classes in languages of the world are
‘adjectives’. I use quotation marks because the term ‘adjectives’ inherently implies that
there is a structural class of adjectives with the semantics of referring to property
concepts. The existence of such a class is however not universal across languages of the
world, which has been widely shown ever since the publication of Dixon's (1977) ‘Where
have all the adjectives gone?’ In order to talk about the semantic class of ‘adjectives’
without the implication that they form a lexical category, I will use the expression
‘property-concept term’ (PCT) (Thompson 1988).
In what follows, I show that there is a tremendous overlap in the kinds of
morphosyntactic constructions that simple (i.e., underived) prototypical verb stems and
simple stems with the semantics of property-concept terms occur in. One such
69
As another example of negating a nominal predicate by means of the quasi-reduplicative suffix, consider
t ‘tiger’ and t - ‘it’s not a tiger’, which was deemed acceptable. However, t with the
syntactic expression of nominal predicate negation via the equational negative copula would be the
more normal way of saying ‘it’s not a tiger.’
119
construction was already discussed above: the prefixation of the nominalizer ke-, which
occurs on PCT roots just as productively as it occurs on prototypically verbal roots.
Table 55 gives an overview of the morphosyntactic constructions that are
discussed in the following sections in an attempt to identify criteria to distinguish
between the semantic category of actions or events (i.e., ‘verbs’) and the semantic
category of PCTs (i.e., ‘adjectives’).
Table 55. Morphosyntactic criteria for the attempt to identify a class of ‘adjectives’
Possible PCT class criteria
May take - - t ‘comparative’
May take - i ‘superlative’
“V t V-suffixes” intensifier construction
“V a-V” adverb construction
Possible verbhood criteria
May take aspectual and modal suffixes
(May function as the predicate of a clause)
May take an imperative / hortative suffix
Noun modification
May modify nouns (ke- ‘NMLZ’ required)
The following subsections are organized as follows. Before discussing the criteria
in Table 55, I provide a list of roots that belong into different semantic types of property
concept terms in §4.2.1. Then I discuss the ‘possible PCT class criteria’ (as listed in
Table 55) in §4.2.2, and ‘possible verbhood criteria’ in §4.2.3, and move on to whether
there are differences in ‘noun modification’ constructions in §4.2.4. The discussion of the
various criteria is summarized in §4.2.5, and §4.2.6 surveys some marginal types of PCTs
with different morphosyntactic properties.
According to Dixon (1977, 2004), the most basic property concept terms fall
under the semantic domains of AGE, DIMENSION, VALUE, and COLOR. Dixon predicts that
even languages with very small, closed adjective classes have adjectives that belong into
these four semantic domains. According to him, if languages have larger adjective
categories, their members will also cover the semantic domains of SPEED, PHYSICAL
PROPERTY, and HUMAN PROPENSITY. In Karbi, roots that belong in six of Dixon’s seven
120
domains share the properties discussed in the following sections. All domains except for
that of HUMAN PROPENSITY are expressed by roots; some basic ones are given as
examples in Table 56.
The domain of HUMAN PROPENSITY is the only one for which simple roots are not
readily found; no such root occurs in the corpus.70 Instead what we find is that derived
stems and complex predication constructions are used to express concepts of HUMAN
PROPENSITY. Table 57 offers examples of derived stems and of so-called ‘psycho-
collocations’ (Matisoff 1986), i.e., predicate constructions that involve an obligatory
noun like ‘heart/mind’ along with a property-concept term root (for further
discussion, see §8.6.2).
70
Exceptions are metaphorically extended roots, such as ‘be sharp (like, e.g., a blade)’, which, just like
English, has a metaphoric sense ‘be smart, clever’ (see, e.g., HK, TR 064).
121
Table 57. Sample complex stems expressing concepts of HUMAN PROPENSITY
Type Stem structure Gloss Translation
Derived stem - - ‘RR-talk-GOOD’ ‘nice to talk to, kind,
generous’
- ‘see-BAD’ ‘be bad, evil’
Psycho- a- ‘POSS-heart/mind be.cold’ ‘be glad, happy, be
collocation grateful’
a- ‘POSS-heart/mind ‘be happy’
be.happy’
Compare this to (24), where the same comparative and superlative suffixes attach
to the underived, propotypical (i.e., non-PCT) verbal stems ‘eat’ and ‘dance’.
122
(24) - t ‘eat-COMPAR’ - t ‘dance-COMPAR’
- ‘eat-SPLT’ - ‘dance-SPLT’
[SiT 130827, elicitation]
The resulting meaning is ‘eat more’, ‘eat the most’, and ‘dance more’ and ‘dance
the most’, i.e., quantification of the action or event (‘dancing more’), or, in a sense,
quantification of the object in transitive events (‘eating more’). The comparative and
superlative constructions are fully productive among roots that may take ke-
‘nominalizer’, i.e., both PCT roots and prototypical verb roots (see also §6.5.5.1).
Grüßner (1978: 107) even provides the example shown in (25), where the superlative -
attaches to a classifier marked diminutive by - , in order to get a reading of ‘the smallest
[of the given cylindrical items]’.
(25) - - -
POSS-CLF:cylindrical-DIM-SPLT
‘the smallest (loaf of bread)’ (Grüßner 1978: 107)
I was able to confirm the construction in (25) as a productive pattern, but only
with classifiers. In order to use this construction with a common noun, a noun phrase as
in (26) needs to be formed so the superlative still occurs on a diminutive marked
classifier. Using a noun root in this construction such as * - - ‘house-DIM-SPLT’ is
unacceptable.
(26) - - -
house POSS-CLF:container-DIM-SPLT
‘the smallest house’
Lastly, also note that there is another superlative construction, which consists of
the suffix - attaching to nominal PCTs, as in (27) (see §5.4.5).
(27) - - -
1EXCL:POSS-older.brother POSS-old.one-SPLT
‘my oldest brother’
123
It turns out then that gradability is not a useful criterion for differentiating a word
class of property-concept terms (PCTs) in Karbi. Comparative and especially superlative
constructions by no means exclusively take PCT roots.
Another construction semantically related to gradability that can be considered in
the search for constructions that distinguish between PCT roots and prototypical verb
roots is the “V=t V-suffixes” copy verb construction (see §12.2.3.2). It has an
intensifying function, which may be more prototypically related to property-concept
terms rather than actions or events. However, (28) and (29) show that not only PCT roots
like ‘be good’, but also prototypical verb roots like ‘reach’ participate in this
construction.
124
It appears that prototypically verbal roots do not participate in this construction.
For example, it is not possible to say * - ‘go CAUS-steer/drive’ to mean ‘go by
driving’ or *ch- ‘RR-talk CAUS-shout’ to mean ‘talk to one another by
shouting’.
However, this construction is not entirely productive with (basic) PCTs either. For
example, t - ‘cook CAUS-be.good’ to mean ‘cook well’ is acceptable, but t -
‘cook CAUS-be.bad’ to mean ‘cook poorly, badly’ is not acceptable.71
71
However, it is possible to say t -lang- ‘cook CAUS-look-BAD’ with that meaning.
125
Even though a range of declarative verbal suffixes occur with simple stems
consisting of PCT roots in the corpus, there is no instance of a PCT root occurring with
directive suffixes such as conative imperative -t , conditioned imperatives - and - ,
prohibitive - , and hortative - and conative hortative - (§11.1.2 and §11.1.3).
Instead, where PCT roots occur with directive suffixes, they also take the causative prefix
pe-~pa- as in (33). In this example, pa- is obligatory.
(34) - - - -
2POSS-mind come-PROH 2POSS-head be.hot-PROH
‘don’t throw up!’ ‘don’t have a fever!’ [SiT 130825]
Besides ‘be hot’ in the expression for ‘have a fever’, the PCT t ‘be big’ also
takes a directive suffix, as in t -t ‘be.big-CON.IMP’, which may be said to children
with the meaning ‘grow up (well)!’. It is not clear whether t in this case is the PCT ‘be
big’ or a case of polysemy, where it is actually a verb with the meaning ‘grow up’.
All in all then, evidence from directive suffixes does point towards a distinction
between ‘proper’ verbs and PCTs. Being more of a semantic rather than a structural issue,
however, this is not the kind of evidence we might want to strongly rely on to talk about a
separate word class of adjectives.
126
(2011), which was based on data in Grüßner (1978)) that the noun modification
constructions of PCT roots and prototypical verb roots show a clear syntactic difference.
While both are marked morphologically the same, i.e., nominalization via ke-, the order
of head noun and modifier appeared to mark a clear difference: PCT-based modifiers are
post-head, whereas prototypical verb-based relative clauses are pre-head. Elicited PCT-
based modifiers and prototypical verb-based relative clauses have shown the same pattern.
It turns out, however, that this clear distinction does not hold up against a large
body of natural data. There are instances in the corpus of pre-head PCT-based modifiers
and there is one potential instance of a post-head relative clause (although it is impossible
to distinguish in purely structural terms a simple (i.e., verb-only) post-head relative
clause from a simple internally-headed relative clause in a verb-final language such as
Karbi).
Noun modification nominalization constructions are discussed in detail in §7.7.1,
but relevant examples are repeated here: (35) shows a ‘standard’ post-head PCT-based
modifier, (36) a ‘standard’ pre-head prototypical verb-based relative clauses, (37) a pre-
head PCT-based modifier, and (38) what can be interpreted as a post-head, but also as an
internally-headed, relative clause.
127
elong adim dosi computer akam klemlo
e-l ng a-d m d -si computer a-k m kl m-l ]
one-CLF:place POSS-place exist-NF:RL computer(<Eng) POSS-work do-RL
‘[…] Lily, the lady who took us there, she stayed in one place and did her computer
work, and she said to us, <enjoy yourselves as much as you want>' [SiT, HF 034]
To summarize the discussion above, consider Table 58, which is based on Table
55 above, but has added columns that show the participation of prototypical verb roots
and PCT roots in each construction.
The majority of PCTs share the properties laid out in the previous sections and
summarized in §4.2.5. But there are a few other, marginal types of PCT constructions that
occur in the corpus and should be mentioned. Compared to the properties shared by basic
PCTs, these PCT constructions appear not to be productive: there are very small sets of
items that can occur in these constructions. They are listed in Table 59, which also
references relevant examples given below. Note that most of these PCTs have semantics
related to SIZE, which is one of the core semantic categories of ‘adjectives’ according to
Dixon (1977).
72
The ‘modifier’ suffix - appears to derive modifiers from PCT roots, but it is not synchronically
productive.
129
(39) as nominal modifier
nangong akleng ahemke nangong
[nang- ng a-kl ng a-h m ke] [nang- ng
2:POSS-maternal.uncle POSS-old.one POSS-house=TOP 2:POSS-maternal.uncle
ki'ikrintile putangho
ke- k-rint le p -t ngh
NMLZ-be.black-equally:PL:S/A=FOC:IRR QUOT-REP
'the family of your older maternal uncle, the family of your uncle who is the eldest son of
the family, all of them are so black (so you shouldn't get your wife from them)' [WR,
BCS 014]
130
akibi abangke Bey Ronghang
[a-ke-b ab ng ke B y Rongh ng]
POSS-NMLZ-be.small NPDL=TOP CLAN CLAN
'there were three Bey brothers, the oldest one was Bey the Black, the second one was Bey
the Fair, and the small one was Bey Ronghang' [WR, BCS 002]
(44) t - ‘big (< t ‘be big’ and - ‘MODIF’) following its head noun
anke ha langso asiluka thepota pulelo
nke h [langs a-s luk the- t ] p -l -l
and.then over.there this POSS-sp.tree(<Asm) be.big-MODIF=ADD:DM say-again-RL
‘and then, there, this big siluka tree replied (said again)’ [KK, BMS 107]
(45) t - ‘big (< t ‘be big’ and -p ‘AUGMENT’) following its head noun and
b ‘very big (<Ind)’ preceding its head noun
la bhari talo thepi nangkekapji kopuloma
l [bhar tal the- ] nang=ke-k p-j kop lo ma
this very.big(<Ind) sea be.big-AUGMENT CIS=NMLZ-cross.water-IRR2 how=FOC=Q
‘how will we be able to cross the huge sea?’ [KK, BMS 100]
131
(46) - ‘long (< ding ‘be long’ and - ‘MODIF’) following its head noun and
‘very big (<Ind)’ preceding its head noun 73
bhari arleng dingpo
bhar arl ng d ng- arl -l
very.big(<Ind) slope be.long-MODIF climb-RL
‘and she climbed a very big slope’ [SeT, MTN 046]
73
Note that there is another occurrence of - ‘big(?)’ in the corpus, where it is a lot less clear whether it’s
a productive use of - , or an idiosyncratic, lexicalized instance. This is ingnar nothongpo ‘deaf elephant’,
a character in a folk story (note also the use of - and - in animal species names in §5.4.1):
132
(50) ng ‘a lot’ following its head noun
[…] photo penang endunlo
[phot en ng n-d n-l
photo(<Eng) a.lot take-JOIN-RL
‘ […] (we) took many photos’ [SiT, HF 035]
4.3. Summary: The Clausal Functions of Nouns, Verbs, and Property-Concept Terms
(PCTs)
Croft (2001: 99) offers a figure that shows a semantic map of English parts of
speech constructions, mapping particular constructions that English nouns, adjectives,
and verbs occur in onto the three basic clausal functions of reference, modification, and
predication. In Figure 13, I have copied Croft’s visual representation for mapping parts of
speech constructions onto clausal functions for Karbi without, however, providing details
of different constructions. Instead I have limited this figure to a very basic, and in that
sense simplified, picture of the clausal functions of noun, PCT, and verb roots in Karbi.
NOUN
Figure 13. Semantic map of basic clausal functions of Karbi parts of speech
(following Croft's (2001: 99) model)
What is striking about Figure 13 is that nouns can assume any clausal function
without overt marking, and that noun, PCT, and verb roots all can function as the
predicate of a clause without overt marking. Furthermore, basic PCT and prototypical
133
verb roots basically pattern the same, although I indicate the different syntactic
tendencies in noun modification by the narrowly dotted line.
While I have shown nouns (§4.1.2) and PCTs (§4.2.3) in predicate function, as
well as PCTs and prototypical verbs marked by ke- ‘nominalizer’ in modification
function (§4.2.4), and a verb marked by a-ke- ‘POSS-NMLZ-‘ in reference function (§4.1.1),
what is still missing are examples of an unmarked noun in modification function and a
PCT root marked by a-ke- ‘POSS-NMLZ-‘ in reference function. These are given in (51)
and (52), respectively.
Second after proper nouns, the largest word class that we can consider a subclass
of nouns are classifiers, which are discussed in §4.4.1. In turn, classifiers create
subclasses of nouns by virtue of being classifiers. Some nouns are not enumerated in a
classifier constructions, however, but instead are directly modified by numerals;
information on this is found in §4.4.2. Section §4.4.3 briefly discusses how nouns with
human/personified referents are treated differently from nouns with non-human/non-
personified referents in Karbi grammar. Section §4.4.4 is about relator nouns, i.e., those
words that function the same way adpositions do in other languages. In §4.4.5, the
subclass of those nouns that need to carry a possessive prefix is investigated, including
body part and kinship terms. Note that relator nouns as well as body part and kinship
terms all need to be considered bound roots since they require a possessive prefix in order
134
to occur in an utterance. Nevertheless, the completely different clausal function of relator
nouns compared to body part and kinship terms advocates fo their discussion in separate
sections. Lastly, §4.4.6 offers a brief discussion of frozen nominal prefixes given by
Grüßner (1978).
4.4.1. Classifiers
135
In this example, the classifier functions just like a noun, evidenced not only by prefixing
a- ‘possessive’, but also by being modified by ‘iron’, and occurring in the relator
noun construction with - ‘LOC’.
kevekponlo
ke-v k-p n-l
NMLZ-steer-take.away-RL
‘and then, he doesn't know how to sit down on that sitting place (seat), and then on the
iron bar (/top tube) he sits and steers the bicycle away’ [SiT, PS 024]
Of course many classifiers have their origins in and still synchronically exist side-
by-side with true nouns, in which case they have a regular noun counterpart. One could
consequently argue that whenever a- occurs on a ‘classifier’, that instance actually
represents the noun and not the classifier. However, t in (54) is a good test case
because it is a classifier borrowed from Khasi, as Joseph (2009) has convincingly argued,
136
and thus does not originate in a native Karbi noun. Still t may occur with a-
‘possessive’ and -t ‘each’, which provides evidence for arguing that classifiers are a
subgroup of nouns.
In addition to classifiers forming a subclass, each group of nouns classified by one
particular classifier of course also forms a subclass of nouns.
Sortal classifiers are ‘true’ classifiers in the sense that they actually ‘classify’ or
‘sort’ nouns into categories, in Karbi specifically only when they are counted (although
see §7.7.2.6.1 for the grammaticalization of an indefinite article from a classifier-numeral
word counting ‘one’ item). As we would expect, the great majority of sortal classifiers
are typical, i.e. fully grammaticalized, classifiers (§4.4.1.1.1), but there is one,
presumably one of the most frequently used classifiers, that is not fully grammaticalized,
which is the human plural classifier , discussed in §4.4.1.1.2.
Table 61. List of sortal classifiers that occur in typical classifier construction (G =
Grüßner (1978: 68-70))
Category Classifier Gloss / Description Classified items Source
Animate n t75 ‘CLF:HUM:SG(<Khs)’ one human being (only C/G
used for singular reference)
‘CLF:family’ families/houses (in a C/G
village)
74
For the items recorded by Grüßner only but not in the present corpus, a characterization of what is
classified is provided instead of a full gloss.
75
From Standard Khasi reconstructed form *shi-ngut or Pnar reconstructed form *chi-ngut (Joseph 2009).
137
76
‘CLF:animal’ animals: dogs, tigers, birds C/G
‘CLF:plant’ referring to the plant as a C/G
whole (e.g., trees)
Physical ‘CLF:long.cylindrical’ long, cylindrical items: C/G
properties arms, legs, table legs,
jambili athons (Karbi
totem, see Figure 4 in
§1.1.3), matches (G)
77
‘CLF:flat’ medium- to large-sized flat C/G
items: knives, shovels,
books, shirts, bamboo mats
‘CLF:small.flat’ small flat items: baked C/G
items,
‘CLF:round’ round items: e.g. round C/G
fruits like oranges, apples,
eggs
78
long/thin79 hair G
‘CLF:line’ long, flat items: strips of C/G
meat
long/thick ropes G
Functional ‘CLF:house’ houses, busses (G), packs C/G
properties of cigarettes (G)
bigger containers water containers made G
from bamboo
smaller containers cups/bowls G
t fruit and animal oranges, animal calls G
calls80
81
tools with handles brooms G
Nature and ‘CLF:road’ roads, paths C/G
82
environment fields paddy fields G
water areas, moving rivers G
waters
grains, seeds rice grains G
76
Perhaps from ‘length of an animal from head to tail’ (Grüßner 1978: 66).
77
This may be cognate with Meithei pak ‘be broad, be wide’ (Chelliah 1997: 335).
78
Presumably from ‘thread’ (Grüßner 1978: 66).
79
See footnote 74.
80
Like with (see below) Grüßner (1978: 68) suggests there is only one classifier t , where we may
want to pose two homophonous classifiers based on the highly diverging semantics.
81
Presumably from ‘handle (n.)’.
82
Presumably from ‘EE:r t (field)’.
138
83
bush, shrub bamboo bush G
Generics/ ‘CLF:thing’ “things” ( ‘thing’), C/G
Abstract words, songs, matters,
items news; also: for kinds/types
of items (e.g. drums),
t ‘CLF:word’ words, matters, issues C/G
84
‘CLF:place’ markets (G), C/G
85
Time ‘CLF:month’ months C/G
The plural human classifier is the only sortal classifier attested so far that is
not fully grammaticalized, as evidenced by the fact that independent instead of bound
numerals are used for ‘two’ and ‘three’ (but not ‘one’, since only t is used to count
‘one person’).
The classifier must have grammaticalized from the noun ‘body’, which
must also be the origin of the indefinite ‘somebody’. Presumably the classifier has
then also given rise to what I call here the ‘noun phrase delimiter’ (see §10.5). The
noun ‘body’ also occurs in a grammaticalized construction, in which it is used with
a personal possessive pronoun instead of just the personal pronoun by itself, seemingly in
order to specifically refer to a person’s physical presence, as in (57).
Unlike sortal classifiers, mensural classifiers are not ‘true’ classifiers in the sense
that they do not ‘classify’ items but only ‘measure’ them. Presumably all languages have
a way of expressing mensural classification, whereas sortal classification is a typological
feature limited to certain parts of the world, Southeast Asia being a prominent one.
Some mensural classifiers are grammaticalized and occur in the ‘typical classifier’
construction that sortal classifiers in Table 61 in §4.4.1.1.1 occur in as well (which is a
good reason to call both types ‘classifiers’). They are listed in §4.4.1.2.1. Others are not
fully grammaticalized, and may even be created ad hoc; these are discussed in §4.4.1.2.2.
Table 62 offers a list of mensural classifiers that appear in the ‘typical’ classifier
construction, i.e., including a head noun and a numeral-classifier word that consists of a
140
classifier and a bound numeral. The classifiers are sorted into semantic categories for
presentational purposes.
86
See footnote 74.
87
Grüßner (1978: 65-66) reports the form instead of for ‘half’.
88
Note that in the case of t ‘(CLF:)piece’, we see a pattern of corresponding low tone verbs and mid tone
classifiers/nominals. For more information on this tone correspondence, see §3.5.10.
141
(58) Classifier for ‘half’
hala arleng ebeng'an t […]
h la arl ng e-b ng- n am tsi a-khal n ingp -v k-l
that slope one-CLF:half-up.to and.then POSS-kd.big.basket open-RES:open-RL
‘she (had climbed) half of the slope and then she opened her khalun basket […]’ [SeT,
MTN 050]
Note that Grüßner (1978: 70-71) offers additional measuring units including ones
for weight, volume, length, area, money, traditional items, and time; he provides some
traditional units and units borrowed from Assamese/Indic and English.
In his list of classifiers, Grüßner (1978) offers the three that I provide below in
Table 63, which occur in the not fully grammaticalized construction involving the head
noun, the classifier, and the independent numeral, e.g., t ‘one set of keys’
(Grüßner 1978: 69). In addition to these, any noun that denotes a container of some sort
can be used ad hoc as a mensural classifier in this construction, e.g., ‘cup’ to
measure cups of tea, for example, or t ‘bucket’ to measure buckets of water
(Grüßner 1978: 70).
A list of all self-referential classifiers attested so far is offered in Table 64. The
term ‘self-referential classifier’ is taken from DeLancey and Boro (in preparation). It
refers to those classifiers that count themselves instead of counting (and being
morphosyntactically dependent on) a head noun (see §7.7.2.1.2).
142
Table 64. Self-referential classifiers
Form Gloss Measured items Source Possible lexical origin
89
Time ‘CLF:day’ days C/G ‘day (noun)’
‘CLF:night’ nights C/G ‘night (noun)’
‘CLF:week’ weeks C/G
‘CLF:times’ once, twice, etc. C
ng C/G
C/G
[ G]
90
Other ‘CLF:step’ steps C ‘step (v.)’
Almost all self-referential classifiers are time units as has also been noted for self-
referential classifiers in Boro (DeLancey and Boro in preparation). However, in Karbi
there is at least counting steps, which also acts as a self-referential classifiers as seen
in (59). For ‘night’ and ‘day’ acting as self-referential classifiers, see (60). Note the
irregular form for ‘one day’ as - - there is no other classifier found so far that
follows this pattern.
89
The enumeration with ‘one’ is irregular: it is - ‘day-one’ for ‘one day’, rather than * - .
90
Note that in the case of k m ‘CLF:step’, we see a pattern of corresponding low tone verbs and mid tone
classifiers/nominals. For more information on this tone correspondence, see §3.5.10.
143
4.4.1.4. Human and Animal Classifiers and Personification
In folk stories, animals frequently act as rational beings. This has ramifications on
the classifier-marked distinction between humans and animals, as the human
classifiers -n t (for singular) and - (for plural) are typically used for animals as well.
Although in (61), the dog t - is counted via the animal classifier , as the
storyteller summarizes that between the one dog and the two people, there were three of
them, she uses the plural human classifier in the classifier-numeral word - t .
vangchomchomchomchomchom
v ng-ch m~ch m~ch m~ch m~ch m
come-a.little~DIST.PL~DIST.PL~DIST.PL~DIST.PL
'”let's go!”, and then one dog, his grandmother and his grandfather, all three, went step by
step by step’ [KK, BMS 099]
An even clearer example is (62), where the human plural classifier is used to
count ants, which represent protagonists in this folk story.
The lines in (63) and (64) give the context for the use of the animal classifier to
refer to a human in (65). In this folk story, tigers celebrate the same ceremony as
humans/Karbis, which requires sacrificing chickens and/or goats, as explained in (63)
(which represents an utterance not by the primary storyteller but by a native Karbi
speaking listener, as indicated by the curly brackets).
144
(63) Context for (65)
{la monitsi kenangpo he, halatum aphanke,
l mon t si ke-n ng-p he h la-tum aph n ke
this man=FOC NMLZ-need-IRR1 AFTERTHOUGHT that-PL NSUBJ=TOP
The line in (64) immediately precedes (65) and mentions the ‘village head tiger’
(where the use of this terminology further underscores the personification), who collects
the humans that are caught by the other tiger as sacrifice for the ceremony.
The line in (65) clearly refers to the tigers requiring one more human being as
sacrifice. The animal classifier anaphorically refers to humans used as sacrifice.
What these examples show is that the classifier use is dynamic and depends more
on the perspective of ‘higher’ vs. ‘lower’ being, rather than objectively human vs. animal.
145
4.4.2. Nouns That Are Counted with Numerals Only (without Classifiers)
Table 65 offers a list of these nouns that are counted with independent numerals
(instead of classifier-numeral words), sorted into ‘monosyllabic’, ‘disyllabic’, or
‘borrowed’. Note that some of the ‘monosyllabic’ and ‘disyllabic’ nouns may in fact be
borrowed as well, which is, however, yet to be investigated.
The reason a distinction is drawn between monosyllabic and disyllabic stems is
that Grüßner (1978: 64-65) suggests that might be a criterion: he suggests that
multisyllabic stems occur with independent numerals, whereas monosyllabic stems occur
with bound numerals. As Table 65 shows, this is not true without exceptions, but it could
very well represent the right tendency.
146
Table 65. Nouns counted with numeral only (without classifier)
Category Form Gloss
Monosyllabic ‘finely woven bamboo basket’
91
‘village’
‘tribe, community’
Disyllabic t ‘kind of bamboo basket’
‘year’
t ‘time’ (as in ‘once’, ‘twice’, etc.)
‘stone, rock’
‘society’
’ ‘festival’
‘bag’
Borrowed t ‘week(<Asm)’
t ‘hour(<Asm)’
‘time(<Asm)’
‘share, portion(<Asm)’
‘reason(<Ind)’
While certain nouns only occur in the direct enumeration construction when
counted, other nouns may occur in the direct enumeration construction while also
occurring in classifier constructions. An example is (69), in which ‘matter’ is
counted with ‘one’ instead of - n ‘one-CLF:thing’.
91
Grüßner (1978: 67) reports that occurs as a self-referential classifier the same way as ‘night’
(§4.4.1.3).
147
(69) ‘matter’ in direct enumeration instead of classifier construction
sita non isi alam abangke
set n n is a-l m ab ng ke
but now one POSS-matter NPDL=TOP
‘but on another matter […]’ [KaR, SWK 097]
This option to occur in the direct enumeration construction may in fact be true of
all nouns that perhaps more typically occur in classifier constructions. According to
Grüßner (1978: 67), it is true of all nouns with inanimate referents; those that refer to
humans or animals have to be counted with the classifiers t ‘CLF:HUM:SG’,
‘CLF:HUM:PL’, or ‘CLF:animal’. Example (70) shows, however, that perhaps at least
mistakenly,92 even nouns with animate referents may occur in the direct enumeration
construction in natural texts, here ‘wife of headman’.
Where there is an option between using the classifier or the direct enumeration
construction, my impression is that younger speakers use direct enumeration more
frequently than older speakers do. In addition, younger native Karbi speakers have been
observed to switch to Assamese numbers in enumeration constructions, which is certainly
a common type of contact influence given that markets are typically Assamese-speaking
places.
It may be fairly universal that languages have some grammatical domains where
human/personified referents are distinguished from non-human/non-personified ones, for
example in English interrogative pronouns ‘who’ versus ‘what’. In Karbi, this distinction
additionally exists in classifiers (though see the discussion in §4.4.1.4), as well as in
92
It was suggested to me that the speaker in (70) mistakenly said instead of - t.
148
‘differential’ plural marking via -t (§7.6) and primary object marking via -
(§10.6.5).
Relator nouns exhibit structural properties of nouns while functioning the same
way adpositions do in languages that have them as a distinct category (Starosta 1985;
DeLancey 1997). The reason why they structurally overlap with nouns is because they
grammaticalize from lexical nouns. In Karbi, the possessive construction that relator
nouns most typically occur in involves a possessor noun followed by an a-
‘possessive/modified’ marked possessed noun, schematically: [N]POSR [a-N] POSD (see also
§7.3 and §7.4); the a- marked possessed noun is the relator noun. An example is (71),
where - is the relator noun, and the expression - ‘at the side of the river’
functions like an adpositional phrase in English, here specifically as a locational oblique.
Relator nouns are bound roots. Although they most frequently occur with the a-
‘possessive, modified’ prefix in the corpus, they may also occur with personal possessive
prefixes. An example is (72).
nangklochomlo dei
nang=kl -ch m-l d i
CIS=fall-together.few.close-RL OK?
‘[…] two grasshoppers, over there, from above us, children of god fell down’ [KTa, TCS
009]
149
A list of so far attested relator nouns is given in Table 66. They are grouped by
semantic domain. Instances of the common pathway of metaphorical shift of ‘locational >
temporal > causal’ are -i ,- , and - , of which at least the latter two have
corresponding lexical nouns with locational semantics that undoubtedly represent their
origins. What may be the lexical source of the relator noun, or anyway a related
lexical/grammatical item is provided where known. Some of the relator nouns given here
are recorded with different tones by Grüßner (1978), which is mentioned in footnotes.
-
95
- ‘near’
- ‘at.bottom.of’
-a ‘inside’
- t t 96 ‘towards’
97
-i ‘towards’ (general
direction)98
- ‘amidst’
- ‘road.inbetween’
Locational/ -i ‘in.middle.of’99
93
Grüßner (1978: 77) reports this item with high tone as y.
94
A text example of this suffix is in the verb - - - ‘RR-see-sideways-RL’ meaning ‘looked
sideways’ (HK, TR 053).
95
Grüßner (1978: 77) reports this item with high tone as .
96
Grüßner (1978: 77) reports this item with high tone on -t .
97
Grüßner (1978: 77) reports this item with low tone as .
98
Grüßner (1978: 77) reports the meaning ‘opposite from’ for . I have put “general direction” in
brackets to indicate that this item has a more general sense than the more specific t t ).
99
The difference in meaning between ‘in.middle.of’ and ‘amidst’ is that locates a
person or thing right in the middle or center of something, whereas more generally locates a person or
thing within a defined area.
150
Temporal - ‘before’ - ‘front’
- ‘after’ - ‘backside’
Temporal100 - t ‘during’
Causal - t ‘because’
-
-
Other101 - ‘LOC’ ng ‘classifier:place’
- ‘for; NSUBJ; GOAL’
- t ‘with(<Asm)’ logot (<Asm) ‘id.’
~- t
- ‘beyond’
-t ‘according.to’
- ‘associated.with’ ‘side’
Relator nouns may occur with =pen ‘with, from’ to add an ablative dimension to
the meaning of the relator noun. Examples are (73) and (74).
(73) Relator noun - ‘high.up’ with =pen meaning ‘from high up’
thengpi angsongpen nangkeklosi {mm} thinilo {mm}
thengp angs ng pen nang=ke-kl -si mm th n -l mm
tree/wood high.up=from CIS=NMLZ-fall-NF:RL AFF be.almost.dead-RL AFF
‘(he) had fallen down from the tree and hurt himself badly, but by trying very hard came
home [HK, TR 196]
(74) Relator noun - t ‘towards’ with =pen meaning ‘from the direction of’
anung anatpen ketheklong <ma> kosonma angno
[[an ng a-n t pen] ke-thekl ng <ma> kos n ma] [[angn
back POSS-direction=from NMLZ-see Q how=Q in.front
100
Grüßner (1978: 77) also lists ‘as long as, until’ as a relator noun. Since this marker only occurs as a
subordinator in the corpus, I discuss it in §4.4.4.7.1.
101
Grüßner (1978: 77) also records the Assamese loan ab ‘against, opposed to’ as a relator noun.
This form does not occur in the present corpus.
151
Finally, a frequent use of relator nouns in discourse is that of discourse connectors
at the beginning of an intonation unit. This construction involves an anaphoric use of the
proximal or distal demonstrative followed by a relator noun, as in (75) and (76).
The general relator noun - ‘LOC’ is the most frequent relator noun in the
corpus, due to its bleached semantic content. It can be used instead of a relator noun with
a more specific semantic content. Examples below show the functional range covered by
- ‘LOC’. First, (77) shows two instances of - ‘LOC’ in its locative function.
152
Examples of other locational uses of - are given below. In (78), - has an
allative function of ‘motion towards’ a place, or location of a person or entity. In (79), -
indicates a relationship where one item is attached to another item. Example (80)
shows that - may be used even when there is a more specific relator noun that
describes a spatial configuration, in this case , which specifically means ‘in’.
kapabon
ke-pa-b n
NMLZ-CAUS-be.attached
‘and then, on this tobacco container, there was a mirror attached’ [HK, TR 026]
In (81), the metaphorical use of - is shown, as the context here is ‘working for
a particular project’. Furthermore, (82) shows that for certain verbs, such as ‘follow’,
- marks the object, i.e., the person being followed.
153
(82) Relator noun - marking O argument of ‘follow’
apiso along chidunkri […]
a- is a-l ng] chV-d n-kr
POSS-wife POSS-LOC RR-follow-follow.closely
‘he followed his wife closely […]’ [KK, BMS 082]
Finally, there are instances where - heads clauses and with the frozen
a- prefix functions as a subordinator. In (83), - functions as the head noun of the
locative relative clause ‘where the birds don’t chirp’. In (84) and (85), a marks
subordinate clauses that indicate simultaneity in (84) and perhaps causality in (85) (where
the elaborate expression ‘rule, be in charge of’ occurs in a parallelism
construction, see §12.1.1).
154
<so'arlosomar atum> arlosomar atum arpu ko'an do?
<s 'arlos -m r a-t m> rlos -m r a-t m arp ko' n d
women-PL POSS-PL woman-PL POSS-PL responsibility how.much exist
'so with running a family and being in charge, how much responsibility do women
have?' [KaR, SWK 026]
Examples of locational relator nouns are given below in (86) - (94). They are
given in the order of Table 66 as repeated in Table 67 for easier reference.
155
(88) Relator noun -i ‘in.front’
korte banghini <angno> angno nangklolo
ort b ng-hin <a-ingn a-ingn nang kl -l
same.gender.siblings CLF:HUM:PL-two POSS-in.front POSS-in.front CIS=fall-RL
‘[…] in front of the two brothers they fell’ [KTa, TCS 010]
156
(94) Relator noun - ‘in’
“ t ” “ longku arlo”
kon t lo? pu h long a-arl 102
where=FOC QUOT over.there cave POSS-in
'”where are they” (the king asked), “over there inside the cave” (the girls said)’ [CST,
HM 111]
anat chevangthulo
a-n t] che-v ng-th -l
POSS-towards RR-come-again-RL
‘and then, he whistled for him, picked up the hat and brought it back’ [SiT, PS 036]
pathikhangjanganang pu amat
[pa-th -khangj ng-lon ng pu am t]
CAUS-die-do.irreversibly-HORT:EMPH QUOT and.then
‘”[…] in the middle of the jungle, if this causes them to die, then let it happen”’ [CST,
RO 049]
(97) - ‘road.inbetween’
bonseta Dobokapen <Hojai> Hojai adak
b nset [[[ obo en <hojai> oj i] a-d ]
but PN=from PN PN POSS-road.inbetween
102
Note that when a- ‘possessive, modified’ attaches to disyllabic roots that start in ar-, one of the two /a/
vowels that end up occurring in a row is often deleted, especially in grammaticalized constructions such as
the relator noun construction (see §3.9.2.1).
157
4.4.4.3. Locational/Temporal Relator Nouns
Relator nouns that encode both locational and temporal concepts most likely
originate in locational lexical nouns. As part of a typical shift from locational to temporal
concepts, they end up encoding temporal relations as well. Table 68 lists three
locational/temporal relator nouns. Below examples are given that show the locational use
of ‘amidst’ in (98) and its temporal use in (99).
nangkethonti
nang=ke-th n-t
CIS=NMLZ-drop-get.rid.off
'and then, she took us here in the middle of the jungle and abandoned us [CST, HM 052]'
So far only one relator noun is attested that is used to express a temporal
relationship without also expressing location: - t ‘during’. An example is given below.
Note that other temporal relator nouns have further grammaticalized to subordinators (see
§4.4.4.7.1 and §9.5.2).
159
atumpen non adin ahut non akai
a-t m pen] [n n a-d n a-h t] [n n a-k i
POSS-PL=with now POSS-day(<Asm) POSS-during now POSS-time(<Asm)
There are three causal relator nouns: - t, -j , and - . They all occur much
more frequently as subordinators (see §4.4.4.7.1 and §9.5.2) than as relator nouns. In fact,
- does not occur as a relator noun in the corpus, but - t and - do, as in (104) and
(105).
akechome […]
a-ke-ch -m ]
POSS-NMLZ-eat-GOOD
‘this fragrant smell is the reason why the curry is tasty […]’ [SiH, KH 015]
160
4.4.4.6. Relator Nouns with Other Functions
Relator nouns that express functions other than locational, temporal, and causal
notions are given in Table 69.
The general relator noun - was already discussed in §4.4.4.1, and the
functions of - are illustrated in §10.6.2. Examples of the other four relator nouns in
this list are given below.
avelo
av -l
not.exist-RL
‘to me, this is nothing but a bad omen’ [KK, BMS 021]
161
api alam ateng amat laso a’osomar
a-p a-l m] a-t ng am t [[las a-os -m r]
POSS-female.animal POSS-matter POSS-according.to then this POSS-child-PL
kimi kiri
ke-m ] ke-r
NMLZ-be.new NMLZ-search
‘on behalf of the Karbi Lammet Amei, we are looking for new (i.e., special or rare) words
and expressions […]’ [KaR, SWK 001]
4.4.4.7.1. Subordinators
162
(110) Subordinator ‘after’ (grammaticalized from relator noun)
[…] laso hem nangkachiri aphi, apenan abang sunjoi […]
[[las h m nang=ke-che-r ] a h a-pen n ab ng s n-j i
this house CIS=NMLZ-RR-search after POSS-husband NPDL descend-quietly
‘[…] after she went back to search for it in the house, the husband quietly came down
[…]' [SeT, MTN 042]
Furthermore, some subordinators clearly look like relator nouns (i.e., they carry
the a- prefix), while they do not require the subordinate clause to be nominalized. They
thus represent an advanced stage of grammaticalization. An example is ‘before,
when’, which occurs in a construction with a negated, non-nominalized verb to convey
163
the meaning ‘before’, as in (113). For more information on subordinators, see §4.8.3 and
§9.5.
4.4.4.7.2. Adverbs
164
4.4.5. Bound Noun Roots
Body part and kinship terms, and inalienably possessed nouns have in common
that they are bound noun roots, i.e., they do not (typically) occur without a possessive
prefix (but see below for an exception). This may be either the general
possessive/modified prefix a- (§5.3.1) or one of the personal possessive prefixes (§4.5.1).
A lot of body part terms are monosyllabic and have clear Tibeto-Burman origins,
e.g., - ‘eye’, - ‘ear’, - ‘tongue’, - ‘tooth’, - ‘hand, arm’, - ‘foot’. Some,
however, are disyllabic, such as - ‘face’ and -nokan ‘nose’. A subset is disyllabic
and has the ing- prefix (see §4.7.1), for example, -i ’ ‘cheek’, -i ‘mouth’, -
‘lungs’, and -i ‘beard’.
165
4.4.5.3. Other Inalienably Possessed Items
In addition to relator nouns, and body part and kinship terms, a few other nouns
obligatorily have a possessive prefix. Among these are ‘border, mark’,
‘circumference’, p ‘foam’, t ‘unbroken rice’, t ‘hole (inhabited by an
animal such as a rat)’, ‘sombody’s turn (in a particular procedural order)’,
‘number’, ‘a little bit’. This presumably has to do with these nouns only ever being
used in a relational sense, i.e., as they relate to another concept.
4.4.6. Other Possible Noun Classes with Frozen Prefixes (Grüßner 1978: 44-6)
Grüßner (1978: 44-6) has sorted disyllabic lexical roots by shared first syllables.
The two by far most frequent ones, ing- and ar-, are discussed below (§4.7). In addition
to those two, Grüßner also offers lists of only disyllabic nouns (i.e., no verbs) that share
the first syllable kar- (seven roots in total), those that share ke- as their first syllable
(seven roots in total), those that share che- as their first syllable (fifteen roots in total),
and those that share me- or ma- as their first syllable (five roots in total). Grüßner
suggests that the respective first syllables are frozen prefixes, whose meaning, however,
is no longer recoverable. In addition to these ‘frozen prefixes’ that according to Grüßner
only occur with nouns, there are other ones that occur with both nominal and verbal
disyllabic roots (see §4.7.3).
There are collective noun roots that indicate reference to a general group of
people, such as ’ ‘the women’ (compare ‘woman, girl’) and ‘the
men’ (compare to ‘married man’).
4.5. Pro-Forms
The pro-forms listed below have the same distribution as noun phrases. In
addition, demonstratives and interrogative pronouns have limited distributional properties
of nouns as well (note that nouns may function as noun phrases), but this is not the case
for personal pronouns. Evidence for this difference comes from the possessive/
166
modification construction, in which both demonstratives and interrogative pronouns may
function as pre-head modifiers with the head marked a- ‘possessive’ (and therefore
behave like other nouns), while personal pronouns have their own possessive prefixes,
thus functioning differently from nouns.
Table 70 shows the forms of personal pronouns. Karbi distinguishes between first
person exclusive and inclusive. The inclusive form requires pluralization via t (see
below) for use as an independent pronoun, but not for use as a possessive prefix. The
exclusive form is the same as the first person singular form, suggesting a shared origin.103
For honorific forms (see §12.4.1) - is added. Possessive forms are prefixes,
which do not have a tonal specification. Honorific possessive prefixes have the same - ,
but are probably best analyzed as disyllabic prefixes rather than a sequence of two
prefixes, which would require us to posit the existence of both a suffix - and a prefix li-
in the Karbi lexicon, when it is clearly the same affix.
Note that the third person possessive has alang(li)- and a- both in square brackets.
This is meant to indicate that there is not actually a dedicated productive third person
possessive prefix in Karbi. Sometimes, alang(li)- is used as a third person possessive
prefix, for example Alanglimen Basapi. ‘Her name is Basapi.’ (with ‘name’).
However, it is not generally used, and instead the a- ‘possessive / modified’ prefix is used,
which receives a third person possessive interpretation from context (and is, in fact
reconstructible to a Proto-Tibeto-Burman third person possessive *a- prefix, see §5.3.1).
For example, consider the possessive clause construction in (117). The possessive clause
103
In the texts of this corpus, and ne- are always glossed as ‘1EXCL’ even if the context makes it clear
that there is a singular rather than a plural reference.
167
construction requires a possessor, a possessed item, and the existential copula (see
§4.6.2.1.1). In (117), only the general possessive prefix a- is acceptable to index the third
person possessor ’ on the possessed item osomar ‘children’.
Plural forms of personal pronouns involve the (bound) plural noun -t (see
§7.6).104 Table 71 gives the example of the first person exclusive independent pronoun
and its possessive prefix counterpart ne- in honorific, plural, and other possessive
constructions.
104
It is impossible to tease apart whether the plural pronoun forms should be phonologically analyzed as a
prefixal or compound construction since all of the pronouns are low tone. Thus, both the phonological
compound interpretation leads to a low-mid tone pattern, e.g., on -t , as does the prefix interpretation
where the toneless prefix ne- would be realized with a low tone before a mid tone syllable (see §3.5.6).
168
4.5.2. Reflexive/Reciprocal Pronouns
Elicited examples that show coreference between A and O are (120) and (121). In
(121), the A argument occurs as a zero anaphora.
169
(121) Coreference between A and O (second person)
[nang- t ( - )] -t =ma?
2:POSS-self POSS-NSUBJ RR-see=Q
‘do (you) see yourself’ [BIK 110205]
Reflexive pronouns are also used to emphatically refer to the subject in Karbi. In
(122), the reflexive pronoun nemethang ‘I myself’ is used with the verb thurdappranglo
‘got up early’ to signal the contrast to the following clause, which states that the subject
also ‘woke up the children’. This contrast is further marked by the use of the additive
particle =t on the corresponding NPs: the subject in the first clause and the object in the
second clause.105
To add yet greater emphasis to the referent of the subject (in contrast to other
referents), the reflexive pronouns - t or - may be combined with a third
reflexive form -m t using a possessive construction, i.e., [REFL.PRO] [a-m t] (following
the possessive construction: [NPOSR] [a-NPOSD], see §7.3), as in (123) and (124).
105
Note that nemethang in (122) is indeed an S argument here and not an O argument, as t ‘wake up’ is
intransitive, compared to transitive in t ‘wake up’ (which, actually, goes against the hypothesis that ing-
has a detransitivizing function, see §4.7.1).
170
banghini atum... o nelimena amatta…
b ng-hin a-t m] o ne-li-men a-m t t
CLF:HUM:PL-two POSS-PL AFF(<Asm) 1EXCL:POSS-HON-self POSS-self=also
‘and then, because of this effort (they make), another few people (are able to go out to
make money)..., I myself also...’ [KaR, SWK 065]
Furthermore note that just like the verb prefix che- (§6.4.3), - t not only
allows a reflexive but also a reciprocal interpretation. In (125), the sense is clearly
reciprocal rather than reflexive, because the people no doubt know their own languages,
they just don’t know each other’s languages.
171
4.5.3. Demonstratives
172
According to Grüßner, the - in the forms used perhaps more frequently as
modifiers is the diminutive suffix - (§5.4.2). However, it is clear that there is no
synchronic diminutive function associated with the use of as opposed to . I do not
know what the - is or what it could be derived from.
The basic forms and frequently occur in place of the third person pronoun
. They follow the personal pronoun pattern of having the plural marker t and the
primary object marker attach directly, as in -t and -t , and - and
- . In addition, occurs in for third person plural inanimate pro-NP
function.
Finally, - , the combination of the demonstrative with the noun phrase
delimiter (§10.5), is used as a pro-form in a co-relative construction (§9.7.3.1.4) in
(127), while Grüßner (1978: 84) also reports this form being used in modifier function.
Grüßner furthermore reports that this form is used with the gender suffixes - and -
(§5.4.1) to mean ‘this woman’ and ‘this man’.
labangke ahoklo
l -ab ng ke a-h k lo
this-NPDL=TOP POSS-truth-RL
‘therefore, whoever can walk over this thread, that one is true’ [CST, HM 096]
Table 73 offers a list of words that contain proximal and/or distal . The
rightmost column lists the combining morpheme(s) where known. Some of the inherently
deictic forms that the demonstratives combine with in this table also occur along with
interrogatives (see §4.5.5).
173
Table 73. Words with demonstrative roots
Function Form Gloss Other morpheme
Deictic Manner la la la ‘this way, like ‘like’, ? ‘say /
this’ quotative’
Time ha ‘(in) the old ‘time’
days’
Place la ‘this side’ ?p
ha ‘that side’
la ‘here’ k ‘here’
‘there’
la tt ‘over there’ natthu ‘direction,
tt towards’
Amount la’ ‘this much’ ‘this much; all’
Discourse connector la ‘therefore, ?si ‘focus’
/ coordinator then’
la ‘and’ =pen ‘with’
106
An often asked question when meeting a familiar person on the road is Konatlo kedam(po)? ‘Where are
(you) going?’, which often gets reduced to simply Na(t)lo?, with the /t/ typically left unpronounced (§3.7.1).
107
For a discussion of the reflexive function of - t, see §4.5.2. Also note that Daai Chin has mat for ‘one’
(So-Hartmann 2009: 128), which might be related.
174
REASON ‘why’ ? ‘what’ + =si ‘FOC’
‘do (v.)’
t t ‘reason’
TIME t ‘when’
MANNER ‘how’ ‘like’
kopu ‘say, quotative’
AMOUNT ’ ‘how much’ - ‘this much; all’ (§7.8.2)
Examples of the other interrogative pronouns and adverbs in the corpus are given
below, in the order of Table 74.
175
(132) Reason interrogative
[…] “Kopisi nang nesopi aphan kipithima?“
o si n ng [ne-oso-p aph n] ke-pV-th ma]
why 2 1EXCL:POSS-child-female NSUBJ NMLZ-CAUS-die=Q
‘[…] "for what possible reason did you kill my daughter?", […]' [RBT, ChM 028]
nangketetroiroidetlo
nang=ke-t t-r i~r i-d t-l
CIS=NMLZ-exit-PL.solid.obj~DIST.PL-PFV-RL
‘if they are my children, how did they come out as pieces of wood?’ [CST, HM 023]
176
‘something, anything’, t- ‘somewhere, anywhere’, etc. An example of t-
‘somebody, anybody’ in a positive clause is (136).
177
co-relative construction (§9.7.3.1.4), as shown in (138), (139), and (140). In (138), the
=ma marked interrogative pronouns functions as a simple relative pronoun, whereas in
(139) and (140), it functions as an indefinite or universal relative pronoun, ‘whoever’ and
‘whatever’, respectively.
lata nangpaklanglo
l t nang pa-kl ng-l
this=ADD:also 1/2:NSUBJ=CAUS-appear-RL
‘while we were going, they also showed us where Patkai College is [SiT, HF 011]
labangke ahoklo
lab ng ke a-h k lo
this=TOP POSS-truth-RL
‘therefore, whoever can walk over this thread, that one is true’ [CST, HM 096]
178
Table 75. Corresponding demonstratives and interrogatives
Function Demonstrative Interrogative Other morpheme
Manner la la ko ‘like’, ? ‘say /
ko quotative”
Place la ko k ‘here’
The pronouns and adverbs of universal quantification, which are used in the
negative indefinite construction (‘nobody’, ‘nothing’, etc.), are listed in Table 76 (for
information on the positive indefinite construction, see §4.5.4.1). Some forms may also
occur in positive clauses (i.e., as ‘all’, ‘everything’, etc.) as indicated in the table.
108
Grüßner (1978: 87) also reports - t ‘over here’ and - t ‘over there’.
179
§7.8.3.1.5). The negative indefinite construction thus consists of a pronoun or adverb of
universal quantification marked by additive =t , followed by a negated verb, as in (141)
and (142).109
Examples for the positive indefinite construction are (143) and (144). Note that in
(143), it is not directly the additive particle that attaches to = , but the longer
form t .
109
Post (2007: 618) calls this a “Dismissive” construction.
180
4.6. Verb Subclasses
Besides the main subclass of verbs denoting actions, states, processes, etc., the
second-largest subclass of verbs are property-concept terms (§4.2). While verbs can be
roughly divided into intransitives, transitives, and ditransitives, these are not salient emic
categories of Karbi verbs. One way to group verbs into subclasses is to compile lists of
all those that may occur with a particular predicate derivation suffix (§6.5), since many
predicate derivations tend to not be fully productive. An example is the predicate
derivation - ‘firmly’, which occurs with the following verbs: ‘tie’, p ‘catch’, t
‘touch, hold’, and t t ‘tie’. In a sense then, semantically rich predicate derivations
classify verbs the same way classifiers classify nouns. There is also a sense in which less
semantically rich verb morphology classifies verbs. For example, cho- ‘auto-
benefactive/malefactive’ only occurs with certain verbs (while, for example, che-
‘reflexive/reciprocal’ appears to occur with any verb). While these are all possible ways
of subdividing the large class of verbs, I limit myself here to only one subsection on
copulas.
4.6.2. Copulas
In addition to a basic existential and locative copula, Karbi also has an additional
copula with a quantifying function.
181
4.6.2.1.1. Positive and Negative
Surprisingly, there are some data that suggest that may additionally function as
an equational copula. One type of data is a construction that nominalizes the main verb
and adds ‘exist’, as shown in (148) (see §9.7.1).
182
(148) ‘exist’ in nominalization construction
[…] kechungkreng dolo, marjeng dolo, lok'hu dolo, lokphlep dolo;
[ke-chungkr ng d -l [marj ng d -l [l k'h d -l ] [l kphl p d -l
NMLZ-be.thin exist-RL be.thin exist-RL be.pale exist-RL be.pale exist-RL
kedothupo
ke-d -th -p ]
NMLZ-exist-again-IRR1
‘[…] they became thin, they became pale, and they were about to die' [CST, RO 022]
lason arjulonghe
[las n arj -l ng he]
that.way hear-GET=AFTERTHOUGHT
‘[…] the women would be the working ones, the men would always play with the
children instead, this is actually what I've heard’ [KaR, SWK 071]
183
lexical verb takes the regular verbal quasi-reduplicative negative suffix (i.e., - ), the
copula has a suppletive negative form, .110 Examples (150) and (151) show in the
locative function and in the possessive construction.
110
There may be other formal/phonological differences that exist despite the surface homophony. For
example, it appears that there is a stress/prominence difference between - meaning either ‘there still
is (more of something)’ or ‘still staying/living (somewhere)’. For the copula, the stress appears to be shifted
to the suffix, whereas for the lexical verb, there is more prominence on the root, as is the case with other
lexical roots when they take - ‘still’. This looks like ongoing grammaticalization of the copula, but
more research is required to provide clear evidence.
184
however, is that the first syllable is the frozen ing- prefix, which in the Karbi song
language typically is replaced by ma- (see §4.7.1), and ma is in fact a very widespread
Tibeto-Burman negative marker that has also been reconstructed to Proto-Tibeto-Burman.
Karbi also has a quantifying existential copula , as shown in (152) and (153).
It is not clear whether also participates in locative or possessive constructions the
way and a do. Note that a suffix - ‘(too) much, many’ has grammaticalized from
this copula.
The copula has also grammaticalized to a verbal suffix - (6.5.5.1) with the
meaning ‘too much’. An apparent cognate of is Daai Chin to:ng ‘too much’ (So-
Hartmann 2009:157).
185
kalilo nesomarlole laho
al -l ] [ne-oso-m r lo le laho]
NEG.EQU.COP-RL 1EXCL:POSS-child-PL-RL=FOC:IRR EXCLAM
‘how strange, they are my children, they are not pieces of wood anymore, they are my
children’ [CST, HM 082]
Karbi has a large number of disyllabic roots with either ing- or ar- as their first
syllable. Among them, there are both verb and noun roots. In compounds, the ing- or ar-
get dropped (see §5.2.1).
186
other Tibeto-Burman languages. A neat piece of evidence also comes from the Karbi
song language (§1.6), which in a considerable number of cases has ma- instead of ing-
(Grüßner 1978: 43).111
On verbs, ing- has been suggested to have an intransitive or medial function
(Grüßner 1978: 42), which is also the proposal for the function of PTB *m- (also
including related functions such as “durative” and “reflexive”) (Benedict 1972: 117 ff.;
Matisoff 2003: 117 ff.).112 Benedict furthermore suggests that *m- on verbs is the same
element as the *m- that occurs on nouns (p.118), where the function on nouns has to do
with inalienable possession, as evidenced by their occurrence on body part or kinship
terms.
In modern Karbi, a number of verb and noun roots with ing- fit into this account:
there are a number of ing- verb roots with intransitive113 or medial function, and there are
a number of ing- noun roots which represent body part and kinship terms. These are
recorded in the above mentioned sources, and Grüßner additionally lists nine ing- roots
with corresponding ing-less monosyllabic counterparts. Examples of ing- in intransitive
and medial verbs include ‘to float’ (perhaps related to ‘water’), ‘to
break (intransitive)’ (compared to ‘to break (transitive)’), and ‘to run
(animals), gallop’. Other examples that could be added in support of the hypothesis that
/ing/ in Karbi still retains a semantic association with middle voice and intransitivity are
t ‘to dawn’, ‘to get intoxicated’, ‘to dissolve’, ‘to fly’,
‘to look proper, well prepared’, ‘be hungry’, or ‘to love’, among others.
There are, however, plenty of other ing- roots that do not obviously fit in, or even
have opposite functions. Nouns that are not body/kinship terms or otherwise
understandable as a part of a whole include, for example, ‘mud’, ‘silkworm’,
‘twenty’, ‘hill’, ‘forest’, ‘elephant’, ‘species of grass’,
111
According to Grüßner, this is only the case in verb roots. The examples he gives are ~
‘call (small animals)’, t ~ t ‘be dark’, and ~ ‘smell’.
112
Matisoff (2003:117) further assumes that the semantics of this prefix as it goes on verbs have to do with
signaling “inner-directed states or actions, including ‘middle voice’ notions like stativity, intransitivity,
durativity, reflexivity.”
113
Note that Daai Chin (Southern Kuki-Chin) still has a productive intransitive velar nasal ng- prefix with
reciprocal and reflexive functions (So-Hartmann 2009: 202 ff.).
187
t ‘sign’, t ‘salt’, or t ‘big bamboo basket’. Sample ing- verbs that are
problematic for a synchronic intransitivity account include ‘push’, ‘carry a
load’, ‘open (something)’, ‘separate’, t ‘go and bring’, or t
‘wake (somebody) up’. This last root t is particularly problematic, because it has a
monosyllabic counterpart t , which has the intransitive sense of ‘wake up’, and thus
seems like it is backward by having the root with ing- be transitive, and the root without
ing- intransitive.
The frozen prefix ar- is the other highly frequent element besides ing- in Karbi. It
is a reflex of a Proto-Tibeto-Burman *r- prefix (Wolfenden 1929: 43-44; Benedict 1972:
109; Matisoff 2003: 127). Although Wolfenden suggests *r- is a “directive” prefix,
Benedict and Matisoff remain agnostic about its function. Likewise in Karbi it appears
impossible to find anything all the noun and verb roots with ar- have in common.
Note that like ing-, the ar- prefix has a different form in the Karbi song language,
where it changes to ra- or ru-, as in the pairs ~ ‘god’ and ~ ‘rain’
(Grüßner 1978: 43).
4.7.3. Other Possible Frozen Prefixes in Disyllabic Noun and Verb Roots
In addition to ing- and ar-, Grüßner (1978: 43) reports tV-, the-, pi-, phe-, se-, and
he- as possible frozen prefixes that occur as the first syllable in disyllabic noun and verb
roots.
4.8.1. Adverbs
A list of day ordinals with respective forms for the day as a whole, only the
morning, and only the evening/night are given in Table 77.
Note that unique expressions for ‘whole day’ terms are limited to today,
tomorrow, and yesterday. For two days into the future or past, the suffix -dik is added,
while further days into the future or past just add the independent numerals that count
how many more days are added ( for ‘one’). The ‘morning’ terms all have the word
‘morning’ in them: as compounds in terms for ‘tomorrow’ and ‘yesterday’, in a
syntactic possessive construction for further days into the future or past, and only the
189
word for ‘this morning’ has just a prefix along with ‘morning. The ‘evening/night’
terms pattern the same as the ‘morning’ terms except for ‘last night’, which also has a
simple disyllabic stem that consists of a prefix along with the second syllable of
‘evening/night’ (in fact, this form is even more reduced then the form for ‘tonight’).
In order to compare the prefixes, consider Table 78, which also includes terms for
year ordinals.
What is quite striking is that although we get three basic prefixes: pV-, tV-, and
mo- (ignoring chu-, which only occurs in the form for ‘last year’), they do not align the
way we might have expected. pV- occurs in all ‘present’ forms, but it additionally occurs
in the elongated expression patumi ‘yesterday’ as well as in ‘tomorrow’. tV- occurs
in all ‘1 before’ forms (except for ‘last year’), but additionally also in todap ‘this
morning’. Finally, mo- occurs in ‘1 after’ forms as well as in mo- ‘tonight’. A
reasonable hypothesis to explain two of the unexpected forms, todap ‘this morning’ and
- ‘tonight’, is that they are typically used around midday, when ‘this morning’
is in the past and ‘tonight’ in the future, which would explain the tV- and the mo- prefix if
we assume they originally mark past and future, respectively. Now, this turns out to be
not correct. For ‘this morning’, either to-dap or pa-dap may be used interchangeably, no
matter whether it is 4:00AM and somebody talks about plans to do at 7:00AM, or if it is
already 4:00PM, and they talk about something that happened at 7:00AM. The same
turns out to be true for pe- and mo- for ‘tonight’, which are used
irrespectively of whether the reference is to the past or the future. Nevertheless, it seems
likely that this is how the forms originated.
The other unexpected forms are the forms for ‘today’, ‘tomorrow’, and
‘yesterday’, in that all three of them contain pV-. Here, it seems likely that pV- was
190
reinterpreted as an element that means ‘day’ rather than indicating present, past, or future,
and that that is the reason why it got extended to the words ‘tomorrow’ and ‘yesterday’.
In addition to day and year ordinals, there are a few other temporal adverbs, such
as ‘now’, t ‘right then’, or relator nouns that are used as adverbs like
‘before, earlier’ or ‘later, afterwards’ (see §4.4.4.7.2). Deictic, interrogative, and
universal quantifier temporal adverbs are listed in §4.5.3, §4.5.4, and §4.5.6, respectively.
There is a subclass of adverbial temporal nouns that take the suffix - ‘every’:
e.g., - ‘each year’ and - ‘each day’. It is likely that this -
suffix is related to the verb ‘come’, considering functionally similar expressions in,
for example, English such as ‘in the coming years’ to mean ‘in future years’. Also note
the possibly related verb suffix - ‘plural:S/A’ (§6.5.5.1).
(156) shows that the deictic temporal adverb ‘(at) that time’ may be
reduplicated as for the purpose of intensification.
(156) Reduplication
<ne hala> hakokota so'arlo atumke hem
<n h la> [[h k ~k t s 'arl a-t m ke h m
1EXCL that that.time~INTENS=ADD:EXH women:COLL POSS-PL=TOP house
191
4.8.1.2. Locative Adverbs
Other adverbs, i.e., those with neither temporal nor locative meaning, include
t ‘anyway’ (with the components a- ‘POSS’, t ‘type(<Asm), ? ‘now’);
‘more’ (perhaps from a- ‘POSS’, ke- ‘NMLZ’, ? ‘reach’); ‘more(<Asm)’, and
furthermore reduplicated forms such as ‘suddenly’ and ‘in.a.hurry’.
There are also several words for ‘like this’, such as , , etc. (see §4.5.5).
4.8.2. Numerals
Numerals from ‘one’ to ‘ten’ are listed in Table 79. For numerals ‘one’, ‘two’,
and ‘three’, there exist bound forms that occur with classifiers (see §4.4.1). Note that the
forms for ‘seven’, ‘eight’, and ‘nine’ are morphologically complex and are based on
addition (in the case of ‘seven’) or subtraction (‘eight’ and ‘nine’).
Particularly the forms of ‘eight’ and ‘nine’ are interesting in the Tibeto-Burman
context, as this uncommon subtractive pattern also exists in Meithei. There the form for
192
‘eight’ is nipan and has the internal structure ni ‘two (< əni)’ + pan ‘subtract’, and
similarly the form for ‘nine’ is məpan, which is mə ‘one (< əmə)’ + pan ‘subtract’
(Chelliah 1997: 85). Since this is a highly unusual pattern that so far has not been attested
in other TB languages, neither genetic inheritance nor parallel innovation are likely
explanations, but instead contact is the probable source. How exactly a contact scenario
between Karbi and Meithei has to be modeled is, however, currently not clear (§1.3).
Table 80 gives a list of numerals over ‘ten’. The interesting ones, again (and
perhaps unsurprisingly so), are throknerkep ‘eighty’ and throksirkep ‘ninety’. Their
respective internal structures may be interpreted in two different ways, as done by
Grüßner and Joseph. According to Grüßner, throk “has taken over the function of
indicating the ‘ten’”, as a result of which the words would be parsed as throk-nerkep and
throk-sirkep. This seems a bit bizarre given that t is the word for ‘six’.
Joseph offers a different account that tries to hold on to t meaning ‘six’. In
order for that to be the case, he parses throkne-r-kep so that the -ni ‘two’ goes with t ,
yielding ‘six’ plus ‘two’ times ‘ten’. While this is very sensical, for throksirkep ‘ninety’,
he has to pose a non-surfacing -ni ‘two’ so that things add up correctly: ‘six’ plus ‘one’
plus ‘two’ times ‘ten’. This seems a bit speculative; an alternative is to propose that
throksirkep was formed in analogy to throknerkep.
A list of coordinators that are currently attested is given in Table 82, along with
an indication of whether a particular form coordinates NPs or predicates/clauses. Note
that Grüßner (1978: 128) additionally lists as coordinators: - - ‘for (causal),
therefore’, t - - ‘for (causal), therefore’, and - ‘if(<Asm)’.
114
St t ‘although’ almost certainly is a combination of -si ‘NF:RL’ and =t ‘ADD’. See §7.8.3.1.4.
194
Examples of the coordinators listed in Table 82 are given below.
195
(162) t ‘otherwise, or’ coordinating clauses
lapen latum haladak... la botor bokan areng
lap n l -t m h lad k [la bot r bok n a-r ng
and.then this-PL there this climate(<Asm) EE:bot r POSS-skin
kangsampik apotlone
ke-ings m-p k] a-p t-l -n
NMLZ-be.cold-very POSS-reason-RL-INDEF
‘and then, they... maybe the reason was that the botor bokan, climate in English, was
good or the reason was seeing the Nagas, (but) everybody was exceedingly happy’ [SiT,
HF 036]
196
CHAPTER V
5. NOMINAL MORPHOLOGY
This chapter discusses nominal word formation. There are very few
morphological constructions that Karbi nouns may occur in. Cross-linguistically common
categories of productive nominal morphology such as role (or case) marking and plural
do not exist as such: While there are two role markers (i.e., - , §10.6.2, and - ,
§10.6.3), as well as a plural marker (i.e., -t , see §7.6), these are syntactic and not
morphological constructions. They represent transparent grammaticalizations that have
arisen out of the possessive construction, as evidenced by their initial a- (i.e., the
possessive/modified a- prefix, see §5.3.1 below as well as §7.3).
Although this chapter only lists a handful of prefixes and a handful of suffixes,
there are some additional elements that may modify noun stems (e.g., =pen ‘with, from’
or =t ‘additive’), which are, however, best considered clitics and therefore discussed in
Chapter VII ‘The Noun Phrase’.
This chapter begins with an overview of the morphology of noun stems and noun
words (§5.1). Next, two types of compounding are discussed in §5.2. In §5.3 and §5.4, an
overview of nominal prefixes and nominal suffixes, respectively, is offered. Lastly, §5.5
briefly discusses functions of noun stem reduplicative morphology.
Nouns are morphologically not very complex in Karbi. There are only two
derivational suffix slots and one inflectional prefix slot.
Generally, a root can function as a stem without being modified and can function
as a word without being modified. However, a noun stem can also be complex in that it is
either a compound (§5.2) or in carrying a derivational suffix. Nominal derivational
suffixes are not frequent in the corpus, but the gender/size suffixes - and - (§5.4.1)
197
may occur in a first suffix slot, and either the human plural suffix - (§5.4.4) or a
reduplicative suffix (§5.5) may occur in a second suffix slot, as illustrated in (163).115
Noun stems can function as noun words without additional morphology, but can
also occur with one of two prefixes to function as noun words: with the general
possessive/modified prefix a- (§5.3.1) or with one of the personal possessive prefixes
(§5.3.2).
5.2. Compounding
The sections below discuss two types of compounds in Karbi. The first falls under
the traditional notion of compounds, where one noun root is combined with either another
noun root or a verb root into a new noun stem. The second type is more specific to the
South/Southeast Asian context and is a type of ‘elaborate expression’ construction.
New noun stems may be formed by compounding a noun root with another root.
Examples of noun-noun and noun-verb compounds are given in Table 83. For further
examples of compounds, see Grüßner (1978: 54-5).
Compounds provide evidence for the fact that frozen prefixes such as ing- and ar-
are peripheral to the core semantic part of disyllabic roots, as they get dropped in
compounds like - ( < ), an- ( < ), and me-t ( <
). For further examples of this productive process that drops frozen prefixes in
compounding (as well as the first syllables of disyllabic roots in other cases), see Grüßner
(1978: 35-6).
115
It is also possible to add the honorific suffix - to nouns that have the human plural suffix - ,
apparently to form vocative forms, see §12.4.1.
198
Table 83. Noun-noun and noun-verb compounds
Type Compound Meaning Root 1 Gloss Root 2 Gloss
N-N ‘molasses’ ‘sugarcane’ ‘water’
t ‘banana (fruit)’ ‘banana leaf’ t ‘fruit’
‘house owner’ ‘house’ ‘grandfather’
N-V ‘long-bearded ‘beard’ ‘be long’
ones, muslims’
‘elder’ ‘head’ ‘be ripe’
‘elder’ ‘head’ ‘be white’
‘cold rice’ ‘rice’ ‘be cold’
t ‘dog’ ‘tail’ t ‘to cut’
‘cloth to carry ‘cloth’ ‘carry baby
baby on back’ on back’
chain ‘cow’ ‘mithun’ ‘plow’
(Grüßner
dictionary ms.)
The following subsections discuss the nominal prefixes of Karbi: the general
possessive or ‘modified’ prefix a-, the set of four personal possessive prefixes (including
their systematically derived honorific forms), and the verbalizer pe- ~ pa-.
200
Examples (166) and (167) show that a classifier-numeral modifier only triggers
the a- prefix on its head noun if it precedes its head noun, as in -t ‘POSS-tiger’ in
(166), but not if it follows it, as in ‘hen’ in (167).
116
Benedict gives the Aimol example rul ə-rmai ‘snake POSS-tail’, which is cognate in all three elements
with the same NP in Karbi: phu - ‘snake POSS-tail’, and the Bahing examples “ ŋ əta-mi ‘calf’
(cow its-child), byar əpwaku ‘sugar-cane’ (cane its-juice).”
201
explanation. One set of occurrences is exemplified by (168), where ‘elephant’ is
marked with a- without there being a modifier and without a- indicating third person
possession. The reason why a- is used here is that this is a reduction of - ‘this
POSS-elephant’ according to my consultants, who explain that this kind of reduction
occurs as part of regular colloquial speech. Remember that a noun is regularly marked
possessive via a- if modified by a demonstrative.
202
discussion of preposed PCT modifiers, see §7.7.1.2). The reason why this may be the
same type of occurrence of a- is because in this case of preposed PCT modification, the
underlying construction may be a possessive construction (§7.5.2), in which the preposed
PCT modifier is nominalized in order to act as the (nominal) possessor or modifier.
(170) Pre-head PCT-based modifier ‘be new’, marked with a-ke- ‘POSS-NMLZ-‘
lasonthotsi Lindata ketheklongpen aning kerong,
[las n-th t si L nd t ke-thekl ng-p n] [a-n ng ke-ar ng]]
that.way-exactly=FOC PN=also NMLZ-see-NF POSS-mind NMLZ-be.happy
Personal possessive prefixes are discussed along with the independent pronouns
they derive from in §4.5.1; Table 84 repeats the forms (with the optional honorific marker
- in parentheses). Note that first person exclusive ne(li)- is also used for first person
singular reference.
The verbalizer pe- ~ pa- (for the allomorphy, see §3.9.2.1) is related to (or
arguably the same morpheme as) the causative pe- ~ pa- prefix (§6.4.2). Since nouns can
function as predicates without additional marking (§4.3), the use of pe- ~ pa- typically
has a causative, or perhaps change-of-state, implication ‘make X (be) [N]’. In (171), for
example, the literal meaning is ‘make the backside (be) the front’, i.e., ‘turn around’.
203
(171) Verbalizer pe- ~ pa- on noun root ‘front’
[…] aphi anatsi chepangnophit
a-ph a-n t si che-pa-ingn -ph t
POSS-backside POSS-direction=FOC RR-VBLZ-front-right.away
‘[…] (he) turned pointing back again right away’ [HI, BPh 006]
nangparisomardunlonglang
nang=pa-ris -m r-d n-l ng-l ng]
CIS=VBLZ-unmarried.boy-PL-JOIN-GET-yet
‘they know how to go on a good road up to a high degree (metaphorically referring to
knowing how to do things properly, how to keep everything clean, etc.), and, because
they know how to keep everything clean and nice, those people back then up until today,
get to stay even more like girls and boys (i.e., young) than we do’ [SiH, CW 017]
204
In (173), pa- occurs on t ‘fruit’ to mean ‘to bear fruit’, and Grüßner also records
-t ‘VBLZ-egg > to lay eggs’. These examples suggest that pe- ~ pa- may also function
to express ‘cause [N] to exist’. Further research will need to investigate the factors that
determine the function of pe- ~ pa- on a particular noun stem (which may turn out to be
largely pragmatic in nature).
The discussion of nominal suffixes starts with the gender suffixes - and - ,
which have further functions as augmentative and modifier-deriving markers (§5.4.1). In
§5.4.2, information on the diminutive - is offered, and §5.4.3 discusses the occurrence
of the diminutive and the two gender and augmentative/modifier-deriving suffixes in the
names of animal/plant subspecies. In §5.4.4, the plural suffix - , which occurs on
nouns with human referents, is shown, and §5.4.5 offers an account of the nominal
superlative suffix - .
Karbi has two gender suffixes, - ‘female’ and - ‘male’. There are a number of
nouns with human referents that may be specified for gender via one of the two suffixes,
for example - ‘daughter’ and - ‘son’, - ‘female friend’ and - ‘male
friend’, or - ‘married woman’ and - ‘married man’.117 ‘Female’ - also occurs
productively with animal referents, whereas ‘male’ - is often replaced by - (which is
117
Note that - is often replaced by the word - in colloquial speech, where the second part
is a borrowing from Assamese. There is no directly analogous female form, but instead a form with both
suffixes, i.e., - - , is used in addition to - . The male form - is still used in ceremonial
speech.
205
commonly used as a noun rather than a suffix, e.g., in - ‘chicken POSS-male >
rooster’). Below I discuss each suffix separately.
The ‘female’ suffix - occurs in several places, where - is not or not typically
used for the male counterpart. One example is clan names. Children carry the clan name
of their father, for example or . But while the clan names of sons (or men) do
not require the ‘male’ suffix, the clan names of the daughters (or women) do, so the
daughter would be called or , but the son would be or , just like
his father.
‘Female’ - is also used on nouns with animal referents, for example -
‘chicken-female > hen’. The lexical noun - still exists, so it is also possible to say -
‘chicken POSS-female > hen’ or - ‘goat POSS-female > female goat’. The noun -
‘female (person/animal)’ is likely related to ‘mother’.
While a form - for male referents, or grammaticalized as a nominalizer, exists
all across Tibeto-Burman, ‘female’ - is not very common. Interestingly, however,
Meithei also has a ‘female’ -pi suffix as in nu-pi ‘female human’ or hənu-bi ‘old woman’
(Chelliah 1997: 244). The corresponding ‘male’ suffix in Meithei is -pa. The shared -pi
suffix may be part of a set of items that occur in both Karbi and Meithei due to language
contact (§1.3).
Polysemically related to the ‘female’ marking function, - also has an
augmentative function. Although it may be cross-linguistically more common to
polysemically relate the male gender with augmentative marking, the ‘female-
augmentative’ link has a parallel in Galo (Tani, Tibeto-Burman), where the ‘female’
suffix -nə also means ‘big’ (for example dog-nə ‘large stone bead at ornament base’ or lə-
nə ‘big toe’ (Post 2007: 873)). Examples of Karbi ‘female’ - as an augmentative are
t - ‘vast plains’, - ‘large mountain’, or the name of a river, (lit.
‘water-AUGMENT’).
The ‘female, augmentative’ - occurs on PCT roots, arguably in its augmentative
function; this is discussed below in §5.4.1.3.
206
5.4.1.2. Male (and Non-Productive Modifier-Deriving) -
The male suffix - is related to the lexical noun - ‘father’, a common root in
Tibeto-Burman with the same semantics of ‘male’ or ‘father’. Examples are -
‘wise old man (borrowed from Indic Brahmin)’ (which also has a female counterpart,
- ), or - , a character in folk stories whose stupidity results in entertaining,
funny stories.
‘Male’ - alternates with ‘female’ - as augmentatives in the noun stem ‘thumb’,
which may be either - or - . This appears to be an exception, however;
generally, only - is used as an augmentative suffix on noun stems.
There are several occurrences that suggest that - has historically derived
modifiers from PCT roots in a synchronically no longer productive way. This is
interesting considering that -pa/-po nominalizers are found in a large number of Tibeto-
Burman languages from different branches of the family (LaPolla 2008: 52). Examples
are (174) and (175), which come from folk stories. (174) shows that the elephant in the
story is characterized as t - ‘deaf’, where t is a compound of - ‘ear’
and t ‘deaf’.
207
(175) - in modifiers - ‘black’ and t - ‘big’
vo ikpo vo thepo do
[v k- ] [v th - ] d
bird be.black-MODIF bird be.big-MODIF exist
‘[…] there is a big black bird’ [SeT, MTN 018]
Examples (176) and (177) demonstrate the equivalent uses of - and - , as they
occur on the PCT root t ‘be big’ to form a modifier of the preceding head noun (see
also §4.2.6).
The fact that - also appears to derive a modifier from the PCT root t ‘be big’
in (177) - just like - does in (176) - is surprising. It is the only instance in the corpus of
- occurring on a PCT root rather than on a noun. One possibility is to consider t -t
‘big sea’ a compound noun stem (parallel to, for example, - ‘long bearded
(person)’, from ‘beard’ and ‘be long’). However, t -t does not occur
without - , so this is not an appealing explanation. A perhaps better interpretation is that
- , possibly in analogy with - , may also function to derive modifiers from PCT roots,
with additional augmentative semantics.
208
5.4.2. Diminutive -
There is only one occurrence of the diminutive suffix - in the corpus, which is
given in (178).
Grüßner (1978: 55) also records other occurrences of the diminutive, such as
- ‘little leaf’ or - ‘small pond’. Another example is (179), where - occurs on
a classifier. It appears that - is used productively to derive diminutive stems of nominal
elements (i.e., nouns and classifiers).
(179) - - -
POSS-CLF:cylindrical-DIM-SPLT
‘the smallest (loaf of bread)’ (Grüßner 1978: 107)
A number of proper nouns that refer to animal (and some plant) subspecies end in
the gender/augmentative suffixes - or - , or the diminutive suffix - . Table 85 offers
a list of examples attested in the corpus (and Grüßner provides more examples in his
dictionary manuscript).
209
Table 85. Proper nouns with - , - , or - suffix
Proper nouns with - Proper nouns with - Proper nouns with -
- ‘pig.sp’ - ‘ant.sp’ - ‘frog.sp’
- ‘bird.sp’ t - ‘dog.sp’
- ‘squirrel.sp’
- ‘plant.sp’
The plural suffix - only occurs on nouns with human referents, e.g., -
‘unmarried boys’, - ‘children’, - ‘married women’, - ‘married
men’. The more general plural marking strategy is a possessive construction involving
the ‘plural noun’ -t (§7.6). As seen in (180), nouns that already carry - in addition
often occur in the -t plural construction.
langdunveretlo...
l ng-d n-ver t-l ...
see-JOIN-INTENS-RL
‘and then, he (the old man) keeps watching the fruits that these children are each
holding, […]’ [SiT, PS 045]
210
(181) Plural -
[…] nangparisomardunlonglang
nang=pa-ris -m r-d n-l ng-l ng]
CIS=VBLZ-unmarried.boy-PL-JOIN-GET-yet
‘[…] (they) get to stay even more like [girls and] boys (i.e., young) than we do’ [SiH,
CW 017]
5.4.5. Superlative -
(183) shows that - may also occur on adverbs - here, an adverb derived from a
relator noun.118
118
It might be worth double-checking that here cannot instead be interpreted as a noun with the
meaning ‘the later/last one’.
211
(183) Superlative - on ‘after, last’
[…] si aphi abangke aphisi vanglo, mh
s a-ph ab ng ke aph -s v ng-l mh
and.then POSS-grandmother NPDL=TOP afterwards-SPLT come-RL DSM
‘[…] and then, the grandmother was the last one who came’ [KK, BMS 074]
5.5. Reduplication
Reduplication occurs more frequently and has more different functions on verb
stems (§6.6) than on noun stems. Nevertheless, there are some cases where reduplication
of noun stems occurs. So far dual and (distributive) plural functions of reduplicated noun
stems have been attested. Note that reduplication of noun and verb stems consists of the
reduplication of the last syllable of the stem, with scope typically over the whole stem,
but see §6.5.1.1.1 with scope differences in verbs.
5.5.1. Dual
Reduplication of the last syllable of a noun stem may indicate duality, as shown in
(184), where refers to ‘two friends’. A plural interpretation (with a different
context) is not acceptable.
Also note (185) from an on-line narration of the pear story (i.e., the speaker is
describing what is going on as he is watching the video clip). Here, the reduplicative dual
form arlososo from the noun arloso ‘woman/girl’ does not mean that there are two girls,
but instead refers to the duality of the boy on the bike colliding with the girl on the bike,
i.e., the duality of the girl and the boy.
212
chetongder amat aphutup klophit
che-t ng-d r am t a-phut p kl -ph t
RR-collide-sound and.then POSS-hat fall-right.away
‘the bicycle..., with the girl he collided and then his hat fell down’ [SiT, PS 028]
In (187), heihai with its sense of ‘different ones’ occurs in a general extender
construction (see §12.2.1).
213
(187) heihai in general extender construction
Europe lapen kaprek kaprek adet
Europe lap n ke-pr k ke-pr k a-d t
Europe(<Eng) and NMLZ-be.different NMLZ-be.different POSS-country
214
CHAPTER VI
6. VERBAL MORPHOLOGY
This chapter discusses verbal word formation. §6.1 offers a short note on
compound verbs. In §6.2, an overview of approximate position classes in the Karbi verb
is provided. The remainder of the chapter follows the order of position classes in the verb
in discussing the respective morphemes involved in verbal word formation, see §6.2 for
the organization of this chapter.
6.1. Compounding
Just like in nominal word formation (§5.2.2), compound verb stems can be
formed based on elaborate expression root pairs (§12.2.2.2). For example, the two roots
t ‘cook’ and ‘put on stove’ form an elaborate expression meaning ‘cook’ in (190).
tun-dangpikang'et nangjilang
t n-d ng-p -k ng- t n ng-j -l ng
cook-put.on.stove-BEN/MAL-leaving-PRF need-IRR2-still
'early in the morning, over there, the women have to cook and do all the household works
from before dawn also for those who stay at home (besides food for themselves)' [KaR,
SWK 074]
Figure 14 lays out approximate position classes in the Karbi verb. There are four
pre-root slots, and six post-root slots. Slot 6 consists of predicate derivations, more than
one of which can occur in a given verb. Note that all pre- and post-root slots are entirely
215
optional: the bare stem (consisting of just the root) may function as the predicate all by
itself.
This chapter discusses these verbal affixes including the proclitics in the order
they occur in as shown in Figure 14. Specifically, this means that the next section deals
with the proclitics (§6.3); followed by the prefixal derivational morphology of slot 2 as
well as the scope-sensitive slots 3 and 4 (§6.4); then the large class of suffixal predicate
derivations in slot 6 (§6.5); followed by functions of reduplication in slot 7 (§6.6); then
the onset-reduplicative negative suffix - in slot 8 (§6.7). Finally, the more inflectional
suffixes are in the remaining four sections. Slot 9 is an ‘aspect’ category (§6.8), whereas
slot 10 contains four different types of suffixes: mood marking suffixes (§6.9);
subordinate marking suffixes (§6.10); one suffix marking aspect (§6.11); as well as non-
declarative speech act suffixes (§6.12).
1 2 3+4 5 6 7 8 9 10
nang= ke- che- pa- ROOT DERIV* RDPL - - t -
ne= pa- che- - -
e= cho- pa- -
(pa- cho-) -si
cho- -ra
che- -pen
pa- -
-
…
216
6.2.1. Derivational and Inflectional Affixes, and the Verb Stem
The elements that occur in pre-root slots are limited to the ones given in Figure 14
above, which include three proclitics (cross-referencing non-subject speech act
participants and marking the cislocative), the nominalizer ke- (and progressive ke-), and
two slots for argument structure influencing prefixes with scope-sensitive ordering
options: pa- ‘causative’, che- ‘reflexive/reciprocal’, and cho- ‘auto-
benefactive/malefactive’. Below I provide examples for the ordering possibilities
between the causative prefix on the one hand, and the reflexive/reciprocal and auto-
217
benefactive/malefactive prefixes on the other hand; the reflexive/reciprocal and auto-
benefactive/malefactive prefixes cannot co-occur on the same verb.
First, (191) offers a verb form with all prefix slots and the proclitic slot filled:
nang= cross-referencing the first person causee (or the first/second person possessor, see
§6.3.1.3), the nominalizer ka-, the causative pa-, and the reflexive/reciprocal che-. In this
verb, the causative precedes the reflexive/reciprocal. This results in the causative having
scope over the stem consisting of the reflexive/reciprocal-marked root, i.e., -[ -t ]
‘cause (somebody) [to meet (somebody)]’.
(191) Verb form with all pre-root slots filled (causative preceding
reflexive/reciprocal)
[…] “nephi aphan nangkapachetongji pule,
[ne-ph aph n [nang=]1 [ke-]2 [pa-]3 [che-]4 t ng-j pu=le
1EXCL-grandmother NSUBJ 1/2:NSUBJ=NMLZ-CAUS-RR-meet-IRR2 QUOT=FOC:IRR
nangpachetongte” […]
nang pa-che-t ng-C
1/2:NSUBJ=CAUS-RR-meet-NEG
‘[…] “you said you will make/help me meet my (should be 'your') grandmother, but you
didn't make/help me meet her”, (said bamonpo) […]’ [KK, BMS 073]
218
akhai ason mane undunjima pusi
a-kh i as n mane n-d n-j ma pusi
POSS-community like I.mean(<Asm) be.able-JOIN-IRR2=Q QUOT.COMP
nelita kamatha
n -l t kV-math
1EXCL-HON=ADD NMLZ-think
‘so the men and women, or the mothers and fathers, if they make themselves equal for
the children, if they take this their own responsibility, then like other peoples' tribes we
will be successful (on a par with other peoples' tribes), is what I think’ [KaR, SWK 052]
nangji
n ng-j
must-IRR2
‘and then, people need to stay and live in a clean environment’ [SiH, CW 001]
219
kachopethepen <abahere> angparke pirthe
kV-cho-pe-th -pen <a-baher > anp r ke] [pirth
NMLZ-AUTO.BEN/MAL-CAUS-be.big-NF:with POSS-beyond(<Asm) besides=TOP world
The structure of post-root slots is more complex than that of pre-root slots. This is
already obvious from the fact that there is a hugely greater number of suffixes than
prefixes.
There are co-occurrence restrictions between slots 8 and 10 that cannot be
represented in a position class diagram such as Figure 14 (repeated here with the post-
roots slots only, as Figure 15). They are discussed in §6.2.3.1. Further complications exist
with respect to - in slot 9 (§6.2.3.2), as well as within slot 10, as - and - appear to
co-occur. This is discussed in §6.2.3.3.
220
5 6 7 8 9 10
ROOT DERIV* RDPL - - t -
- -l
-si
-
-
…
” { }
ko j rp pu mm
buddy:VOC friend QUOT AFF
‘”and there won't be anything (any difficulties, problems, dangers), my friend?”’ [HK,
TR 140]
221
6.2.3.1. Cooccurrence Restriction between Negative - and Irrealis -j and - (Slots
8 and 10)
The negative suffix - may only co-occur with the irrealis suffixes - and - if
perfective - t intervenes. Therefore, t- - t- = ‘enter-NEG-PFV-IRR2=Q’ (RBT,
ChM 042) and - - t- ‘be.fine-NEG-PFV-IRR1’ (SH, CSM 049) are perfectly
acceptable. However, * - - ‘*go-NEG-IRR2’ and * - - ‘*go-NEG-IRR1’ are
unacceptable.
The experiential suffix - only occurs with - ‘realis’ of slot 10, but no other
slot 10 suffixes. It also has mostly been found to occur in conjunction with negative -
in the sense of ‘have never V-ed’. For more details, see §6.8.2.
The proclitic slot on the verb consists of nang= indexing speech act participants
in non-subject roles (as well as, in some dialects, ne= < ‘1EXCL’ and e= < e- ‘1INCL’),
and cislocative nang= (defined as a marker of motion towards a reference point). Based
on the different functions, we may synchronically consider the person cross-referencing
nang= a different morpheme from the cislocative marking nang=. Historically, however,
all the evidence suggests that there is just one nang= proclitic, which likely goes back to
the second person pronoun . This is further discussed in the summary §6.3.3.
This section discusses the function of nang= (and ne= and e=) to cross-reference
non-subject speech act participants (SAPs). In §6.3.1.2, an overview is provided of the
frequent cross-referencing of SAPs that function as primary objects, and §6.3.1.3 offers
examples of instances where SAPs in other, non-core, roles are cross-referenced; §6.3.1.4
summarizes this section.
223
6.3.1.1. Introduction
In (198) and (199), nang= occurs on verbs that mark events in which a first
person A argument acts on a second person O argument, which may be both singular as
in (198), or plural as in (199). The second person O argument may occur as an
independent pronoun in addition to the preverbal clitic, as in nangphanke in (198), or
only in the form of the preverbal clitic (i.e., with the independent pronoun being a zero
anaphor in the clause), as in (199).
224
(199) First person acting on second person (1>2)
[…] nangkecharjulo, peipen po
nang=ke-che-arj -l p i pen p
1/2:NSUBJ=NMLZ-RR-ask-RL mother=with father
‘[…] we are asking you, mother and father’ [CST, HM 117]
(200) shows that nang= also occurs when there is a third person acting on a
second person. The second part of this example also shows, however, that this cross-
referencing via nang= is not obligatory. Although the two clauses ([[nangphan
nangkelang] inut] donangji ‘there needs to be somebody to look after you’ and [[kevan
kepon] inut] donangji ‘there needs to be somebody to bring you and to take you’) are
parallel in their structure and in their reference to the second person, nang= is only used
on the verb in the first clause and not the second.
225
po pu nang “
p pu nang=ke-h ng a-b ng ke-d -l pu
father QUOT 1/2:NSUBJ=NMLZ-call POSS-CLF:HUM:PL NMLZ-exist-RL QUOT
‘and then, Bokolapo didn't know that it was his own home, and then Bokolapo, “how
strange!, who is here to call me 'father'?”’ [HI, BPh 012]
(203) shows that nang= also cross-references the standard of comparison in the
comparative construction, which is also indicated by the - ‘non-subject’ marked
independent first person exclusive pronoun, - .
(204) First person inclusive primary object (3>1INCL) marked with e= ‘1INCL:NSUBJ’
[…] itum aphanke ha nampi namdur alongsi
e-t m aph n ke h n mp n md r al ng si
1INCL-PL NSUBJ=TOP over.there big.forest EE:namp LOC=FOC
ekethondamlo
e=ke-th n-d m-l
1INCL:NSUBJ =NMLZ-drop-GO-RL
‘[…] (the witch) abandoned us over there in the deep forest’ [CST, HM 076]
226
(205) First person exclusive primary object (2>1EXCL) marked with ne=
‘1EXCL:NSUBJ’
mh nephan nechiriphetnoi
mh ne- h n ne=che-r p-h t-n i
DSM 1EXCL-NSUBJ 1EXCL:NSUBJ=RR-hold.firmly-firmly-INFRML.COND.IMP
‘”hold (yourself) firmly onto me”’ [KK, BMS 046]
Although it may be dialectal variation that underlies the preference to use ne= or
e= to cross-reference first person non-subjects instead of nang=, there is an even more
interesting sociolinguistic dimension to this, as even the same speaker, in the same text,
and even in the same utterance, may switch between using e= or nang= for a first person
inclusive O argument. This is shown in (206), where the beginning of the utterance,
ethapkangdetpen aphi, is part of a tail-head linking construction (§12.1.2), which repeats
the verb from the previous utterance, which in fact was t […] t glo,
i.e., marked with e= to cross-reference the argument also indicated by the primary object
marked first person inclusive pronoun itum aphanke. After this beginning in (206), even
though the O argument remains constant (while the A argument changes), the speaker
switches to t ’ , using nang= to cross-reference the first person
inclusive argument.
(206) First person inclusive primary object (3>1INCL) marked with e= ‘1INCL:NSUBJ’
and nang= ‘1/2:NSUBJ’ in the same utterance
ethapkangdetpen aphi, hi'ipi abangke
[e=th p-k ng-d t-p n aph ] h 'ip ab ng ke
1INCL:NSUBJ=put.inside-leaving-PFV-from after witch NPDL=TOP
aphan kelangun'e
aph n ke-l ng- n-C
NSUBJ NMLZ-see-be.able-NEG
'after (our mother) put us in the cradle, the witch could not tolerate (< see) us, she also
couldn't tolerate our mother' [CST, HM 051]
Proclitic nang= may also cross-reference SAPs in non-core roles. In (207) and
(208), first and second person possessors are cross-referenced by nang=.
nangenri"
nang= n-r
1/2:NSUBJ=take-PROH
'"Oh no, king, don't take my life!"' [RBT, ChM 041]
We may label examples such as (208) and (207) instances of ‘possessor raising’,
i.e., ‘promoting’ the possessor to a core role. However, (209) provides evidence that the
use of nang= cannot be generalized in syntactic terms, but that semantic/pragmatic
principles determine whether a SAP is ultimately affected. That is why nang= in (209)
can cross-reference the second person argument, which, syntactically, is the subject of a
relative clause that modifies the O argument of the main verb. (Note that is indeed
the subject of the relative clause rather than a possessor of the nominalized verb.)
228
(209) Cross-referencing a relative clause subject (1>2)
{nang kedo adim ne nangchinike}
{[[n ng ke-d ]REL a-d m] n nang=chin ke
you NMLZ-stay POSS-place 1EXCL 1/2:NSUBJ=know=TOP
‘{I know the place where you are staying}’ [HK, TR 105]
6.3.1.4. Summary
Procliticized nang= (and, via dialectal variation, also ne= and e=) cross-
references SAP in non-subject roles depending on pragmatic as well as syntactic factors.
If this was a strictly syntactic construction, we would expect that we could define, and
predict, which syntactic role gets cross-referenced in this way, and that that particular
syntactic role gets cross-referenced every time it occurs in a clause. Example (200) above,
however, shows that nang=, (perhaps) just like independent pronouns, may be left out as
a zero anaphor. And (209) above shows that it is not just a particular syntactic role that is
cross-referenced, but that it instead appears to be a non-agentive, affected SAP that is
pragmatically cross-referenced via nang=.
Table 87 gives an overview of the person interactions that allow (but not
automatically trigger) cross-referencing via nang= (or ne=/e=). Instead of the typical
A>O format of such tables, I used the notation A>’non-A’ to highlight that it is not just
first and second person O arguments that are cross-referenced in this construction.
229
and V’ (§6.3.2.2); and metaphorical extensions of the cislocative function (§6.3.2.3). A
summary is offered in §6.3.2.4.
6.3.2.1. Directional
The directional cislocative function occurs on motion verbs and indicates that this
motion is directed towards a reference point or deictic center. An example of a manner
motion verb is in (210), where ‘ride, straddle’ occurs with nang= to express ‘ride
(a bicycle) towards (the reference point)’. This is further (i.e., redundantly) indicated here
through the construction that nangardon occurs in. This verb is marked non-final via -si,
and the sentence ends with the lexical cislocative verb ng ‘come’.
230
(212) Cislocative nang= on ‘fall’
bang hantharsi nemoi nangklodup,
[b ng hanth r=si ne-m i nang=kl -d p]
CLF:HUM:PL vegetable.sp=FOC 1EXCL-back CIS=fall-falling.sound.from.high.solid.obj
In (213) and (214), nang= occurs with its cislocative function on - ‘RR-
return’ and - ‘RR-reach’. Both stems carry the reflexive/reciprocal prefix che-, which
here additionally emphasizes that the motion is directed back towards the starting point
(also serving as the reference point). In (215), nang= is used with t t ‘exit’ for the motion
out of the womb, into the world (which acts as the reference point).
hem nangchelelo
h m nang=che-l -l
house CIS=RR-reach-RL
‘we were coming back and at eight o'clock at night we arrived at home’ [SH, CSM 071]
nangketetroiroidetlo
nang=ke-t t-r i~r i-d t-l
CIS=NMLZ-exit-PL.solid.obj~DIST.PL-PFV-RL
‘if they are my children, how did they come out as pieces of wood?’ [CST, HM 023]
231
6.3.2.2. Associated Motion (‘come and V’)
(216) = ‘CIS=keep > come (to reference point) and keep/put (there)’
hala ejon ateketa vangpo laso angchin
h la e-j n a-tek t v ng-p las a-ingch n
that one-CLF:animal POSS-tiger=ADD:also come-IRR1 this POSS-iron
(217) nang= ‘CIS=drink > come (to reference point) and drink (there)’
[…] alanglike latum ahemsi nangvursi sa
al ng-l ke l -t m a-h m si nang=v r-si s
3-HON=TOP this-PL POSS-house=FOC CIS=drop.in-NF:RL tea(<Ind)
ajat nangjunlo
aj t nang=j n-l
GENEX CIS=drink-RL
‘[…] it was him, at their house we stopped by and had tea and everything’ [SH, CSM
067]
(218) and (219) are further examples of nang= on non-motion verbs with an
implied preceding motion towards a reference point. In (218), the preceding motion is
indicated by ablative =pen ‘from’. (219) is given to demonstrate that even ‘stay’ may
occur with cislocative nang= in this sense of ‘come and stay’.
(218) nang= ‘CIS=see > come (to reference point) and see (there)’
det arlo angbongpen laso nangkelangdunta
d t arl angb ng pen las nang=ke-l ng-d n t
country inside middle=from this CIS=NMLZ-see-along=ADD:also
‘from within the country they came to watch this (i.e., there were domestic tourists)’
[SiT, HF 039]
232
(219) nang= ‘CIS=stay > come (to reference point) and stay (there)’
amatsi apenan abang pulo, “ai sarpi!
[am tsi a-pen n ab ng p -l ] ai sarp
and.then POSS-husband NPDL say-RL how.strange! old.woman
In some occurrences of nang=, the cislocative function rooted in space has been
metaphorically extended to the domain of time. The result of this is a perfect-like
interpretation of events that occurred over time up until a (temporal) reference point. In
233
(222), nang= thus indicates that the ‘times changing’ is a process that occurred until the
moment of utterance. Similarly, the verbs at the end of (223) encode duration up until the
moment of utterance, as also indicated by the suffix - ng ‘still’.
nangparisomardunlonglang
nang=pa-ris -m r-d n-l ng-l ng]
CIS=VBLZ-unmarried.boy-PL-JOIN-GET-still
‘they know how to go on a good road up to a high degree (metaphorically referring to
knowing how to do things properly, how to keep everything clean, etc.), and, because
they know how to keep everything clean and nice, those people back then up until today,
get to stay even more like girls and boys (i.e., young) than we do’ [SiH, CW 017]
Figure 16 summarizes what are described above as the basic and semantically
extended functions of cislocative nang=. In the upper portion of the figure, we have the
two functions of the prototypical directional and associated motion.
In the lower portion of Figure 16, the two metaphorically extended functions are
included. They are represented as being extensions off of the directional cislocative
function, which is defined as motion in the direction of a spatial orientation. This
234
representation allows for the simplest analysis since only one semantic change has to
have occurred in both cases. In the case of marking ‘Spatial orientation only’ (examples
(220) and (221) above), the original motion component has disappeared. In the case of
‘Temporal orientation and ‘motion’’ (examples (222) and (223) above), the
commonplace metaphorical extension from space to time has occurred such that the new
reference point is temporal in nature (‘now’ rather than ‘here’), and the ‘motion’ is time
passing on, a common metaphor as well.
nang= ‘cislocative‘
6.3.3. Summary
The two functions of non-subject speech act participant marking and cislocative
marking are synchronically differentiated in Karbi: The person-marking nang= has the
variants ne= and e= for first person non-subject marking, whereas the cislocative nang=
does not have any variants. While this evidence bears the caveat that only some speakers
use the variants, the fact that the variants were able to develop (assuming that they are
indeed a later development) demonstrates that the person and the cislocative marking
functions are functionally individuated enough that they offer fertile ground for structural
differentiation.
Independent of the synchronic difference of the two functions of nang=, however,
there are a number of occurrences of nang= in which elements from both functions can
be recognized, thus representing possible bridging contexts of the type that would have
235
given rise to the initial polysemy. Perhaps the most obvious bridging context is a clause
with a motion verb whose goal or endpoint is a speech act participant. For example, in
(212) above, repeated for convenience as (224), nang= occurs on the motion verb
‘fall’. The vertical motion is thus marked as being oriented towards a reference point,
which is a body part of a first person participant, nemoi ‘my back’.
(224) nang= marked motion verb with speech act participant body part goal
bang hantharsi nemoi nangklodup,
[b ng hanth r si ne-m i nang=kl -d p]
CLF:HUM:PL vegetable.sp=FOC my-back CIS=fall-falling.sound.from.high.solid.obj
(225) nang= marked motion verb with speech act participant goal
kevang akoke ne nanglong nangdunjuilo
[ke-v ng ak ke] [n nang-l ng nang=d n-j i-l ]
NMLZ-come when=TOP 1EXCL 2POSS-LOC CIS=follow-away-RL
Note that in both (224) and (225), nang= is glossed as cislocative, because it
cannot be replaced with e= or ne= for a first person goal interpretation.
The bridging context of motion verbs with SAP goals offers a link to tie the two
functions historically together. Further evidence that we are in fact dealing with related
236
functions that are best considered as having a common historical origin comes from
typological parallels.
Cross-linguistic parallels for the syncretism between non-subject SAP and
cislocative marking exist in several entirely unrelated languages. For example, in the
(unrelated) North American West Coast languages Nez Perce and Shasta, cislocative
markers have taken on the function of indicating speech act participant objects (Mithun
1996). Likewise, in Old Babylonian, the cislocative marker is used for speech act
participant objects (N. J. C. Kouwenberg 2009). As Mithun (1996: 418) notes, “it would
be a short semantic step to reanalyze a verb like ‘Pass it here’ to ‘Pass it to me’.” Now, in
Karbi, the form nang= suggests that the (second) person function, rather than the
cislocative function, is primary, as ŋ is a second person form all across the Tibeto-
Burman family.119
There also are interesting parallels to Karbi nang= inside TB, and specifically in
the Kuki-Chin branch. In Sizang (or Siyin) Chin (Northern Kuki-Chin), there is a
preverbal cislocative (h)ong, which, in fact, parallel to Karbi, also marks non-subject
speech act participants (Stern 1984; DeLancey 2001:132-3).
Furthermore, in Purum (Northern Kuki-Chin), a second person prefix also marks
speech act participant objects in general.
For a more detailed discussion of nang= in a typological context, see Konnerth
(under review).
The nominalizer ke- (with allomorphs ki- and ka-; see §3.9.2.1) is discussed in
Chapter IX.
119
A second person form nang is found in Kuki-Chin languages such as Hakha Lai (Peterson 2003: 411), in
the Bodo-Garo language Garo (Burling 2004: 215), and in Bodic languages such as Baram and Thangmi up
in Nepal (Kansakar et al. 2011: 59). Matisoff (2003: 639) reconstructs it for Proto-Tibeto-Burman as one of
two second person forms, the other one being the also very similar form na.
237
6.4.2. Causative pe- ~ pa-
The causative prefix pe- ~ pa- occurs on all verb stems including those based on
prototypical verb roots and those based on property-concept term (PCT) roots, and
arguably the same prefix occurs as a verbalizer on nominal stems (§5.3.3). As it attaches
to PCT-based stems, it may be part of an adverb construction (§8.3.2). For the
morphophonological properties of this prefix, see §3.9.1.4 and §3.9.2.1.
An example of pe- ~ pa- is (226), where - ‘CAUS-fall’ means ‘cause (the fruit)
to fall (down)’.
Causative pe- ~ pa- is also used with a permissive function (i.e., ‘let somebody do
something’), as shown in (227).
Note that Grüßner (1978: 93-4) reports the acceptability of a double causative, as
in - - ‘CAUS-CAUS-be.good > make somebody improve something’, - -t
‘CAUS-CAUS-die > make somebody kill somebody’.
Matisoff (2003: 132) suggests that in Karbi (and in other Northeast Indian
languages with similar forms, such as Dimasa (Bodo-Garo), Angami Naga, as well as
Kuki-Chin languages such as Khumi and Maraa (Lakher)), causative pe- ~ pa- has
238
grammaticalized from the verb ‘give’ . While this could be the case, it would have to be
a fairly old instance of grammaticalization, because more recent grammaticalizations
seem to end up in suffix slots on the verb, such as the (arguably) more recent
grammaticalization of ‘give’ to - ‘benefactive’ (§6.5.5.2.1).
Besides marking the reflexive (in a broader than typologically expected way) and
the reciprocal, che- also occupies part of the functional territory of a middle as defined by
Kemmer (1993), and to some degree overlaps with cho- ‘auto-benefactive/malefactive’
(§6.4.4). For the morphophonological properties of this prefix, see §3.9.1.1 and §3.9.2.
First, for the reflexive function, see (228), where che- indicates the ‘standard’
reflexive that has the A and O argument of a transitive verb referring to the same
participant, here -t ‘RR-put.inside’ > ‘put oneself inside (a hole in the tree)’. Note
that while there are reflexive/reciprocal pronouns (§4.5.2), their presence is not
obligatory, and che- may be the only indicator of reflexivity/reciprocity.
t t
pe-kl -b p-p ] b ng ke ch -t ng-l j n-t ng-l
CAUS-fall-sd.of.falling.down-IRR1 CLF:HUM:PL=TOP eat-finish-RL drink-finish-RL
‘and then, the husband, when it had gotten bright, he would put himself there in (the hole
in the tree trunk), and then she dropped the rice bundle there (into the hole), and he ate
and drank (everything)' [SeT, MTN 023]
A clause with a che- marked verb may, however, also have an O argument that is
a different participant but is possessed by the A argument. This reflexive-marked
coreferentiality between the A and the possessor of the O can be considered a form of
possessor raising. Nevertheless, since an O argument may be present, reflexive che- does
239
not change the valence of a verb (but see Chapter X and specifically §10.1.2 and §10.6 on
why valence as a strictly syntactic concept is generally problematic in Karbi).
In (229), the same reflexive-marked verb as above, -t ‘RR-put.inside’, has
to be interpreted as ‘put one’s (own; here: tobacco container) inside (here: a bag)’
In (230), (231), and (232), reflexive che- is used with human O arguments that are
possessed by the A argument. The relationships are either kinship (children or wife in the
examples) or friendship.
chevarkangchorlo
che-v r-k ng-ch r-l
RR-throw.away-just-RES:away-RL
'he got rid of his children' [CST, RO 050]
In the ditransitive clause in (233), che- indicates the friendship between the
referents of the R argument and the A argument.
240
(233) Reflexive che-: possessive relationship = friendship
an laso a'oso abang thesere pumni hala
n [las a-os ab ng] [theser p m-n ] [h la
and.then this POSS-child NPDL fruits CLF:round-two that
As illustrated in (234), che- can also occur on motion verbs, here ‘go’. In this
example, the reflexive marking on the verb cross-references amethang atovar ‘their own
road’. Similarly, - ‘RR-go’ also occurs in clauses that contain ‘house, home’
as the goal of the motion, as in (235).
241
(236) che- with reciprocal function
y
pangr pangd n-r ng-l nke h aph ke [[l B y
reconcile even.out-instead-RL and.then over.there after=TOP this CLAN
che'oitanglo
che- i-t ng-l
RR-be.sad-finish-RL
‘they got married, and then quite some time later, Bey the Fair and Bey the Black got
upset with each other’ [WR, BCS 017]
There are several pieces of evidence that che- is an old morpheme. First, there are
phonological reasons. It is a prefix, and like other prefixes in Karbi, it is phonologically
small, i.e., it only has an onset (but no coda) consonant and a weak vowel that is
sometimes deleted. It also has morphophonological effects on the following root tone
(§3.9.1.1). Second, there are a few verb roots whose first syllable is clearly the che-
prefix because the semantics fit perfectly, but the portion without the che- is no longer
242
used as a root independently, for example t ‘meet, run into’, as shown in (238),
where t by itself is not a root.
Since che- is relatively old, there are likely to be (apparent) cognates in other
Tibeto-Burman languages. In fact, in Northern Kuki-Chin, a number of languages have a
ki- reflexive/reciprocal prefix (Henderson 1965; Krishan 1980; Stern 1984; see Konnerth
(2009: 120-3) and So-Hartmann (2013)). 120
anat damlo
a-n t d m-l
POSS-direction go-RL
‘in order to go and sell bananas there he went towards the shop’ [HI, BPh 005]
The much rarer case of expressing a negative effect on the subject, i.e., the auto-
malefactive, is represented in (240), where - means ‘get sick’.
120
There also is a kê- prefix in Tenyidie (formerly known as Angami; belonging to the Angami-Pochuri
group of Central Tibeto-Burman), which, however, combines reciprocal marking and verbalization (Kuolie
2006), so it may or may not be related to the Kuki-Chin and Karbi prefixes.
121
Grüßner suggests the function of cho- is to indicate that “the doer himself carries out the action” (“dass
der Täter selbst die Handlung ausführt”).
243
(240) Auto-malefactive function of cho-
sok sang hem chevan'etke Bamonpo abang
[s k s ng h m che-v n- t=ke] [[Bam np ab ng
paddy raw.rice house RR-bring-PRF=TOP PN NPDL
- - ka-cho-t
child-PL POSS-NSUBJ book NMLZ-AUTO.BEN/MAL-tell
'I teach children (for a living)'
6.5.1. Overview
In a complex verb stem that consists of a root and a predicate derivation followed
by the negative suffix, i.e. schematically: ‘V.ROOT-P.DER-NEG’, there are two options for
the scope of negation. It could be either ‘[[V.ROOT-P.DER]-NEG]’ or ‘[V.ROOT-[P.DER-
122
Note also that Bey (2010:29-42), in a booklet on suggestions for how to standardize Karbi orthography,
offers a list of predicate derivations with sample host roots for the particular suffixes.
245
NEG]]’, that is, the scope of negation could be over the complex stem as a whole, or
specifically just over the predicate derivation. While the result is often ambiguity
between the two scope possibilities, it appears possible to sort a subset of predicate
derivations into those that attract the scope of negation such that only the predicate
derivation ends up negated, and those that form such a tight bond with the verb root or
stem they attach to that negation can only ever scope over the entire complex stem.
Predicate derivations that attract the scope of negation include certain highly
productive derivations that quantify over the event or the O argument, such as - ‘much’
in (245).
Predicate derivations that are not able to be singled out under the scope of
negation include some low productivity (§6.5.1.1.3) degree or extent derivation ones. An
example is - , which occurs with t ‘be clean’ to indicate a higher degree of
cleanliness, ‘very clean’ or ‘thoroughly clean’. However, if the negative suffix is added,
the whole word as in (246) can only ever mean ‘be not clean (at all)’, and not ‘be clean
but not very clean’. That is, the scope of negation can only be over the whole verb stem,
but not specifically over the suffix.
Examples of result derivation, which are ambiguous when the negative suffix is
added, are (247) and (248). In both examples, the scope of negation may be specifically
over the result derivation, such that the action denoted by the verb root occurred but
without the result denoted by the suffix. Or the scope of negation may be over the verb
stem as a whole such that neither the action denoted by the root occurred nor,
consequently, the result denoted by the suffix.
246
(247) Scope of negation ambiguous (result derivation)
t - -
kick-RES:fall.over-NEG
‘(s/he) kicked (it) but (it) didn’t fall over’ or ‘(s/he) didn’t kick it (and it didn’t fall over)’
[KT 111208]
(248)
- - t-
sit-JOIN-RES:affecting.inflated.object-NEG
‘s/he sat down without affecting an inflated object’ or ‘(s/he) didn’t sit down (and didn’t
affect an inflated object)’ [KT 111103]
In addition, (249) shows that more scope issues arise when both a reduplication
suffix and the negative suffix are added. The reduplication suffix, which here indicates
plurality, in combination with the negative suffix, could have scope over the entire verb
stem, i.e., over the action plus the result, with both being negated; or it could similarly
have scope over the entire verb stem, but with only the result negated, which then leads to
the interpretation of ‘repeated jumping without breaking anything’.
Finally, note that reduplication in some instances may also only have scope over
the suffix, as in (250), where the reduplication functions as an intensifier for - ‘a little
bit’.
(250)
t -
know.how-little.bit~INTENS
‘know just a tiny little bit’
247
6.5.1.1.2. Discontinuous Predicate Derivations
A small number of predicate derivations are expressed via a suffix pair that occurs
on two subsequent repetitions of the verb stem. The three so far attested discontinuous
predicate derivations are listed in Table 88, and a text example of -t …- ‘here and
there’ is provided in (251).
6.5.1.1.3. Productivity
Predicate derivations (PDs) vary with respect to productivity, i.e., the flexibility to
occur with different verb roots. Another way to look at that is to consider the various
suffixes located on a grammatical-to-lexical continuum. While some predicate
derivations have a purely grammatical function and can seemingly occur with any verb
root, others are closer to the lexical end of the spectrum, and only occur with verb roots
from a particular semantic field, or a subset of those, or even just a single one.
Sample PDs that are highly productive are given in Table 89. Functional
categories include certain quantifying (or degree or extent indicating) ones such as -
‘much’; argument structure modifying ones such as - ‘benefactive/malefactive’; or
aspectual ones such as - t ‘perfect’.
248
Table 89. Sample highly productive PDs
Highly productive PDs
- ‘much’
- ‘BEN/MAL’
- t ‘PRF’
Table 90. Sample P s that mean ‘quite’ (productive and non-productive PDs)
PD Host verb root
- i [productive]
- ‘be tired, exhausted’
- k ‘be heavy’
-t ng ng ‘be light’
The origins of predicate derivations (perhaps with the exception of the ideophonic
ones) lie in lexical items that in some cases still co-exist in the language along with the
suffixes. In the majority of cases, however, no corresponding lexical items can be found
(anymore). Table 92 shows some sample predicate derivations with their apparent lexical
origins.
250
structure (§6.5.5), and aspect/aktionsart and time derivations (§6.5.6). Several other
derivations that do not sort neatly into one of these categories are discussed in §6.5.7.
6.5.2. Manner
Table 94 shows several sample ideophonic manner derivations, which imitate the
sound of different types of movements.
251
Table 94. Sample ideophonic manner derivations
Form Gloss Sound of… Sample host root
-sir ‘sd.spinning’ spinning t ng ‘spin’
- p ‘falling.sd.from.high.solid.obj’ falling (small, heavy object) ‘fall’
- ‘sd.very.quickly’ running very quickly t ‘run’
Table 95. Sample PDs that indicate a considerable degree (‘quite’) (productive and
non-productive PDs)
PD Sample host root
- [productive]
- ‘be tired, exhausted’
- k ‘be heavy’
-t ng ‘be light’
252
lengvaretmati, sarbura" pu
l ng-var t mati, s rbur pu
be.fat.HUM-INTENS=CG old.man QUOT
‘and then he would return, "why are you so fat/healthy, man? (That's very strange!)"’
[SeT, MTN 025]
There are several highly productive suffixes shown in Table 97 that function as
argument quantifiers (§6.5.5.1) with transitive and ditransitive verbs, but indicate the
degree or extent on intransitive verbs.
For example, ‘eat’ with - ‘much’ becomes ‘eat much’, where - quantifies
the O argument, but ‘be good’ with - ‘much’ becomes ‘very good’. Non-adjectival
intransitive verbs also occur with - , for example, - ‘go-much’, in which case the
scale is one of frequency: ‘go a lot, go often’.
253
Comparative - - t and superlative - - are used with adjectival as
well as non-adjectival verbs, as discussed in §4.2.2. Examples of both suffixes are
repeated below. (See also §10.2.2.5 on comparative constructions.)
(254) Comparative - - t
anke ejon nangtetphlut <a> nangthemuchot
nke e-j n nang t t-phl t nang th -m ch t
and.then one-CLF:animal CIS=exit-suddenly.big.A/O CIS=be.big-COMPAR
‘and then, one (tiger) came out (of the jungle or some area in the Rongker ground) and he
was bigger (than expected and than the previous one)’ [HK, TR 172]
(255) Superlative - -
akethenei akehoineilo tangho […]
a-ke-th -n i akeho -n i-l t ngh
POSS-NMLZ-be.big-SPLT powerful.person-SPLT-RL REP
‘he was the biggest and the most powerful one (so they say) […]’ [HK, TR 033]
6.5.3. Result
Table 98 offers some sample result derivations along with verb roots the suffixes
may occur with. A corpus example with - ‘RES:little.wound’ is provided in (256).
nanglutchok nangarkerakrakdetkema?"
ne-n nang l t-ch k nang ark -r ~r k-d t ke ma]
my-ear CIS=enter-disappearing CIS=scratch-RES:little.wound~DISTR.PL-PFV=TOP=Q
' “O Voarbipi, what were you thinking, coming into my ears and scratching and
wounding me?!"’ [RBT, ChM 034]
254
6.5.4. Direction, (Associated) Motion, Path
255
Let us consider other examples of - n - ‘JOIN’ and - n - ‘on.the.way’.
Another example of - n - ‘JOIN’ is (258), where the suffix indicates that the food
items are being wrapped in order to be taken to the field. (Note that - n - ‘JOIN’ also
occurs in contexts that do not involve motion events, see §6.5.7.1.)
(258) Acting in the context of expected motion: - n - ‘JOIN’
[…] ok paka paka han paka paka lopen
[ k pak pak h n pak pak l pen
meat very.good very.good curry very.good very.good banana.leaf=with
256
(260) Motion away or occurring during ongoing motion away: - n -
‘on.the.way’
saikel vekponbom dambomlo
[saik l v k-p n-b m d m-b m-l ]
bicycle(<Eng) steer-take.away-CONT go-CONT-RL
Another suffix that also indicates the translocative is - - ‘GO’, see (261).
123
Associated motion categories indicate that the event denoted by the verb occurs against the background
of a motion event (Guillaume 2013).
257
Finally, -t - ‘leave behind’ indicates another associated motion
category, which signals in a sense the opposite of - n - ‘JOIN’. While, as pointed
out above (as well as below in §6.5.7.1), - n - ‘JOIN’ is about the involvement in
another event, -tek - ‘leave behind’ is about the lack of involvement in another
event. The difference, however, lies in where motion comes in. In verbs marked
by - n - ‘JOIN’, the motion is associated with the ‘other’ event and not with the
subject’s event (i.e., in the example above, giving water to the runners, the runners are in
motion but not the giver). In verbs marked by -t - ‘leave behind’, it is the
subject that moves and leaves the location of the event, thus signaling the lack of
involvement (i.e., due to moving on). An example is (263).
inutvetpo
e-n t-v t-p
one-CLF:HUM:SG -only-IRR1
‘and then, who would we have given her to, she would have been alone’ [SH, CSM 063]
Note that it appears generally possible to determine which syntactic role a given
predicate derivation quantifies over, although further study is required to confirm that this
is indeed syntactically fixed and not pragmatically flexible.
Examples of both -t -t i ‘all’ and - i ‘many:S’ occur in (264). Note that -t
in this example occurs after a separate construction to indicate universal quantification,
which is the use of ‘all’ in t ’ t (§7.8.2).
124
See §6.5.1.1.2.
259
In (265), an instance of - ‘plural:S/A’ is shown. This suffix is interesting
because it is a negative polarity item, which only occurs in combination with
negative - .
6.5.5.2.1. Benefactive/Malefactive -
(266) Benefactive/malefactive -
laso aphike asitin akhei aphanta
las aph ke [a-is -t n a-kh i a-ph n t ]
this after=TOP POSS-one-each POSS-community POSS-NSUBJ=ADD:EXH
260
Although - acts as a differential O marker (§10.2.1.2), which could be taken
as evidence that - is an applicative that ‘promotes’ an oblique benefactee to argument
status, there is evidence presented in §10.2.3.2 which suggests that this ‘promotion’ is not
marked by - but only ‘highlighted’ by - : The evidence consists in a sentence with a
benefactee participant marked by - , without - occurring on the verb.
(267) Instrumental -
motorsaikel=pen=si t - m-
1EXCL motorcycle=with=FOC market(<Asm) NMLZ-go-with
'I went to the market on a motorcycle' [SiT 090223]
(268) Comitative -
kosonlo, mandu dopo, rit along sitame pinsomar
kos n lo mand d -p rt a-l ng setam [pins -m r
how=FOC field.hut stay-IRR1 field POSS-LOC nevertheless married.man-PL
Note that many argument-classifying predicate derivations are not productive, but
are restricted to occurring with verbs from a particular semantic field. For example, -
‘appearing.small:S’ in Table 101 occurs with t t ‘squat’, as do - and -t , which
classify larger items. However, if we change the verb root to t ‘hold, touch’, we can still
use - and -t for larger items, but for smaller items instead of - we need to
use - . In order to speak about holding a flat object (e.g., paper), another classifying
predicate derivation used with t ‘hold, touch’ is - m.
262
6.5.5.4. Argument Structure Changing
6.5.6.1. Overview
125
The aspect derivations -t and - t are labeled ‘perfective2’ and ‘perfective3’, leaving the simple label
‘perfective’ for the much more frequent - t, which goes in a verb position class following the negative
suffix (see §6.8.1).
263
Table 102. Aspect/aktionsart and time derivations
Type Form Gloss Productivity
Aspect / - t ‘perfect (PRF)’ [productive]
aktionsart - - ‘continuative (CONT)’ [productive]
-t ‘perfective2 (PFV2)’126 [productive]
- ‘completive (COMPL)’ [limited]
- t ‘perfective3 (PFV3)’ [limited]
- ‘durative (DUR)’ [limited]
Temporal - ‘for.first.time’ [productive]
-t ‘again’ [productive]
- - ‘again’ [productive]
Examples of the aspect/aktionsart suffixes listed in Table 102 are given in the
following subsections.
6.5.6.2. Perfect - t
The perfect - t is very frequently used. For example, it occurs in the common
Karbi greeting question shown in (271).
(271) - t- =
rice/food eat-PRF-RL=Q
‘have you eaten?’
6.5.6.3. Continuative - -
(272) Continuative - -
lasonsi juibom juibom juibomlo
las n si j i-b m j i-b m j i-b m-l
that.way=FOC:RL play-CONT play-CONT play-CONT-RL
'this way, they played and played and played' [CST, HM 058]
126
What could be glossed as ‘perfective1’ is - t (§6.8.1), which is, however, glossed as simply ‘perfective’
because it is very frequent as well as in a different slot in the verbal position-class.
264
Continuative - - is likely a reflex of a form reconstructed to Proto-
Tibeto-Burman as *bam~*pam by Benedict (1972: 125), based on, among similar forms
in other languages, Meithei pham ‘sit’ and Lepcha bam ‘remain’. In Lepcha, bam has
futher grammaticalized as a progressive marker (Plaisier 2007:119).
6.5.6.4. Perfective2 -t
Perfective2 -t is productively used with any verb root, but it is not as frequent
as - t ‘perfective’ (§6.8.1). An example of -t is (273), where it is used on a
nominalized verb, of which there are, in fact, a number of other instances. (On the other
hand, it appears that - t ‘perfective’ only occurs on predicates, which makes sense
considering that -t can be analyzed as being derivational and - t as inflectional
(§6.2.1).)
(273) Perfective2 -t
ante kolo nangkeneptang, nangpaklangtha lason
nte ko=lo nang=ke-n p-t ng, nang pe-kl ng-th
OK.then ITROG=FOC CIS=NMLZ-catch-finish 1/2:NSUBJ=CAUS-appear-CON.IMP
'if so, which one have you already caught? You have to show me!, […]’ [HK, TR 090]
265
and.then water NMLZ-take.bath NMLZ-finish-NF:with NMLZ-come and.then king
6.5.6.5. Completive -
Completive - occurs just nine times in the corpus. There are six occurrences
with ‘reach’ and three occurrences with t ‘get rid off’. One of the occurrences with t
‘get rid of’ is offered in (276).
(276) Completive -
Rengsopen Onso aphanke tidamroklo chevanvedetlo
R ngs pen ns a-ph n ke t -d m-r -l che-v n-C -d t-l
NAME=with NAME POSS-NSUBJ=TOP get.rid.off-GO-COMPL-RL RR-bring-NEG-PFV-RL
'"I don't know at all (what to do). Go, in the future, you also become gods (praying to
god)", he had gotten rid of Rengso and Onso' [CST, RO 054]
127
This particular story that (277) is taken from is actually about the impossibility of returning from the
place of the dead. Somewhat resembling the Greek story of Orpheus and Eurydice, the wife dies and the
husband (the bamonpo) in the story travels to Chom arong to try and get his wife to come back to the world
of the living with him. But even though the bamonpo does not make a particular mistake like Orpheus, his
wife still ends up having to go back to Chom arong.
266
6.5.6.6. Perfective3 - t
Perfective3 - t only occurs once in the corpus, with t ‘finish’, as shown in (278).
(278) Perfective - t
amat jutletlo
am t j t-l t-l
and.then finish-PFV3-RL
‘and then, it (the story) is finished’ [SiT, PS 047]
6.5.6.7. Durative -
Like perfective3 - t, durative - also only occurs once in the corpus, see
(279).
(279) Durative -
[… ] mendu chikimra <sok nang arlu> hadak doklungnang […]
[mend che-k m-ra <s k n ng arl > h d k d -kl ng-n ng]
field.hut RR-build-NF:IRR paddy you weed there stay-DUR-HORT
'"[…] let's build us a field hut, and stay there for a long time […]' [KTa, TCS 007]
The next two examples (281) and (282) show, however, that - ~- ‘JOIN’ can
also be used in contexts that do not necessarily involve motion. The larger and more
abstract function of - ~- ‘JOIN’ thus has to do with indicating that the event is
conceptualized as occurring in the context of (or intervening in) an already established
event, which may be a motion event, but does not have to be one.
In (281), a procedural text about traditional Karbi cooking with alkaline foods by
using ashes, the speaker explains that after the fire has burned down the field, it is
necessary to pick up the ashes quickly. The idea is that picking up the ashes quickly
intervenes in a process, which is culturally known to occur otherwise, which is that the
ashes that need to be collected will become wet in the morning dew or will be blown
away by the wind, or will in some other way become unavailable.
elitum humdunji
e-li-t m h m-d n-j
1PL.INCL-HON-PL pick.up-JOIN-IRR2
‘after that, on the hill fields where we have cleared the vegetation, we have to set a fire
and then these ashes from the immature bamboo we have to pick up together’ [SiH, KH
003]
268
In (282), the context of a war between England and Japan is already established.
The use of - ~- ‘JOIN’ on verbs such as t ‘die’ or ‘fight’ indicates that these
events are supposed to be understood in the already established context of this war.128
t t
a-regiment d - ] [las a-regiment k l ~l
POSS-regiment exist-much this POSS-regiment which~DIST.PL
Finally, (283) shows that the involvement suffix - ~- ‘JOIN’ can also be
used with an additive function, in which case it corresponds to the occurrence of the
additive particle on a noun phrase in the clause.
More evidence for this analysis of - n~- n comes from its idiomatic use with
certain verbs such as thàk 'answer', 'listen', and dèng 'accept'. The semantics of these
verbs inherently imply that there is a context in which these actions are performed:
answering requires prior asking, listening requires prior talking or any other existence of
sound, and accepting requires prior giving.
6.5.7.2. Formal -
(284) Formal -
si aphrangsi nanglimen chethan asonte <e>
s a-phr ng-s nang-li-m n che-th n as n-t
therefore POSS-first-SPLT 2POSS-HON-name RR-tell like-COND
me'iksenji
m - k-s n-j
be.good-FRML-INTENS-IRR2
‘so first, if you could tell us your name, that would be wonderful’ [KaR, SWK 004]
(285) Formal -
nangli aphrangphrang, hakoko alam sitame
n ng-l aphr ng~phr ng hak ~k a-l m sit m
you-HON first~DIST.PL that.time~DIST.PL POSS-matter although
arju'iklong tahailo
arj - k-l ng tah i-l
listen-FRML-GET DUBIT-RL
'you most probably have heard the different matters about the old days' [KaR, SWK 059]
270
6.6. Reduplication
Full reduplication of the last syllable of the verb stem (either with or without
vowel change, see §3.8.6.1 and §3.8.6.2) indicates a habitual reading, plurality of an
argument or the iterative, or intensification, as shown in the respective sections below.
6.6.1. Habitual
In (286), reduplication of the last syllable of the verb stem - ‘eat-JOIN’, i.e.,
- ‘eat-JOIN~HAB’ indicates the habitual.
nangkechodundun
nang=ke-ch -d n d n
CIS=NMLZ-eat-JOIN~HAB
‘to us you would also give us, on the ground always you would serve us (food), I also
used to eat like that’ [KK, BMS 060]
271
kal ]] [arl ng a-ch t pen si] ke-d p] a-phl k
NEG.EQU.COP stone POSS-piece=with=FOC NMLZ-cover POSS-split.off.pieces
the'o'a la theklonglo
th - l thekl ng-l
be.big-very~PL:S this see-RL
‘neither with thatch nor with tin, but with slabs of stone they cover (their roofs), the slabs
or stone are very big, those also we got to see’ [SiT, HF 050]
6.6.3. Intensification
272
(290) Reduplication indicating intensification
anungpen damjoijoisi berdamphlutlo tangho
an ng-p n d m-j i j i-si b r-d m-phl t-l t ngh
back-from go-quietly~INTENS-NF:RL press.down-GO-miss/fail-RL REP
{mm} berdamphlut
{mm} b r-d m-phl t
AFF press.down-GO-miss/fail
‘from behind he was very secretly approaching and tried to jump on him, but failed’
[HK, TR 052]
The onset reduplicative negative suffix represents the main verbal negation
construction (for nominal negation, the negative equational copula is used, see
§4.6.2.2 and §8.1.1). For (morpho-)phonological details, see §3.8.6.3.
Morphosyntactically, the negative suffix forms the ‘right edge’ of the derivational verbal
complex or verb stem (§6.2.1). An example of the negative suffix in a complex verb is
(291).
Perfective - t occurs with high frequency in the corpus (at a total of 134
occurrences). There are three semantic components to it: perfectivity, exhaustiveness, and,
though much less prominently and less consistently, a stance component of a negative
evaluation of the event by the speaker (in the sense of ‘did X but should not have’).
273
The perfective component is the most consistent semantic component of - t. It is
particularly clear in examples such as (292), where a negated past event is expressed, i.e.,
something that did not happen.129 Keep in mind that - ‘realis’ (which on non-negated
stems typically has past implications) results in a future reading if directly attached to a
negated stem (e.g., - - ‘go-NEG-RL’ > ‘won’t go’, see §6.9.1.2). Therefore, in
order to express a negated past event, perfective - t is required, following which -
‘realis’ again has past implications even on negated stems.
“ ; t M
p -l ch k p si ch k-C -d t-p m -s n-l ke
say-RL be.fine why be.fine-NEG-PFV-IRR1 be.good-INTENS-RL=TOP
”
n-n i pu
take-INFRML.COND.IMP QUOT
'”is it fine?”, they asked me, and I replied, “it's fine, why would it not be fine?, it's okay,
take (the interviews)”' [SH, CSM 049]
129
Another morphosyntactic type of occurrence of - t that especially highlights its perfectivity is in
subordinate clauses with subordinator ‘after’. An example is - t ‘after watching’ (SH, CSM
060).
274
native speakers that participated in a pilot study using the Fish Film experimental
protocol (Tomlin 1995), in which speakers repeatedly describe events of fish swallowing
up entire other fish.
Finally, there also appears to be a stance component associated with - t that at
times surfaced in elicitation (although it remained difficult to pin down), such that the use
of - t suggested that the speaker evaluated the occurrence of the - t marked event in a
negative way, implying that it would have been better for the event not to actually have
occurred.
6.8.2. Experiential -
In the corpus and in most elicitation data, - only occurs on negated verb stems
and thus appears to be a negative polarity experiental marker ‘never’. An example is
keningjejepinpi in (294), which occurs inside a relative clause that marks exclamative
force here.
(294) - ‘never’
mh “an akai adin ne nangkapangreng
mh [ n a-k i a-d n n nang=ke-pangr ng
DSM that.much POSS-time(<Asm) POSS-day(<Asm) 1EXCL 1/2:NSUBJ=NMLZ-rear
275
6.9. Mood: Realis and Irrealis
6.9.1. Realis -
Previous research has ascribed some notion of past tense to - . Jeyapaul (1987:
113) calls it a past tense marker which becomes optional in the presence of an explicit
time adverbial. In other contexts, he ascribes the function of perfect aspect to it claiming
that it occurs if an action starts in the past but continues up to the present (p.114).
Grüßner (1978: 99) calls the category marked by this suffix the “narrative past.” He states
that - marks a completed action, therefore setting the stage for a new action, and thus
giving rise to a clear sequence of events expressed by verbs that carry this suffix.
In the present account of the verbal suffix - , I will argue that - is not a tense
marker. Certain verb forms with - clearly refer to the future. Instead, I will argue that
the frequent past reference of verb forms marked with - is a consequence of the
aspectual-pragmatic functions of this marker.
The most obvious interpretation for many uses of Karbi - is perfective132, and,
to a more limited extent, perfect aspect. There are several examples in which -
resembles a perfective in languages that employ systematic marking of this category.
Likewise, the concept of perfect marking plays a role, but in a broader sense than
typically associated with this notion. However, - is used in other contexts as well.
130
In fact, because my main, fully annotated corpus (with a total of approximately 13,000 words) mostly
contains narrative texts, there are over 1,000 occurrences of - ‘realis’ in it.
131
An intriguing hypothesis is that Karbi - might possibly be related to the Daai Chin (Southern Kuki-
Chin) ‘inceptive aspect’ and change of state marking lo (So-Hartmann 2009:110-1).
132
A standard definition of perfectivity is 'the view of a situation as a single whole, without distinction of
the various separate phases that make up that situation' (Comrie 1976:16).
276
Considering the morphosyntactic context of Karbi - , as it occurs in close
paradigmatic opposition with the two irrealis suffixes - and - (§6.9.2), I gloss - as
‘realis’. Clearly, a vague label such as ‘realis’ cannot explain the functional range of such
a highly frequent grammatical category. The following subsections thus aim to break
down this functional range into particular semantic and pragmatic contexts in which - is
used. These are:
Examples in the following subsections are mostly taken from two stories, a
personal narrative (SH, CSM) and a traditional story (HK, TK). The former was told by
Sashikola Hansepi, a middle-aged woman from Diphu, and is about a trip to the
Chomangkan Festival, a celebration to honor a family's relatives that have passed away.
The traditional story was told by famous singer Harsing Kro, a middle-aged man from
West Karbi Anglong, and is about an orphan and a tiger (hence referred to later as the
Orphan and Tiger Story; see Appendix C).
In perhaps its most frequent usage, - appears on action verbs in the narration of
past event sequences.133 For example, (295) is the beginning of a simple personal
narrative.
133
This function is also the most salient one to native speakers in the sense that everybody I have talked to
will say that - marks past tense - which, however, likely reflects the attempt to impose English categories
on Karbi.
277
Almost all predicates that appear as the sequence of events in narrative texts
contain - , and just hearing the sequence of verbs with - would give the listener a fairly
complete picture of what happened. This is similar to the use of perfective marking for
completed events, and has probably given rise to interpreting it as a past tense marker in
the literature.
As we will see below, stative or adjectival verbs in narratives often occur without
- , which represents evidence that - does not mark past tense as it occurs on event
sequences in narratives. But also in the case of action verbs, - does not always
correspond to past tense. In (296), a negated bare stem without - in (a) is contrasted
with the same form with - in (b) yielding a future sense. Likewise, the statement in (297)
refers to the future.
(296) (a) - k - (b) - k - -
pig-meat eat-NEG pig-meat eat-NEG-RL
'(I) don't/didn't eat pork' '(I) won't eat pork anymore'
(297) ng -t -t -
3SG come-again-NEG-RL
's/he won't come again' SiT 090220
Both (296) and (297) show that - does not mark past tense. Here again, - is
similar to a perfective marker, in that perfectives typically yield a future sense in
combination with a present (or non-past) tense verb. Karbi - is different, however, as
this is only the case with negated stems. Because Karbi does not mark tense at all, the
interaction of - with tense is clearly different from traditional aspect languages such as
Russian, which has different forms for past and non-past verbs.
One way we can interpret the function of - in these two examples is to consider
it marking a change of state - which links it back to a perfective function, but in a
different light. Of course, 'eat' and 'come' are action verbs, but since they are negated in
these two examples, they resemble states more than actions. In (296), the (implied)
previous state of actually eating meat or usually eating meat changes to the state of not
eating meat at the time of utterance. Likewise, the state of 'him/her not coming again' in
278
(297) begins at the moment of utterance, and so applies to the immediate or distant future,
again implying that before the moment of utterance, this statement was not true.
In both examples, we can interpret - as marking the beginning of the state of the
negated action statement. This could be seen as a purely temporal relationship, in which
case 'currently or usually doing something' temporally precedes 'not doing something
anymore right now or in the immediate or more distant future'. However, probably more
often than not, there is an (implicit or explicit) logical relationship that explains the
temporal relationship. In other words, usually, 'now, (I) won't eat pork anymore' implies
some reason why this statement has become true, such as 'I have eaten so much pork
already' or 'I have converted to Islam.' Even more so in the case of (297), 'not coming
again', we probably automatically interpret that this 'now' - the beginning of the negated
action state - refers to a reason that rules out the possibility of 'him/her coming again.'
Given the other contexts in which - is used (especially the ones discussed in §6.9.1.4),
it makes sense to consider this implied logical relationship an important component of the
function of - .
In (a), the bare stem of the negative existential copula is used. This sentence
would be used when, for example, a seller in the market does not sell meat, or a particular
meal was vegetarian. The statement in (b), however, implies that meat is just sold out, or
the particular meal included meat, but all the meat is already eaten up. Therefore, this
function marked by - in (298) is to denote a change of state.
279
This change-of-state sense of - also explains (299), which is from the very
beginning of the Orphan and Tiger Story, and is part of introducing the protagonist. The
implication is that large families with a number of children are the rule in a village setting.
Therefore, being all alone and not having any close relatives is likely to mean that this
orphan lost not only his parents, but also brothers and sisters - as opposed to never having
had any siblings.
tovar henopik
tov r hen -p k
road bad-very
‘but between Doboka and Hojai the road is very bad’ [SH, CSM 017]
280
anke <tovar> tovar longle adukta dopiklo
nke <tov r> tov r [l ngl a-d k t ] d -p k-l
and.then road road earth POSS-dust=ADD exist-very-RL
‘and then, there is also a lot of dust’ [SH, CSM 018]
In (300), - does not occur on the first predicate, because 'being bad' is a general
statement about the road. But the second predicate 'being very dusty' has to occur with - ,
because of the logical connection between both statements: Because the road is bad
(meaning that it has many potholes), the road gets dusty (as dust accumulates in the
potholes).
Note that the second statement in isolation would not be marked with - . Just
remarking that the road was very dusty, the most natural thing to say would be tovar
longle aduk dopik. If one were to say it with - in isolation (outside the context of this
story), i.e., tovar longle aduk dopiklo, it would actually translate more as 'the road was
dustier now' implying a comparison with an earlier condition of the road (which might
have been more or less dusty, but in any case different from the present condition). This
implication is not present in (300). Instead of marking a change of state, maybe one could
say that - here marks a conditioned state, or a conditioned quality as opposed to an
intrinsic quality.
Another example of - marking a conditioned state is (301), where the cause of
ening arongpiklo '(we) were very happy' is mentioned just before (i.e., reaching just in
time to see the own clan perform a certain ritual).
ening arongpiklo
e-n ng ar ng-p k-l
1PL.INCL-mind be.happy-very-RL
‘it was the Amri people doing the Rongketong, and then we are also from Amri, and so
281
we happened to reach just in time for it, and (so) we were very happy’ [SH, CSM 030]
(302)
chingjor epak {chingjor epak?!} hak isi do
chingj r e-p k chingj r e-p k h k is d
shovel one-CLF.flat shovel one-CLF.flat finely.woven.bamboo.basket one exist
'one shovel, ...' {'one shovel?!'} one basket he had’ [HK, TR 016]
The storyteller also mentions that the orphan had a small tobacco container with
him, which will be of importance for the story at a later point, (303).
(303)
atema ajerjer do mati ho
a-tem ajerj r d mat h
POSS-tobacco.container(<Asm) small exist CG EMPH:INTERACT
he had a small tobacco container, OK? [HK, TR 018]
In both (302) and (303), ‘exist’ occurs without - . Along the lines of the
previously discussed analysis of - on stative verbs, we would indeed not expect to
find - here, because we are not dealing with a change of state or a conditioned state. So
(302) and (303) are not problematic. However, the excerpt (304) contains essentially the
same statements as (302) and (303), but here all instances of ‘exist’ occur with - .
134
Note that curly brackets in examples from the HK/TR traditional story are used for utterances (usually
questions, affirmative interjections, or repeated parts of previous sentences) by a Karbi native speaker who
was listening to the storyteller for a more natural storytelling situation.
282
nopakta dolo dak {lahe} mm
nop k-t d -l d k lah mm
dao-ADD:also exist-RL here that.way? AFF
'he did have a dao also there.' {'is it like that?'} 'Mm.' [HK, TR 023]
What caused the storyteller to repeat those statements he just made was that the
listener asked specifically whether the orphan did not also have a knife with him (as it is
normal to carry a knife along when you leave the village and are on your own). So the
storyteller needs to correct himself, or rather, clarify: Yes, the orphan had a knife, and a
shovel, and a basket - and then he also had a tobacco container. It is this pragmatically
marked context of clarifying or correcting wrong assumptions that is the trigger for the
use of - in this case, as native speakers have expressed that - sometimes functions to
emphasize statements.135
Similarly, consider (305), where the realis marked existential copula - is used
in the context of denying an allegation of lying.
135
In this story, of course only the possession of a knife (and not the other items) is a clarifying or
correcting statement - however, it seems reasonable that in this context the storyteller just clarifies overall
what all things the orphan is carrying along.
283
6.9.1.6. Summary
284
The fact that verbal - and nominal - overlap to some degree but not entirely in
their functional range could be taken to consider them two different morphemes.
However, consider that irrealis - and - also occur on nominal predicates in examples
like (307), as do other verbal affixes more generally, which is shown in §4.1.2.
inutvetpo
e-n t-v t-
one-CLF:HUM:SG-only-IRR1
‘but who will (we) give (her) to, (she) will be alone’ [SH, CSM 063]
In light of the larger word class agnosticism underlying the fact that elements
from any of the larger word classes can function as predicates in Karbi (§4.3), it makes
most sense to propose only one morpheme - ‘realis’, which occurs in both verbal and
nominal predicates. (I do recognize there to be another lo morpheme, which does not
occur on predicates, which is =lo ‘focus’, discussed below.)
A number of language consultants have suggested that there also exists a form
, which may be related to - . An example of this form is offered in (308).
However, occurs in a restricted number of contexts, almost all of which
contain the demonstrative . This form appears to be a more formal or
otherwise special variant of the simple realis. It is glossed here as ‘realis:emphatic’
(‘RL:EMPH’). A compatible account of is offered by Grüßner (1978: 132), who
parses into he and - , with he analyzed as a pragmatic sentence-final type particle
(see §12.3.9 on =he).
(308)
bai pei nangingjinso'un'e nangtum aphan,
b i p i nang ingj ns - n-C nang-t m a-ph n
how.mean! mother 1/2:NSUBJ=have.pity-be.able-NEG you-PL POSS-NSUBJ
285
EXCLAM this-RL:EMPH that.way be.thin NMLZ-exist be.thin
Also note (312), where a sentence involving focus realis =si produced by one
speaker (not the main storyteller here, as indicated by the curly brackets) is repeated as a
286
way of confirming it by the main storyteller. Crucially, however, in repeating it, the
storyteller uses =lo instead of =si, which demonstrates that the two clitics have
equivalent focus-marking functions here.
There are two irrealis markers, - ‘irrealis1’ and - ‘irrealis2’, which both cover
cross-linguistically typical irrealis categories associated with futurity. While - is used
in the context of a more immediate, definite future/irrealis situation, - is used in more
general, intentional, or indefinite future/irrealis contexts.
For example, if one says ‘I will come over tomorrow’, they will be more likely to
use - as compared to a statement such as ‘I will come over tomorrow at 2pm’, in which
case the use of - is more likely. That said, the suffix - ‘definitely’ can be used in
conjunction with - as well as with - , and it has proven difficult for my language
consultants to imagine contexts in which only one of the two suffixes is truly acceptable
and the other not acceptable.
An illustrative example from the corpus is (313), where a question using the more
indefinite or general - ‘irrealis2’ is replied to using the more definite - .
287
EXCM 1EXCL 1/2:NSUBJ=take.away-know.how-IRR1 EMPH grandfather:VOC
A similar example with - ‘irrealis2’ in the question, however the bare stem in the
reply is (314).
“ ; t M […]”
p -l ch p si ch k-C -d t-p m -s n-l ke
say-RL be.fine why be.fine-NEG-PFV-IRR1 be.good-INTENS-RL=TOP
'”is it fine?”, they asked me, and I replied, “it's fine, why would it not be fine?, it's okay,
[…]”' [SH, CSM 049]
The following discussion offers examples of the various future irrealis functions
of - and - . In §6.9.2.2, examples of past habitual marking, which apparently only
occurs with - , are offered.
Future irrealis contexts in which both - and - are used include simple future
(§6.9.2.1.1); habitual marking in procedural texts (§6.9.2.1.2); hypotheticals and
counterfactuals (§6.9.2.1.3); epistemic contexts of expressing uncertainty (§6.9.2.1.4);
deontic contexts of expressing necessity and obligation (§6.9.2.1.5); desideratives
(§6.9.2.1.6); and, finally, purpose clauses (§6.9.2.1.7).
Examples (315) and (316) are from the same text and refer to the same situation,
talking about the future as indicated by the time adverb ‘tomorrow’. Perhaps
somewhat unexpectedly, the first instance in (315) is marked by - ‘irrealis1’, which
288
generally is the more definite, immediate future irrealis marker, and the following
instances in (316) are marked by - ‘irrealis2’, the generally more indefinite marker,
including, in the last of three repetitions, - in conjunction with - ‘definitely’. Perhaps
the use of - here emphasizes the intention of the speaker.
In the excerpt (317), three subsequent intonation units from a procedural text are
offered, all of which have the final verb marked with - , as is typical.
t
las takir pen j ng-
therefore spindle(<Asm)=with spin-IRR1
then we spin/make (the thread) with a spindle [KST, PSu 005]
Irrealis2 - occurs less frequently in procedural texts, but there are several
instances, such as the ones offered in excerpt (318).
The two irrealis markers are used in both (present) hypothetical situations, as well
as in counterfactual situations (in which a hypothetical alternative to a past event is
expressed). First, let us examine examples of - . In (319), the speaker says ‘it is okay’,
which is a realis assertion, but then adds ‘why would it not be okay?’, i.e., imagining an
irrealis alternative to the realis situation, which is indicated by the use of - .
290
In (320), which is from the same personal narrative as (319), a text about a trip to
the site of a festival, which was recorded the day after the trip, the speaker imagines a
hypothetical alternative to a realis situation from the trip the day before. This is thus a
counterfactual situation that is marked by - , as the hypothetical alternative is about a
past event.
inutvetpo
e-n t-v t-p
one-CLF:HUM:SG-only-IRR1
‘and then, who would we have given her to, she would have been alone’ [SH, CSM 063]
Similar to the above examples of - in (present) hypotheticals and (past)
counterfactuals, (321) and (322) offer analogous examples of - .
nangpinkhattap nangpinkhatphru
nang pinkh t-t p nang pinkh t-phr
1/2:NSUBJ=advise-here.and.there.1/2 1/2:NSUBJ=advise-here.and.there.2/2
'since we are alive, (how can) you give so many pieces of advice as if we could reach (the
place where my wife has gone after she died, when in fact we cannot)' [KK, BMS 031]
291
Expressing uncertainty, probabilities and venturing guesses is another common
situation where future irrealis markers are used in Karbi, as well as cross-linguistically. In
(323), - is used in conjunction with mene mena ‘maybe’ to indicate that this is an
assumption.
The deontic verb ‘must, need’ almost always occurs with - ‘irrealis2’, as in
(325), and there are a few instances in the corpus where it occurs with - ‘irrealis2’,
such as (326). Realis - cannot be directly attached to , i.e., * - is
unacceptable, but in conjunction with the intensifier - , which also occurs in (325), it
292
yields an acceptable verb form, i.e., - - (although there are only three
occurrences of this form in the corpus).
(325) - on ‘must’
{lang nangkokjike}
l ng n ng-k k-j =ke}
see must-INTENS-IRR2=TOP
‘{we must see! (if there is a tiger there) ’ [HK, TR 086]
(326) - on ‘must’
[…] nangpo […]
[pin pen ke n n ng pen n ke che-k k n ng-p ]
today=from=TOP 1EXCL you=from 1EXCL=TOP RR-part must-IRR1
‘[…] “from today I will need to part from you” […]’ [SeT, MTN 039]
While these are the only two examples in the corpus, elicitation suggests that -
‘irrealis1’ can be used in this construction preceding -te ‘conditional’ as well.
293
6.9.2.1.7. Subordinate Purpose Clause Marking
nangkechodundun
nang=ke-ch -d n ~d n
CIS=NMLZ-eat-JOIN ~HAB
‘to us you would also give us, on the ground always you would serve us (food), I also
used to eat like that’ [KK, BMS 060]
6.9.2.3. Summary
294
guessing; expressing necessity and obligation; occurring in the conditional desiderative
construction; and marking purpose clauses, the more common irrealis marker is - .
There are three non-final suffixes in Karbi: -si and -ra (§6.10.1), and -pen
(§6.10.2), as well as a conditional suffix -te (§6.10.3), and an indirect question marker -
(§6.10.4).
There is a realis non-final marker -si and an irrealis non-final marker -ra, which
are used in forming clause chains, see (331) and (332). Clause chaining is discussed in
§11.2.1.1.
(331) Chained clauses marked with -si ‘non-final:realis’
e anke apaita <m> pharla dam,
e [ nke a-p i t <m> pharl d m]
DSM and.then POSS-mother=ADD:DM HESIT outside.part.Karbi.house go
"dojoinoi, po!"
[d -j i-n i p ]
stay-quietly-INFRML.COND.IMP father
‘and then, the mother went and unloaded the firewood in the Pharla (Veranda), then went
inside the house, sat down in the Hongkup, gave the child the milk, (and said) '”be quiet,
daddy”' [KK, CC 015]
295
this POSS-road-exactly come-NF:IRR this=with here
The use of realis -si or irrealis -ra is determined by the final verb, as shown in
§11.3. Generally, in a particular context, only the realis form or the irrealis form is
acceptable. However, some of my language consultants found a peculiar construction
acceptable (and produced it spontaneously), whereby in a series of more than one non-
final clause, the non-final markers are used in alternation. An example is (333), where a
declarative, non-negated (and therefore regularly realis) final verb is preceded by two
non-final clauses: the first one marked by realis -si, as we would expect, but the second
one marked by irrealis -ra. Peculiar as it may seem, this is robust and fully acceptable for
some of my language consultants.
anat chevangthulo
a-n t che-v ng-th -l
POSS-direction RR-come-again-RL
‘and then, he whistled, picked up the hat and brought it back’ [SiT, PS 036]
In addition to (ir)realis-sensitive -si and -ra, there is a third non-final marker -pen,
which has presumably grammaticalized from a source construction of =pen ‘with, from’
296
(§7.8.1) occurring with nominalized verbs. Synchronically, -pen (still) frequently occurs
on nominalized verbs as in (334), although there are also many instances of -pen on verbs
not marked by ke- ‘nominalizer’, as in (335).
keong angbong
ke- ng angb ng
NMLZ-be.much middle
and then, that way we were joining them and reaching there, the people of the house
served as really really well, amongst so many people [SH, CSM 039]
297
Note that -pen and realis non-final -si may co-occur, as in (336).
The aspectual suffix - , glossed as ‘still,’ occurs in the same temporal sense
English ‘still’ has in instances such as (339). The context in this folk story is that before
the speaker(s) can stay with their biological father, they still need to ask their adopted
(tiger) parents.
Finally, - is used for events that are expected to take little time or when there
is a sense of the event being one step amongst several consecutive ones. For example, if
there is a plan to get going to the market, but the speaker wants to go and wash her hands
before leaving, she could say - - ‘hand wash-GO-still’, with a
299
translation along the lines of ‘I’m just gonna go wash my hands real quick (and then we
can go)’.136
Table 103 offers an overview of non-declarative speech act suffixes along with
references to sections in which the various constructions involving these suffixes are
discussed.
136
In German, noch ‘still’ is used in the same way: ‘Ich gehe mir nur gerade noch schnell die Hände
waschen (und dann können wir los).’
300
CHAPTER VII
The topic of this chapter is the structure of the noun phrase in Karbi. In §7.1, the
elements of the Karbi noun phrase are introduced and an overview of the noun phrase
structure is offered. Evidence for the noun phrase template is provided in §7.2.
The topic of §7.3 is the diachronic significance of the possessive construction
consisting of a modifying nominal preceding an a- ‘possessive’ marked head noun. This
construction has been the source construction for the grammaticalization of both syntactic
categories as well as grammatical markers. The a- marking of head nouns is further
discussed in §7.4. This a- possessive marking of head nouns with preceding modifiers is
also particularly significant as it occurs in more complex noun phrases with several
modifiers. Here, the a- prefix contributes evidence to an analysis of either a flat or
hierarchical/embedded structure of the preceding modifiers.
The following three sections discuss modifiers that always precede their head
noun (§7.5), the plural -t , which always follows the head noun (§7.6), and property-
concept modifiers, relative clauses, and enumeration constructions, which may occur
before or after the head noun (§7.7). Finally, §7.8 discusses the clitics that may occur at
the end of a noun phrase.
The Karbi noun phrase structure is shown in Figure 17. Evidence for this template
is offered in §7.2 below.
(PCT modifier)
(RC) HEAD
(DEM) (NUM) ([NP]POSR) (RC) (PL)
(PCT modifier) NOUN
(NUM)
Figure 17. Karbi noun phrase structure
301
Figure 17 sorts six different types of modifiers into position classes inside the
Karbi noun phrase. Of these, two only occur before the head noun: demonstratives (DEM)
(§7.5.1) and possessor or modifier NPs ([NP]POSR) (§7.5.2). (Two additional types of
marginal preposed modifiers are illustrated in §7.5.3.) On the right edge of the noun
phrase, there is one other modifier: the plural with -t (§7.6). Finally, modifiers that
may occur on either side of the head noun include relative clauses (RC) and modifiers
derived from propert-concept terms (PCT) (§7.7.1), as well as enumeration constructions
(§7.7.2). Note that the order of these three modifier types is not only variable with respect
to pre- vs. postposed to the head noun. In cases of a preposed enumerator and a preposed
relative clause or PCT modifier, the more common ordering in the corpus is for the
enumerator to occur first, but the reverse order is attested as well. Finally, note that in
postposed position vis-à-vis the head noun, the occurrence of more than one modifier of
the three types NUM, RC, and PCT is not attested in the corpus. There also is no attested
instance of preposed RC and PCT modifiers co-occurring.
While the head noun slot is typically occupied by a common noun, there are three
types of coordination constructions which can go inside the head noun slot as well: a
simple juxtaposition construction, and constructions involving coordinators =pen ‘with’
or ‘and’. In addition, quotative pu may immediately follow the head noun.
7.1.2.1.1. Juxtaposition
In the juxtaposition construction, two nouns are placed side by side without an
overt coordinator, such as nepiso neso in (341) and ahor ahan in (342). This cannot be
treated as an instance of root compounding, since any required morphology is repeated on
both nouns, i.e., the first person exclusive possessive pronoun ne- in (341) and the
general possessive/modified prefix a- in (342).
302
(341) NP coordination by juxtaposition: nepiso neso ‘my wife and my children’
e nang nepiso neso aphan
e n ng ne- is ne-os HN aph n]
DSM 2 1EXCL:POSS-wife 1EXCL:POSS-child NSUBJ
kachekipphakma" pulo
ke-che-k p-ph k ma pu-l
NMLZ-RR-pour.out-almost.completely=Q QUOT-RL
‘”[…] this rice beer and curry that I was carrying for my mom to let her taste it, why
is almost all of it it coming out by itself?” (she was thinking)' [SeT, MTN 049]
ephu aphan ” pu
e- h a-ph n] pu
1PL.INCL-grandfather POSS-NSUBJ QUOT
'”take them, look, (take them,) our grandmother and grandfather!”’ [KK, BMS 106]
303
(344) Coordination construction with la n ‘and’ in head noun slot
[…] la Bey Ke'et Bey Ronghang lapen Bey Ki'ik
[l B y ke- t B y Rongh ng la n B y ke- HN
this CLAN NMLZ-be.yellow CLAN CLAN and CLAN NMLZ-be.black
atum angjirta do pu
a-t m] a-ingj r t d pu
POSS-PL POSS-sister=ADD:also exist QUOT
‘[…] that Bey the Fair, Bey Ronghang, and Bey the Black, they also had a sister, it is
said' [WR, BCS 007]
Note that the same conjunctive coordination construction with =pen ‘with’ may
also occur in the possessor or modifier NP slot, as in (345), as well as function to
coordinate separate noun phrases, as in (346). Further research into the structural
properties of coordination at the noun phrase level is required.
7.1.2.2. Quotative pu
304
examples, pu simply occurs after the head noun. However, while in (347), the head noun
and name Kache Kropi also functions as an NP by itself, in (348), pu occurs after the
head noun t ‘fruit’, which, in turn, is followed by the noun phrase delimiter
(§10.5). In (348), pu therefore specifically occurs following the head noun, not following
the NP as a whole.
(348) Quotative pu after the name of a fruit, with noun phrase delimiter following
lasi la suho athe pu abangke pe ri
las [l s h a-th pu ab ng ke] [[p r
therefore this thorny.plant.sp POSS-fruit QUOT NPDL=TOP cloth EE:p
kechok aphan'iklo
ke-ch k] a-ph n- k-l ]
NMLZ-wash.clothes POSS-NSUBJ-FRML-RL
‘then, the so-called ‘suho fruit’ is for washing clothes’ [SiH, CW 007]
The position of pu following the head noun is also illustrated in (349), where both
the plural and the noun phrase delimiter still follow. (Note that in the context of the story,
it is more plausible to consider y ’ t a name and simply a head noun, rather than a
head noun with a following PCT modifier.)
305
7.1.3. The Noun Phrase Delimiter
The noun phrase delimiter marks the right edge of the noun phrase. For a
discussion of structural properties and functions of , see §10.5.
First, a flat structure analysis may be appropriate, such that both the
demonstrative and the possessor noun equally modify the head noun, i.e., [DEM] [NP]POSR
[HN].
Second, however, a hierarchical or embedded structure analysis may be more
appropriate, such that the demonstrative and the possessor noun form a new noun phrase,
which then modifies the head noun: [[DEM] [NP]POSR]NP [HN].137
While the absence of a- ‘possessive’ on the possessor noun may be taken as
evidence that a flat structure analysis is more appropriate (again, see §7.4), a
comprehensive modeling of the NP in Karbi as being essentially flat or hierarchical has to
remain inconclusive at this point and requires further research.
7.2.1. Overview
Table 104 gives an overview of noun phrase structures found in the corpus, with
respective examples offered below in §7.2.2 as listed in the table.
137
In following binary branching analyses, a third possibility is [ DEM [[NP]POSR HN]NP]. This analysis is not
further discussed here, as there is currently no evidence for a tighter unit between the possessor NP and the
head noun.
306
Table 104. NP structure (DEM=demonstrative; NUM=numeral or classifier-numeral-
word; RC=relative clause; PCT=property-concept term modifier; POSR=possessor;
PL=plural)
(PCT) (PCT)
HEAD
(DEM) (RC) ([NP]POSR) (RC) (PL) Example #
NOUN
(NUM) (NUM)
+ +NUM + (350)
+ +RC + (351)
+ + + (352)
+ + +NUM (353)
+NUM +PCT + (354)
+NUM +RC + (355)
+RC +NUM + (356)
+NUM + + (357)
+NUM + +PCT (358)
+RC + + (359)
+ + + (360)
+ +PCT + (361)
7.2.2. Data
307
(352) NP structure: DEM - [NP]POSR - HN138
amat la apiso abangke la jangthu abong
am t l a-pis ab ng ke l j ngth a-b ng
and.then this POSS-wife NPDL=TOP this oil POSS-bottle
138
For a discussion of preposed modifier embedding, and differential analyses of these kinds of examples
as [DEM - [NP]POSR - HN] or [[DEM - NP]POSR - HN], see §7.4.
308
kirim kibi isi ahem do laso
ke-r m ke-b RC [is ]NUM a-h m HN d las
NMLZ-put.in.one.place NMLZ-keep one POSS-house exist this
langdamlonglo
l ng-d m-l ng-l
see-GO-GET-RL
‘[…] there is one house where they put everything they got from the time of when the
British were fighting against Japan, this we got to go and see’ [SiT, HF 056]
ba arong asarthelo
b a-r ng a-sarth -l
I.mean(<Asm) POSS-village POSS-village.headman-RL
‘a long time ago, there was one village Gaonbura, I mean, one village headman’ [CST,
RO 003]
309
(360) NP structure: [N]POSR - HN - PL
[…] halabangso ahut elitum Karbi akhei atum
[h la-b ng-s ah t e-li-t m arb a- h i a-t m
that-CLF:HUM:PL-small during 1PL.INCL-HON-PL PN POSS-community POSS-PL
abiha kedam pu
a-bih ke-d m pu
POSS-trade NMLZ-go say
‘that way, even today also, as the world is mature, the new generations get to tell each
other a saying, "Bokolapo is going to the market" (meaning instead of going to your
destination, you're going the other way)’ [HI, BPh 020]
310
and - (§10.6.3). Moreover, the restrictive focus marker - t (§10.7.6.3) must have
grammaticalized in the same way.
Functionally, it makes sense to treat some of the categories that have emerged,
such as the plural marker -t or the emphatic reflexive - t, as postposed modifiers, i.e.,
modifiers that follow their head noun. Structurally, however, they themselves are the
head noun, the same way that in English NPs like ‘the crowd of the students’ or ‘the
students’ crowd’, ‘the crowd’ is the head noun and ‘the students’ acts as a modifier.
This is analogous to relator nouns, where in an NP like ‘the top of the table’ or
‘the table’s top’ (then grammaticalizing to ‘on (top of) the table’), the head of the NP is
‘the top’ and not ‘the table’.
A good illustration of the significance of the possessive construction in providing
fertile ground for grammaticalization is offered in (362), where in a subordinate clause of
seven words, the a- ‘possessive’ prefix occurs five times.
Within the NP Naka akhei atum, the first occurrence of a- on marks the
synchronic possessive construction (i.e., ‘the tribes of the Nagas’); the second occurrence
of a- is part of the plural construction with -t . The third a- occurs on the non-subject
marker - , which marks the role of the NP Naka akhei atum.
The next word is the adverbial adunghetpen ‘from very near’, which carries a-
because it has grammaticalized from relator noun -d ‘near’. Finally, the conjunction
that marks this is as a subordinate clause, t ‘because’, must also have a history of a
relator noun, to explain why a- occurs here.
311
7.4. Possessive a- Marking of Head Nouns
Head nouns with postposed modifiers are not marked in any special way. Head
nouns with preposed modifiers, however, are generally marked by a-. Those modifiers
that may occur before or after the head noun follow this rule: if they are preposed, the
head noun is marked by a-, if they are postposed, the head noun is not marked by a-.
In complex noun phrases with more than one modifier, the presence or absence of
a- on nominal elements can help analyze the underlying (hierarchical or flat) syntactic
structure.
First, consider the a- ‘possessive’ prefix in the first NP in (363), la apiso abangke
‘the wife’, which occurs on ‘wife’ due to the presence of the demonstrative as a
preposed modifier.
In the second NP in (363), la jangthu abong, there are two modifiers, the
demonstrative and the nominal possessor/modifier, t ‘oil’. The question is
whether the structure of this NP has to be analyzed as either flat, such that both the
demonstrative and the possessor modify the head noun in a linear way:
[DEM - [N]POSR - HN], or as hierarchical or embedded, such that the demonstrative
modifies the possessor rather than the possessed head noun: [[DEM - N]POSR - HN].
The answer is that it has to be analyzed as a flat structure, because if the
demonstrative were modifying the possessor rather than the head noun, the possessor
would have to have the a- prefix attached to signal that it is being modified.
It appears that the alternative, hierarchical or embedded structure of a
demonstrative-possessor-head noun sequence is the only correct analysis in (364). Here,
312
the demonstrative laso has to be analyzed as modifying the possessor Karbipi ‘Karbi
woman’ rather than the head noun osomar ‘children’, because Karbipi is marked as being
modified by a preposed element, i.e., laso ‘this’, via a-.
ba arong asarthelo
b a-r ng a-sarth -l
I.mean(<Asm) POSS-village POSS-village.headman-RL
‘a long time ago, there was one village Gaonbura, I mean, one village headman’ [CST,
RO 003]
In this example, we know that the classifier-numeral word inut modifies the head
noun gaonbura rather than the possessor, because it contains the human singular
classifier n t. It therefore cannot be ‘(the) headman of one village’, but has to be ‘one
headman of (a) village’. Nevertheless, even though the classifier-numeral word modifies
the head noun rather than the possessor ‘village’, the form still occurs with the a-
prefix, i.e., arong.
There are two implications of the previous discussion. First, an NP of the form
[[MODIF] [N]POSR [a-HN]] has to be analyzed as [DEM - [N]POSR - HN], because if the
313
possessor or modifier noun does not carry a-, then it is definitely not being modified.
Second, however, if the NP has the form [[MODIF] [ a-N]POSR [a-HN]], with the possessor
or modifier noun occurring with a-, then the underlying structure is ambiguous. It could
be [[DEM - N]POSR - HN], or it could still just be the flat structure [DEM - [N]POSR - HN].
There is more evidence that a- sometimes occurs on a noun without a preposed
modifier present. Specifically, a- typically occurs on preposed (but not postposed) PCT
modifiers, presumably because they need to be ‘more’ nominalized (and/or more
referential) when they occur before the head noun, essentially functioning as a nominal
possessor or modifier.139 An example of a preposed PCT modifier with a- is (366).
(366) NP structure: NUM - PCT - HN; preposed PCT modifier with a- ‘possessive’
laso ahut amat inut akaprek amonit
las a-h t am t [e-n t a-ke- r a-mon t
this POSS-during and.then one-CLF:HUM:SG POSS-NMLZ-be.different POSS-man
Note also that there are some exceptions to the rule that preposed modifiers
require possessive a- marking of the head noun. For example, in the noun phrase hala
y ’ t in (367), the head noun y ’ occurs without a- although is a
preposed modifier that would typically require the head noun to occur with a-. The
absence of a- here could be hypothesized to be due to y ’ being a name.
thik'etlomati […]
th k- t-l mati
be.okay(<Asm)-PRF-RL=CG
‘and then, Bey the Black and his family had already arranged everything, you know,
[…]’ [WR, BCS 018]
139
For further discussion of a- marked preposed PCT modifiers, see the discussion of examples (169) and
(170) in §5.3.1, as well more examples of preposed PCT modifiers in §7.7.1.2.
314
Likewise, in (368), the same preposed demonstrative is used without adding
a- to the head noun, although the reason here is that the head noun ‘slope’ begins
with ar-, as a result of which morphophonological fusion occurs (§3.9.2.1).
315
7.5.1. Demonstratives
7.5.2. Possessives
316
nangkelangdun do'o domestic touristta do'o
nang=ke-l ng-d n] d - ] domestic tourist t d -
CIS=NMLZ-see-along exist-much domestic tourist=ADD:also exist-much
‘there were many people who came from different countries to come and see, and there
were also many domestic tourists’ [SiT, HF 038]
317
7.5.3. Other Preposed Modifiers
pu, lata
pu l t
QUOT this=ADD:DM
'and then, the squirrel... got mad, “who is the wise one, who is the smart one, who is the
strong and mighty one, who jumped on my ladder so it broke?” it (the squirrel) (said)'
[RBT, ChM 019]
318
(377) ‘where’ functioning as nominal modifier
[ - = ]
today where POSS-market(<Ind)=FOC
‘today is which market (lit., a where-market)? / where is market day today?' (OH
121009.001)
319
non'alom ke'enthapvaret, lahai abangta chokchelo
n n-al m ke- n-th p-var t] [lah i ab ng t ch k-C -l ]
now-while NMLZ-take-mindlessly-INTENS these NPDL=ADD:also be.fine-NEG-RL
‘[…] soap, and whatever, shampoo from wherever, are now constantly used, and it is not
okay (to use) all these things (without knowing much about them)’ [SiH, CW 003]
7.5.3.2. Adverbials
320
7.5.3.3. Clausal Modifiers
This plural construction with -t often has an associated plural reading. In (385),
y ’ t refers to the family of y ’ . In (386), ’ t does not refer to a
plurality of older brothers, but to the older brother and his wife.
321
(385) Associative plural function of -t
hala Bey Ki'ik atumke lo han
[h la B y ke- k a-t m ke] [l h n
that CLAN NMLZ-be.black POSS-PL=TOP banana.leaf curry
thik'etlomati […]
th k- t-l mati
be.okay(<Asm)-PRF-RL=CG
‘and then, Bey the Black and his family had already arranged everything, you know,
[…]’ [WR, BCS 018]
chomathalone, […]
cho-math -l -n ]
AUTO.BEN/MAL-think-RL-INDEF
'the frog and the ant thinking up whatever (bad things) […]' [RBT, ChM 072]
140
See Grüßner (1978) and Konnerth (2011) (which was based on data in Grüßner (1978)).
322
difference. While both are marked morphologically the same, i.e., nominalized via ke-,141
the order of head noun and modifier appeared to mark a clear difference: PCT-based
modifiers are post-head, whereas prototypical verb-based relative clauses are pre-head.
Elicited PCT-based modifiers and prototypical verb-based relative clauses have shown
the same pattern.
It turns out, however, that this supposedly clear distinction does not hold up
against a larger body of natural data. There are instances in the corpus of pre-head PCT-
based modifiers and there also is one potential instance of a post-head relative clause.
Below I will first show examples of the more typical pattern: post-head PCT-
based modifiers and pre-head prototypical verb-based relative clauses (§7.7.1.1), and then
discuss examples of pre-head PCT modifiers and the potential instance of a post-head
relative clause (§7.7.1.2).
In (388), the PCT modifier ‘good’ follows its head noun ‘plate’ and
‘bowl’, respectively (with the idiomatic meaning ‘brass plate’ and ‘brass bowl’).
141
See Chapter IX on deverbal property-concept term modifiers and relative clauses in the context of
nominalization functions in Karbi. Also note, however, the inconsistency with which ke- actually occurs, as
discussed in §9.8.
323
In (389), the relative clause nelitum aphan nangkejapon ‘(who) took us there’
whose verb nangkejapon carries the ke- nominalizing prefix, precedes its head noun
’ ‘lady’, which is marked as being modified via a- ‘possessive’.
Thus in these examples, both PCT and prototypical verb roots need to carry ke-
‘nominalizer’, but PCT-based modifiers follow their head noun, whereas prototypical
verb-based modifiers precede them. In addition, however, a preceding relative clause
requires its head noun to occur with a- ‘possessive’, whereas a following PCT-based
modifier has an unmarked head noun. This follows the general pattern that head nouns
with preceding modifiers are marked possessive (or modified) by a-, but head nouns with
following modifiers remain unmarked (§7.4).
There also are a number of instances in the corpus in which a PCT modifier
precedes its head noun. Examples (390), (391), and (392) show pre-head PCT modifiers
that themselves are marked with the a- ‘possessive’ in addition to ke- ‘nominalizer’.
(390) Pre-head PCT-based modifiers ‘be red’ and ‘be black’, marked with a-
ke- ‘POSS-NMLZ-‘
akeer pusetame akiik pusetame
a-ke- r set m a-ke- set m
POSS-NMLZ-be.red likewise POSS-NMLZ-be.black likewise
akala
a- al r p -p
POSS-color(<Eng) give-IRR1
‘red or black color we will add’ [KST, PSu 009]
324
(391) Pre-head PCT-based modifier ‘be new’, marked with a-ke- ‘POSS-NMLZ-‘
lasonthotsi Lindata ketheklongpen aning kerong,
[las n-th t si L nd t ke-thekl ng-p n] [a-n ng ke-ar ng]]
that.way-exactly=FOC PN=also NMLZ-see-NF POSS-mind NMLZ-be.happy
(392) Pre-head PCT-based modifier ‘be good’, marked with a-ke- ‘POSS-NMLZ-‘
la'an akemesen atovar kedamtheksi
[la- n a-ke-m s n a-tov r ke-d m-th k-si]
this-that.much POSS-NMLZ-be.good POSS-road NMLZ-go-know.how-NF:RL
a'ansose nangpa'okorjangdunlonglang
a- n-sos nang pa-okorj ng-d n-l ng-l ng
POSS-that.much-more CIS=VBLZ-girl-JOIN-GET-yet
nangparisomardunlonglang
nang pa-ris -m r-d n-l ng-l ng]
CIS=VBLZ-young.man-PL-JOIN-GET-yet
‘they know how to go on a good road up to a high degree (metaphorically referring to
knowing how to do things properly, how to keep everything clean, etc.), and, because
they know how to keep everything clean and nice, those people back then up to today, get
to stay even more like girls and boys (i.e. young)’ [SiH, CW 017]
In (393) and (394), we can see that pre-head PCT-based modifiers do not,
however, obligatorily take a- ‘possessive’.
325
(394) Pre-head PCT-based modifier ‘be different’
Europe lapen kaprek kaprek adet,
Europe lap n [ke-pr k ke-pr k a-d t]
Europe(<Eng) and NMLZ-be.different NMLZ-be.different POSS-country(<Ind)
The conditions for the presence or absence of a- on pre-head PCT modifiers has
to remain a topic for further research.142 However, some implications from the difference
in ordering between pre- and post-head PCT modifiers do emerge from the preceding
examples: In the marked order of pre-head PCT modifiers, there is a greater focus on the
property rather than the noun that is ascribed the property.143 In (390) above, because two
colors are contrasted, clearly the specific colors are what is important here. In (391),
akimi ahormulo, the head noun is ‘thing’, which of course has very general
semantics and therefore the PCT ‘be new’ is the crucial information. In (392), the
’ ‘this much’ in ’ akemesen ‘this good (a way/manner)’ may be considered
evidence that the PCT is focused. In (393), the fact that the suffix -t t ‘as (much) V
as it can be’ is added to the PCT root suggests that this is the semantic head. Finally, (394)
has a reduplicated PCT ‘be different’ (and not a reduplicated ‘head noun’ t
‘country (<Ind)’), which makes it the more prominent element.
The hypothesis that pre-head order of PCT modifiers s is indeed used to focus on
the PCT semantics is supported by (395) and (396), which are subsequent lines in a folk
story, in which a present Karbi native speaking listener interrupts the storyteller from
time to time to ask for clarifications (indicated in the examples by curly brackets).
142
Post-head PCT modifiers may also be marked by a- ‘possessive’, see §9.2.
143
Grüßner (1978:123/4) also reports on two different ‘adjective‘ constructions, with differences in greater
semantic weight on either the head noun or the ‘adjective‘ (PCT modifier). However, in Grüßner’s account,
this difference is what is marked by the presence or absence of a- on the ‘adjective’. Grüßner does not
report on ordering differences between pre- and postposed PCT modifiers.
326
(395) Post-head mesen ‘be good’
“ba ko jirpo {mm} pinike ne
[ba ko j rp ] mm [pin ke n
SURPRISE(<Asm) buddy:VOC friend AFF today=TOP 1EXCL
In (395), the storyteller reports the protagonist of the story saying ‘I got to hear
good news’, where ‘good news’ occurs in the standard order of the PCT-based modifier
mesen following its head noun ‘news(<Ind)’. In (396), the Karbi speaker listening
to the storyteller wants to make sure he is following the story correctly and asks to clarify:
‘he said “good news”?!’, which likely corresponds to what in English would be stressing
the word ‘good’ to put it under focus.
Leaving behind PCT-based modifiers, there also are several examples in the
corpus which look like post-head relative clauses, such as (397) and (398). It is important
to note, however, that there are a number of internally-headed relative clauses in the
corpus (§9.3.2), and so it is not exactly clear whether we are dealing with a post-head or
an internally-headed relative clause in these instances.
327
(398) Post-head relative clause or internally-headed relative clause
ansi phelo-bisir pu Karbi atum kabonai do
nsi hel -bis r pu] [ arb a-t m] ke-bon i
then alkaline-funnel.for.filtering.ashes QUOT PN POSS-PL NMLZ-make(<Asm)
do
d
exist
‘and then, there is the so-called 'phelo bisir' (funnel-like instrument for filtering the ashes)
that the Karbi people make’ [SiH, KH 004]
Differences in head noun and modifier order are functionally exploited in some
Tibeto-Burman languages. In Angami Naga, for example, a language spoken just east of
Karbi Anglong into Nagaland, (derived) adjectives only occur post-head, whereas relative
clauses may be pre- or post-head. The functional difference is an “inherent or internal”
interpretation if post-head (like adjectives), or an “circumstantial or external”
interpretation if pre-head (Herring 1991:58).
329
(399) Typical classifier construction
laphan aning ingsamsi, thesere pumni tekanglo
[l -ph n a-n ng ings m-si] theser m-n tek ng-l ]
this-NSUBJ POSS-mind be.cold-NF:RL fruits CLF:round-two leave.for-RL
‘[…] he was grateful to him and gave him two pieces of fruit’ [SiT, PS 039]
330
abangke aseme dokoklohe
ab ng ke] a-sem d -k k-l he
NPDL=TOP POSS-oath exist-firmly-RL=DM
‘and then, (between) the two brothers, the oath remains, you know’ [WR, BCS 023]
kephopon do
ke-ph -p n d
NMLZ-reach-in.passing exist
‘[…] one the road inbetween / up to there, there were one or two tiny villages that we had
driven past’ [SiT, HF 017]
Another example from elicitation in (403) shows that in this construction, the
indepedendent numeral may be removed from its head noun. In this example, the head
noun is topicalized and left-dislocated, with the numeral in the main clause functioning
anaphorically.
331
(403) Numeral functioning anaphorically in ‘non-classifier construction’
[ ] [ t hoton=si ]
b.basket=TOP one exist but b.basket=FOC not.exist
‘as for hak bamboo baskets, there is one, but there is no hoton bamboo basket’
[Elicitation SiT 130905]
vangchomchomchomchomchom
v ng-ch m~ch m~ch m~ch m~ch m
come-a.little~ITER~ITER~ITER~ITER
'”let's go!”, and then one dog, his grandmother and his grandfather, all three, went step by
step by step’ [KK, BMS 099]
332
(405) Postposed enumerator: specifying number after story character is introduced
hako ahut hedi Bey atum korte bangkethom do tangho
hak ah t hed [B y a-t m] [ ort b ng- eth m] d t ngh
that.time during DM TITLE POSS-PL brother CLF:HUM:PL-three exist REP
'in the old days, you know, right?, there were three Bey brothers, they say' [WR, BCS
001]
(408) Two juxtaposed classifier-numeral words: inut banghini ‘one or two (people)’
[…] Boithalangso along neli inut banghini amonit
Boithalangs a-l ng n -l [[e-n t b ng-hin ] a-mon t
TOWN POSS-LOC 1EXCL-HON one-CLF:HUM:SG CLF-two POSS-man
334
(410) Two juxtaposed classifier-numeral words: bangpho bangthrok ‘many (people)’
o nelimena amatta lapu'an bangpho
o [ne-li-men am t t lap - n [b ng- h
AFF 1EXCL:POSS-HON-self self=also like.this-up.to CLF:HUM:PL-five
335
(412) ‘Another’ additive construction with ‘now’
non ejon nangalang, neta {mm}
n n e-j n n ng-j -l ng n t mm
now one-CLF:animal need-IRR2-still 1EXCL=ADD:also AFF
‘one more I need’ [HK, TR 067]
336
(415) - t ‘one-CLF:HUM:SG‘ as head noun ‘somebody’ of relative clause
[…] nangphan nangkelang inut do nangji […]
[nang- h n nang=ke-l ng e-n t d n ng-j
you-NSUBJ 1/2:NSUBJ=NMLZ-see one-CLF:HUM:SG exist need-IRR2
‘[…] there needs to be somebody to look after you […]’ [SH, CSM 066]
pacho'et pajun'et
pa-ch - t pa-j n- t
CAUS-eat-PRF CAUS-drink-PRF
'I myself also after feeding this (lapu'an) many children and children's relatives, the
whole family, […]’ [KaR, SWK 067]
Lastly, postposed ‘one’ enumeration can be also be used to indicate ‘the same’. In
the two examples (417) and (418) from the same story, rong isi means ‘the same village’.
337
sarpita ajo mek janglo, bamonpita
sarp t a-j m k j ng-l ] [bam n-p t
old.woman=ADD:DM POSS-night eye fall-RL wise.person(<Ind)-female=ADD:DM
abangta dunjuilo
ab ng t d n-j i-l ]
NPDL=ADD:DM join-away-RL
‘and then people from the same village, at the time when she died, it happened right at
that moment, and then also that old woman slept at night, the bamonpi slept, and then she
had gone along (with the person from the village who has died)’ [KK, BMS 118]
There are two slots for noun phrase clitics as shown in Figure 18. The first slot
contains the comitative/instrumental/ablative clitic =pen (§7.8.1) and the nominal
quantifier clitic = (§7.8.2), whereas the second slot contains information structure
clitics including additive =t (§7.8.3). Note that Figure 18 only lists the NP (as described
in the previous sections of this chapter), role markers, and the clitics; another element not
listed here is the noun phrase delimiter (§10.5), which, however, may occur before or
after the role markers and is therefore not easily represented in a linear fashion.
The clitic =pen has comitative and instrumental, as well as ablative functions. In
(421) and (422), comitative participants are marked by =pen. In (422), this comitative
participant is additionally projected by the verb being marked by - ‘with’ (§6.5.5.2.2).
The comitative function of =pen also underlies its occurrence in coordination
constructions (§7.1.2.1.2).
339
(423) Instrumental NP marked by =pen; verb marked with - ‘with’
motorsaikel=pen=si t - -
1EXCL motorcycle=with=FOC market(<Asm) NMLZ-go-with
'I went to the market on a motorcycle' Elicitation SiT 090223
In (424), =pen marks a spatial ablative adverbial, ‘from here’, while in (425),
=pen marks a participant (arguably) projected by the verb chekak ‘part (from somebody)’.
In the same example (425), the first adverbial also occurs with =pen, here indicating a
temporal ablative: ‘from today on’.
The case illustrated with (427) (of which there are other similar examples in the
copus), however, is not as clearly connected semantically to the other ablative examples.
The temporal adverbial adap ‘morning’ here refers to a point in time, rather than a time
340
period for which ablative notions such as ‘from … on’, or ‘since’ more readily make
sense. The occurrence of =pen here could be explained, however, in that the point in time
was in the past, hence there is still an implied time period, which would be from the
moment in the past when the event happened until the present.
341
(429) Quantifying clitic = ‘this much; all’ directly following NP
methan atum'anke abangke ha kasu keme harlung
[[meth n a-t m= n=ke] ab ng ke] h [kas ke-m harl ng
dog POSS-PL=all=TOP NPDL=TOP over.there plate NMLZ-be.good bowl
kemepen langta
ke-m pen] [l ng t j n-l ng n t ch -l ng,]
NMLZ-be.good=with water=ADD:COORD drink-GET rice=ADD:COORD eat-GET
(430) Quantifying clitic = ‘this much; all’ in modifier construction with ‘exist’
alang kepon athesere do'anta klolaplo
[[[al ng ke-p n] a-theser ] d = n t ] kl -l p-l
3 NMLZ-take.away POSS-fruits exist=all=EXH fall-completely-RL
‘all of the fruit that he was taking away fell out’ [SiT, PS 030]
342
In (432), = attaches to the demonstrative ‘this’, parallel to the interrogative
amount or degree adverb ’ ‘how much’ (§4.5.4).
7.8.3.1. Additive =t
Karbi =t occurs in contexts that overlap with English ‘also’ or ‘too’, as well as
in a number of other contexts. Nevertheless, from a cross-linguistic point of view, =t
covers an expected functional range for an additive particle. For a more detailed
discussion of =t , see Konnerth (forthcoming).
Karbi =t functions like other additive particles to "express that the predication
holds for at least one alternative of the expression in focus." (Krifka 1998)
In (433), teke atumta ‘the tigers (also)’ occurs with the additive particle, and the
‘alternative expression’ for which the predicate Rongker pu do ‘celebrate (lit., have) the
Rongker’ holds as well is culturally implied, as the Rongker is a common festival among
the Karbis.
343
(433) =t meaning 'also' with topic marker =ke in the same clause
hako arnike… teke <pu> atumta Rongker pu do tangho
hak arn ke [tek <pu> a-t m t ] [R ngk r pu] d t ngh
that.time day=TOP tiger QUOT POSS-PL=ADD FESTIVAL QUOT exist REP
‘at that time (in the old days), tigers also (like humans) celebrated the Rongker [HK, TR
035]
344
7.8.3.1.4. Scalar Additive ‘Even’
(438) =t ‘any’
ansi elitum pakrengdunpo, anke horpentame
nsi e-li-t m pa-kr ng-d n-p nke h r pen t m
then 1PL.INCL-HON-PL CAUS-be.dry-JOIN-IRR1 and.then liquor=with=any
345
jattame ingti patippo tokklingpo
j t t m ingt pat p-p tokkl ng-p
GENEX=any salt mix-IRR1 pound.until.tight-IRR1
‘and then, we need to dry it, and then either with liquor or something else we need to mix
it with salt and pound until it's tight/crushed to a paste’ [SiH, KH 013]
Like several other clitics, =t also occurs in a copy verb construction (§12.2.3.2),
in which a preposed verb copy is marked with the clitic, as in (439). With =t , the copy
verb construction functions as an intensifier construction in declarative clauses.
346
CHAPTER VIII
This chapter discusses predicate constructions beyond the simple verbal predicate.
It offers an overview of monoclausal, single-event predicate constructions. Clause-
combining constructions are discussed in Chapter XI.
8.1. Overview
Most commonly, the predicate consists of at least one verb root, including both
prototypical verb roots and property-concept term roots (§4.2). This chapter gives an
overview of complex predicate constructions that (appear to) involve more than one word.
Specifically, §8.2 offers a discussion of markers at the monoclausal endpoint of the
complementation scale in Karbi, which may or may not be grouped with the
heterogeneous category of predicate derivations (§6.5). In §8.3, adverbial constructions,
including non-final marked constructions, are discussed. Two periphrastic constructions
based on copulas are the topic of §8.4. In §8.5, a complex motion construction is
mentioned, which requires further study. Finally, §8.6 outlines the various predicate
constructions that involve noun-verb pairs.
The remainder of this section is dedicated to a brief note on non-verbal predicate
constructions and on verbal and nominal predicate negation.
As discussed in §4.1.2, §4.1.3, and §4.3, items from a large range of different
syntactic categories can function, just as they are, as the predicate of a clause. As shown
in those above sections, this includes nominal stems, adverbs, interrogative pronouns and
adverbs, and interjections, and it is not clear that any particular element could not
function as a predicate in Karbi.
An example of a noun phrase functioning as the predicate is (440); for further
examples, see the respective sections in Chapter IV.
347
(440) Noun phrase functioning as predicate
halata kidukthektik amonitlo […]
[h la t ]NP [ke-d k-thekt k a-mon t-l ]PRED(<NP)
that=ADD NMLZ-be.poor-as.much.V.as.it.can.be POSS-man-RL
‘that one also is an unimaginably poor man’ [HK, TR 128]
The nominal negation construction can also be used to negate nominalized rather
than just nominal predicates, as in (443).
348
However, the verbal negation construction can also be used with nominal
predicates, as in (444), where a derived indefinite pronoun ‘something’ is derived from
interrogative ‘what’, then reduplicated, and then negated with - .
” { }
ko j rp pu mm
buddy:VOC friend QUOT AFF
'”and there won't be anything (any difficulties, problems, dangers), my friend?”' [HK, TR
140]
And the nominal negation can also be used with verbal predicates, as in (445).
totdamlunchot kali
t t-d m-l n-ch t al
squat-go-big:AO-only NEG.EQU.COP
‘[…] maybe there's a field hut or a tree house (i.e., places to rest and sit), but it's not
about just sitting around there' [KaR, SWK 075]
The difference between verbal and nominal negation is therefore not a strictly
syntactic one. It can instead be exploited for pragmatic purposes. While further research
is required to work out the details of the pragmatic functions of these different negation
strategies, it appears that nominal negation is generally more emphatic, while verbal
negation is specifically non-emphatic.
Note also that the different negation constructions can be used together with
differences in scope. In (446), the verbal negation construction has scope inside the
participant nominalization: kechokche therefore means ‘the one who is not good’ or ‘the
guilty one’. The nominal negation with can then be used to negate the predicate:
kechokche kali ‘(be) not the guilty one’.
349
(446) Verbal negation inside participant nominalization, nominal negation of
predicate
"pot nele kechokche kali […]
p t [n le ke-ch k- al
reason 1EXCL=FOC:IRR NMLZ-be.fine-NEG NEG.EQU.COP
therefore, I'm not the guilty one, go and ask the pig, you're the only one (who can ask and
find out), he said it again [RBT, ChM 052]
8.2. Modal and Other Markers at the Monoclausal End of the Complementation
Scale
8.2.1. Overview
350
morphophonological evidence that can contribute to the problem is outlined in §8.2.4. In
§8.2.5, then, the more common ones of these markers are discussed individually.
This section thus aims to highlight the morphosyntactic properties of these
markers, of which especially (-)n ‘need, must’, (-) ‘get’, - t ‘know how’, and
- ‘be able’ are important modals in the language (other markers with similar
properties are discussed in §8.2.5.5). The goal is to highlight their particular
morphosyntactic properties vis-à-vis (non-modal) predicate derivations (as well vis-à-vis
biclausal complementation constructions). Note, however, that it might ultimately be
most practical to consider them part of the (already heterogeneous) category of predicate
derivations.
While the exact details of the complementation scale in Karbi are outside the
scope of this dissertation, four points on this clause union continuum are illustrated in the
following discussion. At the biclausal endpoint, we have a fully finite complement clause
marked with the quotative complementizer pusi, as in (447).
t
a-so-m r ab ng math -l
POSS-child-PL NPDL think-RL
‘the children thought that he had already gone to the market’ [HI, BPh 010]
351
phatang abang vansi... la phatang saikel along
[phat ng ab ng] v n-si... [[l phat ng] [saik l al ng]
kd.basket NPDL bring-NF:RL this kd.basket bicycle(<Eng) LOC
One step closer to clause union, (449) offers an example of the complement-
taking verb ‘start’. The complement clause verb ‘go’ directly precedes
the main verb, but is marked dependent via nominalization with ke-.
Finally, at the mono-clausal end of the continuum, we have the markers of interest
in this section. They occur under the same intonation contour as the preceding main verb
stem, for example - t ‘know how’ in (450). Compared to predicate derivations such
as - - ‘JOIN’, however, which occurs both in (450) as well as in (451) below, the
modal markers show some signs of morphosyntactic independence. This is discussed in
the next section.
352
anborpenpen chongho abang ingnidunpret
n-b r p n~p n chongh ab ng ingn -d n-pr t
rice-wrapped.bundle=with~DISTR.PL frog NPDL sit-JOIN-acting.on.inflated.object
‘[…] and as (the ant) was passing through between the frog’s legs, with all its rice
bundles, the frog sat down (pressing down the rice bundles)’ [RBT, ChM 016]
There are three morphosyntactic tests we can use to better describe the structural
properties of the modals. While the first test shows the relative degree of morphological
boundedness of these markers, the other two tests provide evidence for their
morphological independence.
First, markers such as - t ‘know how’ are under the scope of negation of the
main verb, which is evidence for their morphological boundedness. In (452), the
adverbial construction kemesenpen kechothek (see §8.3.4.1) is embedded into a
nominalized adverbial clause headed by - ‘POSS-reason’ and therefore nominalized.
(452) - t -)t ‘know how’ under scope of nominalization along with main
verb root
kemesenpen kechothek ajoine apotsi
[ke-m s n-pen ke-ch -th a-join a-p t si
NMLZ-be.good-NF:with NMLZ-eat-know.how POSS-reason POSS-because=FOC:RL
‘it's the reason why they know how to eat nicely’ [SiH, CW 022]
kedam kechenglo
ke-d m ke-ch ng-l
NMLZ-go NMLZ-begin-RL
‘after the car came, we started going from here to Hongkram’ [SH, CSM 008]
353
8.2.3.2. Follow Adverbial Construction [V] main verb [pa-V] adv?
Part of the evidence that the modal markers of interest have some
morphosyntactic independence is that they can modify complex adverbial constructions.
As we can see in (454), ‘need’ can follow and modify the adverbial construction
kan pame ‘dance well’.
Finally, there is evidence for some of the modals that, within the same predicate,
they may be followed a predicate derivation suffix that modifies the modal rather than the
main verb root. An example is (456), where - ‘unintentionally’ modifies the ‘non-
control’ marker (-) ‘GET’ (see §8.2.5.2.1 for the functions of (-) ) rather than
‘take’, and - couldn’t, in fact, occur without (-) ‘GET’.
145
For example, following low and mid tone verb roots such as ‘go’ and ‘eat’, the mid tone form
-t occurs: -t and -t . However, after a high tone verb root such as ‘(take) rest’, the
low tone form - t occurs: -t .
355
Similarly, the mid tone form of the modal - ‘GET’ only occurs in
conjunction with a preceding verb stem (of any tonal specification), while the high tone
form of the independent verb ‘get’ occurs if there is no other verb stem.
This tonal distinction between morphosyntactically independent and bound forms
provides evidence that certain constructions are ambiguous: for example, in (458), -
‘GET’ modifies an adverbial construction, and according to my language consultant, both
bound, mid tone (- ‘GET’ and independent, high tone ‘get’ are used here.
(Interestingly, my language consultant feels that the high tone form is grammatically
more correct, but that either one is acceptable.)
The deontic verb (-)n ‘need, must’ (for argument structure properties, see
§10.2.2.2) may occur under the scope of nominalization of the main verb stem as in (459),
although this construction is not perfectly natural to my language consultant as indicated
with the question mark. This is the case both with and without the addition of - ‘irrealis2’
following - .
(459) - ‘need, must’ under scope of nominalization along with main verb root
? [[ - - - ] ng] -
NMLZ-dance-need-IRR2 NPDL go-RL
‘the one who needs to dance left’ [SiT 140127]
356
In (460) and (461), - modifies an adverbial construction rather than a
simple main verb stem.
nangji
n ng-j ]
must-IRR2
‘and then, people need to stay and live in a clean way’ [SiH, CW 001]
However, note also that in (462), (-) ‘need’ occurs inside the adverbial
construction keklem parsik ‘do thoroughly’ (which is embedded in a relative clause that
precedes its head noun ‘work’). Specifically, (-) occurs on the nominalized
main verb. This shows that the constructions that (-) occurs in are flexible, rather
than syntactically fixed.
357
(463) (-) ‘need’ modified by following predicate derivation - ‘absolutely
required’
{duma mongponbom nangkokjike} […]
dum m ng-p n-b m n ng- -j ke
tobacco smoke-in.passing-CONT need-absolutely.required-IRR2=TOP
‘it's necessary to keep smoking […]’ [HK, TR 074]
8.2.5.2.1. Function
kemepen langta
ke-m pen] [l ng t j n-l ng n t ch -l ng,]
NMLZ-be.good=with water=ADD:COORD drink-GET rice=ADD:COORD eat-GET
(468) - ‘GET’ under scope of nominalization along with main verb root
[[ - - ] ng] -
NMLZ-dance-GET NPDL go-RL
‘the one who got to dance left’ [SiT 140127]
360
8.2.5.3. Skillful Ability - t -)t ‘know how’
The ‘skillful ability’ marker - t -)t ‘know how’ also does not get
separately nominalized but is under the scope of negation of the preceding main verb root
in (472) and (473).
(472) - t -)t ‘know how’ under scope of nominalization along with main
verb root
[[ - -t ] ng] -
NMLZ-dance-know.how NPDL go-RL
‘the one who knows how to dance left’ [SiT 140127]
(473) - t -)t ‘know how’ under scope of nominalization along with main
verb root
kemesenpen kechothek ajoine apotsi
[ke-m s n-pen ke-ch -th a-join a-p t si
NMLZ-be.good-NF:with NMLZ-eat-know.how POSS-reason POSS-because=FOC:RL
‘it's the reason why they know how to eat nicely’ [SiH, CW 022]
In (475), - t -)t (and not the main verb ‘speak’) is modified by the
following (reduplicated) predicate derivation - ‘little bit’.
361
(476) t ‘be skilled, knowledgeable’ as an independent verb root
lasi la Hingchong musoso atomoke
[las l Hingch ng musos a-tom ke
therefore this CONSTELLATION 2.siblings.of.diff.gender POSS-story=TOP
anke nangpekengpon'iknoi
[ nke nang pa-k ng-p n- k-n i]
and.then 1/2:NSUBJ=CAUS-be.straight-take.away-FRML-INFRML.COND.IMP
‘thus, they settled down and lived together until the end of their lives, and then that was
the story of Hingchong musoso, Mister Rongphar, I'm not an expert, and then make it
clear (i.e. correct it)’ [CST, HM 120]
The morpheme (-) - ‘be able’ is about physical or emotional ability rather
than skillful ability. An example from an on-line narration of the pear story is (477),
where the boy on the bike has an accident as he is riding over a large stone.
t
arl ng al ng t ng-d r
stone LOC collide-sound
‘and then he went just a bit further and then he couldn't hold himself anymore, he hit
one stone’ [SiT, PS 029]
(478) (-) - ‘be able’ under scope of nominalization along with main verb root
[[ke-k - ] ng] -
NMLZ-dance-be.able NPDL go-RL
‘the one who can dance left’ [SiT 140127]
362
In (479), (-) - ‘be able’ modifies the adverbial construction kan pame
‘dance well’.
(481) - - ‘dare’ under scope of nominalization along with main verb root
[[ - - i] ng] -
NMLZ-dance-dare NPDL go-RL
‘the one who dares to dance left’ [SiT 140127]
363
(482) - - ‘dare’ following an adverbial construction
n [[ pa- ] - ]
rice/food eat CAUS-be.good dare-NEG
‘(s/he) didn’t dare to eat well’ [SiT 140127]
The verb t ‘forget’, on the other hand, while also occurring under the scope
of nominalization of - ‘NMLZ-dance’ in (483), cannot modify an cho pame ‘eat well’
in (484) - which may, however, be a semantic rather than morphosyntactic problem here;
further research is required.
(483) - t ‘forget’ under scope of nominalization along with main verb root
[[ - -t ] ng] d -
NMLZ-dance-forget NPDL go-RL
‘the one who forgot to dance left’ [SiT 140127]
364
Finally, a verb with modal semantics as well but with a higher degree of
morphosyntactic independence is t ‘desire’, which may be individually
nominalized as shown in (486).
than kangtung
th n ke-ingt ng
tell NMLZ-desire
'I want to tell just a little bit about when we went to Kohima yesterday' [SiT, HF 001]
8.3.1. Overview
There are two proper adverbial constructions (ADVCs) in Karbi, as well as two
non-final adverbial constructions (NF-ADVCs) that may function and be structured
similarly to the proper adverbial constructions, and are therefore discussed here, too. All
four constructions consist of a main verb and a derived adverbial element. In the two
proper ADVCs, the adverbial element consists of a property-concept term (PCT; see §4.2)
verbal stem; it always follows the main verb, as seen in Table 107.
In the ‘Causative ADVC’, the main verb is unmarked, while the adverbial element
consists of a causative pa- marked PCT stem. In the ‘Nominalization ADVC’, the main
verb is nominalized with ke-, while the adverbial element consists of a bare stem PCT.
There is no apparent functional difference between these two proper ADVCs. Both occur
in elicitation when the task is to translate from English into Karbi.
365
In addition to these two proper ADVCs, there are also two non-final adverbial
constructions (NF-ADVCs), which resemble proper ADVCs functionally and structurally.
As seen in Table 108, in NF-ADVCs, there also is an adverbial element that occurs
juxtaposed to the main verb and modifies it like an adverbial. However, since in NF-
ADVCs, the adverbial element is a non-final marked verb, it has to precede the main verb.
While the NF-ADVC with -pen appears to be used with both PCT and prototypical
(non-PCT) roots in the adverbial element, the NF-ADVC with -si ‘non-final:realis’ is so
far only attested to occur with prototypical verb roots in the adverbial element.
NF-ADVCs thus resemble the proper ADVCs, but they of course also resemble
clause chaining constructions (§11.2.1.1).
In the ‘Causative ADVC’, the main verb is unmarked, while the following
adverbial element consists of a causative pa- marked PCT root and may take any
inflectional suffixes. This is shown in (487), where the main verb ‘eat’ is followed by
the causativized PCT root ‘be good’, which further has the realis suffix - attached.
Finally, there are a small number of instances in the corpus in which it looks like
a causativized PCT root occurs in a causative adverbial construction with an ellipsed,
contextually retrievable main verb, as in (489). Here it is plausible to interpret the
causativized ‘be late’ as the adverbial element of the ellipsed main verb ‘come’
mentioned in the previous clause.
367
Another example is (491), where the nominalization adverbial construction occurs
inside a nominalized adverbial clause headed by - ‘POSS-reason’ (as well as an
elaborate expression construction ‘live’ < ‘take’ and ‘eat’). The fact that the PCT
stem t r is nominalized as well (as opposed to being under the scope of
nominalization of the ke- on the preceding main verb) represents evidence that we are
dealing with two separate grammatical verbs.
Non-final -pen and -si may occur on a nominalized verb in order to function as an
adverbial element preceding the main verb.
In this non-final adverbial construction, the adverbial element takes -pen ‘non-
final:with’. In (492), the adverbial element is furthermore nominalized with ke- (while the
main verb -t ‘know how to eat’ is also nominalized due to being in a nominalized
adverbial clause construction headed by - ‘POSS-reason’).
368
are quoted: In (a), this construction with the preposed adverbial element marked with pe-
‘causative’ and -pen ‘non-final:with’ is illustrated. In (b), a semantically equivalent
version, the causative adverbial construction is employed (§8.3.2).
(493) Preposed adverbial construction with pe- ‘causative’ and -pen ‘non-final:with’
(a) - - -pen t -
CAUS-be.clear CAUS-be.good-NF:with tell-RL
‘(s/he) told clearly and well’
(b) t - - -
tell CAUS-be.clear CAUS-be.good-RL
‘id.’ (Grüßner (1978:89)
Finally, consider (494), in which nominalized, non-final -pen marked verbs do not
constitute a single clause with the final verb damlo because of the lack of a single
intonation contour as well as the intervening adverb lason ‘this way’. Also, note that the
non-final verbs here are all prototypical verbs rather than PCTs. This example also
demonstrates the gradient nature of constructional categories.
(494) Clause chaining construction with non-final verbs marked with ke- ‘NMLZ’
and -pen ‘non-final:with’
[…] amatsi netum chepenangpen kangnekpen
am tsi ne-t m che-pen ng-pen ke-ingn k-pen
and.then 1EXCL-PL RR-make.fun-NF:with NMLZ-laugh-NF:with
Lastly, prototypical verbs (i.e., non-PCT roots) can be marked non-final with -si
‘non-final:realis’ and (typically) nominalized with ke- in order to express a simultaneous,
and sometimes specifically manner indicating, event. In (495), ‘choose’ occurs
369
with nominalizer ke- and non-final -si indicating the manner in which the main event
‘eat’ occurred.
kechongvailoksi kecho"
ke-cho-ingv i-l -si ke-ch
NMLZ-AUTO.BEN/MAL-choose-only-NF:RL NMLZ-eat
'and then, "how mean, mother, was it always for you like this?", "o father, like this we
(can) always eat only what we pick and choose"' [CST, RO 037]
In (496), ‘put on stove’ (here: ‘cook tea’) is nominalized and marked non-
final with -si, indicating simultaneity with the main verb t t ‘squat’.
abangke
ab ng ke
NPDL=TOP
‘[…] outside the Karbi house, preparing himself tea, he sat there, the husband' [SeT,
MTN 040]
Finally, (497) suggests that the non-final marked adverbial element does not need
to be nominalized, as ‘ride’ clearly indicates the manner in which the directional
motion verb occurred, but there is no ke- prefix on . Further research is
370
required for a more detailed account of non-final marked clauses functioning as adverbial
elements.
A -si non-final marked main verb may be followed by the existential copula to
form a construction with a progressive reading. An example is (498), which is from an
on-line narration of the Pear Story. Since the speaker is telling the story as he is seeing it
unfold in the video clip, there are many occurrences of progressive and imperfective
constructions in this text.
371
juirekraksi do
j i-re r -si d
play-silently~DIST.PL-NF:RL exist
‘nearby, there were a few children, they were there playing silently’ [SiT, PS 031]
The beginning of the same text also serves as a good example to show the use of
imperfective ke- (§9.7.3.2) in the same context as the - construction (499).146
A corpus example is (501), where the - quantification suffix on the copula has
scope over the subject of the nominalized clause, ’ atum ‘the women’.149
148
While it is the simpler analysis to consider monit kevang an event nominalization, it is technically also
possible to consider it an internally headed relative clause, with monit as the head noun.
149
Despite being structurally different from English due to the use of argument quantification predicate
derivations, there may be a similar pragmatic function of this kind of construction in both Karbi and
English, such that the nominalization construction with the existential copula serves a presentational
function (i.e., the difference between ‘many women try to get a job’ and ‘there are many women trying to
get a job’).
373
Similarly, in (502), the parsing of nangkelelesi suggests an interpretation of this
word as a participant nominalization and therefore the =si as the focus marker. Another
possibility is to interpret this as a non-final construction such that there is no focus
marker =si but the realis non-final marker -si.
There are complex motion constructions that require further research. An example
is vekponbom dambomlo in (503). The fact that - ‘continuative’ occurs twice
suggests that this is a sequence of two independent verbs, but the fact that inflectional -
‘realis’ only occurs once at the end suggests that we are dealing with a juxtaposition of
two stems.150
150
Note that this kind of stem juxtaposition also occurs in the parallelism type of elaborate expression
construction (§12.2.2.2).
374
point associated with translocative motion, so further research into their exact functions is
required to understand the implication of this construction that apparently conjoins the
two.
While (503) above illustrated a direct juxtaposition of verb stems in the complex
motion construction, (504) illustrates a case where a noun intervenes. What is
furthermore different in this example, is that while - - ‘on the way’ is involved
here as well, it is the element that is repeated across both verb stems. Note that there is
still only one final inflectional element - ‘still’ at the end of the second part junpon of
this construction.
junponlang
j n- n-l ng
drink-in.passing-still
‘and then, from here we went, we stopped by in Dimapur and just had tea’ [SiT, HF 009]
375
Table 109. Noun incorporation
Noun incorporation expression Incorporated noun Semantic role of noun
t ‘dry in sun’ ‘sun’ Force, Instrument
chinglu ‘take bath’ ‘water’ Instrument
‘be hungry’151 ‘rice; food’ ?Patient
t ‘be born’152 t ‘world’ ?Goal
(b) Klirbon [ - = = ] -
NAME water NMLZ-be.hot=with=FOC:RL water NMLZ-take.bath
‘Klirbon is taking a bath with hot water’ [BIK 110205]
The optionality of ‘water’ in (b) shows that this is still a somewhat loose
noun incorporation construction, which is also evidenced by the fact that it is not possible
to fully integrate the noun into the verb stem: it is not acceptable to say *ka-lang-chinglu
‘*NMLZ-water-take.bath’ or *ka-che-lang-inglu ‘*NMLZ-RR-water-take.bath’.
Further evidence that these noun incorporation constructions are not entirely
grammaticalized and semantically bleached comes from (506). This is from a folk story,
and there are two abandoned babies that are crying because they are hungry. The
storyteller first says an kangchir for ‘be hungry’, and then seemingly decides that there is
151
In addition to r for ‘be hungry’, one can also say -p k i r with - ‘stomach’ as the
incorporated noun, see §8.6.2.
152
For the expression t as well, there is a possessed noun incorporation construction that can be
used instead (§8.6.2). Using t ‘world’ in an expression meaning ‘be born’ also exists in German auf
die Welt kommen ‘lit.: come onto the world > be born’.
376
still too much semantics associated with ‘rice, food’ that he adds the non-final marker
-si and specifies: mok kangchir ‘be hungry (for breast milk)’.
kangchir
ke-ingch r
NMLZ-be.hungry
‘[…] the two of them (i.e., babies), were crying loudly, they were hungry, they were
hungry for milk' [CST, HM 027]
153
In a way, this type of construction can be seen as the converse to possessor-raising constructions: instead
of generalizing to an affected possessor, the psycho-collocation (and other possessed noun incorporation
with incorporated body part terms, as discussed further below in this section) specifies the ‘possessed’
locus of the impact of an event, which in psycho-collocations is - ‘heart, mind’, but may be other body
part terms.
154
Karbi - goes back to Proto-Tibeto-Burman * ŋ ‘heart’ (Matisoff 2003). Apparent cognates in
Meithei are -niŋ ‘wish to V’ and a homophonous root with the meaning ‘head/mind’ (Chelliah
1997:215;333;512).
377
Table 110. Sample psycho-collocations
Form Meaning Meaning without -
‘be happy’ n/a
‘be sad’ n/a
t ‘be mad’ <‘be short’
‘be glad, be grateful’ <‘be cold’
- ‘be calm, well-tempered’ <‘be good’
‘be troubled’ < ‘be difficult’
t ‘hope’ <‘be big’
‘want’ n/a
‘throw up’ <‘come’
378
transitively as ‘be grateful to (somebody)’ in (508). Also, in (509), the reciprocal prefix
che- is used with - ‘be sad’ to express ‘be upset with one another’.
Two of the possessed noun incorporation expressions are illustrated in (510) and
(511).
155
Beyond a notion of ‘salient involvement’, it is not possible to characterize the role of the referent with
respect to a particular semantic role (which is generally possible with nouns in non-possessed noun
incorporation expressions (§8.6.1)).
379
(510) Noun incorporation construction: - ‘wake up’
lengvaretmati, sarbura" pu
l ng-var t mati s rbur pu
be.fat.HUM-INTENS=CG old.man QUOT
‘and then he would return, "why are you so fat/healthy, man? (That's very strange!)"’
[SeT, MTN 025]
380
8.6.3. Light Verb Construction
In (516), the light verb t , which does not occur by itself, is used with the noun
‘vow’.
Cognate object constructions are those in which the verb and the object noun are
derived from the same etymological root. Examples in Karbi are ‘sing a song’
(517) and ‘place a ladder’ (518).
381
(518) Cognate object construction: ‘place a ladder’
dondon chedonsi... anke amonit abang
dond n che-d n-si...] [ nke [a-mon t ab ng]
ladder RR-place.ladder/bridge-NF:RL and.then POSS-man NPDL
382
discussed in Chapter XII: the general extender construction (§12.2.1); complex predicates
based on elaborate expressions (§12.2.2); and copy verb constructions (§12.2.3).
383
CHAPTER IX
9. NOMINALIZATION
In its perhaps most basic function, ke- derives nouns from verbs. In (520),
‘crow’ undergoes event nominalization via ke-, and then functions as a noun and
furthermore as a noun phrase, as it takes on the role of the S argument in the clause akiku
’ ‘his crowing is very nice to hear’. In addition to ke-, the a- ‘possessive’ prefix
is attached. This prefix occurs in a wide range of grammatical contexts (§5.3.1; §7.3;
384
§7.4). Here, it appears to mark the third person possessive, referring to the rooster as the
‘possessor’.
In (521), ke- functions as a participant nominalizer on the PCT root ‘be small’.
Here again, a- occurs in addition to ke-, but with a different function that in (520). Here,
a- appears to be contributing to the noun-hood or referentiality of akibisi ‘the youngest
one’ (see §5.3.1).
Example (522) shows that ke- may also occur without a- ‘possessive’ as the sole
element deriving the noun kakirla ‘change’ from the verb ‘turn over’.
dochetlo […]
d -ch t-l
exist-a.bit-RL
'but nowadays, there's some change, […]’ [KaR, SWK 064]
385
Finally, (523) and (524) are instances of nominalized verbs occurring with the
demonstrative ‘this’, which serves as an additional indicator of the noun-hood of the
event nominalizations. In (523), ’i ‘struggle’ is derived from ’ ‘try hard, make
an effort’ and in (524), the event nominalization is laso kekoi abangke ‘this rubbing’, also
featuring the noun phrase delimiter (§10.5).156
anijom asonlo
a-nij m as n-l ]
POSS-procedure(<Asm) like-RL
‘and then, like this we rub the cloth, this rubbing, I mean.. it creates a lot of foam, like
using Shree soap’ [SiH, CW 006]
156
Interestingly, a- ‘possessive’ does not occur on ’ or kekoi, even though nouns modified by
preceding elements such as demonstratives generally take a- (see §7.4 and §7.5.1). This needs to be
addressed more in future research.
386
to modify ‘plate’ and ‘bowl’. While post-head noun order is more common,
pre-head noun order also occurs, as detailed in §7.7.1.
Most PCT modifiers in the corpus are not complex and do not have any affixes in
addition to nominalizer ke-. However, it is possible to add predicate derivation suffixes
(§6.5), as in the preposed PCT modifier kidukthektik ‘unimaginably poor’ in (526).
As pointed out in §7.7.1.2, PCT modifiers preposed to their head noun are
sometimes marked by a- ‘possessive’. In the corpus, this occurrence of a- is only found
on preposed PCT modifiers but not on postposed ones. However, elicitation examples
such as (527), as well as data reported by Grüßner (1978:123-4) demonstrate that PCT
modifiers following their head noun may also be marked by a- ‘possessive’; further
research is needed to determine what functional difference there may be in adding or not
adding a- on PCT modifiers.
387
(527) PCT modifiers following their head noun, marked with a- ‘possessive’
a-ke- - a-ke- - -
flower POSS-NMLZ-look-be.good flower POSS-NMLZ-look-be.good-NEG
‘a pretty flower’ ‘an ugly flower’ [SiT 090301]
9.3. Relativization
In the standard relativization construction, the head noun occurs external to the
relative clause, with the relative clause preceding. The only instances in which a relative
clause looks like it is following its head noun (§7.7.1.2) are instances that may instead be
interpreted as being internally-headed (discussed in the next section §9.3.2).
388
9.3.1.1. Relativization on Different Clause Participants
(528) S relativization
tennis <a> kapathu abol lapenke bet
[[[tennis <a> ke-path ] a-b l lap n ke b t]
tennis(<Eng) POSS NMLZ-play POSS-ball(<Eng) and.then=TOP bat(<Eng)
389
(529) A relativization (with O-high)
Lily, la nelitum aphan nangkejapon aosopi,
Lil l [ne-li-t m a h n nang=ke-j - n]RC a-oso ]]HN
NAME this 1EXCL-HON-PL NSUBJ 1/2:NSUBJ=NMLZ-lead-take.away POSS-lady
adungan nanglelo
ad ng= n] nang l -l
near=up.to CIS=reach-RL
‘[…] dragging along a female goat, close up to this fruit picking man he reached’ [SiT,
PS 010]
390
arlosi lahai kethap lapen arum kevan
arl si] [lah i] ke-th p]] lap n [ar m ke-v n]]
inside=FOC these NMLZ-put.inside and down NMLZ-bring
'he's placed himself a ladder... and then, the man, into one bag sown from white cloth he
puts these (fruits), and then brings them down [SiT, PS 003]
There are only a few ditransitive constructions in the corpus and none involving
relativization. Therefore the above shown S, A, and O relativization constructions
represent all attested instances of the relativization of particular syntactic roles. However,
it is also possible to relativize on other types of (oblique) clause participants. For example,
(533) and (534) show that it is possible to relativize on locative NPs.
In (533), [[hala] [ka-ngni] a-dim]] is ‘that place where on sits’.
391
kirim kibi isi ahem do […]
ke-r m ke-b RC is a-h m HN d
NMLZ-put.in.one.place NMLZ-keep one POSS-house exist
‘[…] there is one house where they put everything they got (from the time of when the
British were fighting against Japan, i.e., the War Museum), […]’ [SiT, HF 056]
However, marking the verb with - is not obligatory, as illustrated in (536), where
- does not occur.
otdunno, […]
t-d n-n ]
touch-JOIN-be.bad
‘the pestle with which the rice beer cake is ground is bad to touch, […]’ [WR, BCS 037]
392
achetpen sarnung kidip ahemta
a-ch t en sarn ng ke-d RC a-h m t HN
POSS-small.piece=with roof NMLZ-cover POSS-house=ADD:also
nelitum thekdamlong
ne-li-t m th k-d m-l ng
1EXCL-HON-PL see-GO-GET
‘from the backside, how (the houses) look, from the front, how they look, and then, we
also went to see the houses, whose roofs (they) cover with slabs of stone’ [SiT, HF 048]
While the relative clause verb typically remains unmarked for aspectual or modal
categories, it is possible to add - ‘irrealis2’ for a future or irrealis reference (see also the
general discussion of irrealis-marked nominalized subordinate clauses in §9.6). For
example, in (538), the speaker refers to the matter she is going to talk about as ne kethanji
alamthe.
nangkachetongdunsi
nang=ke-chet ng-d n-si]
CIS=NMLZ-meet-JOIN-NF:RL
‘the matter that I will talk about... after meeting here at the festival’ [KK, CC 002]
Only a few sentences later in (539), however, she refers to that same matter she is
about to narrate as ne kethan atomo, without using - . This shows that the simple relative
clause verb (not marked with - ) has a wide range of default interpretations, including
irrealis contexts, and that using - is an optional way of specifically highlighting a future
or irrealis reference.
393
pachekengdamthekthedet tahai meneta
[pe-che-k ng-d m-th k-C -d t tah i] [men t
CAUS-RR-be.straight-go-know.how-NEG-PFV DUBIT maybe=ADD:even
nangthanpo
nang th n-p ]
1/2:NSUBJ=tell-IRR1
‘the story I'm telling now, maybe I can't tell it perfectly (straight), but maybe I will still
tell’ [KK, CC 008]
The noun ‘habit’ occurs in a construction in which it takes a relative clause but,
surprisingly, is additionally marked by a personal possessive prefix rather than just the
general a- possessive prefix. An example is (540).
Two analyses are possible based on interpreting the existential copula as being
more intransitive-like (‘X exists’) or more transitive-like (‘X has Y’). In the intransitive
analysis, the example needs to be bracketed as follows: [[[nangtum kachekoi] nangbe] S
doji] ‘lit., your habit of you accusing each other will exist’. In the transitive analysis, it
would instead be: [[nangtum] A [[kachekoi] nangbe] O doji] ‘lit., you will have your/a
habit of accusing each other’.
The second analysis as a more transitive-like possessive construction may be
considered preferable because it follows the basic possessive construction that requires
the ‘O argument’ to be marked possessive (§10.2.2.3.2).
394
9.3.2. Internally-Headed (or Post-Head) Relativization
395
parallelism between the two sentences, the relativized noun nangso kithike occurs in the
same subject position as the topic =ke marked NP consisting of the simple possessed
noun neso ‘my children’ in the following clause.
In (543), the construction again only consists of the head noun, i.e, the S argument
that is being relativized on, and the RC verb. The S argument is nangpiso, Karbipi
asomar, aso, where there are two appositional constructions: first, nangpiso, Karbipi
‘your wife, the Karbi woman’, and second, asomar, aso ‘(her) children, (her) child’.157
9.4. Complementation
Complement clauses (CCs) are typically nominalized, although there also are
(formally) non-nominalized CC constructions, as discussed in §11.2.2. Also see §8.2 for
a discussion of modal and other markers at the monoclausal endpoint of the
complementation scale.
Examples (545) through (549) illustrate complement clauses functioning as O
arguments of complement-taking verbs and , both meaning ‘start’; t,
t , and t , all meaning ‘finish’ or ‘conclude’; and - ‘see-JOIN’ > ‘watch’.
158
This notation indicates that this is a complement clause (CC) functioning as the O argument of the
matrix verb.
397
(546) Nominalized complement clause of ‘start’
t t
gar pen v ng-d t aph si ne-t m d en ong r m
car(<Asm)=with come-PFV after=FOC:RL 1EXCL-PL here=from PLACE
kedam kechenglo
ke-d m ke-ch ng-l
NMLZ-go NMLZ-begin-RL
‘after the car came, we started going from here to Hongkram’ [SH, CSM 008]
399
(552) Complement clause functioning as S argument
neli kachoklemke Habepi ahabekongiklo
[n -l ke-cho- l m e] CC (>S) [Habep a-habek ng- k-l ] PRED
1EXCL-HON NMLZ-AUTO.BEN/MAL-do=TOP DISTRICT POSS-main.headman-HON-RL
'what I work as is Habekong of the Habepi district (of Rongkhang)' [SeT, MTN 003]
Finally, in (553), the complement clause laso ahormu abangke kadokave akheita
kacharlidun is marked as a topic with =ke and functions pragmatically as a conditional
(i.e., ‘it would be good if everybody learned this’) (see Haiman (1978) for a discussion of
the functional similarity between conditionals and topics).
400
9.4.2.1. Irrealis-Marked Complement Clauses with Purpose/Non-Subject Marker -
401
9.4.2.2. Irrealis-Marked Complement Clause with Noun Phrase Delimiter
Besides , the noun phrase delimiter can also mark the right edge of
an irrealis complement clause. While inherently has the semantics of marking a
goal, purpose, or intention and also marks adverbial clauses with that semantic range (see
§9.5), represents a structural-only, semantically empty marker of the right edge of
the noun phrase (see also the discussion in §10.5).
An example is (556), where the irrealis CC functions as the O argument of
t ‘know how’.
402
9.5. Adverbial Subordination
403
(559) Non-nominalized anteriority adverbial clause
, netum dakpen Hongkram
[gar pen v ng-d t aph si] ne-t m d k pen Hongkr m
car(<Asm)=with come-PFV after=FOC:RL 1EXCL-PL here=from PLACE
kedam kechenglo
ke-d m ke-ch ng-l
NMLZ-go NMLZ-begin-RL
‘after the car came, we started going from here to Hongkram’ [SH, CSM 008]
In the following discussion of semantic types, all those constructions are included
that are based on a relator noun-derived subordinator, even though in some instances, the
adverbial clause verb is no longer nominalized with ke-, as discussed in the preceding
section §9.5.1 (but see also the discussion in §9.8 on the overall inconsistent occurrence
of ke- on nominalized subordinate clause verbs). As shown in Table 112, nominalized
adverbial subordination covers a wide-range of interclausal semantic relations, including
the categories of place, time, and causality, as well as several other ones. The
terminology for the semantic types listed in Table 112 is from Kortmann (1996:138),
except for the last type, ‘topic’.
Table 112 shows the relator noun-derived subordinators used to express the
particular semantic types of adverbial clause, as well as lists references for the examples
that illustrate each type.
404
There is one subordinator, ‘before’, which is different from the other
subordinators in that it requires the adverbial clause verb to be negated. Obligatory verbal
negation in posterior adverbial clauses is typical in Tibeto-Burman; it exists in languages
as diverse as Kurtoep (Hyslop 2011:633-4), Galo (Post 2007:828),159 and Burmese
(Konnerth 2008).160
In the following discussion, each semantic type of nominalized adverbial
subordination is illustrated by an example from the corpus, as referenced in Table 112.
In (560), a (non-nominalized and negated) locative adverbial clause is marked by
‘locative’.161
159
According to Post, there is no form in Galo that directly expresses posteriority. The form quoted here to
occur with obligatory verbal negation is a “combination of subordinate clause predicate negation and
achievement marking”, which, however, according to Post, comes closest to expressing posteriority in
Galo.
160
There is, of course, also a clear functional motivation for negating a ‘before’ clause, since the inherent
nature of this event type consists in not being realized (yet) in relationship to the event expressed by the
main clause.
161
The relator noun - that this subordinator is derived from covers a wide range of semantics (§4.4.4.1),
and also represents one of the two basic role markers (§10.6.3).
405
In (561), ‘after’ marks an anteriority relationship to the following main
clause event.162
162
The relator noun - is also used in the locative sense of ‘back(side)’ (§4.4.4.3).
163
The same root - or - for ‘time’ appears to be part of the word , with the distal demonstrative
formative (§4.5.3) being the other part. This word occurs in the folk story introductory phrase hako
ahut, translatable as the English fairy tale introductory phrase ‘once upon a time’.
406
nangthondunnoi pulo tangho
nang th n-d n-n i p -l ] t ngh
1/2:NSUBJ=drop-JOIN-INFRML.COND.IMP say-RL REP
'<when we came, I followed you along far away, I now won't find my way back, (so)
come along and drop (the bamonpi)!>, (the bamonpo) said, as they say' [KK, BMS 097]
chetonglok […]
che-t ng-l k
RR-meet-happen.to
‘[…] while he was looking for more, he happened to run into the orphan […]’ [HK, TR
048]
407
(566) Causality adverbial clause with t ‘because’
[…] thoisi kedo apot rit pine akam
th i=si ke-d a t CAUSE [r t p -n a-k m
plains=FOC:RL NMLZ-stay because jhum.field what-INDEF POSS-work
408
sarbura atum ha ritlo vopi
s rbur a-t m h r tl v -p
old.man POSS-PL over.there inhabited.field chicken-female
chopangrengdamlo
cho-pangr ng-d m-l
AUTO.BEN/MAL-rear-GO-RL
‘[…] once upon a time, because they had many children, the old woman and the old man
went over there to the field, to rear hens' [SeT, MTN 004]
Purpose adverbial clauses are marked by , which has the same origin as the
non-subject marker -ph (§10.6.2). An example of is (569), where it is embedded
into a simultaneity duration adverbial clause marked by t.
409
thangbaksi keleduntam thekji ason
th ngb si ke-l -d n-t m th -j as n] COMPAR/SIMIL
as.if=FOC:RL NMLZ-reach-JOIN-impossible know.how-IRR2 like
nangpinkhattap nangpinkhatphru
nang pinkh t-t p nang pinkh t-phr
1/2:NSUBJ=advise-here.and.there.1/2 1/2:NSUBJ=advise-here.and.there.2/2
'since we are alive, (how can) you give so many pieces of advice as if we could reach (the
place where my wife has gone after she died)' [KK, BMS 031]
164
Compare this construction to complementation with the noun phrase delimiter discussed in
§9.4.2.2, as well as irrealis-marked nominalized subordinate clauses discussed more generally in §9.6.
410
(572) Topical adverbial clause with ‘NPDL’
“t t t
las t t ngt ne- h a- h n
therefore OK if 1EXCL:POSS-grandmother POSS-NSUBJ
t t ”
nang=ke- a-chet ng-j ab ng e pa-the' ng-sin ng
1/2:NSUBJ=NMLZ-CAUS-meet-IRR2 NPDL=TOP CAUS-be.bright-si.HORT
'and then, "okay then, in order to make you meet my grandmother, let's wait a little while
until it's bright (lit., make it bright)"' [KK, BMS 062]
The verbs inside nominalized subordinate clauses typically only consist only of
the stem with nominalizer ke-. The one exception is that subordinate clauses of all three
types (relative, complement, and adverbial clauses) may be marked with - ‘irrealis2’. If
marked irrealis, the meanings expectedly change to intentional futures, purposives, or
other types of hypotheticals.
Relative clauses marked by - may indicate an intentional future event pertaining
to the head noun, as in (573), where ne kethanji alamthe refers to ‘the matter that I will
talk about’ or ‘want to talk about’. Note that this marking is not obligatory in the case of
future reference; as discussed in §9.3.1, a few intonation units later, another relative
clause is produced by the same speaker about the same topic (i.e., with the same future
reference), but without - .
411
(573) Irrealis-marked relative clause indicating intentional future
ne kethanji alamtheke jo dak rong'aje along
[n ke-th n-j a-lamth ke] [j d k r ng'aj al ng
1EXCL NMLZ-tell-IRR2 POSS-matter=TOP see here festival LOC
nangkachetongdunsi
nang=ke-chet ng-d n-si]
CIS=NMLZ-meet-JOIN-NF:RL
‘the matter that I will talk about... after meeting here at the festival…’ [KK, CC 002]
Similarly, in (575), the irrealis marked clause [aso mok ka-chepechu-ji] ‘(to)
breast-feed their children’ represents another nominal complement purpose construction,
here with the head noun ‘time’.
412
kachepechuji apor ave tangho
ke-che- a-ch -j a- r av t ngh ]
NMLZ-RR-CAUS-suck-IRR2 POSS-time not.exist REP
‘they just kept grinding the rice for the rice beer cake, the whole time they didn't get to
take rest, they were grinding (for) the rice beer cake, they didn't even have time to breast-
feed their children’ [WR, BCS 030]
413
(578) Irrealis-marked adverbial clause indicating purpose
phinu chojordamji aphan hu kulat
hin cho-j r-d m-j a- h n h kul t
banana AUTO.BEN/MAL-sell-GO-IRR2 POSS-PURP over.there shop(<Asm)
anat damlo
a-n t d m-l
POSS-direction go-RL
‘in order to go and sell bananas there he went towards the shop’ [HI, BPh 005]
Instead of a simple verbal predicate, there are some instances in the corpus where
speakers choose to use a nominalization construction involving the existential copula .
An example is (579), where the nominalization plus copula construction is further
embedded in a declarative intensifier copy verb construction of the structure ‘V=ke V-
suffixes’ (§12.2.3.1).
414
(579) Nominalization plus copula construction
Naka anglong pu bihek akopenta arjulong
N k a-ingl ng pu b -h k ak pen t arj -l ng
TRIBE POSS-hill QUOT be.small-small then=from=ADD:even listen-GET
dohe
d =he
exist= AFTERTHOUGHT
‘since my childhood, I got to hear about the Naga hills, the Naga people, but from both
the women and men, (only some), like that, I got to meet in a place or two’ [SiT, HF 004]
In (579), instead of kechetong doke dohe, a simple verbal predicate could have
been used in the same copy verb construction as well: chetongke chetonghe. Although we
would expect that the nominalization construction is somehow more emphatic than using
a simple verbal predicate, the pragmatic difference is if anything very subtle, as my
language consultants did not perceive a clear functional difference.
Similarly, in (580), instead of the simple verbal predicate chungkrenglo, a
nominalization construction is used: kechungkreng dolo. Oddly enough, the speaker does
not use ke- on the remaining three PCT predicates (for further discussion of this, see
§9.8).
415
kedothupo
ke-d -th -p ]
NMLZ-exist-again-IRR1
‘[…] they became thin, they became pale, and they were about to die' [CST, RO 022]
In the second line of the same example (580), t ‘die’ occurs in an elaborate
expression (EE) construction with . The verb pair further occurs in a nominalization
construction: kithita kedothupo and kejangta kedothupo, instead of thipo and jangpo.
Here, however, it seems that the nominalization construction serves the structural purpose
of allowing the speaker to coordinate the EE verb pair with the use of =t ‘additive’ on
the preposed verb copies (§11.5.1.2).
In another type of nominalization plus copula construction, there appears to be an
underlying structural ambiguity. Consider (581), where under one analysis, the
nominalized clause could be interpreted as a direct S argument of the existential copula,
i.e., [[laso adak isi hini achitchit arong kephopon] S do]. Alternatively, it could be
interpreted an internally-headed relative clause, with the head noun being the S argument
of the existential copula: [[[laso adak [isi hini achitchit arong] HN kephopon] RC ] S do].
kephopon do
ke-ph -p n d ]
NMLZ-reach-in.passing exist
‘and then we went and at one o'clock, we reached Kohim Town..., one the road
inbetween / up to there, there were several tiny villages that we had crossed’ [SiT, HF
017]
416
this analysis is therefore less preferable, all things being equal. On the other hand,
however, the RC analysis makes sense functionally, as this appears to be a kind of
presentational construction used to introduce the head noun. Consider the intonation unit
following (581), which is offered as (582).
Since this following intonation unit comments on the ‘several tiny villages’, this
suggests that the nominalization construction in (581) served the information-structural
function of introducing the ‘several tiny villages’ to the interlocutor/listener.
For another example, consider the excerpt in (583), which contains both the
intonation unit (IU) with the nominalization construction and the subsequent IU.
do
d
exist
‘and then, there is the so-called 'phelo bisir' (funnel-like instrument for filtering the ashes)
that the Karbi people make’ [SiH, KH 004]
417
Here again, functional considerations suggest that the relative clause analysis is
more appropriate than the event nominalization analysis, as this construction serves the
function of introducing the participant ‘the so-called phelo bisir’. This function can be
deduced from the next intonation unit, which comments on this participant, specifically
on how this phelo bisir is used as a tool in the process of preparing alkaline food by using
ashes.
The relative clause analysis implies an underlying structure that can be
schematically represented as: [[phelo-bisir pu] HN Karbi atum kabonai ] S do]. Note that
here, the construction is again structurally ambiguous between an internally-headed or a
post-head relative clause, as discussed in §9.3.2. It could be internally-headed:
[[[[phelo-bisir pu] HN Karbi atum kabonai] RC ] S do], or post-head relativization:
[[[phelo-bisir pu] HN [Karbi atum kabonai] RC ] S do].
In addition to the instances discussed here, the function of a presentational
construction might also be underlying the copula argument quantification construction
discussed in §8.4.2.
There are three subtypes of focus constructions. First, there is the general
argument focus construction, which may occur in all clause types including non-
418
declarative speech acts (§9.7.3.1.2). Second, specifically in content questions, the
interrogative pronoun or adverb is typically marked as being under focus (although
another element can be under focus as well, then belonging to the first type of focus
construction) (§9.7.3.1.3). Third, there is a co-relative construction that is best analyzed
as a focus construction as well, with the co-related elements across the two clauses being
under focus (§9.7.3.1.4).
All focus constructions have two structural properties in common: First, there is
an element in the clause that occurs with one of three focus clitics: mostly with =si if the
verb is realis, or with =le if it is irrealis (§11.3), but in a few cases also with =lo
(§6.9.1.7). Second, in a substantial number of occurrences (though not in all, see the next
section §9.7.3.1.1), the verb is marked with ke- without there being a copula to render the
clause finite. Since these are not imperfective constructions (see §9.7.3.2 below)
(although some may be as well, see §9.7.3.3), there has to be a different historical
explanation for why the verb is diachronically nominalized in the focus construction. The
typologically most plausible explanation is to assume that the Karbi focus construction
originates in a cleft construction. This historical development is sketched out in
§9.7.3.1.5.
419
9.7.3.1.2. General Argument Focus Construction
ki-t ]
NMLZ-cook
'I don't/won't cook pork, I (will) cook chicken' or: ‘it is not me cooking meat, it is (me)
cooking chicken’ [Elicitation SiT 090303]
Here, the first clause is negated and the element under contrastive focus is hence
marked with =le, while the second clause is asserted and the element under contrastive
focus therefore marked with =si. In both clauses, the main verb is marked with ke-.
Corpus examples that also illustrate the co-occurrence of focus marked elements
with ke- marked verbs follow: with realis focus marker =si in (585); with irrealis focus
marker =le in (586); and with focus marker =lo in (587). Again, note that these examples
only illustrate a subset of focus construction instances; more than half of the instances
occur without ke- on the verb.
(585) General argument focus construction with realis focus marker =si
amatsi itum aphanke dak habit angbongsi
am tsi e-t m a-ph n ke d k hab t angb ng=si
because 1PL.INCL-PL POSS-NSUBJ=TOP here jungle in.middle.of=FOC
nangkethonti
nang=ke-th n-t
1/2:NSUBJ=NMLZ-drop-get.rid.off
'and then, she took us here in the middle of the jungle and abandoned us’ [CST, HM 052]'
420
(586) General argument focus construction with irrealis focus marker =le
[…] “itum nangpeile kedo kalilo” […]
[i-t m nang-p i=le ke-d kal -l ]
1PL:INCL-PL 2:POSS-mother=FOC:IRR NMLZ-exist NEG.EQU.COP-RL
‘ […] “we don't have your mother anymore” […]’ [CST, RO 008]
(587) General argument focus construction with realis focus marker =lo
bang nekengdak arumloklo kedam
b ng ne-k ng-d k ar m-l k=lo ke-d m
CLF:HUM:PL 1EXCL:POSS-foot-road.inbetween down-just=FOC NMLZ-go
‘the other people simply walked through between my legs’ [RBT, ChM 015]
Interrogative pronouns and adverbs often occur with a focus clitic: mostly with
=si but sometimes also with =lo. Looking at those instances of content questions in the
corpus that have a focus-marked interrogative pronoun or adverb, about two thirds have
verbs marked with ke- in the clause. Examples that illustrate this are (588) and (589).
(Note that content questions may have another focus-marked element in addition to a
focus-marked interrogative pronoun or adverb; see the discussion of (832) in §11.3.2).
(588) Content question focus construction with realis focus marker =si
nesomar pule kosonsi thengpi abeng
ne-oso-m r pu=le os n si thengp a-b ng
1EXCL:POSS-child-PL QUOT=FOC:IRR how=FOC:RL tree/wood POSS-piece
nangketetroiroidetlo
nang=ke-t t-r i~r i-d t-l
CIS=NMLZ-exit-PL.solid.obj~DIST.PL-PFV-RL
‘if they are my children, how did they come out as pieces of wood?’ [CST, HM 023]
(589) Content question focus construction with realis focus marker =lo
ne kopilo kevipo laho <m>
n kop =lo ke-v -p l -ho
1EXCL what=FOC NMLZ-do-IRR1 this-EMPH:INTERACT
'what should I do?' [CST, HM 013]
422
historical scenario would explain how the two structural pieces of the focus construction,
i.e., the focus clitic and the ke- on the verb, go together.
(591) General argument focus construction with irrealis focus marker =le
[…] “itum nangpeile kedo kalilo” […]
[i-t m nang-p i=le ke-d kal -l ]
1PL:INCL-PL 2:POSS-mother=FOC:IRR NMLZ-exist NEG.EQU.COP-RL
‘ […] “we don't have your mother anymore” […]’ [CST, RO 008]
This account for the historical development of the focus construction in Karbi is
well supported by cross-linguistic case studies. Focus markers grammaticalizing from a
cleft construction involving a copula are attested in many languages of the world
including the Uto-Aztecan language Cora (Casad 1984); the Afro-Asiatic languages
Lamang and Rendille (Heine and Reh 1984); Japanese (Harris and Campbell 1995); and
165
The reanalysis might have been facilitated if the original copula was most commonly used in this
construction. Since a simple juxtaposition construction for equational clauses might have always been an
option, the copula would have only been used in pragmatically marked contexts, of which a cleft would be
a prime example.
423
Papiamentu (S. Kouwenberg and Muysken 1995; Holm 1988); see also Heine and Kuteva
(2002).
In addition to the typological support for this reconstruction, there is Tibeto-
Burman internal evidence as well, in particular for the =si realis focus clitic. The
evidence comes from Central Kuki-Chin languages, where there is a sii equational copula
in Hakha Lai (Peterson 2003: 424), and a si equational copula in Falam Lai (King 2010),
filling in comparative evidence for Karbi =si originating in an equational copula.
Moreover, in Falam Lai, there also is a si focus marker that occurs with interrogative
pronouns and adverbs the same way Karbi =si does (§9.7.3.1.3). Falam Lai has therefore
both the reanalyzed focus marker si as well the copula source form si, providing direct
evidence that this development has occurred in this language.
As for the other two, much less frequent focus clitics in Karbi, =le and =lo, the
evidence is not as clear. However, it could very well be that =le can be historically linked
to the second syllable in the negative equational copula , since there is comparative
evidence that the velar-initial first syllable can be linked to negative forms in Tibeto-
Burman, and the lateral-initial second syllable to copular forms (§4.6.2.2). While the
vowels do not match, this is still a strong hypothesis for functional reasons, because it can
explain the irrealis-sensitivity of the focus marker =le with the negative polarity
association of .
An investigation into the historical origin of =lo will need to consider realis -
(§6.9.1.7), but whether both forms ultimately reconstruct back to an equational copula is
not clear at present.
424
imperfective construction. Examples of ke- occurring on PCT-based verbs are offered in
(592) and (593).
In (594) and (595), ke- occurs on active verbs t ‘pound, grind’ and t ‘dig’ in
constructions that further highlight their imperfectivity: in (594) with the suffix ‘only’
that translates as ‘keep V-ing (without doing anything else)’, and in (595) in a
construction that repeats the same verb a number of times to indicate the durative nature
of the event.
425
kachepechuji apor ave tangho
ke-che-pa-ch -j ] a-p r] av t ngh ]
NMLZ-RR-CAUS-suck-IRR2 POSS-time not.exist REP
‘they just kept pounding the rice for the rice beer cake, the whole time they didn't get to
take rest, they were pounding the rice beer cake, they didn't even have time to breast-feed
their children’ [WR, BCS 030]
Imperfective ke- occurs frequently in one particular text in the corpus, which is an
on-line narration of the pear story, where the speaker is commenting on the video clip as
he is watching it. Examples are (596) and (597).
426
A typologically well-supported way to model the historical development of the
imperfective construction from a nominalization construction is to assume a locational
construction as the source construction. As sketched out in Figure 20, the locational
construction at Stage 1 would include an S argument, a locational NP that structurally
consists of a nominalized clause, and the locational copula, which is the same as the
existential copula in Karbi, . This hypothesized source construction for the Karbi
imperfective is structurally analogous to the source construction of the English
progressive (i.e., ‘S be on/at V-ing’), except that in English, a preposition ‘on’ or ‘at’ was
required, where in Karbi nothing but the locational copula is required.
In some instances, such as (598), there is an ambiguity as to what triggers the ke-
prefix on the main clause verb, as this is an imperfective context but there is also a
focused element = ‘mirror FOC:RL’ in the clause.
427
kapabon
ke-pa-b n
NMLZ-CAUS-be.attached
‘and then, on this tobacco container, there was a mirror attached’ [HK, TR 026]
Similarly, in (599), which is from the on-line narrated pear story (see (596) and
(597) in the preceding section §9.7.3.2), the ke- can be interpreted as being the fossilized
nominalizer from the focus construction evoked by =si ‘focus:realis’, but it can also be
interpreted as marking imperfective aspect, as the speaker is commenting on the event as
it is occurring in the video clip.
Note that in examples like these, ke- is glossed as ‘nominalizer’ because it is the
more neutral label and better reflects the ambiguity between the two possible
interpretations.
This section aims to draw attention to the inconsistency with which ke-
‘nominalizer’ occurs in synchronically nominalized constructions, across all types of
synchronically nominalized constructions (though less so in nominalized
complementation constructions, as discussed further below).
In (600) and (601), ke- is absent from PCT modifiers, both preposed and
postposed. Note that (600) is an elicited sentence (produced when asked for a translation
of the English sentence), which strongly suggests that the absence of ke- cannot be
accounted for as being due to fast colloquial or hypo-articulated speech.
428
(601) Lack of ke- ‘NMLZ’ on postposed PCT modifier
ba ko jirpo {mm} pinike ne
ba ko j rp mm pin ke n
SURPRISE(<Asm) buddy:VOC friend AFF today=TOP 1EXCL
In (602), there is a lack of ke- on a relative clause verb, and similarly, in (603), a
construction that can be interpreted as a relative clause verb or a participant
nominalization also occurs without ke-.
… hormu hortar
[[[b ] a-b ng] av ]] h m [horm hort r]
keep POSS-CLF:HUM:PL not.exist house thing EE:horm
rikcho rikhaplo pu
r k-ch r k-h p-l pu
be.scattered-everything.neg.1/2 be.scattered-everything.neg.2/2-RL QUOT
‘[…] the plates and bowl and everything (are scattered) because nobody kept them in
place, everything is scattered’ [KK, BMS 093]
429
In (604), (605), and (606), adverbial clause verbs occur without ke-. Note that in
(606), there are three parallel nominalized, non-final marked verbs, but only the last two
are marked with ke-.
anat damlo
a-n t d m-l
POSS-direction go-RL
‘in order to go and sell bananas there he went towards the shop’ [HI, BPh 005]
(605) Lack of ke- ‘NMLZ’ on PCT root functioning as an adverbial clause verb
lasi laso adakke tovarta mesen apot
las [[[las a-d k ke] tov r t m -s n] ap t]
therefore this POSS-road.inbetween=TOP road=ADD:DM be.good-INTENS because
Finally, complement clauses have a different status, because there are a number of
different complementation constructions that can be situated on a scale of clause union
(§8.2.2). Therefore, the presence or absence of ke- may (or should) be interpreted as an
actual functional difference between two constructions, which is different from the
absence of ke- on relative and adverbial clause verb, which have to be interpreted as
430
being nominalized independent of whether ke- is actually used on the verb or not.166 That
said, there is one nominalized complementation construction that structurally resembles
relative and adverbial clauses due to the noun phrase delimiter functioning as a
complementizer (which is an element structurally equivalent to the head noun of relative
clauses and the subordinator of adverbial clauses). As (607) shows, however, ke- may
also be absent in this case of a nominalized complementation construction.
It is not clear at this point what may be behind the inconsistent occurrence of ke-
on synchronically nominalized verbs. The two most plausible reasons do not actually
fully explain it: first, it does not seem to be a (purely) phonological issue, such that ke-
does not occur if the verb stem already has a certain number of syllables, because there
are instances such as (606), where the trisyllabic chepenangpen occurs without ke-, but
instances such as (528), where an equally trisyllabic chingkoidup occurs with ke-.
Second, it is not (only) an issue of colloquial versus careful speech, since the lack of ke-
also occurs in elicitation as in (600) above. This issue has to be left to future study.
166
Note, however, that it is also possible to interpret the absence of ke- on adverbial clause verbs as an
indicator of advanced grammaticalization (and thus as a functional element), as discussed in §9.5.1.
431
CHAPTER X
This chapter deals with clause participants, both arguments and obliques, which
are argued not to be two discrete types of participants but to exist on a continuum in
Karbi. It discusses their expression and types of marking, depending on their functions
and roles. Starting with brief sections on terminological and conceptual issues in §10.1,
§10.2 offers an overview of what is to be said about the expression of clause participants
from the perspective of the predicate. Here I discuss argument roles in typical declarative
clauses as well as constructions that deviate from the typical patterns. While this chapter
is focused on the discussion of clause participant structure in declarative clauses, §10.2
also surveys what can be said about grammatical relation constructions in domains other
than declarative clauses.
In §10.3, an overview is provided of the ways participants are expressed and
marked, which leads into the following four sections: §10.4 discusses argument
expression (as lexical noun phrases, pronouns, or zero anaphora). §10.5 offers an
overview of how the Karbi noun phrase delimiter marker interacts with participant
marking. In §10.6, syntactic and semantic role marking is discussed. Here, the three
major types of marking participants (i.e., unmarked and marked by - ‘non-subject’
or - ‘locative’) are discussed in a section each to demonstrate the range of contexts in
which they occur. Other issues in role marking are also discussed in this section. Finally,
§10.7 offers an overview of the four main information structure clitics as well as several
other information structure constructions.
10.1. Preliminaries
10.1.1. Terminology
This section gives an overview of how terminology is used in this chapter. While
the noun phrase is the smallest unit that is relevant in this chapter, in §10.4 I distinguish
432
between pronouns and lexical noun phrases, where ‘lexical noun phrase’ is intended to
refer to a noun phrase with a lexical head noun, i.e., in opposition to a pronoun.
In order to be able to independently refer to the structural and the functional
elements of relevance in this chapter, I use the following terminology. Functionally, we
can distinguish between arguments, which are required, or projected, by the predicate,
and obliques (also called adjuncts), which are not required by the predicate but instead
offer additional information. (As labels intended to be purely functional or conceptual,
their application to clause participants will naturally be controversial at times.)
Structurally, we can distinguish between noun phrases, which are unmarked for local
‘case’ marking, and ‘relator noun phrases’, which are marked by a relator noun. There is
also one postposition =pen, and noun phrases marked with this postposition I will refer to
as ‘postpositional phrases’.
In §10.1.2 and throughout the chapter, I argue that there is no strict divide
between arguments and obliques. Nevertheless I find the syntactic primitives or
macroroles (depending on the viewpoint) of S, A, O, R, and T useful labels for the
description of ‘argument’ structure. The syntactic roles of S, A, O, R, and T define the
number of arguments in a given construction, i.e., one, two, or three arguments, with
(attempted) minimal reference to semantics, originally going back to Comrie (1978) and
Dixon (1979).167 Within this framework, S is the single argument of an intransitive
predicate; A is the more agent-like argument of a transitive clause, and O (or P) is the
other argument of a transitive clause; and in ditransitive clauses, R is the recipient-like
argument and T is the theme-like argument.
10.1.2. The Argument-Oblique Continuum and the Syntax, Semantics, and Pragmatics
in Role Marking
The distinction between participants that are required by the predicate and those
that are not, that is, the distinction between arguments and obliques, is not
straightforward in Karbi.168 First, there is a practical challenge in examining argument
167
Note that Comrie actually used the label ‘P’ instead of ‘O’.
168
In fact, it is not clear how a theory that posits a strong division between core arguments and obliques
would hold up to cross-linguistic examination. Croft (2001: 272 ff.) in his ‘Radical Construction Grammar’
433
structure in Karbi (shared by, in fact, many and presumably the majority of languages
across the world), which is that Karbi pervasively uses zero anaphora. The consequence
is that large amounts of texts have to be considered in order to get a sense for what
typical patterns are, and what patterns may be exceptional or irregular. Nevertheless,
what we find after consideration of large amounts of text is that Karbi does not have clear,
syntactic case marking, such that each syntactic role of S, A, O, R, and T is always
marked the same way. The only one of these syntactic roles that is consistently
represented is the S argument, which is always unmarked for role. However, that does not
necessarily mean that there is an S argument category in Karbi. Instead, within the system
as a whole, it makes sense to also interpret the consistently unmarked status of S
arguments pragmatically, i.e., as being due to the fact that S arguments by definition are
the only argument in a clause, hence the pragmatic principle of ‘don’t mark an argument
if it’s obvious what its role is with respect to the predicate’.
Even more broadly, it is not true that a direct relationship exists between a
particular type of marking or not-marking and core versus non-core or oblique status.
Unlike in English, where at least in transitive clauses the generalization holds fairly well
that core arguments are unmarked while obliques occur as prepositional phrases, no such
generalization works in Karbi. As shown in §10.6.1, an unmarked NP may take on any
‘syntactic’ or semantic role. Nevertheless, there is a sense that unmarked NPs, NPs
marked by - ‘non-subject’, and NPs marked by - are more likely to have, and
therefore more typically have, core status in the sense of a non-semantically-specific
status. It seems, however, that in Karbi, the non-semantic end of the continuum is more
strongly influenced by pragmatic factors such as contextual role assignment, rather than
syntactic factors such as valence.
Semantics plays a role in local marking as O-high169 arguments (see §10.2.1.2),
which are marked by , tend to have human referents while O-low arguments, which
remain unmarked, do not have human referents. Similarly, goal arguments with human
referents are always marked by , while those with non-human referents are more
examines the distinction in English, and finds arguments in favor of a sharp distinction insufficient. He
concludes that “under scrutiny, arguments and adjuncts as global syntactic roles evaporate.”
169
Karbi exhibits differential O marking. O-high arguments are those marked by - ‘non-subject’, while
O-low arguments are those not marked by - ; see §10.2.1.2 and §10.6.5.
434
typically unmarked (§10.2.2.1). The generalization in these cases appears to be that
arguments with human referents are always marked, whereas arguments with non-human
referents are typically unmarked, but may be marked if pragmatically important or salient.
These issues are further discussed in §10.6.
This section discusses the generalizations we can make about how predicates
defines their argument roles. §10.2.1 discusses the common mono-, bi-, and trivalent
constructions in typical declarative clauses. §10.2.2 discusses other common
constructions that deviate from the patterns discussed in §10.2.1, including the ‘need’
construction, existential and possession constructions, and motion constructions. §10.2.3
discusses predicates with derivationally changed argument structures. In §10.2.4, the only
type of ‘agreement’ in Karbi is discussed, which is the proclitic nang= as a marker of
non-subject speech act participants. Finally, §10.2.5 discusses (largely, the absence of)
constructions in other grammatical domains whose structures vary as a function of
argument structure differences.
435
the predicate may further specify that one of these roles be marked in a way that deviates
from the default pattern (§10.2.2).
Monovalent predicates require one core argument only, which never receives any
kind of role marking, but may occur with ‘noun phrase delimiter’ (§10.5) and/or
one of the information structure clitics (§10.7). In (608), the predicate ’ ’
based on the PCT ‘be small’ requires exactly one argument, which is filled by nesomar
‘my children’.
(609) and (610) offer examples of monovalent clauses that feature an oblique
participant in addition to the S argument. In (609), the oblique is a topic, and specifically
the possessor of the S argument (see §10.2.2.4), and in (610), the oblique is a topic as
well, but specifically a participant to whom the proposition applies
436
10.2.1.2. Bivalent Predicates: Marking of A and O Arguments
(612) Typical bivalent predicate with O-high: A unmarked and O marked with
‘primary object’
hala bamonpopen bamonpita
[NP]A
[h la bam n-p pen bam n-p t ]
that wise.person(<Ind)-male=with wise.person(<Ind)-female=ADD:EXH
437
Other possibilities in marking transitive A and O arguments are as follows. The
‘A’ argument may be marked by - ‘locative’ in the possession construction
(§10.2.2.3), or by - ‘non-subject’ in the ‘need’ construction (§10.2.2.2). The O
argument may be marked by - ‘locative’ in the motion construction, particularly if
the goal is a human referent (§10.2.2.1).
438
(613) Trivalent predicate with -marked R argument
e nang nepiso neso aphan
[NP]A [NP ]R
e n ng [[ne-pis ne-os ] aph n]
INTERJ 2 1EXCL-wife 1EXCL:POSS-child PAT
439
combination of a non-human T argument and a human R argument, because this
constellation occurs in the R-centered trivalent construction (§10.2.1.3.1). A recurring
verb in these constructions is t , which is glossed as ‘drop’, but more exactly translates
as ‘take T to R’ or ‘leave T with/at R’, where both T and R may or may not have human
referents (although it may be a requirement that at least one of the two have a human
referent). Note that this verb also occurs in the R-centered trivalent construction. Another
verb that occurs in this construction is t ‘put inside’, where R is the location or item in
which T is put.
Let us look at the data. (616) and (617) are from the same story, talking about the
same event of a witch abandoning children in the forest. Therefore, in both clauses, the T
argument has a human referent, while the R argument is a non-human location. The T
argument is marked by - in both instances. The R argument is marked by the general
locative marker - in (616), but it is marked by the more semantically specific relator
noun - ‘in the middle of’ in (617).
ekethondamlo
e=ke-th n-d m-l
1INCL:NSUBJ =NMLZ-drop-GO-RL
‘[…] (the witch) abandoned us over there in the deep forest’ [CST, HM 076]
nangkethonti
nang=ke-th n-t
1/2:NSUBJ=NMLZ-drop-get.rid.off
'and then, she took us here in the middle of the jungle and abandoned us’ [CST, HM 052]'
440
(618) is from the same story as the previous two examples, but is from earlier in
the story where the T argument is a third person instead of the first person. Interestingly,
the storyteller first uses the non-subject marker - on the T argument, then mentions
the locative R argument, which is rather complex and lengthy, and then again mentions
the T argument, this time without - . The reason for this change in marking could be
that the storyteller corrected himself and wanted to leave the T argument unmarked,
perhaps because he decided to use - to indicate the mother of the children as the
malefactee and hence primary non-subject argument. However, I think another
explanation is that the storyteller simply wants to remind the listeners of the identity of
the T referent, and leaves it unmarked for predicate role, following the pragmatic
principle of ‘if the role of an argument is obvious, leave it unmarked’ (see §10.1.2, §10.6).
441
In (619), the verb is again t ‘drop; take T to R’, and the T argument again has
a human referent while the R argument again has a non-human referent. This example
demonstrates, however, that the R argument may remain unmarked, as does here.
(620) and (621) show that both the T and the R argument may have non-human
referents, here with the predicate based on the verb t ‘put inside’. In both instances, T
remains unmarked. R is marked by the general locative - in (620), and by the more
semantically specific ‘inside’ in (621).
442
ajamborong arlosi lahai kethap lapen arum kevan
[NP]T
a-jambor ng arl si] [lah i] ke-th p]] lap n [ar m ke-v n]]
POSS-bag inside=FOC these NMLZ-put.inside and down NMLZ-bring
‘[…] and then, the man, into a bag sown from white cloth he puts these (fruits), and then
brings them down [SiT, PS 003]
Lastly, (622) is the only example in the corpus of recorded texts, in which both T
and R have human referents. Although T is only overtly expressed in a preceding non-
final clause, we at least can note that in this scenario, the R argument is marked by the
general locative marker - .
443
as the R-centered construction and ‘reduced’ the T-centered construction to a bivalent or
transitive construction with an additional goal argument. If we did this, the verb t n
would be problematic, because it occurs in both constructions. In the end, it seems
impossible to describe Karbi with just one true trivalent construction, because with a
+human R argument, both constructions occur. It is just not the case that a human R
argument makes for a true trivalent construction while a non-human locative or goal
argument is simply a bivalent or transitive motion construction with A, O, and a goal
argument.
444
10.2.2. Declarative Clause Constructions with Non-Typical Role Marking of
Arguments
Motion verbs such as ‘follow; join’, ‘go’, ‘reach’, or ‘drop in; stop
by’, often occur with overtly expressed goal participants.
However, it depends on the particular verb, how overt goal participants are
marked, or whether we may even consider them goal arguments (i.e., conceptually
required participants). And overt goal participant marking also depends on the referent,
particularly whether it is human or non-human. Finally, pragmatic factors also appear to
play a role.
First, let’s consider ‘follow; join’. (623) has a non-human referent goal,
which remains unmarked, while (624) has a human referent goal, which in both clauses is
marked by the general locative - .
The same pattern of leaving a non-human referent goal unmarked but marking a
human referent with - is found with ‘go’ in (625) and (626).
445
(625) ‘go’ with unmarked non-human goal
t “ bojar t ”[…]
am tsi Bokol p ab ng ke boj r GOAL/?O d m- t-l
and.then NAME NPDL=TOP market(<Asm) go-PRF-RL
‘[the children thought that] Bokolapo had already gone to the market’ [HI, BPh 010]
(627) shows that with ‘drop in, stop by’, the location where one ‘stops by’ has
a non-human referent here and is also left unmarked.
junponlang
j n-p n-l ng
drink-in.passing-still
‘and then, from here we went, we stopped by Dimapur and just had tea’ [SiT, HF 009]
(628) and (629) once again show that with ‘reach’, we get the same pattern of
marking a human referent goal with - while leaving a non-human referent goal
unmarked.
446
However, (630) shows that a non-human referent goal may also be marked
with - ‘locative’, not just +human referent goals, here also with ‘reach’.
The same is true for the goal of ‘come’, where non-human referent goals
may also be marked with - as in (631), or unmarked as in (632).
It is perhaps surprising that ‘come’ allows the same options for marking
goals as more inherently goal-oriented motion verbs, considering that is inherently
more source-oriented and more frequently occurs with overtly expressed source
participants in the text corpus, which, however, are then marked with ablative =pen. This
suggests we can preliminarily generalize that goals are more important semantic roles in
Karbi than sources, independent of the verb.
While the required argument status of goals for motion verbs has been argued for
other Tibeto-Burman languages such as Galo (Post 2007: 417), in Karbi any motion verb
takes unmarked goal participants, and even non-motion verb may take unmarked locative
participants. This is further shown in §10.6.1.7.
447
10.2.2.2. ‘Need’ Construction
The verb ‘need’ projects two arguments: the participant who needs
something, which I will refer to as the A argument, and the participant that is needed,
which I will refer to as the O argument. As shown in the elicited sentence in (633), the A
argument may be marked by - ‘non-subject’, but this is in fact optional in the sense
that some speakers will mark the A argument of ‘need’ with - , and others will
170
not. It is an open question at this point, whether this is a dialectal difference. According
to my consultants, it is ‘more correct’ to use - here.
There are several constructions indicating location and possession that involve the
existential copula , with the suppletive negative form (§4.6.2.1.1). Besides the
simple locative construction, there are two different possession constructions that differ
in how A and O are marked, with unclear effects on the semantic or pragmatic
interpretation.
The simple locative construction expresses ‘X is.at Y’, where ‘Y’ represents the
location. The ‘subject’ X is always unmarked for role marking, while the location Y may
be unmarked as well or marked by either the general locative - or by a more
170
There is no example of ‘need’ with overt A and O arguments in the corpus of recorded texts. The
following example demonstrates, however, an unmarked O argument of despite having a human
referent:
Since the verb used in the simple locative construction is the existential copula, it
perhaps comes as no surprise that the existential construction may be structurally
identical, if a locative participant is added (although we should then conceptualize the
locative participant as an oblique, i.e., as not required by the predicate). (636) and (637)
show that existential clauses with added locatives are parallel to simple locatives in that
the locative may also be either unmarked or marked by a locative relator noun.
449
vo thepo do
v th -p d ]
bird be.big-MODIF exist
‘[…] by the wild jackfruit tree, there is a <tr(ee)..> huge black bird’ [SeT, MTN 018]
“nangke nangdin ”
n ng e]A/POSR [nang-d n]O/POSD d -l ng]
2=TOP 2POSS-day(<Asm) exist-still
'although she said, “you shouldn't follow me”, he didn't agree, “you still have your life to
live (lit. days)” (he said)' [KK, BMS 084]
(639) Possession construction with unmarked A and 1st person exclusive possessive-
marked O
[…] amat neta neri ave nekeng ave […]
am t [n t ]A/POSR [ne-r ]O/POSD av [ne- ng]O/POSD av
and.then 1EXCL=ADD:DM 1EXCL:POSS-hand not.exist 1EXCL:POSS-foot not.exist
‘[…] and then also, I don't have hands or feet […]’ [RBT, ChM 030]
450
(640) Possession construction with unmarked A and 3rd person possessive-marked O
[…] jongsi phakok abang athu do pulote…
j ngsi h - ab ng A/POSR a-th O/POSD d pulote
if pig-meat NPDL POSS-fat exist if
‘[…] if there is fat on the pork…’ [PI, BPR 002]
mandamlo apot
m n-d m-l ap t]
become/happen-go-RL because
‘therefore, he asked his children, “O mothers, so then, what else could we do?”, We don't
have your mother anymore, because your mother has gone to become god’ [CST, RO
008]
451
10.2.2.3.3. Predicational Possession Construction with Locative-Marked A and
Unmarked O Argument
452
apran endet amui endetta bang
a-p an n-d t a-m i n-d t-t b ng
POSS-life take-PFV POSS-EE:pr n(<Ind) take-PFV-if CLF:HUM:PL
Note that in this construction type, the possessor does not necessarily have to be
strongly affected. In the last line in (646), halata refers to ‘Bey the Fair’, and the
following proposition is about his children: asomar thelo dinglo ‘his children grew up’.
That is, the possessor construction can also be used to keep the possessor a topic, not only
if the possessor is conceptualized as being affected.
453
halata asomar thelo dinglo
[[h la t ]POSR [a-oso-m r]S th -l d ng-l ]
that=ADD POSS-child-PL be.big-RL be.long-RL
'his children were growing up and then Bey the Fair also did the same and he got married
and so with respect to him also, his children grew up' [WR, BCS 006]
Interestingly, the -te in aphante in (647) may be the same element as the second
syllable in t ‘compared to’ in (649), which marks the standard of comparison here.
The first syllable could be related to - ‘comparative’.
454
pu
pu
QUOT
‘[…] compared to the men, it was the women who did much more, (I heard) […]’ [KaR,
SWK 070]
If the causative pe-~pa- occurs on monovalent verb such as t ‘die’ in (650), the
original S argument becomes the new causee, which in Karbi appears to be treated like an
O argument. In (650), the causee r is marked non-subject via - .
455
However, in (652), the causee and former A argument is not marked by - , but
is unmarked. A caveat is that this may be due to this being a jussive construction that
involves the hortative - , though if that was actually the case, it would not be clear
why.
The text example (653) and elicitation based on this example provided in (654)
show that further argument marking options exist, perhaps specifically with the verb
t ‘meet’ and a human referent O argument. (653) shows that the former O
argument may also be marked by - , and it is unclear how the former A argument
would be marked in that case. Alternatively, as shown in (654), the former O argument
may be marked by =pen ‘with’ (§10.6.4.1.1), while the former A argument may receive
marking via - . It is impossible, however, to mark both arguments with - .
456
10.2.3.2. Benefactive/Malefactive (Affective) -
Adding the suffix - to the verb indicates that there is an argument that is
affected by the particular event described by the predicate; the affected argument may be
a benefactee or a malefactee (§6.5.5.2.1). If overtly expressed, the benefactee or
malefactee is non-subject-marked with - , as shown in (655).
han cholo
h n ch -l
prepared.vegetables eat-RL
‘[…] there, they prepared meals for all the Diphu people, there they had prepared the
meals for us, there we had our meal’ [SH, CSM 054]
However, (656) shows that the affected argument may also remain unexpressed
via zero anaphora (§10.4.3). We know that this is a case of zero anaphora because the
verb is marked by - . Context makes it clear that here the malefactee (rather than a
benefactee) is the mother of the children, la Karbipi, mentioned at the beginning of this
intonation unit.
457
ingrengre voso ingrengre along
ingr ng-C v s ingr ng-C al ng]
call(small.animals)-NEG EE:v t k call(small.animals)-NEG LOC
171
Note, however, that this example was judged grammatically unproblematic by two independent native
speakers, in the course of working through the analysis of this folk story.
458
10.2.4. Person: Non-Subject Speech Act Participant Indexing on the Verb
The only arguments (or, actually, participants, as they are in fact not always
arguments) cross-referenced on the verb are speech act participants (SAPs) in non-subject
roles. The details about this phenomenon are discussed in §6.3.1.
It should also be noted that there are instances of nang= that follow in line with
evidence presented by Bickel (2001) on different agreement relations in Tibeto-Burman.
An example is (658), where the benefactee NP ‘(for) all of the Diphu people’ is cross-
referenced on the verb by nang=, which through context is disambiguated to be first
person plural ‘we’. This is similar to Bickel’s ‘partitional’ agreement, “where the NP
denotes a subset of the referents denoted by the agreement marker” (p.586), though in an
inverted sense, because here, the ‘agreement’ nang= represents the subset of the referent
denoted by the NP.
han cholo
h n ch -l
prepared.vegetables eat-RL
‘[…] there, they prepared meals for Diphu people, there they had prepared the meals for
us, there we had our meal’ [SH, CSM 054]
459
(659) nang= cross-referencing possessor
[…] ne banghini aphanta nangpran nangen'etji
n [b ng-hin a h n t ]POSR [nang-pr n]O nang= n- t-j
1EXCL CLF:HUM:PL-two NSUBJ=ADD:EXH 2:POSS-life 1/2:NSUBJ=take-all:S/O-IRR2
‘[…] I will take both of your lives' [CST, RO 019]
Note that the verbal proclitic nang= does not ‘agree with’ a particular syntactic
role; the only generalization that holds (so far) is that it does not occur with subjects. It
therefore does not represent a grammatical relations construction.
It also occurs with transitive verb stems such as pivir ‘destroy’ in (661), in which
case the scope is over the O argument, i.e., the item destroyed.
460
a reminder that the unsolved status of grammatical relations in basic declarative clauses
makes it difficult to talk about grammatical relations in other grammatical domains.
nangle'etlo {oi}
nang l - t-l i
CIS=reach-all:S/O-RL yes
‘the barika and the sarthe and all the other important tigers have already arrived there’
[HK, TR 186]
Nevertheless, predicate derivations are still the only clear part of the grammar
where there appears to be a consistent and non-neutral alignment of arguments across
both intransitive and transitive constructions (in particular, alignment of S with O).
Other grammatical domains or constructions often cited to pay attention to
grammatical relations, such as agreement (§10.2.4), clause-chaining and clause
coordination or pivots (§11.2.1 and §11.5), or relativization (§9.3) do not treat arguments
differently based on their syntactic roles.
For each argument required or projected by the predicate (§10.2), choices need to
be made as to whether or how it should be expressed, depending on a host of pragmatic,
syntactic, and semantic factors. Figure 21 offers a representation of the options speakers
have in argument expression. The arrows between slots 2 and 3, i.e., between the optional
use of the noun phrase delimiter and the role markers, indicate that either order
occurs, i.e., 2 before 3 but also 3 before 2.
461
1 2 3 4
[Participant] NPDL Role Information structure
[LEXICAL NP] a -ø -ø
[PRONOUN] - ‘NSUBJ’ =ke ‘TOP’
- ‘LOC’ =t ‘ADD’
[Other RNs/ =si ‘FOC’
=pen] =le ‘FOC:IRR’
[ZERO ANAPHORA-------------------------------------------------------------------]
The remainder of this chapter discusses the argument expression options as laid
out in Figure 21. We start out with a participant in slot 1. The speaker has to decide to
leave the argument unexpressed, i.e., as a zero anaphora, or to use a pronoun or a lexical
noun phrase; this is discussed in §10.4. Then, Karbi has an element that may or may not
be added, the ‘noun phrase delimiter’ , discussed in §10.5. §10.6 discusses the
options in role marking, and §10.7 discusses the options in information structure marking.
The choice of using either a lexical noun phrase, a pronoun, or zero anaphora
depends on how the speaker models the information and activation status of the argument
in the addressee’s mind. While this is a highly complex process, in what follows, I give
text examples to demonstrate this three-way distinction in argument expression.
A clause participant is expressed via a lexical noun phrase if the speaker assumes
the addressee cannot uniquely identify its referent. This is the case when a referent is not
activated - either because it is newly introduced or reintroduced - or if it is contrasted
with another referent.
462
The term ‘lexical noun phrase’ is underspecified as to how much information is
provided. A lexical noun phrase may consist of just a noun, or it may include a
combination of modifiers (see Chapter VII). I use the term ‘lexical noun phrase’ in
contrast with just ‘noun phrase’ in order to exclude pronouns. In (663), two new referents
are introduced, one is the oblique participant ‘in a king’s village’, the other is the S
argument ‘orphan’.
(663) Two newly introduced participants: locative inut arecho arong and S
argument jangreso
hako arni tangho {mm} inut arecho
h k arn t ngh {mm} [[e-n t a-rech ]
that.time day REP AFF one-CLF:HUM:SG POSS-king
10.4.2. Pronoun
Pronouns are in a sense intermediate between a lexical noun phrase, which occurs
when a participant is first introduced, and zero anaphora, which represents the absence of
an overt reference to the argument. (665) starts out with a lexical noun phrase: ‘among
the Bey, there were three brothers’. In the following intonation unit, this participant is
463
first referred to by the pronoun latum ‘they’, and then left out via zero anaphora in the
next clause.
(665) Full NP Bey atum / korte bangkethom changing to pronoun latum (and to zero
anaphora)
Bey atum korte bangkethom do aklengsi abangke
B y a-t m ort b ng- eth m d ] [a-kl ng-s ab ng ke
CLAN POSS-PL brother CLF:HUM:PL-three exist POSS-old.one-SPLT NPDL=TOP
464
‘frog’ is mentioned again at the end of the second intonation unit,
expressed via a lexical NP marked as topic via =ke.
kethondam
ke-th n-d m
NMLZ-drop-GO
‘the ant was on the way to drop off rice for his older brother’ [RBT, ChM 010]
465
asopi abang ha langhe lang kachinglu...
a-oso-p ab ng h l ngh l ng ke-chingl ]
POSS-child-female NPDL over.there washing.place water IPFV-take.bath
The ‘noun phrase delimiter’ (NPDL) occurs very frequently in the corpus of
recorded texts. There are approximately 300 instances, depending on whether some
occurrences are counted as - ‘POSS-CLF(:HUM:PL)’ instead (§10.5.2), which likely is
the source form of the grammaticalized . It marks noun phrases and relator noun
phrases in any clausal role and with any information status, although it most frequently
occurs on noun phrases unmarked for clausal role and either unmarked for information
structure status or marked with topic =ke; it thus appears to correlate with subject and
topic to a large degree. Its function has to do with marking contrast between clausal or
discourse participants. The position of the noun phrase delimiter is the right edge of an
NP, either preceding or following any role markers, and always before any information
structure clitics.
Examples of are given in (668), where both an extraclausal possessor NP
and the clausal S argument are marked by .
466
asomar abang etpik tangho
[a-oso-m r ab ng] t-p k t ngh
POSS-child-PL NPDL be.yellow-very REP
'and then, with respect to the so-called Bey the Fair and his family, his children were very
fair' [WR, BCS 010]
In the story that (668) is taken from, is used particularly frequently. The
story is about three brothers from the Bey Clan in mythological times, and gives an
account of why each brother started a subclan and how the relationships between the
subclans came about, which is a common genre in Karbi oral literature (§1.6.1). What
appears to trigger the frequent use of then is that there are three protagonists,
which are constantly contrasted. A further example that highlights this is (669).
467
also with pronouns, such as ‘he’, t ‘we (exclusive)’, and t ‘we (inclusive;
formal)’. It also occurs with uniquely referential proper names, e.g., Linda abangke.
The ‘noun phrase delimiter’ most typically occurs on noun phrases, but there are
some instances where occurs on adverbs (or on (nominalized) subordinate clauses,
see §10.5.2). Its occurrence on adverbs generally leads to a (nominal-like) interpretation
of adverbs as referring to bounded entities, which can then be contrasted with other
bounded entities. For example, (670) is about a character in the story, who takes his
grandmother’s advice and changes his mind about which of his two uncles’ daughters he
wants to marry. As a consequence, (670) states that at the house of the uncle that had
been his original choice for marriage, referred to as hadak abangke ‘there’, all the food
that had already been prepared for the wedding got rotten, because the wedding never
took place. The expression of interest, hadak abangke ‘there’, clearly contrasts the
original house/family with the newly chosen house/family to marry into. It appears that
adding results in the vague adverb ‘there’ receiving a more specific
interpretation of referring to the original house/family (i.e., literally, ‘at the there one (i.e.,
house)’).
468
and to Bey the Fair's house, you know, right?, there (that man's) daughter, to the daughter
of Bey the Fair, you know, right? he went to ask for her as a wife instead, there he went
to get married instead' [WR, BCS 016]
labangke ahoklo
l -ab ng e a-h k-l
this-NPDL=TOP POSS-truth-RL
‘therefore, whoever can walk over this thread, that one is true’ [CST, HM 096]
469
as ‘possessive-somebody’, because is also used as an indefinite pronoun ‘somebody’
(see also bang-pak in §4.5.6).
(674) shows that abang can also head a referential relative clause. In this case, it
is parsed and glossed as ‘possessive-classifier:human’.
(675) shows, however, that may also occur with action nominalizations, in
which case it is clearly not a syntactic head for a relative clause.
470
appears better analyzed as the noun phrase delimiter, because (677) shows that ng
does not fulfil a syntactic head requirement as it occurs on adverbial subordinate clauses.
t t ”
nang=ke-pa-chet ng-j ab ng e pa-the' ng-sin ng
1/2:NSUBJ=NMLZ-CAUS-meet-IRR2 NPDL=TOP CAUS-be.bright-si.HORT
'and then, "okay then, in order to make you meet my grandmother, let's wait a little while
until it's bright (lit., make it bright)"' [KK, BMS 062]
nangponpo pulo
nang p n-p ] p -l
1/2:NSUBJ=take.away-IRR1 say-RL
'”[…] (there) my grandmother will come and so I will take you there”, (the dog) said
[KK, BMS 064]
Most commonly, occurs with NPs unmarked for role (§10.6.1). It rarely
occurs with ‘non-subject’ - marked NPs or ‘locative’ - marked NPs (see §10.6.2
and §10.6.3). There are only four instances of occurring with the non-subject
marker in the corpus of recorded texts, one of which is given in (678). The typical order
of preceding the roler marker is reversed in one of these four instances, where
instead of as in (678), the order is , with no clear semantic or
pragmatic difference.
humponpera pareponpera
h m-p n-C -r par -p n-C -r
pick.up-on.the.way-NEG-NF:IRR use.tool-on.the.way-NEG-NF:IRR
si bor'ilonang hedi?
[s bor' -lon ng hed ?]]
therefore try-HORT:EMPH okay?
'so as much as we can, regarding our talking to each other, without picking up another
language, without speaking another language, in Karbi as much as we can, to speak in
Karbi to each other let's try, okay?' [KaR, SWK 016]
472
(679) above also exemplifies the use of on (nominalized or non-
nominalized) subordinate clauses without a head noun or subordinator, in ’
‘as much as we can’. This is best analyzed as an instance of a topic-marked subordinate
clause (§9.5.1).
The noun phrase delimiter mostly occurs with NPs unmarked for information
structure and NPs marked with =ke ‘topic’, of which there were several examples in the
preceding discussion because those are the most common information structure statuses
occurs with. However, also occurs with additive =t as in (680), with focus
=si as in (681), and with irrealis focus =le as in (682).
473
10.6. Role Marking
I use the term ‘role marking’ to refer to a type of marking or non-marking that
relates a participant to the verb. This term intentionally blurs the line between syntactic
and semantic roles, as well as between core and oblique participants. Figure 22 shows the
four basic possibilities in role marking on a continuum of the parameter of ‘syntacticity’
and/or, in fact, ‘pragmaticity’ (least semantic information and most likely to be ‘core
argument’) and ‘semanticity’ (most semantic information and most likely to be ‘oblique’).
These four possibilities are: 1) leave the NP unmarked; 2) the ‘non-subject’ marker - ;
3) the ‘locative’ marker - ; and 4) one of a set of (semantically specific) relator nouns
(§4.4.4) or the comitative/instrumental/ablative =pen.
474
role an NP has. In particular, instead of offering concrete semantic information, the use of
an unmarked NP means that the speaker relies on the addressee to be able to
unambiguously figure out the role of the NP. The two middle points on the continuum, -
and - may be analyzed as having some syntactic basis, because - generally
never marks S/A (except for, arguably, in the semantic marking of arguments of
‘need’, §10.6.2.4), and - generally never marks S/A (except for, again arguably, in
the semantic marking of possessor, §10.6.3.4), and never O or T, pending the analysis of
motion/goal constructions (§10.2.2.1, §10.6.3.2, §10.6.3.3).
Unmarked noun phrases in a clause may fill any ‘syntactic’ or ‘semantic role’, i.e.,
they may fill in for ‘required’ core arguments as well as any type of additional ‘oblique’
participant. It therefore makes sense to think of them as truly unmarked, rather than
‘zero-marked’, i.e., leaving them without a marker is not to be understood in opposition
or in a paradigm with the actual markers, but as not specified for a syntactic or semantic
role. Instead, then, unmarked is best understood as a pragmatic default, which represents
the assumption on part of the speaker that the addressee is able to identify the syntactic or
semantic role the unmarked noun phrase takes in the clause (see also §10.1.2 and §10.6).
‘Unmarked’ then instructs the listener to interpret the noun phrase as the most obvious or
expected syntactic or semantic role. In addition, ‘unmarked’ NPs predominate where the
semantic role is non-salient or most typical. For example, a locative NP that could receive
the general locative marker - is more likely to remain unmarked than a locative NP
whose locational properties are unexpected or salient, such as - ‘high up’
(§10.6.1.7). Examples that demonstrate the occurrence of unmarked noun phrases in all
types of syntactic and semantic roles follow.
10.6.1.1. S Argument
(683) and (684) show that NPs unmarked for role can function as S arguments
both of predicates consisting of PCT roots and those consisting of active verbs.
475
(683) Unmarked S argument (PCT root)
jangrengso abangta repik tangho {mm}
jangr ngs ab ng t r -p k t ngh mm
orphan NPDL=ADD:DM be.smart-very REP AFF
‘[but] the orphan is very smart’ [HK, TR 064]
10.6.1.2. A Argument
chelangledetlohe {mm}
che-l ng-C -d t-l =he
RR-see-NEG-PFV-RL=AFTERTHOUGHT
‘this orphan, as he was digging and digging and digging... he didn't look around’ [HK,
TR 051]
(686) and (687) show that O arguments both with non-human and human
referents may remain unmarked. Unmarked O arguments are defined as O-low arguments
476
(§10.2.1.2), so per definition, only O-low arguments are unmarked, while O-high
arguments are marked by - ‘non-subject’.172
chirithupo pu […]
che-r -th -p pu
RR-search-again-IRR1 QUOT
'<from anywhere, I will find another wife again> (he thought by himself), […]' [CST, RO
009]
10.6.1.4. T Argument
172
Note that in (687), the reason why the O argument nechor is unmarked is not because the verb is marked
reflexive by che-; O arguments of reflexive-marked verbs can still be O-high (see §6.4.3).
477
(689) Unmarked R argument in T-centered trivalent construction
[…] apai abangsi mane kechengpenke
[NP]A
a-p i ab ng si mane ke-ch ng pen ke
POSS-mother NPDL=FOC I.mean(<Asm) NMLZ-begin=from=TOP
Both the verb ‘go’ and the verb ‘sit’ in (690) take goal participants.
While we may consider ‘go’ to be a motion goal verb and ‘sit’ a motion manner verb,
both take equally unmarked goal/locative participants: ‘outside part of Karbi
house’ and ‘house’ in the case of ‘go’, and ‘entrance area of Karbi
house’ in the case of ‘sit’.
t hem damsi,
[th ng a-khangr -k k-si] h m d m-si]
firewood POSS-basket.for.firewood leave-in.a.fixed.place-NF:RL house go-NF:RL
hongkup
hong ingn -l n-si] [m k che-pa-ch -si]
entrance.area.Karbi.house sit-big:AO-NF:RL breast RR-CAUS-suck-NF:RL
"dojoinoi, po!"
[d -j i-n i p ]
stay-quietly-INFRML.COND.IMP father
‘and then, the mother went and unloaded the firewood in the Pharla (Veranda), then went
inside the house, sat down in the Hongkup, gave the child the milk, “be quiet, father”’
[KK, CC 015]
478
10.6.1.7. ‘Oblique’ or Semantic Role
As shown in §10.6.1.6, both goals of motion goal verbs, which should therefore
be considered arguments, i.e., required roles, and locatives of motion manner verbs are
often unmarked. Another example of a different motion manner verb, ‘jump’, with
an unmarked goal participant is offered in (691).
cho
ch ]
eat
‘[…] if your father takes his meal in the hemtap, you eat in the mandu>' [CST, RO 017]
The fact that in the same sentence, the locative NP ‘(in) the field hut’ is
unmarked, but the other locative NP hemtap angsong ‘high up in the treehouse’, provides
further evidence that ‘unmarked’ really does mean ‘unmarked’ in the sense that a role
does not need to be marked if it is unambiguously and expectedly defined by context (as
479
judged, of course, by the speaker). Here, the unusual and salient location ‘up in the tree
house’ is marked by the specifically semantic relator noun - g, while the ordinary
and default location on the ground ‘(in) the field hut’ remains unmarked.
Finally, (693) and (694) demonstrate that NPs in unusual and typically oblique
roles such as ‘instrument’ and ‘path’ may be unmarked if the speaker judges the context
to satisfactorily disambiguate their roles.
The ‘non-subject’ marker - may occur with any roles except for S and A
arguments, which is why it is glossed as ‘non-subject’. The only instance where -
marks an argument that might be argued to have A-like properties is the ‘subject’ of
‘need’ (§10.6.2.4). In its more syntactic functions, - marks O arguments (§10.6.2.1);
R arguments in the R-centered trivalent construction (§10.6.2.2); and T arguments in the
T-centered trivalent constructions (§10.6.2.3); finally, - marks oblique participants
(§10.6.2.6).
480
10.6.2.1. O-High Argument
However, there are three instances of O-arguments with non-human referents, all
with the same referent in the same recording: a procedural text which is mostly about
how traditional clothes are made. One of these instances is in (696). The O-high
argument in three different sentences is pindeng sumpot a collective elaborate expression
that refers to ‘dress and ornaments’, which is the topic of the larger discourse or text, and
perhaps because of that is treated as an O-high argument.
kosonsonsi keklemma
kos n~s n si ke-kl m ma
how~DIST.PL=FOC NMLZ-do=Q
'in the old days, the people of the Karbi tribe, the clothing and ornaments, during the old
days, what are the (different) ways in which we the Karbi tribe would make/produce
dresses and ornaments?, (that's what I will talk about)' [KST, PSu 002]
481
(697) - marked participant of predicate ‘be glad; be grateful’
lapenke la phutup <kapa..> kipidunthu apot
lap n ke [l phut p ke-p -d n-th ap t]
and.then=TOP this hat NMLZ-give-JOIN-again because
tekangbuplo
tek ng-b p-l
leave.for-RES:gone-RL
‘she left her child with the grandmother’ [KK, CC 011]
482
(699) - -marked human T argument and a -marked locative R/goal argument
[…] itum aphanke ha nampi namdur alongsi
[NP ]T [NP a ]R
e-t m aph n ke h n mp n md r al ng si
1PL.INCL-PL NSUBJ=TOP over.there big.forest EE:namp LOC=FOC
ekethondamlo
e=ke-th n-d m-l
1INCL:NSUBJ =NMLZ-drop-GO-RL
‘[…] (the witch) abandoned us over there in the deep forest’ [CST, HM 076]
avelo <bu>
av -l
not.exist-RL
‘to me, this is nothing but a bad omen’ [KK, BMS 021]
484
(705) a as subordinator with adverbial clause
<aphatang along'an saikel...> <a> saikel along'an
<a-phat ng al ng- n saik l...> [saik l al ng- n
POSS-kd.basket LOC-up.to bicycle(<Eng) bicycle(<Eng) LOC-up.to
173
This example then also provides evidence for the gradient nature of categories as caused by
grammaticalization (DeLancey 1997), here specifically as - being ‘in between’ an NP marking relator
noun (with a syntacticized function) and a subordinator.
485
10.6.3. Functions of ‘Locative’ -
486
t
t ng-d r
collide-IDEOPHONE
‘and then he went just a bit further and then he couldn't hold himself anymore, he hit one
stone’ [SiT, PS 029]
While (711) in the previous section has already demonstrated that goal/O-like
arguments may be marked by ‘locative’ - , further examples of goals or locatives are
offered below. The goal of the verb ‘go’ has a human referent that is marked
by - in (712), and, similarly, the locative of the verb ‘stay’ has a human referent
that is marked by - in (713).
487
(712) - marked goal of ‘go’
t t
[n h n n t n-p aj t-p ] [chir -r -n i]
1EXCL curry rice cook-IRR1 GENEX-IRR1 cry-NEG.IMP-INFRML.COND.IMP
nangphihai along ]
[[nang- h -h i al ng] d m-n i d i]
2POSS-grandmother-HON LOC go-INFRML.COND.IMP OK?
I will cook food and everything, don't cry, go to your grandmother, ok? [KK, CC 016]
pu tangho
chir -r -n i pu t ngh
cry-NEG.IMP-INFRML.COND.IMP QUOT REP
"stay with your grandmother, don't cry" [KK, CC 017]
There is one type of possessor construction, in which the ‘A’ argument is marked
locative by - . There is only one example of this construction in the corpus of
recorded texts, offered in (714). This is further discussed in §10.2.2.3, and specifically
also in §10.2.2.3.3.
488
10.6.4. Semantically Marked Participants
In what follows, I discuss relator noun phrases (and the one postpositional phrase
of Karbi, with =pen ‘with, from’) other than those with - ‘non-subject’ (§10.6.2) and
- ‘locative (§10.6.3). These participants marked with elements other than -
and - may also occur both as core roles (§10.6.4.1 below) and as obliques (§10.6.4.2
below), in the (not easily operationalizable) sense of required or not required by the
predicate. This is evidence that they are not, in fact, categorically different from -
and - , which is why the various role markers are represented as being on a continuum
in §10.6. Nevertheless, there is a gradient difference between - ,- and the more
semantically specific relator nouns (and =pen), in two interrelated senses. First, -
and - are semantically bleached, which makes them applicable in a greater variety of
contexts. Second, this leads to a higher frequency with which - and, to a lesser extent,
- are used, compared to the lower frequency that each of the more semantically
specific relator nouns are used. This in turn leads me to the hypothesis that to the extent
that we can differentiate between core arguments and obliques, - and - more
often mark core arguments, and the more semantically specific relator nouns more often
mark obliques.
There are two constructions in which core arguments are marked by elements
other than - and - . One is with =pen ‘comitative, instrumental, ablative’, the
other is with various specific locative relator nouns.
The predicates in both (715) and (716) project a comitative argument: in (715)
through the predicate derivation - ‘together (with)’, and in (716) through the
comitative/instrumental predicate derivation - . In both clauses, however, the comitative
argument is marked by =pen.
489
(715) Projected comitative NP marked by =pen
apot “la nangpopen chorappetlongle... […]”
[ap t l nang- en ch -r -p t-l ng-C ]
because this 2POSS-father=with eat-together-all-GET-NEG
'because of that, “you don't get to / must not eat together with your father, […]”' [CST,
RO 019]
The same, seemingly obligatory, marking with =pen occurs on the instrumental
NP in (717), where, again the verb stem contains the predicate derivation - , which
indicates the core status of the instrument role.
Finally, the verb stem chekak ‘part/separate (from)’ in (718) also projects a core
ablative or comitative role, which is fulfilled by the second person pronoun marked by
=pen.
It thus appears that while comitative, instrumental, and ablative roles may be
unmarked (§10.6.1.7), if they are marked, they are always and only marked by =pen.
Even with the ‘applicative-like’ (though not actually applicative) - suffix, comitatives
and instrumentals are never ‘promoted’ to being marked by - ‘non-subject’. This
suggests that there is a sense in which comitative, instrumental, and ablative roles are
490
never afforded a syntactically high status (while they can arguably be afforded a
pragmatically high status if left unmarked)
nangkethonti
nang=ke-th n-t
1/2:NSUBJ=NMLZ-drop-get.rid.off
'and then, she took us here in the middle of the jungle and abandoned us’ [CST, HM 052]'
Semantically specific relator nouns presumably most often mark obliques, i.e.,
roles not required or projected by the predicate. Examples of the semantically specific
491
roles marked by the various relator nouns are offered in §4.4.4; a sample instance
of - ‘high up’ marking a semantically specific location is provided in (721). In this
example, the non-specific or non-salient locative role of ‘(in) the field hut’
remains unmarked (see also §10.6.1.7).
mandule cho
[mand =le] ch ]
field.hut=FOC:IRR eat
‘[…] if your father takes his meal in the hemtap, you eat in the mandu>' [CST, RO 017]
492
(722). But in any event, it clearly shows that differential O marking may be triggered
solely by pragmatic factors, because these two examples are semantically identical.
ako chedamlo
ak che-d m-l ]
again(<Asm) RR-go-RL
‘no! because our mother and father are there, let's still go and ask our parents, and then
they went to the tigers’ [CST, HM 112]
Below I discuss the constructional component in argument structure. That is, the
same verb stem can, to some extent, participate in different argument structure
constructions, as shown below with two different verb stems: che(-)t ‘(RR-)meet’ and
- ‘ask-go’.
10.6.6.1. -t ‘RR-meet’
493
so t is a lexicalized unit. As the following examples show, the argument structure
projected by t is not fixed, but there is variation, which we can ascribe to some
degree to differences in construal. In (724), t is used intransitively with a plural S
argument, here t ‘the public’, in the sense of ‘meet’ or ‘gather’.
(725) and (726) show that alternatively, and more often in the corpus of recorded
texts, t is used transitively. In (725), the O argument is marked by =pen
‘comitative’, and in (726), the O argument is marked by - ‘non-subject’.
chetongvekpo <ne>
chet ng-v k-p <n >
meet-definitely-IRR1 1EXCL
‘I will definitely meet my grandmother later’ [KK, BMS 028]
494
10.6.6.2. - ‘ask-GO’
peklodup athema?"
pe-kl -d p ath ma
CAUS-fall-falling.sound.from.high.solid.obj because
'over there he went and asked the squirrel, "so then, squirrel, why did you gnaw on the
Hanthar and thus made it fall down on the back of the pig?"' [RBT, ChM 066]
495
10.7. Information Structure Marking
The four sections below discuss each of the four information structure clitics:
topic =ke in §10.7.1; additive =t , which acts as a topic-switch marker, in §10.7.2;
furthermore, (realis) focus =si in §10.7.3; and irrealis focus =le in §10.7.4. Noun phrases
unmarked for any of these information structure statuses are taken to simply be
unspecified with regard to the four information statuses. Note, however, that information
structure marking is not restricted to these four clitics. §10.7.6 discusses several other
information structure constructions including a preliminary account of constituent order
(§10.7.6.1).
In calling =ke a topic marker, I follow the traditional notion of ‘topic’ as ‘what
the sentence is about’. Since this does not serve as an operationalizable definition, this
section instead offers examples to provide an overview of typical occurrences of topic
=ke, which coincide with typical topic functions in the linguistic literature. Note that in
all instances, =ke indicates an element at the beginning of the clause, only following any
discourse connectors (or another element marked by =ke). NPs marked by =ke may be
oblique participants that are not projected by the verb, or they may be core arguments that
are projected by the verb. NPs marked (or unmarked) for any syntactico-semantic role
(§10.6) may be marked with =ke. =ke also occurs on adverbs, as well as on entire clauses.
Topic =ke optionally marks the S argument of equational clauses, as it does in
(729).
Topic =ke marks framing elements at the beginning of clauses, which indicate the
setting of the event expressed in the clause. In (730), a temporal NP is marked with =ke,
while in (731), a temporal adverb occurs in the same construction.
496
(730) Topic =ke on framing NP
[…] laso arni abangke hala osomar atum mandu kecho […]
las arn ab ng e [h la os -m r a-t m mand ke-ch ]
this day NPDL=TOP that child-PL POSS-PL field.hut NMLZ-eat
‘[…] that day, those children were eating in the mandu, […]' [CST, RO 030]
(732) and (733) show that there may be more than one topic-marked element, in
which case both topics occur at the beginning of the clause. In (732), the second person
pronoun , which is an oblique or non-core person participant of the nominal
predicate ‘(these) are new people’, represents the first topic; the locative adverb ‘here’
represents the second topic.
In (733), the temporal postpositional phrase laso arnipenke ‘from this day on’
represents the first topic. The O argument of the verb stem ‘cause to not exist’, hala
’ ‘that witch’, represents the second topic, which is reactivated here after having
been an important character in the story before.
In another example of the contrastive topic construction in (735), the clauses are
not structured exactly parallel to one another, but the contrast is still very clear: ‘you have
already reached’, and ‘we don’t know how to get there’.
t t t
ne-t m e d m-th k-C -l nang-ph pen
1EXCL-PL=TOP go-know.how-NEG-RL 2POSS-grandmother=with
nang-ph ke]
2POSS-grandfather=TOP
‘it's your house and property, you've already reached, (but) we don't know how to go,
your grandmother and your grandfather’ [KK, BMS 096]
The next example is part of a story, in which a dog takes his owner to Chom
Arong ‘Chom’s Village’, the place in Karbi mythology where the dead people reside. In
(736), the dog and its owner have reached a huge body of water on their journey. In this
sentence, the dog is speaking, and =ke occurs on the first person pronoun subject. It
appears that the function of =ke here is to indicate the despair of the dog over the
difficulty of the task; =ke may thus serve to evoke a set of candidates more likely to
succeed in the challenge.
498
(736) Topic =ke evoking a set of more likely candidates?
kopusi neke nangkepaparponpoma
kop si n e nang=ke-pa-p r-p n-p ma
how=FOC 1EXCL=TOP 1/2:NSUBJ=NMLZ-CAUS-cross(<Asm)-take.away-IRR1=Q
‘how will I (of all ‘people’) take you across the water?’ [KK, BMS 045]
Finally, while the above examples suggest that =ke is only used with given and
referential, or definite, participants, (737) shows that the derived indefinite pronoun
komatne ‘somebody’ may also occur with =ke.
10.7.2. Additive =t
499
(738) =t signaling the perspective-switch from asking to answering (converses)
“ ” ,
[[ch k-j m ] p ne-ph n arj -l ]
be.fine-IRR2=Q QUOT 1EXCL-NSUBJ ask-RL
t t “ ”
[am tsi n t p -l ch k]
and.then 1EXCL=ADD:DM say-RL be.fine
'<is it fine?>, they asked me, and I replied, <it's fine>' [SH, CSM 049]
In the other construction type, =t marks a reactivated topic that contrasts with
the current topic. This function of =t is argued to indicate discourse continuity, i.e., as
telling the listener to understand the utterance as being strongly connected to the previous
discourse despite the fact that the topic has changed (cf. also Kaplan's (1984: 514)
explanation of obligatory English 'too' as "to emphasize the similarity between
contrasting constituents"). This strong connection is typically afforded by a causal,
resultative, or reactive relationship to the previous event, as demonstrated with the
following examples.
In the folk story Chongho-kaloso lapen Miso-rongpo ‘The Frog and the Ant’ (see
Appendix D), there are numerous instances of topic-switch =t due to the structure of the
story, which is built on the idea of a chain reaction of events: after a fight between an ant
and a frog, the ant bites the frog, the frog is mad and jumps around and destroys a
squirrel’s ladder, the squirrel gets mad and gnaws a fruit off a tree, which then falls on
the back of a pig, and so on. In the end, a rock kills the king’s daughter, and as the king
tries to find out who he can hold responsible, he traces the chain reaction back to the frog
and the ant and punishes them. This story that is built on this chain reaction of events
offers a number of natural topic-switch situations that are marked by =t . For example, in
(740), the first clause states that the frog, ellipsed via zero anaphora (§10.4.3), is in rage
and jumps on the squirrels ladder, which then breaks. As a consequence, the squirrel,
500
marked by =t , gets angry. It is this topic switch to the squirrel as it reacts to the frog’s
action that is marked by =t .
t t ” ta ningrilo
ke-d -th k ap tlo ma] [bam n-p t ningr -l ]
NMLZ-stay-know.how should= Q wise.person(<Ind)-male=ADD:TS worry-RL
'she returned to Chom, and so Bamonpo worried: “how bad! How should I be able to
live on?”' [KK, BMS 015]
The next example is from a story about an orphan who encounters a tiger. Here,
over the last few intonation units, the tiger has just threatened the orphan and said that he
will take away the orphan. In (742), the storyteller switches back to the orphan
protagonist. This kind of topic-switch after the end of direct speech (i.e., topic-switch to
the addressee of the direct speech) is very commonly marked by =t on the new topic.
501
As a last example, consider (743), where the topic-switch does not involve a
subject-switch as has been the case in the previous examples. This is from a story where a
king’s children are abandoned in a jungle without the king ever knowing they existed.
Here, at the end of the story, the children return and the king recognizes that they are in
fact his children. In (743), the first clause is the direct speech of the king proclaiming to
the children that he considers them indeed his children. In the second clause, the topic
switches from the king to the children, as the action of the king taking his children home
consequentially follows from the recognition that they are his children.
174
It thus may be the case that is a cross-linguistically ‘normal’ that additive particles are used as topic-
switch devices, and that Western Indo-European languages are cross-linguistically odd in this respect.
502
10.7.3. Realis Focus =si
Realis focus =si occurs in affirmative, declarative main clauses and appears to be
in complementary distribution with =le ‘irrealis focus’, which occurs in a number of
irrealis-type constructions (see §10.7.4 on =le, and see §11.3 on the notion of realis vs.
irrealis in focus clitics). I call =si and =le focus markers because they both mark
contrastive/corrective focus, schematically ‘not X=le, but Y=si’ (§10.7.5). However, =si
also occurs in a range of other, more or less focus-like, constructions.
(744) offers an example of corrective/contrastive focus from natural text, which
can be schematically represented as ‘not X=ADD, not Y=ADD, but Z=si’.
[ - t= =si] ki-
stone POSS-piece=with=FOC NMLZ-cover
‘neither with thatch nor with tin, but with pieces of stone they cover (their roofs)’ [SiT,
HF 050]
(745) offers an example of restrictive focus: the orphan ‘all by himself’ performed
the action. For a discussion of additional and more specific restrictive focus
constructions, which optionally involve =si, see §10.7.6.3.
In (746), which is from the same story about the orphan, =si occurs on ‘mirror’
in a clause that describes a mirror being attached to a tobacco container. This mirror ends
up being instrumental in the orphan’s successful lie to a tiger, which helps him reverse
his role from being a victim to scaring the tiger off. The =si marking could thus be
signaling the importance of this referent, as a way to indicate that this referent should be
paid attention to.
503
(746) Focus =si to indicate an important referent?
anke laso athongkup along lujisi kapabon […]
nke las a-thongk p al ng luj =si ke-pa-b n
and.then this POSS-tobacco.container LOC mirror=FOC NMLZ-CAUS-be.attached
‘and then, on this tobacco container, there was a mirror attached […]’ [HK, TR 026]
There are a range of elements which often have =si attached, while the force of
the focus is weak. One such type of element is discourse connectors (‘and then’) like
t or t , and apparently also in the forms and nsi (§12.1.3). Another type of
elements are adverbs meaning ‘like this’, such as in (748).
chethanbom
che-th n-b m
RR-tell-CONT
‘and so that's how people continue to tell each other the story of Bokolapo’ [HI, BPh 021]
504
narration of a sequence of activities: ‘after we drank tea, we watched the drumming’.
There is clearly no sense of any special status of this sequence of events.
-t -t - -
1EXCL-PL drum NMLZ-beat.drum see-JOIN-RL
‘after we drank tea and everything, we watched the drumming’ [SH, CSM 041]
Finally, there are also a number of instances in which =si occurs on locative NPs,
as in (750). Future study will need to address whether these can be somehow subsumed
under the general focus marking function, or whether they need to be accounted for
otherwise.
Focus =si may have its origin in a copula, as apparent copula si(i) cognates exist
in Kuki-Chin. This then would also explain the common occurrence of ke- ‘nominalizer’
on predicates in clauses that have an NP marked with =si. The diachronic nominalization
scenario that accounts for copulas grammaticalizing to focus markers is discussed in
§9.7.3.1. Another (or perhaps in the end the same) possibility for the etymology and
apparent cognates of =si is to link it to demonstrative si in Meithei (Chelliah 1997: 81).
Note that there are a few instance of a =lo focus marker in the corpus, which must
be linked historically to the realis predicate suffix - . As discussed in §6.9.1.7, judging
from the few occurrences of =lo in the corpus, there is no functional difference to realis
focus =si. As further research is required, however, =lo will not be further discussed here.
505
10.7.4. Irrealis Focus =le
Irrealis focus =le occurs in complementary distribution with realis focus =si. The
irrealis contexts in which =le occurs include non-declarative speech acts (§11.1),
subordinate clauses, deontic predicates that involve ‘need’, and negated predicates.
For a discussion of irrealis clause types, see §11.3.
In (751), =le occurs in the first clause with a negated predicate and in the second
clause with an imperative. The focus marking here suggests that there is a set of
alternatives to the focus marked element, and that a previous false preconception is being
corrected. In the first clause, the first person pronoun is focus-marked. While it is clear
that there is somebody who is guilty in this context, the speaker asserts - and corrects the
wrong belief - that it is not him. In the second clause of this example, the focus marking
again highlights the existence of a set of alternatives. This time, one alternative is the
speaker himself and that alternative is contrasted and corrected with the focus-marked
participant, the Vo’arbipi bird.
The next example shows that it is, however, not always concrete contrastive focus
that corrects a misconception, which is marked by =le. In this story, the mother of two
little children dies. In (752), the father desperately addresses his children.175 One thing he
says is itum nangpei=le kedo kalilo ‘we don’t have your mother anymore’, where =le
occurs on nangpei ‘your mother’. It is clear that ‘your mother’ does not contrast with any
concrete alternative here, in the sense that there was any belief that somebody else could
have died. It does make sense, however, to think of the function of =le here as
175
Note that it is common in Karbi to address one’s children as ‘mother’ and ‘father’.
506
emphasizing how compared to the death of any person, the loss of the mother is the most
difficult loss to cope with.
mandamlo apot
m n-d m-l ap t]
become/happen-go-RL because
‘therefore, he asked his children, “O mothers, so then, what else could we do?”, We don't
have your mother anymore, because your mother has gone to become god’ [CST, RO
008]
See the next section, §10.7.5, for more information on how =le relates to, and
interacts with, the other information structure clitics.
A possibility for the etymology of =le is to relate it to the negative equational
copula (§4.6.2.2). As is hypothesized to be historically parsable into two
morphemes, where the first, velar onset syllable is the negative morpheme and the second,
lateral-onset syllable is the copula, this second morpheme in could very well be
historically the same morpheme as =le. Copulas often grammaticalize to focus markers in
cleft constructions, which, in fact, likely happened with =si as well (§9.7.3.1). The fact
that it is specifically (the copula portion of) the negative equational copula that may be
the grammaticalization source of =le would have further explanatory force given that =le
marks focus in irrealis contexts only.
507
10.7.5. Relationships between Information Structure Clitics
We can start out with a preliminary generalization that both =ke ‘topic’ and =t
‘additive’ mark topic, while =si ‘focus’ and =le ‘irrealis focus’ mark focus.
As mentioned above, both =ke and =t occur in the (affected) possessor
construction (see §10.2.2.4), as in (753) and (754).
“nangke nangdin ”
n ng e A/POSR nang-d n O/POSD d -l ng]
2=TOP 2POSS-day(<Asm) exist-still
'although she said, “you shouldn't follow me”, he didn't agree, “you still have your life to
live (lit. days)” (he said)' [KK, BMS 084]
(754) Possession construction with unmarked A and 1st person exclusive possessive-
marked O
[…] amat neta neri ave nekeng ave […]
am t n t A/POSR ne-r O/POSD av ne- ng O/POSD av
and.then 1EXCL=ADD:DM 1EXCL:POSS-hand not.exist 1EXCL:POSS-foot not.exist
‘[…] and then also, I don't have hands or feet […]’ [RBT, ChM 030]
In a different construction, =ke and =t can co-occur in the same clause as they
mark different constituents. This is perhaps not surprising considering that =ke can occur
on two different elements in the same clause as shown above in §10.7.1 (note, however,
that topic-switch =t does not occur twice in the same clause). (755) shows that within
the same clause, the - ‘non-subject’ marked R argument may be marked by =t and
the unmarked A argument by =ke. The context of this folk story that allows this
coocurrence of =t and =ke is as follows. Bokolapo, a folk story fool character, asks his
children why they are calling him their father, as Bokolapo thinks that he is in a different
village. (755) directly follows Bokolapo’s direct speech talking to his children, and =t
thus occurs after direct speech, which is typical (see §10.7.2). After this sentence, which
explains how Bokolapo mistakenly talks to his children not knowing that they are his
508
children, the topic does indeed switch to the children. In this sentence, however,
Bokolapo still is the topic, and is marked as such by =ke.
(755) Additive topic-switch =t and =ke marking two arguments in the same clause
t
[a-so-m r aph n t ] [Bokol -p ab ng=ke] las n si l m
POSS-child-PL NSUBJ=ADD:TS NAME-male NPDL=TOP that.way=FOC word
thakdunronglo
th k-d n-r ng-l
answer-JOIN-instead-RL
‘that way, his (own) childred Bokolapo mistakenly answered like this’ [HI, BPh 013]
Now let us look at the relationship between the two focus clitics. (756) is an
elicited example that shows their complementary distribution in corrective focus
statements, which schematically are: ‘not X=le, but Y=si’.
(756) Contrastive/corrective with irrealis focus =le and realis focus =si
[ - =le -t ] [v - =si ki-t ]
1EXCL pig-meat=FOC:IRR NMLZ-cook NEG.EQU.COP bird-meat=FOC NMLZ-cook
'I don't/won't cook pork, I (will) cook chicken' [Elicitation SiT 090303]
Furthermore, =si and =le can co-occur (on different constituents) in the same
clause. The construction in which this is possible is content questions. Here, the content
question word is often marked by =si, while another element in the same clause can be
marked as focus as well, then receiving irrealis focus marking via =le (see also §11.3.2).
An example is (757).
(757) Realis focus =si on content question word; irrealis focus =le on other element
[…] nonke methan-sibongpopen banghinivetle kopusi pirthe
n n=ke meth n-sib ngp =pen b ng-hin -v t=le kop =si pirth
now=TOP dog.sp=with CLF:HUM:PL-two- how=FOC world
509
Finally, let us examine data that document how topic and focus marking can
interact. As shown in (758), both contrastive topic and contrastive focus can be marked
within the same sentence. The example consists of two parallel clauses, which have the
same verb ‘eat’, while both the A arguments and the locatives are contrasted. The
contrasting A arguments are nangpo ‘your father’ and nangtum ‘you (both)’. The
contrasting locatives are hemtap angsong ‘up in the tree house’ and mandu ‘(in) the field
hut’. While the two participants are marked as we would expect in the second clause, i.e.,
the A as topic and the locative as focus, the first clause has the A marked as focus and the
locative unmarked. This may be because the speaker had not planned the whole sentence
as he was producing the first clause. The second clause does give us evidence, however,
of how topic and focus marking can interact in an expected pattern. Note that irrealis
focus =le occurs in both clauses for different reasons: in the first clause because it is a
(subordinate) conditional clause, in the second clause because it is an unmarked
imperative.
(758) Contrastive topic and contrastive focus: =le ‘irrealis focus’ and =ke ‘topic’
nangpole hemtap angsong chote nangtumke mandule cho
[[nang-p =le h mt p angs ng ch -t ] nang-t m=ke mand =le ch ]
2-father=FOC:IRR tree.house high.up eat-if 2-PL=TOP field.hut=FOC:IRR eat
‘ “[…] if your father takes his meal in the hemtap, you eat in the mandu”' [CST, RO 017]
Lastly, (759) shows that =t , =ke and =le can occur within the same clause. Here,
we have two topics marked by =t and =ke as also seen above in (755), and in addition a
focused adverb marked by =le.
510
chodunnoi […]
ch -d n-n i]
eat-JOIN-INFRML.COND.IMP
'and then, “the rice bundle, today, eat them there!, go, you too eat together […]”’ [CST,
RO 017]
While all of the data presented above demonstrate that we can group topic
markers =ke and =t , as well as focus markers =si and =le, there is a single construction
that suggests a different grouping, which is the intensifier copy verb construction. Here,
the realis construction uses =t (§12.2.3.2), while the irrealis construction uses =le
(§12.2.3.3).
This section offers some evidence for the pragmatic basis of constituent order
variation. Karbi is verb-final like most Tibeto-Burman languages. However, it is
relatively easy to find clauses with arguments placed after the verb if they represent old
or given information. The following examples demonstrate this for mono-, bi-, and
trivalent clauses. In (760), the S argument occurs after the verb as it represents old
information, which is evidenced by the preceding question with verb focus rather than
argument focus.
511
In (761) and (762), the A argument and the O argument, respectively, occur after
the verb as they represent old information. In (761), this is evidenced by the use of the
distal demonstrative , which is frequently used for discourse-old information. In
(762), the evidence lies in how the speaker corrects herself after first saying ‘men and
women’ to saying ‘mothers and fathers’, thereby clearly making the O argument
‘children’ given information, because the terms ‘mothers and fathers’ versus ‘women and
men’ imply that this is about children.
(761) OVA
damchot aphi.... phutup kitirok theklongloklo
[NP]O V
[d m-ch t a-ph ....] [phut p ke-t -r k] th kl ng-l k-l
go-a.bit POSS-after hat NMLZ-leave.behind-CMPLV see-just-RL
(762) AVO
lasi pinso arlo ba apai apota
[NP]A
las [pins arl ] b [a-p i a-p t ]
therefore married.man woman or(<Asm) POSS-mother POSS-father=ADD
512
nelita kamatha
n -l t ke-math
1EXCL-HON=ADD NMLZ-think
‘so the men and women, or the mothers and fathers, if they make themselves equal for the
children, if they take this their own responsibility, then like other peoples' tribes we will
be successful (on a par with other peoples' tribes), is what I think [KaR, SWK 052]
In (763), the R argument in this trivalent clause is given information, because the
T argument aphutup is marked possessive and the R argument is the possessor, so the T
argument makes the R argument old or given information.
(763) TVR
lapenke aphutup pidetlo laphan
[NP]T V [NP]R
lap n ke a-phut p p -d t-l l -ph n
and.then=TOP POSS-hat give-PFV-RL this-NSUBJ
‘and then, he gave him his hat’ [SiT, PS 038]
Finally, although zero anaphora (§10.4.3) is very common and clauses with all
arguments overtly expressed are rare in the corpus of recorded texts, the default is for A
arguments to go in sentence-initial position, presumably because A arguments tend to be
topics. Whether there is a default for positioning of R and T arguments in trivalent
constructions is unclear (see §10.2.1.3). Due to the seeming default of sentence-initial A
arguments and the verb-final structure of Karbi, however, we can say that there is a
default for bivalent clauses with overtly expressed NPs to be AOV. However, there
certainly are examples where that is not the case, as already shown above with A and O
occurring after the verb. Another example of a bivalent clause that is not AOV is (764),
where the order instead is OAV. The first two lines offer the context for the OAV clause:
The king has two children, which, however, he never knew, because they were taken by a
witch right after they were born. After many years, the king finally meets his children and
finds out the truth. The example below starts with an exclamation by the king: ‘They are
indeed my children!’
513
(764) OAV
apot laso aricho abang mathalo, “ai!
ap t [las a-rech ab ng] math -l i
because this POSS-king NPDL think-RL how.strange!
nesomarlo laho ” pu
ne-oso-m r-l l ho pu
1EXCL:POSS-child-PL-RL EXCLAM QUOT
‘therefore, the king thought, “How strange! They are indeed my children!”' [CST, HM
084]
The next intonation unit describes the subsequent action by the king, which is to
take his children home. At this point, the king has been the topic, and the fact that the
children are his children (marked possessive by a-) is new information. This could
explain why we get OAV order here.
There are at least six restrictive focus markers, which differ in morphological
status, syntactic distribution, and semantic and pragmatic functions. Table 114 gives an
overview of the forms and distributions; the abbreviation ‘CLF/NUM/QUANT’ stands for
‘classifiers/numerals/quantifiers’. Note that all markers are glossed as ‘only’ in examples.
With respect to morphological status, there is one full word, ; one element
that occurs as a full word t when it syntactically modifies a noun or predicate but
occurs as a suffix on classifiers, numerals, and quantifiers; and finally, there are four
markers that only occur as suffixes. Full words occur as post-head modifiers.
514
Table 114. Distribution of restrictive focus markers176
Occurs on/with > CLF/NUM/QUANT Noun Verb
- + +
t~ - + +
- t + - -
- t + - -
- - + +
- t - + -
- t ? - +
As for their syntactic distribution, the markers differ in whether they attach to
classifiers/numerals/quantifiers, nouns, predicates, or a combination of these, as shown in
the Table 114. Note that t - t is the only marker whose form differs depending on
what type of head it modifies. Finally, note that is the only element for which a
likely historical origin still exists in the language: the verb ‘be naked’. While t
looks like a relator noun, there is no homophonous lexical item in modern Karbi that
could represent its grammaticalization source element. The - t suffix is homophonous
with exhaustive perfective - t (§6.8.1), and the exhaustive semantic component does
make a relationship between the two suffixes likely, as does the fact that restrictive focus
- t does not occur on verbs.
Table 115 shows which restrictive focus marker indicate which particular function,
of which four have so far been identified.
516
kachepechuji apor ave tangho
ke-che-pa-ch -j ] a-p r] av t ngh ]
NMLZ-RR-CAUS-suck-IRR2 POSS-time not.exist REP
‘they just kept pounding the rice for the rice beer cake, the whole time they didn't get to
take rest, they were pounding the rice beer cake, they didn't even have time to breast-feed
their children’ [WR, BCS 030]
jonnivetlo jonphlivetlo
j n-n -v t-l ] [j n-phl -v t-l ]
CLF:animal-two-only-RL CLF:animal-four-only-RL
that way, he kept eating and eating, and from the one hundred hens and one hundred
roosters, two hens and two roosters only, only four (were left) [SeT, MTN 028]
Lastly, the function ‘not meeting the full expectation’ with respect to a referent or
event presupposes a contextually defined semantic and pragmatic scale of referent and
events that fulfill a particular expectation more or less. In the folk story example in (769),
a witch has previously taken the Karbi woman’s children and abandoned them in the
jungle, while putting two pieces of firewood in the cradle, which were meant to imitate
the shape of the children. In (769), the witch says to the father, the king, that the Karbi
woman had given birth to ‘only’ or ‘merely’ two pieces of firewood. Pieces of firewood
clearly do not fulfill the expectation of actual human babies and would be ranked lower
in an intuitive scale on what a woman should give birth to.
517
kehacheke thengpi abeng angse jaho hini... pu
ke-hach ke thengp a-b ng angs jaho hin pu
NMLZ-be.born=TOP tree/wood POSS-piece only look.there! two QUOT
518
CHAPTER XI
This chapter offers an overview of different main and subordinate clause types. It
does not discuss nominalized and historically nominalized clause types, as they are dealt
with in Chapter IX. Likewise, it does not discuss declarative main clause types;
intransitive, transitive, and ditransitive clauses - or mono-, bi-, and trivalent verbs - are
discussed in §10.2.
The structure of this chapter is as follows. It begins with non-declarative main
clause types in §11.1: interrogatives, imperatives, and hortatives. In §11.2, subordinate
clause types and constructions are discussed, again, excluding any discussion of
nominalization. In §11.3, an overview of irrealis clause types is provided, which emerge
as a function of irrealis-sensitive non-final and focus markers. The next section §11.4
deals with (non-nominalized) insubordination constructions, i.e., formally subordinate
clauses functioning as main clauses. In §11.5, the grammatically less prominent domain
of clause coordination (as compared to clause chaining and other subordination) is
discussed. Finally, §11.6 offers a few remarks on the lack of a grammaticalized syntactic
pivot in Karbi to determine participant role continuity across clauses.
Non-declarative main clause types are at the core of a set of constructions that
have a property in common referred to as ‘irrealis’ in this dissertation (see §11.3.2).
Non-declarative main clause types discussed here include interrogatives (§11.1.1),
imperatives (§11.1.2), and hortatives (§11.1.3). Another non-declarative main clause type
discussed in Chapter XII is exclamatives (§12.3.8).
11.1.1. Interrogatives
519
with respect to the degree of the speaker’s expectation that the reply will confirm the
truth of the proposition. On this continuum, polar interrogatives (§11.1.1.4) are neutral
and open, while tag questions (§11.1.1.7) indicate a strong expectation towards the truth
of the proposition. Inbetween we find disjunctive interrogatives (§11.1.1.5), which are
closer to polar interrogatives, and interrogative assumptions (§11.1.1.6), which are closer
to tag questions. The last subsection is dedicated to the feedback request construction
(§11.1.1.8).
The question particle =ma occurs mostly in polar interrogatives (§11.1.1.4.1) and
disjunctive interrogatives (§11.1.1.5), as well as, in a pragmatically marked way, in
content questions (§11.1.1.2.2).
A polar interrogative particle of this form ma is found in a large number of
Tibeto-Burman languages. As pointed out by Thurgood (1983), there is a clear historical
link between the interrogative ma and the common *ma Sino-Tibetan preverbal negative,
such that the interrogative must have developed from the negative. The reconstruction
takes the common ‘A-not-A’ disjunctive question construction177 as a starting point. In
this construction, a polar question like ‘do/did you eat?’ is expressed as ‘eat not-eat’. In
order for the negative to be reanalyzed as a polar interrogative particle, all that has to
happen is for the second repetition of the verb to be dropped.
When eliciting content questions via translation from English, the verb is not
marked by the question particle =ma, and also in texts, it appears that not using =ma is
pragmatically more neutral.
(770) and (771) are content questions in which the verb occurs without the
question particle =ma. In both examples, the verb is nominalized (§9.7.3.1). In (770), the
content question word si ‘why’ apparently diachronically contains =si ‘focus’ (see also
§11.3.2) and, presumably as a result, cannot occur with either =si ‘focus’ or =lo ‘focus’,
which is otherwise common for interrogative pronouns and adverbs. In (771), t ‘where’
is marked by =lo ‘focus’.
While most questions have the content question word in clause-initial position,
this is not necessarily the case, as demonstrated in (771). Note that (771) is not an echo
question that asks the addressee to reconfirm something already said, but the question
asks for previously not addressed information. It is said by somebody listening to a
storyteller tell this folk story (as indicated by the curly brackets), and the question is
getting at the location of two protagonists of the story. It could therefore be the case,
521
however, that the speaker in (771) knew the answer to his question, as he might have
known the plot of the folk story.
If the question particle =ma occurs at the end of a content question, then the
question becomes rhetorical and is either not intended to elicit an answer, or at least
signals that there is no ‘correct answer’ to the question that would satisfy the person
asking. An elicited minimal pair is (772). Whereas (a) without =ma is pragmatically
neutral and simply asks for new information, (b) may be asked in a situation where
everybody had been worried about the absence of the addressee.
(772) Content question minimal pair with and without =ma ‘question particle’
(a) t t=si ke- m- ? (b) t t=si ke- m=ma?
yesterday where=FOC NMLZ-go-RL yesterday where=FOC NMLZ-go=Q
‘where did you go yesterday?’ ‘where did you possibly go yesterday?’
(Elicitation KT 111208)
Examples from the corpus show the same function of content questions with =ma.
In (773), from a folk story, the king asks a rock why it had killed his daughter. Since it
was an accident, which the king is aware of, there is of course no satisfactory answer to
the question, so =ma is used here.
Similarly, in the folk story from which (774) is taken, the question ‘why didn’t
you get up?’ is not intended to elicit an answer, because it is followed up by a command
that the addressed group of tigers should hurry up.
523
(776) Polar interrogative with =ma and with answer
“ ” ,
ch -j m p ne-ph n arj -l ]
be.fine-IRR2=Q QUOT 1EXCL-NSUBJ ask-RL
t t “ ”
[am tsi n t p -l ch
and.then 1EXCL=ADD say-RL be.fine
'”is it fine?”, they asked me, and I replied, “it's fine”' [SH, CSM 049]
(777) shows that this type of interrogative with =ma also allows the verb to be
negated.
524
a'ik avelo ate avelo
a- k av -l a-t av -l
POSS-older.brother not.exist-RL POSS-elder.sister not.exist-RL
‘he didn't have a mother anymore, he didnt have a father anymore.... he didn't have any
brothers or sisters anymore’ [HK, TR 004]
{enutvetlo?}
e-n t-v t-l
one-CLF:HUM:SG-only-RL
‘{he was alone?}’ [HK, TR 005]
enutvetlo {to}
e-n t-v t-l t
one-CLF:HUM:SG-only-RL OK
‘he was alone {OK}’ [HK, TR 006] 178
Figure 23 offers the spectrogram and waveform of the question enutvetlo? and the
answer enutvetlo, with the blue line representing F0 and the yellow line representing
intensity (figure made with Praat (Boersma and Weenink 2013)).179
178
The audio file for HK, TR 005-6 is available under the DOI name 10.7264/N3736P5Z, see Appendix B.
179
Note that in both speakers’ productions, both alveolar stops are elided as is common in hypoarticulated
speech (§3.7.1). Segmentically, they therefore both produce “ nuvelo.”
525
While the first enutvetlo, which represents the question, shows the expected rise
in F0, it is perhaps surprising that the answer closely mirrors the interrogative F0, as more
clearly seen in Figure 24, which only displays F0 over time (also produced with Praat).
As we can see in Figure 24, F0 goes up to just over 200 Hz in the question and to
just below 200 Hz in the answer, and these are both male speakers.
0
300
250
200
Pitch (Hz)
150
100
80
e n u v e l o e n u v e l o
0 1.051
Time (s)
Figure 24. F0 contour of interrogative and subsequent declarative enutvetlo ‘he was
alone’ ( TR 005-6)
526
11.1.1.5.1. Type 1: A=ma A-NEG
Examples of the second type, ‘A=ma B’, are provided in (780) and (781), which
both have nominal predicates.
jangrengsolo
jangr ngs -l
orphan-RL
‘it was an orphan with no parents left’ [HK, TR 012]
180
However, while in Mandarin Chinese, the construction involves simply an affirmative predicate
followed by its negated counterpart, both Karbi and Garo require the first affirmative predicate to be
additionally marked by the question particle =ma, i.e., ‘A=ma A-NEG’.
527
However, verbal predicates may also occur in this construction, for example -
= - ‘eat-GOOD=Q eat-BAD > is it edible or not?’. It is not clear whether verbs
can only occur in this construction if they denote opposites (in which case they mirror the
‘A=ma A-NEG’ construction shown above).
], [town - , - ]
be.much=Q town POSS-LOC=FOC be.much=Q town POSS-LOC
‘the people that don’t behave well, is it mostly the village people or the town people,
from the rongsopi?’ [JB, SWK 174: SiT]
528
has more knowledge and asks for the same information more directly, rephrasing it as a
disjunctive question. He stops, however, after the coordinating ma, leaving the
alternative of the disjunctive question unexpressed.
Kamrup a-deng=pen=ma, …
DISTRICT POSS-district=from=Q Q
‘which district are you from? From the Kamrup district, or…?’ [JB, SWK 021: SiT]
The ‘interrogative assumption’ clitic =bo marks statements whose proposition the
speaker assumes is true while still eliciting a reply that is expected to confirm the truth of
the proposition.
An example of a =bo marked interrogative assumption is (785), in which the
curly brackets indicate a native Karbi speaker different from the main storyteller of this
text. In this example, there are two =bo marked interrogative assumptions, both with a
third person subject. In both cases, they are confirmed: the first question via the
affirmative interjection a, and the second question via affirmatively repeating the
predicate.
pulobo?} kekatchonlo
p -l =bo} ke-k t-ch n-l
like.this-RL=ITROG.ASSUM NMLZ-run.HUM-very.quickly-RL
‘{and so, he is not waiting, huh? is he running away} he is running away’ [HK, TR 116]
In (786), the subject of the =bo marked clause is second person. Here also, a full
reply follows the interrogative assumption.
529
(786) Interrogative assumption =bo
ke methan-sibongpoke “hem chirimkangetlobo,
k meth n-sib ngp ke] [h m che-r m-k ng- t-l =bo
and.then dog.sp=TOP house RR-keep.in.order-leaving-PRF-RL=ITROG.ASSUM
Tag questions marked by are used to seek confirmation for the truth of a
proposition that the speaker strongly believes in. In (787), a tag question marked by is
confirmed in a reply with the same adverbial predicate.
Whereas d ‘question tag’ occurs after realis clauses and elicits confirmation of
the truth of a proposition, ‘OK?’ occurs after irrealis clauses and represents a request
to the addressee to confirm their ‘having taken notice.’ Irrealis contexts in which is
used for feedback request include imperatives (§11.3.2), as in (788), and deontic clauses
with ‘need’ (§11.3.4), as in (789).
530
chinglunoi juinoi pu amat […]
chingl -n i j i-n i pu am t
take.bath-INFRML.COND.IMP play-INFRML.COND.IMP QUOT self
‘[…] stay here, okay?, here do everything, and take your bath and play” he said, and then
[…]’ [CST, RO 049]
181
I have heard this used by somebody who was talking to a four-year old child.
531
(790) Bare stem imperative
“ako nangpole mandu cho tangte...
ak nang-p -le mand ch t ngt
on.the.other.hand(<Asm) 2:POSS-father-FOC:IRR field.hut eat if
In this part of the story, the king is looking for somebody he can hold responsible
for the death of his daughter. The elephant, who is speaking here, explains why he is not
responsible, and suggests the king ask the Voarbipi bird instead as somebody who would
be able to provide further information. It is this implication of an expected consequence
that will follow from the suggested action that is encoded in - (and - ).
182
It is admittedly still surprising that an informal form is used in talking to the king. Note, however, that
the first person pronoun in the previous clause also occurs without the honorific suffix - (§4.5.1),
which suggests that this interaction with the king is indeed occurring on a rather informal level.
532
Another example is (792), where the function of - is to highlight a negative
consequence if the suggested action is not performed. In the given part of this folk story,
a stepmother is speaking to her stepchildren, who she has been mistreating. Here, she
wants to prevent them from eating together with their father, so their father will not
notice what bad food she has them eat.
Note also in (792) that the bare stem imperative (§11.1.2.1) occurs in the last
clause of this intonation unit, thus appearing functionally equivalent to - here.
Conative imperative -t is used when the speaker suggests the addressee try and
do something and see what happens. Compared to - i and - n (§11.1.2.2,
§11.1.2.3), -t is less direct or immediate, although they may be used in the same
context suggesting that their functions are not very different. This can be seen in the
example. In this folk story, a father desperately sees no other way to deal with a difficult
situation than to abandon his children. In (794), he has taken them to a place in the jungle
and tells them to try and stay there, using -t ; and that they should take baths and play
there, using - .
juinoi” […]
j i-n i
play-INF.COND.IMP
‘he took them to the place of a very big rock, and then, “mothers, stay here, okay, here
take your bath and play” […]’ [CST, RO 049]
183
Examples of when -t rather than one of the other imperative constructions would be used centered on
moral imperatives, such as ‘(You must) love your parents!’ or ‘(You must) love the poor!’ Using, for
534
reach home, which may be why this imperative construction is used here, rather than one
of the other ones.
(795) Imperative -t
t t ]
d -j aj t le d -j t [m nang-h m chV-d m
stay-IRR2 GENEX=FOC:IRR stay-IRR2=ADD future 2:POSS-house RR-go
ahomoike lapu t t t t
a-hom i ke lap che-v ng-th -ra] [langs a-tov r-th t
POSS-time(<Asm)=TOP this.side RR-come-again-NF:IRR this POSS-road-exactly
chepaletu
che-pa-l -t ]
RR-CAUS-reach-UNCOND.IMP
and then, you will stay with my grandmother one night and one day or however long, and
later when you go home, again come like this, come on the exact same road, and then go
to the familiar place (i.e. where the dog is staying) and then go and make yourself reach
your house [KK, BMS 089]
Elicitation sessions conducted for this research suggest that Grüßner's (1978: 96-7)
classification of imperative suffixes on a politeness continuum is a secondary (and
therefore less consistent) pragmatic dimension to these constructions. The primary
functional differences appear to be as follows. First, if the speaker suggests for the
addressee to do something given particular circumstances and resulting consequences,
then the conditioned imperative suffixes - and - are used. Second, if the
implication is that the addressee should try to do something and see what will happen,
then the conative imperative suffix -t can be used instead. Third, if the addressee
should do something unconditionally, the imperative suffix -t is the most appropriate.
example, the conative imperative -t (§11.1.2.4) instead in this context, the meaning would be ‘Love them
and see what happens / what the consequences are.’
535
11.1.2.7. Prohibitive -
(796) Prohibitive -
ai nepran neenri, nemui
i [ne-pr n ne n-r ] [ne-m i
how.bad! 1EXCL:POSS-life 1EXCL:NSUBJ=take-NEG.IMP 1EXCL:POSS-EE:pr n(<Ind)
neenri, richo
ne n-r ] rich
1EXCL:NSUBJ=take-PROH king
'"please, don't take my life, king"' [RBT, ChM 068]
”
ne-oso-m r-l
1EXCL:POSS-child-PL-RL
the king said, 'don't go then! you are my children' [CST, HM 115]
536
The form vungritha! in (798) is used in a context, in which the speaker asks the
addressee to suspend the action just for a moment, not in general. It thus appears that
combining prohibitive - with imperative suffixes serves to convey the pragmatic
distinctions inherent in the different imperative markers, which are otherwise
undifferentiated in the single prohibitive - .
11.1.3. Hortatives
The general hortative marker that indicates that the speaker suggests they
themselves together with the addressee, i.e., first and second person, perform an action
(also cross-linguistically referred to as ‘cohortative’) is the suffix - .184 Several
instances of - ‘hortative’ occur in (799).
(799) Hortative -
ne kedam aling nangdunnoi setame, chonang junnang!
[n ke-d m al ng nang d n-n i set m ] [ch -n ng j n-n ng]
1EXCL NMLZ-go INDEF CIS=join-INF.COND.IMP nevertheless eat-HORT drink-HORT
ha nephi along!
h ne-ph al ng]
over.there 1EXCL:POSS-grandmother LOC
'Join me, wherever I go, but let's eat and drink and let's keep everything nicely (here, i.e.
put everything in order before we leave), let's go there, to my grandmother's place' [KK,
BMS 033]
184
Grüßner (1978:97) lists this suffix with a low tone. According to my language consultants, there is a low
tone form of this suffix, which, however, is used to convey a conative sense, see below.
537
The negative hortative is formed with the verbal negative suffix - (§6.7). An
example is (800).
“thatnangne ti sarbura”
th t-n ng- ti s rbur ]]
slaughter-HORT-NEG EMPH old.man
‘ “[…] let's kill us a hen tonight”, he said, (but) the wife said, “let's not, old man (and I
won't change my mind)”' [SeT, MTN 007]
Interestingly, the hortative can be turned into a question by adding the question
particle =ma (§11.1.1). For example, - = ? ‘go-HORT=Q’ means ‘should we
go?’
A change in tone can be used for a change in pragmatic function. While the
regular mid tone form - carries the connotation of a more immediate ‘let’s (do) right
now!’, a low tone form - is used for a more general and/or conative interpretation,
i.e., ‘let’s (in general) (do) (or try to do)’, as in (801). (See also §11.1.3.3 on the extended
hortative form - , which also carries a conative connotation.)
538
Peterson 2003:415), which may be cognate with the Karbi hortative, in which case we
have to assume a much larger time depth.
nangpai an patunnang” pu
nang-p i n pa-t n-n ng pu]
2:POSS-mother rice CAUS-cook-HORT QUOT
and then, the grandmother, and then "come, daddy, let your mother cook (rice)" [KK, CC
018]
As (803) and (804) show, verbs marked with pa- ‘causative’ plus - ‘hortative’
may also be directed towards speech act participants, as the verbs here are marked with
the SAP non-subject marker nang= (§6.3.1). The examples also show that this
construction is negated via negation of the verb stem, followed by - ‘hortative.’
nangpatoklonglenang pu
nang=pa-t k-l ng- -n ng pu
1/2:NSUBJ=CAUS-pound-GET-NEG-HORT QUOT
'”we don't get to pound the rice for the rice beer, even in the future when the world is
mature, we, the Bey Ronghang shouldn't be made to have to pound the rice for rice beer
cake”, (he) said' [WR, BCS 034]
539
(804) Hortative - on negated verb stem, directed towards first person plural
t t t y t
m pirth ke-ingt ng t n ng B y ke- t pen
future world NMLZ-be.strong=even you CLAN NMLZ-be.yellow=with
185
Grüßner (1978:97) additionally reports a form - , which is, however, not known to consultants to
this research. It also needs to be noted that Grüßner spells what is here called the ‘conative hortative’
as - , and suggests that its function is that of an immediate hortative rather than a conative hortative.
540
aphanke rira jonghe setame, theklonglelo”
a-ph n ke r -r jongh set m thekl ng-C -l
POSS-NSUBJ=TOP search-NF:IRR as.much.as nevertheless see-NEG-RL
‘[…] “let's keep working and eating (i.e., let's live as usual), even if we keep searching
for my grandmother, we won't see her”' [KK, BMS 017]
chorappetsinang
ch -r p-p t-sin ng
eat-together-all-HORT:CON
‘no, o mothers, today let's eat together now (and see what's going to happen)!’ [CST, RO
027]
The children (as the addressee) disagree with this suggestion in (806), and, as a
consequence, in (807), the father, now emphatically using - , makes an executive
decision saying that they should have their meal together.
186
A conative hortative is also expressed via tone change. If the low tone instead of the mid tone is used on
the basic hortative - , then it also carries a conative connotation, see §11.1.3.1 above.
541
11.2. Non-Nominalized Subordinate Clause Types
There are three suffixes in Karbi that indicate the non-final status of a verb: -si
‘non-final:realis’ (§6.10.1), -ra ‘non-final:irrealis’ (§6.10.1), and -pen ‘non-final:with’
(§6.10.2). While -si and -ra are realis/irrealis-sensitive counterparts of one another
(depending on the realis/irrealis specification of the final verb, §11.3), -pen historically
originates in the postposition =pen ‘with; from’ (§7.8.1). Non-final -pen is younger than
-si and still shows traces of nominal affiliation in that it very often (but not obligatorily)
attaches to nominalized verbs. The most frequent non-final marker is -si with 104
occurrences in the corpus. Non-final -pen only has 73 occurrences. Irrealis non-final -ra
only occurs 21 times.
The realis non-final marker -si occurs in typical clause chains that denote
subsequent events. In (808), four non-final -si marked clauses occur in temporal sequence.
Note, however, that is very rare to find such long clause chains. It is most common to
only find one non-final clause marked by -si, followed by a final clause. Nevertheless,
while both -si and -ra may, if rarely, occur in chains of several clauses, -pen never occurs
on more than one non-final clause to denote a temporal sequence.
542
theng akhangra ksi, hem damsi,
[th ng a-khangr -k k-si] [h m d m-si]
firewood POSS-basket.for.firewood leave-in.a.fixed.place-NF:RL house go-NF:RL
"dojoinoi, po!"
[d -j i-n i p ]
stay-quietly-INF.COND.IMP father
‘and then, the mother went and unloaded the firewood in the Pharla (Veranda), then went
inside the house, sat down in the Hongkup, gave the child the milk, (and said) '”be quiet,
daddy”' [KK, CC 015]
In addition, non-final clauses may bear other semantic relationships to the final
clause, besides temporal sequence. In (809), -si occurs on a manner motion verb
‘steer’ (here: ‘ride a bike’), while the final verb ‘come’ indicates the deictic
direction towards the reference point.
In (810), -si marks a subordinate clause that has a causal relationship to the final
clause.
543
The high degree of clausal integration exhibited by clause chaining with -si is
shown in (811) and (812). Here the O argument of the final verb is placed at the
beginning of the sentence, thus appearing to be structurally under the scope of the non-
final clause.
thaptekanglo
th -te ng-l
put.inside-leaving-RL
‘so (the mother) fixed the cradle, and went to wash clothes, and then, she fixed the cradle
for the two children, and put them inside’ [CST, HM 010]
All three non-final markers may also occur on negated verbs. An example is with
-si in (813), where the semantic relationship between the non-final and the final clause is
one of causality.
544
In the excerpt in (814), -si marks the ‘head’ part of the tail-head linkage
construction (§12.1.2). Both -si and -pen commonly occur in the tail-head linkage
construction in this way.
aphanta ingthumponlo
a-ph n t ingth m- n-l
POSS-NSUBJ=ADD go.and.bring-take.away-RL
‘so we went there to pick up my elder sister and brother-in-law as well’ [SH, CSM 010]
ha kedamlo
h ke-d m-l
over.there NMLZ-go-RL
'and then, the wife put the basket rope around her head and was just leaving to go over
there (to her parents' place)' [SeT, MTN 041] 187
187
The audio file for SeT, MTN 041 is available under the DOI name 10.7264/N3639N04, see Appendix B.
545
The prosodic extra high tone to indicate ‘non-final’ is sometimes also used in
addition to ‘non-final’ suffixes, in which case it is typically the syllable preceding the
‘non-final’ suffix that receives the extra high tone, specifically - ‘in a fixed place’ and
- ‘big:AO’ in (816), and - ‘right then’ in (817).
"dojoinoi, po!"
[d -j i-n i p ]
stay-quietly-INF.COND.IMP father
‘and then, the mother went and unloaded the firewood in the Pharla (Veranda), then went
inside the house, sat down in the Hongkup, gave the child the milk, (and said) '”be quiet,
daddy”' [KK, CC 015]188
188
The audio file for KK, CC 015 is available under the DOI name 10.7264/N3FN14GV, see Appendix B.
189
The audio file for HK, TR 101 is available under the DOI name 10.7264/N3ZK5DXG, see Appendix B.
546
11.2.2.1. Verb Juxtaposition
In (818) and (819), the verb t ‘know how’ occurs in the CC juxtaposition
construction. In most instances of this construction, it looks like a serialization
construction, in which the root of the CC verb and the root of the CC-taking verb appear
directly serialized with no other elements intervening. However, (819) shows that a more
complex predicate in an adverb construction (§8.3.2) may also be followed by
complement-taking t ‘know how’.
547
11.2.2.2. Indirect Questions
In (822), three separate indirect questions are all marked with - ‘indefinite’ and
then anaphorically picked up on in the following topic NP laso alam ‘(about) this matter’.
Note that in the third indirect question CC kolosontong alang kemunthilone ‘how
is she thinking about it’, the interrogative pronoun ‘how’ has the suffix -t
‘indirect interrogative pronoun’ attached to it, see §4.5.4.2.
548
11.2.2.3. Topic =ke Marked Complement Clauses
Topic =ke may also mark CCs, as in (823). However, this construction is very
rare in the corpus.
Both the simple quotative pu and the more complex form pusi, which is derived
from -si ‘say-NF:RL’,190 function as complementizers. In (824), both pu and pusi occur
within the same intonation unit, both times marking the end of direct speech. First, we
have pu occurring after the CC nepo kajoklu ‘our father is a fool’, which represents direct
speech as evidenced by the use of the first person exclusive possessive prefix ne- rather
than the general and third person possessive prefix a-. Similarly, further along, we have
pusi occurring after the direct speech CC […] ‘our father
[…] has come back after going to the market’.
190
Analogous quotative complementizers derived from a non-final marked verb ‘say’ are found in various
subbranches of Tibeto-Burman, for example, in the Tamangic language Chantyal (Noonan 2006: 4), the
Tani language Galo (Post 2007: 848), and the Kuki-Chin language Daai Chin (So-Hartmann 2009: 321).
549
kevanglo pusi asomar abang mathalo
ke-v ng-l ] pusi] [a-so-m r ab ng] math -l
NMLZ-come-RL QUOT.COMP POSS-child-PL NPDL think-RL
‘but the children didn't know at all “our father is a fool”, they thought, “our father has
come back after going to the market”' [HI, BPh 014]
In the same story, talking about the same event, the same construction using pusi
is also used to indicate the end of indirect, rather than direct, speech, where instead of
nepo ‘our father’, the same referent is referred to by his character name Bokolapo, see
(825).
t
a-so-m r ab ng math -l
POSS-child-PL NPDL think-RL
‘the children thought that he had already gone to the market’ [HI, BPh 010]
550
11.2.3.2. Purpose Clauses with Quotative Complementizers
Purpose clauses may be nominalizations (§9.5), but they may also be finite
clauses followed simply by quotative pu or by the quotative complementizer pusi, as in
(827).191
11.3.1. Overview
192
‘Irrealis’ is a commonly applied descriptive label, but has also been argued not to constitute a (single)
grammatical category (Bybee, Perkins, and Pagliuca 1994; Mithun 1995; Chafe 1995; Bybee 1998) - a
claim these Karbi data certainly lend support to if we compare the irrealis categories of -ra and =le as well
as - ‘irrealis1’ and - ‘irrealis2’ (§6.9.2). The best explanation for the functional overlap and differences
between the irrealis ranges of the four markers no doubt lies in their differential diachronic developments,
as a general principle also argued by Cristofaro (2012) and Mauri and Sansò (2012).
193
Preliminary attempts to elicit -ra in deontic and conditional subordinate clauses suggest that -ra can be
used in those contexts as well.
552
native speakers to be the correct form and other native speakers prefer using realis non-
final -si.
Table 117. Irrealis contexts for -ra ‘NF:IRR’ and =le ‘FOC:IRR’
Grammatical context -ra ‘NF:IRR’ =le ‘FOC:IRR’
Non-declarative speech acts Questions X X
Imperatives X X
Hortatives X X
Negation (X) X
Deontic clauses (with ‘need’) - X
Conditional subordinate clauses - X
Both the irrealis non-final marker and the irrealis focus marker occur in all three
types of non-declarative speech acts: questions, imperatives, and hortatives. Respective
examples of -ra are offered in (829), (830), and (831).
553
(831) Irrealis non-final -ra; hortative-marked final verb
hormu horton chirim chibikangvetra, dunnang,
horm hort n chV-r m chV-b -k ng-v t-r ] [d n-n ng
thing EE:horm RR-put.in.one.place RR-keep-give.leave-nicely-NF:IRR join-HORT
ha nephi along!
h ne-ph a-l ng]
over.there 1EXCL:POSS-grandmother POSS-LOC
'Join me, wherever I go, but let's eat and drink and let's keep everything nicely (here, i.e.
put everything in order before we leave), let's go there, to my grandmother's place' [KK,
BMS 033]
Examples of =le in all three types of non-declarative speech acts are given in
(832), (833), and (834). Note that (832) is interesting, because in questions, both focus
markers may occur: realis focus =si only ever occurs on interrogative pronouns, like
‘how’ here, whereas irrealis focus =le may occur on an additional focused element, here
banghinivet ‘the two (of us)’.
554
In (835), both markers co-occur: =le occurs on the subject NP in a non-final
clause marked by -ra, where the final verb is a hortative form marked by -
‘hortative’.
(835) Irrealis focus =le and irrealis non-final -ra; hortative-marked final verb
“dah! etumle arveng chepachap chepachapra,
d h e-t m=le arv ng che-pa-ch p che-pa-ch p-ra
go! 1PL.INCL-PL=FOC:IRR feather RR-CAUS-pile.up RR-CAUS-pile.up-NF:IRR
11.3.3. Negation
For non-final suffixes, negative contexts are ambiguous. Some native speakers are
happy to use either -ra, as in elicited (837), or -si, while others only accept -si and ban -ra
entirely from negative contexts.
555
(837) Irrealis non-final -ra; negated final verb
ng k - n t n-ra ch - -
3 meat POSS-curry cook-NF:IRR eat-much-NEG
'he cooked food but didn't eat much' [SiT 090302]
Preliminary elicitation suggests irrealis non-final -ra can be used as well if the
final verb is a complex predicate involving ‘need, must’.
cho ”
ch ]
eat
‘ “[…] if your father takes his meal in the hemtap, you eat in the mandu!”' [CST, RO
017]
556
In this context as well, preliminary elicitation suggests irrealis non-final -ra can
be used if a non-final claused is linked to a following conditional subordinate clauses
(schematically, ‘[Vx-ra]Clause.1 [Vy-COND]Clause.2, then…’).
While (historically) nominalized main clauses are discussed in §9.7, this section
deals with insubordination constructions that are not based on nominalization. Following
Evans (2007:367), insubordination is defined here as “the conventionalized main clause
use of what, on prima facie grounds, appear to be formally subordinate clauses”.
11.4.1. Main Clauses Marked with =ke ‘topic’: Background Information Construction
Main clauses marked with =ke ‘topic’ provide a background against which a
statement just mentioned (or about to be mentioned) is meant to be understood (see
§12.3.10). Since clauses marked with =ke may function as subordinate clauses
(§11.2.2.3), main clauses marked with =ke are best treated as instances of
insubordination.
A =ke marked clause may follow or precede a main clause that it is contextually
linked to, as in (840) and (841).
(840) ‘Topic’ =ke marked clause following its contextually related clause
- … - =ke
be.cold-a.bit rain NMLZ-fall=TOP
'It's a bit cold… it's raining, that's why' [OH 121011: 001]
(841) ‘Topic’ =ke marked clause preceding its contextually related clause
apok ingchirdukke…
a-p k ingch r-d k=ke
POSS-stomach be.hungry-INTENS=TOP
557
choridamlo tangho […]
cho-r -d m-l t ngh
AUTO.BEN/MAL-search-go-RL REP
‘he was suffering from hunger, and so, he went here and there to look for wild
vegetables […]’ [HK, TR 014-5]
If the context is provided by a question, a =ke marked main clause may also be
used in the answer, as in (842).
Indirect question complement clauses such as ‘[I do not know [whether he leaves
or stays]CC]’ are discussed in §11.2.2.2. They are typically based on a disjunctive or polar
question ‘X or not X’, or ‘X or Y’. The two clauses X and Y are then marked by -
‘indefinite’.
In the following two examples, indirect question complement clauses are used
without a main clause, making them instances of insubordination. In (846), the function is
to ask a polar question (for Karbi polar and disjunctive question structure, see §11.1.1.4
and §11.1.1.5) in an indirect way, as can be done in English with ‘I wonder if you heard
of this’ instead of the direct ‘Have you heard of this?’194
194
Less direct questions are a common politeness (because face-saving) strategy; insubordination
constructions cross-linguistically often serve this kind of function (Evans 2007).
559
(846) Using stand-alone indirect question instead of direct polar question
si nanglitumta lason arjulongne
s nang-li-t m t las n arj -l ng-n
therefore 2POSS-HON-PL=ADD that.way hear-GET-INDEF
Another example is (847), where the ellipsed main clause has to be understood as
‘we don’t know’ rather than ‘I wonder’.
Clause coordination, i.e., coordinating two main clauses into one unit, is not
common in Karbi, as in clause-chaining languages in general. However, there are a few
strategies available, which are discussed below. Note that it is a cross-linguistic problem
to define coordination as involving two truly ‘equal’ clauses with none showing any signs
of dependency on the other (Haspelmath 2004). In Karbi, it is still up to future research to
investigate this further, including, for example, the question of whether or not the
constructions discussed below allow for the subject to change across coordinated clauses.
560
11.5.1. Conjunctive Coordination
195
This is a matter for future research. That said, it is cross-linguistically often difficult to distinguish
between clause and VP coordination, and therefore sometimes useful to group them as ‘verbal conjunction’
(Haspelmath 2004, 2005).
561
(849) Clause coordination via additive-marked NPs
[…] langta junlong anta cholon […]
[l ng t j n-l ng n t ch -l ng
water=ADD drink-GET rice=ADD eat-GET
‘[…] they got to drink water and they got to eat rice, […]’ [KK, BMS 056]
In the other construction, the additive particle may be used in a copy verb
construction, in which a copy of the verb root occurs with =t , followed by the verb root
with optional addition of morphology (see also §12.2.3 for other copy verb constructions
that have discourse functions). In (850), the two events ‘cook much’ and ‘eat much’ are
coordinated via this construction.
Note that conjunctive coordination via a copy verb construction with the additive
particle also exists in Hakha Lai (Central Kuki-Chin) (Peterson and VanBik 2004:348).
562
ahem arit dolo
a-h m a-r t d -l ]
POSS-house POSS-field exist-RL
‘[…] Bey the Black had his (own) house and property, Bey the Fair likewise had his
(own) house and property, and Bey Ronghang, the young one, also had his (own) house
and property' [WR, BCS 004]
Examples from personal narratives are (852) and (853). (852) connects S
arguments while repeating the verb. (853) is a more complex case involving negation,
which requires translation into English as ‘neither… nor…’
dunji pulo
d n-j ] pu-l
join-IRR2 QUOT-RL
‘Sampri, Rasinja, and Linda all were going to join’ [SH, CSM 004]
the'o'a la theklonglo
th - ~ l thekl ng-l
be.big-very~DIST.PL this see-RL
neither with thatch nor with tin, but with pieces of stone they cover (their roofs), the split
off pieces are quite big, those also we got to see [SiT, HF 050]
564
11.6. Lack of a Syntactic Pivot
565
CHAPTER XII
196
Acoustically, parallel structures are often accompanied by prosodic marking, which is what I understand
much of Solnit’s notion of “rhythmic effect” in his definition to refer to.
566
text genre. Below I discuss first EE parallelism and then biclausal NP coordination
parallelism.
EE constructions may be built on nouns or verbs. In example (857) from a folk
story, the EE t ‘birds’ is the basis for the parallelism in the relative clause votek
ingrengre voso ingrengre along. A similar example from a personal narrative is (858),
where the parallelism is based on the verbal EE ‘prepare’, which occurs in an
adverb construction (§8.3.2).
nangdangpi pame
nang d ng- a-m
1/2:NSUBJ=put.on.stove-BEN CAUS-be.good
‘[…] and so they had nicely prepared and gotten ready (some) tea and food and drink’
[SH, CSM 040]
567
example is (859), where the coordinated neri nekeng ‘my hands and feet’ occur in two
different clauses, both ending in the negative existential copula .
568
(861) Tail-head linkage: identical repetition
anke latum thelo dinglo piso some enlo tangho
nke la-t m th -l d ng-l s s m n-l t ngh
and.then this-PL be.big-RL be.long-RL wife EE: s take-RL REP
'and then, they grew up and got married, so they say' [WR, BCS 003]
The excerpt in (862) starts with a sentence that is much longer and more complex.
For the tail-head linkage construction, the storyteller only repeats a simplified version of
the predicate, i.e., shortening ’ tt ’ t at the “tail” to simply thondamtilo
at the “head”, without including any NPs.
569
ingrengre voso ingrengre along osomar
ingr ng-C v s ingr ng-C al ng os -m r
call(small.animals)-NEG EE:v t k call(small.animals)-NEG LOC child-PL
ponpidam'et thondam'et
n- -d m- t th n-d m- t
take.away-BEN/MAL-go-PRF drop-go-PRF
'and then, after the Karbi woman has gone to the water place, these Hingchong sisters,
over there, she went to carry the children to a place where the roads cross, where the
birds don't sing, and went left them there [CST, HM 014]'
Discourse connectors are elements that typically occur at the beginning of a new
paragraph in narratives, like English ‘and then’. An overview of so far attested forms in
Karbi is offered in Table 118. They are sorted by being based on either ‘this much; all’
(§7.8.2) or the demonstrative (§4.5.3). Apparent combining elements are listed in the
right-hand column. Glosses are not given; they all carry out equivalent functions
translatable into English as ‘and then’.
570
The placement of discourse connectors in narratives either at the end of a
paragraph, the beginning of a new paragraph, or in a neutral position in-between two
paragraphs merits further study. It appears that all types of placement occur, but it is
currently not clear what functions may be associated with the differences. Another
interesting aspect of discourse connector placement also in need of further investigation
is that they occur in second position following the subject in several clauses in the corpus.
(863) and (864) exemplify this construction.
The fact that both instances of post-subject occurrence of discourse connectors are
immediately following direct speech may be a coincidence. There is, however, a
correlation in topic-switch often occurring after the end of direct speech, which is one of
the typical instances where topic-switch is marked by additive =t (§10.7.2), as is the
case, in fact, in (863) in the subject t - =t . I suspect that discourse
connectors are further markers of topic-switch as they occur following a (subject)
argument.
There are two markers that are particularly involved in structuring discourse. I
have glossed them both as ‘discourse structuring marker’ or ‘DSM’, but I describe their
571
distributions and functions in more detail below. They both have unusual phonological
shapes that ally them with interjections: one is e with a long vowel that typically hosts a
salient intonation contour; the other is ’ , which is similar to (but shorter and with a
lower voice than) the English backchanneling ‘mm’ - essentially a glottal stop initial,
voiced /m/. The e is more frequent than the ’ in the corpus.
The e frequently occurs in narratives by different speakers at the beginning of
direct speech, as in (865) and (866).
apiso banghini”
[a-pis b ng-hin ]]
POSS-wife CLF:HUM:PL-two
'”we have a father, though we have a father, he has two wives”' [CST, HM 046]
(867) shows that e also occurs inside direct speech, not only at the beginning.
572
(868) further shows that while e most frequently occurs in direct speech, this is
not exclusively the case. In this example, however, e also occurs at the beginning of an
intonation unit, and specifically at the beginning of a narration (which follows an
introduction by the speaker).
(869) and (870) offer examples of the use of the discourse structuring marker ’ .
In both cases, ’ occurs either right after or right before an important event in the
storyline is told. After uttering ’ , the storytellers pause for effect. In (869), the ’ is
uttered after mentioning of the Karbi woman giving birth to the two children that end up
being the protagonists of the story.
arlo
arl ]] ‘mh
inside DSM
‘after her husband had gone for meetings, the Karbi woman gave birth to children there
inside the house’ [CST, HM 008]
In (870), the storyteller utters ’ and pauses after mentioning one of the
protagonists as topic via =ke, then goes on to say that she passed away. In the course of
the story, her husband will follow her and temporarily succeeds in bringing her back to
the living.
573
(870) Discourse structuring marker mh in middle of sentence, before narrating an
important event
amat aphu kesopen amat bamonpi
am t a-ph ke-s -p n am t bam n-p
and.then POSS-head NMLZ-hurt-NF and.then wise.person(<Ind)-female
General extenders are expressions like English ‘and so on’, ‘etcetera’, ‘and stuff’,
‘and everything’, ‘or anything’, ‘or whatever’, as discussed by Overstreet (1999).
Overstreet shows that the intuitive idea that general extenders in English function to
indicate that there is a larger set involved is typically correct. For example, in ‘apples and
oranges and stuff’, the general extender ‘and stuff’ is indeed usually interpreted to set up
a natural category (e.g., ‘fruit’) and thus include further members of this category.
However, this does not actually explain why general extenders are used. Overstreet
shows that general extenders are often used because they are built on intersubjective
knowledge of what else may belong to a category invoked by general extenders, which
represents a politeness or in-group marking strategy. Furthermore, general extenders are
also often used in particular emphatic constructions, where they may not even imply that
there are further category members. For example, one such construction takes the form X
and everything, but Y (as in, ‘she is nice and everything, but I still don’t like her’). Here,
574
the purpose of the general extender ‘and everything’ is to assert X to build up to the
contrast with Y (Overstreet 1999: 91).
In Overstreet’s study, English general extenders most frequently occur in the
construction “[1 item + general extender],” for example, ‘apples and stuff’, although a
three-parted construction [2 items + general extender] as in ‘apples and oranges and stuff,’
and longer expressions also occur (p.25-7). In Karbi, the construction is always binary:
one noun phrase or verb plus general extender, i.e., ‘[NP/V] [GENEX].’ The general
extender is typically t, but may also be (and dialectal variants of the same word)
for NPs.
The typical general extender t has grammaticalized from t ‘type, thing’ -
which also occurs as an indefinite pronoun, see §4.5.6 - and the possessive/modified a-
prefix.197
In (871), the predicate occurs in the t general extender construction. While the
main verb only takes perfect - t, the general extender t repeats the
perfect suffix and also adds realis - . The general extender here invokes a category of
grooming activities.
In (872), the verb and general extender are marked the same, with - ‘IRR1’.
Here, it is not clear whether the general extender invokes a category of activities that
involve cooking. There could be a category of household chores, but there is no mention
in this text of anything other than cooking. The general extender may then rather serve as
197
Galo (Tibeto-Burman; Tani) also has a general extender construction based on an indefinite marker that
occurs in the negative indefinite construction, although in Galo it is the same marker as the interrogative
pronoun of content ‘what’, (Post 2007: 344 ff.).
575
an intensifier construction, which contrasts the obligation of the mother to cook with the
infant’s crying that prevents her from doing so.
A few intonation units further along in the same text, the speaker picks up again
on the same issue, stating that the mother couldn’t cook because the baby was crying. In
(873), the speaker uses an elaborate expression construction (§12.2.2) instead of the
general extender. Since the same content is communicated in (872) and (873), this
suggests that the functions of general extender and elaborate expression constructions
overlap in Karbi.
danglonglelo […]
d ng-l ng- -l [a-os ke-chir a-join ]
put.on.stove-GET-NEG-RL POSS-child NMLZ-cry POSS-reason(<Ind)
‘that mother couldn't cook at all, because the child was crying […]’ [KK, CC 026]
In (874), the verb is marked imperfective with the ke- prefix and t is unmarked.
The general extender construction here invokes a category of mad or disturbed elephant
behavior, which is clear from the context, which tells us that a bird had just scratched the
elephant’s ears. This example shows then that categories invoked by general extenders
are always context-dependent, which is also what Overstreet has found for English and
has called the ‘intersubjective’ property of general extender constructions.
576
harlong turpur
harl ng t r-p r
stone kick-move.over
'this elephant was roaring and everything and kicked around mindlessly, and kicked over
a rock' [RBT, ChM 025]
The general extender t also occurs with NPs, as in (875). In this instance, it is
not clear whether the general extender implies a category of other items that caused
people getting sick with a cold besides all the dust on the roads (a common cause for
getting sick). The general extender construction may rather serve an intensifying function
here.
Finally, (876) and (877) show that (and its dialectal variants) also functions
as an NP general extender. In (876), this invokes a category of snacks for the road, which
here likely just replaces a tedious enumeration at an irrelevant level of specificity. In
(877), however, saying America heihaipen may be parallel to Overstreet's (1999: 83)
notion of general extenders functioning to indicate “a maximum extreme”, i.e., the least
likely item or fact from a category. In this interpretation of (877), America is presented as
the least likely country to have visitors from, among the category of countries that
visiting foreigners may be from. The general extender would then have a scalar additive
function translatable with ‘even’.
577
(876) General extender construction with ; function: invoking set of snacks
anke thesere haihuita namponlo <kecho> tovar
nke theser h i h i t n m-p n-l <ke-cho> tov r
and.then fruit these~PL=also buy-take.away-RL NMLZ-eat road
578
12.2.2.1. Forms of Elaborate Expressions
In Karbi, EEs always have their head first, followed by the EE complement, as in
ing ‘laugh’, where is the head and is the EE complement. The head
also exists as an independent item, outside the EE construction (with so far only one
exception), i.e., ‘laugh’ is a full lexical verb by itself. Some EE complements also
exist as independent items, although the majority of complements from EEs in the corpus,
are items that only occur as EE complements and do not actually have a meaning by
themselves. I gloss EE complements that are independent items with their existing gloss,
e.g., in the EE t , I gloss as ‘beat with something flexible’ and t as ‘beat
with something solid’. Other EE complements I just gloss as ‘EE:X’, where X is the EE
head. In some cases, the EE complement is homophonous with an independent item that
has no semantic relationship with the EE head, for example in the EE
‘laugh’, is homophonous with the verb ‘to sit’. Since there is no apparent semantic
relationship between laughing and sitting, I have glossed the EE complement as
‘EE:ingn k’.
Both monosyllabic and disyllabic roots participate in EEs. In EEs based on
monosyllabic roots, either a prefix or suffix often occurs on both the head lexeme and the
EE complement, which results in the characteristic four-syllable structure, although two-
syllable structures, as in (878) occur (as well as structures with more than four syllables
in total, e.g., six syllables); EEs are always symmetrical in terms of syllable structure
such that the head lexeme and the EE complement always have the same amount of
syllables.
kechok aphan'iklo
ke-ch k] a-ph n- k-l ]
NMLZ-wash.clothes POSS-PURP-FRML-RL
‘therefore, the suho fruit is for washing clothes’ [SiH, CW 007]
579
Table 119 represents a subset of the 55 EE complements so far attested (that are
not also independent items), which show the range of formal relationships between head
lexeme and EE complement. It begins with monosyllabic and then lists disyllabic (and
two trisyllabic) nouns and verbs, and then offers some sample items from other word
(sub-)classes that also have EE complements. I represent different syllables (between
head and complement) by different letters (‘A’ through ‘E’) and identical syllables by
using the same letter; quasi-reduplication (e.g., using the same syllable but switching a
vowel) is indicated by adding an apostrophe (e.g., A’). The right-most column ‘Meaning
of EE (if changed)’ notes if a particular EE has a consistent semantic effect (for EE
functions, see §12.2.2.3).
In the corpus, the most frequent pattern for monosyllables is A-B, and the most
frequent pattern for disyllables is AB-AC, followed by AB-AB’. The other patterns are
very rare, and identical syllables in AB-BC and AB-CB may, in fact, be coincidence.
Table 119 shows that in both monosyllables and disyllables, there are instances of
quasi-reduplicated EE complements with vowel switch (§3.8.6.2), such as ‘rice’ or
t t ‘banana’. This vowel switch quasi-reduplication is a frequent pattern; other
quasi-reduplication are not frequent and may be coincidental, such as t ‘assembly’,
which I have put in square brackets due to this reason.
In the subset of forms given in Table 119, there is only one instance where the EE
complement represents a borrowing, i.e., t ‘stay’, but there are other forms that
follow the same pattern (of using a borrowing as EE complement), which was also
reported to be the case in Lahu by Matisoff (1973: 83).198
One item, ‘thing’, occurs with two different EE complements in the corpus,
t r and t , although t is more typically used, according to my consultants.
Curiously, the same speaker in the same text uses first t and then t . Generally,
EE heads only occur with one specific EE complement.
There are a few instances, where using an EE instead of just the head by itself has
a consistent semantic effect, which is recorded in the right-most column in Table 119.
Specifically, there is a tendency for noun EEs to convey a collective interpretation.
The verb suffix EE - - is the only instance, in which the ‘head’ is not used
independently outside the EE (which calls into question whether it should really be
considered the ‘head’).
198
One neat example of a modern borrowed EE complement that in fact goes with a borrowed EE head is
t ‘car’, where both mot and are ultimately borrowed from English, though through
Assamese.
581
EEs may be formed with any word (sub-)class, as shown in Table 119. Text
examples (879) and (880) show elaborate expressions based on a relator noun and on a
verb suffix, respectively.
harlong turpur
harl ng t r-p r
stone kick-move.over
'this elephant was roaring and everything and kicked around mindlessly, and kicked over
a rock' [RBT, ChM 025]
The same options are available in forming elaborate expressions off verbs.
Consider the following three examples based on the elaborate expression t
‘cook’. While in (883), the elaborate expression occurs in a parallelism construction such
that the imperative suffix - is repeated on both roots, the roots form a compound stem
in (884). Note that (885) is somewhat of an intermediate type, as the derivational
complex occurs on both roots, but the inflectional imperative suffix - only occurs on
the second root.
583
akopen hem kedo atum aphanta
ak pen [h m ke-d a-t m a-ph n t ]
before=from house NMLZ-stay POSS-PL POSS-NSUBJ=also
tun-dangpikang'et nangjilang
t n-d ng-p -k ng- t n ng-j -l ng
cook-put.on.stove-BEN/MAL-leaving-PRF need-IRR2-still
'early in the morning, over there, the women have to cook and do all the household works
from before dawn also for those who stay at home (besides food for themselves)' [KaR,
SWK 074]
584
(886) Elaborate expressions: nangpran__nangmui; nangsapji__nangthengji
“[…] pinike nangpran damji nangmui damji,
pin ke nang- r n d m-j nang-m i d m-j ]
today=TOP 2:POSS-life go-IRR2 2:POSS-EE: r n( nd) go-IRR2
<nangsapdamji> nangsapji
<nang s p-d m-j > nang s -j
1/2:NSUBJ=beat.w/sth.flexible-go-IRR2 1/2:NSUBJ=beat.w/sth.flexible-IRR2
nangthengji!”
nang th ng-j
1/2:NSUBJ=beat.w/sth.solid-IRR2
'[…] today your life will definitely go, I will beat the hell out of you!” [RBT, ChM 033]
A similar level of intensity can also be argued to underlie (887), where two babies
abandoned in the jungle by a witch are saved by a tiger couple. In (887), the female tiger
says to the male tiger that they should not leave the two crying babies in the jungle, but
that they should take them with them and take care of them. The reason why they should
do that, according to the tigress, is because the babies are monit monor ‘human beings’.
The use of the EE evokes a number of (perhaps, noble) connotations, which is why the
tigress can use it as a reason for why they should go through the trouble of taking care of
them.
orapondetnang”
or -p n-d t-n ng
take.care-take.away-PFV-HORT
‘[…] “because they are human beings, let's carry them away and take care of them”'
[CST, HM 033]
A matter for future research is the relationship between EEs and general extenders.
There is a functional overlap between the two in that both may add an intensifying force
to the utterance (§12.2.1). Note (888), which demonstrates that a predicate can
simultaneously occur in the elaborate expression construction and in the general extender
construction, such as nangcharkok nangcharpheklo ajatlo here.
585
(888) Elaborate expression and general extender used in combination
t t
[ n nsi a-h m a-r t chV-v ng-l ] [[a-h m a-r t
that.much after.that POSS-house POSS-field RR-come-RL POSS-house POSS-field
There are a range of copy verb constructions, which consist of a preposed copy of
the verb root that is marked with one of the information structure clitics (see §10.7) or the
question clitic (see §11.1.1.2.2), followed by the actual main verb, which may carry
derivational and inflectional morphology. Their functions have to do with adding force to
the predicate, which we can consider iconic given the phonological force created through
repetition of the verb.199
In addition to the four copy verb constructions presented in what follows below,
there is another copy verb construction that has the function of coordinating clauses, as
discussed in §11.5.1.2.
The assertive copy verb construction consists of the root of the main verb marked
with =ke ‘topic’, followed by the full main verb with additional suffixes if applicable,
schematically ‘V=ke V-suffixes’. The function of this construction is to strongly assert a
199
Interestingly, a similar type of copy verb construction is found in Kurtoep (Tibeto-Burman; East Bodish)
(Hyslop 2011: 680-3).
586
proposition. As a result, the construction is often employed to set up a contrast, as in X is
indeed the case, but Y, which is evidenced by the fact that this construction is frequently
followed by a concessive conjunction.200 An example is (889), where the assertive copy
verb construction occurs in the clause epoke doke do ‘we do have a father’, which sets up
a contrast to ‘he has two wives’ as marked by the concessive do setame - a contrast
presumably in the sense that although there is one father, the maternal side is unusual,
because there are two women.
(889) Assertive =ke copy verb construction followed by contrast (do setame)
e epo do, epoke doke do, do setame,
e [e-p d ] [e-p ke d e d [[d setam ]
DSM 1INCL-father exist 1INCL-father=TOP exist=TOP exist exist nevertheless
apiso banghini
[a-pis b ng-hin ]]
POSS-wife CLF:HUM:PL-two
'we have a father, though we do have a father, he has two wives' [CST, HM 046]
Similarly, in (890), there is a contrast built up by kechetong doke dohe ‘(I) have
indeed met (some of them)’, which follows in the next intonation, offered in (891), which
unsurprisingly starts with t ‘but’.
(890) Assertive =ke copy verb construction followed by contrast in next IU (seta)
Naka anglong pu bihek akopenta arjulong
N k a-ingl ng pu b -h k ak pen t arj -l ng
TRIBE POSS-hill QUOT be.small-small then=from=ADD:even hear-GET
200
While English does not have a dedicated marker for this kind of strong assertion in anticipation of a
contrast, German has zwar, which always has to be followed by aber ‘but’. The particle zwar has the telling
etymology es ist wahr ‘it is true’.
587
pusetame elong longni lason kechetong doke
p set m e-l ng l ng-n las n ke-chet ng d e
likewise one-CLF:place CLF:place-two that.way NMLZ-meet exist=TOP
dohe
d =he
exist=AFTERTHOUGHT
‘since my childhood, I got to hear about the Naga hills, the Naga people, from both the
women and men, in one or two places, like that, I actually have met (some of them)’
[SiT, HF 004]
This assertive copy verb construction does not necessarily occur in a larger
contrast construction. In (892), there is no contrast involved. The speaker states that she
saw a particular ceremony performed in a different region that has somewhat different
customs. She says she saw it for the first time and that she liked it. Then she says prekke
prekchot ‘it really is slightly different’, referring to the ceremonial customs in that region.
Apparently, the assertive copy verb construction is used here because the speaker just
stated that she saw it for the first time and so it was perhaps a surprise to find out that the
ceremony is indeed performed in a slightly different way in that part of Karbi Anglong.
588
chinidun'o'e setame ning arongpiklo
chin -d n- -C set m n ng ar ng-p k-l
know-JOIN-much-NEG nevertheless mind be.happy-very-RL
The intensifier declarative copy verb construction consists of the root of the main
verb marked with =t ‘additive’, followed by the full main verb with additional suffixes
if applicable, schematically ‘V= t V-suffixes’. The function of this construction is
intensification, which is evidenced by the fact that the main verb typically carries
quantifying and intensifying suffixes, such as - ‘(too) much’, - ‘much, very’, or -
‘very’.
In (893), the context is that one of the tigers tells the other tigers that they need to
be quick in performing a ceremony and explains why that is of tremendous importance.
(893) occurs after the direct speech and states that the tigers did indeed do everything
very quickly.
In (894), a grandmother talks to her infant grandchild and says that the child has
cried too much, using the intensifier construction.
(894) Intensifier copy verb construction, main verb with - ‘(too) much’
chiruta chiru'ongchotlo nang piba chevanra
chir t chir - ng-ch t-l n ng pib che-v n-ra
cry=ADD:INT cry-too.much-a.bit-RL 2 baby.carrying.cloth RR-bring-NF:IRR
589
hong vannoi ho
h ng v n-n i ho
outside bring-INFRML.COND.IMP definitely
‘(you) have cried too much, bring your piba and bring it outside’ [KK, CC 022]
In (895), an evil stepmother mistreats her stepchildren, and so the storyteller says
this sentence, using the intensifier construction to say just how evil the stepmother was.
In (896), the speaker talks about travel by car to a place relatively far away. The
first portion is the way up to Bokolia, as mentioned in this sentence. She says that
because the road is good, they reached Bokolia really early, employing the intensifier
construction.
The available data on the =le ‘focus:irrealis’ construction suggest that it also has
an intensifier function, like the copy verb construction with =t (§12.2.3.2), but that it
occurs specifically with non-declarative speech acts.
In (897), the construction is used to intensify an imperative ‘follow me!’, which
has further emphatic marking through use of the interactive emphatic particle ho
(§12.3.6).
590
(897) Intensifier copy verb construction with =le, imperative main verb
“ t t
n nang-p n-th k-p t ph
EXCLAM 1EXCL 1/2:NSUBJ-take.away-know.how-IRR1 EMPH grandfather:VOC
(898) [ - - ] [ ’ - - ] [ -t = -t ]
left RR-see-NEG right RR-see-NEG NMLZ-write=Q NMLZ-write
‘without looking left or right, s/he keeps writing’ [Elicitation SiT 090228]
In Grüßner’s examples (899) and (900), no context is offered, but his translations
suggest that the construction indicates durativity.
(899) = -
look.for.lice=Q look.for.lice-RL
‘she was looking for lice for awhile’ (Grüßner 1978: 130)
(900) - = - -
RR-complain=Q RR-complain-RL
‘they were complaining for awhile’ (Grüßner 1978: 130)
Lastly, Grüßner’s example (901) shows that the second verb may be marked by
additive =t , i.e., ‘V=ma V=t ’. Unsurprisingly, the additive marking results in a
591
concessive meaning that combines with the durative function to then translate as ‘despite
persevering in V-ing, (something else happened)’.
(901) [ = =t ] - - t-
search=Q search=ADD meat get-NEG-PFV-RL
‘however much he was searching (for it), he didn’t get the meat’ (Grüßner
1978: 129)
There are two constructions that employ the negative equational copula ; the
disagreement construction, which does not require a nominalized verb; and the intensifier
construction, which does require nominalization of the main verb.
kanghonso kali
ke-ingh n-s al
NMLZ-love-INTENS NEG.EQU.COP
‘the dog loved the bamonpo and bamonpi so much!’ [KK, BMS 004]
In (905), the construction occurs with the PCT root t . It is further combined
with an elaborate expression construction (§12.2.2) based on the - suffix element of
this construction.
593
(905) Nominalization-based intensifier construction with t ‘be big’ (involving
elaborate expression construction)
[…] t t t
l ng t ke-th -s ke-th -v r al ] [langs
water=ADD:even NMLZ-be.big-INTENS NMLZ-be.big-EE:-s NEG.EQU.COP this
atalo bhari
a-tal bhar ]
POSS-sea very.big(<Ind)
‘[…] there was an unbelievably big water body, a huge sea’ [KK, BMS 044]
201
The audio file for KK, CC 012 is available under the DOI name 10.7264/N3KD1W5R, see Appendix B.
594
12.3. Particles
12.3.1. Quotative pu
Quotative pu, grammaticalized from ‘say’, follows after, and indicates the end
of, direct speech, as in the reported exchange in the intonation unit in (906).
595
(906) Quotative pu marking the end of direct speech
“ t t ”
[[[chin -C -d t-l p ] [t m ne chen k-var t-j ma p ]
know-NEG-PFV-RL father if future 1EXCL:NSUBJ=torture-INTENS-IRR2=Q father
pu, “ t “ pu
pu] [[nang chen k-C p i a-t m] pu]
QUOT 1/2:NSUBJ=torture-NEG mother POSS-PL QUOT
'"I don't know, father, if (you take a new wife), will she keep torturing us, father"', '"she
won't torture you, mothers"' [CST, RO 010]
angjirta do pu
a-ingj r t d pu
POSS-sister=ADD:also exist QUOT
'and then, Bey Ronghang, the youngest, he also got married and his children grew up, and
then, the oldest one, that Bey the Fair, Bey Ronghang, and Bey the Black, they also had a
sister, it is said' [WR, BCS 007]
596
ki'ikrintile putangho
ke- k-rint le u-t ngh
NMLZ-be.black-equally:PL:S/A=FOC:IRR QUOT-REP
'the family of your oldest maternal uncle, the family of your uncle who is the eldest son
of the family, all of them are so black (so you shouldn't take your wife from them)' [WR,
BCS 014]
597
12.3.2. Reportative t
598
The disyllabic form of t suggests that diachronically there are two
morphemes involved. In fact, there is one instance in the corpus where apparently the
first syllable is used independently of the second syllable. In (913), -t is used as a
verbal suffix in a predicate that is marked as a question via =ma. Apparently, -t here
has the same function of indicating that the requested information is not directly known
by the addressee, but is information that the addressee only knows if at all as reported
information.
(913) Reportative t
the'o'otangma atibukta} mm
th - ~ -t ng=ma a-tib k t mm
be.big-much~DIST.PL-REP=Q POSS-earthen.pot=ADD:DM AFF
{are the earthen pots big enough?} [HK, TR 177]
(914) - - -t = c - =ma?
POSS-skin NMLZ-be.white POSS-PL=TOP rice eat-REP=Q
‘are white people said to eat rice?’
(915) - - t-
fall-NEG-PFV-IRR2 DUBIT
'it probably won't rain' [SiT 090221]
nangthanpo
nang th n-p ]
1/2:NSUBJ=tell-IRR1
‘the story I'm telling now, maybe I can't tell it perfectly (straight), but maybe I will still
tell’ [KK, CC 008]
arju'iklong tahailo
arj - k-l ng tah i-l
hear-FRML-GET DUBIT-RL
'you probably have heard the different matters about the old days' [KaR, SWK 059]
600
12.3.4. Always t t
(918) ‘Always’ t t
so'arlo atumsi keklem abang dopo
[s 'arl a-t m si [ke-kl m a-b ng] d -p ]
women:COLL POSS-PL=FOC NMLZ-do POSS-CLF:HUM:PL exist-IRR1
lason arjulonghe
[las n arj -l ng he]
that.way hear-GET=AFTERTHOUGHT
‘[…] the women would be the working ones, the men would always play with the
children instead, this is actually what I've heard’ [KaR, SWK 071]
In (919), t t surprisingly occurs with what may be -si ‘NF:RL’, or possibly =si
‘focus’. It is currently not clear how to analyze titisi here, but this demonstrates that t t
belongs to the subset of particles that may occur with morphological marking (§12.3).
(Also note that there is no doubt that titisi is indeed t t plus a /si/ element, because the
habitual function is additionally indicated by the following, reduplicated main verb
nangkechodundun.)
nangkechodundun
nang=ke-ch -d n~d n
CIS=NMLZ-eat-JOIN~HAB
‘to us you would also give us, on the ground always you would serve us (food), I also
used to eat like that’ [KK, BMS 060]
601
The etymology of t t could be a reduplication of emphatic ti (see §12.3.5), which
would make sense given that cross-linguistically, habituality often correlates with
reduplicative marking.
12.3.5. Emphatic ti
The emphatic particle ti occurs at the end of a sentence and can be translated into
English via lexical emphatics such as ‘really’ or ‘definitely’, or the do-emphatic
construction. It occurs in positive and negative declarative clauses, as well as in non-
declarative clauses, such as imperatives and hortatives.
In (920), the first line represents a question that expresses the lack of confidence
of the person asking (i.e., the owner of the dog, or ‘grandfather’) that the dog might
actually be able to take him (across a huge body of water). The question starts with
ahokma? ‘is it true?’, and then asks neponthekjima? ‘will you be able to take me?’. In his
reply, the dog tries to reassure his owner that he will indeed be able to take him across the
water, and he does that by adding ti. He then also uses a copy verb construction,
nangdunle nangdunnoi, which is another type of emphatic construction (see §12.2.3.3).
602
(921) Emphatic ti with declarative, positive verb (with ho ‘EMPH:INTERACT’)
o.... neta lasi ho kiribom ti ho
o.... n t las h ke-r -b m t h
AFF 1EXCL=ADD thus EMPH:INTERACT NMLZ-search-CONT EMPH EMPH:INTERACT
‘oh, I see. I also am searching for somebody, just like you’ [HK, TR 065]
603
chothatnang” pulo amat hala apiso abang pulo
cho-th t-n ng] p -l ] [am t h la a-pis ab ng p -l
AUTO.BEN/MAL-slaughter-HORT say-RL and.then that POSS-wife NPDL say-RL
“thatnangne ti sarbura”
th t-n ng- t s rbur ]]
slaughter-HORT-NEG EMPH old.man
‘ “[…] let's kill us a hen tonight”, he said, (but) the wife said, “let's not, old man (and I
won't change my mind)”' [SeT, MTN 007]
(926) and (927) offer examples of ho occurring together with the feedback or
back-channel requesting (§11.1.1.8).
{mm} dei ho
mm d i h
AFF OK? EMPH:INTERACT
‘in the bag, in the old days, they used to keep a so-called luji, a mirror (container), you
know..., right?’ [HK, TR 017]
In (928), ho occurs along with the ‘common ground’ marker mati. This shows that
ho may emphasize information that the speaker expects the listener to know. The use of
ho here implies that the tobacco container will play a key role in the story, so the
storyteller wants to make sure the listener pays attention to the mentioning of the tobacco
container.
In (929), ho marks information that the speaker knows will be surprising to the
listener and may be taken with disbelief. In anticipation of that, ho emphatically asserts
the information.
12.3.7. Vocative
The vocative particle is used before the name or kinship term when calling for
somebody. For example, somebody with the name ‘Rasinza’ is usually called by saying
605
Rasinza!, while the mother is usually called by saying , where is the
lexical noun for ‘mother’.
The irrealis focus marker =le (see §10.7.4) may occur at the end of a sentence to
convey exclamative force, as in (930) and (931).
In both (930) and (931), the exclamation is a result of surprise over a novel insight.
This is probably the more common cause for exclamations. However, the surprise
element is not necessary for the use of sentence-final =le, as shown in (932). The fact that
the members of the uncle’s family all have a dark skin complexion is not something the
speaker just found out. The exclamation here serves the purpose of adding force to the
argument that the addressee should not marry that uncle’s daughter.
606
(932) Irrealis focus =le with exclamative function (not expressing a surprise)
nangong akleng ahemke nangong
[nang- ng a-kl ng a-h m ke] [nang- ng
2:POSS-maternal.uncle POSS-old.one POSS-house=TOP 2:POSS-maternal.uncle
ki'ikrintile putangho
ke- k-rint =le p -t ngh
NMLZ-be.black-equally:PL:S/A=FOC:IRR QUOT-REP
'the family of your oldest maternal uncle, the family of your uncle who is the eldest son
of the family, all of them are so black (so you shouldn't take your wife from them)' [WR,
BCS 014]
607
hala asupo aphanlohe, asopi
h la a-su- a h n-l =he] a-oso-
that POSS-grandchild-male NSUBJ-RL= AFTERTHOUGHT POSS-child-female
The second instance of =he in (933) is in hala asupo aphanlohe ‘(to) the
grandson’, which adds or clarifies the O argument of the previous clause anlo aphi abang
kipu tangho ‘the grandmother said’. The third instance follows right after the second
instance, where the storyteller clarifies which grandson it is when she says asopi aso
aphanlohe ‘(to) her daughter’s son’.
While =he more frequently occurs on NPs in the corpus, there are some instances
where it occurs on full clauses that are added as an afterthought to the previous utterance.
An example is (934), where the speaker makes a strong statement by adding t t ‘always’
at the end. She then apparently decides that she does not want to take responsibility for
this strong claim, and adds lason arjulonghe ‘that is (anyway) what I’ve heard’.
608
lason arjulonghe
las n arj -l ng he
that.way hear-GET= AFTERTHOUGHT
‘[…] the women would be the working ones, the men would always play with the
children instead, this is what I've heard’ [KaR, SWK 071]
Particle =he further occurs more broadly than just in afterthought contexts in the
folk story WR, BCS (see Appendix E). In this text, =he appears to functionally overlap
with (§12.3.12), which in turn is best analyzed as a combination of =he and the tag
question marker (§11.1.1.7). An example of =he functioning as a marker of narrative
style analogous to is (935).
piso hangdamlohe
p s h ng-d m-l =he
wife call-GO-RL=EMPH
‘and then, therefore, among the two brothers, their sister went to the so-called ‘Bey the
Black’, to his house, to ask for a wife, you know’ [WR, BCS 011]
Clauses marked with =ke ‘topic’ provide a background against which a statement
just mentioned (or about to be mentioned) is meant to be understood. Typically, the
semantic link is one of causality, with the =ke marked main clause commonly providing
a reason. This construction is discussed as an instance of insubordination in §11.4.1.
Two examples, in which a =ke marked clause follows a main clause and is
semantically linked to it via causality, are (936) and (937).
(936) ‘Topic’ =ke marked clause following its contextually related clause
- … - =
be.cold-a.bit rain NMLZ-fall=TOP
'It's a bit cold… it's raining, that's why' [OH 121011: 001]
609
(937) ‘Topic’ =ke marked clause following its contextually related clause
- t ng- - = ,
rain fall-still be.dark-much-RL=TOP look!
'It's still raining, (so) it's very dark, look' [AT 121011: 002]
In (938), the first clause apok ingchirdukke is in its own intonation unit and is thus
prosodically marked as a main clause. However, since it is marked with =ke, it signals to
the listener that this is background information providing context for an event that is
more relevant to the storyline.
(938) ‘Topic’ =ke marked clause preceding its contextually related clause
apok ingchirdukke…
a-p k ingch r-d k=ke
POSS-stomach be.hungry-INTENS=TOP
In the excerpt in (939) and in the following intonation unit from the same text in
(940), =ke marked main clauses are used as answers to ‘why’-questions. It appears that
using the background information construction rather than a direct ‘because’-clause is a
more indirect way of answering. This makes sense in the context here, because both the
frog and ant are accused by an enraged king of ultimately being responsible for the death
of his daughter.
202
The audio file for HK, TR 014-5 is available under the DOI name 10.7264/N33B5XDS, see Appendix B.
610
adon chonraimati?" pu
a-d n ch n-r i mati] pu
POSS-bridge/ladder jump-solid.obj.breaking=as.you.know QUOT
'he asked the frog, "Why did you jump on the ladder of the squirrel?"' [RBT, ChM 074]
Finally, in excerpt (941) from an interview, the interviewee first speaks and says
that she is a Terang family member. The interviewer takes his turn and essentially repeats
the information he was just given, using tag questions, which occur after main clauses
marked by topic =ke.
611
Here, the tag questions are clearly not intended to elicit confirmation of the truth
of the proposition, as they represent a repetition of already provided information,203
which is marked by clause-final=ke ‘topic’.
t mati
ch k-C -d t-l =mati
be.fine-NEG-PFV-RL=CG
‘[…] and the reason why Bey the Fair and Bey the Black can't tolerate each other is that
<that wife…>, that grandmother made that mistake’ [WR, BCS 019]
(943) tells the reason why the orphan protagonist of a story is leaving the village,
which is because he is hungry and so needs to go look for vegetables. =mati here
indicates that the orphan being hungry is not new information: because he is an orphan,
of course he will be hungry because he doesn’t have a family to provide for him. The use
of the question tag suggests that the storyteller is making sure that the orphan being
203
In fact, the video of this interview shows that the interviewer is not even looking to the interviewee as he
is uttering the tag questions, but instead is looking down to his notes.
612
hungry can indeed be considered common ground, and a logical reason for the orphan to
leave the village.
In (944), =mati occurs in a question, which at first glance poses a problem for the
interpretation that =mati marks common ground. It appears, however, that =mati here
marks the common ground, or presupposition, that the hanthar fruit fell on the back of the
pig, and the information sought by the question is only why it did that.
613
luji pu hedi {luji do} luji do {mm}
luj pu hed luj d luj d mm
mirror QUOT Q.TAG mirror exist mirror exist AFF
‘[…] it had an aina on top; it had what we call a mirror, right? - {it had a mirror?} - it had
a mirror’ [HK, TR 019-20]
In the western Karbi Anglong Amri dialect, is highly frequent in folk stories,
where it occurs after topical noun phrases, as in (947). In that use, it apparently also
functions as a feedback requesting device, but with less force. In this dialect, it has
grammaticalized as a marker of narrative style.204
Note that just =he functions quite similarly to as a marker of narrative style
in this dialect (§12.3.9).
204
When I recorded the folk story that example (947) is taken from, I was staying in Umswai in West Karbi
Anglong. After the recording was finished, then 10-year old Platinum Hanse suggested I record him telling
a folk story as well, which we did. Although the folk story that (947) is taken from struck me as abounding
with , in Platinum’s narration, he probably used twice as much. He clearly knew that was a
marker of narrative style in the local dialect, and so made sure to show that he knew it and that he was
telling a folk story the way folk stories are supposed to sound.
614
12.4. Honorific and Formality Marking
There are three markers for a honorific or formal speech style. First, there is a
suffix - , which occurs on pronouns and addressing words (§12.4.1). Second, there is a
verbal suffix - (§12.4.2). And lastly, there is an honorific suffix - - , which
occurs on kinship terms (§12.4.3).
615
me'iksenji
m - -s n-j
be.good-FRML-INTENS-IRR2
‘so first, if you could tell us your name, that would be wonderful’ [KaR, SWK 004]
kechok aphan'iklo
ke-ch k] a-ph n- -l ]
NMLZ-wash.clothes POSS-PURP-FRML-RL
‘therefore, the suho fruit is for washing clothes’ [SiH, CW 007]
In the corpus of recorded texts, - does not occur in folk stories, but it does occur
in procedural texts. There is one exception, where - does, however, occur in a folk story
in the corpus, which is provided in (950); this intonation unit represents the last one of
this recording.
616
’ ’helo Rongphar asangho kethekthe anke
lap - k-hel Rongph r a-sangh ke-th k-C nke
like.this-FRML-RL:EMPH CLAN POSS-mister:VOC NMLZ-know.how-NEG and.then
nangpekengpon'iknoi
nang pa-k ng-p n- -n i
1/2.NSUBJ=CAUS-be.straight-take.away-FRML-INFRML.COND.IMP
‘thus, they settled down and lived together until the end of their lives, and then that was
the story of Hingchong musoso, Mister Rongphar, I'm not an expert, so make it clear (i.e.
correct it)’ [CST, HM 120]
What the use of - here suggests is that the storyteller says the last sentence of the
story not as part of actually telling the story, but as a statement about the story, i.e., on a
meta-level.
12.5. Interjections
617
Table 120. Interjections
Domain Form Gloss Use
Replies to ‘yes’ reply to polar interrogatives, expressing
questions ‘NEG.EQU.COP’ (dis-)agreement205
‘ ’ [ʔmmʔ] ‘no’
‘yes’
t ‘yes, okay, I see’
o ‘AFF’ backchanneling
reply when called by name
Directives ‘(let’s.)go!’ typically a hortative, literally meaning
‘let’s go!’; but also used when directed
to addressee only, as ‘go!’
‘leave.me!’ when speaker is held by somebody
‘look.here!’ directing addressee to look at something
right in front
‘look.there!’ directing addressee to look at something
further away
how ‘here.you.go!’ when handing something to addressee
Surprise, ‘surprise’ expressing surprise
fear, shock, ‘how.bad!’ expressing desperation, disapproval, fear
etc. ‘how.bad!’
‘EXCLAM’ ‘exclamation’ (only used by one
particular speaker, at the end of
exclamatory sentences)
‘EXCLAM(<Asm)’ ‘exclamation’, used by many speakers,
after the constituent that is stressed or
emphasized
‘really?’, ‘is it reaction to (sometimes only mildly)
so?’ surprising information
(951) ‘exclamation(<Asm)’
anke lapusi ekdom nangkangthuloklo
nke l p s kd m nang kangth -l k-l
and.then like.this=FOC EXCM(<Asm) CIS=bounce-only-RL
and then, like this we were constantly bouncing up and down [SH, CSM 019]
205
Note that a very typical way to reply to polar interrogatives is to repeat the verb (while, as the case may
be, adding negation).
618
Generally, interjections do not occur embedded into a clause. In (952), however,
‘no’ and ‘yes’ are embedded. Here, kali kalang kejekthek is an idiomatic
expression, ‘know what to do and what not to do’.
apotsi
ap t si
because=FOC
‘because they know what to do and what not to do’ [SiH, CW 019]
This section offers description and examples of how hesitation and correction
words are used in Karbi.
619
anke kene aphi sarpi abang kipu tangho
[ nke en a-ph sarp ab ng ke-p t ngh ]
and.then HESIT POSS-grandmother old.woman NPDL NMLZ-say REP
‘[…] the eldest brother, because he was called Bey the Black... and then, um, he
grandmother, the old woman said’ [WR, BCS 012]
(954) and (955) show that may also function as a “pro-lexeme” similar to
what Post (2007: 678) has described for the Galo hesitation word mə . This means that
can substitute for any noun or verb. This is evidenced by the fact that in (954),
takes the the (nominal) focus clitic =si, and in (955), it takes the (verbal) hortative suffix
- .
along
al ng l -C ak ab ng ke...]
LOC reach-NEG before NPDL=TOP
‘on a very steep slope, there is a place for them to park cars, and then, there..., the place
where we reached first, I mean, we saw... how was it, before we reached the festival
grounds...’ [SiT, HF 026]
620
pusi pu tangho tharve athe endamnang
pusi p t ngh tharv a-th n-d m-n ng
QUOT.COMP say REP mango POSS-fruit take-GO-HORT
'and then, "today, younger brother, let's, let's go and get mangos, the mangos are very
ripe", (the older brother) said, "let's go and get mangos"' [KTa, TCS 041]
Karbi has borrowed the Assamese hesitation word mane. An example of mane
used in Karbi is (956). Here the speaker apparently is trying to remember the word for
‘hill field’ while saying mane, then just uses a descriptive possessive construction, then
says kopunelo ‘what do you call it?’, but still does not remember and leaves it at that.
t
ko-pu-n -l l ingl ng a-r t p n-p
WH-QUOT-INDEF-RL this hill POSS-field clear.vegetation-IRR1
‘I mean... up there on the fields on the hill, what do you call it?, we clear the vegetation
from the hill fields’ [SiH, KH 002]
622
(959) Corrective (<Assamese) after unintentional use of an Assamese borrowing
aphrang ahut... inut arong agaonbura
a-phr ng ah t e-n t a-r ng a-g onbur
POSS-front during one-CLF:HUM:SG POSS-village POSS-village.headman(<Asm)
ba arong asarthelo
b a-r ng a-sarth -l
I.mean(<Asm) POSS-village POSS-village.headman-RL
a long time ago, there was one village Gaonbura, I mean, one village headman [CST, RO
003]
623
APPENDIX A
13. ABBREVIATIONS
625
TOP topic
TS topic-switch
UNCOND unconditional
VOC vocative
626
APPENDIX B
There are three types of supplementary audio files that are available:
1) particular text examples, whose phonological/prosodic properties are discussed in
this dissertation
2) the five entire texts from Appendices C through G: HK_TR; RBT_ChM;
WR_BCS; SiT_PS; and, SiH_KH
3) sample phonetic files of roots that illustrate tone minimal pairs.
627
APPENDIX C
This is a traditional Karbi folk story about an orphan who leaves the village to
look for vegetables. He is surprised by a tiger who wants to take him along so the orphan
can be sacrificed the next day during a Rongker celebration. The orphan, however,
outsmarts the tiger. He has a tobacco container with a mirror on top that he shows to the
tiger as proof that the orphan has been going around catching tigers. The tiger gets scared
and runs off, while the orphan returns to his village and tells everything to his friend. The
two then go to the Rongker celebration of the tigers on the next day, and get to take
earthen pots filled with money from the tigers.
This story was narrated by Harsing Kro, while Sikari Tisso was listening and
occasionally commenting on the story or asking questions for clarification, as well as
doing a lot of back-channeling with mm. Speech by Sikari Tisso is indicated in the text
through {curly brackets}.
The audio file for the entire text HK, TR is available under the DOI name
10.7264/N32B8W8D at https://scholarsbank.uoregon.edu/xmlui/handle/1794/13657
{enutvetlo?}
e-n t-v t-l
one-CLF:HUM:SG-only-RL
‘{he was alone?}’ [HK, TR 005]
enutvetlo {to}
e-n t-v t-l t
one-CLF:HUM:SG-only-RL OK
‘he was alone, {OK}’ [HK, TR 006]
629
{dukpik} dukpik
d k-p k d k-p k
be.poor(<Asm)-very be.poor(<Asm)-very
‘{he was very poor?!} he was very poor’ [HK, TR 008]
{jangreso ma jangrengsolo?}
jangr s ma jangr ngs -l
single.parent.child Q orphan-RL
‘{was it an orphan with one parent left or no parents at all?}’ [HK, TR 011]
jangrengsolo
jangr ngs -l
orphan-RL
‘it was an orphan with no parents left’ [HK, TR 012]
jangrengso {mm} o
jangr ngs mm o
orphan AFF AFF
‘an orphan with no parents left? {Mm} Oh.' [HK, TR 013]
630
apok ingchirdukke {mm}
a-p k ingch r-d k ke mm
POSS-stomach be.hungry-INTENS=TOP AFF
‘so he was suffering from hunger, (and then...)’ [HK, TR 014]
{mm} dei ho
mm d i h
AFF OK? EMPH:INTERACT
‘in the bag, in the old days, they used to keep a so-called luji, a mirror (container), you
know..., right?’ [HK, TR 017]
631
{thongkup} thongkup ason {mm} athakke
thongk p thongk p as n mm ath k ke
tobacco.container tobacco.container like AFF on.top=TOP
aina do {mm}
ain d mm
mirror(<Ind) exist AFF
‘the tobacco container was of such a kind that it had a mirror on top’ [HK, TR 019]
632
anke hala duma kemong athongkupta
nke h la dum ke-m ng a-thongk p-t
and.then that tobacco NMLZ-smoke POSS-tobacco.container-ADD:also
dolo {mm}
d -l mm
exist-RL AFF
‘and then, he also had that tobacco container so he could smoke later’ [HK, TR 025]
{lujisi kapabon} mm
luj si ke-pa-b n mm
mirror=FOC:RL NMLZ-CAUS-be.attached AFF
‘and then, on this tobacco container, there was a mirror attached’ [HK, TR 026]
633
hu arni kangsam apor abangke {mm}
h arn ke-ings m a-p r ab ng ke {mm}
then.later sun NMLZ-be.cold POSS-time NPDL=TOP AFF
vanglo {mm}
v ng-l mm
come-RL AFF
‘he was the biggest and the most powerful one (so they say) {jeez!}... he came’ [HK, TR
033]
634
hako arnike teke <pu> atumta Rongker pu do tangho
hak arn ke tek <pu> a-t m t R ngk r pu d t ngh
that.time day=TOP tiger QUOT POSS-PL=also FESTIVAL QUOT exist REP
‘at that time (i.e. in the old days), the tigers also (like humans/Karbis) celebrated the
Rongker’ [HK, TR 035]
635
{laanke long'etlo} mm
l - n ke l ng- t-l mm
this-that.much=TOP get-all:S/O-RL AFF
‘{he had already collected everything?!} Mm.' [HK, TR 040]
636
akhei ponbomlo akaprek
a-kh i p n-b m-l a-ke-pr k
POSS-community take.away-CONT-RL POSS-NMLZ-be.different
{mm} berdamphlut
{mm} b r-d m-phl t
AFF press.down-GO-miss/fail
‘from behind he was very secretly approaching and tried to jump on him, but failed’ [HK,
TR 052]
638
chelangrailo hala jangrengso abangke {mm}
che-l ng-r i-l h la jangr ngs ab ng ke mm
RR-see-sideways-RL that orphan NPDL=TOP AFF
639
“pap kedo keda nangphanke
p p ke-d ke-d nang-ph n ke
fault NMLZ-exist NMLZ-EE 2-NSUBJ=TOP
nangkeponpo” {“nangponpo”}
nang=ke-p n-p nang p n-p
1/2:NSUBJ=NMLZ-take.away-IRR1 1/2:NSUBJ=take.away-IRR1
‘ “there is no sin or anything, but I'll carry you away” ’ [HK, TR 059]
“ t R ” {mm}
e-t m pen p lo R ngk r-p mm
1PL.INCL-PL tomorrow=FOC FESTIVAL-IRR1 AFF
‘ “tomorrow we will have to/definitely will celebrate the Rongker” ‘ [HK, TR 061]
ejonnatlo” {mm}
e-j n-n t-l mm
one-CLF:animal-only-RL AFF
‘ “this Rongker festival.. our number (i.e. the number of people we have) is not OK, it's
just one (that is missing)” ’ [HK, TR 062]
640
jangrengso abangta repik tangho {mm}
jangr ngs ab ng t r -p k t ngh mm
orphan NPDL=ADD:DM be.smart-very REP AFF
‘(but) the orphan is very smart’ [HK, TR 064]
{angtan along...}
a-ingt n al ng...
POSS-outside LOC
‘{outside....}’ [HK, TR 069]
641
ingtong ardikkluilo
ingt ng ard k-kl i-l
big.bamboo.basket.roughly.woven be.heavy-quite-RL
cheponpedetlo he {mm}
che-p n-C -d t-l he mm
RR-take.away-NEG-PFV-RL you.know AFF
‘the bamboo basket was (/had become) quite heavy, so he hadn't carried it with him’
[HK, TR 070]
bikangkoklo {mm}
b -k ng-k k-l mm
put-leave-in.a.fixed.place-RL AFF
‘the bag.. he kept it away}, he kept it away’ [HK, TR 071]
642
mongponbom nangkokjike
m ng-p n-b m n ng-k k-j ke
smoke-in.passing-CONT need-absolutely.required-IRR2=TOP
‘it's necessary to keep smoking (while digging yam)’ [HK, TR 074]
“ t t t t ” {mm}
n t las n thakth k t h mm
1EXCL=ADD:also that.way same EMPH EMPH:INTERACT AFF
‘ “same here (in trying to find animals for sacrifice)!” ‘ [HK, TR 076]
“ t t duma kemong
muth j n n t l k- ng-si dum ke-m ng
but(<Asm) see! now=ADD be.tired-too.much-NF:RL tobacco NMLZ-smoke
“ pulo ho....
[ nke ar ke-r -d m-l -p p -l h
and.then and/again(<Asm) NMLZ-search-go-again-IRR1 say-RL EMPH:INTERACT
643
“non ” putangho {mm}
n n e-j n d -l ng p -t ngh mm
now one-CLF:animal exist-yet QUOT-REP AFF
{“nangphantong nangkenepchektong
n ng-ph n-t ng nang=ke-n p-ch k-t ng
you-NSUBJ-instead.of 1/2:NSUBJ=NMLZ-catch-firmly-instead.of
kemepoma ”}
ke-m -p ma
NMLZ-be.good-IRR1=Q
‘ “is it good to just catch you (i.e. instead of keeping looking around more for another
one)?”}’ [HK, TR 080]
644
teketa manghutailo tangho{manghutailo}
tek t mangh -t i-l t ngh mangh -t i-l
tiger=ADD:DM get.surprised-for.a.moment-RL REP get.surprised-for.a.moment-RL
{lang nangkokjike}
l ng n ng-k k-j ke
see need-absolutely.required-IRR2=TOP
‘{(we) must see! (if there is a tiger there)}’ [HK, TR 086]
645
“kali nang lam ningjebomte, neke
kal n ng l m ningj -b m-te n ke
NEG.EQU.COP 2 word speak-CONT-if 1EXCL=TOP
nangkebengpo ” {“nangkebengpo”}
nang=ke-b ng-p {nang=ke-b ng-p
1/2:NSUBJ=NMLZ-lock-IRR1 1/2:NSUBJ=NMLZ-lock-IRR1
‘ “on the other hand, if you keep speaking like this, I will lock you in” “will lock you
in”} ’ [HK, TR 089]
646
thongkup ulaiphlutlo
thongk p ul i-phl t-l
tobacco.container take.out(<Asm)-suddenly.big.A/O-RL
‘he opened his bag and took the big tobacco container out in a hurry’ [HK, TR 093]
t “ chelangtha ” {mm}
luj d ath k jo che-l ng-th mm
mirror exist on.top see RR-see-CON.IMP AFF
‘the mirror is on it, “see, look yourself!” ' [HK, TR 095]
ansose dakke” pu
n-sos d k ke pu
that.much-more here=TOP QUOT
' “here is the place where I locked him, this one (tiger) here is more (i.e. bigger, stronger)
than you” ' [HK, TR 096]
{chethekloklo}
che-th k-l k-l
RR-see-happen.to-RL
‘that one (the tiger) looked this side, see, that tiger peeped, and got to see his image’ [HK,
TR 097]
647
“opeija nephanta halasi nangsarlang ” <both laughing>
opeija ne-ph n t h la si nang s r-l ng
my.goodness! 1EXCL-NSUBJ=ADD:even that=FOC:RL 1/2:NSUBJ=be.old-still
‘ “my goodness! this one is still older than me!” ’ (both laughing) [HK, TR 098]
ingplongphitlo {ingplongphitlo}
ingpl ng-ph t-l ingpl ng-ph t-l
run.away.animal-right.away-RL run.away.animal-right.away-RL
ingplongphitlo {mm}
ingpl ng-ph t-l mm
run.away.animal-right.away-RL AFF
‘as he was looking, he of course saw the image and he ran away’ [HK, TR 101]
648
“komantupo aRongkerjike ” {mm}
komant -p a-R ngk r-j ke mm
when-IRR1 POSS-FESTIVAL-IRR2=TOP AFF
‘ “when will it be, the Rongker?” ‘ [HK, TR 103]
nangchinike {mm}
nang chin ke mm
1/2:NSUBJ=know=TOP AFF
‘{“I know the place where you stay!”}’ [HK, TR 105]
649
{“nangkokjike”}
n ng-k k-j ke
need-absolutely.required-IRR2=TOP
do tangho {o} mm
d t ngh o mm
exist REP SURPRISE AFF
‘uhm.. that day, if (i.e. because) it's a Teke Rongker, there are silver coins, there are
earthen pots full of money‘ [HK, TR 110]
650
along tangka atibuk nunrenpo lasi
al ng t ngk a-t b k n n-r n-p las
LOC money POSS-earthen.pot place.container-in.a.row-IRR1 therefore
‘they (i.e. the tigers) distribute/offer the money pots for the god, for all the gods, {on their
plates (of offerings)}, and they will put the earthen pots in one long row’ [HK, TR 111]
651
pu tangho {mm} “vangvekji” o
pu t ngh mm v ng-v k-j
QUOT REP AFF come-definitely-IRR2 DSM
‘and then, he keeps shouting, “tomorrow I will come, when it's a new day again
tomorrow, I will come, definitely I will come” he said, “definitely I will come” ' [HK, TR
115]
652
hem nanglengphitlo lata
h m nang l ng-ph t-l l t
house CIS=go.long.distance-right.away-RL this=ADD:DM
653
{esonbak klemdunpo} “esonbak klemdunlo tangte....
e-s n-b k kl m-d n-p e-s n-b k kl m-d n-l t ngt ....
one-CLF:thing-about do-JOIN-IRR1 one-CLF:thing-about do-JOIN-RL if
{kidukthektik}
ke-d k-thekt k
NMLZ-be.poor-as.V.as.it.can.be
‘that one also is an unimaginably poor man’ [HK, TR 128]
654
“dak vangtha, ko jirpo ” {mm}
d k v ng-th ko j rp mm
here come-CON.IMP buddy:VOC friend AFF
‘ “come here, my friend!” ’ [HK, TR 129]
655
<rong> teke atum <a> rongkerji” {teke rongkerji} mm
<rong> tek a-t m r ngk r-j tek r ngk r-j mm
village tiger POSS-PL FESTIVAL-IRR2 tiger FESTIVAL-IRR2 AFF
‘ “the news is that, my friend, tomorrow the tigers will have a Rongker” ‘ [HK, TR 135]
656
“t t
te m p -n ~pin -C -d t-j ma
and.then/therefore future what-INDEF~EE-NEG-PFV-IRR2=Q
” { }
ko j rp pu mm
buddy:VOC friend QUOT AFF
‘”and there won't be anything (any difficulties, problems, dangers), my friend?”' [HK, TR
140]
{pudun'etlo}
p -d n- t-l
say-JOIN-PRF-RL
‘ “nothing will happen, friend, I've already done the talking (I've made the necessary
preparations, taken the necessary precautions)” ‘ [HK, TR 141]
damnangji” {mm}
d m-n ng-j mm
go-must-IRR2 AFF
‘ “Okay, then!”, that friend followed him, “in the very early morning we need to go”
“need to go very early” “we need to go very early, we need to go before anybody gets
there” ‘ [HK, TR 142]
657
“damdapprangnangji” {mm} mm to
d m-d ppr ng-n ng-j mm mm t
go-early-must-IRR2 AFF AFF OK
‘ “in the early morning we need to go” ’ [HK, TR 143]
658
<laso> aningkanvang <along> Rongker <adim> apirda {mm}
<las > a-ningk n-v ng <al ng> R ngk r <a-d m> a-pird mm
this POSS-year-every LOC FESTIVAL POSS-place POSS-place.of.worship AFF
kithirthipke mati{mm}
ke-th r-th p ke=mat mm
NMLZ-be.firmly.fixed/unvariable(<Asm)-firmly=TOP=CG AFF
vangvelang {mm}
v ng-C -l ng mm
come-NEG-yet AFF
‘those, the ones performing the teke rongker have not yet come’ [HK, TR 150]
659
latumke le'etlo {le'etlo}
l -t m ke l - t-l l - t-l
this-PL=TOP reach-PRF-RL reach-PRF-RL
‘these (i.e., the orphan and his friend) have already reached’ [HK, TR 151]
tangho, “ t ” {mm}
t ngh ] [pe-pr p-n ng t ko-mar-l h ] mm
REP CAUS-be.quick-HORT EMPH buddy:VOC-HUM:PL-HON EMPH:INTERACT AFF
‘and then, that tiger that had encountered (the orphan), that is the barika, he kept saying,
“let's make it quick, friends!”' [HK, TR 153]
660
siksakpo”
siks k-p
be.difficult-IRR1
‘ “afterwards if any living being (i.e. the orphan and his friend) comes, it will be difficult
for us” ’ [HK, TR 156]
{o}
o
INTERJ
‘ “we have to make it quick, we have to make it quick” ‘ [HK, TR 157]
{nangledunrokte kopulo} o
nang l -d n-r k-t kop lo o
CIS=reach-along-completed-COND what.will.happen? AFF
‘ “ ‘we also will come’ he already said” {if they reach, what will happen?} yes’ [HK, TR
159]
661
<pi> a'um akhangrata tangka
<p > a- m a-khangr t t ngk
what POSS-cage POSS-basket.to.carry.firewood=ADD:also money
662
{latum vangve ako} vangve ako {mm}
l -t m v ng-C ak v ng-C ak {mm}
this-PL come-NEG before come-NEG before AFF
‘{before they (i.e., the tigers) came?!} - before they came’ [HK, TR 165]
663
{ha Chomang penak Keche penak dopiklo} pine
h chom ng pen k Kech pen k d -p k-l p -n
over.there Khasi actually EE:Chom ng actually exist-very-RL what-INDEF
klemtotlo {oi}
kl m-t t-l i
tremble-RL yes
‘{actually, there are many Khasis and plains people}- (yes,) there are ones from so many
different tribes; but the friend was very scared so he was trembling’ [HK, TR 170]
{“ t t ? achimi dokok”}
nang th k-l k-t kop lo a-chim d -k k
1/2:NSUBJ=see-happen.to-COND what.will.happen? POSS-claw exist-firmly
‘{“if they happen to see us, what will happen?, they surely have claws”}’ [HK, TR 171]
664
halake kaphere'ong pu'amatlo {mm}
h la ke ke-pher - ng pu am t lo mm
that=TOP NMLZ-fear-be.much QUOT=and.then=FOC AFF
{the'o'otangma, atibukta?} mm
th - ~ -t ng ma a-tib k t mm
be.big-much~DIST.PL-REP=Q POSS-earthen.pot=ADD:DM AFF
‘{are the earthen pots big?}’ [HK, TR 177]
665
ingjonglunlanlo karhanghunglo {karhanghunglo} mm
ingj ng-l n~l n-l k r-h ng~h ng-l k r-h ng~h ng-l } mm
move-big:AO~DIST.PL-RL burn-quite~DIST.PL-RL burn-quite~DIST.PL-RL AFF
‘and then, they came, everything has been arranged nicely and then they looked big and
were moving simultaneously and they looked like fire (because of the tiger stripes)’ [HK,
TR 180]
{nangle'etlo}
{nang l - t-l
CIS=reach-all:S/O-RL
‘they are just going to perform the worship but they still haven't reduced the number of
the men collected for even one (i.e. hadn't sacrificed anybody yet); it' s already all been
arranged, the earthen pots have reached’ [HK, TR 181]
666
anke lake phereboplo {mm}
nke l ke pher -b p-l mm
and.then this=TOP fear-to.death-RL AFF
‘and then, this one (i.e., the friend) was scared to death’ [HK, TR 184]
nangle'etlo {oi}
nang l - t-l i
CIS=reach-all:S/O-RL yes
‘the barika and the sarthe and all the other important tigers have already arrived there’
[HK, TR 186]
vang'etlohe
v ng- t-l =he
come-all:S/O-RL=AFTERTHOUGHT
667
chirip thekthedetlo nangklodurlo
che-r p th k-C -d t-l nang kl -d r-l
RR-hold.firmly know.how-NEG-PFV-RL CIS=fall-fall.down.bigger.than.expected-RL
nangklo
[nang kl ]
CIS=fall
‘the friend was very scared and then from above this tree he didn't know how to hold on
to it firmly and he fell down’ [HK, TR 188]
{mm klodamduplo} mm
mm kl -d m-d p-l ] mm
AFF fall-go-falling.sound.from.high.solid.obj-RL AFF
{nangklodurlo} mm
[nang kl -d r-l ] mm
CIS=fall-fall.down.bigger.than.expected-RL AFF
‘right in the middle of them he fell down, yeah he fell down...he fell down...’ [HK, TR
189]
t atumta
kl -d p-p n ke tek a-t m t
fall-falling.sound.from.high.solid.obj-NF:with=TOP tiger POSS-PL=ADD:DM
668
langphrong hala abarikake <...> “lake komarli”
l ng-phr ng h la a-barik ke l ke ko-mar-l
see-PL:S that POSS-OFFICER=TOP this=TOP buddy:VOC-PL:HUM-HON
‘they saw them and that barika tiger said, “see friends!” ‘ [HK, TR 192]
abangke <lake> mm
ab ng=ke <l ke> mm
NPDL=TOP this=TOP AFF
' “from yesterday I've told you and from this morning I've told you, hurry up!”, that
friend…’[HK, TR 193]
669
amat la jangrengso atum abangke jirpopo laso atangka
am t l jangr ngs a-t m ab ng ke j rp ~p las a-t ngk
and.then this orphan POSS-PL NPDL=TOP friend~DUAL this POSS-money
la'anhelo tangho
l n=hel t ngh
this=that.much=RL:EMPH REP
‘that’s it’ [HK, TR 199]
670
jumepik phu! nang tomo kethan
arj -m -p k ph n ng tom ke-th n
listen-GOOD-very grandfather:VOC you story NMLZ-tell
671
APPENDIX D
Folk story about a chain reaction of events that sets off when a frog blocks the
road and sits down on an ant that in turn bites the frog; from there on, one animal suffers
from being disturbed or hurt by another animal, and as a consequence accidentally
disturbs or hurts another animal, and so on: the ant disturbs the frog, the frog disturbs the
squirrel, the squirrel disturbs the pig, the pig disturbs the bird, the bird disturbs the
elephant, the elephant kicks a rock, which then kills the daughter of the king. The king
traces back the chain reaction to the frog and the ant, and punishes both of them.
The audio file for the entire text RBT, ChM is available under the DOI name
10.7264/N3XK8CTJ at https://scholarsbank.uoregon.edu/xmlui/handle/1794/13657
673
arnisi
[arn -s ]
day-one
‘it's like this, the story of when the ant fought with the frog, one day’ [RBT, ChM 007]
674
ai tovar nangpektha ti ko jirpo,
i tov r nang p k-th ti ko j rp ,
how.bad! road 1/2:NSUBJ=give.way-IMP:CON EMPH buddy:VOC friend
chingkiponklung atum'anta
che-ingk -p n-kl ng a-t m- n t ]
RR-talk-CARRY-echoing.sound POSS-PL-all=ADD:EXH
‘ “you should pass through between my legs, buddy, just awhile earlier, all of these
people that are talking there have done the same” ‘ [RBT, ChM 014]
675
anborpenpen chongho abang ingnidunpret
n-b r p n~p n chongh ab ng ingn -d n-pr t
rice-wrapped.bundle=with~DISTR.PL frog NPDL sit-JOIN-acting.on.inflated.obj
‘the frog said, and as the ant was passing through, with all its rice bundles, the frog sat
down (pressing down the rice bundles)’ [RBT, ChM 016]
676
pu, lata
pu l t
QUOT this=ADD:DM
‘and then, the squirrel... got mad, “who is the wise one, who is the smart one, who is the
strong and mighty one, and sat down on my ladder so it broke?” it (the squirrel) (said)’
[RBT, ChM 019]
677
joningkoi jonthomkep kedoji, lata
j n-ingk i j n-th m-k p ke-d -j [l t
CLF:animal-twenty CLF:animal-three-ten NMLZ-exist-IRR2 this=ADD:DM
aso kaluksonlo
a-s ke-luks n-l
POSS-child NMLZ-lose(<Asm)-RL
‘and then, as a consequence, this pig startled and then it also was lifting up things with its
snout here and there, and then it lifted and uprooted the plantain plant with its snout, (the
plantain plant) beat the nest of a Voarbi and then there will be many many offspring, this
one's (the bird's) offspring were also lost’ [RBT, ChM 022]
678
harlong turpurlo, aharlong ingplonglo, richo
[harl ng t r-p r-l ], [a-harl ng ingpl ng-l ], [rich
stone kick-move.over-RL POSS-stone run.away.animal-RL king
harlong baplam
harl ng b p-l m
stone press.down-RES:paste.like
'and then, “even though it's already past mid-day, why has my child having gone to the
washing place still not come back?”, said the king and all the subjects went to bring (the
king's daughter)... and she was flattened by the stone' [RBT, ChM 027]
679
nesopi aphan kipithima?”, “O, ne
ne-oso-p a-ph n ke-pV-th ma] [o n
1EXCL:POSS-child-female POSS-NSUBJ NMLZ-CAUS-die=Q SURPRISE 1EXCL
kali.”
kal ]
NEG.EQU.COP
'and then the king asked the stone, "for what possible reason did you kill my daughter?",
"o, it wasn't me" ' [RBT, ChM 028]
“komatlo tangte”
kom t lo t ngt
who=FOC if
'"who then?"' [RBT, ChM 029]
vangnangkoklo”
v ng-n ng-k k-l ]
come-must-absolutely.required-RL
‘ “the deaf elephant lifted up my butt, and then also, I don't have hands or feet, and like
this, I had to come (rolling)” ’ [RBT, ChM 030]
680
‘ “and then, I knew your child was there, but I also couldn't control myself and then I
unintentionally took your daughter's life”.’ [RBT, ChM 031]
damji, <nangsapdamji>
d m-j ] <nang s p-d m-j >
go-IRR2 1/2:NSUBJ=beat.w/sth.flexible-go-IRR2
nangsapji nangthengji ”
nang s p-j nang th ng-j
1/2:NSUBJ=beat.w/sth.flexible-IRR2 1/2:NSUBJ=beat.w/sth.solid-IRR2
' “what did you kick the rock for? The life of my daughter has gone. Today your life and
everything will go, I will beat you, I will make you suffer!” ' [RBT, ChM 033]
681
voarbipi, akam kechomathale neno
v arb p a-k m ke-chomath le ne-n
bird.sp POSS-work NMLZ-think.with.bad.intentions=FOC:IRR 1EXCL:POSS-ear
nanglutchok nangarkerakrakdetkema?"
nang l t-ch k nang ark -r k~r k-d t ke ma]
CIS=enter-disappearing CIS=scratch-RES:little.wound~DISTR.PL-PFV=TOP=Q
'he said and then, the deaf elephant answered, 'O Vo arbi, what were you thinking,
coming into my ears and scratched there; that's why" [RBT, ChM 034]
turpurlo"
t r-p r-l ]
kick-move.over-RL
'and then, "I also didn't know how to stay or how to rest, my ears were itching, my ears
are deaf, and then it's true I was kicking here and there, and I kicked the rock"' [RBT,
ChM 035]
682
voarbipi aphanle arjudamnoi"
v arb -p a-ph n le] arj -d m-n i]
bird.sp-female POSS-NSUBJ=FOC:IRR ask-go-INF.COND.IMP
'"therefore, I'm not the guilty one, king, go ask that Voarbipi!"' [RBT, ChM 037]
nangenri"
nang n-r
1/2:NSUBJ=take-PROH
'"Ai, king, don't take my life!"' [RBT, ChM 041]
683
“pisi ne ingnar nothongpo ano lutledetjima ”
p si n ingn r noth ng-p a-n l t-C -d t-j ma
why 1EXCL elephant deaf- MODIF POSS-ear enter-NEG-PFV-IRR2=Q
‘ “how could I’ve not gone into the deaf elephant's ears?” ' [RBT, ChM 042]
684
"O lobong, chonghoisi nang... voarbipi atar
o lob ng, chongh i si n ng... v arb -p a-t r
VOC plantain reason=FOC:RL 2 bird.sp-female POSS-nest
’ t "
nang pa-v r- t-j sala th r-d m-l
1/2:NSUBJ=CAUS-lose-all-IRR2 damn.you! threaten-go-RL
'"O plantain, why the heck did you break the nest of the Vo Arbipi, your life will go, I
will destroy your tribe and your species, damn you!", he went to threaten (the plantain)'
[RBT, ChM 046]
lobong chethakdunlo
lob ng che-th k-d n-l
plantain RR-answer-JOIN-RL
'the plantain answered' [RBT, ChM 047]
685
ingnilun anat”
ingn -l n a-n t]
sit-big:AO:not.suddenly POSS-only
'"and then, I also couldn't stay still, I also just sat down big as I am"' [RBT, ChM 050]
Nangpithitheiji!"
nang pi-th -th i-j
1/2:NSUBJ=CAUS-die-all:S/O-IRR2
'"today, I will destroy your tribe and kind, I will kill all of you!"' [RBT, ChM 054]
686
“Nangtum kopisi lobong katimurthap katimurphru? <ar>
[nang-t m kop si lob ng ke-tim r-th p ke-tim r-phr ]
2:POSS-PL why plantain NMLZ-root-mindlessly NMLZ-root-EE:-th p
687
ha hanthar aphan arjudamlo... "Kopisi nang
h hanth r a-ph n arj -d m-l kop si n ng
over.there vegetable.sp POSS-NSUBJ ask-go-RL why 2
nepivirri!"
ne pi-v r-r
1EXCL:NSUBJ=CAUS-lose-PROH
'"ai, king, don't kill the ones of my kind, don't kill my tribe!"' [RBT, ChM 062]
688
neta ahoklo phakbelengpi amoi
[n t a-h k-l ph kbel ngp a-m i
1EXCL=ADD:DM POSS-truth-RL pig.sp POSS-back
chonduplo”
ch n-d p-l ]
jump-falling.from.high:solid.obj-RL
' “the squirrel gnawed off my hands and then, it's true, I jumped on the pig's back” '
[RBT, ChM 064]
peklodup athema?"
pe-kl -d p ath ma
CAUS-fall-falling.sound.from.high.solid.obj because
'over there he went and asked the squirrel, "so then, squirrel, why did you gnaw on the
Hanthar and thus made it fall down on the back of the pig?"' [RBT, ChM 066]
689
“ai nepran neenri, nemui
i ne-pr n ne n-r ne-m i
how.bad! 1EXCL:POSS-life 1EXCL:NSUBJ=take-NEG.IMP 1EXCL:POSS-EE:pr n(<Ind)
neenri, richo ”
ne n-r rich
1EXCL:NSUBJ=take-PROH king
'"please, don't take my life, king"' [RBT, ChM 068]
chomathalone, nedon
cho-math -l -n ] [ne-d n
AUTO.BEN/MAL-think-RL-INDEF 1EXCL:POSS-ladder
690
nedon kachithu”
ne-d n ke-chi-th ]
1EXCL:POSS-bridge/ladder NMLZ-RR-cut
' ‘the frog and the ant doing whatever mischief, they just jumped on my ladder and
destroyed it; because of that I was just collecting the materials to (re-)build my ladder’ '
[RBT, ChM 072]
adon chonraimati?" pu
a-d n ch n-r i mati] pu
POSS-bridge/ladder jump-RES:solid.obj.breaking=as.you.know QUOT
‘(the king) asked the frog, "Why did you jump on the ladder of the squirrel?"' [RBT,
ChM 074]
691
o <nang> mota nangtum kachekoi
o <nang> m t nang-t m ke-che-k i
INTERJ you future=ADD:EXH you-PL NMLZ-RR-accuse
nangbe doji
nang-b d -j
2:POSS-habit exist-IRR2
'"o, you will (continue to) in the future have a habit to accuse each other"' [RBT, ChM
077]
pechengran
pe-ch ng-r n
CAUS-be.narrow.inbetween-delicate/about.to.break
‘(the king) took the ant on its hair and tied it up on its waist' [RBT, ChM 078]
pevangphrok
pe-v ng-phr k
CAUS-come-bulging.out
‘and with a bundle of jamir they beat the frog so his skin got swollen’ [RBT, ChM 079]
ansi ingtonlo
nsi ingt n-l
after.that conclude-RL
'and that's it’ [RBT, ChM 080]
692
APPENDIX E
This folk story gives an account of subdivisions within the Bey clan. The first part
of the folk story focuses on the two brothers y ’ (Bey the Black) and y ’ t
(Bey the Fair) and their falling out over changing marriage plans as a result of the
grandmother’s advice. The second part of the story focuses on members of the Bey
Ronghang clan and gives an account of why in modern times they are not allowed to
grind rice for rice beer. The story’s account is that in mythological times, the Bey
Ronghang women kept grinding rice for rice beer without doing anything else, even
forgetting to breast-feed their children, which subsequently died from starvation.
The audio file for the entire text WR, BCS is available under the DOI name
10.7264/N3JD4V2T at https://scholarsbank.uoregon.edu/xmlui/handle/1794/13657
693
anke latum thelo dinglo, piso some enlo tangho
nke [la-t m th -l d ng-l ] [p s s m n-l ] t ngh
and.then this-PL be.big-RL be.long-RL wife EE:p s take-RL REP
'and then, they grew up and they got married, so they say' [WR, BCS 003]
694
halata asomar thelo dinglo
h la t a-oso-m r th -l d ng-l
that=ADD:also POSS-child-PL be.big-RL be.long-RL
'his children were growing up and then Bey the Fair also did the same and he got married
and so with respect to him also, his children grew up' [WR, BCS 006]
695
piso hangdamlo anke kene, ‘Bey Ki'ik‘ pu
p s h ng-d m-l nke ken B y ke- k pu
wife call-GO-RL and.then filler CLAN NMLZ-be.black QUOT
t t ‘Bey
n las la-t m kort b ng-hin a-l ng pen [B y
and.then therefore this-PL brother CLF:HUM:PL-two POSS-LOC=from CLAN
piso hangdamlohe
p s h ng-d m-l he
wife call-GO-RL=EMPH
‘and then, therefore, among the two brothers, their sister went to the so-called ‘Bey the
Black’, to his house, to ask for a wife, you know’ [WR, BCS 011]
696
sarpi dolang, anlo aphi abang kipu tangho,
sarp d -l ng] [ nlo a-ph ab ng ke-p t ngh ]
old.woman exist-still and.then POSS-grandmother NPDL NMLZ-say REP
697
Ki'ik ahemsi nang piso hangdamrong”
ke- k a-h m si n ng p s h ng-d m-r ng]
NMLZ-be.black POSS-house=FOC:RL 2 wife call-GO-instead
‘ “it's to your second-born (maternal uncle), you know, to Bey the Fair (that you should
go and ask for a wife), at your second-born maternal uncle's house, his daughters are all
equally beautiful, you know, father, you went there to the house of Bey the Black to ask
for a wife instead” ‘ [WR, BCS 013]
ki'ikrintile putangho
ke- k-rint le p -t ngh
IPFV-be.black-equally:PL:S/A=FOC:IRR QUOT-REP
' “the family of your oldest maternal uncle, the family of your uncle who is the eldest son
of the family, all of them are so black (so you shouldn't get your wife from them)” ‘ [WR,
BCS 014]
698
ke'etrinti” pu apot, asupo
ke- t-rint pu ap t] [a-su-p
IPFV-be.yellow-equally:PL:S/A QUOT because POSS-grandchild-male
699
piso hangdamrong hadaksi pangri pangdonrong tangho
p s h ng-d m-r ng h d k si pangr pangd n-r ng t ngh
wife call-GO-instead there=FOC:RL reconcile even.out-instead REP
‘because it was his grandmother's advice, he agreed, and so he went to Bey the Black's
house to ask for a wife and do all the formalities, and prepare the banana leaves and the
curry (for the wedding) and everything, they prepared all of the necessary things for the
wedding, and then there all the food got rotten, (because) he had followed his
grandmother's advice and had gone to Bey the Fair's house, you know; there (that man's)
daughter, to the daughter of Bey the Fair, you know, he went to ask for her as a wife
instead, there he went to get married instead' [WR, BCS 016]
y
pangr pangd n-r ng-l nke h aph ke l B y
reconcile even.out-instead-RL and.then over.there after=TOP this CLAN
che'oitanglo
che- i-t ng-l
RR-be.sad-finish-RL
‘they got married, and then after this, Bey the Fair and Bey the Black got upset with each
other’ [WR, BCS 017]
700
Ke'et ahem chepangri chepangdonronglo; anke
ke- t a-h m che-pangr che-pangd n-r ng-l [ nke
NMLZ-be.yellow POSS-house RR-reconcile RR-even.out-instead-RL and.then
t t
ch k-C -d t-l mati
be.fine-NEG-PFV-RL=CG
‘up to today, Bey the Fair and Bey the Black can't tolerate each other (among) the Karbis,
and the reason why Bey the Fair and Bey the Black can't tolerate each other is that <that
wife>, grandmother made a mistake, you know’ [WR, BCS 019]
t “nangong, Bey
[a-su-p a-ph n p -d n-d t] [[nang- ng B y
POSS-grandchild-male POSS-NSUBJ say-JOIN-PFV 2:POSS-maternal.uncle CLAN
t ”
ke- t a-h m si] [a-oso-p ] [arj ke-m ]]
NMLZ-be.yellow POSS-house=LOC POSS-child-female appearance IPFV-be.good
701
t , hadak chepangridet apot hedi;
pu p -d n-d t-pen h d k che-pangr -d t ap t hed
QUOT say-JOIN-PFV-NF:with there RR-reconcile-PFV because EMPH
“ t t t y t
m pirth ke-ingt ng t n ng B y ke- t pen
future world NMLZ-be.strong=even you CLAN NMLZ-be.yellow=with
ke latum t
ke la-t m b ng-hin ab ng ke kort b ng-hin
and.then this-PL CLF:HUM:PL-two NPDL=TOP brother CLF:HUM:PL-two
702
Ki'ik chi'une
ke- k che- n-C
NMLZ-be.black RR-be.able-NEG
‘and so up to today, Bey the Fair and Bey the Black can't tolerate each other’ [WR, BCS
024]
atumlo tangho
a-t m lo t ngh
POSS-PL-RL REP
‘and then, the (children of the) so-called ‘Bey Ronghang’ were the youngest’ [WR, BCS
026]
703
anke latum abangke hedi thap ketoklok tangho
nke la-t m ab ng ke hed th p ke-t k-l k t ngh
and.then this-PL NPDL=TOP NP.EMPH cake.for.rice.beer IPFV-pound-only REP
‘and then, they kept pounding the rice to prepare the cake for rice beer, it is said’ [WR,
BCS 028]
t t t gle,
[th p ke-t k-l k] [j arn t s ng-l ng-C ]
cake.for.rice.beer IPFV-pound-only night day=ADD:EXH take.rest-GET-NEG
704
anke latumta hedi seme tangdetlo
nke la-t m t hed sem t ng-d t-l
and.then this-PL=ADD:also NP.EMPH vow(<Khs) LV-PFV-RL
‘and then, they also, you know, took a vow’ [WR, BCS 032]
toklongle
t k-l ng-C ]
pound-GET-NEG
‘they can tolerate anybody (i.e. can tolerate both Bey the Black and Bey the Fair), the
youngest brother's clan ones, you know, Bey the Fair's clan members they can tolerate,
and Bey the Black's clan members they can tolerate, but with respect to their vow, they
shouldn't pound the rice for the rice beer cake’ [WR, BCS 033]
“t t t t t
th p t k-l ng-C m pirth ke-ingt ng set l
cake.for.rice.beer pound-GET-NEG future world NMLZ-be.strong but this
nangpatoklonglenang” pu
nang pa-t k-l ng-C -n ng pu
1/2:NSUBJ=CAUS-pound-GET-NEG-HORT QUOT
' “we don't get to pound, even in the future when the world is mature, we, the Bey
Ronghang shouldn't be made to have to pound the rice for rice beer cake”, (he) said'
[WR, BCS 034]
705
lasi pini-arni'anta hedi la Bey Ronghang atumke
las pin -arn - n t hed l B y Rongh ng a-t m ke
therefore today-day-till=ADD:even NP.EMPH this CLAN CLAN POSS-PL=TOP
706
kekoinangji bongpen, kotpenhe pu
ke-k i-n ng-j ] [b ng pen k t pen he]] [p
IPFV-rub-must-IRR2 bottle=with paper(<Asm)=with= AFTERTHOUGHT like.this
707
APPENDIX F
This text is an online narration of the Pear Story (Chafe 1980). The speaker was
asked to describe what was happening as he was watching the video clip.
The audio file for the entire text SiT, PS is available under the DOI name
10.7264/N3P55KRP at https://scholarsbank.uoregon.edu/xmlui/handle/1794/13657
vo kiku
v ke-k
chicken IPFV-crow
'a rooster is crowing' [SiT, PS 001]
708
arlosi lahai kethap lapen arum kevan
arl si] [lah i] ke-th p]] lap n [ar m ke-v n]]
inside=FOC:RL these NMLZ-put.inside and down NMLZ-bring
'he's placed himself a ladder... and then, the man, into one bag sown from white cloth he
puts these (fruits), and then brings them down’ [SiT, PS 003]
ki'iksi kachingchoi
ke- k si ke-chingch i]
NMLZ-be.black=FOC:RL NMLZ-wear
‘this man also has a cloth on his neck, on his head he wears a hat, he wears a black shirt’
[SiT, PS 006]
709
kethepar
ke-th -p r]
NMLZ-be.big-very
‘there's a ladder, the tree is very big; this fruit tree..., the voice of a goat is also very loud’
[SiT, PS 007]
710
kelik amonit adung'an nanglelo
ke-l k a-mon t ad ng- n] nang l -l
NMLZ-pluck POSS-man near-up.to CIS=reach-RL
‘over there, far away, all the way down the road, one hat-wearing man, dragging along a
female goat, close up to this fruit picking man he reached’ [SiT, PS 010]
kedun mo
ke-d n m
NMLZ-join hesitate
‘this person also... like this carries the goat, this goat is hesitating to go along’ [SiT, PS
011]
chithurponbomlo
chith r-p n-b m-l
drag-take.away-CONT-RL
'it's as if the goat wants to graze, but that man without stopping for awhile keeps dragging
(the goat) away' [SiT, PS 012]
711
chonghoipen kecho amonit abangke... puthot
cho-ingh i-pen ke-ch a-mon t ab ng ke]... puth t
AUTO.BEN/MAL-do-NF:with NMLZ-eat POSS-man NPDL=TOP next/again
712
phatang hini plengchongchong... <a> thesere kiliktang
phat ng hin pl ng-ch ng~ch ng... theser ke-l k-t ng
kd.basket two be.full-slightly.high~INTENS fruits NMLZ-pluck-PFV
713
phatang abang vansi... la phatang saikel along
[phat ng ab ng] v n-si... l phat ng saik l a-l ng
kd.basket NPDL bring-NF:RL this kd.basket bicycle(<Eng) POSS-LOC
kethap
ke-th p
NMLZ-put.inside
‘and then he puts them on the bicycle in the front’ [SiT, PS 023]
kevekponlo
ke-v k-p n-l
NMLZ-steer-away-RL
‘and then, he doesn't know how to sit down on that sitting place (saddle), and then on the
iron bar he sits and steers the bicycle away’ [SiT, PS 024]
714
saikel vekponbom dambomlo atheta
[saik l v k-p n-b m d m-b m-l ] a-th t
bicycle(<Eng) steer-away-CONT go-CONT-RL POSS-fruit=also
715
amat laso damchet amat chekhang'un'elo isi arlong
am t las d m-ch t am t che-kh ng- n-C -l is arl ng
and.then this go-a.bit and.then RR-keep-be.able-NEG-RL one stone
t
a-l ng t ng-d r
POSS-LOC collide-IDEOPHONE
‘and then he went just a bit further and then he couldn't hold himself anymore, he hit a
stone’ [SiT, PS 029]
juirekraksi do
j i-rek~r k-si d
play-silently~DIST.PL-NF:RL exist
‘nearby, there were a few children, they were playing silently’ [SiT, PS 031]
716
aphatang along thesere thapdunlo rapdunlo, laphan
a-phat ng a-l ng theser th p-d n-l r p-d n-l l -ph n
POSS-kd.basket POSS-LOC fruits put.inside-JOIN-RL help-JOIN-RL this-NSUBJ
‘(there’s) one child who’s holding a tennis ball and a bat, and then for the person who had
fallen down, they put with him the fruit in the basket, they helped him’ [SiT, PS 032]
717
damchot aphi.... phutup kitirok theklongloklo hala
d m-ch t aph .... phut p ke-t -r k th kl ng-l k-l [h la
go-a.bit after hat NMLZ-leave.behind-COMPL see-just-RL that
anat chevangthulo
a-n t che-v ng-th -l
POSS-direction RR-come-again-RL
‘and then, he whistled, picked up the hat and brought it back’ [SiT, PS 036]
718
an laso a'oso abang thesere pumni hala ajirpo
n [las a-os ab ng] [theser p m-n ] [h la a-jirp
and.then this POSS-child NPDL fruits CLF:round-two that POSS-friend
t t
che-pe-kl ng-d m-pen... l -t m ke-d m-th pangch ng-l
RR-CAUS-appear-GO-NF:with this-PL NMLZ-go-again start-RL
‘after going to show them, they again started to go’ [SiT, PS 041]
719
aphatang along thesere chetheklongledetlo
a-phat ng a-l ng theser che-th kl ng-C -d t-l
POSS-kd.basket POSS-LOC fruits RR-see-NEG-PFV-RL
‘and then, he again put more fruit in the baskets, but in one basket he didn't see any fruit’
[SiT, PS 044]
kamatha
ke-math
NMLZ-think
‘and then he kept watching them for a long time, "could they have taken them all away,
look at that", like this he was thinking’ [SiT, PS 046]
amat jutletlo
am t j t-l t-l
and.then finish-PFV3-RL
‘and then, (the story) is finished’ [SiT, PS 047]
720
APPENDIX G
This procedural text discusses how alkaline food (called kangmoi ahan) is
prepared. The procedure involves burning jhum fields and using the ashes. This type of
food has a high significance in Karbi culture.
The audio file for the entire text SiH, KH is available under the DOI name
10.7264/N3ST7N4P at https://scholarsbank.uoregon.edu/xmlui/handle/1794/13657
t ’helo
ke-cho-t n a-r n-ke las n- k-hel
NMLZ-AUTO.BEN/MAL-cook POSS-custom-TOP that.way-FRML-RL:EMPH
‘the custom of our Karbi people's way of cooking ourselves vegetables with alkaline is
like this’ [SiH, KH 001]
kopunelo… t
ko-pu-n lo l ingl ng a-r t p n-p
ITROG-QUOT-INDEF=FOC this hill POSS-field clear.vegetation-IRR1
‘I mean... up there on the fields on the hill, what do you call it?, we clear the vegetation
from the hill fields’ [SiH, KH 002]
721
elitum humdunji
e-li-t m h m-d n-j
1PL.INCL-HON-PL pick.up-JOIN-IRR2
‘after that, on the hill fields where we have cleared the vegetation, we have to set a fire
and then these ashes from the immature bamboo we have to pick up’ [SiH, KH 003]
do
d
exist
‘and then, there is the so-called 'phelo bisir' (funnel-like instrument for filtering the ashes)
that the Karbi people make’ [SiH, KH 004]
722
tun'ikpo
t n- k-p ]
cook-FRML-IRR1
‘and then, we pour water and then when (the bisir) has filtered (the ashes), we test the
taste, and after finding it matching, we cook the curry in a small earthen pot’ [SiH, KH
007]
kechome'aretke aseng
ke-ch -m -ar t ke a-s ng
NMLZ-eat-GOOD-INTENS=TOP POSS-condiments
kejok'ikke lapuhelo
ke-j k- k ke lap -h lo
NMLZ-drop.into.container-FRML=TOP like.this-RL
‘to become very tasty, condiments are added like this’ [SiH, KH 009]
kechojoknangjike nempo
ke-cho-j k-n ng-j ke nemp
NMLZ-AUTO.BEN/MAL-add-must-IRR2=TOP sesame
‘the condiment that we have to add is sesame’ [SiH, KH 010]
723
peme
pe-m
CAUS-be.good
‘after that, you need to pound the dried fish in the Langpong well’ [SiH, KH 011]
anke nemphruphlungpo
nke nemphr -phl ng-p
and.then have.sweet.smell-spreading-IRR1
‘and then it will develop a sweet smell’ [SiH, KH 014]
724
akechomeke lason kejok
a-ke-ch -m ke las n ke-j k
POSS-NMLZ-eat-GOOD=TOP that.way NMLZ-drop.into.container
jokpo
j k-p
drop.into.container-IRR1
‘this fragrant smell is the reason why the curry is tasty; in order to make the dish tasty,
you have to add many condiments like this, because of that you have to add tiha’ [SiH,
KH 015]
jokprepo
j k-pr -p
drop.into.container-spread.all.over-IRR1
‘and then add whatever, for example riho’ [SiH, KH 016]
725
kene elitum Karbi atumke jat chojun
ken [e-li-t m Karb a-t m ke j t [ch j n
HESIT 1PL.INCL-HON-PL PN POSS-PL=TOP type CELEBRATION
karkli kaphojo
ke-arkl ke-phoj ]
NMLZ-bewail NMLZ-EE:arkl
‘so... for us Karbi people, consider any celebration or anything, if we don't cook alkaline
curry, for us, the celebration is not adequate’ [SiH, KH 023]
726
chojunta hanmoi doji ajat doji, lasi
ch j n t hanm i d -j aj t d -j las
CELEBRATION=ADD:also ingmoi.curry exist-IRR2 GENEX exist-IRR2 then
kenangsot
ke-n ng-s t
NMLZ-need-be.required
‘there is alkaline curry at the Chojun also, and other things also (at the Chojun), therefore
for us, wherever there are customs performed, alkaline curry is required for us’ [SiH, KH
026]
727
lasi kangmoi ahan kitunnangji
las ke-ingm i a-h n ke-t n-n ng-j
therefore NMLZ-cook.with.alkaline POSS-curry NMLZ-cook-need-IRR2
t ’an'ikvetlo
ke-t n-n ng-j as n ke lap - n- k-v t-l
NMLZ- cook-need-IRR2 like=TOP like.this-that.much-FRML-only-RL
‘like this it is all about how we need to cook the curry with alkaline’ [SiH, KH 030]
728
APPENDIX H
20. METADATA
Table 121. Metadata of texts that have formed the main corpus for this dissertation
Speakers Recording Recording full name Genre206 Date Format207 Equipment208 hh:mm:ss # words
CST HM Hingchong Musoso FS 100331 A/V M2,A2,V 00:10:43 1,140
CST RO Rengsopen Onso FS 100331 A/V M2,A2,V 00:06:08 649
HI BPh Bokolapo Phinu aBiha Choklem FS 100402 A/V M2,A2,V 00:03:04 226
HK, SiT TR Teke Rongker FS 100410 A/V M2,A2,V 00:11:59 1,516
KaR, BT SWK Status of Women in Karbi Society I/C 111123 A/V M4,V 00:12:19 1,128
KK CC Crying Child N 100401 A/V M2,A2,V 00:03:41 338
KK BMS Bamonpo lapen Methan Sibongpo FS 100401 A/V M2,A2,V 00:12:52 1,317
KsT PSu Pindeng Sumpot PT 100402 A/V M2,A2,V 00:01:42 131
KTa TCS Terang Clan Story FS 090202 A M1,A1 00:09:45 1,006
PI BPR Bamboo Pork Recipe PT 090123 A M1,A1 00:01:49 118
RBT ChM Chonghokaloso lapen Misorongpo FS 100401 A/V M2,A2,V 00:05:57 733
SeT MTN Monit Thinlangno FS 100401 A/V M2,A2,V 00:06:05 710
SH CSM Chomangkan Story Mother PN 090226 A M1,A1 00:08:38 827
SiH KH Kangmoi aHan PT 100402 A/V M2,A2,V 00:03:13 291
SiH CW Cleaning and Washing PT 100402 A/V M2,A2,V 00:03:53 371
206
Genre abbreviations - PT: Procedural Text; FS: Folk Story; PN: Personal Narrative; N: Narrative; I/C: Inteview/Conversation; OEN: on-line elicited narrative
207
Format abbreviations - A: .wav; V: .mts (for further information, see §2.2.2)
208
Equipment abbreviations - M1 (Microphone1): Audio Technica AT3032; M2 (Microphone2): AT813a; M3 (Microphone3): Audio Technica AT8010; M4
(Microphone4): Audio-Technica AT831b; M5 (Microphone5): Shure SM10A; A1 (Audio recorder1): Marantz PMD 660 audio recorder; A2 (Audio recorder2):
Zoom H4n digital audio recorder; V: Video recorder Sony Vixia HF S10
729
Speakers Recording Recording full name Genre Date Format Equipment hh:mm:ss # words
SiT HF Trip to Hornbill Festival PN 111206 A A2 00:08:43 852
SiT PS Pear Story OEN 121205 A M5, A2 00:05:52 502
WR BCS Bey Clan Story FS 111013 A A2 00:05:15 630
TOTAL 02:01:38 12,485
Table 122. Brief descriptions of texts that have formed the main corpus for this dissertation
Recording full name Topic
Hingchong Musoso Folk story about two girl twins (hingchong musoso) that are born to the human wife of a king, who also has a
second wife who is a witch; the witch is jealous and abandons the girls, who are subsequently brought up by
tigers; after having grown up, they return to their father’s kingdom and tell their story; the father learns the
truth, the witch dies, and the reunited family lives happily ever after
Rengsopen Onso Folk story about step-parenthood: the mother of two children (with the names Rengso and Onso; note -
‘diminutive’ suffix, §5.4.2) dies and the father marries again; the new stepmother is evil and mistreats her step-
children, whereupon the father does not know what to do and both abandons his children in the jungle and kills
his wife
Bokolapo Phinu Folk story about a common fool character, who does stupid things (bokolapo); in this story, he wants to go to
aBiha Choklem the market but by switching sides of the basket he is carrying also accidentally switches the direction in which
he is walking; he ends up walking back to his own house without realizing it
Teke Rongker Folk story about an orphan who encounters a tiger; by using a container with a mirror on top, he manages to
convince the tiger that he (the orphan) has caught a large tiger in that container; the tiger is scared and the
orphan ends up managing to steal gold from the tigers’ Rongker (= the name of a festival in Karbi traditional
culture) due to his trick with the mirror container
Status of Women in Interview/conversation about the status of women in Karbi society; the interviewer is a man, the interviewee is
Karbi Society a woman; the interviewee points out the ways in which women have traditionally played a very important role
in Karbi society
Crying Child Narration about the (traditional) everyday situation of a mother having a baby to take care of while having to
perform other chores such as getting firewood and cooking (not considered a true folk story by language
consultants)
730
Bamonpo lapen Folk story about an elderly couple with a dog; the wife dies and the dog takes the husband to Chom arong (i.e.,
Methan Sibongpo the village of the dead); the wife does not want to leave, but the husband forces her; upon return to their
village, soon somebody else in the village dies, and the wife has to die and go back to Chom arong with the
other person
Pindeng Sumpot Procedural text / non-fictional narration about traditional clothing items, weaving, and applying colors
Terang Clan Story Folk story about the origin of the division into Terang subclans starting with two Terang brothers, the younger
one, Dili and the older one, Rongchetcho (also referred to as Rongchercho); the older one causes a lot of
hardship for the younger one, finally, the younger one, Dili, takes a vow to officially form separate subclans
(with various societal implications)
Bamboo Pork Procedural text about the recipe for pork with fermented bamboo shoots
Recipe
Chonghokaloso Folk story about a chain reaction of events that sets off when a frog blocks the road and sits down on an ant
lapen Misorongpo that in return bites the frog; from there on, one animal suffers from being disturbed or hurt by another animal,
and as a consequence accidentally disturbs or hurts another animal, and so on.
Monit Thinlangno Folk story about a couple, in which the husband tricks his wife in order to get her to give him meat to eat; in
the end, the wife leaves the husband
Chomangkan Story Personal narrative about a trip to the Chomangkan festival, narrated by the mother of the family that took the
Mother trip
Kangmoi aHan Procedural text about cooking alkaline food by burning jhum fields and using the ashes
Cleaning and Procedural text / non-fictional narration about the important status of cleanliness in Karbi culture
Washing
Trip to Hornbill Personal narrative about a trip to the Hornbill festival in Nagaland
Festival
Pear Story On-line narration / commentary of the Pear Story
Bey Clan Story Folk story about the origin of the division into Bey subclans; y ’ and y ’ t have a falling out over
changing marriage plans as a result of the grandmother’s advice; members of the Bey Ronghang clan are not
allowed to grind rice for rice beer, because in mythological times, the Bey Ronghang women kept grinding rice
beer without doing anything else, even forgetting to breast-feed their children, which subsequently died from
starvation
731
Table 123. Metadata of speakers of texts that have formed the main corpus for this dissertation
Speaker Full name Gender From (near) Living Other languages spoken, Dialect area Age
initials in (if in brackets: only a little bit)
BT Bhudeswar M Bamuni Karbi gaon Assamese, English, Hindi 39
Timung
CST Chandra Sing M (Assamese) Amri ~60
Teron
HI Hem’ari Ingjai M Pharkong Abi (Assamese) ~40
HK Harsing Kro M Borkok (Boythalangso) (Assamese) Amri ~60
KaR Kare Rongpipi F Ran Rongki Habe Arong Assamese, Hindi 48
KK Kache Kropi F Balijuri (Assamese) Socheng ~60
(Nowgong) Dhenta
KsT Kasang Teronpi F Dingso Terang (Dengaon) Assamese 45
KTa Kahan Terangpi F Diphu Assamese ~75
PI Puspa Engtipi F Diphu Assamese ~50
RBT Rongbang Teron M Ujandongka (Dongkamoka) (Assamese) Rongkhang ~50
SeT Seng Tisso M Ujandongka (Dongkamoka) (Assamese) Rongkhang ~60
SH Sashikola Hansepi F NC Hills Diphu Assamese, (English) Amri ~50
SiH Sika Hansepi F Men Terang (Dengaon) (Assamese) ~45
SiT Sikari Tisso M Bhoksong Diphu Assamese, English, Hindi, Bengali, Amri 52
Tiwa
WR Welisbon F Umswai Model Assamese, Tiwa Amri ~40
Ronghipi
732
APPENDIX I
The English and Karbi versions of the consultant consent form are attached below.
The translation of the English version into Karbi by Mr. Dharamsing Teron is gratefully
acknowledged.
733
Language Consultant Consent Form
I am Linda Konnerth, a graduate student in Linguistics at the University of Oregon. My research
project is to write a grammatical description of the Karbi language. This will be a contribution to
the field of linguistics, and is intended to serve as the basis for the development of practical school
materials in the future. The information I collect will also contribute to my PhD dissertation. You
are invited to help me with my research by serving as a consultant and sharing your knowledge of
the language.
If you decide to participate, I will ask you questions about words and sentences in your language.
Later in the project we may ask you to tell some simple stories, or to have a conversation with
another speaker. I will audio-record these words, sentences, stories and conversations, and will
continue to use these audio-recordings in future research about the language. Other copies of
these audio-recordings may be shared with the community’s language program.
In case it's OK for me to video-record your performance, please tell me which of the following you
willingly agree to:
___I agree to be video-recorded for the purpose of linguistic analysis only without making the
recording available to the public.
___I agree to be video-recorded and allow the recording to be available to the public (as may be
helpful for the development of school materials).
If at any time you feel that I may have audio-/video-recorded anything which you would
prefer that I not keep, tell me and that recording will be erased.
I intend that the results of my research will be published as a book or an article in a professional
journal. If you like, I will include your name in this publication as someone who helped with the
project. If you prefer to remain anonymous, then your name will not be included. Please tell me
which you would prefer:
___I would like to be acknowledged by name in any publication that is a direct result of this
project.
___I agree to be recorded for purposes of this project but do NOT want my name to appear in any
publication that is a result of this project.
I also need to know who will be permitted access to read, listen to, and/or see your performance.
Please tell me if and/or how you want to restrict access to your performance, and what type of
representation of your performance you are OK with to share:
734
I agree to make audio-recordings of my performance available to:
( ) nobody
( ) people authorized by the Karbi Lammet Amei
( ) public access
( )___________________________________________________________________
I agree to make video-recordings of my performance available to:
( ) nobody
( ) people authorized by the Karbi Lammet Amei
( ) public access
( )___________________________________________________________________
In signing this form you are making no commitment to this project. You are always free to
withdraw from this project at any time and for any reason. If you would like to be compensated for
your work on this project, the appropriate level will be determined by the KLA.
If you have any questions about this project, you can contact me by e-mail at
lkonnert@uoregon.edu or by telephone at +1-541-337-2352. Please also feel free to contact my
faculty advisor, Professor Scott DeLancey, who also is the Department’s Human Subject
representative, by email at delancey@uoregon.edu or by phone at +1-541-346-3901. If you have
questions regarding your rights as a research subject, and wish to contact someone outside of the
Linguistics Department, contact the Office of Human Subjects Compliance, University of Oregon,
Eugene, OR 97403, +1-541-346-2510.
Your signature indicates that you have read and understand the information provided above, that
you willingly agree to participate, that you may withdraw your consent at any time and discontinue
participation without penalty, that you have received a copy of this form, and that you are not
waiving any legal claims, rights or remedies. Please request a copy of this form if you would like to
keep it for your records.
This research project is funded by the National Science Foundation, Doctoral Dissertation
Research Improvement Grant # 0951749.
Date ______________________
735
anguage onsultant onsent Form
Nelimen ke Linda Konnerth lapen Oregon University along Linguistics ahamphang arlo charlibom
enut a-‘graduate’. Neli Karbi lamtasam aputhaksi tarlipbom. Laso aketarlip ajok linguistics aphan
akerap dopo lapen lo kethan aphanta kenangsot akerap longponpo. Neli ketarlip pen kelong pon
alamthe ke neli kecharlibom PhD aphanta akerap longponpo. Nangli ta labangso aketarlip along
‘rap-bang’ enut nangmandunra Karbi alam kepachan along nangcherap dunpo pusi choningri iklo.
Nangli nangcherapdunlo pulote nangliphan Karbi lamthe lapen lamseng aputhak karjupon doji.
Laso aphi, Karbi atomo lapen enut hejan kachingki heihui ta bangso aketralip aphan kenangsotji.
Nangli lamthe-lamseng, tomo lapen enut hejan kachingki heihui anta kepangsip si bipo lapen mo
aphi aketarlip aphan ta lahai kenangsot anke enponbom po. Lahai kepangsip tang ason amung an
Karbi alam kepachan along ta enpon unpo.
Jongsi nangli lam-arje kepangsip aputhak kachekroi do pulote, la aber ateng pen pachini ik tha:
___Karbi alam ketarlip aphan ke neli lam-arje kepangsip along neli chekroi lo bonta la rat angno
paklang nangne.
___Neli lam-arje kepangsip rat aphan paklang tame (athe lake lo kethan ahut kacherap doji).
Kepangsip tang lam-arje anbong ning kephophe kopine do pulote lahai kepapraiji aphan
neliphan pachini non.
Neli ning kehang ke lahai ketarliptang ason-amung an ta ardi kedo alo-athui asonsi paklangpo.
Nanglining oi’e pulote, laso alo athui along nanglimen ta paklang pon po. Bonta nanglimen
paklangpon nangne pulote, paklang ponpe lo. Laso aputhak nanglining kehang alam chpaklang ik
tha:
___Ketarlip pen kelongpon amo an ajat akangchap alongle klangpon setame nelimen kehumripon si
nelining kehang.
___Laso aketarlip apharman neli lamthe kepangsip aputhak neli kachekroikre ave bonta mo aphi
kepaklangji alo-athui along nahokta nelimen paklangpon longle.
Lapen komatsi lahai ason-amung kelang, karju ajakong longji ma, la neliphan pachinimek nangji.
Neliphan pachini tha jongsi lahai kelang-karju aputhak kekhang do pulote kodak lapen kopi
angdeng along si lason akekhang kedoma lapen kekhang ave pulote Kodak/kopi angdeng along ma.
( ) maduta longle
( ) Karbi Lammet Amei pen kangvaitang abangphu/phi
( ) jokta long
( )________________________________________________________________
736
Neli lam kepangsip tang an karjulongji ke:
( ) maduta longle
( ) Karbi Lammet Amei pen kangvaitang abangphu/phi
( ) jokta long
( )___________________________________________________________________
( ) maduta longle
( ) Karbi Lammet Amei pen kangvaitang abangphu/bang phi
( ) jokta long
( )___________________________________________________________________
Bangso alo along nanglimen nangchipidunpen nangli bangso aketralip along thang akachechak ave.
Nanglining hanglet aling ahut seta nangli bangnso aketarlip along pen nangchehekdunjok at chok.
Nangli nangkacherapdun apharman kopine anam kenangji pulote lake KLA pensi cherai ponpo.
Bangso aketarlip aputhak kopine kachini kangtung dolang pulote neli e-mail lkonnert@uoregon.edu
kalite lamrihing +1-541337-2353 along pachini ik tame. Lapen neli bidisar Prof. Scott DeLancey,
alangli ke neli kecharli ahamphang along ‘Human Subject’ ahin enut abangphu lapen alangliphan ta
delancey@uoregon.edu mate lamrihing +1-541-346-3901 along arju setame. Lapen anrparta bangso
aketralip aputhak Linguistics ahamphang angtan ta jongsi nangli enut kacherap abangphu/phi
ahinpen kopine karju nangji dothukok pulote ‘Office of Human Subject Compliance, University of
Oregon, Eugene, OR 97403, +1-541-346-2510 along nanglilam pale ik tame.
Nanglimen ladak nangkachethap dunke athak kepachinitang alam an nangchekroidun angthek lo,
lapen nanglimethan ning kehang atengsi nangchedondunlo, lapen ajat ahut tame nanglining hanghe
pulote ajat tu nangnepin pen nangche’o dun unlo, lapen laso kachekroi alo epak nanglirideng dolo,
lapen nangli nangkachekroiduntang alamthe aputhak mo aphi lamcherui avelo. Laso kachekroi alo
nangliphan ta ekenangsot pulote, epak nanglipharman chehang ta choklo.
Labangso a-ketarlip ke National Science Foundation pen kacherap Doctoral Dissertation Research
Improvement Grant # 0951749 akerap pensi kepachun
Niphai______________________
737
22. REFERENCES CITED
Austin, Peter K., and Julia Sallabank, ed. 2011. The Cambridge Handbook of Endangered
Languages. Leiden: Cambridge University Press.
Balawan, Fr. Michael. 1978. A Mikir-English Dictionary with Some Tit-Bits of Mikir
Grammar. Shillong: Khasi Jaintia Presbyterian Press.
Bauman, James. 1976. “An Issue in the Subgrouping of the Tibeto-Burman Languages:
Lepcha and Mikir”. Paper presented at the 9th International Conference on Sino-
Tibetan Languages and Linguistics, October 22-24, Copenhagen.
Bey, Chitro Kumar. 2010. Karbi Alam Ketok Kapacheton: Isi-Hini Aron. Amlapatty,
Diphu, Karbi Anglong, Assam, India: Hiddhinath Rongpi.
Bird, Steven, and Mark Liberman. 2001. “A Formal Framework for Linguistic
Annotation.” Speech Communication, no. 33: 23–60.
Bird, Steven, and Gary Simons. 2003. “Seven Dimensions of Portability for Language
Documentation and Description.” Language 79 (3): 557–82.
Boersma, Paul, and David Weenink. 2013. Praat: Doing Phonetics by Computer (version
5.3.59). http://www.praat.org/.
738
Burling, Robbins. 1983. “The Sal Languages.” Linguistics of the Tibeto-Burman Area 7
(2): 1–32.
———. 2004. The Language of the Modhupur Mandi (Garo). Vol. 1. Bibliophile South
Asia.
———. 2013. “The ‘Sixth’ Vowel in the Boro-Garo Languages.” In North East Indian
Linguistics, 5:271–82. New Delhi: Foundation / Cambridge University Press India.
Bybee, Joan L. 1998. “‘Irrealis’ as a Grammatical Category.” Anthropological Linguistics
40 (2): 257–71.
Bybee, Joan L., Revere D. Perkins, and William Pagliuca. 1994. The Evolution of
G : t M ty t uages of the World. Chicago:
University of Chicago Press.
Chafe, Wallace L. 1980. The Pear Stories: Cognitive, Cultural, and Linguistic Aspects of
Narrative Production. Norwood, NJ: Ablex.
739
De Vries, Lourens. 2005. “Towards a Typology of Tailhead Linkage in Papuan
Languages.” Studies in Language 29 (2): 363–84.
———. 2001. “Lectures on Functional Syntax” presented at the LSA Summer Institute,
UC Santa Barbara. http://darkwing.uoregon.edu/~delancey/sb/fs.html.
———. 2012. “On the Origins of Bodo-Garo.” In North East Indian Linguistics, 4:3–20.
New Delhi: Foundation / Cambridge University Press India.
Demeke, Girma A., and Ronny Meyer. 2008. “The Enclitic-Mm in Amharic:
Reassessment of a Multifunctional Morpheme.” Linguistics 46 (3): 607–28.
Dixon, Robert M. W. 1977. “Where Have All the Adjectives Gone?” Studies in
Language 1 (1): 19–80.
Evans, Nicholas. 2007. “Insubordination and Its Uses.” In Finiteness: Theoretical and
Empirical Foundations, edited by Irina Nikolaeva, 366–431. Oxford; New York:
Oxford University Press.
740
Genetti, Carol. 1986. “The Development of Subordinators from Postpositions in Bodic
Languages.” In Proceedings of the Annual Meeting of the Berkeley Linguistics
Society, 12:387–400. Berkeley, CA: University of California, Berkeley.
———. 2011. “The Tapestry of Dolakha Newar: Chaining, Embedding, and the
Complexity of Sentences.” Linguistic Typology 15 (1): 5–24.
Genetti, Carol, Ellen Bartee, Alec Coupe, Kristine Hildebrandt, and You-Jing Lin. 2008.
“Syntactic Aspects of Nominalization in Five Tibeto-Burman Languages of the
Himalayan Area.” Linguistics of the Tibeto-Burman Area 31 (2): 97–143.
Georg, Stefan. 1996. Marphatan Thakali. LINCOM Studies in Asian Linguistics 02.
Munich/Newcastle: Lincom Europa.
Gippert, Jost, Nikolaus P. Himmelmann, and Ulrike Mosel, ed. 2006. Essentials of
Language Documentation. Berlin: Mouton de Gruyter.
Goddard, Cliff. 2005. The Languages of East and Southeast Asia: An Introduction.
Oxford: Oxford University Press.
Good, Jeff. 2011. “Data and Language Documentation.” In The Cambridge Handbook of
Endangered Languages, edited by Peter Austin and Julia Sallabank, 212–34.
Leiden: Cambridge University Press.
Grierson, George Abraham. 1903. Linguistic Survey of India. Vol. 3, parts 1–3, Tibeto-
Burman family. Calcutta: Superintendent of Government Printing, India.
http://www.joao-roiz.jp/LSI/search.
Grüßner, Karl-Heinz. 1978. Arleng Alam, Die Sprache Der Mikir: Grammatik Und Texte.
Beiträge Zur Südasienforschung 39. Wiesbaden: Franz Steiner Verlag.
———. “Karbi Dictionary”. Tübingen.
741
Guillaume, Antoine. 2011. “Subordinate Clauses, Switch-Reference, and Tail-Head
Linkage in Cavineña Narratives.” In Subordination in Native South-American
Languages, 97:109–39. Typological Studies in Language.
Amsterdam/Philadelphia: John Benjamins.
Harris, Alice C., and Lyle Campbell. 1995. Historical Syntax in Cross-Linguistic
Perspective. Cambridge: Cambridge University Press.
———. 2005. “Nominal and Verbal Conjunction.” In The World Atlas of Language
Structures, edited by Martin Haspelmath, Matthew S. Dryer, David Gil, and
Bernard Comrie, 262–65. Oxford: Oxford University Press.
Heine, Bernd, and Mechthild Reh. 1984. Grammaticalization and Reanalysis in African
Languages. Hamburg: Buske.
Henderson, Eugénie J.A. 1965. Tiddim Chin: A Descriptive Analysis of Two Texts. Vol.
15. London Oriental Series. London: Oxford University Press.
742
———. 2006b. “The Challenge of Segmenting Spoken Language.” In Essentials of
Language Documentation, edited by Jost Gippert, Nikolaus P. Himmelmann, and
Ulrike Mosel, 253–74. Berlin: Mouton de Gruyter.
Hockett, Charles F. 1967. “The Quantification of Functional Load.” Word, no. 23: 320–
39.
Holm, John A. 1988. Pidgins and Creoles: Theory and Structure. Vol. 1. Cambridge:
Cambridge University Press.
———. 2013. “The Kurtöp -Si Construction: Converbs, Clause-Chains and Verb
Serialization.” In Functional-Historical Approaches to Explanation: In Honor of
Scott DeLancey. Vol. 103. Typological Studies in Language. Amsterdam: John
Benjamins.
Jenny, Matthias. 2012. “The Far West of Southeast Asia - ‘Give’ and ‘Get’ in the
Languages of Myanmar”. Paper presented at the Mainland Southeast Asia - state
of the art, November 29 - December 1, 2012, MPI Leipzig.
Jeyapaul, Vethamuthu Yesunesiah. 1987. Karbi Grammar. CIIL Grammar Series 15.
Mysore, India: Central Institute of Indian Languages.
Joseph, U.V. 2009. “The Numeral ‘One’ in Khasi and Karbi.” In North East Indian
Linguistics, 2:149–61. New Delhi: Foundation / Cambridge University Press India.
Joseph, U.V., and Robbins Burling. 2006. The Comparative Phonology of the Boro-Garo
Languages. Manasagangotri, Mysore: Central Institute of Indian Languages.
Kansakar, Tej R., Yogendra P. Yadava, Krishna Prasad Chalise, Balaram Prasain, Dubi
Nanda Dhakal, and Krishna Paudel. 2011. A Grammar of Baram. Linguistic and
Ethnographic Documentatation of the Baram Language, Central Department of
Linguistics, Tribhuvan University.
743
Kaplan, Jeff. 1984. “Obligatory Too in English.” Language, 510–18.
Konnerth, Linda. under review. “Linking Second Person, Speech Act Participant
Indexation, and Cislocative: Evidence from Karbi (Tibeto-Burman)”
———. 2009. “The Nominalizing Prefix *gV- in Tibeto-Burman”. M.A. thesis, Eugene,
OR: University of Oregon.
744
Kouwenberg, Silvia, and Pieter Muysken. 1995. “Papiamento.” In Pidgins and Creoles:
An Introduction, edited by Jacques Arends, Pieter Muysken, and Norval Smith,
15:205–18. Creole Language Library. Amsterdam/Philadelphia: John Benjamins.
Krifka, Manfred. 1998. “Additive Particles under Stress.” In Proceedings of SALT, edited
by D. Strolovitch and A. Lawson, 8:111–28. Ithaca, NY: CLC Publications.
Kro, Bidor Sing, ed. 2009. Akimi Karbi Lamthe Amarjong. Second edition. Rongmongve,
Diphu, Karbi Anglong, Assam, India: Kangthir Book House and Publisher (Bidya
Sing Rongpi).
Lewis, Paul M., Gary F. Simons, and Charles D. Fennig, ed. 2013. Ethnologue:
Languages of the World. Seventeenth edition. Dallas, Texas: SIL International.
http://www.ethnologue.com.
Li, Charles N., and Sandra A. Thompson. 1981. Mandarin Chinese: A Functional
Reference Grammar. Berkeley and Los Angeles, CA: University of California
Press.
Mariae, Fr. John. 2007. Karbi Self-Taught. Second edition. Diphu, Karbi Anglong,
Assam, India: Don Bosco Catholic Literature Committee.
———. 1973. The Grammar of Lahu. Vol. 75. Berkeley, CA: University of California
Press.
745
———. 1986. “Hearts and Minds in South-East Asian Languages and English: An Essay
in the Comparative Lexical Semantics of Psycho-Collocations.” Cahiers de
Linguistique Asie Orientale 15 (1): 5–57.
Mauri, Caterina, and Andrea Sans . 2012. “What Do Languages Encode When They
Encode Reality Status?” Language Sciences 34 (2): 99–106.
Mithun, Marianne. 1995. “On the Relativity of Irreality.” In Modality in Grammar and
Discourse, edited by Joan L. Bybee and Suzanne Fleischman, 367–88.
Amsterdam/Philadelphia: John Benjamins.
Ohala, John J. 2010. “The Relation between Phonetics and Phonology.” In The Handbook
of Phonetic Sciences, edited by W. J. Hardcastle, J. Laver, and F. E. Gibbon, 2nd
ed., 653–77. Oxford: Wiley-Blackwell.
Öpengin, Ergin. 2013. “Topicalisation in Central Kurdish: Additive Enclitic and Other
Means”. Paper presented at the Information Structure in Spoken Language
Corpora (ISSLaC), Bielefeld, Germany.
Overstreet, Maryann. 1999. Whales, Candlelight, and Stuff like That: General Extenders
in English Discourse. Oxford: Oxford University Press.
746
———. 2010. “Khumi Elaborate Expressions.” Himalayan Linguistics 9 (1): 81–100.
Peterson, David A., and Kenneth VanBik. 2004. “Coordination in Hakha Lai (Tibeto-
Burman).” In Coordinating Constructions, edited by Martin Haspelmath, 58:333–
56. Typological Studies in Language. Amsterdam/Philadelphia: John Benjamins.
———. 2009. “Predicate Derivations in the Tani Languages: Root, Suffix, Both or
Neither?” In North East Indian Linguistics, 2:175–97. New Delhi: Foundation /
Cambridge University Press India.
Robinson, William. 1849. “Notes on the Languages Spoken by the Various Tribes
Inhabiting the Valley of Assam and Its Mountain Confines. Part II.” Journal of
the Asiatic Society of Bengal, no. 18: 310–49.
Scott, James C. 2009. The Art of Not Being Governed: An Anarchist History of Upland
Southeast Asia. New Haven: Yale University Press.
So-Hartmann, Helga. 2009. A Descriptive Grammar of Daai Chin. Berkeley, CA: Sino-
Tibetan Etymological Dictionary and Thesaurus Project.
Stack, Edward, and Sir Charles Lyall. 1908. The Mikirs. London: David Nutt.
747
Starosta, Stanley. 1985. “Relator Nouns as a Source of Case Inflection.” Oceanic
Linguistics Special Publications, no. 20: 111–33.
Stern, Theodore. 1984. “Sizang (Siyin) Chin Texts.” Linguistics of the Tibeto-Burman
Area 8 (1): 43–58.
Stewart, R. 1855. “Notes on Northern Cachar.” Journal of the Asiatic Society of Bengal,
no. 24: 656–75.
Sun, Jackson TS. 1993. “The Linguistic Position of Tani (Mirish) in Tibeto-Burman: A
Lexical Assessment.” Linguistics of the Tibeto-Burman Area 16 (2): 143–88.
Surendran, Dinoj, and Partha Niyogi. 2006. “Quantifying the Functional Load of
Phonemic Oppositions, Distinctive Features, and Suprasegmentals.” Amsterdam
Studies in the Theory and History of Linguistic Science Series 4 279: 43.
Taro, Sarklim, ed. 2010. Karbi-English Dictionary. Guwahati, Assam, India: Assam
Institute of Research for Tribals and Scheduled Castes.
———. , ed. 2012. Karbi Studies: Memories, Myths, Metaphors. Second edition. Vol. 1.
Guwahati, Assam, India: Assam Book Hive.
Teron, Longkam. 2005a. Karbi Lamtasam. Second edition. Rontheang, Diphu, Karbi
Anglong, Assam, India: Karbi Lammet Amei.
———. 2005b. Sar-Lamthe. Second edition. Rontheang, Diphu, Karbi Anglong, Assam,
India: Karbi Lammet Amei.
———. 2006. Karbi Lambeng. Rontheang, Diphu, Karbi Anglong, Assam, India: Karbi
Lammet Amei.
———. , ed. 2008. Karbi Lamlir Achili. Second edition. Diphu, Karbi Anglong, Assam,
India: Longkam Teron.
Teron, Manik, and Rajendra Tumung, ed. 2007. Karbi Lamarjong. Diphu, Karbi Anglong,
Assam, India: Kamrup Karbi Lammet Amei (Longki Phangcho).
748
Thurgood, Graham. 1983. “Morphological Innovation and Subgrouping: Some Tibeto-
Burman Notes.” In Proceedings of the Annual Meeting of the Berkeley Linguistics
Society, 9:257–65. Berkeley, CA: Berkeley Linguistics Society.
Tomlin, Russell S. 1995. “Focal Attention, Voice, and Word Order: An Experimental,
Cross-Linguistic Study.” In Word Order in Discourse, edited by Pamela A.
Downing and Michael Noonan, 517–54. Amsterdam/Philadelphia: John
Benjamins.
Tosco, Mauro. 2010. “Why Contrast Matters: Information Structure in Gawwada (East
Cushitic).” In The Expression of Information Structure: A Documentation of Its
Diversity across Africa, edited by Ines Fiedler and Anne Schwarz, 91:315–48.
Typological Studies in Language. Amsterdam/Philadelphia: John Benjamins.
Van Driem, George. 1991. “Tangut Verbal Agreement and the Patient Category in
Tibeto-Burman.” Bulletin of the School of Oriental and African Studies 54 (3):
520–33.
Van Putten, Saskia. 2011. “Marked Topics and Contrast in Avatime”. Paper presented at
the Information structure and spoken language: cross-linguistic comparative
studies, Boulder, CO.
Walker, George David. 1925. A Dictionary of the Mikir Language. Shillong: Assam
Government Press.
Yap, Foong Ha, Karen Grunow-Hårsta, and Janick Wrona, ed. 2011. Nominalization in
Asian Languages: Diachronic and Typological Perspectives.
Amsterdam/Philadelphia: John Benjamins.
749