The Language of Rotokas, Papua New GUINEA

Split Intransitivity in Rotokas,
a Papuan Language of Bougainville

c 2011, Stuart Robinson
Cover design: Harald Pieper, In-Zicht Graphic Design, Nijmegen
Typeset with L
A
T
E
X
Printed and bound by Ipskamp Printers, Nijmegen
Split Intransitivity in Rotokas,
a Papuan Language of Bougainville
Een wetenschappelijke proeve
op het gebied van de Letteren
Proefschrift
ter verkrijging van de graad van doctor
aan de Radboud Universiteit Nijmegen
op gezag van de rector magnicus prof. mr. S.C.J.J. Kortmann,
volgens besluit van het college van decanen
in het openbaar te verdedigen op dinsdag 5 april 2011
om 15.30 uur precies
door
Stuart Payton Robinson
geboren op 28 oktober 1973 Bad Cannstatt (Duitsland)
Promotor: Prof. dr. Stephen C. Levinson
Copromotor: Dr. Michael Dunn (Max Planck Institute for Psycholinguistics, Nijmegen)
Manuscriptcommissie:
Prof. dr. Leon Stassen
Prof. dr. Ulrike Mosel (Christian-Albrechts-Universit at zu Kiel)
Prof. dr. Robert van Valin (University at Buffalo, The State University of New York)
The research reported in this thesis was supported by a grant from the Max-Planck-Gesellschaft
zur F orderung der Wissenschaften, M unchen, Germany.
Acknowledgements
This dissertation would never have been realized without the generous support of numerous
individuals and organizations.
First and foremost, I must acknowledge the the generous funding of the Max Planck Society
and the institutional support of the Max Planck Institute for Pyscholinguistics in Nijmegen, The
Netherlands. I also should thank Penelope Brown and Stephen Levinson for encouraging my
return to linguistics after a three-year long hiatus as a software engineer. Although building web
sites for business-to-business automative resale and for personal wealth management taught me
a great deal about the art and craft of programming, my intellectual life has been more fullling
since my return to the linguistic sciences.
I also wish to thank the various individuals who read drafts of this dissertation (in part or
whole) and provided feedback: Michael Dunn, Nick Eneld, Tracy King, Stephen Levinson,
Ger Reesink, Gunter Senft, and Angela Terrill. Needless to say, I am solely responsible for its
shortcomings. Special thanks go to Saskia van Putten and Annemarie Verkerk, who translated
my samenvatting from English, and to Edith Sjoerdsma, whose patient assistance with the nal
bureaucratic hurdles of dissertation publication was an absolute lifesaver. Hartelijk bedankt.
I owe a debt to numerous individuals in Papua New Guinea, without whom the eldwork
underlying this work would have been impossible. First and foremost, I would like to thank
Samuel Akoitai for providing accomodations in Togarao and to the rest of his family for their
hospitality, especially his brothers, Samson, Paul, and Thomas Akoitai, who made Togarao feel
like a home away from home. Second, this work would not have been possible without the
assistance of the Rotokas-speaking community. Timothy Taureviri and Sera Mon worked full-
time as language consultants, and their assistance in the transcription and analysis of Rotokas
was invaluable. Vearovira rutu! Finally, the friendly support of the staff at the National Research
Institute in Port Moresby deserves mention, especially that of James Robins.
My various ofcemates over the last few years also deserve acknowledgement for putting
up with my ranting and raving (usually directed at my computer when it refused to cooperate
but also sometimes over the frustrations of life as an expatriate in Nijmegen): rst and foremost,
Claudia Wegener, who shared an ofce with me during the majority of my time at the MPI, but
also Sylvia Tufvesson and Clair Hill.
I also want thank my family for their support throughout the entire Ph.D. process. It was
an eventful few years for my mother, Marsha; my father, Michal; and my sister, Emily. I wish
I could have been a less distant presence. My adopted family (Mike, Surinder, Cindy, Annup,
and Sanjay) has been incredibly supportive. My wife, Rupe Singh, is the most deserving of
vii
acknowledgement for her love, support, and forebearance during a hectic four years divided
between three continents. But it is my deceased grandfather, George Delbert Middleton, to
whom I wish to dedicate this work. His was a virtuous life of a service to family, community,
and country, and I am deeply saddened that he did not live to see this work come to fruition.
Contents
I A Grammatical Sketch of Rotokas 1
1 Introduction 3
1.1 Aims and Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Fieldwork and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Language Background 9
2.1 Bougainville and Island Melanesia . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.1 History of the Region . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 The Rotokas Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Prior Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.2 The Speakers of Rotokas . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.3 Dialectal Variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.4 The Languages of Bougainville . . . . . . . . . . . . . . . . . . . . . 16
2.2.5 Genetic Afliation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3 Phonology 25
3.1 Segmental Phonology (Phoneme Inventory) . . . . . . . . . . . . . . . . . . . 25
3.1.1 Vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.2 Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.1.3 A Practical Orthography . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Suprasegmental Phonology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1 Syllable Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.2 Word Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.3 Reduplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4 Word Classes 37
4.1 Root, Stem, and Word Classes . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 An Inventory of Rotokas Word Classes . . . . . . . . . . . . . . . . . . . . . . 39
ix
4.2.1 Nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2.1.1 Gender and Noun Subclasses . . . . . . . . . . . . . . . . . 40
4.2.1.1.1 Class 1 . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.1.1.2 Class 2 . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2.1.1.3 Class 3 . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.1.1.4 Class 4 . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.1.1.5 Class 5 . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2.2 Noun Classiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2.3 Pronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.2.3.1 Personal Pronouns . . . . . . . . . . . . . . . . . . . . . . . 53
4.2.3.1.1 S . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.3.1.2 A . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.3.1.3 O . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.3.1.4 Oblique . . . . . . . . . . . . . . . . . . . . . . . 56
4.2.3.2 Resumptive Pronouns . . . . . . . . . . . . . . . . . . . . . 56
4.2.3.3 Possessive Pronouns . . . . . . . . . . . . . . . . . . . . . . 58
4.2.3.4 Demonstratives . . . . . . . . . . . . . . . . . . . . . . . . 59
4.2.4 Verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.2.5 Adjectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.2.5.1 Predication . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.2.5.1.1 Event Predicate . . . . . . . . . . . . . . . . . . . 63
4.2.5.1.2 Property or Quality Predicate . . . . . . . . . . . . 64
4.2.5.1.3 Locational Predicate . . . . . . . . . . . . . . . . . 64
4.2.5.2 Attribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.2.6 Adverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.2.7 Postpositions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.2.8 Interrogatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.2.9 Conjoiners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.2.10 Exclamatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5 Morphology 79
5.1 Nominal Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.1.1 Prexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.1.1.1 Order 1 Prex: Reexive/Reciprocal Marker . . . . . . . . . 80
5.1.1.1.1 Reexive . . . . . . . . . . . . . . . . . . . . . . 80
5.1.1.1.2 Reciprocal . . . . . . . . . . . . . . . . . . . . . . 81
5.1.1.1.3 Emphatic/Contrastive . . . . . . . . . . . . . . . . 81
5.1.1.2 Order 1 Prex: Specier . . . . . . . . . . . . . . . . . . . . 81
5.1.1.2.1 Specier with Noun . . . . . . . . . . . . . . . . . 81
x
5.1.1.2.2 Specier with Classier . . . . . . . . . . . . . . . 81
5.1.2 Sufxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.1.2.1 Order 1 Sufxes: Derivational . . . . . . . . . . . . . . . . . 82
5.1.2.2 Order 2 Sufxes: Number/Gender . . . . . . . . . . . . . . 84
5.1.2.3 Order 3 Sufxes: Possession . . . . . . . . . . . . . . . . . 84
5.1.2.4 Order 4 Sufxes: Diminutive . . . . . . . . . . . . . . . . . 84
5.1.2.5 Order 5 Sufxes: Alternative . . . . . . . . . . . . . . . . . 85
5.1.2.6 Order 6 Sufxes: Denite/Indenite . . . . . . . . . . . . . 89
5.1.2.6.1 Indenite Sufx . . . . . . . . . . . . . . . . . . . 89
5.1.2.6.2 Denite Sufx . . . . . . . . . . . . . . . . . . . . 89
5.1.3 Enclitics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.1.3.1 Postpositional Enclitics . . . . . . . . . . . . . . . . . . . . 90
5.1.3.1.1 Locative/Instrumental . . . . . . . . . . . . . . . . 90
5.1.3.1.2 Allative . . . . . . . . . . . . . . . . . . . . . . . 91
5.1.3.1.3 Benefactive . . . . . . . . . . . . . . . . . . . . . 91
5.1.3.1.4 Ablative . . . . . . . . . . . . . . . . . . . . . . . 91
5.1.3.2 Topic Marker . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2 Verbal Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.2.1 Prexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.2.1.1 Order 1 Prex: Reexive/Recriprocal . . . . . . . . . . . . . 93
5.2.2 Sufxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.2.2.1 Order 1 Sufxes: Causative . . . . . . . . . . . . . . . . . . 94
5.2.2.2 Order 2 Sufxes: Modiers . . . . . . . . . . . . . . . . . . 94
5.2.2.2.1 -raga only/just . . . . . . . . . . . . . . . . . . 94
5.2.2.2.2 -irao really . . . . . . . . . . . . . . . . . . . . 95
5.2.2.3 Order 3 Sufxes: Continuous . . . . . . . . . . . . . . . . . 96
5.2.2.4 Order 4 Sufxes: Resultative Sufx . . . . . . . . . . . . . . 97
5.2.2.5 Order 5 Sufxes: Dependent Verb Morphology . . . . . . . . 98
5.2.2.6 Order 5 Sufxes: Person/Number/Gender . . . . . . . . . . 99
5.2.2.6.1 Person/Number/Gender . . . . . . . . . . . . . . . 99
5.2.2.7 Order 6 Sufxes: Tense/Mood . . . . . . . . . . . . . . . . . 101
5.2.2.7.1 Realis . . . . . . . . . . . . . . . . . . . . . . . . 102
5.2.2.7.2 Irrealis . . . . . . . . . . . . . . . . . . . . . . . . 106
5.2.2.7.3 -peira . . . . . . . . . . . . . . . . . . . . . . . . 111
5.2.2.7.4 -veira . . . . . . . . . . . . . . . . . . . . . . . . 111
5.3 Reduplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.4 Morphophonemics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.4.1 Identical Vowel Sequences . . . . . . . . . . . . . . . . . . . . . . . . 113
xi
5.4.2 Deletion and Insertion Rules . . . . . . . . . . . . . . . . . . . . . . . 113
5.4.2.1 o-deletion . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.4.2.2 e-deletion . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.4.2.3 i-insertion . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
6 Syntax 119
6.1 Noun Phrases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.1.1 Nominals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.1.1.1 Demonstrative-Noun . . . . . . . . . . . . . . . . . . . . . 120
6.1.1.2 Possessor-Noun . . . . . . . . . . . . . . . . . . . . . . . . 120
6.1.1.3 Adjective-Noun . . . . . . . . . . . . . . . . . . . . . . . . 120
6.1.1.4 Numeral-Noun . . . . . . . . . . . . . . . . . . . . . . . . . 120
6.1.1.5 Noun-Classier . . . . . . . . . . . . . . . . . . . . . . . . 120
6.1.1.6 Noun-Possessive Pronoun . . . . . . . . . . . . . . . . . . . 120
6.1.1.7 Noun-Relative Clause . . . . . . . . . . . . . . . . . . . . . 120
6.1.2 Possession . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.1.2.1 Possessive Pronoun . . . . . . . . . . . . . . . . . . . . . . 121
6.1.2.2 Possession Marking on Possessed Noun . . . . . . . . . . . 121
6.1.2.3 Ownership . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6.1.2.4 Inalienable Possession (Body Parts) . . . . . . . . . . . . . . 122
6.1.2.5 Kinship . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6.1.2.6 Possession Marking on Dummy Pronoun . . . . . . . . . . . 123
6.1.3 Quantication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.1.3.1 Quantiers . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.1.3.2 Numerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.1.4 Nominal Conjunction . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.2 Intraclausal Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6.2.1 Constituent Order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6.2.2 Displacement of O . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
6.2.3 Interrogatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
6.2.4 Negation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
6.2.4.1 Noun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.2.4.2 Pronoun . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.3 Interclausal Syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.3.1 Complementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6.3.2 Verb Phrases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
6.3.2.1 Dependent Verbs . . . . . . . . . . . . . . . . . . . . . . . . 138
6.3.2.2 Aspectual Verbs . . . . . . . . . . . . . . . . . . . . . . . . 139
6.3.3 Coordination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
xii
6.3.3.1 Coordination of NPs . . . . . . . . . . . . . . . . . . . . . . 140
6.3.3.2 Coordination of Obliques . . . . . . . . . . . . . . . . . . . 141
6.3.3.3 Coordination of Adverbs . . . . . . . . . . . . . . . . . . . 141
6.3.3.4 Coordination of Temporal Nouns . . . . . . . . . . . . . . . 141
6.3.3.5 Coordination of Dependent Verbs . . . . . . . . . . . . . . . 141
II Verb Classes in Rotokas 145
7 Verb Classes in Rotokas 147
7.1 Two Verb Classes: and . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
7.1.1 Morphological Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . 147
7.1.1.1 Verbal Subject Agreement . . . . . . . . . . . . . . . . . . . 148
7.1.1.2 Tense/Aspect/Mood . . . . . . . . . . . . . . . . . . . . . . 150
7.2 Firchows Problem: What is the / Distinction? . . . . . . . . . . . . . . . . 153
7.3 Transitivity and Valency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
7.3.1 Transitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
7.3.2 Valency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
7.4 First Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
8 Valency in Rotokas 161
8.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
8.2 Monovalent One-Place Predicates . . . . . . . . . . . . . . . . . . . . . . . . 166
8.2.1 Agreement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
8.3 Monovalent Two-Place Predicates . . . . . . . . . . . . . . . . . . . . . . . . 170
8.3.1 Agreement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
8.3.1.1 -Agreement . . . . . . . . . . . . . . . . . . . . . . . . . 171
8.3.1.2 -Agreement . . . . . . . . . . . . . . . . . . . . . . . . . . 172
8.3.3 Oblique Marking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
8.3.3.1 Verbs that Select the Enclitic =ia . . . . . . . . . . . . . . . 174
8.3.3.2 Verbs that Select the Enclitic =re . . . . . . . . . . . . . . . 175
8.3.3.3 Verbs that Select the Enclitic =pa . . . . . . . . . . . . . . . 175
8.3.3.4 Verbs that Select the Enclitic =va . . . . . . . . . . . . . . . 176
8.4 Bivalent Two-Place Predicates . . . . . . . . . . . . . . . . . . . . . . . . . . 177
8.4.1 Agreement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
8.5 Three-Place Predicates: Bivalent or Trivalent? . . . . . . . . . . . . . . . . . . 178
xiii
8.5.1 Agreement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
8.5.3 Indirect Object Construction . . . . . . . . . . . . . . . . . . . . . . . 180
8.5.4 Double Object Construction . . . . . . . . . . . . . . . . . . . . . . . 181
8.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
9 Valency-Changing Derivations 185
9.1 Valency-Increasing Derivations . . . . . . . . . . . . . . . . . . . . . . . . . . 185
9.1.1 Ambivalent Verb Roots . . . . . . . . . . . . . . . . . . . . . . . . . . 185
9.1.1.1 Labile Ambivalent Verb Roots . . . . . . . . . . . . . . . . 186
9.1.1.2 Fixed Ambivalent Verb Roots . . . . . . . . . . . . . . . . . 187
9.1.2 Morphological Causative . . . . . . . . . . . . . . . . . . . . . . . . . 188
9.2 Valency-Decreasing Derivations . . . . . . . . . . . . . . . . . . . . . . . . . 193
9.2.1 Reexives/Reciprocals . . . . . . . . . . . . . . . . . . . . . . . . . . 193
9.2.1.1 Construction 1: Verb Marking . . . . . . . . . . . . . . . . . 193
9.2.1.2 Construction 2: Pronoun Marking . . . . . . . . . . . . . . . 195
9.2.1.3 Construction 3: Noun Marking . . . . . . . . . . . . . . . . 196
9.2.1.3.1 Incorporation . . . . . . . . . . . . . . . . . . . . 197
9.2.1.3.2 Reexive/Reciprocal . . . . . . . . . . . . . . . . 197
9.2.2 Noun Incorporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
9.2.2.1 Incoporating Verbs Show Agreement . . . . . . . . . . . . 200
9.2.2.2 No Intervening Material Between Verb and Incorporated Noun 201
9.2.2.3 No Morphology or Oblique Marking on Incorporated Nouns 202
9.2.3 Resultatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
9.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
10 The Semantics of Split Intransitivity in Rotokas 209
10.1 Semantic Classes of Split Intransitivity . . . . . . . . . . . . . . . . . . . . . . 209
10.1.1 Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
10.1.2 Bodily Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
10.1.3 Sound Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
10.1.4 Inferred Causation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
10.1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
10.2 Thematic Roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
10.2.1 Agent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
10.2.2 Theme/Patient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
10.2.3 Experiencer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
10.2.4 Source and Goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
10.2.5 Recipient/Addressee . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
xiv
10.2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
10.3 Split Intransitivity from a Theoretical Perspective . . . . . . . . . . . . . . . . 230
10.3.1 Syntactic Accounts of Split Intransitivity . . . . . . . . . . . . . . . . 232
10.3.2 Semantic Accounts of Split Intransitivity . . . . . . . . . . . . . . . . 236
10.3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
11 Conclusion 245
11.1 Directions for Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . 245
11.1.1 Tense/Aspect/Mood . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
11.1.2 Loan Verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
11.1.3 Comparative Evidence . . . . . . . . . . . . . . . . . . . . . . . . . . 247
11.2 Typological Implications of Rotokas . . . . . . . . . . . . . . . . . . . . . . . 249
11.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
A An Inventory of Verb Classes 253
A.1 1 [SUB] (Total: 385) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
A.2 1 [SUB] (Total: 66) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
A.3 2 [SUB, OBL] (Total: 55) . . . . . . . . . . . . . . . . . . . . . . . . . 259
A.4 2 [SUB, OBL] (Total: 35) . . . . . . . . . . . . . . . . . . . . . . . . . 260
A.5 2 [SUB, O OBL] (Total: 5) . . . . . . . . . . . . . . . . . . . . . . . . 261
A.6 2 [SUB, O] (Total: 482) . . . . . . . . . . . . . . . . . . . . . . . . . . 261
B A Finite State Transducer for Rotokas Morphology 269
B.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
B.2 Source Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
C Sample Texts 281
C.1 Matevu, Version 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
C.2 Matevu, Version 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Bibliography 305
Morpheme Index 320
Samenvatting (Dutch summary) 325
Curriculum Vitae 328
xv
xvi
List of Tables
1.1 Fieldwork Dates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Authors Rotokas Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 Prior Literature on and in Rotokas . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Where Rotokas Dialects Are Spoken (Allen and Hurd, 1963) . . . . . . . . . . 21
2.3 The Languages of Bougainville . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Papuan Languages of Bougainville (Allen and Hurd, 1963; Tryon, 2005) . . . . 23
2.5 Cognate Percentage in the Languages of Bougainville (Allen and Hurd, 1963:21) 23
3.1 Five Most Common Vowels in UPSID . . . . . . . . . . . . . . . . . . . . . . 26
3.2 Minimal Pairs for Vowel Length . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 Rotokas Consonants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4 Allophonic Variants of Rotokas Consonants . . . . . . . . . . . . . . . . . . . 27
3.5 Minimal Pairs for Voicing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.6 Practical Orthography for Rotokas . . . . . . . . . . . . . . . . . . . . . . . . 29
3.7 Rotokas Syllable Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.8 Bisyllabic Words Stressed on the Second Syllable . . . . . . . . . . . . . . . . 32
3.9 Trochaic Feet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.10 Metrical Structure for Bisyllabic and Trisyllabic Words . . . . . . . . . . . . . 33
3.11 Metrical Structure of HL and HH Words . . . . . . . . . . . . . . . . . . . . . 34
3.12 Metrical Structure of HL, HH, and LH Stems . . . . . . . . . . . . . . . . . . 36
3.13 Metrical Structure of Reduplicated HL, HH, and LH Stems . . . . . . . . . . . 36
4.1 Rotokas Nouns Borrowed from Tok Pisin . . . . . . . . . . . . . . . . . . . . 40
4.2 Noun Classes and Number Inection . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Class 1 Nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.4 Class 2 Nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.5 Class 3 Nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.6 Class 4 Nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.7 Class 5 Nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
xvii
4.8 Shaped-Based Noun Classiers (Firchow, 1987:36) . . . . . . . . . . . . . . . 50
4.9 Noun Classiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.10 Personal Pronoun Paradigm . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.11 Resumptive Pronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.12 Possessive Pronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.13 Demonstrative Pronoun Paradigm . . . . . . . . . . . . . . . . . . . . . . . . 60
4.14 Hypothetical Segmentation of Demonstrative Pronoun Paradigm . . . . . . . . 60
4.15 Rotokas Verbs Borrowed from Tok Pisin . . . . . . . . . . . . . . . . . . . . 62
4.16 Rotokas Time Words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.17 Rotokas Postpositions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.18 Interrogatives in Rotokas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.19 Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.20 Exclamatives in Rotokas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.1 Dependent Verb Marking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.2 Subject Agreement Sufxes . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.3 Rotokas Tense Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.4 Segmentation of Realis Sufxes . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.5 Rotokas Irrealis Mood Categories . . . . . . . . . . . . . . . . . . . . . . . . 106
5.6 Segmentation of Irrealis Sufxes . . . . . . . . . . . . . . . . . . . . . . . . . 107
6.1 Elements of the Noun Phrase . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.2 Rotokas Numerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
7.1 Realis Tense Markers By Verb Classication ( vs. ) . . . . . . . . . . . . . 150
7.2 Individuation of O: Relevant Features . . . . . . . . . . . . . . . . . . . . . . 156
7.3 Inventory of Grammatical Functions of the NP . . . . . . . . . . . . . . . . . . 158
7.4 Grammatical Roles and Verb Inection . . . . . . . . . . . . . . . . . . . . . 159
8.1 Predicate Types, Valency, and Subject Agreement in Rotokas . . . . . . . . . . 162
8.2 Some Monovalent Verb Roots that Show Inection . . . . . . . . . . . . . . 167
8.3 Some Monovalent Verb Roots with Subject Agreement . . . . . . . . . . . . 168
8.4 Illustrative Monovalent Two-Place Predicates with Inection . . . . . . . . . 171
8.5 Verbal Agreement and Oblique Marking in Monovalent Verb Roots . . . . . . 173
8.6 Verbs that Select the Postpositional Enclitic =ia . . . . . . . . . . . . . . . . . 174
8.7 Verbs that Select the Postpositional Enclitic =re . . . . . . . . . . . . . . . . . 175
8.8 Verb Roots that Select the Postpositional Enclitic =pa . . . . . . . . . . . . . . 176
8.9 Verb Roots that Select for the Postpositional Enclitic =va . . . . . . . . . . . . 177
8.10 Relationship Between Predicate Type and Valency . . . . . . . . . . . . . . . . 182
8.11 Relationship Between Valency and Verbal Inection . . . . . . . . . . . . . . . 183
xviii
8.12 Grammatical Roles and Verb Inection . . . . . . . . . . . . . . . . . . . . . 183
9.1 Labile Ambivalent Verb Roots . . . . . . . . . . . . . . . . . . . . . . . . . . 187
9.2 Morphological Causatives Derived From and Monovalent Verb Roots . . . 189
9.3 Examples of Causative Sufx -pie . . . . . . . . . . . . . . . . . . . . . . . . 190
9.4 Mappings of Arguments in Causatives Derived from Bivalent Verb Stems . . . 191
9.5 Monovalent Verb Roots Capable of Occuring with the Reexive/Reciprocal
Marker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
9.6 Properties Relevant to the Individuation of O (Hopper and Thompson, 1980:253) 199
9.7 Relationship Between Verb Root Classes and Valency-Changing Derivations . . 207
10.1 Monovalent Verb Roots of Motion in Rotokas . . . . . . . . . . . . . . . . . . 211
10.2 Bodily Process Verbs in Rotokas . . . . . . . . . . . . . . . . . . . . . . . . . 213
10.3 Sound Emission Verbs in Rotokas . . . . . . . . . . . . . . . . . . . . . . . . 215
10.4 Monovalent Verbs of Internal Causation . . . . . . . . . . . . . . . . . . . . 216
10.5 From Verb-Specic Semantic Roles to Grammatical Relations: Adapted for
Rotokas from Van Valin Jr. and Wilkins (1996) and Van Valin Jr. (2005) . . . . 219
10.6 Alignment Possibilities for S, A, and O . . . . . . . . . . . . . . . . . . . . . 230
10.7 Analysis of Verb Classes By Semantic Features . . . . . . . . . . . . . . . . . 238
10.8 RRG Denitions of Thematic Roles . . . . . . . . . . . . . . . . . . . . . . . 240
10.9 Revised Macro-Role Hierarchy from Foley (2005) . . . . . . . . . . . . . . . . 243
10.10The Auxiliary Selection Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . 243
11.1 Grammatical Phenomena Associated with and Inection . . . . . . . . . . 249
B.1 Coverage for Shoebox Dictionary Example Sentences . . . . . . . . . . . . . . 270
xix
xx
List of Figures
1.1 Rotokas Native Speaker Consultants: Timothy Taureviri (left) and Sera Mon
(right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Irwin Firchow (left), David Akoitai (middle), and Caleb Karuru (right) . . . . . 6
2.1 Languages of Island Melanesia . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Dialects of Rotokas [based on Allen and Hurd (1963)] . . . . . . . . . . . . . 15
2.3 Languages of Bougainville . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Shared Vocabulary Percentages of Allen and Hurd (1963) as Neighbor Joining
Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5 Bougainville Branch of the East Papuan Phylum (Wurm, 1972, 1975a) . . . . . 19
3.1 Word length in Rotokas Lexicon . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.1 Breakdown of Words from (18) . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2 Paradigmatic Structure of Personal Pronouns . . . . . . . . . . . . . . . . . . 55
4.3 Postpositional Enclitics and Their Associated Semantic Roles . . . . . . . . . . 77
5.1 Nominal Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.3 Paradigmatic Structure for Person Marking in Personal Pronouns versus Verbal
Agreement Sufxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.2 Verbal Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
7.1 Subject Agreement Sufxes . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
7.2 Semantic Parameters of Transitivity (Hopper and Thompson, 1980:252) . . . . 155
8.1 Distribution of Valency Classes (left) and Inection Classes (right) in Rotokas
Lexicon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
xxi
Abbreviations
The abbreviations and glossing conventions used here are consistent with the guidelines set out
in Lehmann (1983).
1 1st Person
2 2nd Person
3 3rd Person
ABS Absolute
ABL Ablative
ADV Adverbial
ALL Allative
ALT Alternate
ANIM Animate
ANTIC Anticipatory
BEN Benefactive
CL Classier
CMPL Completive
COMP Complementizer
CONT Continuous
DELIM Delimiter
DEM Demonstrative
DEP Dependent (Verb)
DERIV Derivational
DF Distant Future
DL Dual
DP Distant Past
EPEN Epenthetical
EXCL Exclusive
EXCLAM Exclamative
F Feminine
FP Free Pluralizer
HAB Habitual
HUM Human
INANIM Inanimate
INCL Inclusive
INDEF Indenite
INTEN Intensier
IP Immediate Past
LOC Locative
M Masculine
MED Medial
INDEF Non-Absolute
NEG Negation
NF Near Future
NP Near Present
N Neuter
PL Plural
POSS Possession
POST Postposition
PRES Present
PROX Proximal
PRO Pronoun
PPRO Personal Pronoun
RES Resumptive
PURP Purposive
RDP Reduplicated
RP Remote Past
RP Resumptive Pronoun
REF Reexive
REL Relativizer
RP Remote Past
RPRO Resumptive Pronoun
RR Reexive/Reciprocal
SG Singular
xxiii
SIM Simultaneity
SUB Subjunctive
xxiv
Part I
A Grammatical Sketch of Rotokas
1
Chapter 1
Introduction
1.1 Aims and Focus
The aim of this thesis is two-fold. First, it provides a reasonably comprehensive grammar of Ro-
tokas, a Papuan (Non-Austronesian) language of Bougainville, Papua New Guinea. Although
some grammatical description of Rotokas does exist (see 2.2.1 for a complete inventory), it
is scattered across numerous smaller publications and can be difcult to follow. Second, this
thesis focuses on a particular area of Rotokas grammar that poses challenges for grammatical
theory: the nature of verbal inectionmore specically, the existence of two mutually ex-
clusive inectional classes for subject agreement and tense/mood marking. Various aspects of
the morphosyntax of Rotokas will be investigated toward the eventual conclusion that Rotokas
possesses a typologically interesting form of split intransitivity. The nature of split intransitivity
in Rotokas has implications for theories concerning split intransitivity more specically and for
theories of transitivity, valency, and the semantics-syntax interface more generally.
1.2 Fieldwork and Data
This thesis is based on materials obtained during four eldwork trips to Bougainville during the
course of a three-year Ph.D fellowship at the Max Planck Institute in Nijmegen, The Nether-
lands. The dates during which these eldwork trips took place are provided below in Table
1.1.
3
Trip Start End
First 7 April 2003 16 July 2003
Second 27 February 2004 3 May 2004
Third 21 June 2004 29 August 2004
Fourth 12 June 2005 06 October 2005
Table 1.1: Fieldwork Dates
A preliminary eldwork trip was made in 2002 by Ger Reesink, who surveyed the status
of the Papuan languages spoken on Bougainville, established a number of contacts, and made
recommendations for potential eldwork sites. During my rst trip to Papua New Guinea, I
followed up on these contacts and in Port Moresby met with the Minister of Parliament for
the central district of Bougainville, the Honorable Samuel Akoitai, who is a native-speaker
of Rotokas and whose father worked with missionary linguists for a number of years (most
intensively during the sixties, but also during the seventies and eighties). He contacted various
members of the Rotokas-speaking community in Bougainville and made arrangements for an
extended stay in his home village of Togarao, a Rotokas-speaking village in the mountains of
central Bougainville with a few hundred inhabitants.
1
It is located in the Wakunai District,
approximately 25 kilometers inland from Wakunai Station (the main access point for the feeder
road that leads up into the mountains).
The choice of Togarao as a eld site was motivated by a number of considerations. One of
these was continuity. Since prior descriptions of Rotokas were based on the variety spoken in
Togarao, basing my own eldwork there would make it possible to utilize existing materials and
assess the degree of change that has occured in the language. Another consideration was that
eldwork would be easier to conduct in a community where there has been prior exposure to
language documentation work and where there are consultants ready, willing, and able to par-
ticipate in this type of work. The only real drawback of Togarao was its relative inaccessibility.
The feeder road that runs from the coast to Togarao was poorly maintained and riddled with
potholes that would wash out during heavy rains.
2
In addition, there was no regularly available
transport, which meant that when no vehicle was available, travel to and from the coast along
the long and sometimes steep roads would have to be carried out on foot. However, the rel-
ative inaccessibility of the village has reduced the amount of language contact that has taken
place, effectively reducing the amount of Tok Pisin and English spoken in the community and
strengthening the position of Rotokas as the primary community language.
During my various stays in Togarao, I worked with native speakers of Rotokas in the local
community to document and describe the grammatical structure of the language. My two main
native speaker consultants were Timothy Taureviri and Sera Mon, shown in Figure 1.1.
1
The village is identied as Togarau on same maps, based on a misidentication of the nal vowel of the word.
2
During my nal trip to Togarao, a road improvement project nanced by the European Union was initiated
which should signicantly improve the quality of the feeder road.
4
Figure 1.1: Rotokas Native Speaker Consultants: Timothy Taureviri (left) and Sera Mon (right)
Timothy and Sera are both in their fties and learned Rotokas as their rst language. They
both speak Rotokas on a daily basis as their primary language. In addition, both consultants
are uent speakers of Tok Pisin (Neo-Melanesian). Timothy Taureviri had previously worked
with Irwin Firchow and Sera Mon is the daughter of Irwin Firchows primary consultant, David
Akoitai. Seras husband is from the mainland of Papua New Guinea and Tok Pisin is their
primary language of communication. Caleb Karuruanother native speaker of Rotokas who
worked with Firchowalso worked with me as a native speaker consultant. He is shown with
Irwin Firchow and David Akoitai in Figure 1.2.
3
The example sentences in this dissertation come either from previously published materials,
in which case they are cited, or from my own materials, which are listed in Table 1.2. The
majority of previously published materials were published exclusively in Rotokas but were sub-
sequently double-checked and given Tok Pisin translations by bilingual Rotokas consultants.
Uncited example sentences come either from eld notes or (more likely) from the authors lex-
ical database (a large Shoebox/Toolbox dictionary described in Appendix A). Citations appear
after the English gloss in square brackets.
1.3 Organization
This thesis is divided into two parts, reecting the twin aims described in 1.1: 1) a grammatical
sketch of Rotokas, and 2) a more detailed analysis of a particular topic of theoretical interest
namely, split intransitivity.
The rst part of this thesis is a grammatical sketch of Rotokas consisting of six chapters.
Chapter 1 introduces the aims and organization of the thesis and provides background informa-
3
The photo was taken during the seventies and was provided by Irwin Firchows widow, Jackie Firchow. Her
generous contribution of the photo is hereby acknowledged.
5
Figure 1.2: Irwin Firchow (left), David Akoitai (middle), and Caleb Karuru (right)
tion concerning the authors eldwork trips to Bougainville between 2003 and 2005. Chapter
2 chapter provides background information concerning the Rotokas language and its speakers.
2.1 covers the recent history of Bougainville as well as the deep history of the region while 2.2
provides important background information on Rotokas: prior documentation of the language,
information about its speakers, an overview of dialectal variation and an inventory of languages
spoken on Bougainville, and a summary of what is known about their genetic afliation. Chap-
ter 3 provides an overview of the phonology of Rotokas, which is typologically unusual for
possessing a very small phoneme inventory. 3.1 describes the phoneme inventory of the lan-
guage (the segmental inventory) while 3.2 describes what is known concerning the languages
suprasegmental phonology. Chapter 4 looks at the word classes, or parts of speech, found in
the language. It distinguishes between roots, stems, and words, and provides a breakdown of
the word classes into a number of categories: nouns, noun classiers, pronouns, verbs, adjec-
tives, adverbs, postpositions, interrogative, conjoiners, and exclamatives. Chapter 5 overviews
Rotokas morphology, which is fairly extensive, looking rst at nominal morphology and then at
verbal morphology. It also covers reduplication and morphophonemic rules. Chapter 6 is con-
cerned with syntax, focussing rst on the noun phrase and then on clausal syntax (both intra-
and inter-clausal). A few aspects of intraclausal syntax are discussed in the rst section: basic
constituent order, departures from the basic (canonical) constituent order in content questions
and when O is displaced, interrogatives, and negation. Interclausal sufx is covered in the nal
6
How Cited Description
title in quotes Recordings of various conversations and folk tales (e.g., C.2)
Robinson and Mon (2006) Trilingual English/Tok Pisin/Rotokas primary school reader
developed with Sera Mon
uncited Shoebox dictionary with 6088 entries and 7152 example sen-
tences
uncited Notes made during eldwork sessions in Togarao
RR:<CLIP>, <CONSULTANT> Descriptions of the Reciprocals video clips, glossed in Tok
Pisin (Evans et al., 2004)
CB:<CLIP>, <CONSULTANT> Descriptions of the Cut and Break video clips, glossed in Tok
Pisin (Bohnemeyer et al., 2001)
by publication Previously published Rotokas materials translated into Tok
Pisin
Table 1.2: Authors Rotokas Materials
section, which looks at complementation, verb phrases, and larger clausal units (sentences with
conjoiners).
The second part of this thesis is a detailed examination of verb classication and split intran-
sitivity. Chapter 7 formulates the basic analytical problem posed by the two classes of verbal
inection found in Rotokas. 7.1 more rmly establishes the formal nature of the distinction and
a clear set of diagnostics for its recognition while 7.2 states the basic problem and puts forward
a tentative hypothesis concerning its solution which is rened in later chapters as the facts of
the matter are established in more detail. Chapter 8 examines the nature of valency in Rotokas
and establishes that there are two main valency types in Rotokas: monovalent verb roots (in-
transitive), which take a single argument, and bivalent verbs roots (transitive), which take
two (or possibly three) core arguments. If a clause possesses two core arguments, it will show
agreement; however, the reverse does not hold true. If a verb shows agreement, it will not nec-
essarilly take two core arguments. This asymmetry owes to the fact that monovalent verb roots
are split according to their form of verbal inection: most show inection but some show
. Chapter 9 overviews the devices for increasing or decreasing the default valency of verb
roots. Valency-increasing derivations are discussed in 9.1 and valency-decreasing derivations
are discussed in 9.2. Valency changing derivations provide little evidence for an underlying
syntactic difference between and monovalent verb roots, since the various valency-changing
derivations are not sensitive to the distinction; however, they do provide additional evidence in
favor of a tight relationship between valency and verbal inection, since a decrease in valency
is associated with inection and an increase in valency with inection. Chapter 10 exam-
ines the nature of split intransitivity in Rotokas in more detail and discusses the implications
of Rotokas for theories of linking (the syntax-semantics interface) specically and grammatical
theory more generally. Chapter 11 provides concluding remarks and lays out a few directions
7
for future research.
The thesis also includes 3 appendices. The rst appendix provides a detailed listing of
verb stems taken from an electronic database of the Rotokas lexicon developed by the author.
The second appendix provides a formal implementation of the analysis of Rotokas morphology
within the framework of nite state morphology using the PARC software toolkit. The third ap-
pendix provides two sample texts with interlinear glossing and translations into English and Tok
Pisin. The sample texts provide two different versions of a traditional folk tale, one documented
by Irwin Firchow and the other by the author during his eldwork.
8
Chapter 2
Language Background
In this chapter, background information about Rotokas is provided in order to place the lan-
guage in a wider sociocultural perspective. Bougainville is situated within the context of Island
Melanesia in 2.1 and background information about Rotokas and its speakers is provided in
2.2.
2.1 Bougainville and Island Melanesia
Rotokas is spoken in the central region of the island of Bougainville, which belongs to a region
that is generally known as Island Melanesia, which lies to the east of mainland Papua New
Guinea and encompasses the larger islands of New Britain, New Ireland, and the Solomon
Island Chains, as well as various smaller islands and atolls that are too numerous to list, as
shown in Figure 2.1.
Figure 2.1 also shows the distribution of the two major groupings of languages spoken in the
region: Austronesian and non-Austronesian (Papuan). Austronesian languages appear in pink
while Papuan languages appear in blue. Casual inspection of the distribution of Austronesian
and Papuan languages shows that Bougainville is somewhat unique in the region to the extent
that it possesses a relatively higher proportion of Papuan languages (see 2.2.4 for details).
2.1.1 History of the Region
Australia-New Guinea has a history of settlement that is known from archaeological evidence
to date back as far as 40,000 BP (White and OConnell, 1982). Archaeological evidence from
the island of Buka (a smaller island immediately north of Bougainville) provides evidence of
inhabitation dating as far back as 29,000 years ago (Wickler and Spriggs, 1988).
Bougainville is named after the French explorer Louis Antoine de Bougainville (de Bou-
gainville, 1772), the rst European to spot the island, when he sailed past it in 1768 during his
9
Figure 2.1: Languages of Island Melanesia
circumnavigation of the globe following the Seven Years War. Contact between Bougainvil-
leans and the West did not begin in earnest for almost one-hundred years, when the German
New Guinea Company established control over Bougainville and Buka, Choiseul, the Short-
lands and Treasury Islands in 1885 (Sack, 2005). Their control over this area did not extend to
the islands farther south in the Solomon Island chain, which came under a British protectorate
in 1893 (with the eastern islands being added in 1899). In 1900, Germany transferred all of
its claims in the Solomons other than Bougainville and Buka to Great Britain while Britain,
in return, withdrew from Western Samoa. During the rst world war, Australia occupied the
island in 1914 and administered it as a League of Nations mandatory power from 1918 until
World War II. It was invaded by the Japanese in 1942 and between 1942 and 1945 was the site
of an intense military campaign in the lead-up to the assault on the Japanese bastion of Rabaul
(Gailey, 1991; Nelson, 2005).
After WWII, Australia resumed control over the island as a United Nations mandatory
power until Papua New Guinea achieved independence in 1975 (Waiko, 1993). When PNG
achieved independence, Bougainvilles copper resources provided an early source of govern-
ment revenue. Bougainville Copper Limited set up and ran the Panguna mine, which at the time
was the worlds largest open-cut copper mine. The mine proved to be politically contentious
due to disputes over land tenure and allegations of environmental damage (Vernon, 2005).
After negotations between landowners and the owners of the mine broke down, Francis Ona
formed the Bougainville Revolutionary Army, which began to sabotage mining operations. The
campaign was successful to the extent that in May 1989 the mine was shut down after the power
cables which supplied its electricity were blown up by a group of indigenous landowners led
10
by Francis Ona. On June 26, 1989, the Papua New Guinean government declared a state of
emergency, and in September, the Papua New Guinea Defense Force (PNGDF) was sent into
Bougainville in order to quell resistance to the mine. Their heavy-handed response enraged
Bougainvilleans and set in motion a chain of events that led to a decade-long military conict,
generally referred to as The Bougainville Crisisor simply The Crisisin which somewhere
between 10,000 and 15,000 people lost their lives, either directly through ghting or indirectly
through other causes (e.g., lack of medical attention).
A full history of The Crisis goes beyond the scope of this short and high-level overview of
the history of Bougainville (see Dorney (1998); Regan and Grifn (2005) for more information).
The main protagonists in this conict were the Bougainville Revolutionary Army (BRA), the
PNGDF, and The Resistance, a paramilitary group that dened itself in opposition to the BRA
and aligned itself with the national government of PNG. Many Rotokas joined The Resistance
after a foiled kidnapping attempt on the Akoitai family was carried out in Togarao by self-
proclaimed BRA members, resulting in the death of one local and a number of BRA members.
Despite their sympathy for the BRA cause, many members of the community feared reprisal as
a result of this incident and felt that the PNGDF provided the best chance of protection from
attack.
After years of hardship and bloodsheed, The Crisis ofcially ended in 1997, thanks in large
part to negotiations brokered by New Zealand. A Peace Agreement nalised in 2000 provided
for the establishment of an Autonomous Bougainville Government, and a referendum in the
future on whether the island should become politically independent. In 2005, elections for the
rst Autonomous Government were held and Joseph Kabui was elected President on June 15.
During the same year, the rebel leader Francis Ona, whose sabotage of the mine set in motion
the events leading to The Crisis, died after a short illness, leaving in question the leadership of
the BRA and its remaining hardcore members who had refused to join the peace process. These
individuals remain heavily armed and in control of the area surrounding the Panguna mine.
The effects of The Crisis on modern Bougainville can hardly be overstated. It has led to
widespread social and economic change, the effects of which will continue to be felt in the
coming years. These include a breakdown in law and order, which is only now beginning to be
properly dealt with. In addition, it has led to a great deal of population displacement, which has
disrupted the transfer of traditional customs and undermined respect for village elders. It has
also led to signicant decline in the eductional system of the island, which was at one point one
of the best in Papua New Guinea (Litteral, 2001). This has led to a somewhat unusual situation
where the older generation is both more steeped in the traditional culture and better educated
than the younger generation.
2.2 The Rotokas Language
This section provides background information on Rotokas in order to situate it within the wider
linguistic scene of Bougainville and Island Melanesia. The prior literature on Rotokas is briey
11
described in 2.2.1; dialectal variation in the language is overviewed in 2.2.3; and the relation-
ship of Rotokas to the other languages spoken in Bougainville is discussed in 2.2.4.
2.2.1 Prior Literature
All of the prior literature on Rotokas is the work of a missionary couple from the Summer
Institute of Linguistics, Irwin Firchow and Jacqueline Firchow, who translated the Bible into
Rotokas and also did some anthropological and linguistic work over the course of three decades
(from the early sixties to the late eighties). A number of publications came out of their work;
they are listed in Table 2.1 along with a brief description of their content.
Reference Description
Firchow and Firchow (1969) Description of the segmental phonology of Rotokas
Firchow (1971) Description of Rotokas nominals
Firchow (1973) Vocabulary of Rotokas (vowel length omitted from orthography)
Firchow (1977) Analysis of Rotokas nominals
Firchow (1974a) Collection of Rotokas songs
Firchow and Akoitai (1974) Collection of Rotokas stories (folk tales)
Firchow (1974b) Descriptions of Rotokas customs in Rotokas
The New Testament (1982) Translation of the New Testament into Rotokas
Firchow (1984) Electronic Shoebox dictionary on CD
Firchow (1987) Grammar sketch
The Old Testament (1993) Translation of the Old Testament into Rotokas
Table 2.1: Prior Literature on and in Rotokas
The primary motivation for their descriptive work was the translation of the Bible into Ro-
tokas and the linguistic descriptions produced by Irwin Firchow reect a limited background
in academic linguistics. For example, postnominal modiers are described as prepositions
in Firchow (1987:85). They are, however, largely accurate empirically and therefore provide a
very useful starting point for more in-depth analysis.
To my knowledge, there has been no documentation of Rotokas carried out since the Fir-
chows ceased working on the language. During the twenty to thirty year gap between the
publication of the above-listed materials and the start of my own research, the language has
undergone some change as a result of various factors. One of the main factors is multilingual-
ism in Tok Pisin/English as a result of increasing language contact. This is reected by the
growing number of loan words and the loss of a great deal of vocabulary associated with the
traditional culture. This vocabulary is typically described by Rotokas speakers as toktok bilong
bipo words from before and is retained only by elderly speakers of the language, who have
rst-hand knowledge of the traditional practices in which that terminology was embedded. For
example, few younger speakers of Rotokas are familiar with the word keroroi lean to, which
12
describes a traditional type of temporarily shelter used in the past during stays in the jungle
(for example, while hunting possum), or with the word toara market, a loan word from Teop,
which described a traditional practice of having bartering markets (presumably with the Teop,
given the borrowing of the term from their language). The loss of some traditional vocabulary
cannot be solely attributed to the loss of traditional practices, suggesting that other processes
are at work (for example, language contact with the Keriaka or between dialects of Rotokas). In
some cases, there is no readily forthcoming reason for a words obsolescence. For example, the
word kare has replaced ragui as a pluralizer for animate entities (animals, sh, etc.) (Firchow,
1987:40) and the word isike has replaced kusike as the generic term for rats.
Three books with monolingual Rotokas texts (collected from various native-speakers
primarily David Akoitaiand edited by Irwin Firchow) were published and these provide a
snapshot of the variety of Rotokas then spoken. Firchow and Akoitai (1974) provides a number
of folk tales and personal narratives while Firchow (1974b) provides descriptions of traditional
customs. During my various trips to Bougainville, native speakers of Rotokas translated these
monolingual Rotokas texts into Tok Pisin and they were systematically interlinear glossed for
analysis in Toolbox, a computer program developed by the Summer Institute of Linguistics for
the development of language resources (e.g., lexicons) and their deployment in linguistic anal-
ysis.
1
The impression of those speakers who translated these materials was that there was some
vocabulary in the texts that was no longer used but that they were overall very similar in form
and structure.
2.2.2 The Speakers of Rotokas
Rotokas speakers are primarily subsistence agriculturalists. Their primary crops are sweet
potatos, yams, taro, and English potatoes. These are supplemented by local fruits and vegeta-
bles (such as coconuts, various varieties of banana, and a local green known as kumul) as well
as some store-bought goods (such as tinned sh, rice, and noodles), paid for by money obtained
through various cash crops, such as cocoa and vanilla. The formal economy of Bougainville
was fairly small during the period when the eldwork described in this thesis was carried out,
having shrunk considerably as a result of The Crisis and the closing of the Panguna Mine. Op-
portunities for employment were limited and therefore government positions, such as that of a
school teacher or local government functionary were highly sought after.
In village of Togarao, where I spent the majority of my time during my eld research,
Rotokas is the primary language of communication. It is the rst language used by children and
it is the preferred language in day-to-day life. It is used in daily conversation, village meetings,
church, and numerous other contexts. However, the situation is somewhat different in Wakunai
Station, a village through which the highway of East Bougainville runs. At Wakunai Station,
the same dialect of Rotokas is spoken as the lingua franca but there are many more members
1
Toolbox is the latest incarnation of Shoebox, and differs from the latter in only a few respects, such as its
support for Unicode data storage and its ability to export data as XML (Robinson et al., 2007).
13
of the community who do not speak Rotokas as well as travellers who pass through the area,
and Tok Pisin is the primarily language of communication with these individuals. Tok Pisin is
therefore more frequently used in and around Wakunai Station than in Togarao. The remoteness
of villages such as Togarao therefore ensures less language contact and therefore less inuence
from Tok Pisin, but there is considerable transit between Wakunai Station and the more remote
inland villages, particularly among young men (who frequently spend time with relatives on the
coast in search of work and/or entertainment).
Although Rotokas remains the primary language of village life, and the rst language
learned by children, bilingualism in Tok Pisin is the norm among adult Rotokas speakers. It
is the lingua franca of Bougainville and is used on the local radio stations (for example, Radio
Bougainville), in the school system, at school meetings, and at political events. Tok Pisin is also
used as a lingua franca among the minority of villagers who, due to unusual circumstances, do
not speak Rotokas. There are a few individuals who have a passive command of Rotokas and
are able to understand the language but not to speak it. For example, the daughter of a local
politican who was raised in an urban environment does not speak Rotokas but is able under-
stand it. When conversing with family members who speak Rotokas, she will speak Tok Pisin
and others will speak to her in Rotokas or in Tok Pisin, depending upon their awareness of her
passive competence of Rotokas and their own level of comfort in Tok Pisin.
There is some familiarity with English in the community, as well. English is the ofcial lan-
guage of instruction after grade three in the Papua New Guinea education system, and therefore
anyone who has received formal education will have some familiarity with it, as well, although
competence in the language varies dramatically and depends largely on levels of educational
attainment, which is now fairly low in general due to the fact that the educational system of
Bougainville deteriorated considerably during The Crisis, as did most of the infrastructure on
the island. As a result of the deterioration of the feeder road connecting Togarao to the coastal
highway, Togarao has been less accessible than in the past and this has had an effect on the
economy and the school system. At present, most students manage to complete their studies
through grade eight but only a small percentage of students continue on to high school.
2.2.3 Dialectal Variation
The rstand onlysystematic survey of the languages and dialects of Bougainville was car-
ried out by the Summer Institute of Linguistics during the early sixties and is reported in Allen
and Hurd (1963). On the basis of lexicostatical comparison, it is claimed that there are four
dialects of Rotokas, named after the geographical regions where they are spoken: Central, Aita,
Pipipaia, and Atsilima. The names of the villages where these dialects are spoken and their
approximate population size at the time of publication (the sixties) are provided in Table 2.2.
2
2
Some of the villages in Table 2.2 are not exclusively Rotokas-speaking. For example, Allen and Hurd (1963)
observes that Teop is spoken in Tiaraka (Tearaka). According to Ruth Spriggs (a native-speaker of Teop collabo-
rating with Ulrike Mosel on its documentation and preservation), there is considerable language contact between
14
These gures must be interpreted with caution, given that they are fairly out-of-date. Migra-
tion and population growth will have changed the size and composition of these villages, and
therefore the total size of the Rotokas-speaking community.
3
The geographical distribution of the dialects recognized in Allen and Hurd (1963) is pro-
vided in Figure 2.2, where colored dots representing the four dialects have been superimposed
on village locations according to the following color scheme: Aita (yellow), Atsilima (green),
Central (blue), and Pipipaia (red).
Figure 2.2: Dialects of Rotokas [based on Allen and Hurd (1963)]
The only dialect of Rotokas described in any detail is Central Rotokas. This dialect is
labelled Rotokas Proper in Allen and Hurd (1963), but this term is eschewed here since it
unnecessarily privileges one dialect over the others. Aita Rotokas is mentioned in Firchow and
Firchow (1969) and its consonant inventory described in passing. Robinson (2006) compares
its segmental phoneme inventory to that of Central Rotokas and, on the basis of a compari-
son of cognate vocabulary in the two dialects, argues that the phoneme inventory of Aita Ro-
tokas is conservative and that the smaller inventory of Central Rotokas arose by collapsing the
voiced/nasality contrast in Aita Rotokas.
Details of the dialects and the criteria by which they are dened are otherwise lacking. No
information is available concerning the Pipipaia dialect. Although Central Rotokas is relatively
better described, information concerning its distribution is questionable. Native speakers of
Rotokas describe another variety of Rotokas which they describe as Red River, suggesting that
additional dialects may need to be recognized.
Rotokas and Teop in the villages of Tiaraka and Teohiupu.
3
Wurm and Hattori (1981) provide a higher gure for the total population of Rotokas speakersviz., 4,320. The
discrepancy between this gure and the gure provided by Allen and Hurd (1963) presumably owes to population
growth, since by 2000 the population of Bougainville had doubled (141,161 according to PNGs 2000 Census).
15
The most intriguing dialect of Rotokas identied in Allen and Hurd (1963) is Atsilima,
which had 112 speakers in the village of Atsilima when it was surveyed, but its current status
is unknown. Atsilima is possibly a dialect of Rotokas, but one that differs so much from it that
Allen and Hurd (1963) describe it as a sub-language of Rotokas: more distant than a dialect
and yet not far enough removed to be a separate language (Allen and Hurd, 1963:2).
4
It is
spoken in a language contact zone between Rotokas and Kereaka and is described by Rotokas-
speakers as a mix of the two languages.
2.2.4 The Languages of Bougainville
Bougainville covers an area of 10,954 km
2
, measuring 120 km in length and between 65 and
95 km in width. Despite its relatively small size, Bougainville possesses an impressive amount
of linguistic diveristya total of approximately 25 languages (Allen and Hurd, 1963; Tryon,
2005). The languages and the approximate geographic area where they are spoken is provided
in Figure 2.3.
5
Figure 2.3: Languages of Bougainville
4
Allen and Hurd (1963) dene sublanguage operationally in terms of the percentage of shared vocabulary
between speech varieties in the available word lists: Speech groups which are 93% to 100% related belong to the
same dialect, speech groups which are 76% to 92% related are different dialects of the same language, and speech
groups which are 65% to 75% related are sub-languages of the same language. (Allen and Hurd, 1963:5)
5
The best way of visualizing the geographical distribution of languages on Bougainville would be to plot each
language variety on a village by village basis. The linguistic boundaries in Figure 2.3 should therefore be viewed
largely as a convenient ction that provides only a very rough impression of where the various languages are
spoken.
16
A full listing of the languages of Bougainvilleincluding the Austronesian languagesis
provided in Table 2.3.
Rotokas is one of eight Papuan languages spoken in Bougainville. The Papuan languages
which make up roughly a third of the languages in Bougainvilleare listed below in Table
2.4.
Documentation of the Papuan languages of Bougainville is quite limited. Only Motuna and
Rotokas have modern descriptive grammars available; Buin has a dictionary with a grammar
sketch; Nasioi and Nagovisi have even more limited materials; Kunua has only one published
description (essentially little more than a vocabulary with some grammatical notes); Eivo and
Keriaka are completely undocumented.
There is some Austronesian-Papuan language contact at the edges of the Rotokas-speaking
areas with Keriaka, another non-Austronesian language in the Rotokas family, and Teop, an
Austronesian language belonging to North Bougainville network of the North-West Solomonic
chain (Mosel, 1991; Mosel and Spriggs, 1999a,b; Mosel and Reinig, 2000; Tryon, 2005; Schwartz
and Mosel, 2006). The degree of language contact between the Rotokas and surrounding lan-
guage groups is difcult to gauge, given the absence of solid ethnographic description (Grifn,
2005). However, recent work by the author in collaboration with Ulrike Mosel has revealed
a reasonable amount of lexical borrowing between Rotokas and Teop, covering a variety of
semantic domains, which provides grounds for believing that contact between the two groups
went far beyond casual contact and involved not only trade but also intermarriage.
The relationship of the Papuan languages to one another is a matter of controversy, as will
be seen in the following section, which takes up the question of the genetic afliation of the
languages of Bougainville.
2.2.5 Genetic Afliation
Rotokas is usually described as belonging to the East Papuan phylum, a somewhat controversial
grouping of non-Austronesian languages rst proposed by Wurm (1975a). Before discussing
this grouping in greater detail, it is worthwhile to step back and examine the prior descriptive
work upon which it is based.
On the basis of an examination of shared vocabulary (lexicostatistics), Allen and Hurd
(1963:20) claim that Rotokas belongs to the Kunua-Keriaka-Rotokas-Eivo stock and to the
Rotokas-Eivo family. (They dene a stock as languages sharing 12% to 28% cognate vocabu-
lary and a family as languages sharing 28% to 81%cognate vocabulary.) A pairwise comparison
of all of the languages within the survey is provided in Table 2.5.
In Table 2.5 a horizontal line divides the two language groups: Austronesian towards the
top from non-Austronesian languages towards the bottom, with a vertical line dividing the
within-group and between-group comparisons. As one would expect, rates of shared vocabulary
are much higher within language groups than between them. In addition, the rates are higher
within Austronesian than within Papuan, which is consistent with the view that the Papuan lan-
17
Kunua
(Rapoisi)
Keriaka
Rotokas
Eivo
Nasioi
Nagovisi
Motuna
(Siwai)
Buin
Figure 2.4: Shared Vocabulary Percentages of Allen and Hurd (1963) as Neighbor Joining Tree
guages have undergone greater diversication due to an earlier settlement date. For example,
the Austronesian languages Teop and Tinputz share 67% cognate vocabulary whereas the Non-
Austronesian languages Rotokas and Kunua share 30% cognate vocabulary. Teop and Rotokas
were found to share only 6% cognate vocabulary (close to chance according to Dunn and Terrill
(submitted)).
The gures provided by Allen and Hurd (1963) can be transformed into distances matrices
to build a distance tree using the neighbor-joining tree method, an algorithm which seeks the
optimal tree that preserves the relative distance between each of the terminal nodes (Saitou and
Nei, 1987). The result is provided in Figure 2.4.
Although there are problems with establishing genetic relationships solely on the basis of
shared vocabulary (Laycock, 1970; McElhanon, 1987), the work of Allen and Hurd at least
gives a rough impression of Rotokas relationship to some of the other language of Bougainville
and will have to sufce until more complete descriptions of the various languages are available.
Despite the sketchy materials available on the languages of Bougainville, a few authors have
put forward tentative genetic groupings for Rotokas.
Based on prior work by Allen and Hurd (1963) and Greenberg (1971), Wurm (1972) pro-
vides the rst explicit postulation of an East Papuan phylum:
The East Papuan phylum which comprises what has until quite recently (Wurm
1971) been regarded as the Bougainville Phylum, the Reef Islands-Santa Cruz
phylum-level Family, and a number of isolates in the New Britain, New Ireland,
Solomon Islands and Louisiade Archipelago areas, has been set up by the present
18
writer (Wurm 1972a) on the basis of his own preliminary assessment of the avail-
able information and materials, and taking into account Greenbergs (1971) ndings
as well . . .
The Bougainville branch of Wurm (1972)s proposed East Papuan phylum is provided as a
tree diagram in Figure 2.5.
Bougainville
E. Bougainville
Nasio Family
Nasioi
Nagovisi
Buin Family
Motuna
Buin
W. Bougainville
Rotokas Family
Eivo
Rotokas
Konua (Rapoisi)
Kereaka
Figure 2.5: Bougainville Branch of the East Papuan Phylum (Wurm, 1972, 1975a)
Ross (2001) questions the validity of Wurm (1975a)s East Papuan phylum, noting that it is
based on phonological similarity in word lists (rather than on regular sound correspondences)
and uses typological similarities to bolster proposed groupings. This is problematic because
such evidence could equally well reect the results of language contact rather than inheritance.
For this reason, Ross (2001) looks at pronouns on the assumption that they are less susceptible
to wholesale borrowing than other parts of speech (for a critical assessment of this assumption,
see Thomason and Everett (2001)). On the basis of the pronominal evidence, Ross (2001)
concludes that there is no good evidence that the West and East Bougainville groupings of
Wurm (1972) are related:
Surprisingly, perhaps, the two groups recognised by Wurm on Bougainville seems
to be unrelated to each other. Matthew Spriggs (pers. comm.) points out that
there has been a good deal of recent population movement on Bougainville, and
that, although the two groups appear contiguous on the map, they were probably
separated in traditional times by a large area of volcanic activity (Ross, 2001:311).
Typological similarity has frequently been invoked in discussions of the East Papuan phy-
lum, but, as observed in Dunn et al. (2002), the languages in the proposed grouping are quite
heterogenous in terms of their typological features. The majority of them do, however, possess
the following features:
constituent order the majority exhibit verb-nal constituent order (typically Papuan), with the
notable exception of Kuot; most also exhibit posessor-possessed order in possessive noun
phrases.
19
pronominal systems an inclusive/exclusive distinction in the rst person non-singular and a
dual number category are both widespread
verbal morphology largely segmentable; nominative/accusative; argument marking through
afxation (with a preference for sufxation)
Given the equivocal status of the evidence in favor of the East Papuan phylum, and the
absence of systematic sound correspondences that would lend themselves to traditional meth-
ods of reconstruction, Dunn et al. (2005) pursue a novel approach to the problem by using
methodologies taken from computational cladistics (Kitching et al., 1998). They constructed
a database of grammatical features for 15 Papuan and 16 Austronesian languages and, using
cladistic algorithms (maximum parsimony and NeighborNet), analyzed the potential phyloge-
netic relationship between these languages. When applied to the Austronesian languages, the
results of the technique provided a very close match to the classications reconstructed using
the traditional comparative method. This provided the basis for extrapolating the technique to
the Papuan languages, where it was found that the classications produced by cladistic algo-
rithms strongly reected geography. This is interpreted as evidence of large-scale genealogical
clustering of the Island Melanesian languages that predates the Austronesian expansion. They
interpret their results as evidence in favor of the idea that the two language groups now located
on the Solomons and Bougainville separated from a common ancestor.
20
Dialect Villages Population (1960s)
Central Total: 1640
Beteriopaia 131
Ibu 104
Keriana 92
Leikaia 68
Lesiopaia 95
Mapioro 132
Okowapaia 128
Ruruvu 129
Sirioripaia 194
Sisivi 190
Teakon 75
Tiaraka 86
Togarao 216
Aita Total: 1003
Koribori 62
Kusi 89
Nupatoro 164
Osiwaipa 146
Pokoia 217
Siribia 93
Tokai 112
Tsubiai 120
Pipipaia Total: 765
Bulistoro 149
Kakaropaia 190
Pipipaia 264
Tutupaia 162
Atsilima Total: 112
Atsilima 112
3520
Table 2.2: Where Rotokas Dialects Are Spoken (Allen and Hurd, 1963)
21
Afliation Name Ethnologue Code Speakers
Papuan Konua (Kunua) / Rapoisi kyx 3,500
Rotokas roo 4,320
Keriaka kjx 1,000
Eivo eiv 1,200
Nasioi (Kieta) nas 10,000
Nagovisi nco 5,000
Siwai (Motuna) siw 6,600
Buin (Telei) buo 30,500
Total 62,120
Austronesian Halia hla 20,000
Haku hao 5,000
Solos sol 3,200
Petats pex 2,000
Saposa sps 1,400
Hahon hah 1,300
Piva tgi 550
Banoni bcm 1,000
Tinputz tpz 3,900
Teop tio 5,000
Papapana paa 150
Torau (Rorovana) ttu 605
Uruava urv EXTINCT
Nehan (Nissan) nsn 7,000
Takuu nho 250
Nukumanu nuq 200
Nuguria nur 200
Total 51,755
Table 2.3: The Languages of Bougainville
22
Name Code Pop. Notes
Konua (Rapoisi) kyx 3,500 M uller (1954)
Rotokas roo 4,320 Firchow and Firchow (1969); Firchow (1987)
Keriaka kjx 1,000 UNDESCRIBED
Eivo eiv 1,200 UNDESCRIBED
Nasioi (Kieta) nas 10,000 Hurd and Hurd (1970a,b); Hurd (1977)
Nagovisi nco 5,000 Hostetler and Hostetler (1975)
Motuna (Siwei) siw 6,600 Onishi (1994, 2002)
Buin (Telei) buo 30,500 Laycock (2003)
Table 2.4: Papuan Languages of Bougainville (Allen and Hurd, 1963; Tryon, 2005)
Teop Austronesian
53 Hahon
67 71 Tinputz
21 16 17 Halia
17 15 16 34 Solos
27 19 15 56 65 Petats
40 36 28 34 30 40 Saposa
32 24 23 25 28 20 20 Banoni
24 24 20 19 20 15 23 70 Nagarige
19 16 15 19 20 16 16 20 17 Torau
37 27 21 25 18 20 35 14 22 29 Papana
19 19 17 27 19 28 20 20 17 14 20 Nissan
21 13 17 23 19 18 19 13 18 14 16 19 Nahoa
11 11 8 5 6 4 2 5 5 10 7 2 4 Kunua Papuan
8 6 4 6 3 3 2 9 5 8 4 2 3 22 Keriaka
6 3 5 7 1 3 3 7 7 4 5 3 2 22 30 Rotokas
8 7 6 8 6 5 5 7 7 11 3 5 5 17 23 35 Eivo
3 4 7 2 6 3 3 9 2 11 7 3 7 8 5 7 9 Nasioi
3 5 4 3 4 4 1 11 7 4 4 1 4 4 4 8 6 50 Nagovisi
3 3 4 2 3 4 0 6 2 7 1 0 4 4 4 6 11 27 17 Siwai
6 4 5 1 2 3 2 8 4 8 4 1 4 6 7 7 4 20 19 34 Buin
Table 2.5: Cognate Percentage in the Languages of Bougainville (Allen and Hurd, 1963:21)
23
24
Chapter 3
Phonology
Published materials on the phonology of Rotokas are few in number. The primary references
are Firchow and Firchow (1969), which describes its segmental phonology, and Firchow (1973),
which covers some of the same ground and provides a few remarks concerning its supragemen-
tal phonology.
3.1 Segmental Phonology (Phoneme Inventory)
The phoneme inventory of Rotokas is one of the worlds smallest, as pointed out by Firchow
and Firchow (1969): The Rotokas languages ranks among those analyzed languages of the
world with the least number of segmental phonemes [emphasis mine] (hereafter referred to
simply as phonemes). It should be stressed that, although Rotokas possesses the smallest
known inventory of segmental phonemes, the language has a vowel length distinction which
effectively doubles its inventory of vowel phonemes (Maddieson, 1984).
3.1.1 Vowels
Rotokas possesses a simple ve-vowel system with a two-way length distinctioni.e., every
vowel has a short and long variant. Firchowand Firchow(1969) describes the vowels as follows:
/a/ low central, open and unrounded
/e/ mid front, varies between mid close and mid open unrounded
/i/ high front, varies between the high close and high open and is unrounded
/o/ mid back, mid close back rounded
/u/ high close back rounded
Unlike its consonant inventory, the vowel inventory of Rotokas is fairly typical from a ty-
pological perspective. As Maddieson (1984:126) notes, an inventory of ve vowels is the most
25
common sort (and represents 21.5% of the languages in the UPSID database). Furthermore,
the vowels in the inventory of Rotokas are the ve most cross-linguistically common vowel
qualities, as shown in Table 3.1.
Vowel Num. of Languages Percentage
/i/ 290 91.5%
/a/ 279 88.0%
/u/ 266 83.9%
/o/ 139 43.8%
/e/ 118 37.2%
Source: Maddieson (1984:125)
Table 3.1: Five Most Common Vowels in UPSID
Vowel length is distinctive (Firchow and Firchow, 1969) and all ve vowels have both a
short and long counterpart. Minimal pairs for vowel length are provided for all ve vowel
qualities in Table 3.2, where long vowels are indicated by the IPA symbol :.
Contrast Minimal Pair
/a/ vs. /a:/ varuto esh, meat
va:ruto deaf-mute
/e/ vs. /e:/ kera species of bird (similar to albatross)
ke:ra to call for, to beckon
/i/ vs. /i:/ pigi to squeeze
pi:gi g
/o/ vs. /o:/ ovato legendary wild man
o:vato type of ground
/u/ vs. /u:/ tupesi second
tu:pesi hoe
Table 3.2: Minimal Pairs for Vowel Length
3.1.2 Consonants
Rotokas has only 6 consonants, the result of crossing three points of articulation (bilabial, alve-
olar, velar) with a voicing distinction (voiced vs. voiceless). These phonemes are listed accord-
ing to their most characteristic allophonic realization in Table 3.3 (see 3.1.3 on the practical
orthography used for Rotokas consonants).
The IPA symbols in Table 3.3 are somewhat arbitrary, given that there is considerable al-
lophonic variation of the consonant phonemes, as described in Table 3.4. This suggests that
manner is only partially specied (or possibly unspecied).
26
Voicing Bilabial Alveolar Velar
Voiceless p t k
Voiced B R g
Table 3.3: Rotokas Consonants
Point of Articulation
Voicing Manner Bilabial Alveolar Velar
Voiceless Stop p t k
Affricate ts
Fricative s
Voiced Stop b d g
Flap R
Liquid l
Fricative B G
Nasal m N
Table 3.4: Allophonic Variants of Rotokas Consonants
Some aspects of the consonant inventory of Rotokas are fairly typical cross-linguistically.
For instance, Maddieson (1984:39) observes that the most common situation among languages
is the possession of two stop series (i.e., two set of stops that share the same manner) and
three places of articulation, and that, if a language has a two stop series, it has a voice onset
time contrast between them: over half (51.1%) of the languages in UPSID possess 2 stop series
(51.1%) and 3 places of articulation (53.9%) and among languages with two stop series, 88.9%
have a voice onset time contrast between them. However, other aspects of the inventory are
atypical, such as the lack of a primary nasal consonant (Ferguson, 1966).
The two-way voicing distinction found in Central Rotokas appears to be the result of a
collapsing of a three-way contrast between voiced, voiceless, and nasal stops, which is still
found in Aita Rotokas (Robinson, 2006). During eldwork in Bougainville, the author worked
with speakers of Aita Rotokas from the village of Kusi. Previous description of Aita (Firchow
and Firchow, 1969) suggested that the voiced stops of Central Rotokas correspond to nasal
stops in Aita. However, the author found that the phoneme inventory of Aita Rotokas includes
both voiced and nasal stops. A comparison of cognate items in the two dialects reveals that
the nasal stops of Aita Rotokas systematically correspond to voiced stops in Central Rotokas,
regardless of their position within the word. However, voiced stops in Central Rotokas do not
always correspond to nasal stops in Aita Rotokas. Given the sound correspondences between the
two dialects, the reconstruction of Proto-Rotokas appears to require the postulation of a sound
change whereby nasality was lost in Central Rotokas (rather than acquired by Aita Rotokas). In
other words, Aita Rotokas is conservative with respect to nasality, whereas Central Rotokas is
27
innovative.
The following minimal pairs demonstrate the voicing distinctions for each point of articula-
tion Firchow and Firchow (1969:273):
Table 3.5: Minimal Pairs for Voicing
Contrast Minimal Pair
/p/ vs. /v/ pore to turn
vore to return
/t/ vs. /r/ tupa to lock
rupa dark
/k/ vs. /g/ kapu tight
gapu naked
The most systematic allophonic variation is between [t] and [s]. The former occurs between
all vowels except /i/ while the latter occurs only before /i/.
3.1.3 A Practical Orthography
A practical orthography for Rotokas was established by Irwin Firchow in collaboration with
the Rotokas-speaking community. This orthography is used in a variety of pre-existing lan-
guage materials already familiar to many native-speakers (The Old Testament, 1993; The New
Testament, 1982; Firchow and Akoitai, 1974; Firchow, 1974a,b). One aspect of the practical
orthography that is typically overlooked by native-speakers is the distinction between short and
long vowels, which is made by doubling a vowel letter.
All examples provided here will be written using this practical orthography.
3.2 Suprasegmental Phonology
In this section the suprasegmental phonology of Rotokas is overviewed. The syllable structure
is described in 3.2.1 and word stress is described in 3.2.2. Since reduplication provides ad-
ditional evidence in favor of the analysis of Rotokas stress assignment as a quantity-sensitive
system, it is also discussed, in 3.2.3.
3.2.1 Syllable Structure
There is considerable cross-linguistic variation in permissible syllable types (Blevins, 1995:217).
The syllable structure of Rotokas is fairly simple, consisting of an optional consonant onset and
a vowel nucleus (with consonant codas prohibited): (C)V. Rotokas therefore falls on the fairly
restrictive end of the continuum, permitting only two of the 9 syllable types listed in Table 3.7.
28
Practical Orthography IPA
a a
aa a:
e e
ee e:
g g
i i
ii i:
k k
o o
oo o:
p p
r R
s s
t t
u u
uu u:
v B
Table 3.6: Practical Orthography for Rotokas
Syllable Type V CV CVC VC CCV CCVC CVCC VCC CCVCC
Permitted? Y Y N N N N N N N
Table 3.7: Rotokas Syllable Types
The two syllable types combine more or less freely to form words of varying length, as
illustrated in (1), where a period (full stop) indicates a syllable boundary.
(1) a. upe ceremonial hat [ u.pe ]
b. aatu ying fox [ aa.tu ]
c. varu meat [ va.ru ]
d. veeta bamboo [ vee.ta ]
e. ketoo seedling [ ke.too ]
f. keetaa jaw [ kee.taa ]
A breakdown of the Rotokas lexicon according to word length (measured in terms of the
number of segmental phonemes) is provided in Figure 3.1.
Figure 3.1 treats long vowels as a single segmental phoneme. Firchow (1987) analyzes
long vowels as a sequence of two short vowels, effectively treating vowel length as an issue
29
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Num. of Segmental Phonemes
N
u
m
.

o
f

E
n
t
r
i
e
s

i
n

L
e
x
i
c
o
n
0
2
0
0
4
0
0
6
0
0
8
0
0
Word Length Type
Roots Stems
Figure 3.1: Word length in Rotokas Lexicon
of syllable types rather than phoneme types. This analysis requires the postulation of complex
vowel nuclei. In other words, according to such an analysis, the syllable structure of Rotokas
would be slightly more complicated, as in (2).
(2) (C)V
i
(V
i/j
)
The treatment of vowel sequences remains an issue for more in-depth investigation. Quite
long vowel sequences can result from morphological processes, such as reduplication, as in
(3), where a six vowel sequence results from the reduplication of the verb stem aio eat, or
cliticization, as in (4).
(3) Ratu,
Ratu
aio-a
food-SG.N
oa
RPRO.3.SG.N
aioaio-pa-ri-vere
eat.RDP-CONT-2SG
-NF
raivaro
road
Ratu, as for food, eat it on the road.
(4) opi-vira
shortcut-ADV
ikau-ri
run-2SG
vo-vaiao=ia
SPEC-road=LOC
Take a shortcut running along this road.
30
3.2.2 Word Stress
Firchow (1973) devotes little attention to suprasegmental phonology, but he does provide some
observations concerning the assignment of stress. He notes that stress is largely predictable on
the basis of word length, measured in syllables.
1
In words consisting of 2-3 syllables, the rst syllable is stressed, as in (5); in words consist-
ing of 4 syllables, the rst and third syllables are stressed, as in (6).
(5) a. k e.pa house
b. .to banana
c. u.ta.ve baylor shell
(6) a. e.to.ka.si re
b. a.ta.ri.to sh
Firchow (1973) also claims that in words consisting of 5 or more syllables, such as those in
(7), the third-from-the-last syllable is stressed most strongly. This observation is questionable
and is further complicated in the case of (7b) by the fact that the third-from-the-last syllable
consists of a vowel sequence (ai).
(7) a. ga.ru.t u.vi.ra slowly
b. po.po.te.p ai.ra.ra white-men
Firchow also notes that there are exceptions to these rules, primarily relating to long vowels,
although he does not clarify the nature of these exceptions. One such class of exceptions are
bisyllabic words in which the rst syllable consists of a single short vowel and the second
syllable consists of a single long vowel, such as those listed in Table 3.8. In these words,
primary stress falls on the second syllable rather than the rst.
This is a systematic class of exceptions, and can be explained if we assume that stress
assignment in Rotokas is quantity-sensitive (Hayes, 1995). According to such an analysis,
word stress in Rotokas is a xed system in the sense that the location of stress is predictable by
general rules. Furthermore, it appears to be governed purely by phonological factors (distance
from word edges, syllable weight, etc.) and not by morphological factors such as the distinction
between roots and sufxes.
Given a few reasonably well motivated assumptions, stress assignment in Rotokas can be
calculated in a fairly straightforward fashion. The rst step in calculating stress assignment in
Rotokas is to parse a candidate word into feet, following the assumptions described in (8).
1
Note that Firchow (1973) does not provide syllable boundaries. These are based on the description of syllable
structure in Firchow and Firchow (1969).
31
Words Gloss
kap o: poor, destitute
kap u: dumb, (not) speaking
ket o: plant which came up from seed
kok e: peek through a blind or crack
kok o: to carry, a plate
kop: die, very ill
kor o: to have hampered speech
kup: pupa of beetle
kur o: penis
kus: cool off
rep o: hiccup
ror o: to suckle, to drink
rug o: think, reason, hope
tav o: wall up with sago palm leaves
tok: tight, binding
tor: bamboo
tur a: sew up sago leaves
vis: you (pl.)
vor o: hunt with dogs
Table 3.8: Bisyllabic Words Stressed on the Second Syllable
(8) foot construction proceeds from left to right
foot construction is quantity-sensitive
the foot is trochaic
primary stress falls on the leftmost foot
Given the assumption of trochaic feet, a proper foot can therefore take one of the forms
provided in Table 3.9, where H stands for heavy syllables, L for light syllables, and stressed
syllables appear in boldface.
Given the previous assumptions, word stress is predictable as follows: primary stress falls on
the rst syllable of the rst foot and secondary stress falls on the rst syllable of all subsequent
feet. Degenerate feet (i.e., syllables that cannot be parsed into a well-formed trochaic foot) are
unstressed.
These principles explain the patterns of word stress observed for the various words men-
tioned by Firchow (1973), as shown in Table 3.10: utave baylor shell in (5) parses into one
trochaic foot consisting of two light syllables; atarito sh parses into two feet, but since the
third syllable cannot by itself form a proper trochaic foot (cf. Table 3.9)), the second foot is
degenerate and therefore cannot receive secondary stress.
32
Light Syllable Heavy Syllable
Ft
L
Ft
H
Table 3.9: Trochaic Feet
LL LLL
PrWd
Ft
ke
L
pa
L
PrWd
Ft
u
L
ta
L
Ft
ve
L
Table 3.10: Metrical Structure for Bisyllabic and Trisyllabic Words
As noted in Firchow and Firchow (1969), vowel length interacts with stress assignment.
This can be seen most clearly in the case of CVCV: roots, which receive stress on the second
syllable rather than the rst. According to the assumptions previously given in (8), this follows
from the fact that such words will necessarily begin with a degenerate foot, as shown by (9)).
(9) PrWd
Ft
to
L
Ft
rii
H
Vowel length in the rst two syllables is decisive in stress assignment. Only words beginning
with a light syllable followed by a heavy syllable will have primary stress on the second syllable.
All other types of words will have primary stress on the rst syllable. This is demonstrated in
Table 3.11 for words beginning with HL or HH.
There are two additional considerations that lend support to this account of word stress in
Rotokas. The rst is that stems and word consist minimally of a trochaic foot. In other words,
33
LL LH HL HH
PrWd
Ft
ke
L
pa
L
PrWd
Ft
to
L
rii
H
PrWd
Ft
pii
H
ro
L
PrWd
Ft
kee
H
Ft
taa
H
Table 3.11: Metrical Structure of HL and HH Words
there are no content words consisting of only a single syllable and only a few function words
consisting of a single syllable.
The exceptions to the trochaic foot minimality requirement are all function words: the com-
plementizer ra, the third person singular masculine pronoun va, and the deictic particle vo
here. Even these minor exceptions can potentially be explained away by analyzing them as
clitics rather than full words. It would then be possible to say that all morphology consisting of
less than a trochaic foot attaches to a minimal word.
The second consideration that supports the postulation of a quantity-sensitive system of
metrical phonology in Rotokas is reduplication, which is analyzed in the following section.
3.2.3 Reduplication
Reduplication provides additional evidence in favor of the postulation of quantity-sensitive met-
rical feet in Rotokas. It is a fairly productive process in the language, particularly for verbal
stems, where it has an iterative meaning, and to a lesser extent for nominal stems, where it has
a distributive meaning.
In the simplest case, reduplication consists of the repetition of an entire stemthat is, full
reduplication. Full reduplication is found with bisyllabic stems in which both syllables are light
and monosyllabic stems in which the single syllable is heavy. The reduplication of the stems
eri dig, roo saw, and gasi break is illustrated in (10) through (12).
(10) Rake
Rake
evao-a
tree-SG.N
erieri-pa-re
dig.RDP-CONT-3SG.M
evao
tree
eri-pa-a=ia
dig-DERIV-SG.N=LOC
Rake is digging that tree with a shovel.
(11) ragai
PPRO.1.SG
kookai
rooster
rooroo-pa-a-voi
saw.RDP-CONT-1SG
-PRES
ra
COMP
rera
PPRO.3.SG.M
ori-sia
cook-DEP.SEQ
I sawed the rooster in order to cook it.
34
(12) Perairi
Perairi
rauru-va
spear-SG.F
gasigasi-re-vo
break.RDP-3SG.M
-IP
uvare
because
oira=ia
PPRO.3.SG.F=LOC
ora-toga-ro-e
RR-spear-3SG.M
-IP
Perairi broke the spear because he speared himself with it.

Longer stems do not reduplicate in full but instead show partial reduplication. For example,
the stems parikou crossed and ragete be weak do not reduplicate in full; only the rst two
syllables are reduplicated, as illustrated in (13) and (14).
(13) Savere
Savere
takei
wall
pariparikou-vira
cross.RDP-ADV
pura-re-voi
make-3SG.M
-PRES
rera
PPRO.3.SG.M
vo-kepa-aro=ia
SPEC-house-POSS=LOC
Savere is making thatched walls on his house.
(14) kakae-to
child-SG.M
rageragete-pie-pa-i-voi
weak.RDP-CAUS-DERIV-3PL
-PRES
riako-va
woman-SG.F
rera
PPRO.3.SG.M
pitu-pa-oro
hold-CONT-DEP.SIM
The women are weakening the child by holding him so much.
At this stage, the generalization appears to be that reduplication consists of copying the rst
two syllables of the reduplicated stems; however, the behavior of stems with a long vowel in
their rst and/or second syllable does not conform to such a simple generalization. When the
rst syllable of a reduplicated stem is long, the reduplicant consists of only the rst syllable, as
illustrated for the verb stem tuusi shake in (15) and the verb stem rookaa distribute in (16).
(15) Tori
Tori
riro-vira
big-ADV
tuutuusi-pa-ro-i
shake.RDP-CONT-3SG.M
-PRES
uriri-pa-oro
be.afraid-CONT-DEP.SIM
Tori is shaking greatly with fear.
(16) Raviata
Raviata
oira-ra=pa
man-PL.N=BEN
aio-ara
food-PL.N
roorookaa-pa-re
distribute.RDP-CONT-3SG.M
Raviata distributed food to everyone.

When the rst syllable of a stem is short and the second syllable long, the long vowel of the
second syllable is shortened, as illustrated for the verb stem rugoo think in (17).
(17) Riopeiri,
Riopeiri
aaro-vira
excessive-ADV
rugorugoo-pa-u
think.RDP-CONT-2SG
Riopeiri, you think too much.

35
Before attempting to produce a generalization that will cover all of the various attested
cases, it pays to revisit the metrical structure of words described in 3.2.2. According to the
rules given in (8), the metrical structure of the three stem patterns illustrated in (15) through
(17) is provided in Table 3.12.
HL HH LH
PrWd
Ft
tuu
H
Ft
si
L
PrWd
Ft
roo
H
Ft
kaa
H
PrWd
Ft
ru
L
Ft
goo
H
Table 3.12: Metrical Structure of HL, HH, and LH Stems
If reduplication is described in terms of the units of metrical phonology, a simple and elegant
generalization of reduplication can be maintained, which is simply that reduplication copies the
rst foot (rather than the rst two syllables). Since in Rotokas a foot consists of either a heavy
syllable or two light syllables (cf. 3.2.2), the reduplication of stems with a heavy syllable falls
out naturally, as can be seen in Table 3.13.
HL HH LH
PrWd
Ft
tuu
H
Ft
tuu
H
si
L
PrWd
Ft
roo
H
Ft
roo
H
Ft
kaa
H
PrWd
Ft
ru
L
go
L
Ft
ru
L
Ft
goo
H
Table 3.13: Metrical Structure of Reduplicated HL, HH, and LH Stems
36
Chapter 4
Word Classes
This chapter provides an overview of the word classes of Rotokas. Although a meaningful
distinction can be drawn between word classes and parts of speech (for example, a given part
of speech may consists of multiple word classes), the term word class is used here in a fairly
loose sense, more or less as a synonym for part of speech. Part of speech systems have received
a great deal of attention among typologists and the discussion here is informed by this literature
(Schachter, 1985; Sasse, 1993; Anward et al., 1997; Vogel and Comrie, 2000; Evans, 2000;
Evans and Osada, 2005).
4.1 Root, Stem, and Word Classes
Before discussing the various word classes found in Rotokas, it is useful to distinguish between
root, stem, and word classes. The distinction between these three units is as follows (Payne,
1997):
Root A root is an unanalyzable form that expresses the basic lexical content of the word.
Stem A stem consists minimally of a root, but may be analyzable into a root plus derivational
morphemes
Word A word is a minimal stand-alone unit, which consists of a stem and possibly inectional
morphemes.
Consider (18). It is a minimal sentence in the sense that none of its elements can be freely
ommitted (direct objects of transitive verbs cannot be freely elided in Rotokas, and are realized
either as a pronoun or a noun phrase) and it consists of only two words: the noun koie and the
verb kopiipieeva.
(18) koie
pig
kopii-pie-e-va
die-CAUS-3SG.F
-RP
She killed the pig.

37
In (18), the word kopiipieeva She killed is morphologically complex. It is based on the
verb stem kopiipie kill, which is ultimately derived from the verb root kopii die; however,
the word koie pig is morphologically simple, consisting of only a single morpheme. It is
therefore a root, a stem, and a word simultaneously. The analysis of the two words into root,
stem, and word is provided in Figure 4.1.
Morphologically Simple Morphologically Complex
Word
..
Stem
..
Root
..
koie
Word
..
Stem
..
Root
..
kopii pie e va
Figure 4.1: Breakdown of Words from (18)
A full account of word classes in Rotokas should provide an analysis of the relationship
between root and stems on the one hand and stems and words on the other. It would provide an
explicit account of why some roots are capable of functioning as a noun or as a verb whereas
others can only function as one or the other, but not as both. For example, the root atari sh
is capable of functioning as a verb, as in (19), or a noun, as in (20).
(19) Jisu
Jesus
Pita
Peter
tavi-re-va
tell-3SG.M
-RP
oa iava
hence
riro-vira
big-ADV
atari-ro-epa
sh-3SG.M
-RP
Jesus told Peter so that he would go catch many sh.

(20) opuruva
canoe
ivara
above
iava
POST
vuvureo-to
ying-SG.M
atari-to
sh-SG.M
site-pa-io-vo
watch-CONT-1PL.EXCL-IP
osa
as
papa-pa-re-vo
y-3SG.M
-IP
toru
wave
kou-ro
CLASS-PL.CL
ivara=ia
above=LOC
From inside of the canoe, we look at the sh as he ies on top of the waves.
Given roots such as atari sh, it is necessary to accept that some roots are underspecied
with respect to their stem class membership. However, it is not the case that all roots are
unspecied for word class membershipi.e., there is a distinction between nouns and verbs
within the lexicon itselfsince there are a number of stems (e.g., kakae child) that can be
used nominally but not verbally. The primary concern of this chapter will therefore be the
relationship between stem and word classes.
The issue of root versus stemwill arise later in the case of labile verbsthat is, ambivalent
verb roots that show two patterns of valency, either taking a single core argument and showing
agreement or taking two core arguments and showing agreement. For example, the verb
root kavau has two meanings, to be born or to give birth. The meaning to be born shows
agreement while the meaning to give birth shows agreement. The question is whether
38
there is a single underspecied root from which the two meanings (and their associated sub-
categorization and agreement) derive. For a more theoretically oriented discussion of this issue
and its syntactic representation, see Chapter 11.
4.2 An Inventory of Rotokas Word Classes
In the following sections, the various word classes of Rotokas are enumerated and described
in turn. Some of these word classes can be grouped together to form major word classes. For
example, nouns, classiers, and pronouns could be grouped together into a broader category of
nominals, as in (21).
(21) Nominal
Noun Classier Pronoun
This suggests that word classes can be organized hierarchically into inheritance classes, such
that the behavior of a parent class (e.g., Nominal) is inherited by a child class (e.g., Pronoun).
There is considerable debate concerning the feasibility of this exercise, since it has been claimed
that any attempt to ground the exercise in purely distributional criteria will produce almost as
many subclasses as items considered in the analysis (Croft, 2006). I will not attempt to construct
such an inheritance hierarchy for all of the word classes discussed here, since it raises a number
of theoretical questions (e.g., Is multiple inheritance permissible?) that go beyond the scope
of the more modest descriptive goals of this section (but see Davis (2001) for a theoretical
approach based on HPSG).
4.2.1 Nouns
The dening feature of nouns is their ability to inect for number and gender. Nouns are an
open class in Rotokas, since numerous words from Tok Pisin have been borrowed into Rotokas
(with varying degrees of phonological transformation as the larger phonological inventory of
Tok Pisin is shoe-horned into the smaller inventory of Rotokas). A handful of Tok Pisin loan
nouns are listed in Table 4.1.
The use of these borrowed nouns is illustrated in (22) and (24). Note that the borrowed noun
tisa teacher occurs with native (i.e., non-borrowed) inectional morphology: the singular
masculine sufx -toa and the indenite sufx -vai (see 5.1).
(22) vovokio=ia
today=LOC
siveri
cement
pura-pa-i-voi
make-CONT-3PL
-PRES
reo
talk
sipo-pa
send-DERIV
kepa=ia
house=LOC
Today theyre laying cement for the telephone building.
39
Noun Gloss Tok Pisin Source
peepa paper pepa
pike gum pike
reeta ladder reta
reviti rabbit reviti
rigi ring ring
sikuru school skul
sipuru spoon sipuru
siveri cement simen
tevoro table tebol
tisa teacher tisa
valusi plane balus
Table 4.1: Rotokas Nouns Borrowed from Tok Pisin
(23) ragai
PPRO.1.SG
sipuru=ia
spoon=LOC
aio
food
toke-pa-ra-i
serve-CONT-1SG
-PRES
kakae
child
vure=pa
ANIM.PL=BEN
I serve food to the children with a spoon.
(24) oisio
COMP
ruipa-pa-i-e
want-CONT-1PL.EXCL-IP
ra
and
tisa-toa-vai
teacher-SG.M-INDEF
ou-pe
get-1PL.EXCL+SUB
ra
and
voeao
PRO.3.PL.M
sikuru-pie-pa-re-ve
school-CAUS-CONT-3SG.M
-SUB
We want to get a teacher to school them [the children].
4.2.1.1 Gender and Noun Subclasses
On the basis of the form of number inection, Rotokas nominals can be broken down into a
number of distinct classes, which are listed below in Table 4.2.
1
Classes 1 and 2 make a semantically motivated distinction between the masculine and fem-
inine based on natural gender (i.e., biological sex). Class 1 nouns may be masculine and/or
feminine, and the distinction between the two genders is preserved for all number categories
(singular, dual, and plural), with the exception of the irregular nouns in 1c. The subclasses of
Class 1 differ from each other in their form of masculine plural marking: Class 1a uses -irara;
Class 1b uses the pluralizer -vure, which is a free form and not a bound morpheme (see 4.2.2
for discussion); and Class 1c uses the pluralizer -ra. Class 2 nouns may also be either masculine
or feminine in the singular and dual, but the distinction is neutralized for the plurali.e., there
1
Firchow (1987:4041) divides Rotokas noun roots into ve classes; however, there are a number of
gender/number-marking sufxes that he does not deal withspecically, the animate (non-human) plural -kare
and the irregular animate plural -vure.
40
Class Class Singular Dual Plural
1a Masculine -to(a) -toarei -irara
Feminine -va -rirei riako
1b Masculine -to(a) -toarei vure
Feminine -va -rirei riako
1c Masculine -to(a) -toarei
-ra
Feminine -va -rirei
2 Masculine -to(a) -toarei
kare
*
Feminine -va -rirei
3 Masculine -to(a) -toarei
-ara 4 Feminine -va -rirei
5 Neuter -a -(a)rei
* the archaic form of kare is ragui
Table 4.2: Noun Classes and Number Inection
is only one form, the pluralizer kare. Classes 3 through 5 have a xed gender. They are formally
distinguishable from Classes 1 and 2 on the basis of their form of plural marking, which is the
sufx -ara.
The assignment of nouns to the various noun classes is largely predictable from semantics.
The nouns in Class 1 are human; the nouns in Class 2 are (non-human) animates; and the nouns
in Class 3, 4, and 5 are almost exclusively inanimate. The role of semantics in noun class
assignment can be seen in the behavior of the noun stem koi pig. When it refers to living pigs,
it occurs in the plural with the pluralizer kare, as illustrated in (25); however, when it refers to
pork, it occurs with the plural sufx -ara, as illustrated in (26).
(25) koie
pig
kare
FP
siku-pa-a-veira
wallow-CONT-3PL
-HAB
rikui=ia
hole=LOC
The pigs wallow in mudholes.
(26) uva
and
evara
DEM.MED.PL.N
rutu
very
oisioa
always
siovara=ia
on=LOC
tou-pa-oro
be-CONT-DEP.SIM
riro-vira
big-ADV
sipei-pa-pe
sweet-CONT-SUB
koie-ara
pig-PL.N
Always being inside of them, the pig meat is very sweet. [Firchow (1974b:81)]
Nominal inection for number/gender is optional in some circumstances. For example,
consider the feminine noun aveke stone. It occurs with the sufx -va in (27) but occurs bare
in (28).
41
(27) riako-va
woman-SG.F
aveke-va
stone-SG.F
peka-e-vo
turn over-3SG.F
-IP
uva
and
rakoru
snake
keke-e-vo
look at-3SG.F
-IP
uva
and
kea-o-e
mistake for-3SG.F
-IP
oisio
as
uo-va
eel-SG.F
The woman turned over the stone and saw a snake but mistook it for an eel.
(28) kaveakapie-vira
insecure-ADV
aveke
stone
tovo-i-vo
place-3PL
-IP
uva
and
kove-o-e
fall-3SG.F
-IP
They placed the stone insecurely and it fell down.

Zero marking is more common for some types of nouns (non-specic and/or inanimate)
than others (human) and some grammatical roles (object) than others (subject). In the case of
noun incorporation, zero marking is obligatory (see 9.2.2). In the following sections, each
noun class is reviewed in greater detail.
4.2.1.1.1 Class 1 The rst class consists of nouns that refer to human beings, as can be
seen from the sample list provided in Table 4.3. It includes various kinship terms, inherently
human nouns, agentive nouns derived from other parts of speech (typically verbs), and various
quasi-human nouns.
2
Class Noun Gloss Notes
Kinship Terms aite father
aako mother
ovii offspring
Human avuka old person
kakae child irregular plural: vure
Derived Agentive ira-pa leader derived from verb ira precede, go ahead
keri-pa enemy derived from verb keri make enemies with
Other tugara spirit
ova mythical dwarf
Table 4.3: Class 1 Nouns
The pattern of inection for Class 1 nouns can be illustrated with the various forms of the
noun stemovii offspring, which occurs in the masculine singular in (29), the feminine singular
in (30), the masculine dual in (31), the feminine dual in (32), the masculine plural in (33), and
the feminine plural in (34).
2
The nouns in this class would be labelled rational according to the terminology used for Dravidian lan-
guages, where a distinction is made between rational (humans, deities) and irrational (animals and everything
else) nouns.
42
(29) aite
father
ovii-to
offspring-SG.M
agesi-pie-re-vere
laugh-CAUS-3SG.M
-NF
The father will make his son laugh.
(30) Vivisori
Vivisori
voki
night
vuuta-ia
time=LOC
takato-pa-ro-e
speak angrily-CONT-3SG.M
-IP
ovii-va=re
offspring-SG.F=ALL
Vivisori spoke angrily to his daughter at night.
(31) Jacob
Jacob
Josepu
Josepu
ovii-toarei-aro
offspring-DL.M-POSS
virako-re-va
bless-3SG.M
-IP
vaiterei=ia
PRO.2.DL.M=LOC
variri-oro
pray-DEP.SIM
Jacob blessed Josephs two boys praying for them.
(32) Uva
and
ovii-rirei
offspring-DL.F
oaesi
PRO.POSS.3.DL.M
aru-pa-si-va
order-CONT-3DL.M-IP
The two of them ordered their two daughters. [C.1]

(33) Tori
Tori
ovii-irara
offspring-HUM.PL
rutu=va
very=COM
ava-ro-e
go-3SG.M
-IP
eisi=re
LOC=ALL
Wakunai
Wakunai
Tori went with all his children to Wakunai.
(34) Raratuiri
Raratuiri
ira
RPRO.3.SG.M
oisoa
always
ovii
offspring
riako-aro
FP-POSS
tarai-pie-pa-re-ve
understand-CAUS-CONT-3SG.M
-SUB
Raratuiri would always teach his daughters.
Whereas ovii offspring is an example of a Class 1 noun stem that can occur in either
the masculine or feminine gender, some Class 1 nouns are inherently gendered and occur in
one gender or the other but not both. For example, the noun stem aite father is inherently
masculine and cannot occur in the feminine gender. Its occurs in the masculine singular in (35),
the masculine dual in (36), and the masculine plural in (37).
(35) kakae-to
child-SG.M
pikopiko-pie-re
whip.RDP-CAUS-3SG.M
aite-to
father-SG.M
uvare
because
kaureo-pa-ro-e
disobedient-CONT-3SG.M
-IP
Father is whipping the boy because he was disobedient.

(36) aite-toarei
father-DL.M
rutu
very
kopii-si-epa
die-3DL.M-RP
oira
PPRO.3.SG.F
virakoi-pie-oro
be.orphan-CAUS-DEP.SIM
uva
and
oira
PPRO.3.SG.F
vaisi-pa-i-veira
call-CONT-3PL
-HAB
oiso
COMP
virakoi-i-va
orphan-3PL
-RP
Both parents died leaving her orphaned and they call her an orphan.
43
(37) vovokio-pa-irara
today-DERIV-HUM.PL
riro
big
kaureo-irara
disobedient-HUM.PL
aite-irara=re
father-HUM.PL=ALL
The people of today are disobedient to their parents.
There are two subclasses that display minor irregularities: Class 1b and Class 1c. Class
1bwhich consists of a single member, the nominal stem kakae childbehaves like a noun
from Class 1 except that its plural marker is an independent word, the free pluralizer vure, as
exemplied in (38). When the modier riro many agrees with the noun kakae vure children
in (38), it takes the expected Class 1 plural sufx -irara.
(38) Kura
Kura
vaio
ANIM.DL
ora
and
Raku
Raku
katai-toarei-vi
one-DL.M-DIM
raga
only
viovoko-toarei
teenager-DL.M
vo-urui-o=ia
SPEC-village-?=LOC
ari
but
riro-irara
big-HUM.PL
rutu
very
kakae
child
vure
HUM.PL
raga
only
Kura and Raku are the only two teenage boys in this village because there are many
children.
(39) oearo-vu
PRO.3.PL.M-ALT
oisoa
always
avui-pa-i-ve
pierce.nose-CONT-3PL
-SUB
voo-va
here=ABL
iruvao-ara
nose-PL.N
kakae
child
vure
PL.M
ora
and
kakae
child
riako
FP.F
Other people would always pierce the noses of the boys and girls.
Class 1c consists of only two members, the nouns oira man and riako woman. Ex-
amples of the singular, dual, and plural form of riako woman are provided in (40) through
(42).
(40) riako-va
woman-SG.F
kakae-to
child-SG.M
roroo-pie-pa-e-vo
nurse-CAUS-CONT-3SG.F
-IP
The woman is nursing the child.

(41) riako-rirei
woman-DL.F
airea
PPRO.RES.3.DL.F
eisi=va
LOC=ABL
urio-ere-i-e
come-3DL.F-EPEN-IP
Kereaka
Kereaka
The two women came from Kereaka.
(42) riako-ra
woman-PL.N
ava-a-e
go-3PL
-IP
sioko
chayote
ou-sia
get-DEP.SEQ
ori-sia
cook-DEP.SEQ
The women went to get chayote in order to cook.
44
Noun Gloss Notes
aatu ying fox
asioko cockroach
atari sh
isike rat
isio spirit believed to reside in the jungle
kaakau dog
kavori craysh
koie pig
kokio bird
koora possum
posiva black ant
4.2.1.1.2 Class 2 The second class of nouns refer primarily to non-human animates (insects,
birds, sh, mammals, etc.), as can be seen from the sample list of Class 2 nouns provided in
Table 4.4.
The noun koie pig exemplies this class of nouns, as can be seen from examples in (43)
through (45), which illustrate its masculine singular, feminine singular, and plural forms.
(43) ragai
PPRO.1.SG
rera-aro
PPRO.3.SG.M-POSS
koie-to
pig-SG.M
ritoko-pa-re-vora
defecate-CONT-3SG.M
-DP
evoa
there
My (male) pig defecated over there.
(44) Ririuto
Ririuto
ora-poisi-ro-e
RR-brace-3SG.M
-IP
koie-va
pig-SG.F
ora-upo-oro
RR-ght-DEP.SIM
aruvea
yesterday
Ririuto braced himself ghting the (female) pig yesterday.
(45) koie
pig
kare
ANIM.PL
urui-a
village-SG.N
vuri
bad
keke-pie-i-vo
look-CAUS-3PL
-IP
va
PPRO.3.SG.N
eri-oro
dig-DEP.SIM
ora
and
ritoko
pig.shit
kou-oro
leave-DEP.SIM
voraro
everywhere
rutu
very
The pigs made the village look bad, digging and defecating everywhere.
The form ragui is an archaic form of the kare which is still found in the speech of some
older speakers. Its use is illustrated below in (46).
(46) paitu
deep
rovu=ia
CL=LOC
oteote
crocodile
ragui
FFP
keke-i-vorao
look at-3PL
-NP
kakae
child
vure
FFP
aruvea.
yesterday
Yesterday the boys looked at crocodiles in the pool.
45
Some Class 2 nouns lack gender/number inection in the singular for one gender, but not the
other. This subclass of nouns consists largely (if not exclusively) of non-human animatesfor
example, the noun rakoru snake has zero marking in the singular feminine, as in (47), but not
in the singular masculine, as in (48). It otherwise behaves like a Class 2 noun, as can be seen
from its plural form in (49).
(47) rakoru
snake
ora-pugo-o-i
RR-roll-3SG.M
-PRES
uvare
because
oira
PRO.3.SG.F
ragi-re-voi
beat-3SG.M
-PRES
Ruruviri
Ruruviri
vurukoa=ia
stick-SG.N=LOC
The snake coiled up because Ruruviri beat her with a stick.
(48) rakoru-to
snake-SG.M
sirava-pa-ro-i
hiss-CONT-3SG.M
-PRES
Tavi=re
Tavi=ALL
The snake is hissing at Tavi.
(49) vo-kaki
SPEC-cave
ua
CLASS
siovara=ia
inside=LOC
rakoru
snake
kare
FPP
tou-pa-i-veira
be-CONT-3PL
-HAB
riro-pa
big-DERIV
kare
FPP
Inside of the hole live many snakes.
4.2.1.1.3 Class 3 The third class of nouns consists almost exclusively of nouns that refer to
inanimate objects, as can be seen from the sample of Class 3 nouns provided in Table 4.3. These
nouns largely refer to things traditionally associated with male culture (e.g., hunting, warfare)
and/or long, thin objects. Aikhenvald (2000:42) observes a similar pattern of classication for
the Manambu (Ndu family), spoken in the East Sepik region of mainland Papua New Guinea,
noting that nouns which denote male humans and higher animates and long and thin objects
are masculine, while those which denote female humans and high animates, and short and round
objects, are feminine.
The noun opita coconut tree is illustrated in its singular and plural form in (50) and (51).
(No example of the dual could be found in the materials available to me.)
(50) kakae-vira
little-ADV
tou-pa-oro
be-CONT-DEP.SIM
roo
DEM.PROX.SG.M
opita-to
coconut-SG.M
pau-ri-va
plant-2SG
-RP
When you were little, you planted this coconut tree.

(51) Kakarera=ia
Kakarera=LOC
uva
and
opita-ara
coconut-PL.N
pau-re-va
plant-3SG.M
-RP
Raupeto
Raupeto
Raupeto planted coconut trees in Kakarera.
46
Noun Gloss Notes
avuo charm belt believed to strenghten and protect children wearing it
govugovu rainbow
kaku ghting club traditionally used in warfare
kato rib
keari long spear traditionally used in hunting possum
kipe scythe used to cut wild grass
koki ear
kupare smoke
opita coconut tree
pakou ghting stick traditionally used in warfare
sigo bush knife
vopa betel nut traditionally used to make pakou ghting stick
4.2.1.1.4 Class 4 The fourth class of nouns refer almost exclusively to inanimate objects,
as can be seen from the partial list of Class 4 nouns provided in Table 4.3. It is unclear what
determines the assignment of inanimate nouns to this class. In general, however, these nouns
tend to refer to tools (bow, axe), containers (basket, pot), and things relating to water (rain, dew,
beach, canoe).
The pattern of inection for gender and number is illustrated by the stem aveke stone,
which is illustrated in the singular (52), dual (53), and plural (54).
(52) riako-va
woman-SG.F
aveke-va
stone-SG.F
peka-e-vo
turn.over-3SG.F
-IP
uva
and
rakoru
snake
keke-e-vo
see-3SG.F
-RP
uva
and
kea-o-e
confuse-3SG.F
-IP
osia
as
uo-va
eel-SG.F
The woman turned over the stone and saw the snake and thought it was an eel.
(53) uva
and
Pauto
God
tavi-ro-iva
tell-3SG.M
-RP
Moses
Moses
airei-vu
TWO-ALT
aire-pa-rirei
new-DERIV-DL.N
aveke-rirei
stone-DL.N
pura-oro
make-DEP.SIM
vairei=va
PRO.3.DL=COM
Pautoa
God
iare
POST
ipa-ro-epa
go up-3SG.M
-RP
pukui-a=ia
mountain-SG.N=LOC
And God told Moses to make two new stones and he went to God with them on the
mountain. [Exodus 34:4]
(54) Pioto
Pioto
ira
RPRO.3.SG.F
aruo-va
mark-SG.F
pura-pa-re-veira
make-CONT-3SG.M
-HAB
aveke-ara=ia
stone-PL.N=LOC
Pioto (a river) always makes a mark on the stones.
47
Noun Gloss Notes
aasi betel nut
aveka beach
aveke stone
evao tree
garoa rattan, cane, vine (generic)
kareko vine
koeta bow
kogo stone axe
opuru canoe
pekuri basket traditional woven variety
pirutu ash ood
pitoka pot traditional clay variety
taetuo childs bow essentially a plaything
4.2.1.1.5 Class 5 The fth class of nouns refer exclusively to inanimate objects, as can be
seen from the partial list of Class 1 nouns provided in Table 4.3.
Noun Gloss Notes
akoro lime
apui ditch
atoi village
raiva road
torara axe generic term (used primarily for gardening)
tetevu sago
voki day
vuku book borrowed from Tok Pisin
vuuta time, space
The neuter inanimate nouns and their pattern of inection is illustrated by the stem urui
village, which is illustrated in the singular (55), dual (56), and plural (57).
(55) Aita=ia
Aita=LOC
tou-pa-i
be-CONT-3PL
urui-a
village-SG.N
oa
RPRO.3.SG.N
vaisi-pa-i
call-CONT-3PL
Kuusi
Kuusi
In Aita theres a village that they call Kuusi.
48
(56) vo-urui-rei
SPEC-village-DL.N
ora-toa-raga-pa-peira
RR-face-just-CONT-HAB
The two villages face each other.
(57) reo-a
talk-SG.N
paru-pie-ri
move-CAUS-2SG
urui-ara
village-PL.N
rutu
very
iare
POST
ra
and
sikua=ia
school=LOC
kovo-sia
work-DEP.SEQ
urio-a-ve
come-3PL
-SUB
Pass the word for everyone to come to work at the school.
Neuter nouns frequently appear without gender/number marking, particularly in the third
person singular, as in (58) and (59).
(58) Teokon
Teokon
urui
village
oa
RPRO.3.SG.N
tou-pa-i
be-CONT-3PL
Wakunai=ia
Wakunai=LOC
ruvara=ia
near=LOC
Teokon village is close to Wakunai.
(59) Ruruvu
Ruruvu
urui
village
arakasi-ei
empty-PRES
rutu
very
viapau
NEG
oira-ra-vai
man-PL.N-INDEF
Ruruvu village is truly empty, there are no people.
Some nouns take the sufx -arei (rather than -rei) to mark the neuter duale.g., vavae
hand, as illustrated in (60).
(60) kakae-to
child-SG.M
vara-vira
low-ADV
voka-pa-re
walk-CONT-3SG.M
aue=ia
CONN=LOC
koko-toarei
leg-DL.N
ora
and
vavae-arei
hand-DL.N
The little boy is walking low on his hands and legs.
4.2.2 Noun Classiers
Nominal classication subsumes a number of distinct grammatical phenomena, including gen-
der markers and noun classiers (Grinevald, 2000; Aikhenvald, 2000). The gender system of
Rotokas was already discussed in the previous section (4.2.1.1). Here the system of noun
classiers in Rotokas will be described.
Grinevald (2000) distinguishes between four types of classiers: numeral classiers, noun
classiers, genitive classiers, and verbal classiers. Of these four types of classiers, all but
verbal classiers are found in the East Papuan languages (Terrill, 2002). However, in Rotokas,
only noun classiers are found, and these consist of two different systems. One system consists
of congurational classiersthat is, classiers that make reference to the shape of the nouns
they classifywhile the other consists of taxonomic classiersthat is, classiers that make
reference to the kind of nouns they classify.
49
Classier Semantic Domain Example
isi round object takura isi egg CLASS
kuio round object (edible) opo kuio taro CLASS
ua narrow object rogara ua sand CLASS
kae long object evao kae tree CLASS
Table 4.8: Shaped-Based Noun Classiers (Firchow, 1987:36)
The shape-based classier system has few members and appears to be a closed class. The
items belonging to this system are provided in Table 4.8.
The classiers in Table 4.8 resemble a gender system, to the extent that the classiers also
occur on modiers of the classied noun, regardless of whether the modier is attributive, as in
(61) and (62), or predicative, as in (63) and (64).
Attributive
(61) gorupasi
strong
isi
CL
rutu
very
karuvera
Singapore
isi
CL
aio-a-voi
eat-1SG
-PRES
I am eating a really strong Singapore fruit.

(62) vearopie
pretty
ua
CL
pua
face
ua
CL
vura-a-vo
look-1SG
-IP
riakova
woman-SG.F
iava
POST
oa iava
hence
oira=pa
PPRO.3.SG.F=BEN
ruipa-ra-e
want-1SG
-IP
I saw the pretty face of the woman and thats why I desired her.
Predicative
(63) riro
big
kuio
CL
rutu
very
vao
DEM.PROX.SG.N
opo
taro
kuio
CL
This taro is a really big taro.
(64) kokovara
unripe
isi
CLASS
opita
coconut
isi
CLASS
The coconut is unripe.
In addition, anaphoric reference to a noun classied by one of these classiers takes the
form of a pronoun co-occurring with the classier, as illustrated in (65) and (66).
(65) Rite=pa
Rite=BEN
opo
taro
isi
CL
oriori-e-voi
cook.RDP-3SG.F
-PRES
aako-va
mother-SG.F
ra
and
va
PPRO.3.SG.N
isi
CL
kae-re-ve
carry-3SG.M
-SUB
vo=re
SPEC=ALL
sikuru
school
Mother scraped taro for Rite and he will carry it to school.
50
(66) Mak
Mark
ira
PPRO.3.SG.M
kiki
ball
isi
CL
gori-re-vo
turn-3SG.M
-IP
va
PPRO.3.SG.N
isi
CL
kiki-oro
kick-DEP.SIM
Mark turned the ball by kicking it.
There is a second classier system which differs from the previous classiers in various
respects. This classier system is not shape-based. Instead, these classiers have a collective
meaning for fairly specic semantic classesfor example, the classier tai refers to a collection
of edible vegetables, such as arua vegetables, ruve aibika, or rereveo wild sugarcane.
A number of these forms are listed below in Table 4.9. It is less clear that these classiers
constitute a closed class; although they are more numerous than the shape-based classiers, no
borrowed forms have been identied to date.
Classier Semantic Domain Notes
kokoo plateful
koota group of rope-like objects
kou heap
kovo garden
ovi liquid
pitu swarm
pota group of at layered objects
rovu body of liquid
tai edible vegetables
tesi group of bamboo tubes
tou container
vasie group of people
viku group of people
vou stranger
Table 4.9: Noun Classiers
Firchow (1987:35-36) describes these classiers under the category of nominal sufxes;
however, this characterization is inaccurate since classiers are not bound to the nouns with
which they co-occuri.e., they can function as the head of a noun phrase, as in (67) and (68).
(67) ruve
aibika
tai
CLASS
ori-e-voi
cook-3SG.F
-PRES
uva
and
riro-vira
big-ADV
ruve-vira
greasy-ADV
irao
INTENS
uvare
because
riro-vira
big-ADV
opita
coconut
kuri-o-i
scrape-3SG.F
-PRES
vo-tai=re
SPEC-CL=ALL
She is cooking aibika and it is very greasy because he is scraping a lot of coconut on it.
51
(68) tatai-va
aunt-SG.F
ruveta
aibika
tai=va
CLASS=COM
kare-o-i
return-3SG.F
-PRES
vo=va
SPEC=ABL
kovo-a
work-SG.N
vo-tai
SPEC-CL
ori-sia
cook-DEP.SEQ
ra
and
va
PPRO.3.SG.N
aio-e-ve
eat-3SG.F
-SUB
Auntie is coming from the garden with the aibika in order to cook it and eat it.
On the basis of Firchows description, Terrill (2002:73) characterizes these forms as spe-
cial pluralizers for different types of objects. These classiers do resemble free (i.e., unbound)
pluralizers (such as riako or vure) in some respects but their differing behavior with respect to
the marking of number speaks in favor of their analysis as separate word classes. For example,
both classiers and free-form pluralizers can function as the head of a noun phrase. This was al-
ready illustrated for the noun classiers in (67) and (68) and is illustrated for the free pluralizers
riako in (69) and kare in (70).
(69) kapoko
kapok
pupupu-ro
cotton-PL.CL
iava
POST
orave-ara
pillow-PL.N
pura-pa-i-veira
make-CONT-3PL
-HAB
vo-riako
SPEC-FP
From kapok cotton the women make pillows.
(70) raageo
green.frog
kare
FP
ou-io-vorao
get-1PL.EXCL-NP
uva
and
vo-kare
SPEC-FP
aio-io-vora
eat-1PL.EXCL-DP
We will get the green frogs and then we will eat them.
Despite their similarities, noun classiers can be distinguished from free-form pluralizers
on the basis of their ability to take number marking.
3
The singular lacks overt number marking,
whereas the dual is marked by -rei and the plural by -ro. For example, the classier kuio occurs
with dual marking in (71) and the classier kou occurs with plural marking in (72).
(71) evo
DEM.MED.N
kuio-rei
CLASS-DL.CL
oarea
RPRO.3.DL.N
pau-re-va
plant-3SG.M
-RP
Those were the two (taro) that he had planted. [Caleb, Matevu]
(72) Kavi
Kavi
iria
PPRO.3.SG.F
isisio
grass
kou-ro
CLASS-PL.CL
guruguru-pa-e-voi
gather.RDP-CONT-3SG.F
-PRES
vara
PPRO.3.PL.N
kasi-sia
burn-DEP.SEQ
Kavi is gathering all of the grass in order to burn it.
3
Firchow (1987:4748) treats classiers and free pluralizers as a single class, but given that they behave dif-
ferently with respect to number marking, they must be distinguished. It is likely, however, that classiers are
the diachronic source of the pluralizers, according to a scenario where number marking on commonly occurring
classiers is lost and the classier comes to have inherent plural semantics.
52
Some inanimate nouns do double duty as classiers (4.2.2). For example, kovo garden
shows the expected pattern of inection for an inanimate noun, as in (73). However, kovo also
functions as a classier for types of gardens (taro, cocoa, etc.), as illustrated by (74).
(73) Raku
Raku
tuuke-re-vo
punish-3SG.M
-IP
sipito
chief
uva
and
kovo-ara
garden-PL.N
tori-re-vorao
run-3SG.M
-NP
The chief punished Raku because he ran away from the gardens.
(74) kakau
cocoa
kovo-ro
CLASS-PL.N
pura-pa-i-voi
make-CONT-3PL
-PRES
oira-ra
man-PL.N
moni
money
ou-pa-sia
get-CONT-DEP.SEQ
Men make cocoa gardens in order to get money.
Classied nouns behave like neuter nouns with respect to subject agreement, as can be seen
from (75) and (76), where classied nouns play the role of subject and show zero agreement on
the verb. In addition, classied nouns co-occur with the form of the subjunctive mood normally
found with neuter subjects (-pe), as in (75) (see 5.2.2.7.2).
(75) kokovara
unripe
isi
CLASS
opita
coconut
isi
CLASS
viapau
NEG
erako-pa--pe
dry-CONT-3SG.N-SUB
The unripe coconut isnt dry.
(76) gaegaere-vira
drift-ADV
roko--voi
go down-3PL.N-PRES
opita
coconut
isi
CLASS
uuko-va=ia
water-SG.F=LOC
The coconuts are going drifting down the water.
4.2.3 Pronouns
In Rotokas, there are four different pronominal paradigms: personal pronouns (4.2.3.1), re-
sumptive pronouns (4.2.3.2), possessive pronouns (4.2.3.3), and demonstrative pronouns (4.2.3.4).
Each will be described in turn.
4.2.3.1 Personal Pronouns
The most basic and commonly occurring pronouns are the personal pronouns. The personal
pronouns are sensitive to person (rst, second, third), number (singular, dual, plural), and gen-
der (masculine, feminine, and neuter), as well as clusivity (inclusive vs. exclusive). The full
paradigm is provided in Table 4.10.
Table 4.10 provides no segmentation of the personal pronouns since no productive segmen-
tation appears to be possible. For example, the rst personal plural inclusive might be analyzed
as the second person singular plus the rst personal plural exclusive. However, if this were an
instance of productive concatenation of morphemes, the vowel of the rst syllable (vi) should
be long (cf. 5.4.1). It is therefore more likely that historically the rst person singular plural
53
Number
Person Singular Dual Plural
1
Incl.
ragai vegei
vigei
Excl. igei
2 vii vei visii
3
M rera vaiterei voea
F oira vairei vairo
N va varei vara
Table 4.10: Personal Pronoun Paradigm
exclusive arose from the fusion of the second person singular and the rst person plural exclu-
sive. Comparative data from the other languages in the Rotokas family may shed some light on
the diachronic origins of the paradigm. In Konua, for example, the rst person plural inclusive
personal pronoun bioga cannot simply be analyzed as the concatenation of the second person
singular and the rst person plural exclusive since the second person singular is biru or bira and
the rst person plural exclusive is ioka (M uller, 1954; Ross, 2001).
The paradigmatic structure for person marking in the pronoun paradigms is somewhat in-
teresting from a typological perspective. Although a clusivity distinction is found in the rst
personal plural, it is neutralized in the rst person dual, as illustrated by (77) and (78). As the
Tok Pisin translations provided by consultants underscore, the addressee is included in (77) but
excluded in (78), yet the same pronoun, vegei, is used in both cases.
(77) ragai=pa
PRO.1.SG=BEN
viru
move
ra
and
vegei
PRO.1.DL
rutu
very
pau-ve
sit-1DL
Move for me and well sit down./Yu surik bai yumi tupela wantaim sindaun.
(78) vei
PRO.2.PL
rogo
begin
rovo-pa-si-ei
start-CONT-2DL-PRES
ikau-oro
run-DEP.SIM
ra
and
vegei
PRO.1.DL
utu-pa-vira
follow-DERIV-ADV
ikau-veare
run-1DL+NF
You two start rst and the two of us will follow running./Yutupela bai stat ron pastaim na
bihain bai mitupela i ron.
Table 4.2 uses the analytical scheme employed in Cysouw (2003)s cross-linguistic survey
of paradigmatic structure to represent the Rotokas pronominal system.
According to Cysouw (2003), this type of congurationwhere there is hymophony
(i.e., a neutralization across cells) along the vertical dimensionis fairly uncommon cross-
linguistically but is nevertheless attested in the literature. Cysouw (2003:218-219) cites ve
other languages that show a similar pattern: the Australian language Kuku-Yalnji (Oates and
Oates, 1964:7); the Tibeto-Burman language Jiaron (Bauman, 1975:131-132,276); and three
54
Group Restricted Group
vigei
vegei
1+2(+3)
1 ragai igei 1+3
2 vii visii vei 2+3
3 rera / oira / va voea vaiterei 3+3
Figure 4.2: Paradigmatic Structure of Personal Pronouns
Papuan languagesTuaripi (Wurm, 1975b:515), Guhu-Samane (Richard, 1975:781), and Ko-
rafe (Farr and Farr, 1975:734-735).
Pronouns are invariant in form across grammatical roles (unlike, for example, English
pronounsI vs. me). This holds true for all of the pronominal paradigms, but can be most
easily illustrated with personal pronouns due to their high frequency of occurence. Therefore,
in examples (79) through (86), the pronoun ragai I, me remains invariant in form despite the
fact that it plays varying grammatical roles.
4.2.3.1.1 S
(79) ragai
PPRO.1.SG
katokato-to
black-SG.M
Im a black man.
(80) ragai
PPRO.1.SG
kasipu-ra-i
angry-1SG
-PRES
Im angry.
4.2.3.1.2 A
(81) ragai
PPRO.1.SG
vii
PPRO.2.SG
ita
again
ou-a-voi
get-1SG
-PRES
Now Im marrying you. [Firchow and Akoitai (1974:71)]

(82) ragai
PPRO.1.SG
vo-siposipo
SPEC-story
pura-a-voi
make-1SG
-PRES
Tarui
name
I, Tarui, am telling this story. [Firchow and Akoitai (1974:36)]
4.2.3.1.3 O
(83) ari
but
eera
DEM.PROX.3.SG.M
raga
only
ragai
PPRO.1.SG
tauva-re-vo
help-3SG.M
-IP
But only this one helped me. [Firchow and Akoitai (1974:53)]
55
(84) uvare
because
ragai
PPRO.1.SG
tavi-irao-re-va
tell-INTEN-3SG.M
-RP
eera
DEM.PROX.3.SG.M
masta
white man
Because he really talked to me. [Firchow and Akoitai (1974:19)]
4.2.3.1.4 Oblique
(85) ava-u
go-2SG
ragai=pa
PPRO.1.SG=BEN
uukoa-vai
water-INDEF
ou-sia
get-DEP.SEQ
You go get water for me. [Firchow and Akoitai (1974:64)]
(86) ragai=re
PPRO.1.SG=ALL
keera-ro-epa
beckon-3SG.M
-RP
oisio
COMP
uro-u-vere
come-2SG
-NF
He beckoned to me, Come here.
4.2.3.2 Resumptive Pronouns
Firchow (1987) recognizes a second pronominal paradigm, whose members he labels relative
pronouns. The full paradigm is provided below in Table 4.11.
Number
1
Incl.
ragao vegoa
vigoa
Excl. igoa
2 vigoa veigoa visiigoa
3
M ira aiterea oea
F iria airea airoa
N oa oarea oara
Table 4.11: Resumptive Pronouns
Firchow (1987)s characterization of these pronouns as relative is based on the fact that
they are used to form relative clauses, as illustrated in (87) and (88).
(87) Aita=ia
Aita=LOC
tou-pa-i
be-CONT-PRES
urui-a
village-SG.N
oa
RPRO.3.SG.N
vaisi-pa-i
call-CONT-PRES
Kuusi
Kuusi
In Aita theres a village that they call Kuusi. [=(55)]
(88) Gara
Gara
uuko-va
river-SG.F
vaisi-aro
name-POSS
iria
PPRO.3.SG.F
tou-pa-i-veira
be-CONT-3PL
-HAB
eisi
LOC
Sisisivi=ia
Sisisivi=LOC
ruvara=ia
near=LOC
Gara is the name of the river that is close to Sisivi.
56
These pronouns agree in person, number, and gender with the head noun: oa agrees with the
third person singular neuter noun uruia village in (87) and iria agrees with the third person
singular feminine noun uukova river in (88).
Given that this pronominal paradigm includes local persons (i.e., rst and second person),
their characterization as relative pronouns is questionable, and the term resumptive pronoun
will be used instead. The resumptive function of these pronouns is illustrated in (89) and (90).
In these examples, a topic is rst established and subsequent references to it are then made using
a resumptive pronoun.
(89) kapokarito
tree
ira
RPRO.3.SG.M
epao
exist
vavo
there
Rarova
Rarova
ira
RPRO.3.SG.M
iava
POST
vavurupa-ara
root-PL.N
ou-a-vorao
get-1SG
-NP
The tree that is in Rarova, I got roots from it.

(90) utave-va
shell-SG.F
Kiki
Kiki
oira-aro
RPRO.3.SG.F-POSS
iria
RPRO.3.SG.F
kavu-re-va
leave behind-3SG.M
-RP
eisi
LOC
Ruruvu=ia
Ruruvu=LOC
iria
RPRO.3.SG.F
oisioa
always
vuvure-pa-re-ve
blow-CONT-3SG.M
-SUB
Kikis shell, the one he would always blow, he left it in Rururvu.
Local person (i.e., rst and second person) resumptive pronouns are fairly rare. They are
illustrated in (91) and (92).
(91) viovoko
teenager
riro
big
vatasioko-to
vagabond-SG.M
vii
PPRO.2.SG
vigoa
RPRO.2.SG
viapau
NEG
oisio
COMP
katai
one
urui-va=ia
village-SG.F=LOC
ora-tou-pie-pa-u-veira
RR-be-CAUS-CONT-2SG
-HAB
Boy, youre a vagabond, you dont stay put in one village.
(92) riro
big
kavikaviru-irara
steal.RDP-HUM.PL
visii
PPRO.2.PL
visigoa
RPRO.2.PL
atari
sh
kare
FP
kaviru-ta-vora
steal-2PL-DP
Youre big thieves, you stole the sh.

Topicalized nouns occur at the left-most boundary of the sentence and subsequent reference
to them takes the form of resumptive pronouns that agree with them in person, number, and
gender. These resumptive pronouns occur in situ, as illustrated in (93) through (96), which
illustrate topicalized nouns serving a variety of grammatical roles.
S
(93) sigo-a
knife-SG.N
vii
PRO.2.SG
va-aro
PRO.3.SG.N-POSS
oa
RPRO.3.SG.N
asikauru-era
rust-DP
Your knife, it rusted.

57
A
(94) kauo-va
aunt-SG.F
iria
PRO.3.SG.F
upiriko
sweet potato
kovo
garden
aruo-pa-e-voi
weed-CONT-3SG.F
-PRES
Your auntie, she is weeding the sweet potato garden.

O
(95) Resi
Resi
iria
RPRO.3.SG.F
agoagoto-raga-pa-re-vo
atter.RDP-ONLY-CONT-3SG.M
-IP
Voipiri
Voipiri
As for Resi, Voipiri is attering him.
Oblique
(96) koeta
bow
iria=ia
RPRO.3.SG.F=LOC
koora
possum
ritaa-pa-a-veira
shot-CONT-3PL
-HAB
ora
and
aue
CONN
tapo
also
kokio
bird
With a bow they shoot possums and birds.
4.2.3.3 Possessive Pronouns
Possessive pronouns are those that substitute for possessors. The full paradigm for the posses-
sive pronouns is provided in Table 4.12.
4
1
Incl.
oaa oave
oavi
Excl. oaio
2 oara oaesi oavisi
3
M oaro oaesi
oaive
F oo oaere
Table 4.12: Possessive Pronouns
Unlike other pronominal paradigms, the possessive pronouns lack a category for the third
person neuter (Firchow, 1987). When the possessor of a noun is neuter, the only option for mark-
ing possession is the possessive sufx -aro, which occurs on the possessed noun (see 5.1.2.3),
as illustrated in (97) and (98).
(97) rasi-a
ground-SG.N
vaisi-aro
name-POSS
oa
RPRO.3.SG.N
vaisi-pa-i
call-CONT-3PL
oisio
COMP
Aperaipa
Aperaipa
The name of the place, they call it Aperaipa.
4
Firchow (1987) provides the form oae for the third person plural.
58
(98) kakau-ara
cocoa-PL.N
vavata-aro
weight-POSS
ate-pa-i-vo
weigh-CONT-3PL
-IP
oiso
COMP
ra
and
vara=IA
PRO.3.PL.N-ABL
vori
money
ou-a-ve
get-3PL
-SUB
They weighed the cocoa so that they could get money from them.
Possessive pronouns agree in person, number, and gender with their possessors, and follow
them, as illustrated in (99) through (101).
(99) kepa
house
oaive
PPRO.3.PL
eva
DEM.3.SG.N
oa
RPRO.3.SG.N
vura-pa-ri
look at-CONT-2SG
Thats their house that youre looking at.

(100) Joseph
Joseph
ira
RPRO.3.SG.M
kovo-pa-ara
work-DERIV-PL.N
oaro
PPRO.3.SG.M
guru-re-voi
gather-3SG.M
-PRES
Joseph is gathering all of his tools.

(101) kepa
house
oaio
PPRO.1.PL.EXCL
eva
DEM.3.SG.N
oa
RPRO.3.SG.N
iare
POST
ava-pa-vi-ei
go-CONT-1DL-PRES
Thats our house which were going into.

4.2.3.4 Demonstratives
Demonstrative pronouns are deictic words that indicate which entities a speaker refers to, and
distinguishes these entities from others (Anderson and Keenan, 1985). The demonstrative pro-
nouns in Rotokas encode three levels of distance: proximal, medial, and distal. The range of
spatial deixis associated with these three categories is characterized in Firchow (1987:43) as
follows: demonstrative pronouns are sub-classied according to the position of the referent
in relation to the speaker, i.e., referent near at hand, referent at a distance, and referent removed
or out of sight. The full paradigm is given in Table 4.13.
These forms may appear to be amenable to further segmentation, along the lines shown in
Table 4.14, since the proximal forms consistently end with o, the medial forms consistently
begin with e, and the distal forms consistently end with ri.
If the common elements for each level of distance are treated as afxes (proximal, -o; me-
dial, e-; distal, -ri), we would expect their hosts to be consistent in form across the levels of
distance. However, the base forms obtained by segmenting out the hypothetical afxes are not
internally consistent. The third person singular masculine and feminine are irregular for all
levels of distance. There is also some irregularity in the masculine and feminine dual proximal
as well as the medial third person plural. There is also no consistent correspondence between
the base forms and any other pronominal paradigm. For example, the third person singular
neuter has a consistent base form across the three levels of distance (va), which corresponds
to the third person singular of the personal pronoun paradigm. But the same cannot be said
59
Number
Distance Gender Singular Dual Plural
Proximal M roo vaitereo voeao
F oo vaireo vairoo
N vao vareo varao
Medial M eera evaiterei evoeao
F eira evairei evairo
N eva/evo evarei evara
Distal M roari vaitereiri voeari
F oari vaireiri vairori
N vari vareiri varari
Table 4.13: Demonstrative Pronoun Paradigm
Number
Distance Gender Singular Dual Plural
Proximal M ro vaitere voea
-o F o vaire vairo
N va varei vara
Medial M
e-
era vaiterei voeao
F ira vairei vairo
N va varei vara
Distal M roa vaiterei voea
-ri F oa vairei vairo
N va varei vara
Table 4.14: Hypothetical Segmentation of Demonstrative Pronoun Paradigm
for the third person singular masculine or feminine. The third person singular feminine distal
appears to be based on the third person singular neuter resumptive pronoun while none of the
third person masculine singular base forms correspond to any other pronominal paradigm. The
demonstratives will therefore be treated as unanalyzed forms in all interlinear glossing.
Demonstratives can occur in isolation as pro-forms, as illustrated for the medial third person
masculine in (102) and the medial third person feminine in (103).
(102) aure
yes
ari
but
eera
DEM.MED.SG.M
ava-ro-e
go-3SG.M
-IP
vokipaua
morning
rutu
very
Yes, but that one went in the early morning.
(103) eira
DEM.MED.SG.F
veu-pa-o-i
be.angry-CONT-3SG.F
-PRES
uvare
because
oira=re
PPRO.3.SG.F=ALL
60
reo-a-e
speak-3PL
-IP
That one is angry because (of the way) they talked to her.
Demonstratives also co-occur with nouns, serving as modiers, as in (104), where the me-
dial third person masculine demonstrative eera occurs with the masculine noun oirato man,
or (105), where the medial third person feminine demonstrative eira occurs with the feminine
noun aiopava ashlight.
(104) eera
DEM.MED.SG.M
oira-to
man-SG.M
riro
big
kaviru-to
steal-SG.M
This man is a big thief.
(105) Savuko
Savuko
oira-aro
PPRO.3.SG.F-POSS
eira
DEM.MED.SG.F
aopa-va
ashlight-SG.F
That ashlight is Savukos.
4.2.4 Verbs
The dening feature of verbs is their ability to inect for person and tense/aspect/mood. For
example, the verb vurivuri move back and forth in (106) shows third person subject agreement
(-ro) and is marked for the remote past (-epa).
5
(106) uva
and
ora-viruviru-raga-pa-ro-epa
RR-move.RDP-ONLY-CONT-3SG.M
-RP
ragai
PPRO.1.SG
uriri-pa-oro
frighten-CONT-DEP.SIM
He just moved himself back and forth frightening me.
Verbs are an open category in Rotokas, judging from the increasing amount of borrowing
from Tok Pisin that occurs in the language. It seems, however, that the rate of verb borrowing
in Rotokas lags behind that of noun borrowing. Table 4.15 lists a few of the more commonly
heard Rotokas verbs that have been borrowed into Rotokas from Tok Pisin.
The use of borrowed verbs is illustrated in (107) and (108).
(107) viapau
NEG
ragai
PPRO.1.SG
rootu-pa-ra-era
attend church-CONT-1SG
-DP
I wasnt going to church in the past.

(108) oire
okay
ora-agesi-pie-a-i
RR-smile-CAUS-3PL
-PRES
voa=va
here=ABL
sikeari-a-epa
shake hands-3PL
-RP
Okay, theyre smiling at each other and shook hands. [RR-Ata:49]

5
Although both independent and dependent verbs can occur with the continuous sufx -pa, it is an equivocal
diagnostic for verbhood, since a homophonous sufx occurs on a number of other parts of speech (e.g., derived
nouns and adverbs).
61
Rotokas Verb Tok Pisin Source
iusi use yusim
kiria clarify kiliaim
pereri befriend peren
rootu worship lotu
sekari shake hands sekhan
siku attend school skul
Table 4.15: Rotokas Verbs Borrowed from Tok Pisin
Verbal subject agreement and tense-marking in Rotokas can be divided into two formally
distinct classes, which will be labelled and in order to avoid prejudging the question of
what motivates the distinction, and verbs can be classied according to which form of person
agreement they take. For example, the verb uusi sleep is whereas the verb upo hit is
. The contrast is illustrated in (109) and (110), where the two verbs show distinct patterns of
verbal inections despite the fact that they both have third person plural subjects and occur in
the present tense indicative.
(109) kakae
child
vure
FP
uusi-pa-a-i
sleep-CONT-3PL
-PRES
koke-va
rain-SG.F
rero-aro
underneath
The children were sleeping under the rain.
(110) uva
and
vii
PPRO.2.SG
upo-pa-i-voi
hit-CONT-3PL
-PRES
oira=pa
PPRO.3.SG.F=BEN
eira
DEM.MED.SG.F
riako-va
woman-SG.F
They are hitting you because of the woman.
Although some verbs are ambivalent (see 9.1.1), most verbs belong uniquely to one of
the two classes, and can therefore be described as or . This assignment is systematically
affected by valency-changing derivations (see Chapter 9). Since this topic is the chief concern
of this thesis and is described in considerable detail in the second part of the thesis, it will not
be discussed in detail here. For a description of verbal morphology, and an inventory of forms,
see 5.2.
4.2.5 Adjectives
There has been a great deal of typological interest in the universality of adjectives (Dixon,
1982, 2004). In typological treatments of adjectives, a distinction is usually drawn between two
functions of adjectives: attributive and predicative (Stassen, 1997). Attributive adjectives serve
to modify the heads of noun phrases whereas predicative adjectives serve as the predicates of
62
clauses. For example, in Rotokas, the stem riro big can function either attributively, as in
(111), where it modies the noun aveke stone, or predicatively, as in (112), where it is the
main predicate and occurs with person/number/gender and tense/aspect/mood marking.
(111) aite-to
father-SG.M
riro-va
big-SG.F
aveke
stone
kae-pie-re-vo
lift-CAUS-3SG.M
-IP
aruvea
yesterday
Dad lifted a large stone yesterday.
(112) oira-ra
man-HUM.PL
gorupasi-vira
strong-ADV
riro-pa-a-i
big-CONT-3PL
-PRES
vovokio=ia
today=LOC
People grow up strong today.
Sentences such as (111) and (112) provide no good evidence of a distinct grammatical cat-
egory of adjectives, since rirova big and riroparoi be big can simply be analyzed as noun
and verb, respectively. Such an analysis would be more parsimonious, since it does not require
the postulation of any new word classes, and places the burden of explanation on a mapping
between root and/or stem classes on the one hand and word classes on the other. This is already
an issue for Rotokas due to the existence stems such as atari sh, which indifferently function
as noun or verb (i.e., without recourse to explicit derivational morphology). Although there is
little evidence of a distinct grammatical category of adjective in Rotokas, it is worthwhile to
examine the way in which predication and attribution are expressed grammatically in Rotokas.
4.2.5.1 Predication
Stassen (1997:13) observes that there are four classes of predication, listed in (113), which
languages carve up differently.
(113) a. Event Predicate Joanna rides.
b. Property or quality predicate Joanna is strong.
c. Class predicate Joanna is a ne horse-woman.
d. Locational predicate Joanna is in the stable.
In Rotokas, event, property or quality, and locational predicates take the form of verbs or
verb phrases, as illustrated in (114) through (116).
4.2.5.1.1 Event Predicate
(114) ragai
PPRO.1.SG
roru-pa-oro
be.happy-CONT-DEP.SIM
kauo-pa-ra-i
jump-CONT-1SG
-PRES
I am jumping with joy.

63
4.2.5.1.2 Property or Quality Predicate
(115) riro
big
goru-pa-i
strong-CONT-PRES
rakari-a
skin-SG.N
The skin is really strong.
4.2.5.1.3 Locational Predicate
(116) oovato
red earth
ira
RPRO.3.SG.M
voo
here
tou-pa-re-veira
be-CONT-3SG.M
-HAB
Tutupaio
Tutupaio
kaki-a
cave-SG.N
siovara=ia
inside=LOC
Red earth is found inside a cave in Tutupaio.
Class predicates, on the other hand, take the form of nouns, as in (117), where the subject
occurs in its usual position, or (118), where the subject is right-dislocated (see 6.2.1). In both
cases, the class predicate is a noun and the subject and predicate are simply juxtaposedi.e.,
there is no copula.
(117) Raviata
Raviata
vearo-pie
good-CAUS
koie
pig
ragui-ro
CL-PL.N
toki-pa-to
care for-DERIV-SG.M
Raviata is a good animal caretaker.
(118) gavaure-a
nice-SG.N
vao
DEM.PROX.SG.N
voki-a
day-SG.N
Today is a nice day.
4.2.5.2 Attribution
The situation is less straightforward where attribution is concerned. There is a formal distinc-
tion between two classes of stems in the case of attribution: those that can directly function
attributively and those that require the sufx -pa to do so. Bivalent verbs (see Chapter 8) sys-
tematically take the sufx -pa when they modify nouns, as illustrated in (119) and (120).
(119) kokotu
chicken
takura-aro
egg-POSS
sipo-sia
send-DEP.SEQ
ava-ro-e
go-3SG.M
-IP
Wesli
Wesley
eisi
LOC
uu-pa
meet-DERIV
tapi
place
Wesli went to sell chicken eggs at market.
(120) Rari
Rari
kotokoto
cargo
ou-sia
get-DEP.SEQ
ava-ro-e
go-3SG.M
-IP
eisi
LOC
Buka
Buka
aio
food
kitu-pa
store-DERIV
kepa
house
iare
POST
Rari went to the store (lit., food-storing house) in Buka in order to get cargo.
Monovalent verbs are split in this respect. For example, the stem uteo cold directly mod-
ies the noun tapi place in (121) and vuuta time, space in (122).
64
(121) Sisivi-a
Sisivia
riro
big
uteo
cold
tapi
place
rutu
very
Sisivi is a very cold place.
(122) o-vuuta-a
SPEC-time-SUB
eva
DEM.MED.SG.N
riro
big
uteo
cold
vuuta
time
It was winter. (Lit., That time was a very cold time.) [John 10:23]
It also functions predicatively as a verb stems, as in (123), where it describes a property
of the environment (being cold), and (124), where it describes the feelings of a human agent
(feeeling cold).
(123) kasirao-vira
hot-ADV
uusi-ra-e
sleep-1SG
-IP
vokiaro,
night
viapau
NEG
riro-vira
big-ADV
uteo-pa-e
cold-CONT-IP
Last night it was really hot sleeping because it wasnt very cold.
(124) Vivura
Vivura
ora-raku-ro-i
RR-cover-3SG.M
-PRES
varo-a=ia
clothing-SG.N=LOC
uvare
because
riro-vira
big-ADV
uteo-pa-ro-i,
cold-CONT-3SG.M
-PRES
uvare
because
koke-va=ia
rain-SG.F=LOC
kare-ro-e
return-3SG.M
-PRES
eisi=va
LOC=ABL
kovo-a
garden-SG.N
Vivura covered up with a jacket because he was really cold because he returned from the
garden in the rain.
However, the stem aire new requires the sufx -pa in order to modify a noun, as in (125),
where it modies the noun kepa house, or (126), where it modies the noun kovovai some
garden. It does not appear to be able to function as a verbal predicate (at least no examples of
such usage are attested anywhere in the materials available to the author).
(125) Kokora
Kokora
ira
PRO.3.SG.M
aire-pa
new-DERIV
kepa
house
pau-pa-re
build-CONT-3SG.M
Kokora is building a new house.

(126) vego-a
bush-SG.N
toe-pa-i
cut-CONT-3PL
oira-ra
man-PL.N
aire-pa
new-DERIV
kovo-vai=re
garden-INDEF=ALL
The people are cutting the bush for the new garden.
4.2.6 Adverbs
Adverbs represent a large and somewhat disparate class of elements in Rotokas that serve as
nonsubcategorized modiers (adjuncts). As Butt et al. (1999:133) observe, Adverbs vary so
65
considerably with regard to syntactic distribution and semantic content that the grammatical
category of adverb is often used as a kind of catch-all category for lexical items that one is at a
loss to dene.
The Rotokas lexicon possesses a large number of adverbs due to the productivity of the
sufx -vira, which derives adverbs from other parts of speech. Many different types of adverbs
can be derived with -vira: sentential, as in (127); directional, as in (128); degree, as in (129);
manner, as in (130); and time, as in (131).
Sentential
(127) sirao-vira
pity-ADV
rutu
very
uvare
because
aako
mother
upo-ri-voi
hit-2SG
-PRES
Sadly, you killed my mother. [Firchow and Akoitai (1974:80)]

Directional
(128) iipa-vira
ascend-ADV
iipa-u
ascend-2SG
voo=re
here=ALL
You come up here.
Degree
(129) vioro
ripe
varavara-vira
near-ADV
tou-pa-i
be-CONT-PRES
It is nearly ripe.
Manner
(130) aavu-va
grandparent-SG.M
gapu-vira
naked-ADV
sisiu-pa-o-i
bathe-CONT-3SG.F
-PRES
eisi
LOC
Ivitu
Ivitu
Grandmother is bathing naked in Ivitu (a river near the village of Togarao).
Temporal
(131) voki-pa-vira
day-DERIV-ADV
ava-pa-ra-i
go-CONT-1SG
-PRES
Togarao
village name
iare
POST
Im going to Togarao tomorrow.
There are also interrogative adverbs, as illustrated in (132).
(132) aavio-pa-vira
when-DERIV-ADV
ora-vasike-pa-u
RR-leave-CONT-2SG
eisi-re
LOC=ALL
Togarao
Togarao
When are you going to Togarao?
66
Time Word Gloss
tuariri long ago
aruvea yesterday
vokipaua morning
vokipakou early morning
vokiarovi afternoon
vokiaro night
ovoiaro afternoon
ovoiarovi late afternoon
Table 4.16: Rotokas Time Words
There are a number of words that do not take the sufx -vira but could nevertheless be
classied as adverbs to the extent that they serve as adjunct modiers. In other words, these
words play an adverbial role but lack any overt morphology indicating their part of speech (i.e.,
the sufx -vira). A number of these terms are listed in Table 4.16.
Like adverbs derived with -vira, these time words are exible in their ordering, although
they generally occur at the sentence peripheryi.e., at the beginning of sentences, as illustrated
by (133), or at the end, as in (134).
(133) aruvea
yesterday
rokoroko
frog
kare
FP
keke-io-vo
look.at-1PL.EXCL-IP
Yesterday we looked at frogs.

(134) Raratuiri
name
ragai-re
PPRO.1.SG=ALL
kasipu-ro-e
angry-3SG.M
-IP
aruvea
yesterday
Raratuiri was angry with me yesterday.
Note that aruvea yesterday occurs bare in (133) and (134); it cannot in fact take oblique
marking. Some time words can occur with peripheral marking. For example, vovokio today
occurs with the oblique marker =ia in (136), and would therefore be analyzed as a noun rather
than as an adverb.
(135) riro-a
big-SG.N
kopii-a
die-SG.N
tou-pa-i-voi
be-CONT-3PL
-PRES
vovokio=ia
today=ENC
Serious illness exists today.
(136) vovokio=ia
today=LOC
oira-ra
man-HUM.PL
uu-pa-a-i
meet-CONT-3PL
-PRES
ora-reo-sia
RR-talk-DEP.SIM
Today people are meeting to talk.
There are two other words that also serve as adjunct modiers to verbs and can therefore be
characterized as adverbs: rutu very, illustrated in (137), and riro big, illustrated in (138).
67
(137) Ruruvu
Ruruvu
urui
village
arakasi-ei
empty-PRES
rutu
very
Ruruvu village is truly empty.
(138) itoo-va
banana-SG.F
riro
big
vavata-pa-o-i
heavy-CONT-3SG.F
-PRES
The bananas are very heavy.

These two elements also serve to modify other adverbs: riro big precedes the adverb
it modies in (139) and rutu very follows the adverb it modies in (140). The two can
simultaneously modify a single adverb, as in (141).
(139) riro
big
kaekae-vira
long-ADV
pau-ra-e
sit-1SG
-IP
uva
and
asisoe-ra-i
numb-1SG
-PRES
I sat down for a long time and now Im numb.

(140) ovoio-vira
last-ADV
rutu
very
kare-ra-e
return-1SG
-IP
atoia=re
village=ALL
uvare
because
ragai
PPRO.1.SG
kavu-i-vo
leave-3PL
-IP
I returned to the village last because they left me.

(141) Asiravi
Asiravi
riro-va
big-SG.F
riako-va
woman-SG.F
iria
PPRO.3.SG.F
riro
big
patura-vira
fat-ADV
rutu
very
tou-pa-e-veira
be-CONT-3SG.F
-HAB
Asiravi is a big woman who is really fat.
The modier rutu provide some evidence for categorizing time words with adverbs, since
it occurs with adverbs, as shown above, as well as time words, as can be seen from (142) and
(143).
(142) vokiaro
night
rutu
very
pou-io-viro
arrive-1PL.EXCL-COMPL
eisi=va
LOC=ABL
vara-vira
descend-ADV
Late at night we arrived from above.
(143) aure,
yes
ari
but
eera
DEM.MED.SG.M
ava-ro-e
go-3SG.M
-IP
vokipaua
morning
rutu
very
Yes, but he went in the morning.
68
4.2.7 Postpositions
There is a class of postnominal modiers which Firchow (1987) labels relator particles due to
the fact that they are used to mark the semantic relation of the nominal with which they occur.
These modiers are analyzed here as postpositions, which can be divided into two subclasses
by phonological weight: monosyllabic, illustrated in (144) and (145), and polysyllabic, illus-
trated in (146) and (147). Due to phonological constraints on stress assignment that require the
minimal phonological word to be a foot (see 3.2.2), the monosyllabic postpositions behave as
clitics while the multisyllabic postpositions are able to act as words.
Monosyllabic
(144) Vago
Vago
aapaapau-vira
visit.RDP-ADV
ava-ro-era
go-3SG.M
-DP
eisi=re
LOC=ALL
Kieta
Kieta
Vago went visiting to Kieta.
(145) Eravaa
Mt.Balbi
iare
POST
kakae
child
vure=va
FP=ABL
iipa-ro-era
go.up-3SG.M
-DP
Paravi
Palavi
evoa
there
voka-sia
walk-DEP.SEQ
Palavi went on top of Mt. Balbi with the children and theyre going walking.
Polysyllabic
(146) ava-pa-ra-i
go-CONT-1SG
-PRES
ragai
PPRO.1.SG
vo-kepa-aro
SPEC-house-POSS
iare
POST
I am going home (literally: to my house).
(147) uva
so
uusi-ro-epa
sleep-3SG.M
-RP
ovi-toa
son-SG.M
tapo
also
urua=ia
bed=LOC
So he slept with the son in bed. [Firchow and Akoitai (1974:50)]
The full list of these forms is provided below in Table 4.17.
6
The glosses provided in Table 4.17 are for the most typical meaning of a particular form
and are therefore not exhaustive. Many of these forms are polysemous and mark more than one
semantic role, as shown in Figure 4.3, making it difcult to nd a satisfactory gloss in some
cases.
Two of the polysllabic postpositions appear to be further segmentable: iare as =ia plus =re
and iava as =ia and =va. Given that the =ia is a generic locative, the forms =iare and =iava
appear to involve further semantic specication in terms of path (source with =va or goal with
=re). The postpositions =va and =re differ from their polysyllabic counterparts in at least two
respects. First, there are a number of contexts where a polysyllabic form is incompatible with
6
According to Firchow (1973), there is also a postposition kerete inside out/reverse, but it was not recognized
by native-speaker consultants in the eld and no examples of it have appeared in the materials available to the
author. It has therefore been excluded from Table 4.17.
69
Type Postposition Gloss
Monosyllabic re to
pa for
va from
ia at
Polysyllabic arova without
iare towards
iava from, about
sirova behind
tapo(ro) also, too, with
Table 4.17: Rotokas Postpositions
its monosyllabic counterpart (e.g., the inalienable possession construction discussed in 10.2.2).
Second, verb roots that take an oblique argument select for a monosyllabic postposition but not
for polysllabic ones (e.g., kuara yell at selects the monosyllabic postposition =va). Finally,
the monosyllabic and polysllabic forms differ with respect to allophonic variation. The third
person singular normally takes the form -to when it occurs without additional morphology, but
it obligatorily takes the form -toa when it occurs with a sufx or enclitic, as in (148) to (149).
(148) oira-toa-re
man-SG.M=ALL
sirava-pa-ro-i
hiss-CONT-3SG.M
-PRES
rakoru-to
snake-SG.M
The snake is hissing at the man.
(149) tavaa-toa=ia
needle-SG.M=LOC
varo-a
clothes-SG.N
turu-pa-e
sew-CONT-3SG.F
Salome
Salome
Salome is sewing up clothing with a needle.
Unlike the locative enclitic =ia , the postpositions iare and iava do not obligatorily co-occur
with the form toa. In some cases, postpositions occur with the form to, as in (150). In other
cases, postpositions occur with the form toa, as in (151).
(150) kakae-to
child-SG.M
iava
POST
girigirio
grill
kapua-o-e
sore-3SG.F
-RP
The boys armpits had sores.

(151) tuuvuu-ra-i
swell.up-1SG
-PRES
koko-toa
leg-SG.M
iava
POST
uvare
because
ora-tugururu-a-e
RR-swell-1SG
-IP
My leg swole up because I bumped it.

The characterization of the monosyllabic forms as particles is questionable, given that the
term particle is usually used for words and these forms show many of the properties typically
associated with clitics or afxes (Zwicky and Pullum, 1983; Zwicky, 1985).
70
As Zwicky (1985) observes, if an element is bound, and especially if it cannot occur in
complete isolation, it should be a clitic. The postpositions in Rotokas are bound morphemes
i.e., they do not occur in isolation but always appear attached to another elementand by this
criterion are more clitic-like than particle-like. They are also the nal element in a noun phrase,
as can be seen in (152) and (153), where enclitics occur rightmost relative to other morphemes
(the possessive marker in (152) and the indenite marker in (153)).
(152) ragai
PRO.1.SG
vato-pa-a-veira
respect-CONT-1SG
-HAB
ragai
PPRO.1.SG
taataa-irara-aro=pa
brother-HUM.PL-POSS=BEN
I always respect my brothers.
(153) Paoro
P.
opita-ara-vai=va
coconut-PL.N-INDEF=COM
urio-u-vere
come-2SG
-NF
Paoro, you will come with some coconuts.
Another commonly-cited characteristic of particles is their ability to occur with full phrases
i.e., to occur at the boundary of a phrase rather than on the head noun. As Zwicky (1985)
observes, clitics are in this respect somewhat indeterminate between afxes and words: In-
ectional afxes combine with stems or full words, whereas words combine with other words
or with phrases. In Rotokas, case markers combine with phrases, as can be seen in (154) and
(155).
(154) kokeva
rain
voki-ara
day-PL.N
rutu-ia
very=LOC
kove-pa-o-i
fall-CONT-3SG.F
-PRES
It rains every day.

(155) Isivairi
Isivairi
koorato
possum
kapara-re-voi
roast-3SG.M
-PRES
eto kasi
re
raga=ia
only=LOC
Isivairi is cooking possum by re alone.
The analysis of the role markers becomes less clearcut where morphological simplicity is
concerned. Zwicky and Pullum (1983) observes that a morphologically complex item is prob-
ably an indepedent word. By this criterion, a few of the polysyllabic relators would qualify as
words, but not the monosyllabic relators.
The analysis adopted here treats both monosyllabic and polysyllabic relators as members
of a single class and attributes differences between them to phonological considerations. Since
there appears to be a foot minimality requirement for phonological words in Rotokas, the cliti-
cization of monosyllabic relators essentially falls out on independent grounds (see 3.2.2).
71
4.2.8 Interrogatives
Interrogatives are listed here as a distinct word class because they have a number of properties
which distinguish them from the word classes to which they might otherwise be assigned (e.g.,
pronoun). Their most salient property is their restriction to clause-initial position, as illustrated
for the interrogatives eake what and apeisi how.
(156) irou-toa
who-SG.M
vii
PPRO.2.SG
vaisi-aro
name-POSS
What is your name?
(157) Kepi,
Kepi
eake=re
what=ALL
ragai=va
PRO.1.SG=COM
paupau-pa-u
race-CONT-2SG
Kepi, why are you racing with me?

The full list of interrogatives is is provided in Table 4.18, where they are divided into two
groups, according to their ability to stand alone as question words.
Type Interrogative Gloss
Free-Standing apeisi how
ovu where
irou who
eake what
Modier aa which
arorea which (person)
ovirovu how many
avoviroa how much
roroa how much
Table 4.18: Interrogatives in Rotokas
The rst group of interrogatives stands alone as replacements for questioned elements while
the second group co-occurs either with other interrogatives or with nouns, as illustrated in (158)
and (159).
(158) apeisi
how
roro-a
much-SG.N
moni-a
money-SG.N
vii
PPRO.2.SG
ruvara=ia
near=LOC
tou-pa-i
be-CONT-3PL
How much money do you have on you?

(159) avoviroa
avoviroa
o-ua-vu
SPEC-CLASS-ALT
varo
clothing
ua
CLASS
vori-aro
price-POSS
How much is the price of one article of clothing?
72
Interrogatives occur with some of the same morphology as nouns, as illustrated in (160),
where an interrogative occurs with the diminutive sufx, or (161) through (162), where inter-
rogatives occur with postpositions.
(160) ra
and
apeisi-vai
how-INDEF
tarai-a-ve
know-3PL
-SUB
And they probably didnt understand how. [Firchow and Akoitai (1974:23)]
(161) eake=pa
what=BEN
vii
PRO.2.SG
upo-re-vo
hit-3SG.M
-IP
Why did he hit you?

(162) ovaiaro-vi
afternoon-DIM
avue
in-law
ovu=re
where=ALL
ava-pa-u
go-CONT-2SG
Afternoon, in-law, where are you going?

The occurence of interrogatives with morphology normally associated with nouns suggests
that interrogatives are nouns, but they showbehavior that makes themat least a distinct subclass.
For example, the interrogative irou who replaces nouns referring to human beings when they
are questioned, but it does not behave like a typical human noun (Class 1cf. 4.2.1.1), given
that it can behave as a masculine, feminine, or neuter noun, as illustrated in (163) through (165).
It takes the masculine singular sufx in (163), the feminine singular in (164), and no sufxes in
(165) (where it also shows the zero agreement associated with neuter subjects).
(163) irou-toa
who-SG.M
eera
DEM.MED.SG.M
Who is he?
(164) irou-va
who-SG.M
eira
DEM.MED.SG.M
Who is she?
(165) irou
who
ragai
PPRO.1.SG
oira-aro
PPRO.3.SG.M-POSS
torara
axe
ou-vo
get-IP
Who took my axe?

4.2.9 Conjoiners
The nal word class is a negatively-dened residual class which consists of whatfor lack of a
better termcould be labelled particles (Zwicky, 1985). These particles are for the most part
monomorphemic (with one possible exception discussed below). The words that fall into this
category are listed in Table 4.19.
Note that two of the forms in Table 4.19 can be analyzed as a derived forms based on the
particle uva: uvare (uva=re) and uvava (uva=va).
73
Particle Gloss Function
ari but Concession
oisio that Complementizer
osia as Comparative
ora and Conjunction
ovusia while Temporal
ra and Complementizer
teapi lest Apprehensional
uva so Conjunction
uvare because Causal
uvava because of/from Conjunction
vosia if/when Conditional
Table 4.19: Particles
(166) Rarasori
Robinson
ragai
PRO.1.SG
va-aro
3.SG.N-POSS
ovoi-pie-revo
enough-CAUS-3SG.M
rp.b
uvare
because
vura-pa-va
look-DERIV-SG.F
vori-re-vo
buy-3SG.M
-IP
ragai=pa
1.SG=BEN
Robinson satisied me because he bought binoculars for me.
(167) rera
PRO.3.SG.M
vara-aro=ia
body-POSS=LOC
veeto-a
cut-SG.N
keke-pa-io-vo
look at-CONT-1PL.EXCL-RP
uvava
???
ora-toe-ro-epa
RR-cut-3SG.M
-RP
We saw the marks on his body from where he cut himself.

There is another form, uvavu, which might also be analyzed as a derived form based on uva;
however, it functions as a noun co-occurring with nominal enclitics, as in (168) and (169), and
is excluded from the list in Table 4.19 since it does not serve a clause-conjoining function.
(168) papa-pa
y-DERIV
kepa
house
keke-ta
look-2PL
vavao
there
viara=ia
up-LOC
oa
RPRO.3.SG.N
uvavu=re
somewhere=ALL
ava-pa-i
go-CONT-PRES
Look at the airplane (literally: ying house) high above that is going somewhere.
(169) oira-to
man-SG.M
kakupie-pa-re-vo
shout-CONT-3SG.M
-RP
uvavu=va
somewhere=ABL
A man is shouting from somewhere.
The use of a few of these particles is illustrated in (170) through (173) (see 6.3.3 for more
detailed discussion of their role in interclausal syntax).
74
(170) Pita
Pita
keekee-pa
break-DERIV
sigo-a
knife-SG.N
ari
but
ragai
PRO.1.SG
vearo-a
good-SG.N
sigo-a
knife-SG.N
Peter has a broken knife but I have a good one.
(171) Pita
Pita
veta-ara
bamboo-PL.N
pariparikou-pa-re
cross.RDP-CONT-3SG.M
raiva=ia
road=LOC
oisio
COMP
teapi
lest
oira-ra-vai
man-HUM.PL-INDEF
vo-raiva-ia
SPEC-road-LOC
voka-pa-i-ve
walk-CONT-3PL
-SUB
Peter put bamboo across the road lest people walk on the road.
(172) apirika-pa-irara
Africa-DERIV-HUM.PL
oea
PRO.3.PL.M
kakare-aro
skin-POSS
oisio
COMP
osia
as
igei
PRO.1.PL.EXCL
rupa-irara
dark-HUM.PL
Africans, their skin is like that of us blacks.
(173) vii
PRO.2.SG
ateatepie-pa-a-voi
wait-CONT-1SG
-PRES
ovusia
while
ira-u
go ahead-2SG
Ill wait for you while you go ahead.

4.2.10 Exclamatives
The nal word class is fairly minor and consists of what can be labelled exclamatives, which
can be dened as words that function soley to mark an utterance as expressing a strong emo-
tional state of the speaker (Sadock and Zwicky, 1985; Michaelis, 2001; K onig and Siemund,
2007). The exclamatives of Rotokas are monomorphemic and occur sentence-initially.
A full list of all known exclamatives is provided in Table 4.20. The glosses provided for
these exclamatives are vague and should be considered very provisional, given that an adequate
characterization of their meaning would require more detailed study of their pragmatic function
(illucutionary force, etc.).
Although the exclamatives are largely monomorphemic, the exclamatives auo and auero
are potentially analyzeable (if not synchronically, then at least diachronically). The exclamative
auo is used exclusively to address females. In (175), it is used by a man in a folk tale who is
addressing a woman who is pursuing him aggressively after being charmed by a magical Jews
Harp.
(174) ae
hey
apa,
wait,
auo
hey
eaka-u
be calm-2SG
Hey, wait, woman, settle down! [Firchow and Akoitai (1974:71)]

75
The exclamative auoro is used exclusively to address males. In (175), it is used by one
brother addressing another in a short story about two brothers who swim across a river.
7
(175) auoro
hey
vore-ve
return-1PL.INCL
Hey, wed better go back! [Robinson and Mon (2006:The River)]
The exclamatives auo and auoro (sometimes pronounced auero or increasingly by the younger
generation as avero) may be morphemically broken down into au and a third person singular
demonstrative, either oo (female) or roo (male). The fact that it is sometimes pronounced as
auero suggests that it is diachronically related to the particle aue, which is used to drawattention
to a constituent (see 6.2.2).
7
(175) is a Rotokas translation of an English sentence, taken from an elementary school reader being developed
by the author for the Wakunai school district.
76
Case Marker
=ia (LOC)
Location Time Instrument Topic
=va (COM)
Comitative Source
=re (ALL)
Goal Addressee
=pa (BEN)
Benefactive Recipient
Figure 4.3: Postpositional Enclitics and Their Associated Semantic Roles
7
7
Exclamative Gloss Notes
aera All right!
ae oh, hey
aika wait
akoea truly
apa hey, eh
asi of course
auo Hey! used to address females
auoro Hey! used to address males
aure Yes!
aviova of course not, since when
eagara Let it be!
eari okay, all right
easi why of course
eaviova no, of course not
ee hey, eh
eesia It isnt!
ie Here take it!
iiu yes
kie Watch out!, be careful!
oire okay, all right
oo oh
ovuvaia No!
paapu no
raa Eh!, so?!
tape Stop!
tepa Hey!
Table 4.20: Exclamatives in Rotokas
78
Chapter 5
Morphology
This chapter provides an overview of the morphology of Rotokas, which shows a strong pref-
erence for sufxation and can be characterized as agglutinative, following Comrie (1989:43)s
denition: a word may consist of more than one morpheme, but the boundaries between mor-
phemes in the word are always clear-cut; moreover, a given morpheme has a least a reasonably
invariant shape, so that the identication of morphemes in terms of their phonetic shape is
also straightforward. The main exceptions to this generalization are the various pronominal
paradigms (see 4.2.3) and the verbal morphology for tense/mood (see 5.2.2.7), where the
morphemic segmentation is somewhat less straightforward.
A distinction is often drawn between two different types of morphology: derivational and
inection. Concerning this distinction, Anderson (1985:162) writes:
The central insight of this opposition is that derivation produces new lexical items
(perhaps complete words, perhaps stems) from other lexical material, with the de-
rived items on a par with simple, underived ones as far as their role in grammar
is concerned; while inection on the other hand serves to complete a word by
marking its relations within larger structures. Inection typically marks categories
which are applicable (at least potentially) to any item in a given word class, rather
than being specic properties of individual lexical items.
For descriptive convenience, inectional and derivational morphology will not be dealt with
separately in this chapter. Although there is very little derivational morphology associated with
nouns, there is a good deal of it associated with verbs, and this is given in-depth treatment in
Chapter 9, which looks at the valency-changing derivations found in the language.
5.1 Nominal Morphology
The template for nominal morphology is provided in Figure 5.1. Morphemes are listed accord-
ing to their order of occurence, which is strictly transitive (i.e., if A > B and B > C, then
79
A > C). The only required morpheme is the nominal root; all other morphemes are strictly-
speaking optional (although noun roots normally take a gender/number sufx).
S
p
e
c
i
e
r
R
o
o
t
N
o
u
n
D
e
r
i
v
a
t
i
o
n
a
l
G
e
n
d
e
r
/
N
u
m
b
e
r
P
o
s
s
e
s
s
i
o
n
D
i
m
i
n
u
t
i
v
e
A
l
t
e
r
n
a
t
i
v
e
D
e
n
i
t
e
n
e
s
s
-1 0 1 2 3 4 5 6
_
_
_
vo
o
ora
_
_
_
N -pa
_
_
a
ara
(a)rei
irara
rirei
to(a)
toarei
va
_
_
-aro -vi -vu
_
i
vai
_
Figure 5.1: Nominal Morphology
5.1.1 Prexes
There are two mutually exclusive sets of prexes that occur with nouns: the reexive/reciprocal
marker ora- and the specier vo-. (There is also a prex, a-, which co-occurs with the alternative
sufx and is analyzed here as a circumxsee 5.1.2.5.)
5.1.1.1 Order 1 Prex: Reexive/Reciprocal Marker
The reexive prex ora- occurs with pronouns as well as verbs (cf. 5.2.1.1). It has three
semantic functions: reexive, reciprocal, or emphatic/contrastive.
5.1.1.1.1 Reexive The reexive function of the prex ora- is illustrated in (176).
(176) ora-vii=pa
RR-PRO.2.SG=BEN
kepa-vai
house-INDEF
pura-ri-vere
build-2SG
-NF
riro
big
goru
strong
kepa-vai
house-INDEF
ora
and
aio-ara-vai
eat-PL.N-INDEF
vatatopo-ri-vere
prepare-2SG
-NF
ora-vii=pa
RR-PRO.2.SG=BEN
You must build a house, a strong house, for yourself and prepare food for yourself.
[Robinson and Mon (2006:Cricket and Grasshopper)]
80
5.1.1.1.2 Reciprocal The reciprocal function of the prex ora- is illustrated in (177).
(177) vo-vokiaro
SPEC-night
uva
so
oisoa
always
ora-vaiterei
RR-DEM.MED.M.DL
ruvara=ia
close=LOC
uusi-pa-si
sleep-CONT-3DL.M
During the night they slept next to each other. [Robinson and Mon (2006:Cricket and
Grasshopper)]
5.1.1.1.3 Emphatic/Contrastive The emphatic/contrastive function of the prex ora- is il-
lustrated in (178).
(178) ora-ragai
RR-PRO.1.SG
raga
only
ava-pa-ra-i
go-CONT-1SG
-PRES
Ruruvu
village
iare
POST
I myself am going to Ruruvu./I am the one going to Ruruvu.
5.1.1.2 Order 1 Prex: Specier
The nominal prex vo- can be described as specier.
1
It occurs with both nouns and classiers,
as illustrated in (179) and (180), but not with pronouns.
5.1.1.2.1 Specier with Noun
(179) oira-to
man-SG.M
ira
RPRO.3.SG.M
vo-riako
SPEC-woman
situe-pa-re
watch-CONT-3SG.M
osia
as
siisiiu-pa-a-i
bathe-CONT-3PL
-PRES
The man is watching the women as they bathe.

5.1.1.2.2 Specier with Classier
(180) Savia
Savia
veeta
bamboo
tou
CLASS
pokopoko-pie-e-voi
pop.RDP-CAUS-3SG.F
-PRES
uvare
because
vo-tou
SPEC-CLASS
tovo-e-vo
put-3SG.F
-IP
tuitui kasi
re
sovara=ia
inside=LOC
Savia made the bamboo pop repeatedly because she put it in the re.
Firchow (1987:34) treats the form o- as an allomorphic variant of vo- which co-occurs with
the alternative sufx -vu, as illustrated in (181).
1
Firchow (1987:34) describes the prex vo- as the specic morpheme.
81
(181) Rarasori
Robinson
vigei=pa
PPRO.1.PL.INCL=BEN
reo-pa
word-DERIV
vuku
book
ariara-pie-re-va
on top-CAUS-3SG.M
-RP
oa iava
therefore
uvui-pa-vi-ei
be able-CONT-1PL.INCL-PRES
ra
and
o-vaisi-ro-vu=ia
SPEC-word-PL.CL-ALT=LOC
tarai-pa-vio
know-CONT-1PL.INCL
Robinson prepared a dictionary for us and thats why we can know about other words.
However, unlike the prex o-, the form vo- in some cases functions as the host (rather than
as a modier to another form) for the role-marking enclitics discussed in 5.1.3.1. In 5.1.2.5,
the form o- is reanalyzed as part of the alternative circumx.
(182) riuriu-vira
dirty-ADV
raga
only
Saro
Saro
kare-ro-viro-i
return-3SG.M
-RES-PRES
vo=va
SPEC=ENC
vegoaro
jungle
Saro came from the jungle dirty.
(183) kakae
boy
vasie
CLASS
varu
meat
tara-sia
nd-DEP.SEQ
ava-a-e
go-3PL
-IP
vo
SPEC
vego-aro
jungle-POSS
The boys are going to nd meat in the jungle.
5.1.2 Sufxes
5.1.2.1 Order 1 Sufxes: Derivational
The sufx -pa derives instrumental and agentive nouns from various parts of speech (noun,
verb, etc.): for example, the agentive noun kavirupato thief is derived from the verb kaviru
steal, while the agentive noun vovokiopairara people of today derives from the temporal
noun vovokio today.
2
(184) kaviru-pa-to
steal-DERIV-SG.M
kepa=ia
house=LOC
paroo-ro-i
go inside-3SG.M
-PRES
vori-ara
money-PL.N
rutu
very
kaviru-sia
steal-DEP.SEQ
The thief is going inside of the house in order to steal all of the money.
(185) vovokio-pa-irara
today-DERIV-HUM.PL
riro
big
kaureo-irara
arrogant-HUM.PL
aite-irara=re
father-HUM.PL=ALL
The people of today are arrogant to their parents.
2
The sufx -pa is described as the instrument-agent marker by Firchow (1987:3536), who observes: The
instrument-agent (agt) marker -pa nominalizes adjectives and verb stems and also signals that a following sufx or
bound stem is manifesting the agent.
82
Instrumental nouns are also derived with this sufx: for example, the noun atepato scale
derives from the verb stem ate weigh, measure, as in (186) (which also illustrates the use of
ate weigh as a verb root).
(186) Maikol
Maikol
ira
RPRO.3.SG.M
kakau
cocoa
vaeke-ro
CL-PL.CL
ate-pa-re-voi
weigh-CONT-3SG.M
-PRES
atepatoa=ia
scale=LOC
Michael weighs the cocoa on a scale.
There are a number of instrumental nouns for which the sufx -pa is optional, such as
eri(pa)to shovel (derived fromeri dig), which occurs with the sufx -pa in (188) but without
it in (187). Elicitation work with native-speakers reveals no difference in meaning between the
two forms.
(187) Riki
Riki
eripatoa=va
shovel=COM
urio-u
come-2SG
apui
hole
teka-sia
dig-DEP.SEQ
Riki, come with a shovel to dig a hole.
(188) avu-va
grandparent-SG.F
eritoa=ia
shovel=LOC
opo
taro
pau-sia
dig-DEP.SEQ
ava-o-e
go-3SG.F
-IP
eisi
LOC
kovo-a
garden-SG.N
Grandma went to plant taro with a shovel in the garden.
The sufx -pa also functions as a derivational sufx for a subset of verbal roots when they
play an attributive role, as illustrated in (189) and (190), where in both cases the noun kepa
house is modied by a verb root with the derivational sufx: ruvaru medicate in (189) and
upia be sick in (190).
(189) ori-pa-to
cook-DERIV-SG.M
Raka
Raka
eisi
LOC
ruvaru-pa
heal-DERIV
kepa
house
Raka is the cook at the medical station.
(190) upia-pa-to
sick-DERIV-SG.M
pau-pa-a=ia
sit-DERIV-SG.N=LOC
pau-pa-ro-i
sit-CONT-3SG.M
-PRES
upia-pa
be sick-DERIV
kepa
house
siovara=ia
on=LOC
The sick man sits down in a chair inside of the medical station.
Finally, the sufx -pa also occurs on verbs as an aspectual marker (see 5.2.2.3 for details),
which is considered here to be a case of homophony. It is unclear which of the two functions
(derivational versus aspectual) is instantiated when -pa occurs on adverbs (see 4.2.6), as illus-
trated in (191).
(191) vuri-a
bad-SG.N
vao
DEM.PROX.3.SG.N
uuko-a
water-SG.N
oa
RPRO.3.SG.N
tuvu-pa-vira
muddy-?-ADV
tou-pa-i
be-CONT-3PL
osa
as
ra
and
va=ia
PRO.3.SG.N=ENC
ukaio-pa-u
drink-CONT-2SG
Water that is muddy is bad, as you cant drink from it.

83
5.1.2.2 Order 2 Sufxes: Number/Gender
The full set of number/gender sufxes will not be exemplied here, since the gender/number
markers were already discussed in detail in 4.2.1.1. There is allomorphic variation in the
realization of the masculine singular: when it is followed by another sufx, its form is -toa
rather than -to, as in (192) and (193).
(192) vii
PPRO.2.SG
viapau
NEG
uvui-pa-u
able-CONT-2SG
ra
COMP
tutuvagi
dark
oira-toa-vai
man-SG.M-INDEF
vura-ri
look-2SG
You cant see a man in the dark.

(193) ikoto-to
taproot-SG.M
ira
RPRO.3.SG.M
osiri
deep
rutu
very
roko-pa-re
enter-CONT-3SG.M
rasi-toa=ia
ground-SG.M=LOC
A taproot goes deep into the ground.
The distribution of -toa is wider than that of -to, since -toa also occurs even when it is not
followed by another sufx, as illustrated in (194).
(194) vii
PRO.2.SG
kopuasi-toa
smart-SG.M
rutu
very
viovoko-to
adolescent-SG.M
Youre a smart boy.
5.1.2.3 Order 3 Sufxes: Possession
The most common form of possession marking occurs on the possessed noun as the sufx -aro
(see 6.1.2 for an overview of possession-marking strategies).
(195) vo-ovi
SPEC-water
vaisi-aro
name-POSS
Eriovi
Eriovi
The name of this water is Eriovi. [Firchow and Akoitai (1974:52)]
(196) ira
RPRO.3.SG.M
aite-aro
father-POSS
ava-ro-epa
go-3SG.M
-RP
voka-sia
walk-DEP.SEQ
vego-aro
jungle-POSS
His father went walking through the jungle. [Story About Children]
5.1.2.4 Order 4 Sufxes: Diminutive
The sufx -vi has a diminutive meaning. It follows the number/gender sufxes, as can be seen
in (197), and precedes the alternative sufx, as illustrated in (198). (Note that in both examples
the diminutive sufx occurs on a modier to the head noun, rather than on the head noun itself.)
84
(197) Puruata
Puruata
oavu
another
gare-a-vi
small-SG.N-DIM
uvuo-a
island-SG.N
oae
PPRO.3.PL.F
isi
CLASS
tou-pa-i-veira
be-CONT-3PL
-HAB
eisi
LOC
Torokira=ia
Torokina=LOC
Puruata is a little island that is in Torokina.
(198) o-resiura-vi-vu
SPEC-four-DIM-ALT
vuruko-ara
log-PL.N
oara
RPRO.3.PL.N
varo-a
clothing-SG.N
ivara=ia
on=LOC
tou-pa-i-vo
be-CONT-3PL
-IP
The other four little logs are on the clothes. [CB:LR]

Firchow (1987:37) notes that the diminutive also has a gurative (non-diminutive) mean-
ing when afxed to pronouns, as exemplied in (199) and (200), where it conveys sympathy
for the referent of the diminutive noun. However, this sympathy reading does not seem to be
restricted to pronouns, judging from sentences such as (201), where it occurs with a common
noun, kopiito dead man.
(199) ragai-vi
PRO.1.SG-DIM
takau-ra-i
tired-1SG
-PRES
vo-avao-rei
SPEC-familyDL.CL
tavi-raga-pa-oro
tell-only-CONT-DEP.SIM
Poor little me is tired of just talking to the two women and their families.
(200) ae
oh
asi
of course
ragai-vi
PPRO.1.SG-DIM
tuutu-pie-pa-i
close-CAUS-CONT-PRES
ra
and
kopii-ra
die-1SG
Oh, poor little me is close to dying now. [Firchow and Akoitai (1974:45)]
(201) kopii-toa-vi
dead-SG.M-DIM
uvu-oro
hear-DEP.SIM
toiva
drum
tatuatu-re-vo
beat-3SG.M
-IP
Sova
Sova
Sova beat the drum, having heard (about) the dead man.
5.1.2.5 Order 5 Sufxes: Alternative
The sufx -vu is described as an alternative marker by Firchow (1987:38) because it normally
functions contrastively, essentially conveying the meaning of other or another. Its use pre-
supposes the existence of a contrastive alternative, whether it is explicitly mentioned or simply
implied. In (202), for example, a specic child (a boy named Rivasiri) is explicitly contrasted
with other (non-specic) children.
(202) vo-voki-ro
SPEC-day-PL.CL
rutu=ia
very=LOC
Rivasiri
Rivasiri
visiko
play
ruipa-pa-ro-veira
want-CONT-3SG.M
-HAB
o-kakae-ro-vu
SPEC-child-PL.CL-ALT
taporo
also
Rivasiri always wants to play with the other children.
85
In (203), however, there is a contrast made between between one side of the body (explic-
itly mentioned) and the other (not explicitly mentioned but obviously implied by real-world
knowledge).
(203) o-varata-vu
SPEC-side-ALT
keke-pa-ri
look-CONT-2SG
ragai
PRO.1.SG
iava
POST
Look at the other side of me.
The alternative marker occurs with a variety of noun subclasses: count nouns, as in (204);
classiers, as in (205); resumptive pronouns, as in (206); question words, as in (207); and
free-form pluralizers, as in (208)
(204) viapau
NEG
uvui-pa
be able-CONT
ra
and
ikau-vira
run-ADV
o-vaisi-vu
SPEC-name-ALT
vaisi-re-ve
call-3SG.M
-SUB
uvare
because
vapavapa-vira
unfamiliar-ADV
reo-pa-ro-veira
talk-CONT-3SG.M
-HAB
He cant say the other word quickly because he speaks strangely.
(205) aisi
CLASS
raga
only
aio-pa-ri
eat-CONT-2SG
ra
and
aisi-vu
CLASS-ALT
aio-ri-vere
eat-2SG
-NF
utupaua
later
Eat one now and the other later. [Robinson and Mon (2006:How Snakes Came to Be)]
(206) opeita
dont
ira-vu
RPRO.3.SG.M
roviriei-pa-ri
judge-CONT-2SG
rera
PPRO.3.SG.M
vo-pitupitu-a-aro=ia
SPEC-customs-SG.N-POSS=LOC
Dont judge another because of their customs.
(207) Pita
Pita
eakea-vu=a
what-ALT=TOP
eva
DEM.MED.SG.N
Peter, what is that (other thing)?
(208) o-kare-vu
SPEC-CLASS-ALT
koie
pig
kare
FP
kou-e-vo
carry-3SG.F
-IP
ita
again
aako-va
mother-SG.F
The mother carried the other pigs (her piglets).
Firchow (1987:38) observes that the alternative marker co-occurs with the prex o-, as il-
lustrated in (209).
(209) o-kakae-vu
SPEC-child-ALT
ita
again
kavau-e
give birth-3SG.F
osia
as
o-kakae-vu
SPEC-child-ALT
voka-pa-vira
walk-CONT-ADV
raga
only
tou-pa-i
be-CONT-3PL
She gave birth to (more) children as the others were walking.

86
However, the prex o- co-occurs with the sufx -vu only on consonant-initial stems. On
vowel-initial stems, the sufx -vu occurs alone (i.e., the prex is null or zero realized), as
illustrated in (205) through (207). Given that the two afxes obligatorily co-occur, they can be
analyzed as a single discontinuous morpheme, the circumx o- . . . -vu, which is subject to the
allophonic variation described in (210).
3
(210) o V
The alternative sufx precedes the indenite sufx, as illustrated in (211) and (212).
(211) sipito
chief
voea
PPRO.3.PL
tavi-re
tell-3SG.M
orekerovu-a-vu-vai
something-SG.N-ALT-INDEF
ou-sia
get-DEP.SEQ
vo=re
here=ALL
atoi-a
village-SG.N
eisi-va
LOC-ABL
kovo-ara
work-PL.N
The chief talked to everyone about getting something from the garden to the village.
(212) pukopuko-to
crippled-SG.M
oira-to
man-SG.M
ira
RPRO.3.SG.M
viapau
NEG
uvui-pa-ro-i
be able-CONT-3SG.M
-PRES
ra
and
o-kovo-a-vu-vai
ALT-work-SG.N-ALT-INDEF
pura-pa-re-ve
make-CONT-3SG.M
-SUB
A man with crippled ngers and/or toes is a man who cant do any work.
The segmentability of the alternative sufx is questionable in some cases, and one possible
explanation is that the sufx has been lexicalized. This is arguably the case for the frequently
occuring form oavuavuvai something. It occurs in (213) and (214), but lacks any clear-cut
contrastive semantics.
(213) ragai=pa
PRO.1.SG=BEN
oavua-vu-vai
something-ALT-INDEF
epao
exist
Is there something for me?
(214) uva
and
viapau
NEG
oiso
COMP
oavua-vu-vai
something-ALT-INDEF
vuri-pa-i
be bad-CONT-PRES
There is not something that is bad (i.e., there is nothing wrong).

While the indenite sufx -vai is readily identiable as a separate morpheme, the analysis
of oavuavu is less clear-cut. This form is capable of functioning alone as a noun, as illustrated
in (215) and (216).
(215) oavuavu
something
uvu-pa-voi
smell-CONT-PRES
oa
RPRO.3.SG.N
eru
stink
geesi-pa-i
smell-CONT-PRES
I smell something that stinks.

3
Michael Dunn deserves credit for suggesting the alternative circumex analysis.
87
(216) Pita
Pita
vaio
CL.DL.HUM
ora
and
Jon
Jon
oavuavu=ia
something=LOC
ogaoga-vira
whisper-ADV
ora-reo-pa-si-e
RR-talk-CONT-3DL.M-IP
Peter and John are talking about something with each other in whispers.
However, its internal analysis is questionable. It appears to derive from the word oavu,
which normally functions as a nominal modier, as in (217) and (218).
(217) vori-a
buy-SG.N
goru-aro
strong-POSS
vara-epa
go down-DP
voo=ia
here=LOC
uvare
because
oavu
some
vori
money
ou-pa
get-DERIV
tupa-piro-pa
close-RES-DP
The strength of money went down here, because a money-getting place (the gold mine)
closed.
(218) Puruata
Puruata
oavu
some
gare-a-vi
small-SG.N-DIM
uvuoa
island
oa
RPRO.3.SG.N
eisi
LOC
tou-pa-i-veira
be-CONT-3PL
-HAB
eisi
LOC
Torokira=ia
Torokina=LOC
Puruata is a little island that is in Torokina.
Although oavu does not agree with the noun that it modies in (217) and (218), it does take
take gender/number sufxes when it functions alone as a noun, as in (219) and (220).
(219) Varei
Varei
teka-re
sharpen-3SG.M
evao-arei,
tree-DL.N
oavu-a
one-SG.N
averu-a
thin-SG.N
ari
but
oavu-a
one-SG.N
vuru-pa
thick-DERIV
He is sharpening two trees, one is thin and the other thick.
(220) oire
okay
oavu-va
some-SG.F
oa
RPRO.3.SG.N
vaisi-pa-i-veira
call-CONT-3PL
-HAB
oisio
as
okaoto-va
talis-SG.F
Okay, one tree, they call it okaoto. [Matevu]
This suggests that the proper analysis of oavuavuvai is the one provided in (221a), but the
absence of clear contrastive semantics suggests that the form oavuavu has been lexicalized, and
the proper analysis is (221b).
4
(221) a. akuku-a
empty-SG.N
viapau
NEG
oavu-a-vu-vai
something-SG.N-ALT-INDEF
voo=ia
here=LOC
Its empty, theres nothing here.
b. akuku-a
empty-SG.N
viapau
NEG
oavuavu-vai
something-INDEF
voo=ia
here=LOC
Its empty, theres nothing here.
4
It might also be possible to treat this as a case of reduplication of oavu (i.e., oavu-avu), invoking (210) to
account for the elision of the initial vowel in the second depuplicant. However, the productive formof reduplication
found in the language (see 3.2.3) involves partial reduplication of the rst rather than the second reduplicant.
88
5.1.2.6 Order 6 Sufxes: Denite/Indenite
5.1.2.6.1 Indenite Sufx The sufx -vai is a marker of indeniteness (von Heusinger,
2002), which occurs on nouns that are non-specic, unidentiable, and/or non-referential.
5
For
example, it occurs on the noun oirato man in (222), which asserts that some unidentied (and
perhaps unidentiable) man will desire the addressee once she is properly adorned.
(222) vii
PRO.2.SG
orito-a-voi
decorate-1SG
-PRES
uva
and
vearo
good
keke-irao-u-ei
look-INTEN-2SG
-PRES
ra
and
oira-toa-vai
man-SG.M-INDEF
vii
PRO.2.SG
riri-re-ve
covet-3SG.M
-SUB
Ill decorate you and youll look really good and some man will covet you.
In (223) and (224), the speaker urges the addressee to tear or cut off a plant leaf, without any
specic one in mind, in order to use it for medicinal purposes, and in both cases the indenite
noun appears with the sufx -vai.
(223) ito
banana
guruva-vai
leaf-INDEF
pako-ri
tear off-2SG
Grab a banana leaf. [Robinson and Mon (2006:Leaves Will Help You)]
(224) oire
okay
ragai=pa
PRO.1.SG=BEN
katai
one
vagai-vai
leaf-INDEF
tosi-ri
cut-2SG
aue
CONN
guru-va
leaf-SG.F
ra
and
vao
DEM.PROX.SG.N
ruu-a
cover-1SG
arua
vegetable
tai
CLASS
Cut one leaf for me and Ill cover these vegetables.
In (225), for example, the noun oirara people occurs with the indenite sufx due to the
fact that it is non-referentiali.e., refers to non-existing entities.
(225) Ruruvu
Ruruvu
urui
village
arakasi-ei
empty-PRES
rutu
very
viapau
NEG
oira-ra-vai
man-HUM.PL-INDEF
Ruruvu village is truly empty, there are no people.
5.1.2.6.2 Denite Sufx The meaning and function of the sufx -i is unclear. It is glossed as
the absolute sufx by Firchow (1987:39), who claims that it conveys certitude and observes
that it occurs only with resumptive pronouns, as in (226) and (227). Although very fewinstances
of it occur in the materials available to me, its occurence is conned to resumptive pronouns,
keeping with Firchows observations concerning its distribution.
5
It is glossed as the non-absolute sufx by Firchow (1987:39), who claims it conveys incertitude or pos-
sibility.
89
(226) varao
DEM.PROX.PL.N
rutu=ia
very=LOC
viato-pie
vacant-CAUS
teapi
lest
oa-i
RPRO.3.SG.N-
kavu-pa-ri
leave.behind-CONT-2SG
Clear everything lest you leave one behind.

(227) vosia
when
koie-a-vai
pig-INDEF
upo-a
kill-1SG
oisoa
always
iria-i
RPRO.3.SG.F-
kuvu-a
ll-1SG
aue=ia
CONN=LOC
veeta
bamboo
When I would kill a pig, I would always put it inside bamboo tubes. [Firchow (1984)]
5.1.3 Enclitics
Two categories of enclitics are discussed below: the various postpositional enclitics (5.1.3.1)
and the topic marker (5.1.3.2). These enclitics are included here, in a section on nominal
morphology, because they typically occur on nouns, but strictly speaking they attach at the level
of the noun phrase and therefore may cliticize to other parts of speech (e.g., the postnominal
noun quantier rutu in (238)).
5.1.3.1 Postpositional Enclitics
The postpositional enclitics are discussed with respect to Rotokas word classes in 4.2.7 and
with respect to verb valency and subcategorization in 8.3.3. When these postpositional encli-
tics occur on adjuncts, they mark broad semantic relations, illustrated in (228) through (234).
5.1.3.1.1 Locative/Instrumental
(228) upia-pa-ra-i
hurt-CONT-1SG
-PRES
kukue
head
iava
POST
oa iava
hence
uusi-pa-ra-i
sleep-CONT-1SG
-PRES
uruu-a=ia
bed-SG.N=IA
My head hurts and thats why Im sleeping in bed.
(229) sikuru-pa-irara
school-DERIV-HUM.PL
rearea-a-e
take vacation.RDP-3PL
-IP
vo-wiki-rei=ia
SPEC-week-DL.N=LOC
The school kids take a vacation during these two weeks.
(230) revasi-vira
blood-ADV
paru-re-voi
ow-3SG.M
-PRES
parura-to
blister-SG.M
vo=va
SPEC=ABL
Kuroi
Kuroi
vavae-aro
hand-POSS
ovusia
while
rera
PRO.3.SG.M
toga-e-vo
pierce-3SG.F
-IP
tava-toa=ia
needle-SG.M=LOC
The blister is owing bloodily from Kurois hand, while he pierced it with a needle.
90
5.1.3.1.2 Allative
(231) oiraopie-pa-irara
believe-DERIV-HUM.PL
eisi=re
LOC=ALL
ava-a-verea
go-3PL
-DF
vuvui
sky
ua
CLASS
The believers are going to heaven.
5.1.3.1.3 Benefactive
(232) kakae
child
vure=pa
CLASS=BEN
veevee-a
story-SG.N
pura-pa-a-voi
make-CONT-1SG
-PRES
aue
CONN
iava
POST
pukui
mountain
tou-to
be-SG.M
Im telling a story for the children about a mountain dweller.
5.1.3.1.4 Ablative
(233) riro-pa
big-DERIV
vikuta-to
whistle-SG.M
eera
DEM.M.SG.MED
Sovire
Sovire
ira
RPRO.3.SG.M
uvui-pa-i
be able-CONT-PRES
ra
and
rera
PRO.3.SG.M
uvu-ri
hear-2SG
tauai=va
far away=ABL
Sovire is a big whistler, who you can hear from far away.
(234) Tavi,
Tavi,
tuitui-a-vai=va
re-SG.N-INDEF=ABL
urio-u
come-2SG
ra
and
tuitui kasi-ve
re-1DL
ori-sia
cook-DEP.SEQ
Tavi, come with some re and well make a re in order to cook.
5.1.3.2 Topic Marker
The sufx -a is analyzed here as an optional topic marker that occurs as an enclitic on nouns
in topic position. The topic position is leftmost within the clause and the noun that occupies it
typically agrees in person, number, and gender with a coreferential relative pronoun that occurs
elsewhere in the clause, as illustrated for a topicalized proper noun in (235) and a topicalized
pronoun in (236).
(235) Skip
Skip
Firchow=a
Firchow=TOP
avirika-pa-to
America-DERIV-SG.M
ira
RPRO.3.SG.M
Rotokasi-pa
Rotokas-DERIV
reo
word
pore-sia
turn-DEP.SEQ
urio-ro-epa
come-3SG.M
-RP
Skip Firchow was an American who came to translate Rotokas.

91
(236) ragai=a
PRO.1.SG=TOP
kerui-to
skinny-SG.M
ragoa=ia
RPRO.1.SG=LOC
viapau
NEG
varu-ara-vai
meat-PL.N-INDEF
tou-pa-veira
be-CONT-HAB.ANIM
ora
and
aue
CONN
tuuga
grease
ragai
PRO.1.SG
vara-aro=ia
body-POSS=LOC
I am skinny, there is no meat on my body and there is grease on my body.
The topic marker occurs at the rightmost boundary of the topic noun phrase and follows all
other nominal morphology, such as the possessive sufx in (237).
6
(237) Rarasori
Rarasori
vatua-va-aro=a
wife-SG.F-POSS=TOP
vaita-va
pretty-SG.F
rutu
very
riako-va
woman-SG.F
Robinsons wife is a very pretty woman.
It therefore typically occurs on the nominal head but can occur on other parts of speech,
provided they are the nal element of the noun phrase, such as the postnominal quantier rutu
in (238).
(238) ee
EXCLAM
vo-kepa
SPEC-house
kaekae-aro
long-POSS
rutu=a
very=TOP
vao=ia
DEM.PROX.3.SG.N=LOC
Thats the length of the house.
The sufx described here as a topic marker is described in Firchow (1987:39) as a relative
pronoun marker due to the fact that subsequent anaphoric reference to the topic usually takes
the form of a relative pronoun, as in (239) and (240).
(239) Ruso-a
Ruso-TOP
aveke-va
stone-SG.F
iria
RPRO.3.SG.F
riro-vira
big-ADV
aviavi-pa-e-veira
light.RDP-CONT-3SG.F
-HAB.ANIM
roro-pa-oro
shine-CONT-DEP.SIM
vavoiso
there
virauaro
ground
Ruso is a stone that shines brightly going there into the ground.
(240) Riepi=a
Riepi=TOP
tokoruo-to
sedentary-SG.M
viapau
NEG
ava-pa-ro-i
go-CONT-3SG.M
-PRES
eisi
LOC
kovo-a
garden-SG.N
Riepi is a sedentary man, he doesnt go to the garden.
The topic marker can be difcult to identify for non-native speakers given that many nouns
(e.g., feminine or neuter singular nouns) end with the same vowel as the sufx, in which case
the only reex of topic marking is vowel lengthening on the nouns nal syllable.
6
Examples such as (237) contradict the claim made in Firchow (1987:39) that this sufx is mutually exclusive
with other nominal sufxes.
92
5.2 Verbal Morphology
There is a good deal of morphology associated with the derivation and inection of verb stems
in Rotokas, as illustrated by the morphologically complex verb in (241), which consists of a
prex, verb root, and 5 sufxes: the morphological causative, an intensier, the continuous
aspect, third person masculine singular, and the present tense realis mood.
(241) Pita
Peter
ora-oruo-pie-irao-pa-ro-i
RR-content-CAUS-only-CONT-3SG.M
-PRES
siope-pa-va
meat-DERIV-SG.F
aio-pa-oro
eat-CONT-DEP.SIM
araisi
rice
Peter really contented himself eating meat-lled rice.
The verb in (241) is broken down into its constituent parts in (242).
(242)
Word
..
Stem
..
ora-
Root
..
oruo -pie
Modier
..
-irao
Inection
..
-pa-ro-i
The template for verbal morphology is summarized diagrammatically in Figure 5.2. Note
that the person/number/gender sufxes and the tense/mood sufxes appear in square bracket;
this is due to the fact that they come in two sets. Since the nature of these two sets is the focus
of the second part of this thesis (see 7), it will sufce for now to label them in a neutral fashion,
as Class and Class .
7
5.2.1 Prexes
5.2.1.1 Order 1 Prex: Reexive/Recriprocal
There is only one verbal prex, the reciprocal/reexive marker, ora- (which, as shown earlier in
5.1.1, also occurs with personal pronouns). Verb stems occurring with the reexive/reciprocal
sufx are invariably , as illustrated by the contrast between the reexive and non-reexive
forms of the verb root upo hit, kill in (243).
(243) a. rera
PPRO.3.SG.M
upo-re-va
kill-3SG.M
-RP
He killed him. [Caleb, Another Togarao Story]

7
Firchow (1987) labels the two classes and but the labels and are used here instead.
93
b. ra
and
rera
PPRO.3.SG.M
raga
only
ora-upo-ro
RR-kill-3SG.M
eke
TAG
Will he kill himself? [Mark 8:22]
In-depth discussion of ora- as a valency-decreasing derivational prex can be found in
9.2.1.
5.2.2 Sufxes
5.2.2.1 Order 1 Sufxes: Causative
The causative sufx -pie is a valency-increasing derivational sufx which consistently derives
stems that show agreement (see 9.1.2 for in-depth discussion). For example, the verb root ka-
pua to have sores normally shows agreement, as illustrated in (244), but shows agreement
when it occurs with the causative sufx, as illustrated in (245).
(244) riako-va
woman-SG.F
kapua-pa-o-i
have sores-CONT-3SG.F
-PRES
uvare
because
vatua-to
husband-SG.M
oira
PPRO.3.SG.F.B
upo-re-voi
hit-3SG.M
-PRES
vuri-vira
bad-ADV
rutu
very
The woman has sores because her husband has beaten her very badly.
(245) oirato
man
kapua-pie-i-vo
have sores-CAUS-3PL
-IP
rera
PPRO.3.SG.M
upo-oro
hit-DEP.SIM
uvare
because
kepa
house
toko-oro
break into-DEP.SIM
koata-ro-e
enter-3SG.M
-IP
torara
axe
kaviru-sia
steal-DEP.SEQ
They injured the man by hitting him because he broke into a house to steal an axe.
5.2.2.2 Order 2 Sufxes: Modiers
There are two order 2 sufxes, which are -raga and -irao. Each will be described in turn.
5.2.2.2.1 -raga only/just The characterization by Firchow(1987) of this sufx as a marker
of indifference is dubious. These sentences tend to be translated by informants using the Tok
Pisin modier nating or with the English focus adverbs just or only.
(246) oire
okay
tara-raga-pa-io-va
search-only-CONT-1PL.EXCL-RP
We just searched. (Na mipela i bin painim nating.) [Abraham Raviata, Long Ago in
Raurau]
94
The sufx -raga has an unbound counterpart, illustrated in (247) through (248).
8
(247) katai-toarei-vi
one-DL.M-DIM
raga
only
kokai
chicken
vaio
ANIM.DL
aiterea
PPRO.3.DL.M
ou-a-vo
get-1SG
-IP
I only got two little chickens.

(248) avae-vira
temporary-ADV
raga
only
tou-pa-peira
be-CONT-1DL+HAB
vo-rasio=ia
SPEC-earth=LOC
Were only temporarily on the earth.
5.2.2.2.2 -irao really Firchow (1987) labels the sufx as a marker of emphasis but it is
probably better characterized as an intensier.
(249) oire
okay
uva
and
riro-vira
big-ADV
rutu
very
rugorugoo-irao-ro-epa
think-INTEN-3SG.M
-RP
rera
PPRO.3.SG.M
aite-to
father-SG.M
oisio
COMP
So, his father really thought hard. [Firchow and Akoitai (1974:78)]
(250) ari
but
vovou
DES
tagoro-vira
hidden-ADV
raga
only
kasipu-irao-pa-ro-epa
angry-INTEN-CONT-3SG.M
-RP
He was really angry and hid, thats all. [Firchow and Akoitai (1974:79)]
The sufx -irao has an unbound counterpart, which means true or real, and presumably
arose by incorporation into the verbal complex (adverb incorporation). The unbound form is
illustrated in (252) through (253).
(251) oira-pa-toa
man-DERIV-SG.M
rutu
very
irao
real
vii
PPRO.2.SG
Youre a true man.
(252) sisiara-pa-toa
greasy-DERIV-SG.M
irao
true
roo
PPRO.3.SG.M
koora-to
possum-SG.M
This possum is truly greasy.
(253) ruve
aibika
tai
CLASS
ori-e-voi
cook-3SG.F
-PRES
uva
and
riro-vira
big-ADV
ruve-vira
big-ADV
irao
true
uvare
because
riro-vira
big-ADV
opita
coconut
kuri-o-i
scrape-3SG.F
-PRES
vo-tai=re
SPEC-CLASS=ALL
She is cooking aibika, and it is truly greasy because she is scraping a lot of coconut on it.
The sufxes -raga and -irao are not mutually exclusive according to Firchow (1987), who
cites (254), where -raga precedes -irao.
8
Note the irregular form of verbal inection in (248)see 5.2.2.6.1 for explanation.
95
(254) ava-raga-irao-pa-ra-erao
go-only-INTEN-CONT-1SG
-NP
I really was just going days ago. [(Firchow, 1987:16)]

Finally, there are a few stems that appear to have lexicalized -irao and therefore display
apparent violations of morpheme ordering. For example, the causative sufx -pie normally
precedes -irao, as in (255); however, the order is reversed in (256) due to the lexicalization of
-irao in the stem kasirao hot.
(255) kepa
house
viato-pie-irao-i-vo
clear.out-CAUS-INTEN-3PL
-IP
auero
everything
vera-oro
remove-DEP.SIM
They really cleared out the house, removing everything.
(256) uuko
water
rovu
CLASS
kasirao-pie
heat-CAUS
eto kasi=ia
re=LOC
Heat up the water on the re.
Further evidence for the lexicalization of -irao in the stem kasirao is the fact that the sufx
-irao can co-occur with the stem kasirao (in which case irao occurs twice), as illustrated in
(257), which describes the syptoms of malaria.
(257) vo-rara
SPEC-later
riro-vira
big-ADV
rutu
very
oira-to
man-SG.M
kasirao-irao-pa-ro
hot-INTEN-CONT-3SG.M
ora
and
uteo-pa-ro
cold-CONT-3SG.M
tapo
also
Hence the man is really hot and really cold. [Firchow (1974b:68)]
5.2.2.3 Order 3 Sufxes: Continuous
The sufx -pa is the only Order 3 sufx. It is found with both independent verbs, as in (258),
and dependent verbs, as in (259).
(258) o-voki-vu=ia
SPEC-day-ALT=LOC
ava-ra-era
go-1SG
-NP
eisi-re
LOC=ALL
Ibu
Ibu
ovusia
while
ora-upo-pa-a-era
RR-hit-CONT-3PL
-NP
One day I went to Ibu while they fought.

(259) riro-vira
big-ADV
rutu
very
roru-pa-oro
happy-CONT-DEP.SIM
kauokauo-pa-ra-i
jump.RDP-CONT-1SG
-PRES
I jumped up and down truly happy.

It is glossed as the progressive action marker in Firchow (1987:17). However, as Chung
and Timberlake (1985:214) observe, the term progressive is typically reserved for a category
that is restricted to dynamic events:
96
More importantly, the opposition between states and process can play a role in the
selection of aspectual morphology, specically the progressive. The progressive
asserts than an event is dynamic over the event frame. By denition, then, processes
but not states can appear in the progressive.
The sufx -pa occurs with a wide variety of verbs, including verbs that denote non-dynamic
events (states), such as tarai know in (260) and (261). For this reason, it is glossed here as
continuous rather than progressive.
(260) viapau
NEG
tarai-pa-ra-i
know-CONT-1SG
-PRES
motokara
car
voka-pie-pa-oro
walk-CAUS-CONT-DEP.SIM
I dont know how to drive a car.
(261) o-kare-vu
SPEC-FP-ALT
rutu
very
vaisi-aro=ia
name-POSS=LOC
tarai-pa-o-i
know-CONT-3SG.F
-PRES
Sera
Sera
Sera knows the names of all the different animals.
5.2.2.4 Order 4 Sufxes: Resultative Sufx
The resultative sufx takes the form of -viro or -piro.
9
The gender of a verbs subject determines
which of the two forms occurs. The sufx -piro occurs on verbs with neuter subjects, as in
(262), while the sufx -viro occurs on verbs with non-neuter subjects, as in (263) (where the
non-neuter subject is masculine noun referring to an inanimate objectnamely, a post).
10
(262) Pita,
Pita,
kaitu-pa-i
tight-CONT-PRES
eva
DEM.MED.SG.N
iroiro
rope
oa iava
hence
toko-piro-i
breakRES-PRES
varo
clothes
tava=va
CL=COM
Peter, that rope is tight and therefore it broke with the clothes.
(263) Evato
Evato
tuuta-to
post-SG.M
roe-re-vo
place-3SG.M
-IP
uva
and
gasi-ro-viro
fall-3SG.M
-RES
Evato placed the post and it fell down.
The resultative sufx precedes the tense/mood sufxes and follows the progressive sufx,
as illustrated in (264).
9
This sufx is described as the complete action marker by Firchow (1987).
10
Firchow (1987) claims that animacy determines the choice of the two forms. Although gender correlates
highly with animacy, there are nevertheless mismatches, primarily with masculine or feminine nouns denoting
inanimate entities (e.g., tuutato post), and these show that it is gender (not animacy) which is the determining
factore.g., see (263).
97
(264) tokaaru
orchid
oavu
another
kokoa
ower
iroa
vine
iava
POST
pura-pa-piro-veira
make-CONT-RES-HAB
An orchid is a ower that appears on the vine.
Firchow (1987:15) observes that the allomorph -piro co-occurs with the sufxes -vere, -
verea, -ve, and -veira, which is unexpected, given that -piro normally occurs with neuter subjects
while the latter sufxes normally occur with non-neuter subjects. This unexpected co-occurence
is exemplied in (265), where a neuter subject occurs with -veira, and in (266), where a neuter
subject occurs with -vere.
(265) uuko-vi
river-DIM
vavo-va
there=ABL
kosikosi-pa-piro-veira
exit.RDP-CONT-RES-HAB
pukui=ia
mountain=LOC
vitu-aro
base-POSS
The river gushes out from the base of the mountain.
(266) vosia
if
katai
one
isi-vai
CL-INDEF
kavu-piro-vere
left behind-RES-NF
ovoi-ei
nish-PRES
ra
and
oisio
COMP
kavu-viro-ve-i-ei
left behind-RES-1DL-EPEN-PRES
If one [seedling] is left behind, okay, the two of us are left behind. [Firchow (1987:64)]
5.2.2.5 Order 5 Sufxes: Dependent Verb Morphology
Independent verbs show subject agreement and tense/mood marking, whereas dependent verbs
lack both and instead take one of one of the dependent-marking sufxes listed in Table 5.1 (see
6.3.2.1 on the syntax of dependent verbs).
Morpheme Gloss
-sia purposive action (in order to)
-oro simultaneous action (while)
-arapa negation/negative polarity (not)
Table 5.1: Dependent Verb Marking
The three dependent-marking sufxes are illustrated in (267) through (269).
(267) erako-sia
collect rewood-DEP.SEQ
ava-pa-i-ei
go-CONT-1PL.EXCL-PRES
Were going to collect rewood.

(268) ogoe-ra-i
be hungry-1SG
-PRES
voka-pa-oro
walk-CONT-DEP.SIM
eisi
LOC
Asitavi
Asitavi
Im hungry walking to Asitavi.
98
(269) asia-pa-ra-i
be disinclined-1SG
-PRES
utu-arapa
follow-DEP.NEG
eisi=re
LOC=ALL
kovo-a
garden-SG.N
I dont want to come along to the garden.
Firchow (1987:19) observes that the sufx -arare is an alternative form of sufx -sia, but
that it is very rare and appears to be archaic. No examples of it are attested in the materials
available to the author.
5.2.2.6 Order 5 Sufxes: Person/Number/Gender
The order 5 sufxes fall into two classes on semantic grounds: dependent verb morphology
and person/number. The two are mutually exclusive. In other words, a verb stem can either
take independent or dependent inection, but not both. If it takes independent inection, it
must take person/number marking, whereas if it takes dependent inection, it cannot take per-
son/number marking and must take one of the dependent marking sufxes. Dependent marking
also precludes tense/mood marking, which is discussed in 5.2.2.7.
5.2.2.6.1 Person/Number/Gender Independent verbs agree with their subjects in person,
number, and gender. Agreement is nominative-accusative, in the sense that the verb always
agrees with either S (subject of an intransitive verb) or A (subject of a transitive verb) (see
7.3 for more in-depth explanation of the terms S, A, and O). However, the form of subject
agreement depends upon the particular person, number, and gender congurationsee Table
5.2. Verbal inection shows distinct forms of subject agreement for some congurations of
person, number, and gender (e.g., third person singular), but not for others (e.g., third person
dual). For example, the verb root uusi sleep shows agreeement while the verb root upo
hit shows agreement. The form of verbal inection for the third person singular feminine
differs for the two verb roots: -o in (270) and -e in (271).
(270) atuu
ying fox
koto-vira
hang-ADV
uusi-pa-o-i
sleep-CONT-3SG.F
-PRES
The ying fox sleeps hanging.

(271) vegei
PRO.1.DL.EXCL
upo-e-voi
kill-3SG.F
-PRES
Shes killing us two! [Firchow and Akoitai (1974:32)]

However, verbal inection for the third person dual masculine shows no distinction between
and agreement. The form of verbal inection for the third person dual masculine is invari-
ant, as shown by (272) and (273).
(272) evoa
there
oisioa
always
uusi-pa-si
sleep-CONT-3DL.M
The two of them always slept there. [Firchow and Akoitai (1974:27)]
99
(273) osia
as
viapau
NEG
oisioa
always
koie
pig
kare-vai
FP-INDEF
upo-pa-si
kill-CONT-3DL.M
But the two did not kill any pigs. [useless dogs.txt:4]
The full set of person-marking sufxes is summarized in Table 5.2, where each conguration
of person, number, and gender is provided.
Person Number Gender
1st Person
Singular -ra -a
Dual -ve
Plural Inclusive -vio
Plural Exclusive -io
2nd Person
Singular -u -ri
Dual M -si
F -ere
Plural -ta
3rd Person
Singular M -ro -re
F -o -e
Dual M -si
F -ere
Plural -a -i
Table 5.2: Subject Agreement Sufxes
The paradigmatic structure of the various pronoun paradigms differs slightly from that of the
person-marking sufxes, due to the collapsing of the distinction between the second and third
person dual in the bound pronouns.
11
This is illustrated in Figure 5.3, where the paradigmatic
structure of the personal pronouns (repeated from Figure 4.2) is contrasted with the verbal
agreement sufxes following the analytical scheme of Cysouw (2003).
11
Cysouw (2003) observes that vertical homophony between the second and third person is typical of the Papuan
languages, citing as an example the mainland Papuan language Korafe. This remains to be substantiated, but it is
worth pointing out that this pattern is not particularly widespread among the East Papuan languages. In fact, it is
found in only 4 of the 15 languages surveyed in a comparison of grammatical features described in Dunn et al.
(2005)namely, Mali, Savosavo, Rotokas, and Y el-Dnye.
100
Group Restricted Group
Personal Pronouns (Free)
vigei
vegei
1+2(+3)
1 ragai igei 1+3
2 vii visii vei 2+3
3 rera / oira / va voea vaiterei 3+3
Verbal Agreement (Bound)
-vi
-ve
1+2(+3)
1 -ra / -a -io 1+3
2 -u / -ri -ta
-si / -ere
2+3
3 -ro / -re -i / -a 3+3
Figure 5.3: Paradigmatic Structure for Person Marking in Personal Pronouns versus Verbal Agree-
ment Sufxes
The use of two different personal pronouns with the same form of verbal agreement is
illustrated in (274) and (275). In both cases, the verbal agreement takes the sufx -si, but the
personal pronoun that plays the role of subject is the second person plural vei in (274) and the
third person dual vaiterei in (275).
(274) vei
PRO.2.PL
rogo
begin
rovo-pa-si-ei
start-CONT-2DL-PRES
ikau-oro
run-DEP.SIM
You two start running rst . . . [=(78)]
(275) vaiterei
PRO.3.DL
ora-uugaa-pa-si-ei
RR-kiss-CONT-3DL.M-PRES
The two are kissing each other.

5.2.2.7 Order 6 Sufxes: Tense/Mood
The order 6 sufxes consist of morphemes that mark tense and/or mood. These morphemes can
be divided into two classes: those that are sensitive to verb stem classication and those that are
sensitive to the gender of the subject. These two formally distinguishable classes correspond
to a basic distinction between two categories of mood: realis and irrealis. The distinction
between these two categories is characterized by Mithun (1999:173) in the following terms:
The realis portrays situations as actualized, as having occurred or actually occurring, knowable
through direct perception. The irrealis portrays situations as purely within the realm of thought,
knowable only through imagination.
Tense/mood marking is obligatory for independent verbs, with two exceptions. First, ab-
sence of marking is interpreted as present tensei.e., the present tense can be null-marked (see
101
5.2.2.7.1). Second, no marking of tense, aspect, or mood is found on imperatives, as illustrated
in (276) and (277).
(276) Visiaevi
Visiaevi
uuko-a-va
water-SG.N=ABL
urio-u
come-2SG
ra
and
ukaio-ra
drink-1SG
Visiaevi, come with some water and Ill drink.

(277) varao
DEM.N.PL
vori-ri
buy-2SG
kotokoto-ara
cargo.RDP-PL.N
Buy these supplies.
5.2.2.7.1 Realis Within the realis mood, Rotokas has a system of metrical tense (Comrie,
1985b). Such systems are fairly rare cross-linguistically. They are found among the Papuan
languages of mainland Papua New Guinea but not among the East Papuan language, with the
notable exception of Y el-Dnye (Dunn et al., 2002) and Rotokas.
The metrical tense system of Rotokas distinguishes between the present tense and four cat-
egories of past tense: immediate, near, distant, and remote. This is summarized in Table 5.3.
Tense
Present -ei -voi
Past Immediate -e -vo
Near -era -vora
Distant -erao -vorao
Remote -epa -va
Table 5.3: Rotokas Tense Categories
Additional segmentations of these forms, where the tenses are analyzed as a combination of
sufxes, is possible. An alternative segmentation is shown in Table 5.4.
102
Tense Surface Form Underlying Form
Present -ei -e -i
-voi -vo -i
Immediate -e -e
-vo -vo
Near -era -e -ra
-vora -vo -ra
Distant -erao -e -ra -o
-vorao -vo -ra -o
Remote -epa -e -pa
-va ???
Table 5.4: Segmentation of Realis Sufxes
This analysis isolates a morpheme -vo that is governed by verb stem classication; it occurs
with (but not ) verb stems. The sticking point for such an analysis is the remote past,
which would have to be analyzed as the combination of -vo with another morpheme. (For
ease of morphological glossing, the more supercial analysispositing sufxes sensitive to a
combination of verb classication and tensewill be provided for all example sentences.)
There is an uncommon verb conjugation described as the anticipatory mode by Firchow
(1987:21) that is potentially relevant here. It suggests the isolability of a sufx -vo that is
sensitive to the distinction between and : the form -iva occurs with inection, as illustrated
in (278), while the form -voiva occurs with inection, as illustrated in (279).
(278) varuere-a-iva
hunt-3PL
-ANTIC
They hunt animals and . . . [Firchow (1987:21)]

(279) kakae-to
child-SG.M
posiposi-re-voiva
dry.RDP-3SG.M
-ANTIC
veruta-va
skin ake-SG.F
kareke-o-i
appear-3SG.F
-PRES
rera
PRO.3.SG.M
vara-aro=ia
body-POSS=LOC
Once the boy had dried, skin akes appear on his body.
An alternative analysis of this form is to treat it as a combination of the present tense sufx
-(e)i/-voi plus -va (possibly identiable with the remote past tense sufx), as in (280).
12
12
Additional insight into the diachronic relationship of these sufxes may come fromthe morphological analysis
of the other dialects of Rotokase.g., Aita Rotokas, which possesses a larger phonemic inventory by conserving a
phonemic distinction that has been collapsed in Central Rotokas (Robinson, 2006)or its almost entirely undoc-
umented sister languages (Eivo, Keriaka, Rapoisi).
103
(280) varuere-a-i-va
hunt-3PL
-PRES
-?
They hunt animals and . . . [=(278)]
Present The marker of the present tense takes one of two forms: -ei and -voi. The sufx
-ei occurs with verbs and the sufx -voi occurs with verbs. This is illustrated with the
ambitransitive verb stem ori cook: the form -ei occurs with subject agreement in (281) and
the form -voi occurs with agreement in (282).
(281) Rave,
Rave,
vii
PPRO.2.SG
ori-pa-u-ei
cook-CONT-2SG
-PRES
oira-ra=pa
man-HUM.PL=BEN
ovusia
while
vii-pa
PPRO.2.SG=BEN
kovo-i-ve
work-3PL
-SUB
Rave, you cook for the men while they work for you.
(282) Ireviri
Ireviri
koorato
possum
siare-aro
innard-POSS
ori-re-voi
cook-3SG.M
-PRES
Ireviri is cooking the possums innards.

When verb stems lack TAM marking, they are interpreted as present tense, as illustrated for
the stem era sing in (283) and for the stem kipe cut in (284). (The missing present
tense sufx is explicitly realized as a null sufx for the purposes of illustration but is simply
omitted elsewhere.)
(283) koova-va=ia
sing-SG.F=LOC
viokeke-vira
whistle-ADV
era-pa-ro-
sing-CONT-3SG.M
-PRES
Kare
Kare
Kare is singing a song whistling.
(284) Pita
Pita
isisio
grass
kou
CLASS
kipe-re-
cut-3SG.M
-PRES
uvare
because
kepa
house
ruvara=ia
near=LOC
tou-pa-i-voi
be-CONT-3PL
-PRES
Peter is cutting the grass because it is close to the house.

Immediate Past The immediate past is used to describe events that took place either on
the same day as the present or one day prior to it. The marker of the immediate past can take
one of two forms: -e or -vo. The form -e occurs with stems, as in (285), while the form -vo
occurs with stems, as in (286).
(285) ora-upo-pa-a-e
RR-strike-CONT-3PL
-IP
oa iava
hence
eera=ia
DEM.MED.SG.M=LOC
evara
DEM.MED.PL.N
tou-pa-i
be-CONT-3PL
tapuku-ara
contusion-PL.N
They fought and thats why there are contusions on him.
104
(286) uva
so
apeisi
how
raga-vira
only-ADV
oira
PPRO.3.SG.F
upo-ri-vo
strike-2SG
-IP
And just how did you kill him? [Firchow and Akoitai (1974:44)]
Near/Distant Past The distant and near past are used to describe events that took place at
least one day prior to the present. The marker of the distant past tense takes one of two forms:
-vora or -era. The form -era occurs with stems, as in (287), and the form -vora occurs with
stems, as in (288).
(287) o-voki-vu=ia
SPEC-day-ALT=LOC
ava-ra
go-1SG
eisi=re
LOC=ALL
Ibu
Ibu
ovusia
while
ora-upo-pa-a-era
RR-hit-CONT-3PL
-NP
One day I went to Ibu while they fought. [=(258)]

(288) Vererire
Vererire
raroe-vira
widowed-ADV
tou-pa-e-veira
be-CONT-3SG.F
-HAB
uvare
because
vatua-to
spouse-SG.M
upo-i-vora
hit-3PL
-NP
uva
and
kopii-ro-era
die-3SG.M
-DP
oira
PPRO.3.G.M
arova
POST
Vererire is a widow because they killed her husband and he died leaving her behind.
The marker of the distant past takes one of two forms: -vorao or -erao. The form -erao
occurs with stems, as in (289) and the form -vorao with stems, as in (290).
(289) aako
mother
riako
FP.F
ora-vatevate-a-erao
RR-give.RDP-3PL
-NP
aue=ia
CONN=LOC
aio
food
The women gave each other food.
(290) Rarasori
Rarasori
oirara
people
vate-re-vorao
give-3SG.M
-DP
vuku-ara
book-PL.N
ra
and
vara
PPRO.3.PL
vuravura-i-ve
look.RDP-3PL
-SUB
vara
PPRO.3.PL
voreri-oro
repeat-DEP.SIM
Robinson gave people books so that they would look at them again and again.
Remote Past The remote past is used to describe events that took place in the historical
or mythological past, which is typically described using the phrase voari tuariri long ago, as
(291) poupou
dust.RDP
kovekove-o-i
fall.RDP-3SG.F
-PRES
uvare
because
Toki
Bagana
pokoro-viro-o-pa
erupt-RES-3SG.F
-RP
voari
back
tuariri.
long ago
Dust is falling because Mt. Bagana erupted a long time ago.
105
(292) tataga
log
evao-va
tree-SG.F
iava
POST
oa
RPRO.3.SG.N
toe-i-va
cut-3PL
-RP
voari
back
tuariri.
long ago
They cut the log from the tree a long time ago.
The marker of the remote past takes one of two forms: -epa or -va. The form -epa occurs
with stems, as in (293), and the form -va occurs with stems, as in (294).
(293) voea
PPRO.3.PL
rutu
very
oira-ra
man-HUM.PL
agasi-a-epa
be full-3PL
-RP
tugoro-pa-toa=ia
holy-DERIV-SG.M=LOC
uraura-to
spirit-SG.M
All of the men lled up with the holy spirit.
(294) kakate
bamboo
agasi-pie-re-va
be.full-CAUS-3SG.M
-RP
aue=ia
CONN=LOC
gau
tear
He lled the bamboo tube with tears.
5.2.2.7.2 Irrealis Within the category of irrealis, Rotokas possesses a number of subcate-
gories: the subjunctive, the future, and the habitual. Each has two markers: a p-inital form
which occurs on verbs with neuter subjects and a v-initial form that occurs with non-neuter
subjects.
13
The various markers for the irrealis categories are listed in Table 5.5.
Subject Gender
Mood Neuter Non-Neuter
Subjunctive -pe -ve
Future Near -pere -vere
Distant -perea -verea
Habitual -peira -veira
Table 5.5: Rotokas Irrealis Mood Categories
It should be clear from Table 5.5 that additional segmentation of the irrealis sufxes is
possible, as shown in Table 5.6.
13
According to Firchow (1987:15), the p-initial form occurs in verbs which have an inanimate subject, and the
v-initial formin verbs with animate subjects. The relevant variable is, however, gender, and not animacy, although
the two largely coincidesee 4.2.1.1.
106
Tense Surface Form Underlying Form
Irrealis -pe -pe
-ve -ve
Near Future -pere -pe -re
-vere -ve -re
Distant Future -perea -pe -re -a
-verea -ve -re -a
Habitual -peira -pe -ira
-veira -ve -ira
Table 5.6: Segmentation of Irrealis Sufxes
The segmentation found in Table 5.6 suggests that there is a basic irrealis category marked
by the sufxes -pe and -ve, which is subject to additional specication. This is particularly clear
in the case of the habitual, which is marked only by -pe or -ve when habituality is indicated
lexically with oisioa always, as in (295) or (296), but by -peira or -veira otherwise.
(295) Asitararia
Australia
oea
PPRO.3.PL.M
oisioa
always
Papua
Papua
Niugini
Niugini
toki-pa-i-ve
care for-CONT-3PL
-SUB
Australia always takes care of Papua New Guinea.
(296) tuariripairara
long ago-DERIV-HUM.PL
oea
PPRO.3.PL.M
oisioa
always
evao-ara
tree-PL.N
kogo-pa-i-ve
cut-CONT-3PL
-SUB
aue=ia
CONN=LOC
aveke-va
stone-SG.F
torara
axe
The ancestors always cut trees with a stone axe.
Subjunctive The subjunctive mode is marked by a sufx that takes one of two forms: -pe
or -ve. The form -pe occurs with neuter subjects, as in (297), whereas the form -ve occurs with
all non-neuter subjects, as in (298). Note that in both cases the grammatical subject is notionally
inanimate.
(297) iroiro
rope
vao
DEM.PROX.3.SG.N
kaitu-pie-ri
tight-CAUS-2SG
ra
and
kaitu-pe
tight-SUB
rutu
very
Tighten the rope and it will be tight.
(298) voo
here
oisioa
always
vegoto
jungle
tou-pa-re-ve
be-CONT-3SG.M
-SUB
ira
RPRO.3.SG.M
toe-i-va
cut-3PL
-RP
vo-urui-o
SPEC-village-?
pura-sia
make-DEP.SEQ
Here is the jungle that they cut in order to make this village.
107
The two forms of the sufx occur with both and verbs, as demonstrated for -ve in (299)
and (300).
(299) voari
back
tuariri
long ago
uva
and
oisioa
always
popote-pa-irara
white-DERIV-HUM.PL
torio-ara=ia
sword-PL.N=LOC
ora-upo-pa-a-ve
RR-hit-CONT-3PL
-SUB
Long ago white people would ght with swords.
(300) uriri-pa-ra-i
be.afraid-CONT-1SG
-PRES
teapi
lest
ragai
PPRO.1.SG
upo-i-ve
hit-3PL
-SUB
I am afraid that they might hit me.
The subjunctive marker occurs in a wide variety of contexts. A few of the contexts in which
it typically occurs are provided below: negation (301), conditionals (302), interrogatives (303),
apprehensionals (304), indirect commands (305), and situations of possibility (306) (Lichten-
berk, 1985; Bugenhagen, 1993; Palmer, 2001).
Negation
(301) regore-vira
bent-ADV
evao-va
tree-SG.F
iipa-erao
go up-NP
uvare
because
viapau
NEG
va
PRO.3.SG.N
viou-pa-re-ve
clean-CONT-3SG.M
-SUB
Ririvasi
Ririvasi
The tree grew crooked because Ririvasi didnt prune it.
Conditional
(302) Pita
Pita
Ruke
Ruke
tavi-pa-re-va
tell-CONT-3SG.M
-SUB
raerae-vira
try-ADV
reoreo-u
talk.RDP-2SG
vosia
if
aite
father
uvui-pa-ro
be able-CONT-3SG.M
ra
and
vigei
PRO.1.DL
uvu-re-ve
hear-3SG.M
-SUB
ra
and
vegei
PRO.1.DL
ato-re-ve
answer-3SG.M
-SUB
ikau-vira
hurry-ADV
Peter told Ruke, you try talking, if dad can hear us, he can reply quickly.
Interrogative
(303) irou-vai
who-INDEF
vao
DEM.PROX.SG.N
kae-ve
carry-SUB
oapa
bag
visii
PPRO.2.PL
vasie
CLASS
iava
POST
Who among you can carry my bag?
Apprehensional
108
(304) visivisi-vira
quite-ADV
reoreo-pa-ro-e
speak.RDP-CONT-3SG.M
-IP
Tavi
Tavi
oisio
COMP
teapi
lest
rera
PPRO.3.SG.M
uvu-i-ve
hear-3PL
-SUB
kaakau
dog
kare
FP
Tavi is speaking quietly lest the dogs hear him.
Indirect Command
(305) Riko
Riko
tavi-re-vo
tell-3SG.M
-RP
Pita
Peter
oisio
COMP
ra
and
kepa
house
pura-re-ve
make-3SG.M
-SUB
rera=pa
RPRO.3.SG.M=BEN
Peter told Riko to build a house for him.
Possibility
(306) evao-ara
tree-PL.N
rutu
very
toe-i-va
cut-3PL
-RP
uva
and
vuatoa
clear-SG.N
pura-piro-pa
make-RES-RP
oa iava
hence
uvui-pa-i
be able-CONT-PRES
ra
and
avaka-va
beach-SG.F
keke-pa-i-ve
see-CONT-3PL
-SUB
eisi
LOC
vara-vira
come down-ADV
They cut all of the trees and a clearing was made so that it was possible that way to look
down and see the beach.
Future Firchow (1987:20) describes a number of sufxes as markers of the future tense.
However, unlike the other tense-marking sufxes (e.g., the present tense), these sufxes are
sensitive to the gender of the subject, and not to the distinction between and inection.
Given this formal distinction between the past and present tense sufxes on the one hand and
the future tense sufxes on the other, it can be argued that the two classes of sufxes should be
assigned to different ontological categories.
Conation of future tense and potential/irrealis mood is fairly common cross-linguistically.
As Chung and Timberlake (1985:243) observe:
The future is thus a category where tense and mood merge. In practice many lan-
guages do not distinguish morphologically between future tense and potential (irre-
alis) mood. Where a difference is made, the future tense is used for events that are
presumed to be certain to occur, and the irrealis mood for events that are potentially
possible but not presumed to be certain.
Near Future The marker of the near future can take one of two forms: -pere or -vere. The
form -vere occurs with verb stems, as in (307), as well as with verb stems, as in (308).
(307) vavoisio
there
ava-pa-i-ei
go-CONT-1PL.EXCL-PRES
aue=re
CONN=ALL
oisio
COMP
ra
and
voa-va
here-ABL
kare-io-vere
return-1PL.EXCL-NF
vokiaro
night
Were going there in order that we come back at night.
109
(308) oire
okay
vii
PPRO.2.SG
va-aro
PPRO.3.SG.N-POSS
vuku-a
book-SG.N
kare-pie-a-vere
return-CAUS-3PL
-NF
Okay, Ill give you your book back.
The form -pere occurs with both verb stems, as in (309), as well as verb stems, as in
(310). In each case the subject of the verb with -pere is neuter.
(309) vasirako-vira
tight-ADV
rutu
very
rakorako-a
rope-SG.N
tuke-re
tighten-3SG.M
opuuruva
canoe
iava
POST
oiso
COMP
teapi
LEST
gavogavoto-pere
loose.RDP-NF
He tightened the rope on the canoe so that it will not loosen up.
(310) rigato-a-vai
write-SG.N-INDEF
veri
worthless
tou-pere
be-NF
The writings will not be worth anything in the future. [Firchow (1984)]
The realization of the near future is irregular for rst person dual subjects (Firchow, 1987:15),
as illustrated in (311).
(311) toaera-vira
work for money-ADV
kovo-pa-veare
work-CONT-1DL+DF
We two will work for money.
Distant Future The marker of the distant future takes one of two forms: -perea and -
verea. The form -verea occurs both with verb stems, as in (312), and with verb stems, as in
(313).
(312) oiraopie-pa-irara
believe-CONT-HUM.PL
eisi-re
LOC=ALL
ava-a-verea
go-3PL
-DF
vuvui
heaven
ua
CLASS
The believers are going to heaven.
(313) rovirovirie-a
judge-SG.N
pura-re-verea
make-3SG.M
-DF
pau-to
God-SG.M
utu-pa
follow-DERIV
voki=ia
day=LOC
vigei
PPRO.1.INCL
vo-pitupituro-aro=ia
SPEC-custom-POSS=LOC
vosia
when
viapau
NEG
vearo-vira
good-ADV
tou-pa-pe
be-CONT-SUB
vo-rasio=ia
SPEC-ground=LOC
God will measure us according to our habits when we arent good on earth.
The form of the distant future is irregular with rst person dual subjects, as illustrated in
(314).
(314) reoreo-a
talk.RDP-N
pura-si-va
make-3DL.M-RP
oisio
like
voo
here
ora-aivaropie-vearea
RR-meet-1DL+DF
The two of them arranged things, (saying) we will meet here.
110
Habitual The habitual mode is marked by a sufx that takes one of two forms, -peira or
-veira: the form -peira occurs with neuter subjects and the form -veira occurs with non-neuter
subjects, as illustrated in (315) and (316).
(315) asiga
type of vine
iro
vine
oa
RPRO.3.SG.N
virivoko-pa-peira
be milky-CONT-HAB
The asiga vine is usually milky.
(316) aapova
ying fox
iria
PPRO.3.SG.F
vokiaro
night
papa-pa-e-veira
y-CONT-3SG.F
-HAB
The ying fox ies at night.
The irrelevance of verb stem classication is evident from the fact that -peira and -veira
occur with both and stems, as illustrated for -peira in (317) and (318) and for -veira in (319)
and (320). Note that in example sentences illustrating the sufx -peira the stem classication is
not immediately obvious due to null subject agreement (typical for neuter subjects). However,
the two sentences illustrate the occurence of -peira with verb stems whose inectional pattern
is xed and known: ugoro cold is whereas tou be is .
5.2.2.7.3 -peira
(317) uuko-ara
water-PL.N
ugoro-pa-peira
cold-CONT-HAB
vavoisio
there
tutue=ia
Balbi=LOC
vosia
when
siisiu-u
wash-2SG
The water is always cold there on Mt. Balbi when you wash.
(318) oire
okay
oisio
COMP
raga-vira
only-ADV
iava
POST
tou-pa-peira
be-CONT-HAB
vo-rasi-toa=ia
SPEC-ground-SG.M=LOC
Okay, just like that they would always be on the ground. [Firchow and Akoitai
(1974:82)]
5.2.2.7.4 -veira
(319) virikoi-to
hatchet-SG.M
vearo-pa-ro-veira
good-CONT-3SG.M
-HAB
take
bamboo
tatu-pa-sia
chop-CONT-DEP.SEQ
A hatchet is always good for chopping bamboo.
(320) Pioto
Pioto
ira
RPRO.3.SG.F
aruo-va
mark-SG.F
pura-pa-re-veira
make-CONT-3SG.M
-HAB
aveke-ara=ia
stone-PL.N=LOC
Pioto (a river) always makes a mark on the stones.
The habitual mood is insensitive to tense, and is used to describe events regardless of tense,
as illustrated in (321), where it describes an event in the past tense, or in (322), where it is used
to describe a situation in the present tense.
111
(321) voari tuariri
long ago
uva
and
oisio
COMP
pura-pa-a-veira
say-CONT-3PL
-HAB
Sirovisia
Sirovisia
koke
rain
pura-pa-to
make-DERIV-SG.M
Long ago they said that Sirovisi was a rain maker.
(322) uuko-ara
water-PL.N
ugoro-pa-peira
cold-CONT-HAB
vavoisio
there
tutue=ia
Balbi=LOC
vosia
when
siisiu-u
wash-2SG
ra
COMP
uteo-u
cold-2SG
rutu
very
The water is always cold there on Mt. Balbi; if you bathe, youll be very cold.
The form of the habitual mood is irregular for rst person dual subjects, as illustrated in
(323).
(323) ari
but
visii
PPRO.2.PL
tauva-pa-veaira
help-CONT-1DL+HAB
The two of us always help you two. [Firchow and Akoitai (1974:53)]
5.3 Reduplication
Reduplication is a productive process for verb roots, where its semantic effect is generally one
of intensication, as illustrated in (324), or of iteration, as illustrated in (325).
(324) Maikol
Maikol
ito-va
banana-SG.F
goo-re-voi
bring down-3SG.M
-PRES
uva
and
raverave-o-i
dry.REDUP-3SG.F
-PRES
Michael brought down the banana and it looks dry.

(325) evao
tree
rao
branch
ruviruvi-re-voi
twirl.REDUP-3SG.M
-PRES
He is twirling the stick in his hands.

Reduplication is compatible with valency-changing derivations, and occurs with both the
reexive/reciprocal prex, as in (326), and the causative sufx, as in (327).
(326) uva
and
oavu=ia
another=LOC
ipa
ridge
ora-vorevoreri-pa-i-era
RR-go up-CONT-1PL.EXCL-RP
We repeatedly went up another ridge.

(327) Savia
Savia
veeta
bamboo
tou
CL
pokopoko-pie-e-voi
explode.REDUP-CAUS-3SG.F
-PRES
uvare
because
vo-tou
SPEC-CL
tovo-e-vo
put-3SG.F
-IP
tuitui kasi
re
sovara=ia
inside=LOC
Savia made the bamboo pop repeatedly because she put it in the re.
Whether reduplication is partial or full depends on the the rst syllable of the reduplicated
root, as previously described in 3.2.3.
112
5.4 Morphophonemics
There are a number of systematic morphophonemic alternations in Rotokas inectional mor-
phology. They can be divided into three groups on the basis of their effect on the form of verbal
conjugations.
5.4.1 Identical Vowel Sequences
Since Rotokas syllables are open (i.e., vowel-nal), the sufxation of vowel-initial sufxes (e.g.,
the neuter single -a and the neuter plural -ara) gives rise to vowel sequences. When the nal
vowel of a stem and the initial vowel of a sufx are identical (i.e., homorganic), the result is a
long vowel. This is not uncommon, given that slightly over half (29/51, 57%) of all sufxes are
vowel-initial. It is illustrated in some of the following words:
(328) a. veera
line up
line up (something)
b. veera-a
line up-SG.N
line
c. veera-ara
line up.RDP-PL.N
rows
d. veeveera
line up.RDP
line up (something) in rows
e. veeveera-a
line up.RDP-SG.N
rows
f. veeveera-ara
line up.RDP-PL.N
rows of rows
5.4.2 Deletion and Insertion Rules
The relationship between underlying and surface forms in verbal morphology is largely one-to-
one, with the exception of a few fairly straightforward deletion and insertion rules.
113
5.4.2.1 o-deletion
Another morphophonemic rule deletes o from the end of a sufx when it precedes another
sufx beginning with e (Firchow, 1987:1516). This is not simply a phonological rule, since
the sequence oe across a morpheme boundary is possible, judging from perfectly grammatical
forms such as those in (329).
(329) a. ava-ro-epa
go-3SG.M
-RP
He went.
b. aio-pa-o-e
eat-CONT-3SG.F
-IP
She ate.
The morphophonemic rule is stated formally in (330) and its effect can be seen in the con-
trast between (331) and (332).
(330)
_
io
vio
_
_
i
vi
_
_
_
_
ei
era
erao
epa
_
_
(331) iro-ara-vai
rope-PL.N-INDEF
ou-ta
get-2PL
ra
and
ava-vio
go-1PL.INCL
erako
rewood
ogata-sia
carry in worksack-DEP.SEQ
Get some ropes and well carry rewood in a worksack.
(332) evao
tree
toe-sia
cut-DEP.SEQ
ava-pa-vi-ei
go-CONT-1PL.INCL-PRES
kepa
house
pura-sia
make-DEP.SEQ
Lets go cut a tree to make a house.
5.4.2.2 e-deletion
Firchow (1987:15-16) states two rules that involve the deletion of e from the beginning of a
sufx when it follows a sufx ending with o or a. These rules are specic to particular mor-
phemes and are not general phonological rules, since sequences of oe and ae across morpheme
boundaries are perfectly grammatical, as already shown for oe in (329) and as shown for ae in
(333).
(333) a. ava-a-e
go-3PL
-IP
They went.
114
b. kovekove-pa-epa
fall.RDP-CONT-RP
It kept falling.
These two morphophonemic rules are given in (334) and (335) (Firchow, 1987:15-16).
(334)
_
_
era
erao
epa
_
_
_
_
ra
rao
pa
_
_
__
o
viro
_
(335) ei i
__
a
o
_
These rules are illustrated below:
(336) ava-pa-ra-i
go-CONT-1SG
-PRES
Buka
Buka
iare
POST
I am going to Buka
(337) poupou
dust.RDP
kovekove-o-i
fall.RDP-3SG.F
-PRES
uvare
because
Toki
Bagana
pokoro-viro-o-pa
erupt-RES-3SG.F
-RP
voari
back
tuariri.
long ago
Dust is falling because Mt. Bagana erupted a long time ago. [=(291)]
5.4.2.3 i-insertion
Firchow (1987:15-16) provides the two rules in (338) to account for the epenthetic i that occurs
when some sufxes are concatenated and the rst ends with e and the second begins with e.
(338)
_
_
era
erao
epa
_
_
_
_
iera
ierao
iepa
_
_
__
ere
ve
_
ei iei / e
The effect of (341) on the form of verbal conjugation is illustrated in (339) and (340):
(339) tavauru-rirei
teenager-3DL.F
eisi-re
LOC=ALL
ava-ere-i-e
go-3DL.F-EPEN-IP
Arawa
Arawa
The two teenage girls went to Arawa.
115
(340) uva
and
ora-vasie-ere-i-epa
RR-depart-3DL.F-EPEN-RP
oira=ia
PPRO.3.SG.F=LOC
era-pa-oro
sing-CONT-DEP.SIM
era-va
sing-SG.F
So the two women departed singing the song. [Firchow (1984)]
The two rules from (338) cannot be collapsed into a single, general rule, as in (341), since
it would wrongly predict epenthesis in cases where it does not in fact occure.g., (342) and
(343).
(341) . . . e-e . . . eie
(342) Sira
Sira
sisiro
mirror
kove-e-voi
drop-3SG.F
-PRES
aveke
stone
ivara
above
iare
POST
Sira dropped the mirror on top of the stone.
(343) aako-va
mother-SG.F
kakae-to
boy-SG.M
kaa-pie-e-voi
choke-CAUS-3SG.F
-PRES
aio-a=ia
food-SG.N=LOC
Mother made the boy choke with some food.
116
R
e
e
x
i
v
e
/
R
e
c
i
p
r
o
c
a
l
R
o
o
t
C
a
u
s
a
t
i
v
e
M
o
d
i
e
r
P
r
o
g
r
e
s
s
i
v
e
R
e
s
u
l
t
a
t
i
v
e
P
e
r
s
o
n
/
N
u
m
b
e
r
/
G
e
n
d
e
r
T
e
n
s
e
/
M
o
o
d
-1 0 1 2 3 4 5 6
ora- -pie
_
irao
raga
_
-pa
_
viro
piro
_
_
_
_
_
ra a
u ri
ro re
o e
io
ve
si
ta
a i
_
_
_
_
_
_
_
_
epa va
era vora
erao vorao
e vo
ei voi
_
_
_
_
pe ve
pere vere
perea verea
peira veira
_
_
_
_
_
_
arapa
sia
oro
_
_
Figure 5.2: Verbal Morphology
1
1
7
118
Chapter 6
Syntax
This chapter covers various aspects of the syntax of Rotokas that are not covered in the more
detailed examination of argument structure provided in the second part of this thesis. The
syntax of noun phrases is discussed in 6.1 while the remaining sections cover clause-level
syntax. The constituent order of declarative and interrogative sentences is discussed in sections
6.2.1 through 6.2.3 while negation is described in 6.2.4. Clause combining is covered in
6.3: 6.3.1 covers complementation, 6.3.2 concentrates on verb phrases, and 6.3.3 examines
coordination in general.
6.1 Noun Phrases
A noun phrase (NP) is a constituent headed by a nominal which behaves as a unit. In the
simplest case, it consists of a bare noun, but the head noun can be modied by a number of
different elements, giving rise to much more complex structures. A summary of the elements
found in Rotokas NPs is provided in Figure 6.1.
_
_
_
_
Adjective
Demonstrative
Numeral
Possessor
_
_
_
_
Noun
_
_
Classier
Possessive Pronoun
Relative Clause
_
_
Table 6.1: Elements of the Noun Phrase
6.1.1 Nominals
Instances of a head noun modied by more than one element are rare and difcult to elicit,
making investigation of the internal constituency of NPs difcult. Examples of nouns modied
by one of the elements in Figure 6.1 are provided in (344) through (350).
119
6.1.1.1 Demonstrative-Noun
(344) sisiarapa-toa
greasy-SG.M
irao
true
roo
DEM.3.SG.M
koora-to
possum-SG.M
Possum is very greasy.
6.1.1.2 Possessor-Noun
(345) vii
PRO.2.SG
vaisi-aro
name-POSS
kiro-ri
write-2SG
Write your name.

6.1.1.3 Adjective-Noun
(346) vego-a
jungle-SG.N
toe-pa-i
cut-CONT-3PL
oira-ra
man-HUM.PL
aire-pa
new-DERIV
kovo-vai=re
garden-INDEF=ALL
The men cut the bush for a new garden.
6.1.1.4 Numeral-Noun
(347) ora-veera-i
RR-line up-PRES
eapu
ant
kare
CLASS
katai
one
raiva
road
raga=ia
only=LOC
voka-oro
walk-DEP.SIM
Ants line up and walk in a single line.
6.1.1.5 Noun-Classier
(348) atari
sh
pitu-ro
CLASS-PL.CL
ata-pa-i-voi
swim-CONT-3PL
-PRES
avaka-va=ia
ocean-SG.F=LOC
ovusia
while
vo-pitu-ro
SPEC-CLASS-PL.CL
tue-pa-io-vo
wait-CONT-1PL.EXCL-IP
The schools of sh swam in the ocean while we waited for them.

6.1.1.6 Noun-Possessive Pronoun
(349) upiriko
sweet potato
kovo
garden
oave
PPRO.1.DL
eva
DEM.3.SG.N
vegei
PRO.1.DL
avukarei
married couple
Thats the sweet potato garden of us two married people.
6.1.1.7 Noun-Relative Clause
(350) tugara-to
spirit-SG.M
riro
big
kuukuuvu-to
lie-SG.M
ira
RPRO.3.SG.M
oira-ra
man-HUM.PL
keakea-pa-re-veira
deceive-CONT-3SG.M
-HAB
Satan is a big liar who deceives people.
120
6.1.2 Possession
There are three different strategies for marking possession in a noun phrase in Rotokas: 1) the
use of a post-nominal possessive pronoun; 2) the use of a possessive sufx -aro on the possessed
noun; 3) and the use of a possessive sufx -aro on a dummy pronoun, which agrees in person,
number, and gender with the possessed noun.
6.1.2.1 Possessive Pronoun
The rst strategy for marking possession is the use of a possessive pronoun that agrees with the
possessor in terms of person, number, and gender (see Table 4.12 for the full paradigm). The
possessive pronoun follows the possessed noun, as illustrated in (351) and (352).
(351) oire
okay
rera
PRO.3.SG.M
ragi-i-va
whip-3PL
-RP
voeao
PRO.3.PL.M
ovii-irara
offspring-HUM.PL
oaa
PPRO.1.SG
osa
as
rera=ia
PRO.3.SG.M=LOC
pitu-pa-a-va
hold-CONT-1SG
-RP
My children, they whipped him as I held onto him.

(352) kepa
house
oaive
PPRO.3.PL.M
eva
DEM.3.SG.N
oa
RPRO.3.SG.N
vura-pa-ri
look at-CONT-2SG
Thats everybodys house that youre looking at.

Firchow (1987:61) notes that the possessor can also be explicitly indicated by a personal
pronoun, in which case it occurs in a prenominal position, as in (353), but I have been unable
to nd instances of this type of construction in the materials available to me.
(353) ragai
PRO.1.SG
vaisi-a
name-SG.N
oaa
POSS.1.SG
my name [Firchow (1987:61)]
This form of possession marking is restricted to animate possessors due to the lack of neuter
possessive forms in the possessive pronoun paradigm (see 4.2.3.3).
6.1.2.2 Possession Marking on Possessed Noun
The most common form of possession marking takes the form of the possessive sufx -aro on
the possessed noun, preceded by the possessor. This form of possession can be described as
head-marking, to the extent that the possessed noun functions as the head of the noun phrase.
The posession marking in this construction is invariant in form, and does not agree with the
possessor in terms of person, number, or gender, as illustrated in (354) and (355).
121
(354) Luk
Luk
vo-kepa-aro
SPEC-house-POSS
pako-pi
slump-SUB
uvare
because
eru-erao
rot-NP
tuuta-ara
pole-PL.N
Lukes house is slumped over because the poles are rotten.
(355) urio-pa-ta-i
come-CONT-2PL-PRES
ragai
PRO.1.SG
vo-kepa-aro=ia
SPEC-house-POSS=LOC
tii
tea
tapi-sia
drink-DEP.SEQ
Come drink tea at my house.
This form of possession marking is the most common and covers various semantic relations,
such as ownership, inalienable possession (body parts), and kinship.
6.1.2.3 Ownership
(356) Raratuiri
Raratuiri
vo-kepa-aro
SPEC-house-POSS
goru-vira
strong-ADV
tou-pa-i-voi
be-CONT-3PL
-PRES
Raratuiris house is strong.

6.1.2.4 Inalienable Possession (Body Parts)
(357) ruruku-vira
underwater-ADV
roko-re-vo
enter-3SG.M
-IP
uuko-va
water-SG.F
sovara-aro
above-POSS
raga
only
ragai
PRO.1.SG
kokoto-aro
leg-POSS
pitu-sia
grab-DEP.SEQ
He swam just under the surface of the water in order to grab my leg.
6.1.2.5 Kinship
(358) Rausira
Rausira
avuka-to
old-SG.M
Siuparai
Siuparai
aite-aro
father-POSS
Rausira is old; he is Siuparais father.
Unlike possession marked by possessive pronouns, it occurs with inanimate possessors, as
(359) vori-a
pay-SG.N
goru-a-aro
strong-SG.N-POSS
vara-epa
go down-RP
voo=ia
here=LOC
uvare
because
oavu
another
vori
pay
ou-pa
get-DERIV
tupa-piro-pa
close-RES-RP
The strength of money went down here, because the gold mine (lit., another money
getter) closed.
(360) torae-aro
height-POSS
tutue
Balbi
pukui
mountain
riro
big
kaekae-a
long-SG.N
The height of Mt. Balbi is really great.
122
Possession is potentially recursive, leading to the left-branching stacking of possessors,
as illustrated by (361) and (362).
(361) aikara
EXCL
ava-ra
go-1SG
eisi-re
LOC=ALL
ragai
PRO.1.SG
vate-va-aro
friend-SG.F-POSS
vo-kepa-aro
SPEC-house-POSS
I will go to my friends house. [Cricket and Grasshopper]
(362) Pita
Peter
aite-aro
father-POSS
vo-kepa-aro-i
SPEC-house-POSS-?
Peters fathers house [Firchow (1987)]
6.1.2.6 Possession Marking on Dummy Pronoun
Another strategy for marking possession resembles the one previously described in 6.1.2.2
(and should be considered a subtype), except that possession is not marked on the possessed
noun itself, but rather on a dummy pronoun, which agrees with the possessed noun in terms
of person, number, and gender. This is illustrated for a masculine (kuvupato shirt), feminine
(torara axe), and neuter (voria money) possessed noun in (363) through (365), respectively.
(363) Pita
Peter
rera-aro
PRO.3.SG.M-POSS
kuvu-pa-to
cover-DERIV-SG.M
pogopogoro-to
oversized.RDP-SG.M
Peters shirt is oversized.
(364) irou
who
ragai
PRO.1.SG
oira-aro
PRO.3.SG.F-POSS
torara
axe
ou-vo
get-IP
Who took my axe?

(365) Samuel,
Samuel,
ragai
PRO.1.SG
va-aro
PPRO.3.SG.N-POSS
vori-a
money-SG.N
tavario-ri
exchange-2SG
Samuel, exchange my money.

In (363) through (365), the possessor immediately precedes the dummy pronoun, which in
turns immediately precedes the possessed noun. The possessor and dummy pronoun form a
syntactic unit, as can be seen in cases where the entire phrase is discontinuous, as in (366) and
(367), where the possessor functions as patient/theme and the possessor and dummy pronoun
appear on the right periphery.
(366) peeka
bad
eera
DEM.3.SG.M
oira-to
man-SG.M
ira
RPRO.3.SG.M
kuvu-pa-to
cover-DERIV-SG.M
kaviru-re-vo
steal-3SG.M
-IP
Pita
Peter
rera-aro
PPRO.3.SG.M-POSS
The man who stole Peters shirt was bald.
123
(367) auo
hey
sikure=va
grass skirt=COM
urio-u
come-2SG
ragai
PPRO.1.SG
oira-aro
PPRO.3.SG.F-POSS
Hey, come here with my grass skirt.
Discontinuous possessive noun phrases of the type found in (366) and (367) also occur in
other grammatical roles, as shown by (368) and (369), where a possessed noun functions as an
oblique argument and oblique marking occurs on the dummy pronoun rather than the possessed
noun itself.
(368) karisi-to
Christ-SG.M
vigei
PRO.1.PL.INCL
vara-aro=ia
PRO.3.PL.N-POSS=LOC
vuri-ara
bad-PL.N
kopii-ro-epa
die-3SG.M
-RP
Christ died for our sins.

(369) Rarasori
Rarasori
riro
big
vaisi
name
ou-re-voi
get-3SG.M
-PRES
rera
PRO.3.SG.M
va-aro=ia
PRO.3.SG.N-POSS=LOC
vearo
good
kovo
work
Robinson has a big name for his good work.
It may seem from examples such as (368) and (369) that this possessive construction is
required for possessed oblique arguments, but this is not the case, judging from sentences such
as (370) and (371), where a possessed noun plays the role of an oblique argument, and no
dummy pronoun is involved.
(370) kapu-a
sore-SG.N
eva
DEM.MED.SG.N
vii
PRO.2.SG
kokoto-aro=ia
leg-POSS=LOC
tou-pa-i
be-CONT-3PL
That sore is on your leg.

(371) ee
EXCL
rera
PRO.3.SG.M
vo-reo-aro=pa
SPEC-word-POSS=BEN
ora-toatoa-pa-u
RR-concede-CONT-2SG
Are you giving in to his talk? [Firchow (1984)]

6.1.3 Quantication
This section covers the various means of quantifying noun phrases in Rotokas. The use of rutu
very as a quantier is described in 6.1.3.1 and Rotokas numerals are described in 6.1.3.2.
6.1.3.1 Quantiers
The intensier rutu very, truly can be used in a noun phrase as a universal quantier that has
scope over the immediately preceding noun phrase. It occurs with both nouns, as in (372), and
pronouns, as in (373) and (374).
124
(372) kokootu
chicken
ruipa-pa-a-veira
want-CONT-3PL
-HAB
oira-ra
man-HUM.PL
rutu
very
uvare
because
vearopie-a
good-SG.N
rutu-a
very-SG.N
varu-a
meat-SG.N
Everyone (literally: all people) wants chicken because it is good meat.
(373) voea
PRO.3.PL.M
rutu
very
ora-pugu-pie-pa-a-epa
RR-waste time-CONT-RP
vate-irara
friend-PL.N
agiagi-pa-oro
greet-CONT-DEP.SIM
ora
and
agesi-pa-oro
laugh-CONT-DEP.SIM
All of them were busy greeting their friends and laughing.
(374) vigei
PRO.1.PL.EXCL
rutu
very
tetevu
sago
turaa-pa-vi-ei
sew-CONT-1PL.EXCL-PRES
kepa
house
iare
POST
All of us are sewing up sago for the house.
When a noun is quantied using rutu, case marking appears as an enclitic on the quantier
(rather than on the head noun itself), as illustrated in in (375) and (376). In essence, the case
marker marks the right boundary of the noun phrase.
(375) voki-ara
day-PL.N
rutu=ia
very=LOC
kovo-pa-sia
work-CONT-DEP.SEQ
ava-pa-ere
go-CONT-3DL.F
Every day the two of them went to work. [Caleb, Matevu]
(376) uva
so
vara
PRO.3.PL.N
rutu=va
very=COM
vore-ro-epa
go back-3SG.M
-RP
He returned with everything. [Firchow and Akoitai (1974:1,10:35)]

6.1.3.2 Numerals
Although the use of Tok Pisin for counting is increasingly widespread among Rotokas speakers,
the language does have an indigenous counting system, which is quinary (based on multiples of
ve), as can be seen in Table 6.2.
125
Number Rotokas Term
one katai
two erao
three peva
four resiura
ve vavae
six katai vatara
seven erao vatara
eight peva vatara
nine resiura vatara
ten katai tau
one-hundred vovoto
one-thousand tuku
one-million ipa
Table 6.2: Rotokas Numerals
Note that the term vavae ve is based on the body part term vavae hand, illustrated in
(377) and (378).
(377) eake-a
what-SG.N
eva
PRO.MED.3.SG.N
vii
PRO.2.SG
vavae-aro=ia
hand-POSS=LOC
Whats that in your hands?
(378) vavae-ara
hand-PL.N
itoro-pie-i-vo
extend-CAUS-3PL
-IP
kakae
child
vure
FP
uvare
because
voea
PRO.3.PL.M
tavi-e-vo
tell-3SG.F
-IP
tisa-va
teacher-SG.F
The children raised their hands because the teacher told them to.
Although quite large numbers can be built up using the numerals in Table 6.2, as illustrated
in (379), the use of Rotokas numerals is waning, particularly among the younger generation.
(379) erao
two
tuku
1000
resiura vatara
nine
vovoto
hundred
vo-peva
SPEC-three
tau
ten
vavae
ve
two-thousand nine-hundred and thirty-ve [Firchow (1987:46)]
Although use of Tok Pisin numerals is increasing, particularly among the younger genera-
tion, Rotokas numerals are still commonly used for smaller numbers (ten or less), as illustrated
in (380) and (381).
126
(380) ora-veera-i
RR-line up-PRES
eapu
ant
kare
FP
katai
one
raiva
road
raga=ia
only=LOC
voka-oro
walk-DEP.SIM
Ants line up and walk in a single line.
(381) vurei-a
feast-SG.N
pura-sia
make-DEP.SEQ
vo-peva
SPEC-three
upo-i-vo
kill-3PL
-IP
koue
pig
kare
FP
ora
and
aue
CONN
kokotu
chicken
kare
FP
voo=va
here=ABL
atoi-a
village-SG.N
In order to have a feast, they killed three pigs and also some chickens in the village.
6.1.4 Nominal Conjunction
Before discussing how nominal conjunction works in Rotokas, it is useful to establish some
terminology. The marking of coordination differs widely across languages. As Haspelmath
(2000) observes, some languages lack an explicit marker of conjunction (asyndetic), others
possess some form of explicit marking of conjunction, either on one of the two elements being
conjoined (monosyndetic) or both of them (bisyndetic). Rotokas is monosyndetic, as illustrated
in (382) and (383).
(382) Rake
Rake
ora
and
Jon
Jon
kaakau
dog
kare
FFP
ou-sia
get-DEP.SEQ
ava-si-e
go-3DL.M-RP
Rake and John went to get the dogs.

(383) Revoi
Revoi
ora
and
Siariviri
Siariviri
tutupie
close
siara
clan
rutu
very
Revoi and Siariviri are members of the same clan.
The coordination of two nouns referring to humans typically involves the use of the particle
vaio animate dual, as can be seen in (384) and (385).
(384) Visaevi
Visaevi
vaio
ANIM.DL
ora
and
Mataila
Mataila
atara-pa-ere-i-ei
sleep-CONT-3DL.F-EPEN-PRES
urua=ia
bed=LOC
Visaevi and Mataila are sleeping together in bed.
(385) Ararai
Ararai
kapokaporo-si-voi
carry-3DL.M-PRES
Visa
Visa
vaio
ANIM.DL
ora
and
Apoka
Apoka
Visa and Apoka are carrying Arari between their shoulders.
In a cross-linguistic survey of coordination, Haspelmath (2000) observes that the explicit
marking of coordination can appear either before the coordinand (prepositive) or after it (post-
positive). In Rotokas, coordination marking is prepositive, as can be seen from coordinated
noun phrases that are discontinuous, as in (386) and (387), where the second coordinand occurs
after the verb with ora. Furthermore, (387) demonstrates that the animate dual particle vaio is
associated with the rst coordinand.
127
(386) Polin
Polin
vaio
ANIM.DL
ora-ou-si-e
RR-get-3DL.M
ora
and
Tovisi
Tovisi
aruvea
yesterday
Polin and Tovisi married yesterday.
(387) Pita
Pita
vaio
ANIM.DL
ora-varovaro-raga-pa-si-ei
RR-compete-only-CONT-3DL.M-
ikau-pa-oro
run-CONT-DEP.SIM
ora
and
Raku
Raku
Peter and Raku competed against each other running.
Some additional examples of what appears to be the same construction type as (386) are
provided in (388) and (389).
(388) Riopeiri
Riopeiri
arao-rei
brother-DL.CL
ora
and
Vaviata
Vaviata
ava-si-e
go-3DL.M-IP
eisi
LOC
Buka
Buka
The brothers Riopeiri and Vaviata went to Buka.
(389) Jon
Jon
vaio
ANIM.DL
evaiterei
DEM.MED.DL.M
ora
and
Raki
Raki
aiterea
RPRO.3.DL.M
urio-pa-si-ei
come-CONT-3DL.M-PRES
John and Raki slept.

6.2 Intraclausal Syntax
This section covers various aspects of intraclausal (i.e., clause-internal) syntax, such as the basic
ordering of constituents, the difference between declarative and interrogative word order, and
negation.
6.2.1 Constituent Order
Firchow (1973:x-xi) provides three templates for what he labels basic sentences, which are
provided in (390) (where parentheses indicate optional elementsi.e., elements that can be
elided when contextually retrievable).
1
(390) INTRANSITIVE (Time) (S) (Location) (Verb) (Adverb) Verb
TRANSITIVE (Time) (A) O Verb (Adverb) (Verb) (Location)
DITRANSITIVE (Time) (A) IO O Verb (Adverb) (Location)
The constituent order provided in (390) represent the typical ordering of elements but al-
ternative orderings of these elements are permissible. For example, the time word (or phrase)
occurs sentence-initially in (391), as predicted by (390), but not in (392), where it occurs after
the intransitive subject.
1
Firchow (1973) uses the undifferentiated term Subject, which have been replaced with S and A in (390) for
the sake of consistency with the terminology used to describe grammatical roles in 7.3.2.
128
(391) koke-va
rain-SG.F
voki-ara
day-PL.N
rutu=ia
very=LOC
kove-pa-o-i
fall-CONT-3SG.F
-PRES
It rains every day. [Firchow (1984)]

(392) aveke=ia
stone=LOC
ora-tuguru-ra-e
RR-bump-1SG
-IP
vokiaro
night
eisi
LOC
raiva
road
I bumped into a rock at night on the road.
The distinction between arguments and adjuncts (see 7.3.2) goes a long way towards ex-
plaining the constituent ordering principles of Rotokas. Arguments are more restricted in their
ordering whereas adjuncts are fairly free. For example, manner adverbs can in fact occur in any
of the logically possible positions of an intransitive or transitive clause. Therefore, all of the
intransitive sentences in (393) are grammatical, as are the transitive sentences in (394).
(393) a. oira-to
man-SG.M
tori-re-va
run away-3SG.M
-RP
gapu-vira
naked-ADV
The man ran away naked.
b. oira-to
man-SG.M
gapu-vira
naked-ADV
tori-re-va
run away-3SG.M
-RP

c. gapu-vira
naked-ADV
oira-to
man-SG.M
tori-re-va
run away-3SG.M
-RP

(394) a. oirato
man-SG.M
koie
pig
kaviru-re-vo
steal-3SG.M
-IP
ikau-vira
quick-ADV
The man quickly stole the pig.
b. oira-to
man-SG.M
koie
pig
ikau-vira
quick-ADV
kaviru-re-vo
steal-3SG.M
-IP

c. oira-to
man-SG.M
ikau-vira
quick-ADV
koie
pig
kaviru-re-vo
steal-3SG.M
-IP

d. ikau-vira
quick-ADV
oira-to
man-SG.M
koie
pig
kaviru-re-vo
steal-3SG.M
-IP

129
Core arguments, however, followmore strict principles. The transitive template is illustrated
for a transitive verb in (395), where its core arguments, A and Oare oirato man and koie pig,
respectively.
(395) oira-to
man-SG.M
koie
pig
upo-re-vo
hit-3SG.M
-IP
The man hit the pig.

Although it is also possible for the subject to occur postverbally, as illustrated by (396),
other logically possible orderings are ungrammatical on the intended reading.
2
(396) koie
pig
upo-re-vo
hit-3SG.M
-IP
oira-to
man-SG.M
The man hit the pig.
All other logically possible ordering are ungrammatical: VAO, as in (397a); VOA, as in
(397b); OAV, as in (397c); and AVO, as in (397d).
(397) a. * uporevo oirato riakova
b. * uporevo riakova oirato
c. * riakova oirato uporevo
d. * oira-to
man-SG.M
upo-re-vo
hit-3SG.M
-IP
riako-va
woman-SG.F
The man hit the woman.
The constituent order of objects is strict compared to that of subjects, with objects occuring
in a xed preverbal position, as illustrated in (398).
(398) oira-to
man-SG.M
vuri-va
bad-SG.F
kaakau
dog
upo-pa-re-voi
hit-CONT-3SG.M
-PRES
The man is hitting the bad dog.

Although the position of O must be lled, it is possible for it to be discontinuous. Compare
(399) with (400), where the NP vuriva kaakau bad dog is split: vuriva bad precedes the
verb and kaakau dog follows it.
(399) oira-to
man-SG.M
vuri-va
bad-SG.F
kaakau
dog
upo-pa-re-voi
hit-CONT-3SG.M
-PRES

(400) oira-to
man-SG.M
vuri-va
bad-SG.F
upo-pa-re-voi
hit-CONT-3SG.M
-PRES
kaakau
dog
2
If A and O have the same features for person, number, and gender, a change in word order may result in a
reversal of meaning rather than ungrammaticality.
130
6.2.2 Displacement of O
Although objects cannot freely move from their preverbal position, there are possibilities for
right-displacement to a post-verbal position, although they are subject to syntactic constraints.
Pronominal objects must occur preverbally in situ, whereas full NP (i.e., non-pronominal) ob-
jects can be dislocated to a postverbal position, either in part, as in (401), or in full, as in (402).
(401) oira-to
man-SG.M
vuri-va
bad-SG.F
upo-pa-re-voi
hit-CONT-3SG.M
-PRES
kaakau
dog
(402) oira-to
man-SG.M
aue
CONN
upo-pa-re-voi
hit-CONT-3SG.M
-PRES
vuri-va
bad-SG.F
kaakau
dog
When the entire NP is right-dislocated, the word aue (glossed as CONN for connector) occurs
as a trace of the right-dislocated argument canonical position. It does not occur, however, when
there is a modier to the right dislocated noun that can be stranded in the canonical position, as
shown by the ungrammaticality of (403) and (404).
(403) * oira-to
man-SG.M
vuri-va
bad-SG.F
aue
CONN
upo-pa-re-voi
hit-CONT-3SG.M
-PRES
kaakau
dog
(404) * oira-to
man-SG.M
aue
CONN
vuri-va
bad-SG.F
upo-pa-re-voi
hit-CONT-3SG.M
-PRES
kaakau
dog
Not all noun phrases behave the same way when right-displaced. Pronouns cannot be right-
displaced and right-displaced classiers behave somewhat differently fromright-displaced com-
mon nouns. The noun and its associated classier function as a unit (a classier phrase), and
right displacement requires the movement of the entire phrase, as shown in (405) and (406), and
it is possible (though not obligatory) for the classier to occur twice, as illustrated in (407).
(405) oira-to
man-SG.M
takura
egg
isi
CLASS
aio-re-va
eat-3SG.M
-RP
The man ate an egg.

(406) oira-to
man-SG.M
aue
CONN
aio-re-va
eat-3SG.M
-RP
takura
egg
isi
CLASS
The man ate an egg.
131
(407) oira-to
man-SG.M
aue
CONN
isi
CLASS
aio-re-va
eat-3SG.M
-RP
takura
egg
isi
CLASS
The man ate an egg.
If a right-displaced argument consists of two coordinated noun phrases, the coordinated
noun phrase is optionally preceded by aue.
(408) oira-to
man
aue
CONN
vori-re-vo
buy-3SG.M
-IP
torara
axe
ora
and
sigo-a
knife-SG.N
The man bought an axe and a machete.
(409) oira-to
man-SG.M
aue
CONN
vori-re-vo
buy-3SG.M
-RP
torara
axe
ora
and
aue
AUE
sigo-a
knife-SG.N
The man bought an axe and a machete.
The use of aue for right displacement of constituents is not limited to objects but appears to
extend to oblique arguments, as well, as illustrated by some of the following sentences.
(410) rotokasi-pa-irara
Rotokas-DERIV-HUM.PL
aue=pa
CONN=BEN
ruipa-pa-a-veira
want-CONT-3PL
-HAB
kaukau
sweet potato
The Rotokas like sweet potatoes.
When the right-displaced argument is normally case-marked, the case-marking occurs on
aue. If the right-displaced noun phrase is a complex coordinated noun phrase, the case-marking
is optional on the coordinated noun phrase.
aue=pa
CONN=BEN
ruipa-pa-a-veira
want-CONT-3PL
-HAB
kaukau
sweet potato
ora
and
sioko
chayote
The Rotokas like sweet potatoes and chayote.
aue=pa
CONN=BEN
ruipa-pa-a-veira
want-CONT-3PL
-HAB
kaukau
sweet potato
ora
and
aue
CONN
sioko
chayote
aue=pa
CONN=BEN
ruipa-pa-a-veira
want-CONT-3PL
-HAB
kaukau
sweet potato
ora
and
aue=pa
CONN=BEN
sioko
chayote
132
Right displacement of constituents is fairly common and occurs in other contexts, as well.
Some examples are provided in (414) and (415) (see also 6.1.2.6).
(414) Rita
name
vearopie-a=ia
pretty-SG.N=LOC
aasii-pa-o-i
wear beads-CONT-3SG.F
-PRES
aasii
bead
ua
CLASS
Rita puts on pretty beads. [Firchow (1984)]
(415) Kate
Kate
siopu-a=ia
soap-SG.N=LOC
sisiu-pa-o-i
wash-CONT-3SG.F
-PRES
Pita
Peter
va-aro
PRO.3.SG.N-POSS
Kate washed with Peters soap.
6.2.3 Interrogatives
Yes-no questions in Rotokas do not differ in form from statements. Content questions are
formed by replacing the questioned constituent with a question word (wh-word). Question
words occupy a sentence-initial position, as illustrated in (416) and (417).
(416) apeisi
how
ora-siovo-pa-u
RR-feel-CONT-2SG
vovokio
today
How do you feel today?
(417) auo
hey
ovu
where
iare
POST
ava-pa-u-ei
go-CONT-2SG
-PRES
Hey, man, where are you going?

As can be seen from (417), question words can be morphologically modied in the same
way as other nouns. For example, the question words eake what and ovu where both occur
with the enclitic =re in (418) and (419).
(418) eake=re
what=ALL
tara-pa-ri
look for-CONT-2SG
What are you looking for? [Firchow and Akoitai (1974:52)]

(419) ovu=re
where=ALL
ava=pa -u-ei
go-CONT-2SG
-PRES
Where are you going?

The sufx -pa occurs with the question word eake what in order to form questions of
reason, cause, or motive, as illustrated in (420) and (421). In such questions, eake sometimes
co-occurs with the indenite sufx -vai, as illustrated in (422).
(420) eake=pa
what=BEN
koikoi-pa-ri
groan-CONT-2SG
Why are you groaning?

133
(421) eake=pa
what=BEN
vii
PPRO.PER.2.SG
upo-re-vo
strike-3SG.M
-IP
Why did he hit you?

(422) eake-vai-pa
what-INDEF=BEN
voeao
DEM.PROX.PL.M
riako-va
woman-SG.F
tova-pa-i
bury-CONT-3PL
Why are they burying the woman? [Firchow and Akoitai (1974:27)]
Question words in some cases appear as other parts of speech, as illustrated in (423), where
the question word apeisi what, how functions as a verb, occuring with the causative sufx
-pie.
(423) ovu=ia
where=LOC
apeisi-pie-pa-i-voi
how-CAUS-CONT-3PL
-PRES
Theyre doing it where why? [Firchow, 1974: 69]

6.2.4 Negation
Negation in Rotokas is expressed by means of the negator viapau not/nothing. In a simple
intransitive clause, negation precedes the verb, as in (424). It is questionable whether nega-
tion can follow the verb, as in (425). While a minority of speakers judge such sentences as
grammatical, they are nevertheless unattested in the materials available to the author.
(424) viapau
NEG
roru-a-voi
happy-1SG
-PRES
I am not happy.
(425) ? roru-a-voi
happy-1SG
-PRES
viapau
NEG
I am not happy.
In simple transitive sentences, negation must precede the verb, but can either occur before
the object, as in (426), or immediately preceding the verb, as in (427).
(426) viapau
NEG
vii
PRO.2.SG
too-a
hit-1SG
I wont hit you.

(427) vii
PRO.2.SG
viapau
NEG
too-a
hit-1SG
I wont hit you.

Negation cannot follow the verb, and (428) would therefore be ungrammatical.
134
(428) * vii
PRO.2.SG
too-a
hit-1SG
viapau
NEG
I wont hit you.
There is another form of negation that takes an entire clause, or sentence, in its scope. It in-
volves the use of viapau with the complementizer oisio at the left periphery of the clause/sentence
(see also 6.3.1).
(429) Kare
Kare
uvagi-to
deaf-SG.M
viapau
NEG
oisio
COMP
ra
and
reo-ara-vai
word-PL.N-INDEF
uvu-pa-re-ve
hear-CONT-3SG.M
-SUB
ari
but
gisipo
mouth
raga=va
only=COM
situe-pa-re-vere
watch-CONT-3SG.M
-?
Kare, the deaf, he doesnt hear talk, but he can read lips.
(430) eaviova
EXCL
viapau
NEG
oisio
COMP
Timoti
Timothy
voo
here
urio-ro-e
come-3SG.M
-IP
No, Timothy isnt coming here.

Constituent negation is also accomplished by means of viapau. This involves the use of
viapau immediately preceding the negated constituent, whether it is a noun, as in (431), or a
pronoun, as in (432).
6.2.4.1 Noun
(431) akuku-a
empty-SG.N
viapau
NEG
oavuavu-vai
something-INDEF
voo-ia
here=LOC
Its empty, theres nothing (literally: isnt something) here.
6.2.4.2 Pronoun
(432) viapau
NEG
rutu
very
iria-vu
PPRO.3.SG.F-ALT
uvui-pa-o-i
be able-CONT-3SG.F
-PRES
ra
and
upe
Upe
ua
CLASS
situe-pa-e-ve
look at-CONT-3SG.F
-SUB
No woman is able to look at the Upe wearers. [Firchow (1974b:23)]
6.3 Interclausal Syntax
This section covers a few aspects of interclausal (i.e., between-clause) syntaxthat is, the syn-
tax of clause combining. Complementation is discussed in 6.3.1; the syntax of verb phrases
is covered in 6.3.2; and the syntax associated with combinations of larger clausal units (up to
and including sentences) is covered in 6.3.3.
135
6.3.1 Complementation
This section looks at complementation, which can be described as the syntactic situation that
arises when a notional sentence or predication is an argument of a predicate (Noonan, 1985).
There are a number of predicates that license full clause arguments.
For example, the verb tavi tell can take an entire clause as an argument, in which case it is
marked by the particle oisio as, as illustrated in (433), where the complement clause reports
an impending event; in (434), where the embedded clause is direct speech; and in (435), where
the embedded clause consists of a non-verbal predicate.
(433) oirao-pa-vira
true-DERIV-ADV
visii
PRO.2.PL
tavi-pa-a-voi
tell-CONT-1SG
-PRES
oisio
COMP
kansol
council
urio-pa
-CONT
vigei=pa
PROG.1.PL.INCL
reo
talk
vate-sia
give-DEP.SEQ
I tell you truthfully that the council is coming to talk to us.
(434) Vare
Vare
igei
PRO.1.PL.EXCL
tavi-re-vo
tell-3SG.M
-IP
oisio
COMP
o-vuuta-vu
ALT-time-ALT
epao
oa=ia
RPRO.3.SG.N=LOC
vore-ra-vere
return-1SG
-NF
visii
PRO.2.PL
keke-sia
see-DEP.SEQ
Vare told us that on another occasion, I will come back to see you guys.
(435) vii
PRO.2.SG
tavi-pa-a
tell-CONT-1SG
oirao-pa-vira
true-CONT-ADV
oisio
COMP
vori-a-aro
buy-SG.N-POSS
vuri-to
bad-SG.M
kopi-a
die-SG.N
raga
only
I tell you truly that the wages of sin is death.
In some cases, the complement clause is marked only by oisio. However, in other cases,
oisio co-occurs with the coordinator ra. This is primarily restricted to cases where the meaning
of the verb tavi is to instruct or to telli.e., directive or jussive contexts.
(436) Rutu
Rutu
Siko
Siko
tavi-e-voi
tell-3SG.F
-PRES
oiso
COMP
ra
and
raverave-vira
weaken.RDP-
rarau
ower
ua
CLASS
pore-e-ve
turn-3SG.F
-SUB
Ruth told Siko to gently bend the ower.
(437) kakae-to
child-SG.M
tavi-e-voi
tell-3SG.F
-PRES
aako-va
mother-SG.F
oisio
COMP
ra
and
goro-ara
snot-PL.N
sii-ere-ve
wipe-3DL.F-SUB
Mother is telling the child that they should wipe away the snot.
The co-occurence of oisio and ra is not specic to the verb tavi or to complementation, since
oisio and ra co-occur outside of the context of complementation, as in(438) through (440).
136
(438) eto
re
tara-pa-ro-e
seek-CONT-3SG.M
-IP
Siape
Siape
oisio
COMP
ra
and
eto
re
kasi-ro
burn-3SG.M
Siape was searching for re in order to make a re.

(439) Samuel=ia
Samuel=LOC
vootu-a-epa
vote-3PL
-RP
oisio
COMP
ra
and
voea=pa
PRO.PL.M=BEN
tore-pa-ro
stand up-CONT-3SG.M
eisi
LOC
pareveri
parliament
They voted for Samuel in order for him to stand up in parliament.
(440) Ruben
Ruben
sikuru-sia
school-DEP.SEQ
ava-ro-e
go-3SG.M
-
eisi
LOC
sikuru-a
school-SG.N
oisio
COMP
ra
and
tarai-a-vai
learn-SG.N-INDEF
ou-re-ve
get-3SG.M
-SUB
Ruben went to school so that he would obtain knowledge.
On the basis of examples such as (436) or (437), one might conclude that oisio ra has
a purposive meaning, and that the common thread between the various usages is purposive
semantics. However, there are clearly instances where the rst clause and the second clause are
conjoined with oisio ra but the situation described does not display clear purposive semantics
i.e., it would be difcult to interpret a sentence such as (441) with a purposive reading (hence
the strageness of a gloss such as Raki is unconscious with the purpose of dying).
(441) Raki
Raki
kokopeko-pa-ro-i
unconscious.RDP-CONT-3SG.M
-PRES
oisio
COMP
ra
and
kopii-ro
die-3SG.M
Raki is unconscious and about to die.

The conjunction ra also occurs by itself, without oisio, as illustrated in (442) and (443).
(442) tupa
door
kapu-pie-a
tight-CAUS-1SG
goru-vira
strong-ADV
rutu
very
ra
and
viapau
NEG
ira-i
RPRO.3.SG.N-?
va
PRO.3.SG.N
karu-re-ve
open-3SG.M
-SUB
I close the door very strongly and nobody can open it.
(443) aue
hey
koetaova-pa-re
arrange marriage-CONT-3SG.M
aite-to
father-SG.M
ra
and
ora-tuutuuko-a-ve
RR-repay-3PL
-SUB
riako-rirei=ia
woman-DL.F=LOC
Hey, father arranged things and they will make a payment exchange for the two women.
For a general overview of clause combining in Rotokas, and further discussion of ra, see
6.3.3.
137
6.3.2 Verb Phrases
There are two constructions that involve more than one verb in a clause without explicit co-
ordination in the form of a coordinator such as ora: dependent verbs (previously discussed in
5.2.2.5) and aspectual verbs.
6.3.2.1 Dependent Verbs
In 5.2.2.6, two patterns of inection were described: independent and dependent. Independent
verbs are fully inected for person, number, and gender as well as tense, aspect, and mood
whereas dependent verbs are inected for neither. The order of independent and dependent
verb relative to one another is fairly exible. Although independent verbs generally precede
dependent verbs, as in (444), the reverse situation is also found, as in (445).
(444) toupievira
still
urio-ra-vere
come-1SG
-NF
vii
PRO.2.SG
keke-sia
see-DEP.SEQ
I will still come to see you.
(445) vii
PPRO.2.SG
keke-sia
see-DEP.SEQ
vore-pa-ra-i
return-CONT-1SG
-PRES
vokipavira
tomorrow
Ill return to see you tomorrow.
The subject is the only argument that is necessarily shared between the independent verb of
a clause and any dependent verbs. Co-reference between two non-subjects requires the use of a
pronoun, as in (446) and (448), where the patient/theme of the independent verb is co-referential
with the patient/theme of the dependent verb and the independent verbs patient/theme is real-
ized as a full NP while the dependent verbs patient/theme takes the form of a coreferential
pronoun.
(446) poris-irara
police-HUM.PL
oira-to
man-SG.M
ou-i-voi
get-3PL
-PRES
rera
PPRO.3.SG.M
tuuke-sia
lock.up-DEP.SEQ
uvare
because
riako-va
woman-SG.F
kopii-pie-re-vora
die-CAUS-3SG.M
-PRES
The police are getting the man to jail him because he killed a woman.
(447) Raviata
Raviata
Terita
Terita
ruvaru-re-voi
treat with medicine-3SG.M
-PRES
rera
PRO.3.SG.M
aavito-oro
cure-DEP.SIM
Raviata treated Terita with medicine, curing him.
Argument sharing does not occur, even between multiple dependent verbs with the same
patient/theme, as in (448), where the object of the dependent verb peopeopaoro is coreferential
with the object of the dependent verb vikipiesia (i.e., realized as a coreferential pronoun in its
second occurence as a direct object rather than simply being shared by the two verbs).
138
(448) oira-to
man-SG.M
peopeo-pa-oro
push-CONT-DEP.SIM
utu-a-e
follow-3PL
-IP
rera
PPRO.3.SG.M
viki-pie-sia
fall-CAUS-DEP.SEQ
eisi
LOC
uuko
water
vaga-pa
fall-DERIV
They followed behind the man pushing him in order to make him fall off the waterfall.
6.3.2.2 Aspectual Verbs
The verb roots rovo start, precede and ovoi nish both take a single argument and show
agreement by default, as can be seen in (449) and (450).
(449) vosia
if
parura-to
blister-SG.M
rovo-pa-ro
start-CONT-3SG.M
ra
and
rera
PPRO.PER.3.SG.M
rata
heat up
aue=ia
CONN=LOC
viivi=ia
betel nut husk=LOC
If a blister starts, heat it up with a betel nut husk.
(450) ovoi-ra-i
nish-1SG
-PRES
ari
but
riro-vira
big-ADV
rutu
very
aio-a-voi
eat-1SG
-PRES
uva
and
vukuu-ra-i
ll up-1SG
-PRES
Im nished but I ate a lot and I lled up.

These verbs also serve to provide aspectual information in a clause, in which case they are
able to co-occur either with a bare verb stem, as in (451), or with a dependent verb, as in (452).
(451) Ibu
Ibu
iava
POST
aapaapau
visit
rovo-ro-epa
start-3SG.M
-RP
He came rst from Ibu to visit.

(452) koko-a
ower-SG.N
rovo-pa-i
start-CONT-PRES
karu-pa-oro
open-CONT-DEP.SIM
The ower is starting to open up.
The form of verbal agreement found on aspectual verbs is dependent upon the classication
of the verb with which they co-occur. Aspectual verbs are if they occur alone, as already seen
in (449) and (450) or if they occur with a dependent verb, as in (453).
(453) ragai
PPRO.PER.1.SG
rovo-pa-ra-i
start-CONT-1SG
-PRES
kiro-pa-oro
write-CONT-DEP.SIM
vukua=ia
book=LOC
I am starting to write in the book.
However, when aspectual verbs occur with bare verb stems, they take the form of agreement
dictated by the bare verb stem. If the aspectual verb occurs with a bare verb stem, it will show
its usual classication, as illustrated in (454) and (455).
139
(454) Tesia
Tesia
avaio-va
rst born-SG.F
iria
RPRO.3.SG.F
kavau
be born
rovo-o-ra
start-3SG.F
-DP
Tesi the rst-born was born rst.

(455) voea
PRO.3.PL
rutu
very
koova
sing and dance
rovo-pa-a-vere
start-CONT-3PL
-NF
pupi-pa-oro
play pipe-CONT-DEP.SIM
All of them will start dancing, blowing the pipes and singsing. [Firchow (1984)]
However, the aspectual verb shows agreement when it occurs with any verb root or stem
that shows agreementfor example, a bare verb root, such as the monovalent verb root tou
be in (456); a labile verb root that takes a direct object, such as aio eat in (457); and a
causative verb stem, such as atepie make wait in (458).
(456) oire
okay
uva
and
rera=re
PRO.3.SG.M=ALL
voreri-vira
return-ADV
keera-a-epa
call-3PL
-RP
roo
DEM.3.SG.M
ira
RPRO.3.SG.M
voosi-vira
blind-ADV
tou
be
rovo-pa-re-ve
start-CONT-3SG.M
-HAB
So for the second time they called the man who had been blind [John 9:24]
(457) Rarairi
Rarairi
varu-ara=ia
meat-PL.N=LOC
ovoi-pa-ro-i
nish-CONT-3SG.M
-PRES
uvare
because
upiriko-ara
sweet.potato-PL.N
aio
eat
rovo-re-voi
start-3SG.M
-PRES
Rarairi is last when it comes to the meat because he started eating sweet potato.
(458) oira-to
man-SG.F
ate-pie
wait-CAUS
rovo-ri
precede-2SG
osia
as
kovo-re-ve
work-3SG.M
-SUB
Wait for the man while he works.
6.3.3 Coordination
Coordination in Rotokas is accomplished by means of various particles discussed previously in
4.2.9. The most basic formof coordination (that is, the coordination of non-clausal constituents
nouns, adverbs, verbs) is accomplished by means of the conjunction ora, which is illustrated
in (459), where two nouns are coordinated; in (460), where two oblique adjuncts are coordi-
nated; in (461), where two adverbs are coordinated; in (462), where two temporal nouns are
coordinated; and in (463), where two dependent verbs are coordinated.
6.3.3.1 Coordination of NPs
(459) sigo-a
knife-SG.N
ora
and
torara
axe
oarea
RPRO.3.PL.N
vearo-vira
good-ADV
tou-pa-i
be-CONT-3PL
The knife and the axe, they are ne.

140
6.3.3.2 Coordination of Obliques
(460) avaraosi
type of grasshopper
kare
FP
oea
RPRO.3.PL.M
voo
here
tou-pa-i-veira
be-CONT-3PL
-HAB
tego-ara=ia
wild.banana-PL.N=LOC
ora
and
vo-garavesi-ara=ia
SPEC-pandanus-PL.N=LOC
tapo
also
Avaraosi grasshoppers, they live on wild banana and also on pandanus.
6.3.3.3 Coordination of Adverbs
(461) kaakauko
type of beetle
kore
insect
kare
FP
iava
POST
oira
PPRO.3.SG.F
iria
RPRO.3.SG.F
iava
POST
vara
body
ua
CLASS
vurivuri-vira
brown-ADV
ora
and
kaapo-vira
white-ADV
tou-pa-i-veira
be-CONT-3PL
-HAB
The bodies of kaakavuko insects are brown and white.
6.3.3.4 Coordination of Temporal Nouns
(462) kaku-va
frog-SG.F
iria
RPRO.3.SG.F
gau-pa-e-veira
cry-CONT-3SG.F
-HAB
ovaiaro-vi
evening-DIM
ora
and
avitoava
afternoon
The frog cries in the afternoon and at night.
6.3.3.5 Coordination of Dependent Verbs
(463) avata-pa-to
spirit house-DERIV-SG.M
tugara-to
ghost-SG.M
ira
RPRO.3.SG.M
oisioa
always
keera-pa-i-ve
beckon-CONT-3PL
-SUB
upo
war
pura-pa-sia
make-CONT-DEP.SEQ
ora
and
aio
food
kovo-ro
garden-PL.CL
pura-pa-sia
make-CONT-DEP.SEQ
The ghosts of the spirit house always call to make war and to make gardens.
Whereas ora is primarily used to conjoin noun and verb phrases, ra (an apparent phono-
logical reduction of ora) is used to conjoin clauses. Loosely speaking, it functions to conjoin
clauses that are tightly connected (see previous discussion in 6.3.1). This includes conditionals
and complement clauses as well as quasi-conditional clauses, where there is some sort of causal
dependency between the rst clause and the second, such as (464) or (465). In both cases, the
rst clause is an imperative and the second clause is a description of the state of affairs that will
result if the addressee accomplishes the action encouraged by the imperative.
(464) ao-a
light-SG.N
rukue-ri
turn on-2SG
ra
and
aviavi-ve
shine-SUB
Turn on the light and it will shine.
141
(465) sipito,
chief
oira-ra
man-HUM.PL
oara
RPRO.3.PL.M
rutu
very
areii-ri
organize-2SG
ra
and
vearo-vira
good-ADV
kareke-a-ve
appear-3PL
-SUB
Chief, organize all the people and they will look good.
Sentences such as (464) and (465) are very similar to conditionals, in which the protasis
(if-clause) and apodosis (then-clause) are conjoined by ra, as illustrated in (466) and (467).
(466) vosia
if
kakae-to
child-SG.M
gau-pa-re-ve
cry-CONT-3SG.M
-SUB
ra
and
rera
PPRO.3.SG.M
tavi-pa-e-ve
tell-CONT-3SG.F
-SUB
aako-va
mother-SG.F
oisio
COMP
tape
shush
If a boy cries, his mother will tell him to be quiet.
(467) vosia
if
Erava
Balbi
poko-viro
erupt-RES
ra
and
vigei
PPRO.1.PL.INCL
rutu
very
raku-e-ve
cover-3SG.F
-SUB
voo
here
Togarao
Togarao
If Mt. Balbi erupts, it will cover all of us here in Togarao.
There is another conjunction found in Rotokas, uva, which is used exclusively for conjunc-
tion at the sentential level. It is typically found in topic chains, where numerous sentences
sharing a single topic are strung together in sequential order, as exemplied in (468) and (469).
(468) a. riro
big
kaekae-vira
long.RDP-ADV
pau-ra-e
sit-1SG
-IP
I sat down for a long time

b. uva
and
asisoe-ra-i
sore-1SG
-PRES
and Im sore.
(469) a. ragai
PRO.1.SG
vaisi-i-vo
call-3PL
-IP
oiso
COMP
pirati
peanut
kaviru-a-vo
steal-3PL
-IP
Siku
Siku
oira-aro
PRO.3.SG.F-POSS
They called me out as I stole Sikus peanuts.
b. uva
and
ragai
PRO.1.SG
kotu-i-voi
court-3PL
-PRES
oira=pa
PRO.3.SG.F=BEN
and they took me to court for it
c. uva
and
ragai=pa
PRO.1.SG=BEN
roroveara-ro-e
clarify-3SG.M
-IP
Sepiri
Sepiri
and Sepiri straightened things out for me.
142
d. uvare
because
vegei
PRO.1.DL
rutu
very
tou-pa-ve-vorao
be-CONT-1DL-NP
because the two of us were both there.

In both (468) and (469), there happens to be a causal relationship between the clauses con-
joined with uva, but this does not seem to be a necessary condition, judging from sentences
such as (470), where there is no causal relationship between the rst clause and the second one
conjoined with uva (i.e., the fact that a particular individual was in reality beaten up did not
cause the misapprehension that he had been pelted with stones).
(470) vavae
hand
vuvuko=ia
st=IA
rera
PRO.3.SG.M
tatu-re-vo
beat-3SG.M
-IP
uva
and
kea-a-e
think mistakenly-3PL
-IP
oiso
COMP
aveke=ia
stone=LOC
rera
PRO.3.SG.M
tatu-re
beat-3SG.M
He hit him with his sts and they thought mistakenly that he hit him with a stone.
Although sentences conjoined by uva typically share a subject, subject-sharing is not a
strict necessity, as shown by (471), where co-reference occurs between the patient in (a) and the
(notional) possessor in (b), or (472), where the subject of (a) and (b) are distinct individuals.
(471) a. Raki
Raki
aau-pie-re-vo
blinded by light-CAUS-3SG.M
-IP
ravireo
sun
The sun blinded Raki
b. uva
and
osirei-to
eye-SG.M
voosi-ro-e
be blind-3SG.M
-IP
and his eyes are blind.

(472) a. Sipi
Sipi
asige-o-e
sneeze-3SG.F
-IP
Sipi sneezed
b. uva
and
oisio
COMP
pura-o-e
say-3SG.F
-IP
Vitera
Vitera
pauto-vi
God-DIM
virako-pa-re
bless-CONT-3SG.M
and Vitera said bless you

The conjunction uvare because is potentially analyzed morphologically as uva plus the
enclitic =re , perhaps via the goal semantics associated with the enclitic (see Figure 4.3). There
is, however, a risk of engaging in confabulation when reading too much into cases of similarity
in form when a language possesses a phonemic inventory as small as that of Rotokas (especially
when it is known to have arisen by collapsing voicing distinctionscf. 3.1.2), and in this thesis
uvare is cautiously treated as a single unanalyzed form.
143
(473) gae-o-e
run-3SG.F
-IP
revasi-va
blood-SG.F
oira-to
man-SG.M
iava
POST
uvare
because
rera
PPRO.3.SG.M
toe-i-vo
cut-3PL
-IP
The mans blood ran because they cut him.

(474) tarausisi
trousers
ragai=ia
PPRO.1.SG=LOC
garo-pa-ro-e
loose-CONT-3SG.M
-IP
uvare
because
riro-toa
big-SG.M
The trousers were loose on me because theyre big.
Another coordinator that appears to be derived from uva is ovusia while, which is illus-
trated in (475) and (476).
(475) oira-to
man-SG.M
reoreo-pa-ro-e
talk.RDP-CONT-3SG.M
-IP
ovusia
while
viovoko-pa-irara
teenage-DERIV-HUM.PL
rera
PPRO.3.SG.M
gori-pie-pa-i-vo
turn from-CAUS-CONT-3PL
-IP
vuri
bad
reo-ro
word-
raga
only
pura-pa-oro
make-CONT-DEP.SIM
The man talked while the teenagers just talked back with bad talk.
(476) Rorisi
Rorisi
agara-pie-e-voi
be startled-CAUS-3SG.F
-PRES
Kepi
Kepi
ovusia
while
ito-va
banana-SG.F
ou-pa-e
get-CONT-3SG.F
Kepi startled Rorisi while she was getting a banana.

The particle teapi is used to conjoin a clause that describes an undesirable situation of some
sort. Such clauses have been variously labelled in the literature as apprehensional (Dixon,
1977) or timitive (Palmer, 2001:22). In Rotokas, they are associated with the subjunctive
mood (see 5.2.2.7.2).
(477) avuki-vira
secure-ADV
monia
money
tovo-a-vo
put-1SG
-IP
benk=ia
bank=LOC
teapi
lest
va
PPRO.3.SG.N
kaviru-i-ve
steal-3PL
-SUB
I am putting money in the bank so that they dont steal it.
(478) roe-vira
above-ADV
koie
pig
kuvu-ro
CLASS-PL.N
tovo-ri
put-2SG
teapi
lest
kaakau
dog
vara
PPRO.3.PL.N
aio-e-ve
eat-3SG.F
-SUB
Put the pig-lled bamboo containers above so that the dogs dont eat them.
There is another particle, ari, used to conjoin clauses whose meaning is less clear-cut
than the previously-mentioned ones. In most cases, it would be naturally translated as but
in Englishfor example, in (479) and (480).
(479) aite
father
vao
PRO.PROX.SG.N
ou-pa-re
get-CONT-3SG.M
ari
but
vii
PRO.2.SG
eva
-
evoa
there
Dad is getting this one but you (are getting) that one.
(480) Tomas
Tomas
vareo
DEM.PROX.DL.N
ou-pa-re
get-CONT-3SG.M
vuku-arei
book-DL.N
ari
but
ragai
PRO.1.SG
katai-vai
one-INDEF
ou-pa-a
get-CONT-1SG
Thomas is buying these two books but Im going to just get one.
144
Part II
Verb Classes in Rotokas
145
Chapter 7
Verb Classes in Rotokas
In 5.2.2.6.1, it was established that the form of verbal subject agreement and of tense/mood
marking consists of two classes, which were labelled simply and . These labels were chosen
for their neutrality; they do not presuppose any particular analysis of what these two inectional
classes represent. The analysis of these two classes of verbal inection is the central concern of
this thesis and the remaining chapters will examine the issue in greater detail.
In this chapter, the basic problem will be formulated. In 7.1, the formal nature of the
distinction will be more rmly established and a clear set of diagnostics for its recognition
will be provided. In 7.2, the basic problem is stated and a tentative hypothesis concerning its
solution is put forward, which will be rened in later chapters as the facts of the matter are
established.
7.1 Two Verb Classes: and
The distinction between and agreement and TAMmarking imposes a two-way classication
on all verb stems for the purposes of verbal inection. This classication is observable for
individual tokens of a verb root or stem, in the sense that most instances of an independent verb
(as opposed to a dependent verbsee 5.2.2.5) can be unambiguously assigned to one of these
two classes. The morphological diagnostics that can be used to identify a particular token as
or will be discussed in 7.1.1. Although the classication of a particular token is generally
straightforward, there are a few complications and exceptions, which will also be discussed in
7.1.1.
7.1.1 Morphological Diagnostics
Since the primary concern of this part of the thesis is the distinction between two forms of
verbal agreement, it pays to be clear about how that distinction is established. What form does
it take and what conditions are involved? The distinction between and verbs is not simply a
147
property of verbal subject agreement, since it divides into two classes not only the verbal subject
agreement markers but also the tense/mood markers. Each will be discussed in turn.
7.1.1.1 Verbal Subject Agreement
The two classes of verbal subject agreement were already introduced in 5.2.2.6 and are re-
peated below for convenience.
Person Number Gender
1st Person
Singular -ra -a
Dual -ve
Plural Inclusive -vio
Plural Exclusive -io
2nd Person
Singular -u -ri
Dual M -si
F -ere
Plural -ta
3rd Person
Singular M -ro -re
F -o -e
Dual M -si
F -ere
Plural -a -i
Figure 7.1: Subject Agreement Sufxes
As Table 7.1 (cf. Table 5.2) shows, the distinction between and agreement is not found
in all congurations of person, number, and gender. In fact, it is found only in the singular and
the third person plural. For example, verbs with a rst person singular subject can be easily
identied as or , as illustrated by (481) and (482).
(481) ava-ra-i
go-1SG
-PRES
eisi
LOC
uuko-vi
water-DIM
sisiu-sia
bathe-DEP.SEQ
Im going to the river to bathe.
(482) ragai
PRO.1.SG
raga
only
asiko-vira
alone-ADV
tou-pa-a-voi
be-CONT-1SG
-PRES
kepa=ia
house=ENC
I am alone in the house.
If a verb occurs with a subject that is not singular or third personal plural, only TAMmarking
reveals the class of the verb. For example, the form of subject agreement is the same in (483)
and (484) since the subject is third person masculine dual in both cases, but the difference in
classication is nevertheless identiable on the basis of TAM marking: -ei for in the case of
(483) and -voi for in the case of (484).
148
(483) Pita
Pita
vaio
DL.ANIM
ora
and
Kariri
Kariri
ava-si-ei
go-3DL.M-PRES
voka-sia
walk-DEP.SEQ
Peter and Kariri are going for a walk.
(484) vaea-vira
same-ADV
vaiterei=a
PRO.3.DL.M=TOP
keke-pa-si-voi
look-CONT-3DL.M-PRES
kuvupa-toarei
shirt-DL.M
These two shirts look the same.
On the basis of (485) or (486) alone, for example, it would not be possible to determine
which class the verb stem tou be belongs to since tense/mood marking is absent (thanks to
the possibility of zero-marking for the present realis).
(485) Tasia
Tasia
ora
and
Vitera
Vitera
tou-pa-ere
be-CONT-3DL.F
aore-pa-vira
different-DERIV-ADV
Tasia and Vitera are different (i.e., belong to different clans).
(486) vo
here
oisioa
always
tou-pa-io
be-CONT-1PL.EXCL
voari tuariri
long ago
igei
PRO.1.PL.EXCL
aao
PRO.POSS.1.SG
opo
taro
kovo
garden
toki-pa-oro
care for-CONT-DEP.SIM
Long ago we were here caring for our taro gardens.
There are three ways in which subject agreement may be lacking on a verb stem. First,
dependent verbs always lack subject agreement, as illustrated in (487) and (488). The lack
of agreement is one of the two criteria for their identication (the other being the lack of
tense/mood markerssee 5.2.2.5).
(487) Tasia
Tasia
aivaro-sia
meet-DEP.SEQ
ava-o-e
go-3SG.F
-IP
Vitera=va
Vitera=COM
Tasia went to meet with Vitera.
(488) Vaeako
Vaeako
riro-vira
big-ADV
pupuraki-o-i
sweat-3SG.F
-PRES
eisi=va
LOC=ABL
kare-pa-oro
return-CONT-DEP.SIM
sikuru-a
school-SG.N
Vaeako sweats a lot returning home from school.
Second, verbs with neuter subjects normally show null agreement (though see 7.1.1.2 on
the use of third person plural agreement with neuter subjects), as illustrated in (489) and (490).
(489) rerio
radio
vori-a-aro
pay-SG.N-POSS
apepa--voi
expensive-3SG.N-PRES
rutu
very
The price of a radio is very high.
149
(490) kikisi
ball
kukuuku-pa--voi
hit ground-CONT-3SG.N-PRES
The ball is hitting the ground.

The third case is when verbs lack both subject marking and TAM marking, which appears
to be restricted to verbs with a third person singular subject (regardless of gender) in the present
realis, as in (491) and (492).
(491) vovokio
today
kakau
cocoa
vori-pa-to
buy-DERIV-SG.M
urio-pa
come-CONT
varao
DEM.PROX.PL.N
rutu
very
vori-sia
buy-DEP.SEQ
kakau-ara
cocoa-PL.N
vigei
PRO.PER.1.INCL
vara-aro
PPRO.3.SG.N-POSS
Today the cocoa buyer is coming to buy all of our cocoa.
(492) viapau
NEG
oira-to
man-SG.M
uvui-pa
be able-CONT
ra
and
va
PPRO.3.SG.N
iava
POST
kopii-ro
die-3SG.M
vao-ia
DEM.PROX.3.SG.N=LOC
kuva
sorcery
oa
RPRO.3.SG.N
pura-pa-i-veira
make-CONT-3PL
-HAB.ANIM
vo-evao
SPEC-tree
iava
POST
A man wouldnt die from the poison that they would always make from the tree.
[(Firchow, n.d.)]
In cases where subject agreement on the verb is uninformative with respect to the classi-
cation of a verb, the marking of TAM marking is usually more revealing.
7.1.1.2 Tense/Aspect/Mood
Rotokas has a number of verbal sufxes marking various categories of tense, aspect, and mood
(TAM), and these are also sensitive to the distinction between and verbal inection, as
shown in Table 7.1 (see 5.2.2.7 for discussion).
Verb Classication
Tense
Present -ei -voi
Immediate Past -e -vo
Near Distant -era -vora
Distant Past -erao -vorao
Remote Past -epa -va
Table 7.1: Realis Tense Markers By Verb Classication ( vs. )
150
TAM in fact provides a better diagnostic of the distinction between the two classes than
verbal agreement, given that it is found for all TAM categories whereas in verbal subject agree-
ment, the distinction is restricted to a subset of the available categories. This can be seen in
(493) and (494), where verbal subject agreement is uninformative but the classication of the
verb stem is nevertheless identiable on the basis of TAM.
(493) asi
of course
evoa
there
tou-pa-si-voi
be-CONT-3DL.M-PRES
Why of course the two of them are over there!

(494) Tavi
Tavi
vaio
ANIM.DL
ora
and
Rake
Rake
tetevu
sago
tutaa-pa-si-ei
sew-CONT-3DL.M-PRES
Rake
Rake
vo-kepa-aro
SPEC-house-POSS
iare
POST
Tavi and Rake are sago-sewing for Rakes house.
In some cases, there is ambiguity concerning the proper segmentation of the agreement and
TAM sufxes, as illustrated in (495) and (496), where the same verb form is found but arguably
possess different underlying morphemes. The sufx -i is a verbal agreement marker for the
third person plural in (495) and (496).
(495) kokio
bird
kare
FP
eraerao-vira
two.RDP-ADV
tou-pa-i
be-CONT-3PL
evao-va=ia
tree-SG.F=LOC
The birds are on the tree in pairs.
(496) tugitugi-ara
room.RDP-PL.N
tou-pa-i
be-CONT-3PL
kepa=ia
house=LOC
riro-ara
big-PL.N
Many rooms are in the house.
In (497) and (498), however, the analysis of the sufx -i is unclear; it could be analyzed as
a marker of plural subject agreement or the present tense realis marker -ei (which reduces to -i
according to productive morphophonemic rulessee (335)).
(497) Teokon
Teokon
urui
village
oa
RPRO.3.SG.N
tou-pa-i
be-CONT-?
Wakunai=ia
Wakunai=LOC
ruvara=ia
near=LOC
Teokon village is close to Wakunai.
(498) sirovie-vira
striped-ADV
rutu
very
tou-pa-i
be-CONT-?
veeta
bamboo
kou
CLASS
Bamboo is striped.
Since (497) and (498) both have singular neuter subjects, the analysis of -i as a marker of
the present tense realis seems reasonable; however, examples of third person neuter subjects
with plural agreement, such as (499) and (500), suggest otherwise.
151
(499) veveto-vira
sharp-ADV
rutu
very
tou-pa-i-voi
be-CONT-3PL
-PRES
sigo-a
knife-SG.N
The knife is very sharp.
(500) kasirao-vira
hot-ADV
tou-pa-i-voi
be-CONT-3PL
-PRES
uuko
water
rovu
CLASS
The water is really hot.
Further evidence in favor of analyzing the sufx -i in (497) and (498) as an agreement
marker (as opposed to the present tense realis) comes from the agreement patterns observed for
other verbs, such as paru ow, as in (501) and (502).
(501) uuko-vi
water-DIM
oa
RPRO.3.SG.N
arasi-vira
nice-ADV
rutu
very
paru-pa-i-veira
ow-CONT-3PL
-HAB
The water ows very nicely.
(502) viarora
wild pitpit
kou
CLASS
raga
just
tuvu-a
mud-SG.N
tupa-vorao
cover-NP
oa iava
therefore
viapau
NEG
vearo-pie-vira
good-CAUS-ADV
paru-pa-i-veira
ow-CONT-3PL
-HAB
uuko-a
water-SG.N
Wild pitpit covers the mud and therefore the water doesnt ow well.
Further support for the interpretation of the sufx -i as an agreement marker in (495) and
(500) comes from instances of the same verbs with null subject agreement but TAM marking,
as in (503) and (504).
(503) siopai-vira
unfamiliar-ADV
rutu
very
tou-pa--voi
be-CONT-3SG.N-PRES
evo
DEM.???.SG.N
reo-pa-a
talk-DERIV-SG.N
That talk is very unfamiliar.
(504) katokato-vira
black-ADV
paru--voi
ow-3SG.N-PRES
uuko-vi
water-DIM
The river is owing black now.
Finally, not all TAM markers are sensitive to the distinction between and verbs. The
sufx -pa is invariant in form, as can be seen in (505) and (506), where it is invariant in form
despite the fact that it occurs with an verb in (505) and a verb in (506).
(505) Reari
Reari
ira
RPRO.3.SG.M
akoro-a=ia
lime-SG.N=LOC
aasi
betel nut
aio-pa-ro-i
eat-CONT-3SG.M
-PRES
Reari is chewing betel nut with lime.

(506) Rarasori
Robinson
kakapiko-a
small amount-SG.N
aio-a
food-SG.N
aio-pa-re-voi
eat-CONT-3SG.M
-PRES
uva
and
rera=pa
PPRO.3.SG.M=BEN
sirao-pa-ro-e
feel sorry-CONT-3SG.M
-IP
Pita
Peter
Robinson was eating little food and Peter feels sorry for him.
152
7.2 Firchows Problem: What is the / Distinction?
The basic problem that will be addressed here could be called Firchows Problem, given that it
was originally recognized by Firchow (1987), who observes that the analysis of the distinction
between and verb morphology poses a number of analytical challenges due to its imperfect
correlation with transitivity. There are essentially two main issues. First, Firchow (1987:22)
observes that the notion of transitivity is somewhat slippery:
The root of the problem is the notion of transitiveness (which is even unclear in
the analysis of English verbs). What are the parameters of transitiveness? Can the
verb to walk be transitive because there is some goal or direction involved? Why
is to walk considered transitive in Rotokas when the verb to return is never
transitive and a goal or direction is more obviously implied in the latter?
What Firchow (1987) had in mind with this observation is that verbs with no obvious dif-
ference in transitivity are nevertheless classied differently. In other words, if transitivity deter-
mines verb classication, why do verbs with the same transitivity show different classication?
For example, the verb kare return is whereas the voka walk is . Yet neither takes an
object and no goal needs to be made explicit, as can be seen in (507) and (508).
(507) kupero-vira
unaware-ADV
raga
just
voka-pa-a-voi
walk-CONT-1SG
-PRES
I have been walking around unaware.

(508) kare-pa-ra-i
return-CONT-1SG
-PRES
atoi
village
iare
POST
I am going back to the village.
Second, Firchow (1987) also observes that intransitive and transitive verbs alike show un-
expected classication:
The problem is that some verbs such as voka to walk are also inected by the
transitive sets of markers (voka-re-va he walked years ago), while some verbs
such as ruipa to desire (something) are inected by the intransitive sets of mark-
ers (ruipa-ro-epa he desired (it) years ago).
This is illustrated for the two verb stems mentioned by Firchow (1987): voka walk in
(509) and ruipa want, like in (510). The contrast between the classication of the two verbs
is readily observable, given that both occur with rst person singular subjects and in the present
tense realis.
(509) kupero-vira
clueless-ADV
raga
just
voka-pa-a-voi
walk-CONT-1SG
-PRES
I was just walking around clueless.

153
(510) oari=pa
DEM.3.SG.F=BEN
ruipa-pa-ra-i
like-CONT-1SG
-PRES
riako-va
woman-SG.N
I like that woman.
On the basis of these considerations, Firchow (1987) suggests that either transitivity is not
the relevant parameter or it must interact with other (not yet identied) parameters. Although
Firchows basic worry is justied, he fails to dene transitivity as clearly as one would wish in
order to state denitively that it is not the relevant distinction governing verbal classication,
nor does he identify other parameters that could potentially do so. In the following section, the
notion of transitivity will be explored in greater depth, contrasting it with the related notion of
valency, in order to provide a full account of Rotokas verb classication.
7.3 Transitivity and Valency
This section distinguishes between transitivity and valency in order to set the stage for the
in-depth discussion of valency and valency-changing derivations provided in Chapter 8 and
Chapter 9, respectively.
7.3.1 Transitivity
The notion of transitivity held by Firchow (1987) is somewhat rough-and-ready, and there has
been considerable work done in rening the notion cross-linguistically (Lakoff, 1977; Hopper
and Thompson, 1980; Giv on, 1984; Kittil a, 2002; Lazard, 2003). This literature helps explain
why a verb such as ruipa to want deviates from the transitive pattern, but there is still a good
deal in need of explanation. In this section, the notion of valency will be pinned down more
precisely and integrated into the typology of argument types.
Before discussing valency, it is worthwhile to draw a distinction between transitivity and
valency. The term transitivity is used ambiguously in the literature. On the one hand,
transitivity refers to a syntactic notion, usually the number of (core) arguments taken by a verb.
According to this sense of the term, it is more or less synonymous with the term valency.
On the other hand, transitivity refers to a more general semantic notion, which has to do with
the extent to which an action carries over from agent to patient (Hopper and Thompson, 1980;
Frawley, 1992), in which case it is a gradient notion, inuenced by a number of different factors,
such as those listed in Table 7.2.
The majority of the transitivity features discussed in Hopper and Thompson (1980) are rel-
atively self-explanatory and do not require additional discussion, but a few merit elaboration
namely, agency and the affectedness and individuation of O.
The parameter of agency refers to the nature of the agent that initiates an action. Although
Hopper and Thompson (1980) do not clarify what is meant by A being high or low in potency,
154
Parameter High Low
A participants two or more one
B kinesis action non-action
C aspect telic atelic
D punctuality punctual non-punctual
E volitionality volitional non-volitional
F afrmation afrmative negative
G mode realis irrealis
H agency A high in potency A low in potency
I affectedness of O O totally affected O not affected
J individuation of O O highly individuated O non-individuated
Figure 7.2: Semantic Parameters of Transitivity (Hopper and Thompson, 1980:252)
it appears from their discussion of the parameter that animacy is the main dimension and that a
human or animate A is considered higher in potency than an inanimate one.
The parameter of Affectedness of O refers to the extent to which O is changed as a result
of the situation described in a clause. While some transitive verbs entails a readily identiable
change of state in the patient/theme (e.g., break), others do not (e.g., hit). For example, in En-
glish, this distinction has been invoked to account for which verbs participate in the conative
alternation or form middles (Fillmore, 1970; Levin and Hovav, 1995b). Verbs that involve a
change of state can form middles whereas predicates that involve only a causually affected O
do not, as illustrated in (511).
(511) a. * The table hits easily.
b. The table breaks easily.
Conversely, verbs that involve a causally affected O participate in the conative alternation,
whereas verbs that do not involve a causally affected O or that entail a change of state in O do
not, as illustrated in (512).
(512) a. The judge hit/hit at the table with his gavel.
b. * The judge broke/broke at the table with his gavel.
According to Hopper and Thompson (1980), the parameter of Individuation of O refers
to the distinctness of O from A and from its own background. The specic contrast to which it
refers are listed below in Table 7.2.
155
More Individuated Less Individuated
proper common
human, animate inanimate
concrete abstract
singular plural
count mass
referential, denite non-referential, indenite
Table 7.2: Individuation of O: Relevant Features
For example, in Tongan, non-referential objects undergo noun incorporation and the sub-
ject takes absolutive rather than ergative agreement, as illustrated in (513).
(513) a. nae
PAST
kai
eat
e
ERG
Sion
John
a
ABS
e
DEF
ika
sh
John ate the sh.
b. nae
PAST
kai
eat
ika
sh
a
ABS
Sione
John
John ate sh. [Hopper and Thompson (1980:257-258)]
As Hopper and Thompson (1980) observe, a prototypical transitive situation will have high
transitivity values for most, if not all, of the parameters identied in Table 7.2. In other words,
these parameters cluster to dene a prototypical transitive situation (Lakoff, 1977; Giv on, 1984;
Kittil a, 2002; Lazard, 2003), and a transitive clause is a simple underived clause that describes
such a situation (Nss, 2006):
a transitive situation is one in which an agent acts upon a patient, where the agent is
volitionally involved in the event, causes or instigates the event, and is not affected
by the event; while the patient is not volitionally involved, does not participate in
the instigation of the event, but is affected by it.
There is some controversy concerning the nature of the prototypical transitive clause which
hinges upon what is taken to be the prototypical object (Nss, 2006). (This is an issue that will
be discussed again later, in Chapter 11.) We turn now to valency.
7.3.2 Valency
In the previous section, transitivity was discussed and established as a semantic notion that
concerns the degree to which an action carries over from agent to patient. Here it is distin-
guished from valency, which is a strictly syntactic notion (Tesni` ere, 1959; Somers, 1987; Mosel,
1991; Payne, 1997). Mosel (1991:241) characterizes valency in the following terms:
156
Valency is the property of the verb which determines the obligatory and optional
number of its participants, their morphosyntactic form, their semantic class mem-
bership (e.g., animate, human), and their semantic role (e.g., agent, patient,
recipient). The valency inherently gives information on the nature of the semantic
and syntactic relations that hold between the verb and its participants.
Valency is an essentially verb-centered notion since it is primarily the verb that determines
the number of arguments present in a clause. The number of possible arguments taken by a
verb is stated to be the verbs valency, possible values ranging from zero to three (avalent=0,
monovalent=1, bivalent=2, and trivalent=3). Unlike core arguments, the number of circum-
stantials is unlimited, ranging from zero to n. Examples of sentences with varying numbers of
circumstantials are provided in (514) through (516).
(514) Rave,
Rave,
vii
PRO.2.SG
ori-pa-u-ei
cook-CONT-2SG
-PRES
oira-ra=pa
man-HUM.PL=BEN
ovusia
while
vii=pa
PRO.2.SG=BEN
kovo-i-ve
work-3PL
-SUB
Rave, you cook for the men while they work for you.
(515) ragai
PRO.1.SG
sipuru=ia
spoon=LOC
aio
food
toke-pa-ra-i
serve-CONT-1SG
-PRES
kakae
child
vure=pa
FP=BEN
I serve food to the children with a spoon.
(516) toisikova=ia
mountainous area=LOC
ava-pa-a-veira
go-CONT-3PL
-HAB
raiva=ia
road-LOC
eisi-re
LOC=ALL
Asitaipa
Aistaipa
They went on the road to Asitaipa.
Crucial to the notion of valency is the distinction between core and non-core arguments on
the one hand, and between arguments and adjuncts on the other.
1
The distinction between a
core argument and a non-core argument is recognized in most grammatical theories, although
its instantiation may differ according to the dictates of a particular framework. Dixon (1994:6)
claims that the distinction between verbs with one core argument and verbs with two core argu-
ments is fundamental and universal:
All languages distinguish between clauses that involve a verb and one core noun
phrase (intransitive clauses [monovalent]) and those that involve a verb and two or
more core NPs (transitive clauses [bivalent], including ditransitive as a subtype).
1
Although some authors use alternative terminology (e.g., actant vs. circumstantial), the basic concept remains
largely the same.
157
On the basis of the distinction between verbs with one or two core arguments, Dixon (1979,
1994) argues for a three-way division of core arguments into S, the single core argument of an
intransitive clause; A, the core argument of a transitive clause that it is prototypically associated
with the agent; and O, the core argument of a transitive clause that it is prototypically associated
with the patient (see Andrews (2007) for discussion and justication).
2
In order to discuss valency sensibly, it is necessary to establish a core set of basic argument
types. The main source for this discussion is Andrews (2007), which is summarized in Figure
7.3.
Grammatical Functions
Internal External
Core Oblique Free Bound
S A O Argument Adjunct
Table 7.3: Inventory of Grammatical Functions of the NP
The rst major division in his classication is between internal (inner) and external (outer)
functions. This distinction is recognized in some way by most theories of grammarfor ex-
ample, in Role and Reference grammar, there is a similar distinction made between core and
periphery (Van Valin Jr. and LaPolla, 1997; Van Valin Jr., 2005). Within internal functions,
core and oblique functions are distinguished. The core functions are further broken down into
S, A, and O, which are dened, respectively, as the single argument of an intransitive verb, the
argument of a transitive verb prototypically associated with the agent, and the argument of a
transitive verb prototypically associated with the patient.
S The single core argument of a one-place predicatee.g., The giant is sleeping.
A The core argument associated with the actor/agent of a prototypical transitive predicate
e.g., The enraged drunk killed the innocent man.
O The core argument associated with the undergoer/patient/theme of a prototypical transitive
predicatee.g., The plumber smashed the PVC pipe with a monkey wrench.
Oblique Adjunct A non-core argument licensed by general semanticse.g., Geeks program
computers for the fun of it.
Oblique Argument A non-core argument licensed by the predicatee.g., France supplied
Iraq with missiles.
2
The universality of S, A, and O has, however, been called into question on various grounds (Durie, 1988; Bhat,
1991; Dryer, 1997; Mithun, 1999) but in Rotokas there is good evidence for the existence of these distinctions, as
will be seen in Chapter 8. The universality of S is addressed in Chapter 10.3.
158
Using these grammatical primitives, a more precise statement of the relationship between
grammatical roles and the two forms of verbal agreement can be formulated and evaluated on
the basis of the evidence. In the following section, a preliminary hypothesis is put forward for
evaluation.
7.4 First Hypothesis
Firchows observed correlation between verb classication and transitivity can be reformulated
in terms of Dixons three primitive core argument types. The hypothesis would be that sub-
ject agreement is simply sensitive to the type of subject: the S of an intransitive verb takes
agreement whereas the A of a transitive verb takes agreement, as in (517).
(517) a. S
b. A
According to this hypothesis, there would be a one-to-one relationship between the primi-
tive grammatical roles of Dixon (1979, 1994) and the classication imposed by the distinction
between and verbs. This is of course only one of a number of logically possible mappings
between the two, which are shown diagramatically in Figure 7.4.
Possible Conguration Role Inection
One-to-One S
A
Split-S S
A
Split-A S
A
Many-to-Many S
A
Table 7.4: Grammatical Roles and Verb Inection
The organization of the second part of this thesis is as follows: In the following chapter, the
nature of valency in Rotokas will be examined in depth, and it will be shown that the simple
hypothesis in (517)which posits a one-to-one relationship between grammatical roles and the
two forms of agreementcannot be maintained since verbs with a single core argument (S) are
split between the two forms of agreement. Although the majority of verbs with a single core
argument take agreement, there is a sizeable minority of verbs with a single core argument
that take agreement. This eliminates the one-to-one and split-A mappings, leaving only the
split-S and many-to-many mappings as viable hypotheses.
159
160
Chapter 8
Valency in Rotokas
This chapter examines the nature of valency in Rotokas in an attempt to evaluate the hypothesis
that verbal inection in Rotokas is sensitive simply to the grammatical role of the subject, such
that S shows agreement while A shows agreement. Underived verb roots represent the
default mappings of semantic roles and grammatical relations in the argument structure of the
language and therefore are a natural starting point for investigation. Here it is established that
there are two main valency types in Rotokas: monovalent verb roots (intransitive), which
take a single argument, and bivalent verbs roots (transitive), which take two (or possibly
three) core arguments. If a clause possesses two core arguments, it will show agreement;
however, the reverse does not hold true. If a verb shows agreement, it will not necessarilly
take two core arguments. This asymmetry owes to the fact that monovalent verb roots are split
according to their form of verbal inection: most show inection but some show .
Crucial to the concept of valency is the distinction between core and oblique arguments
(see 7.3.2). In Rotokas, core arguments can be distinguished from oblique arguments on the
basis of a few different considerations. First, verbal agreement for person, number, and gen-
der is sensitive to (i.e., controlled by) the subject. The core argument that plays the role of
subject determines the choice of agreement marking on the verb and the presence of a second
core argument (a direct object) automatically triggers agreement. Second, core arguments
are relatively more restricted in their constituent ordering than other types of arguments or ad-
juncts (e.g., adverbs) (see 6.2.1). Third, core arguments are necessarily present either by way
of verbal agreement in the case of subjects or by way of realization as a nominal (a pronoun
or a lexical NP) in the case of direct objects. Finally, core arguments are unmarked (i.e., occur
as bare NPs) whereas non-core arguments take some form of oblique marking. As Andrews
(2007:153) observes, Languages in which the core/oblique distinction corresponds to that be-
tween bare NPs and those carrying a marker are not uncommon.
The term predicate type is used here to describe the number of subcategorized arguments
taken by a verb, which may exceed the valency of a verb since valency includes only core
arguments while some verbs are subcategorized for oblique argumentsi.e., some two-place
predicates are considered monovalent here. For example, the verb stem tara search for, seek
161
requires (i.e., is subcategorized for) two arguments, but it is not bivalent, since one of its argu-
ments is an oblique, which is marked by the role-marking enclitic =re, as in (518) and (519).
(518) Agiosi
Agiosi
aakova=re
mother=ALL
tara-pa-e-vo
look for-CONT-3SG.F
-IP
Agiosi looked for (her) mother.

(519) oira-ra
man-PL.N
ava-pa-a-i
go-CONT-3PL
-PRES
varu
meat
kare-vai=re
FP-INDEF=ALL
tara-sia
look for-DEP.SEQ
eisi
LOC
vegoaro
jungle
The men are going to look for game in the jungle.
Given the distinction between predicate type, valency, and verbal inection ( vs. ), six
different verb root classes can be distinguished, as shown in Table 8.1.
Predicate Type Valency Agreement Example
1-Place 1 uusi sleep
1-Place 1 gau cry
2-Place 1 ruipa want
2-Place 1 tara look for
2-Place 2 upo strike
3-Place 2/3 vate give
Table 8.1: Predicate Types, Valency, and Subject Agreement in Rotokas
An extensive listing of verb roots in Rotokas is provided in Appendix A, which classies
all of the known verb roots in the Rotokas lexicon. This data comes from a lexical database
of Rotokas under development by the author which contains a wide variety of information,
including the valency, number and type of non-core arguments, and the form of agreement for
verb roots and stems. It is based on a Shoebox dictionary developed by Irwin Firchow (Firchow,
1973, 1984) and substantially rened during the course of my own eldwork, on the basis of
native speaker consultation and the analysis of interlinearized texts.
The relative proportion of verb roots according to their valency type and the overall number
of versus verb roots are provided as a bar graph in Figure 8.1 (see Appendix A for more
information concerning the Shoebox/Toolbox dictionary from which this data was obtained as
well as an exhaustive list by class of verb roots and verb stems found in the dictionary).
162
1
A
1
B
1
O
b
l
A
1
O
b
l
B
2
B
0
100
200
300
400
500
600
Monovalent (Alpha)
Monovalent (Beta)
Monovalent, Oblique (Alpha)
Monovalent, Oblique (Beta)
Bivalent (Beta)
385
66
55
35
487
Alpha Beta
0
100
200
300
400
500
600
440
588
Figure 8.1: Distribution of Valency Classes (left) and Inection Classes (right) in Rotokas Lexicon
8.1 Background
In Rotokas, verb roots can be broadly grouped into two main valency types: monovalent roots,
which take a single core argument (and possibly a second oblique argument marked by one
of the case-marking enclitics described in 8.3.3), and bivalent roots, which take two core ar-
guments. The difference between the two valency types concerns objecthood. While both
monovalent and bivalent verb roots require a subject, with which they agree in terms of person,
number, and gender, only bivalent verb roots take an additional core argument, a direct object,
which occurs in a xed position (see 6.2.1) and cannot be freely elided.
For example, the verb root uusi sleep is monovalent and takes only a single core argument,
as in (520). It cannot take an object, except through valency-changing derivations, as illustrated
in (521).
(520) Avaisisi
Avaisisi
ira
RPRO.3.SG.M
kei
leaning
kepa=ia
house=LOC
uusi-pa-ro-veira
sleep-CONT-3SG.M
-HAB
vegoaro
jungle
Avaisisi is sleeping in a lean-to in the jungle.
(521) aako-va
mother-SG.F
kakae-to
child-SG.M
uusi-pie-e-voi
sleep-CAUS-3SG.F
-PRES
evao
tree
ruvaru-va
relief-SG.F
rero-aro
underneath-POSS
oiso
COMP
ra
and
kovo-e-ve
work-3SG.F
-SUB
The mother put her son to sleep underneath the tree so that she could work (in the
garden).
163
By contrast, the verb root tario chase is bivalent. It takes two arguments: a subject and an
object. The verb agrees in person, number, and gender with the subject but not with the object.
The subject can be elided when contextually inferrable and/or non-emphatic, as in (523), but the
object is obligatory and cannot be freely elided. It occurs in a relatively xed preverbal position
(see 6.2.1), either as a noun phrase, as in (522), or as a pronoun, as in (523).
(522) Pita
Peter
gapu-to
naked-SG.M
oira-to
man-SG.M
tario-re-voi
chase-3SG.M
-PRES
Peter is chasing the naked man.

(523) sora-to
poisoner-SG.M
isisio
grass
kou
CLASS
puri-oro
lay down-DEP.SIM
tori-re-vo
ee-3SG.M
-IP
uvare
because
rera
PRO.3.SG.M
tario-i-vo
chase-3PL
-IP
eisi
LOC
Rarova
Rarova
The poison man ed, making the grass lie down, because they chased him in Rarova.
Some verb roots are compatible with more than one syntactic frame (i.e., valency or sub-
categorization frame). For example, the verb stem reoreo talk (the reduplicated form of
reo) occurs with a varying number of arguments. In (524), it occurs with only a single core
argumentnamely, the subject.
(524) Alice
Alice
gae-o-ra
be startled-3SG.F
-NP
uva
and
viapau
NEG
reoreo-pa-o-ra
talk-CONT-3SG.F
-NP
Alice was startled and couldnt talk.

In (525) through (527), the verb stem reoreo occurs with an additional argument, a non-core
(i.e., oblique) argument, but the presence of this additional argument has no effect on the form
of agreement. It shows agreement even when it occurs with an addressee marked by =re
in (525) or with an interlocutor marked by =va in (526). The same is true when the topic of
conversation is marked by =ia in (527).
1
(525) teapi
PROH
ragai=va
PPRO.1.SG=ABL
voroko-pa-u
arrogant-CONT-2SG
osia
as
Addressee
..
vii=re
PPRO.2.SG=ALL
reoreo-pa-ra
talk-CONT-1SG
Dont be arrogant with me as Im talking to you.

(526) viapau
NEG
reoreo-o-e
talk-3SG.F
-IP
Interlocutor
..
igei=va
PRO.1.PL.EXCL=COM
uvare
because
oira
PPRO.3.SG.F
tavi-i-vo
tell-3PL
-IP
She didnt talk with us because they told her (not to).
1
It is questionable whether interlocutor is the best characterization for the argument marked by =va in (526).
A better gloss may be conversational partner. This raises the issue of the number and nature of thematic roles,
which is addressed in 10.2.
164
(527) viapau
NEG
rorio-ra-e
be clear-1SG
-IP
ovusia
while
Topic
..
va=ia
PPRO.3.SG.N=LOC
reoreo-pa-ro-e
talk-CONT-3SG.M
-IP
uva
and
rera
PPRO.3.SG.M
ake-a-vo
ask-1SG
-IP
I was unclear about it while he was talking about it and I asked him.
Neither the number of non-core arguments nor the form of oblique marking in (524) through
(527) has an effect on the form of verbal inection. This can be further illustrated with the
verb root tavi tell, which is labile (see 9.1.1). It takes a single core argument and shows
agreement in (528).
(528) Potaki=va
Potaki=COM
kuara-pa-i-vo
yell.at-CONT-3PL
-IP
ovusia
while
tavi-pa-ro-e
tell-CONT-3SG.M
-IP
They are yelling at Potaki while he talks.

In (529) through (531), however, tavi shows agreement when it occurs with an object, as
in (529); with an object and a dependent verb phrase, as in (530); and an object and complement
clause, as in (531) and (532).
2
(529) vavae-ara
hand-PL.N
itoro-pie-i-vo
raise-CAUS-3PL
-IP
kakae
child
vure
FFP
uvare
because
Addressee
..
voea
PPRO.3.PL.M
tavi-e-vo
tell-3SG.F
-IP
tisa-va
teacher-SG.F
The children raised their hands because the teacher told them to.
(530) Pita
Peter
Addressee
..
Jon
John
tavi-pa-re-va
tell-CONT-3SG.M
-RP
kokotoa
leg
rupu-pie-sia
be.submerged-CAUS-DEP.SEQ
Peter told John to stick his leg in the water.
(531) tisa-to
teacher-SG.M
Addressee
..
kakae vure
child FFP
tavi-pa-re-va
tell-CONT-3SG.M
-IP
oisio
COMP
opeita
PROH
taku-vira
bend over-ADV
pau-pa-ta
sit-CONT-2PL
ovusia
while
reoreo-pa-ra
talk.RDP-CONT-1SG
The teacher told the children not to sit bent over while hes talking.
2
The complement clause is marked by the complementizer oisio in (531) and by oisio ra in (532).
165
(532) sipito
chief
Addressee
..
oira-ra
man-HUM.PL
tavi-re-vo
tell-3SG.M
-IP
oisio ra
COMP
ava-a-ve
go-3PL
-SUB
vuruko-a
log-SG.N
taroro-sia
pry out-DEP.SEQ
The chief told people that they should go pry out the logs.
Some verbs select very specic subcategorization frames that are not found among other
verb roots. For example, the verb root kea mistake for, think mistakenly selects two oblique
arguments: the thing mistaken for something else, marked by the benefactive enclitic =pa (see
4.3 for discussion of its semantics) and the thing it is mistaken for, marked by oisio (which
also functions as a complementizersee 6.3.1). The verb is illustrated with both of its oblique
arguments in (533) and with only the thing mistaken for something else in (534).
(533) votoue-toa=pa
ant nest-SG.M=BEN
kea-ra-e
mistake-1SG
-IP
evao
tree
rao=ia
branch=LOC
oisio
COMP
koora-to
possum-SG.M
ira
RPRO.3.SG.M
pau-pa
sit-CONT
evao
tree
rao=ia
branch=LOC
I mistook the ant nest on that tree for a possum sitting on a branch.
(534) riako-va
woman-SG.F
aveke-va
stone-SG.F
peka-e-vo
turn over-3SG.F
-IP
uva
and
rakoru
snake
keke-e-vo
see-3SG.M
-IP
uva
and
kea-o-e
mistake-3SG.F
-IP
oisio
COMP
uo-va
eel-SG.F
The woman turned over the stone and saw a snake but mistakenly thought it an eel.
The examples in (524) through (532) underscore a number of important points regarding
the nature of verbal inection in Rotokas. First, the classication of a verb as or as is
not determined exclusively by the verb root. There is an interaction between the syntactic
construction (subcategorization frame) in which a verb appears and its form of verbal inec-
tion. Second, some syntactic constructions are consistently associated with a particular form of
verbal inection. For example, if a verb takes a direct object, its inectional form is entirely
predictableviz., it will be . This relationship is, however, unidirectional, since the reverse
does not hold truei.e., if a verb shows inection, it may not possess a direct object, as we
will see in the following section when we consider the behavior of monovalent verbs.
8.2 Monovalent One-Place Predicates
A monovalent verb root is one that takes only a single core argumentnamely, the subject, with
which the verb agrees in terms of person, number, and gender. The subject is normally realized
as either a full NP or it is elided when contextually retrievable. For example, the verb root uusi
sleep takes a single core argument, which takes the form of a full NP in (535) but is elided in
(536).
166
(535) atuu
ying fox
koto-vira
hang-ADV
uusi-pa-o-i
sleep-CONT-3SG.F
-PRES
The ying fox sleeps hanging.

(536) kakae-to
child-SG.M
karavuru-ro-e
get dusty-3SG.M
-IP
uvare
because
poupou=ia
dust=LOC
uusi-pa-ro-e
sleep-CONT-3SG.M
-IP
visiko-pa-oro
play-CONT-DEP.SIM
The child got dusty because he slept in dust while playing.
8.2.1 Agreement
Monovalent verbs can be divided into two classes on the basis of their form of agreement:
or . The majority of monovalent verb roots belong to . In Table 8.2, a partial list of
monovalent verb stems is providedsee Appendix A for a complete listing.
Stem Gloss Notes
ava go
era sing
gapu be naked
goagoara be boiling inherently reduplicated
kokoro crazy, foolish
ogoe be hungry
opesi end, nish
revasi bleed
riro grow up
upia in pain, sick
urio come
uusi sleep
uvagi be deaf
vearo be good, ne, well
vioro ripen, mature literally: be green
voosi blind
vuri be bad, spoiled, wrong
Table 8.2: Some Monovalent Verb Roots that Show Inection
All of the verbs in Table 8.2 show the same pattern of agreement, as illustrated for the verb
root uusi sleep in (537) and (538).
167
(537) upia-pa-ra-i
hurt-CONT-1SG
-PRES
kukue
head
iava
POST
oa iava
therefore
uusi-pa-ra-i
sleep-CONT-1SG
-PRES
uru-a=ia
bed-SG.N=LOC
My head hurts and thats why Im sleeping in bed.
(538) uva
so
uusi-ro-epa
sleep-3SG.M
-RP
ovi-toa
offspring-SG.M
tapo
also
urua=ia
bed=LOC
So he slept with his son in bed. [Firchow and Akoitai (1974:50)]
Although the majority of monovalent verbs show agreement, there is also a class of mono-
valent verbs that show agreement. These verb stems are fewer in number than the mono-
valent verbs (see Figure 8.1). A few of these are listed in Table 8.4 (see Appendix A for a
complete inventory).
Stem Gloss Notes
aata swim generic
gau cry
ikau run, speed
kapere swim on the waters surface
opoko defecate generic term used for humans
puu fart
roko go inside
viviko urinate
voka walk
vusi rush out, erupt
Table 8.3: Some Monovalent Verb Roots with Subject Agreement
The verbs in Table 8.4 show the same form of agreementnamely, , as illustrated for the
verb root gau cry in (539) and (540).
(539) kuuo
owl
iria
PPRO.REL.3.SG.
gau-pa-e-veira
cry-CONT-3SG.F
-HAB
vokiaro
night
The owl, he cries at night.
(540) avi
light
ua=ia
CLASS=LOC
kokai
chicken
kare
PL
gau-i-vo
cry-3PL
-IP
In the morning the roosters cried out.

Monovalent verb roots cannot take a direct object without recourse to derivational morphol-
ogy. This is true for those that show agreement as well as those that show agreement.
168
For example, the monovalent verb roots uusi sleep and gau cry can only take a direct ob-
ject using the morphological causative -pie (see 9.1.2 for more detailed discussion). This is
illustrated for uusi sleep in (541) and for gau cry in (542).
(541) Rua
Rua
sikeo
infant
uusi-pie-pa-e-voi
sleep-CAUS-CONT-3SG.F
-PRES
Rua put the child to sleep.

(542) kakae-to
child-SG.M
oaa
PPRO.1.SG
gau-pie-pa-ri-veira
cry-CAUS-CONT-2SG
-HAB
rutu
very
rera=va
PRO.3.SG.M=COM
ugaa-pa-oro
kiss-CONT-DEP.SIM
You make our child cry by kissing him. [Firchow and Akoitai (1974:80)]
Monovalent verbs show the same possibilities of constituent ordering regardless of their clas-
sication as or . The subject of a monovalent verb occurs either before the verb, as in
(543a), or after the verb, as in (543b).
(543) a. oira-to
man-SG.M
uusi-ro-epa
sleep-3SG.M
-RP
The man went to sleep.

b. uusi-ro-epa
sleep-3SG.M
-RP
oira-to
man-SG.M
The man went to sleep.
Similarly, the subject of a monovalent verb occurs either before the verb, as in (544a), or
after the verb, as in (544b).
(544) a. Patiriki
Patrick
gau-pa-re-voi
cry-CONT-3SG.M
-PRES
Patrick is crying.
b. gau-pa-re-voi
cry-CONT-3SG.M
-PRES
Patiriki
Patrick
Patrick is crying.
169
8.3 Monovalent Two-Place Predicates
Monovalent two-place predicates are verb roots whose meaning involves two participants (i.e.,
have two actants in their logical structure) but take only a single core argument. The two
participants of these verb roots are realized as a subject and as an oblique argument marked by
one of the role-marking postpositional enclitics described in 4.2.7. A given verb root selects
for a particular postposition, and the choice of postposition is not fully predictable (see 8.3.3).
For example, the verb root tara seek, nd, search for, look for selects for =re, as illustrated in
(545), and its oblique argument cannot be marked by another enclitic, such as =ia, =va, or =pa,
as illustrated by the ungrammaticality of (546a) through (546c). (It is possible for the oblique
marking to be absent in the case of noun incorporationsee 9.2.2.)
(545) Patriki
Patrick
sigo-a=re
knife-SG.N=ALL
tara-pa-re-vo
look for-CONT-3SG.M
-IP
Patrick looked for (his) knife.

(546) a. * Patriki sigo-a=ia tara-pa-re-vo
b. * Patriki sigo-a=va tara-pa-re-vo
c. * Patriki
Patrick
sigo-a=pa
knife-SG.N=COM
tara-pa-re-vo
look for-CONT-3SG.M
-IP
Patrick looked for (his) knife.

A few roots permit more than one type of marking for their oblique arguments (as already
seen for reoreo talk in 8.1). For example, Firchow (1984) furnishes two possible forms of
oblique marking for the verb root tagava salute, either the postpositional enclitic =re or =va,
as in (547).
(547) a. Kukurai
Kukurai
keapi=va
kiap=COM
tagava-re-voi
salute-3SG.M
-PRES
Kukurai salutes the kiap.

3
b. Kukurai
Kukurai
keapi=re
kiap=ALL
tagava-re-voi
salute-3SG.M
-PRES
Kukurai salutes the kiap.

In some cases, it is not clear whether an oblique constituent associated with a verb represents
a subcategorized argument, rather than an adjunct. For example, the verb root voki get dark
normally takes only a single argument, as in (548), but it also occurs with a second argument,
as in (549).
3
The word keapi is a borrowing into Rotokas from Tok Pisin, where the word kiap refers to the patrol of-
cers who served as travelling police ofcers during the period when Papua New Guinea was under Australian
administration (Sinclair, 1981; Kituai, 1998).
170
(548) Rasii
Rasii
roro-pa-va
light-DERIV-SG.F
ruku-e-voi
light-3SG.F
-PRES
uvare
because
voki-ei
night-PRES
Rasii lit the lamp because it was getting dark.

(549) uva
and
voki-epa
night-IP
vaiterei=re
PRO.3.DL=ALL
The night fell on the two of them. [Firchow (1984)]
8.3.1 Agreement
Monovalent two-place predicates are not uniform with respect to verbal inection. Although
most of these verbs show inection, there are also quite a few that show inection.
Inection Verb Stem Oblique Marking
vari threaten =re
roroo suckle on =ia
kaureo disagree with =va
taea deceive =pa
tara seek, nd, look for, search for =re
oruo diligent about =ia
veku bark at =va
vato respect, honor =pa
Table 8.4: Illustrative Monovalent Two-Place Predicates with Inection
8.3.1.1 -Agreement
Monovalent two-place predicate verbs that show agreement are exemplied in (550) through
(553). Each verb stem that takes an oblique argument selects for a specic type of postposition,
and all four enclitics are attested: =re in (550), =ia in (551), =va in (552), and =pa in (553).
(550) ragai=re
PPRO.1.SG=ALL
vari-ro-i
threaten-3SG.M
-PRES
torara=ia
axe=LOC
He threatens me with an axe.
(551) kakae-to
child-SG.M
aakova=ia
mother-SG.F=LOC
roroo-pa-ro-i
suckle-CONT-3SG.M
-PRES
The child is suckling on his mother.

(552) riro-vira
big-ADV
oisoa
always
rera=va
PPRO.3.SG.M=ABL
kaureo-pa-a-ve
disagree-CONT-3PL
-SUB
They were always in much disagreement with him.
171
(553) voea=pa
PPRO.3.PL=BEN
taea-ro-epa
accuse-3SG.M
-RP
He accused them.
8.3.1.2 -Agreement
Monovalent two-place predicate verbs that show agreement are exemplied in (554) through
(557). Each verb stem that takes an oblique argument selects for a specic case-marking en-
clitic, and all four enclitics are attested: =re in (554), =ia in (555), =va in (556), and =pa in
(557).
(554) Agiosi
Agiosi
aako-va=re
mother-SG.F=ALL
tara-pa-e-vo
seek-CONT-3SG.F
-IP
Agiosi looked for mother.

(555) rera
RPRO.3.SG.M
vo-kovo-aro=ia
SPEC-work-POSS=LOC
oruo-pa-re
be satised-CONT-3SG.M
He is satised with his work.

(556) kakau
dog
kare
FFP
ragai=va
PPRO.1.SG=COM
veku-i-vo
bark-3PL
-IP
eisi
LOC
Sikoriara
Sikoriara
Dogs bark at me in Sikoriara.
(557) ragai
RPRO.1.SG
vato-pa-a-veira
respect-1SG
-HAB
ragai
RPRO.1.SG
taataa-irara-aro=pa
brother-HUM.PL-POSS=BEN
I always respect my brothers.
The oblique arguments of monovalent two-place predicate verbs are fairly free with respect to
constituent ordering (regardless of their form of verbal inection), and in this respect they differ
signicantly from direct objects, whose constituent order is xed. (Direct objects can be ex-
tracted from their xed position in the clause, but only through specic grammatical devices
see 6.2.2.)
The oblique arguments of verb roots are fairly free with respect to constituent ordering,
occuring in a wide variety of positions, although an immediately preverbal position is the most
common. For example, the oblique argument of kasipu angry occurs before the verb in (558)
and after the verb in (559).
172
(558) Raratuiri
Raratuiri
OBL
..
oirara=re
people=ALL
V
..
kasipu-ro-erao
angry-3SG.M
-HAB
uvare
because
aue-ro
DEM.???-PL.CL
poko-pie-pa-i-veira
explode-CAUS-CONT-3PL
-HAB
Raratuiri is angry at everyone because they blew things up.
(559) riako-va
woman-SG.F
ora-kaa-o-pa
RR-hang-3SG.F
-RP
V
..
kasipu-pa-oro
angry-CONT-DEP.SIM
OBL
..
vatua-toa=re
husband-SG.M=ALL
The woman hanged herself while she was mad at her husband.
The oblique arguments of verb stems are also fairly free with respect to constituent order-
ing, occuring in a wide variety of positions, although an immediately preverbal position is the
most common. For example, the oblique argument of koroto meet occurs before the verb in
(560) and after the verb in (561).
(560) ragai
PPRO.1.SG
ava-pa-ra-i
go-CONT-1SG
-PRES
OBL
..
Pita=re
Peter=ALL
V
..
koroto-sia
meet-DEP.SEQ
eisi
LOC
raivaro
road
Im going to meet Peter on the road.
(561) vii
2.SG
V
..
koroto-pa-ri
meet with-CONT-2SG
OBL
..
Sera=re
Sera=ALL
Youre going to meet with Sera.
8.3.3 Oblique Marking
Although verb roots that select an oblique argument are mixed with respect to their form of
agreement (i.e., some show agreement while others show ), the form of agreement does not
appear to be predictable simply on the basis of the form of oblique marking. In other words,
verb roots with agreement co-occur with the same set of oblique markers as verb roots with
agreement. In the following sections, each of the four postpositional enclitics that verb roots
select for will be examined and shown to be completely orthogonal to verb root agreement
classication, as illustrated in Table 8.5.
Form of Oblique Marking
Verbal Agreement =ia =va =re =pa
Class
Class
Table 8.5: Verbal Agreement and Oblique Marking in Monovalent Verb Roots
173
The choice of postpositional enclitic is not obviously predictable on semantic grounds. If
the choice of postpositional enclitic were made on purely semantic grounds, one might expect
similar roles in verbs with similar meanings to select the same postpositional enclitic, but this
is not always the case. For example, the verb root koroto meet selects the postpositional
enclitic =re while aivaro meet selects the postpositional enclitic =va, despite having very
similar meanings.
8.3.3.1 Verbs that Select the Enclitic =ia
A number of verb roots that take an oblique argument select for the postpositional enclitic =ia.
A few representative examples of these verbs are listed in Table 8.6.
Class Verb Stem Gloss
erava sing
kavorou covet, keep something intended for another, intercept
vuipa think, visualize something continually
aveavero incite to anger
kuga bump into, nudge
tu fasten, strap on the back
Table 8.6: Verbs that Select the Postpositional Enclitic =ia
Some of the verbs that select =ia show subject agreement, as illustrated for the verb root
kavorou keep, hold on to in (562) and the verb root vui think about, visualize in (563).
(562) eake=re
what=ALL
ragai
PPRO.1.SG
va-aro=ia
PPRO.3.SG.N-POSS=LOC
kavorou-u-ei
keep-2SG
-PRES
monia
money
Why are you keeping my money? [Firchow (1984)]
(563) rera=ia
PRO.3.SG.M=LOC
vui-pa-u
visualize-CONT-2SG
You are constantly visualizing him thinking about him. [Firchow (1984)]
Other verbs that select =ia show subject agreement, as illustrated for the verb root kuga
bump into in (564) and the verb root tuu fasten in (565).
(564) vii=ia
PRO.2.SG=LOC
kuga-pa-a-voi
bump into-CONT-1SG
-PRES
I am bumping into you. [Firchow (1984)]

(565) ora-vaiterei=ia
RR-PRO.3.DL.M=LOC
garo-a=va
rattan vine-SG.N=COM
tuu-si-va
fasten-3DL.M-RP
The two of them fastened themselves together with rattan vine. [Firchow (1984)]
174
8.3.3.2 Verbs that Select the Enclitic =re
A number of verb roots that take an oblique argument select for the postpositional enclitic =re.
Class Verb Gloss
aite call father
isiva turn back towards
kasipu be angry
keerapa signal for meeting
koroto meet together
tara look for, search for
Table 8.7: Verbs that Select the Postpositional Enclitic =re
Some of the verbs that select =ia show subject agreement, as illustrated for the verb root
kasipu be angry with in (566) and the verb root isiva turn back towards in (567).
(566) Areipiri
Areipiri
kasipu-pa-ro-i
be angry-CONT-3SG.M
-PRES
oira-ra=re
man-HUM.PL=ALL
ora
and
riako-ra
woman-HUM.PL
Areipiri is angry at the men and women.
(567) ragai=re
PPRO.1.SG=ALL
isiva-u
turn back-2SG
ava-oro
go-DEP.SIM
You turn your back towards me as you go.
Other verbs that select =re show subject agreement, as illustrated for the verb root uvui
measure in (568) and the verb root tara seek, look for, nd in (569).
(568) uva
and
uvui-si-epa
measure-3DL.M-RP
kovo
work
pitupitu-aro=re
custom-POSS=ALL
The two of them measured the work. [Firchow and Akoitai (1974:27)]
(569) Agiosi
Agiosi
aako-va=re
mother-SG.F=ALL
tara-pa-e-vo
look for-CONT-3SG.F
-IP
Agiosi is looking for (his) mother.

8.3.3.3 Verbs that Select the Enclitic =pa
A number of verb roots that take an oblique argument select for the postpositional enclitic =pa.
175
Class Verb Gloss
agigio respect
koruou sacrice
tauo offer in ceremony
kaviko love intensely
vato respect (treat in the appropriate way according to custom)
Table 8.8: Verb Roots that Select the Postpositional Enclitic =pa
Some of the verbs that select the case marker =pa show subject agreement, as illustrated
for the verb root agigio respect in (570) and the verb root avivike honor in (571).
(570) oira-ra
man-PL.N
rutu
very
pautoa=pa
God=BEN
agigio-pa-a-veira
respect-CONT-3PL
-HAB
Everyone respects God.
(571) tuariri-pa-irara
long ago-DERIV-
oisoa
always
tugara
spirit
kare=pa
FP=BEN
koruou-pa-a-ve
sacrice-CONT-3PL
-SUB
aue=ia
CONN=LOC
koie
pig
People of long ago would always sacrice to the bush spirits with pigs.
Other verbs that select the case marker =pa show subject agreement, as illustrated for the
verb root kaviko to love in (572) and the verb root vato to respect, pay honor in (573).
(572) ira
RPRO.3.SG.M
ovii-toa=pa
offspring-SG.M=BEN
oisoa
always
kaviko-irao-pa-re-ve
love-INTEN-CONT-3SG.M
-SUB
He always intensely loved his son. [Firchow (1984)]
(573) eera=pa
DEM.3.SG.M=BEN
avue
in-law
vato-pa-a-veira
respect-CONT-1SG
-HAB
uva
and
viapau
NEG
rera
PRO.3.SG.M
vaisi-pa-a
call-CONT-1SG
I always respect my in-law here and I dont say his name.

4
8.3.3.4 Verbs that Select the Enclitic =va
A number of verb roots that take an oblique argument select for the postpositional enclitic =va.
4
There is a name avoidance taboo in Rotokas culture which applies to in-laws as well as cross-sex siblings.
The cross-sex sibling taboo is even stronger, since it militates against usage of the second person singular form,
requiring substitution of the second personal plural.
176
Class Verb Stem Gloss
kaureo contradict, disagree, be stubborn or rebellious
keri make enemies with, reject friendship
oive shout, yodel, yell
aivaro meet with, bump into
kuara yell at
veku bark at
Table 8.9: Verb Roots that Select for the Postpositional Enclitic =va
Some of the verbs that select the case marker =va show subject agreement, as illustrated
for the verb root kaureo be stubborn or arrogant in (574) and the verb root oive shout to in
(575).
(574) riro-vira
big-ADV
oisoa
always
rera=va
PPRO.3.SG.M=COM
kaureo-pa-a-ve
stubborn-CONT-3PL
-SUB
They were always so stubborn with him.
(575) Terita
Terita
Salome=va
Salome=COM
oive-pa-ro-e
yell-CONT-3SG.M
-IP
Terita is yelling to Salome.

Some of the verbs that select the case marker =va show subject agreement, as illustrated
for the verb root aivaro meet, bump into in (576) and the verb root veku bark at in (577).
(576) Ruri=va
Ruri=COM
aivaro-a-vo
meet-1SG
-IP
eisi
LOC
raivaro
road
I met Ruri on the road.
(577) kakau
dog
kare
FP
ragai=va
PPRO.1.SG=COM
veku-i-vo
bark-3PL
-IP
eisi
LOC
Sikoriara
Sikoriara
Dogs bark at me in Sikoriara.
8.4 Bivalent Two-Place Predicates
Bivalent verb stems take two core arguments, a subject and an object, both of which are realized
as noun phrases or pronouns (i.e., they do not take oblique marking). The bivalent verb stems
can be further subdivided into two subclasses on semantic grounds: two-place predicates and
three-place predicates. The vast majority of bivalent verbs are two-place predicates (as opposed
to three-place predicates). Their agreement pattern and constituent order will be discussed in
turn.
177
8.4.1 Agreement
Bivalent verb roots invariably show inection, as illustrated by (578) and (579).
(578) vii
PPRO.2.SG
upo-pa-a-voi
strike-CONT-1SG
-PRES
Im going to hit you. [Firchow and Akoitai (1974:52)]

(579) uva
so
rakoru
snake
upo-re-voi-va
strike-3SG.M
-PRES
-RP
oira-to
person-SG.M
eira
DEM.MED.SG.F
raga
only
rasi-to
ground-SG.M
vuripie-e-va
ruin-3SG.F
-RP
So that man killed the snake that screwed up the ground. [Firchow and Akoitai
(1974:82)]
There are two permissible constituent orders for a transitive clause. The canonical constituent
order is AOV, as illustrated in (580), but postverbal subjects are relatively common, giving rise
to OVA constituent order, as illustrated in (581). This canonical constituent order is the same
regardless of which normal subtype A and O belong to (pronoun, proper noun, common noun,
classier, etc.).
(580) oira-to
man-SG.M
riako-va
woman-SG.F
upo-re-vo
hit-3SG.M
-IP

(581) riako-va
woman-SG.F
upo-re-vo
hit-3SG.M
-IP
oira-to
man-SG.M
8.5 Three-Place Predicates: Bivalent or Trivalent?
Rotokas has a number of verbs that are generally characterized as ditransitives in the typological
literature, such as give or put. These verbs subcategorize for an oblique argument. The verb
root vate give selects for an oblique argument marked by the benfactive, as in (582), while
the verb root tovo put selects for an oblique argument marked by the locative, as in (583).
(582) Rosiovi
Rosiovi
ira
RPRO.3.SG.M
akuku-vira
free-ADV
kokai
chicken
vate-re-vo
give-3SG.M
-IP
ragai=pa
PRO.1.SG=BEN
Rosiovi gave me a chicken for free.
178
(583) Savia
Savia
veeta
bamboo
tou
CLF
poko-pie-e-voi
explode-CAUS-3SG.F
-PRES
uvare
because
vo-tou
SPEC-CLF
tovo-e-voi
put-3SG.F
-PRES
tuitui kasi
re
sovara=ia
inside=LOC
Savia is making the bamboo explode because he put it in the re.
This oblique argument of these three-place predicates is optional, as illustrated for the verb
root vate give in (584) and for the verb root tovo put in (585).
(584) ravireo
sun
riro-a
big-SG.N
rutu
very
aau-a
light-SG.N
vate-pa-re
give-CONT-3SG.M
The sun gives strong light.

(585) kaveakapie-vira
insecure-ADV
aveke
stone
tovo-i-vo
put-3PL
-IP
uva
and
kove-o-e
fall-3SG.F
-IP
They placed the stone insecurely and it fell down.

The three-place predicate vate give potentially takes three arguments: the giver, the gift,
and the recipient. The number of core arguments associated with three-place predicates is
debatable, at least in the case of vate give. The recipient is optional, as shown by sentences
such as (584) or (586).
(586) Tatu
Tatu
gare-pa-visivi
small-DERIV-ADV
moni-a
money-SG.N
vate-re-voi
give-3SG.M
-PRES
Tatu is giving a small amount of money.

The optionality of the recipient suggests that the recipient is not a core argument and that
there is no need to posit the existence of trivalent verb roots in Rotokas. On this assumption,
three-place predicates would represent a subtype of bivalent verb stems in Rotokas, in keep-
ing with the observation of Dixon (1994:6): All languages distinguish between clauses that
involve a verb and one core noun phrase (intransitive clauses) and those that involve a verb and
two or more core NPs (transitive clauses, including ditransitive as a subtype). There are two
construction types associated with three-place predicate verbs: the indirect object construction,
discussed in 8.5.3, and the double object construction, discussed in 8.5.4.
8.5.1 Agreement
The form of agreement found on trivalent verb stems is -agreement, the same type found on
bivalent stems with a direct object, as can be seen from (587) and (588).
(587) sirao-vira
pity-ADV
rutu
very
uvare
because
aako
mother
upo-ri-voi
strike-2SG
-PRES
Sadly, you killed my mother. [Firchow and Akoitai (1974:80)]

179
(588) sirao-vira
pity-ADV
rutu
very
uvare
because
viapau
NEG
ragai=pa
PPRO.1.SG=BEN
kakae-toa-vai
child-SG.M-INDEF
vate-pa-ri-veira
give-CONT-2SG
-HAB
Sadly, you have not given me children. (Behold, thou hast given me no offspring.)
[Genesis 15:3]
The canonical three-place predicate is vate give, which shows two patterns of constituent
ordering, depending on whether or not the recipient is case-marked. We will simply refer to
these two patterns as constructions and set aside temporarily the question of which is basic and
which derived.
For ease of discussion, we will describe the arguments of a trivalent predicate in terms of
their semantic roles in a dative verb (e.g., give): donor (the agent, the giver), the gift (the person
or thing given), and recipient (the person or thing the theme is given to).
8.5.3 Indirect Object Construction
When the recipient is case-marked with the sufx -pa, constituent order is more exible. The
theme occurs in its usual xed preverbal position, as shown by the grammaticality of (589),
where the theme occurs immediately before the verb, and the ungrammaticality of (590), where
the theme occurs immediately following the verb.
(589) Raratuiri
name
kaakau=pa
dog=BEN
opita
coconut
isi
CLASS
vate-re-vo
give-3SG.M
-IP
Raratuiri gave a coconut to the dog.

(590) *Raratuiri
name
kaakau=pa
dog=BEN
vate-re-vo
give-3SG.M
-IP
opita
coconut
isi
CLASS
There is considerable exibility in the position of the recipient, as shown by the grammat-
icality of the alternative constituent orderings found in (591) through (593). Although these
constituent orders are deemed grammatical by speakers, they are infrequent and texts show few
departures from the order found in (589).
(591) kaakau=pa
dog=BEN
Raratuiri
name
opita
coconut
isi
CLASS
vate-re-vo
give-3SG.M
-IP

180
(592) Raratuiri
name
opita
coconut
isi
CLASS
kaakau=pa
dog=BEN
vate-re-vo
give-3SG.M
-IP

(593) Raratuiri
name
opita
coconut
isi
CLASS
vate-re-vo
give-3SG.M
-IP
kaakau=pa
dog=BEN
The rightward displacement of the subject is possible, as in (594), but the occurence of a
postverbal theme remains ungrammatical with rightward displacement of A, as in (595) and
(596).
(594) kaakau=pa
dog=BEN
opita
coconut
isi
CLASS
vate-re-vo
give-3SG.M
-IP
Raratuiri
name
Raratuiri gave me a coconut.
(595) * kaakau=pa
dog=BEN
vate-re-vo
give-3SG.M
-IP
opita
coconut
isi
CLASS
Raratuiri
name
(596) * vate-re-vo
give-3SG.M
-IP
kaakau=pa
dog=BEN
opita
coconut
isi
CLASS
Raratuiri
name
8.5.4 Double Object Construction
When the recipient appears as a bare NP (i.e., without oblique marking), it occupies the position
normally held by the object, and the theme occurs postverbally, as illustrated by (597) and (598).
(597) Raratuiri
name
ragai
PRO.1.SG
vate-re-vo
give-3SG.M
-IP
opita
coconut
isi
CLASS
(598) uva
so
aako-va
mother-SG.F
vate-e-va
give-3SG.F
-RP
rera
PRO.PER.3.SG.M
ovii-to
son-SG.M
He gave the boy to his mother. [Firchow and Akoitai (1974:79)]
Elicitation conrms that the theme cannot occur preverbally, as in (599).
(599) * Raratuiri
name
ragai
PRO.1.SG
opita
coconut
isi
CLASS
vate-re-vo
give-3SG.M
-IP

181
The rightward displacement of the subject is possible with three-place predicates, as in
(600). A preverbal recipient and theme remain ungrammatical with rightward displacement of
the subject, as illustrated by (601).
(600) ragai
PRO.1.SG
vate-re-vo
give-3SG.M
-IP
opita
coconut
isi
CLASS
Raratuiri
name
(601) * ragai
PRO.1.SG
opita
coconut
isi
CLASS
vate-re-vo
give-3SG.M
-IP
Raratuiri
name
There is mixed evidence with respect to the status of the postverbal theme. Although the
lack of oblique marking suggests that it is a core argument, there is evidence in favor of its
oblique statussee 9.1.2 on three-place predicates derived from two-place predicates through
causativization.
8.6 Conclusion
In this chapter, the valency of verb roots in Rotokas was overviewed. On the basis of a number
of cross-cutting distinctions (the number of participants, the number of core arguments, and the
formof verbal inection), six verb classes were recognized. These ve classes are the product of
the interaction between these various factors. The relationship between predicate type (i.e., the
number of actants/participants in a verbs logical structure) and the number of core arguments
(i.e., the number of core arguments a verb requires) is summarized in Table 8.10.
Participants
Core Arguments 1 2 3
1
2
Table 8.10: Relationship Between Predicate Type and Valency
Table 8.10 shows that the number of core arguments is only weakly predictable on the basis
of the number of participants associated with a predicate. The number of core arguments is
always less than or equal to the number of participants and there is very little evidence in favor
of positing the existence of clauses involving more than two core arguments.
The relationship between valency and verbal inection ( versus ) is summarized in Table
8.11.
182
Verbal Inection
Core Arguments
1
2
Table 8.11: Relationship Between Valency and Verbal Inection
Table 8.11 shows that verbal inection is partially predictable on the basis of the number
of core arguments, but not necessarily vice-versa: if a verb takes two core arguments, it neces-
sarily shows inection, but if a verb shows inection, it does not necessarily take two core
arguments.
The ndings described in this chapter do not support the simple hypothesis that there is a
one-to-one relationship between valency and verbal agreement. Verb roots that take two core
arguments consistently shows agreement whereas verb roots that take a single core argument
are split: some show agreement while others show agreement. The evidence from underived
verb roots therefore rules out a one-to-one relationship, which leaves three possibilities for the
mapping between grammatical roles and verbal agreement, as shown below in Table 8.12.
Possible Conguration Role Inection
One-to-One S
A
Split-S S
A
Split-A S
A
Many-to-Many S
A
Table 8.12: Grammatical Roles and Verb Inection
Before attempting to formulate a generalization that correctly predicts the distribution of
and agreement on verbs, it is necessary to examine valency-changing operationsthat is, the
various mechanisms available in Rotokas for changing (or simply re-arranging) the default va-
lency pattern of verb stems. The behavior of valency-changing derivations provides further ev-
idence of a tight relationship between valency and verb classication, since valency-decreasing
derivations typically derive verb stems whereas valency-increasing derivations typically de-
rive verb stems, but it also introduces a number of complexities that must be resolved before
a comprehensive statement of the distribution of verbal inection can be formulated.
183
184
Chapter 9
Valency-Changing Derivations
In the previous chapter, the valency of verb roots in Rotokas was described and a number of
different verb root classes were identied. It was established that verbal inection is partially
predictable from valency, in the sense that bivalent verb roots (which take two core argument)
uniformly show agreement. However, if a verb shows agreement, it will not necessarily be
bivalent, since monovalent verbs (which take a single core argument) are split into two classes
those that take agreement and those that take agreement. Furthermore, verbal inection is
also not predictable on the basis of the number of participants (i.e., subcategorized obliques),
since two-place monovalent predicates are split between and inection.
In this chapter, we overview the devices for increasing or decreasing the default valency of
verb roots. Valency-increasing derivations are discussed in 9.1 and valency-decreasing deriva-
tions are discussed in 9.2. Valency changing derivations provide little evidence for an under-
lying syntactic difference between and monovalent verb roots, since the various valency-
changing derivations are not sensitive to the distinction; however, they do provide additional
evidence in favor of a tight relationship between valency and verbal inection, since a decrease
in valency is associated with inection and an increase in valency with inection.
9.1 Valency-Increasing Derivations
There are two means of increasing the valency of verb roots in Rotokas: either through zero
derivation in the case of labile verbs (9.1.1) or through sufxation of -pie in the morphological
causative construction (9.1.2). In both cases, the derived bivalent verb stem invariably shows
agreement.
9.1.1 Ambivalent Verb Roots
The majority of verb roots show only a single pattern of valencythat is, a given verb root can
function only as a monovalent verb stem or as a bivalent verb stem. However, a minority of
185
verb roots are capable of functioning either as monovalent stems or as bivalent stems, and these
will be labelled ambivalent. Although the term ambitransitive is more commonly used to
describe such verbs, the term is eschewed here due to the insistence on a distinction between
transitivity (a semantic notion) and valency (a syntactic notion), as previously discussed in 7.3.
9.1.1.1 Labile Ambivalent Verb Roots
Labile verbs are of two types: those where the S of the monovalent verb corresponds to the A
of the bivalent verb and those where the S of the monovalent verb corresponds to the O of the
bivalent verb. Following Dixon (1994), the former will be referred to as S=A verbs and the
latter as S=O verbs.
The verb stem kavau be born/give birth is representative of the S=O variety, as illustrated
in (602), where it takes only one core argument and shows agreement, and in (603), where it
takes two core arguments and shows agreement.
(602) uva
and
riro
big
tarai-irara
know-HUM.PL
aaviko
star
keke-i-va
look.at-3PL
-RP
ovusia
while
Jisu
Jisu
kavau-ro-epa
be born-3SG.M
-RP
The wisemen looked at the star when Jesus was born.

(603) Kivui
Kivui
kaakau
dog
kare
FFP
kavau-e-voi
give birth-3SG.F
-PRES
tupereo-vira
one after another-ADV
Kivui gave birth to puppies one after another.
The verb stem sisiu wash, bathe is representative of the S=A variety, as illustrated by
(604), where it takes only one core argument and shows agreement, and (605), where it takes
two core arguments and shows agreement..
(604) aavu-va
grandparent-SG.F
gapu-vira
naked-ADV
sisiu-pa-o-i
wash-CONT-3SG.F
-PRES
eisi
LOC
Ivitu
Ivitu
Grandmother is bathing naked in the river Ivitu.
(605) riako-va
woman-SG.F
kakae-to
child-SG.M
sisiu-pa-e-voi
wash-CONT-3SG.F
-PRES
uukovi=ia
water=LOC
The woman is washing the child in the river.
Table 9.1 provides a partial listing of labile verb roots in Rotokas, broken down in terms of
the distinction between S=A and S=O. Verbs belonging to the S=O type predominate.
186
Type Verb Stem Monovalent () Bivalent ()
S=A agesi laugh laugh at
aio eat eat
oe vomit vomit
sisiu wash, bathe wash
tavi tell tell
vura look, see look, see
uvu hear, listen, smell hear, listen, smell
S=O aku be salted cook with salt
kaa gag strangle
kaki be cracked open crack open
kasi build re burn
kavau be born give birth
ori cook cook
papu be extinguished extinguish
pau sit plant, build
pura say make
rovo precede start
uvui be able measure, enable
vatatopo ready prepare
vera leave remove
Table 9.1: Labile Ambivalent Verb Roots
The general rule is for these verbs to show agreement when they behave as monovalent
verb stems and to show agreement when they behave as bivalent verb stems.
9.1.1.2 Fixed Ambivalent Verb Roots
There is one class of verb roots that are an exception to the general rule that ambitransitives show
inection as monovalents and inection as bivalents. These ambivalent verb roots show a
xed pattern of inection, in the sense that they uniformly show inection. For example, the
verb root vura look, see is labile, but consistently shows agreement. In other words, as we
would expect, it shows agreement when it occurs with two core arguments, as in (606) and
(607).
(606) ora-ruvu-ro-e
RR-jump-3SG.M
-IP
uvare
because
rakoru
snake
vura-re-vo
see-3SG.M
-IP
He jumped because he saw the snake.

187
(607) tuuta
post
vura-pa-a-voi
look at-CONT-1SG
-PRES
poori-vira
crooked-ADV
oa
RPRO.3.SG.N
tovo-re-vo
place-3SG.M
-IP
Vaisiri
Vaisiri
va=ia
PRO.3.SG.N=LOC
kepa
house
pura-sia
make-DEP.SEQ
I am looking at the post that crookedly Vaisiri put up to make the house.
However, vura also shows agreement when it occurs with a single core argument, as in
(608) and (609).
(608) kaaki-to
one eye-SG.M
katai-toa
one-SG.M
iava
POST
osirei-to
eye-SG.M
vura-pa-re
see-CONT-3SG.M
A one-eyed man sees out of one eye.

(609) voosi-to
blind-SG.M
vearo-pie-re-va
good-CAUS-3SG.M
-RP
Jisu
Jesus
voari
long
tuariri
ago
oa
RPRO.3.SG.N
iava
POST
vura-re-va
see-3SG.M
-RP
Jesus healed a blind man long ago and he could see.

This appears to be a property of verbs of perception, to the extent that it is also true of the
verb roots uvu hear, smell and siovo feel. For example, the verb root uvu hear, smell, taste
functions as a monovalent verb stem in (610) and as a bivalent verb stem in (611), but shows
agreement in both cases.
(610) vii-a
PRO.2.SG-SUB
kaureo-to
stubborn-SG.M
viapau
NEG
uvu-pa-ri-veira
hear-CONT-2SG
-HAB
Youre stubborn, you dont listen.
(611) pokopoko-ara
explode.RDP-PL.N
uvu-pa-a-vo
hear-CONT-1SG
-IP
uvare
because
Patriki
Patrick
pitokava
saucepan
ragiragi-pa-re-vo
beat.RDP-CONT-3SG.M
-IP
I heard the banging because Patrick beat repeatedly on the saucepan.

9.1.2 Morphological Causative
The causative construction has received a great deal of attention within morphosyntactic typol-
ogy and has been the subject of numerous studies (Dixon, 2000; Comrie, 1975; Shibatani, 1976;
Comrie, 1976, 1985b, 1989; Song, 1996). The prototypical causative construction conforms to
the denition provided by Dixon and Aikhenvald (2000) in (612).
(612) applies to an underlying intransitive [monovalent] clause and forms a derived tran-
sitive [bivalent] clause;
188
the argument in underlying S function goes into O function in the causative;
a new argument is introduced, in A function;
there is some explicit formal marking of the causative construction
Rotokas has a morphological causative construction that ts the prole provided in (612) to
the extent that:
it applies to underlying monotransitive verb root to derive a bivalent verb stem;
the original subject of the monotransitive verb root plays the role of O in the derived stem;
a new argument, the causer, is added to the clause and takes over the role of subject;
the verb root is marked by the sufx -pie (which occurs in Slot 1see 5.2.2.1)
The alternation can be illustrated using the verb stem uriri be frightened, a monovalent
verb stem whose base form is illustrated in (613). A morphological causative stem which shows
agreement can be derived from it with the sufx -pie, as illustrated in (614).
(613) uva
so
rei-vira
large-ADV
uriri-ra-epa
be scared-1SG
-RP
I was really scared. [Firchow and Akoitai (1974:19)]

(614) rera
PRO.3.SG.M
uriri-pie-re-va
be scared-CAUS-3SG.M
-RP
aue=ia
CONN=LOC
kuuvu-va
fake-SG.F
rakoru
snake
He frightened him with a pretend snake. [Reader, Ahu]
Causativization applies to a wide variety of verb root types. In fact, it applies to all of the
various predicate types identied in Table 8.1, with the exception of vate give. In other words,
it is not restricted either by valency (monovalent vs. bivalent) or by agreement types ( vs. ).
Some examples of bivalent verb stems derived from monotransitive verb roots with -pie are
provided in Table 9.2.
Inection Monovalent Verb Root Derived Causative Verb Stem ()
kopii die kopiipie kill
tarai understand taraipie teach
kare return, go back karepie return something
agasi be full agasipie ll up
aata swim aatapie make swim
papa y vusipie y
tugisi defecate (dog) tugisipie make defecate
voka walk vokapie operate
vusi gush out vusipie make gush out
Table 9.2: Morphological Causatives Derived From and Monovalent Verb Roots
189
The use of -pie with a monovalent verb root has already been illustrated in (614). Its use
with a monovalent verb root can be illustrated with the verb root tugisi defecate: its default
behavior as a verb root is illustrated in (615) and a bivalent verb stem derived from it with the
sufx -pie is illustrated in (616).
(615) kaakau
dog
evoa
there
tugisi-e-vo
defecate-3SG.F
-PRES
The dog pooped there.

(616) Pita
Pita
kaakau
dog
tugisi-pie-re-voi
defecate-CAUS-3SG.M
-PRES
uvare
because
oira
PRO.3.SG.F
upo-re-voi
hit-3SG.M
-PRES
Peter made the dog defecate because he hit him.

Use of the causative sufx is not restricted to monovalent verb roots, as can be seen from
Table 9.3, which lists a number of bivalent stems that occur with -pie.
Bivalent Stem Causative Stem
aio eat aiopie feed
keke look kekepie show
kae carry kaepie lift, hoist, raise
guvi reveal guvipie expose, reveal
pura make purapie use
ura chew urapie make chew betel nut
Table 9.3: Examples of Causative Sufx -pie
The use of the morphological causative with the bivalent verb root aio eat is illustrated in
(617), where the verb stem aio eat is causativized; the prederivational O (the notional theme)
can either be omitted, as in (617a), or appear as an oblique, as in (617b).
(617) a. kakae
child
vure
FFP
aio-pie-i-va
eat-CAUS-3PL
-RP
aako
mother
riako
FP
The mother is feeding the boy.
b. aako-va
mother-SG.F
kakae-to
child-SG.M
aio-pie-e-vo
eat-CAUS-3SG.F
-IP
itooa=ia
banana=LOC
The mother is feeding the boy banana.
As Comrie (1989) observes, cross-linguistically, there are three basic possibilities for the
syntactic treatment of causativized bivalent verbs, listed below in (618).
(618) 1. the original O retains its status and the causee is peripheral
190
2. the original O is peripheral and the causee functions as O
3. two objects are permitted: the causee and the original O
The four logical possibilities for the remapping of the arguments of a bivalent verb in a
causative construction are listed below in Table 9.4.
Label Causer Causee Theme
1 Double object A O O
2 Double oblique A OBL OBL
3 Causee-as-O A O OBL
4 Theme-as-O A OBL O
Table 9.4: Mappings of Arguments in Causatives Derived from Bivalent Verb Stems
Rotokas appears to possess two of these four types. It has the causee-as-O construction,
where the causee is O and the theme is oblique, as well as a second construction, where the
causee is O and the theme is less clearly oblique. The two construction types are illustrated
using the causative verb stem aivaropie to introduce, which is derived from the monovalent
verb root aivaro to meet, a monovalent verb root that takes an oblique argument marked by
the postpositional enclitic =va, as illustrated in (619).
(619) oira-to
man-SG.M
riako-va=va
woman-SG.F=COM
aivaro-re-vo
meet-3SG.M
-IP
uva
and
oira
PRO.3.SG.F
piiuu-re-vo
grab-3SG.M
-IP
oira=va
PRO.3.SG.M=COM
vuri-a
bad-SG.N
pura-sia
make-DEP.SEQ
The man met up with the woman and grabbed her in order to do bad with her.
When a bivalent verb stem is derived from aivaro meet with the causative sufx -pie,
the causee plays the role of O and the patient/theme occurs as an oblique, marked either by
the postpositional enclitic that it normally selects for (-va in this case) or by the postpositional
enclitic =ia, as in (620).
(620) a.
A=Causer
..
aite-to
father-SG.M
O=Causee
..
ovii-va
daughter-SG.F
aivaro-pie-re
meet-CAUS-3SG.M
OBL=Theme
..
oira-toa=ia
man-SG.M=LOC
The father introduces his daughter to the man.
b.
A=Causer
..
aite-to
father-SG.M
O=Causee
..
ovii-va
daughter-SG.F
aivaro-pie-re
meet-CAUS-3SG.M
OBL=Theme
..
oira-toa=va
man-SG.M=COM
191
It is also possible for the patient/theme to occur in a postverbal slot, where it normally
(though not necessarily) is followed by tapo also, as in (621).
(621)
A=Causer
..
aite-to
father-SG.M
O=Causee
..
ovii-va
daughter-SG.F
aivaro-pie-re
introduce-3SG.M
O=Theme
..
oira-toa
man-SG.M
tapo
also
The status of the postverbal argument in the this second construction is somewhat equivocal,
but probably best qualies as a double object construction. On the one hand, tapo also could
be analyzed as an oblique marker, given that it is often found introducing adjunct noun phrases
into the clause, as illustrated in (622) through (624).
(622) ruve-pa-i
slimy-CONT-PRES
arua
vegetable
tai
CLF
uvare
because
ruve
aibika
tai
CLF
tapo
also
vara
PRO.3.PL.N
ori-a-vo
cook-1SG
-IP
The vegetables are slimy because I cooked them with aibika. [Firchow (1984)]
(623) kakae
child
vure
FP
tou-pa-i-vo
be-CONT-3PL
-IP
aite-to
father-SG.M
tapo
also
osia
as
aako-va
mother-SG.F
kopii-o-e
die-3SG.F
-IP
The children were with father when mother died.

(624) Rarasori-a
Robinson-SUB
pogarapa-to
white-SG.M
oira-to
man-SG.M
ira
RPRO.3.SG.M
tapo
also
kovo-pa-e-veira
work-CONT-3SG.F
-HAB
Sera
Sera
Robinson is a whiteman who Sera works with.
However, tapo is optional for a causatives postverbal argument, which differs from other
oblique arguments by occupying a xed position in the clause. Any deviations from its postver-
bal position give rise to ungrammaticality, as in (625).
(625) * aite-to
father-SG.M
ovii-va
daughter-SG.F
oira-toa
man-SG.M
tapo
also
aivaro-pie-re
introduce-3SG.M

There is a certain symmetry here, in that direct objects occupy a preverbal position whereas
second objects occupy a postverbal position. Also note that the position occupied by the theme
in the double object construction resembles the position occupied by the theme in the double
object construction of three-place predicate verb roots (e.g., vate give) (see 8.5.4) and could
arguably be considered a single construction.
192
9.2 Valency-Decreasing Derivations
There are three valency-changing derivations in Rotokas that derive verb stems that show
agreement: the reexive/reciprocal construction (9.2.1), noun incorporation (9.2.2), and the
resultative construction (9.2.3).
9.2.1 Reexives/Reciprocals
There is no formal distinction between reexives and reciprocals in Rotokas, as can be seen
from(626), where the reciprocal marker ora- derives a reexive/reciprocal verb fromthe causative
verb stem kopiipie to kill (derived from the verb root kopii die). It is ambiguous between a
reexive and a reciprocal reading.
(626) ora-kopii-pie-pa-a-i
RR-die-CAUS-CONT-3PL
-PRES
They are killing themselves./They are killing each other.

The only explicit means of distinguishing formally between a reexive and a reciprocal is
through use of the adverb oisiaropavira (for some speakers, oisiopavira), which means mutu-
ally or reciprocally.
1
(627) oisiaropavira
reciprocally
ora-kopii-pie-pa-a-i
RR-die-CAUS-CONT-3PL
-PRES
They are killing each other. (= They are killing themselves.)

There are three main reciprocal constructions in Rotokas, which differ formally according
to where the prex ora- occurs in the clause. We will look at each separately.
9.2.1.1 Construction 1: Verb Marking
The primary reexive/reciprocal construction in Rotokas involves the prexation of ora- to the
verb stem. This reciprocal construction shows subject agreement and can be characterized as
a valency-decreasing derivation to the extent that it is available for all bivalent verb roots and
stems (i.e., verbs with two core arguments, a subject and a direct object). By way of illustration,
compare the bivalent clause in (628a) with the derived reciprocal clause in (628b).
(628) a. oira
male
kakae-ro
child-PL.CL
riako
female
kakae-ro
child-PL.CL
tario-pa-i-voi
chase-CONT-3PL
-PRES
The little boys are chasing the little girls.

1
The reciprocal adverb oisiaropavira is morphologically complex and consists of three morphemes: the base
form oisio or oisiaro, which appears to be related to the complementizer for comparisons of manner; the deriva-
tional sufx -pa; and the adverbial sufx -vira.
193
b. oira
male
kakae-ro
child-PL.CL
ora
and
riako
female
kakae-ro
child-PL.CL
ora-tario-pa-a-i
RR-chase-CONT-3PL
-PRES
The little boys and girls are chasing each other.

The verb-marking reciprocal construction also occurs with verbs that are associated with
three participants, such as vate give, as illustrated in (629). Note that the verb shows
inection and that the object has been demoted to an oblique argument. This can be understood
as a consequence of the fact that the reciprocal construction is intransitive and permits only one
core argument, forcing any others into the periphery.
ora
and
aita-pa-irara
Aita-DERIV-HUM.PL
(oisiaropavira)
reciprocally
ora-vatevate-pa-a-i
RR-give.RDP-CONT-3PL
-PRES
aasi-ara=ia
belet.nut-PL.N=LOC
The Rotokas and Aita men are giving each other betel nut.
The demotion of the theme in ditransitive-derived reciprocals resembles the demotion of the
theme in ditransitive-derived morphological causatives, already observed in 9.1.2.
The verb-marking reciprocal construction applies productively to bivalent roots/stems, but
it does not occur exclusively with such verb stems. There are also a number of monovalent verb
roots that enter into it. These verbs can occur with or without the reexive/reciprocal prex
ora- with no obvious change in meaning. For example, the reduplicated verb stem tupetupereo
line up can occur with or without ora-, as in (630).
(630) balusi-ara
plane-PL.N
(ora)tupetupereo-pa-i
RR-line.up.RDP-CONT-PRES
rere-pa-oro
land-CONT-DEP.SIM
rasito-a=ia
ground-SG.N=LOC
The planes lined up (with each other) as they landed on the ground.
A number of verb stems of this type are listed below in Table 9.5.
2
Some of these would
arguably qualify as natural reciprocals (Haiman, 1985) or symmetric predicates (Langen-
doen, 1992).
2
This list is not exhaustive and simply lists those verbs stems that were readily identiable in the authors lexical
database of Rotokas (Robinson and Mon, 2006).
194
Verb Stem Gloss Notes
paupau race
pekapekara line up
riga spread, scattered
sekari shake hands Tok Pisin loan
takato argue
tava sun bathe
topogo be reckless or careless
tupetupereo in pairs
uugaa kiss
virato segregated, rened
viru move
Table 9.5: Monovalent Verb Roots Capable of Occuring with the Reexive/Reciprocal Marker
9.2.1.2 Construction 2: Pronoun Marking
There is a second reexive/reciprocal construction type that differs from the rst to the extent
that the prex ora- occurs on an oblique-marked pronoun, and not on the verbal complex. For
example, in (631), reciprocal marking occurs on the third person plural masculine pronoun,
which is an oblique argument of the verb stem reasi dislike.
(631) oira
male
kakae-ro
child-
ora
and
riako
female
kakae-ro
child-
(oisiaropavira)
reciprocally
ora-voea=pa
RR-3.PL.M=BEN
reasi-pa-a-i
dislike-CONT-3PL
-PRES
Little boys and girls dislike each other.

Prexation of ora- to the verb stem is not possible for the verb reasi, as shown by the
ungrammaticality of (632).
(632) * oira
male
kakae-ro
child-
ora
and
riako
female
kakae-ro
child-
ora-reasi-pa-a-i
RR-dislike-CONT-3PL
-PRES
Little boys and girls dislike each other.

The difference between verb and pronoun marking reciprocal constructions has to do with
the distinction between core and oblique argument (Andrews, 2007). The prex ora- occurs on
the verb when a verb takes a direct object (core argument) but on an oblique-marked pronoun
when a verb takes an oblique argument. It does not matter whether the verb with an oblique
argument shows or agreement. For example, the verb root tara look for also takes an
oblique argument but shows agreement, as illustrated in (633). Yet reciprocals based on this
verb are pronoun marking, as can be seen in (635).
195
(633) oira-ra
man-HUM.PL
riako-ra=re
woman-HUM.PL=ALL
tara-pa-i-voi
seek-CONT-3PL
-PRES
The men are looking for the women.

(634) * oira-ra
man-HUM.PL
ora
and
riako-ra
woman-HUM.PL
oisiaropavira
reciprocally
ora-tara-pa-a-i
RR-seek-CONT-3PL
-PRES
The men and women are looking for each other.

(635) oira-ra
men-HUM.PL
ora
and
riako-ra
women-HUM.PL
(oisiaropavira)
reciprocally
ora-voea-re
RR-PRO.3.PL.M=ALL
tara-pa-a-i
seek-CONT-3PL
-PRES
The men and women are looking for each other.

As we might expect given the previously described core/oblique distinction, the prex ora-
also occurs on pronominal adjuncts, as in (636).
(636) vo-vokiaro
SPEC-night
uva
and
oisoa
always
ora-vaiterei
RR-PRO.3.DL.M
ruvara=ia
near=LOC
uusi-pa-si
sleep-CONT-3DL.M
During the night they slept next to each other.
9.2.1.3 Construction 3: Noun Marking
Whereas the difference between the two previous constructions has to do with the distinction
between core and oblique arguments, this third reciprocal construction type is more semanti-
cally restricted. It is only found when the reexive or reciprocal action is performed on body
parts, as illustrated in (637).
(637) riako-va
woman-SG.F
ora
and
oira-to
man-SG.M
aitereia
PPRO.3.DL.M
pau-pa-si-ei
sit-CONT-3DL.M-PRES
paupaa=ia
chair=LOC
uva
and
ora-vavae-aro
RR-hand-POSS
tapatapa-pa-si-ei
hit.RDP-CONT-3DL.M-PRES
A man and a woman, the two of them sit in a chair and hit each others hands. [RR:#54,
TT]
Although it may seem that this construction involves the incorporation of a body part term
into the verbal complex, much like object incorporation (see 9.2.2), there are good reasons to
reject such an analysis. First, incorporated objects are bare stems, devoid of any morphology,
as shown by the contrast between the transitive clause (638a) and its counterpart with object
incorporation in (638b), which lacks the classier and postpositional enclitic.
(638) a. ragai
PRO.1.SG
opita
coconut
isi-re
CLASS=ALL
tara-pa-a-voi
seek-CONT-1SG
-PRES
Im looking for a coconut.

196
b. ragai
PRO.1.SG
opita
coconut
tara-pa-ra-i
seek-CONT-1SG
-PRES
Im looking for coconuts.

However, in the noun marking reciprocal constructions, body part nouns take possessive
morphology (-aro), as already seen in (637).
Second, whereas noun incorporation consistently displays inection, the noun marking
reciprocal construction does not show a consistent form of agreement, as can be seen in (639),
where either or agreement is possible. (It is unclear at present whether there is any semantic
or syntactic difference between the and noun marking reciprocals.)
(639) a. ora-kagave-aro
RR-face-POSS
upo-pa-si-ei
hit-CONT-3DL.M-PRES
They are hitting each other in the face.

b. ora-kagave-aro
RR-face-POSS
upo-pa-si-voi
hit-CONT-3DL.M-PRES

Third, whereas adverbials cannot intervene between an incorporated noun and the incorpo-
rating verb stem, as in (640), no such constraint operates with body-part reciprocals, as shown
by the contrast between object incorporation in (640) and noun-marked reciprocals in (641).
9.2.1.3.1 Incorporation
(640) a. ikau-vira
run-ADV
opita
coconut
kuri-pa-ra-i
scrape-CONT-1SG
-PRES
I am quickly coconut-scraping.
b. * opita
coconut
ikau-vira
run-ADV
kuri-pa-ra-i
scrape-CONT-1SG
-PRES
I am quickly coconut-scraping.
9.2.1.3.2 Reexive/Reciprocal
(641) a. ora-kagave-aro
RR-face-POSS
oisiaropavira
reciprocally
upo-pa-si-ei
hit-CONT-3DL.M-PRES

b. oisiaropavira
reciprocally
ora-kagave-aro
RR-face-POSS
upo-pa-si-ei
hit-CONT-3DL.M-PRES

197
The distribution of reciprocal construction types in Rotokas is predicted by a combination of
factors. Unless the reciprocal action in the clause is performed on a body part, in which case the
noun marking construction will be found, the general rule is: a reciprocal situation described
by a verb with two core arguments will be verb marking whereas one described by a verb with
a single core argument will be pronoun marking. However, there are unexplained exceptions
to the general rule. For example, the verb reo talk is a monovalent verb root which takes
subject agreement and encodes the addressee as an oblique argument, as in (642).
(642) pisipisia-vira
different-ADV
Rarasiori
Robinson
reo-pa-ro-e
talk-CONT-3SG.M
-IP
kakae
child
vure=re
FP=ALL
Robinson speaks differently to the children.
Although we would expect the prex ora- to occur on a pronominal oblique argument, this
is not in fact what happens, as can be seen from (643), which shows verb-marking.
(643) Pita
Pita
vaio
ANIM.DL
ora
and
Jon
Jon
oavuavu=ia
something=LOC
ogaoga-vira
whisper.RDP-ADV
ora-reo-pa-si-e
RR-talk-CONT-3DL.M-IP
Peter and John are whispering to one another about something.

This could be treated as a lexical idiosyncracy, thereby preserving the general rule; how-
ever, this raises some questions concerning the nature of the difference between verb marking
and pronoun marking reciprocals. It may prove to be the case that a grammatical generaliza-
tion couched in terms of the distinction between core and oblique can be derived from lexical
semantics via some sort of linking algorithm (Levin and Hovav, 2006; Van Valin Jr., 2005),
and a full account of this mapping might better explain the distribution of construction types.
Whatever the nal analysis proves to be, these considerations underscore the challenges recip-
rocals pose for an account of transitivity and argument structure not just in Rotokas but also
cross-linguistically (Evans et al., 2007).
9.2.2 Noun Incorporation
Noun incorporation is a phenomenon where a noun occurs tightly bound or attached to the
verb, rather than in its canonical position (Anderson, 1985; de Reuse, 1964; Mithun, 1984,
1986; Rosen, 1989; Sadock, 1986; Sapir, 1911). In Rotokas, both direct objects and oblique
arguments are capable of being incorporated. The incorporation of a direct object is illustrated
with the bivalent verb root ou get in (644) and the incorporation of an oblique argument is
illustrated with the monovalent verb root tara search, look for nd in (645).
(644) Raupeto
Raupeto
oisoa
always
rasi-va
dirt-SG.F
kasi-pa-re-ve
burn-CONT-3SG.M
-SUB
oiso=re
COMP=ALL
ra
and
revasi-vira
red-ADV
kareke-pa-o
appear-CONT-3SG.F
ra
and
oira=ia
PRO.3.SG.F=LOC
vori
money
ou-pa-ro
get-CONT-3SG.M
Raupeto cooks dirt in order to turn it red and he gets money from it.
198
(645) eto
re
tara-pa-ro-e
search for-CONT-3SG.M
-IP
Siape
Siape
oisio
COMP
ra
and
eto
re
kasi-ro
burn-3SG.M
Siape was searching for re in order to make a re.

Verb stems that have incorporated arguments consistently show inection, regardless of
the form of verbal inection they would normally take with non-incorporated arguments. A
bivalent verb root with an incorporated object is illustrated in (646) and a labile verb root with
an incorporated object is illustrated in (647). Although these verb stems would normally show
agreement when they take a direct object (a second core argument), they show agreement
when they have incorporated objects.
3
(646) teapi
lest
varo
clothing
ou-pa-u
get-CONT-2SG
vao=ia
DEM.PROX.SG.N=LOC
moni-a
money-SG.N
ari
but
araisi
rice
ou-sia
get-DEP.SEQ
eva
DEM.MED.SG.N
moni-a
money-SG.N
Dont go clothes-buying with this money, because that money is for getting rice.
(647) Sirikoiri
Sirikoiri
ratao
door
pura-ro-i
make-3SG.F
-PRES
kepa=ia
house=LOC
aire-pa
new-DERIV
kepa
house
rera
PRO.3.SG.M
vo-kepa-aro
SPEC-house-POSS
ra
and
va=ia
PRO.3.SG.N=LOC
uusi-ro
sleep-3SG.F
Sirikoiri is door-making for his new house, the house in which sleeps.
Incorporated arguments have a number of semantic properties that are identied in Hopper
and Thompson (1980) as features of less individuated objects (see 7.3.1 for discussion). The
specic features discussed in Hopper and Thompson (1980) are listed below in Table 9.6.
Parameter Free Nouns Incorporated Nouns
Specicity specic generic
Referentiality referential nonreferential
Deniteness denite indenite
Properness proper common
Animacy animate inanimate
Humanness human nonhuman
Volitionality volitional nonvolitional
Control control non-control
Table 9.6: Properties Relevant to the Individuation of O (Hopper and Thompson, 1980:253)
3
It might be argued that the verb root pura is not really labile (see 9.1.1) in the sense that its monovalent and
bivalent usages do not have a systematically-related meaning (as e.g. aio eat or kavau give birth clearly do).
It is therefore possible to claim that there are simply two verb roots that happen to be homophonous: one that
functions as a monovalent verb stem and shows agreement and another that functions as a bivalent verb stem and
shows agreementi.e., pura
1
[] to say versus pura
2
[] to make, do. However, the prevalence of this type
of homophony in the region suggests that there may be a systematically-related meaning.
199
In Rotokas, incorporated objects are non-specic/non-referential/indenite (in the sense that
they do not refer to a specic, identiable obect). It is presumably for this reason that they can-
not be proper nouns (as revealed by elicitation with native-speaker consultants). The transitivity
parameters of Volitionality and Control do not appear to be relevant, nor do animacy or human-
ness, since inanimate, animate (animals, etc.), and human nouns all function as incorporated
objects. Examples of inanimate incorporated objects were already provided in (646) and (647).
An animate incorporated object is provided in (648) and a human incorporated object is pro-
vided in (649).
(648) koie
pig
kovasi-o-i
be pregnant-3SG.F
-PRES
iria
RPRO.3.SG.F
kakae
child
kavau-pa-o-i
give birth-CONT-3SG.F
-PRES
rara.
later
The pig is pregnant and she will bear children later.
(649) asao-va
sterile-SG.F
riako-va
woman-SG.F
iria
RPRO.3.SG.F
viapau
NEG
kakae
child
kavau-pa-o
give birth-CONT-3SG.F
A sterile women is one who doesnt bear children.

Noun incorporation in Rotokas is identiable on the basis of a number of formal criteria,
which are listed and briey described in (650).
(650) Agreement the incorporating verb root consistently shows agreement (regardless of
its default classication)
No Morphology the incorporated noun is a bare noun stem, with neither sufxes (pos-
sessive, diminutive, etc.) nor enclitics (oblique marking)
Verbal complex the incorporated noun is tightly bound to the verbal complex, permitting
no intervening material
Each of the criteria in (650) is discussed in more detail in 9.2.2.1 through 9.2.2.3.
9.2.2.1 Incoporating Verbs Show Agreement
As previously established in Chapter 8, bivalent verbs (i.e., verbs with two core arguments)
invariably show agreement, as illustrated for the labile verb root aio eat in (651) and (652).
The verb root aio eat (something) takes a classied noun, oveu kue breadfruit, as its direct
object in (651) and a modied third person singular neuter noun kakapikoa aioa little (amount
of) food as its direct object in (652).
(651) urakava
ying.fox
oveu
breadfruit
kue
CLASS
aio-pa-e-vo
eat-CONT-3SG.F
-IP
vokiaro
night
The ying fox ate breadfruit at night.
200
(652) Rarasori
Robinson
kakapiko-a
little-SG.N
aio-a
food-SG.N
aio-pa-re-voi
eat-CONT-3SG.M
-PRES
uva
and
rera=pa
PRO.3.SG.M=BEN
sirao-pa-ro-e
feel sorry-CONT-3SG.M
-IP
Pita
Peter
Robinson is eating little food and Peter feels sorry for him.
However, verbs with an incorporated object invariably show agreement, as illustrated for
the incorporated objects in (653) and (654).
(653) avuka-va
beach-SG.F
iria
PRO.3.SG.M
atope=ia
coconut shell=LOC
arua
greens
aio-pa-o-i
eat-CONT-3SG.F
-PRES
The old woman is eating greens from a coconut shell.

(654) Reari
Reari
ira
RPRO.3.SG.M
akoroa=ia
betel net=LOC
aasi
betel.nut
aio-pa-ro-i
eat-CONT-3SG.M
-PRES
Reari is chewing betel nut with lime. [=(506)]

Noun incorporation is difcult to identify in dependent verbs since dependent verbs lack
subject agreement or tense/mood marking (see 6.3.2.1). It is, however, identiable with verbs
that normally take oblique arguments, since they occur as bare nominals (i.e., without classiers
or postpositional role-marking encliticssee 8.3.3).
9.2.2.2 No Intervening Material Between Verb and Incorporated Noun
Adverbials are normally free to occupy a wide variety of positions within a clause, even inter-
vening between a verb and its direct object (see 6.2.1), as shown in (655) or (656).
(655) oira-to
man-SG.M
koie
pig
ikau-vira
quick-ADV
kaviru-re-vo
steal-3SG.M
-RP
The man quickly stole the pig. [=(394)]

(656) Savere
Savere
takei
wall
pariparikou-vira
crossed-ADV
pura-re-voi
make-3SG.M
-PRES
rera
PRO.3.SG.M
vo-kepa-aro=ia
SPEC-house-POSS=LOC
Savere made criss-crossed the wall on his house.
However, the tight association of incorporated nouns and their associated verbs is evident
from the fact that adverbials cannot intervene between them (cf. (655)), as shown by the un-
grammaticality of (657).
(657) * oirato
man
koie
pig
ikau-vira
quick-ADV
kaviru-pa-ro-epa
steal-CONT-3SG.M
-RP
The man quickly stole the pig. [=(394)]

201
9.2.2.3 No Morphology or Oblique Marking on Incorporated Nouns
Another indication of the tight association between incorporated objects and their associated
verbs is that arguments that normally appear case-marked appear as bare noun roots when in-
corporated. For example, the verb root tara search for, look for normally shows agreement,
as illustrated in (658); however, when the oblique argument is incorporated, oblique marking
is not found, as shown by (659). (Also note the absence of a classier with the incorporated
noun.)
(658) ragai
PRO.1.SG
opita
coconut
isi=re
CLASS=ALL
tara-pa-a-voi
seek-CONT-1SG
-PRES
Im looking for a coconut.

(659) ragai
PRO.1.SG
opita
coconut
tara-pa-ra-i
seek-CONT-1SG
-PRES
Im looking for coconuts.

It is more difcult to identify noun incorporation with verbs that normally take agreement
since there is no tell-tale change in verbal inection; however, the lack of oblique marking
provides a subtle clue, as can be illustrated with the the verb stem ruipa want, which normally
takes an oblique argument marked by the enclitic =pa, as illustrated in (660) and (661).
(660) oari=pa
DEM.DIST.SG.F=BEN
ruipa-pa-ra-i
want-CONT-1SG
-PRES
riako-va
woman-SG.F
I like that woman.
(661) pepa-ara=pa
paper-PL.N=BEN
ruipa-pa-a-veira
want-CONT-3PL
-HAB
oira-ra
man-HUM.PL
rutu
very
voeao
DEM.PROX.PL.M
oa
RPRO.3.SG.N
sivuka-pa-a-veira
smoke-CONT-3PL
-HAB
They always want paper, those men who smoke.
When the oblique arguments of verbs are incorporated, they occur as bare nominals without
oblique marking. For example, the verb root ruipa want normally selects for the benefactive
postpositional enclitic =pa, but no such oblique marking is found in (662) and (663).
(662) uva
and
riro-vira
big-ADV
uuko
water
ruipa-pa-ra-i
want-CONT-1SG
-PRES
I really want water.

(663) ragai
PRO.1.SG
kavori-pa-ra-i
craysh-CONT-1SG
-PRES
uuko-va
water-SG.F
sirova
behind
uvare
because
riro-vira
big-ADV
kavori
craysh
ruipa-pa-ra-i
want-CONT-1SG
-PRES
I am craysh-hunting on the waters edge because I really want craysh.

202
Thanks to the absence of case marking on incorporated nouns, object incorporation is iden-
tiable even in dependent clauses with no person/number/gender marking, provided the verb
stem takes an oblique argument marked by a particular postpostional enclitic. For example,
the verb root tara seek, search for, look for shows agreement and selects for the allative
postpositional enclitic =re, as illustrated in (664).
(664) Agiosi
Agiosi
aako-va=re
mother-SG.F=ALL
tara-pa-e-vo
look for-CONT-3SG.F
-IP
Agiosi is looking for mother.

When tara functions as a dependent verb, it shows no agreement for person/number/gender
but its oblique argument still occurs with the usual enclitic (=re), as illustrated in (665).
(665) oisio
COMP
ruipa-pa-ra-i
want-CONT-1SG
-PRES
ra
COMP
vore-ta
return-2PL
sigo-a=re
knife-SG.N=ALL
tara-sia
nd-DEP.SEQ
oa
RPRO.3.SG.N
viki-ta-vo
lose-2PL-IP
I want you guys to return and nd the knife that I lost.

However, when the verb root tara seek functions as a dependent verb with an incorporated
object, no oblique marking is present, as illustrated in (666) and (667), where the patient/theme
atari sh occurs as a bare nominal without the enclitic =re.
(666) vegei
PRO.1.DL
roko-pa-ve
go inside-CONT-1DL
eisi-re
LOC=ALL
avaka-va
beach-SG.F
atari
sh
tara-sia
seek-DEP.SEQ
Well go to the beach to seek sh.
(667) kakae
child
vasie
CLASS
varu
meat
tara-sia
seek-DEP.SEQ
ava-a-e
go-3PL
-IP
vo-vegoaro
SPEC-jungle
The boy are going meat-nding in the jungle.
9.2.3 Resultatives
In addition to the various constructions that clearly qualify as valency-changing derivations,
there is another derivational sufx, -piro or -viro, that also systematically affects verb classi-
cation.
4
It is illustrated in (668) and (669). The form -piro is found with neuter subjects, as in
(668), whereas the form -viro is found with non-neuter subjects, as in (669) (see 5.2.2.4).
(668) epusi
cat
ragai
PRO.1.SG
gagarike-e-vo
scratch-3SG.F
-IP
uva
and
gagoago-ara
scratch-PL.N
pura-piro
make-RES
A cat scratched me and left a sore.
4
Firchow (1987) describes the two forms as -(u)viro and -piro. It is unclear why he attributes u to the sufx
since there is no evidence of its presence in Firchows materials or my own.
203
(669) Rusire
Rusire
perete
plate
gaveru-e-voi
lose grip-3SG.F
-PRES
uva
and
pege-o-viro-i
break-3SG.F
-RES-PRES
Rusire lost her grip on the plate and it is broken.

Firchow (1987) characterizes the sufx that marks this construction as the completive
sufx but this characterization is questionable given that the form co-occurs with the continuous
sufx (see 5.2.2.3), as illustrated in (670) and (671).
(670) gesio-pie-vira
taste-CAUS-ADV
rutu
very
aio-pa-piro-i
eat-CONT-RES-PRES
arua
vegetable
tai
CLASS
uvare
because
va
PRO.3.SG.N
kuvu-e-vo
pack-3SG.F
-IP
aue=ia
CONN=LOC
veeta
bamboo
The vegetables are tasty (literally, eat tastily) because he packed them in bamboo.
(671) kakae
child
vure
FP
kosikosi-pa-viro-i
come out.RDP-CONT-RES-PRES
kepa
house
sovara
inside
iava
POST
The children have come outside of the house.
These sufxes are consistently associated with verbal inection. The effect of the sufx
on verbal inection can be illustrated with the labile verb root ori cook (see 9.1.1). It shows
agreement when it takes a single core argument, as in (672) and verb agreement when it
takes two core arguments, as in (673).
(672) Vitera
Vitera
okote-sia
crab-DEP.SEQ
ava-o-e
go-3SG.F
-IP
igei=re
PRO.1.PL.EXCL=ALL
kasipu-pa-oro
angry-CONT-DEP.SIM
uvare
because
viapau
NEG
ori-i-e
cook-1PL.EXCL-IP
Vitera went to collect crabs, mad at us because we didnt cook.

(673) Ireviri
Ireviri
koorato
possum
siare-aro
innards-POSS
ori-re-voi
cook-3SG.M
-PRES
Ireviri is cooking the possums innards.

Because ori cook is a labile verb of the S=A type, its subject corresponds to the semantic
role of actor/agent. However, when the verb root occurs with the resultative sufx, its subject
corresponds to the semantic role of patient/theme and the verb shows agreement, as illustrated
in (675).
(674) Rusire
Rusire
arua
vegetable
tai
CLASS
ori-e-vo
cook-3SG.M
-IP
akurovu=ia
salt=LOC
uva
and
vearo-pie-vira
good-CAUS-ADV
rutu
very
ori-piro
cook-RES
Rusire cooked vegetables in salt and they cooked well.
204
(675) araisi
rice
tovure-vira
soggy-ADV
ori-o-viro-i
cook-3SG.F
-RES-PRES
The rice was cooked soggy.

Because of the remapping of semantic roles that occurs with this sufx, it is tempting to
analyze it as an agentless passive. However, the characterization of this sufx as a valency-
decreasing derivation is questionable, given that the objective resultative construction also oc-
curs with monovalent verb stems. For example, the verb kare return normally shows agree-
ment, as in (676), and this does not change when it is used in the objective resultative construc-
tion, as in (677).
(676) Pera
Pera
turituri-vira
direct.RDP-ADV
kare-ro-e
return-3SG.M
-IP
eisi=va
LOC=ABL
sikuru-a
school-SG.N
Pera returned directly from school.
(677) riuriu-vira
dirty-ADV
raga
only
Saro
Saro
kare-ro-viro-i
return-3SG.M
-RES-PRES
vo-va
SPEC=ABL
vegoaro
jungle
Saro returned from the jungle dirty.
The occurence of the resultative sufx with monovalent verb stems is not simply an idiosyn-
cracy that could be explained away in terms of lexicalization, since it occurs with a number of
other monovalent verb stems in addition to kare return. For example, it occurs with the
monovalent root kosi come out in (678) and with the monovalent root pou arrive in (679).
(678) avavarao-pa-vira
dizzy-CONT-ADV
raga
only
kosi-ro-viro
come out-3SG.M
-RES
kove-sia
fall-DEP.SEQ
vo-garasi
SPEC-grass
ivara=ia
above=LOC
He went outside dizzily and fell down on top of the grass.
(679) Riki
Riki
ora-vikiviki-irao-ro-i
RR-jump.RDP-INTEN-3SG.M
-PRES
roru-pa-oro
happy-CONT-DEP.SIM
ovusia
while
aite-to
father-SG.M
pou-ro-viro-i
arrive-3SG.M
-RES-PRES
kotokoto-ara=va
cargo-PL.N=COM
Riki jumped up and down happy when father arrived with cargo.
This construction is also found with bivalent verbs. Its use with the bivalent verb stem ori
cook was already provided in (675). Since ori cook is labile (see 9.1.1), it is useful to
provide a less equivocal example, such as the verb stem poroporo shatter. It is a bivalent
stem which normally shows agreement, as in (680), but shows agreement when used in the
objective resultative construction, as in (681).
(680) Pita
Peter
siveri
cement
poroporo-pa-re-voi
shatter-CONT-3SG.M
-PRES
torara=ia
axe=LOC
Peter is shattering the cement with a rock.
205
(681) Pita
Peter
siveri
cement
vuro-re-voi
throw at-3SG.M
-PRES
avike=ia
rock=LOC
uva
and
poroporo-o-viro-i
shatter-3SG.F
-RES-PRES
Peter threw rocks at the cement and it shattered.

Although the resultative sufx is found with bivalent verb roots, it does not occur with
bivalent verb stems derived with the causative sufx -pie. In this respect, the resultative differs
from other valency-reducing sufxes (e.g., the reexive/reciprocal), which can be fed by the
causative sufx (cf. (626)).
The verb forms marked by -piro and -viro would be characterized as objective resultatives:
the underlying subject of the state (which is expressed by the surface object of the stative
predicate) is co-referential with the underlying subject of the preceding action, while in the
case of the objective resultative it is co-referential with the underlying object of the latter
(Nedjalkov and Jaxontov, 1988:9). Nedjalkov and Jaxontov (1988:6) characterize resultatives
as verb forms that express a state implying a previous event, distinguishing them from statives
as follows: the stative expresses a state of a thing without any implication of its origin, while
the resultative expresses both a state and the preceding action it has resulted from.
The objective resultative does not co-occur with other valency-changing derivations. It is
incompatible with the causative sufx and the reexive sufx.
9.3 Conclusion
Although one of the strongest generalizations made concerning verbal inection in verb roots
i.e., that a verb stem that has a direct object will show agreementis couched in terms
of valency, the evidence from valency-changing derivation is more equivocal. The behavior
of valency-increasing derivations supports this generalization and the reexive/reciprocal con-
struction provides additional evidence for a fundamental distinction between monovalent and
bivalent verb stems, but noun incorporation and the resultative construction are not as obviously
syntactic. In fact, the distinction between core and non-core arguments does not appear to be
relevant to noun incorporation, since both direct objects and oblique arguments are able to in-
corporate and the reexive/reciprocal and resultative construction do not apply exclusively to
bivalent verb roots.
206
D
e
f
a
u
l
t
I
n
e
c
t
i
o
n
V
a
l
e
n
c
y
A
r
g
u
m
e
n
t
s
C
a
u
s
a
t
i
v
e
R
e
e
x
i
v
e
/
R
e
c
i
p
r
o
c
a
l
N
o
u
n
I
n
c
o
r
p
o
r
a
t
i
o
n
R
e
s
u
l
t
a
t
i
v
e
1 1 S,
2 1 S,
3 1 S, Oblique ?
4 1 S, Oblique ?
5 2 A, O
6 2 A, O, Oblique ?
Table 9.7: Relationship Between Verb Root Classes and Valency-Changing Derivations
In the following chapter, more in-depth analysis of semantic roles and their realization in
Rotokas will be provided in order to pursue the idea that the distinction between and agree-
ment cannot be formulated in terms of simple grammatical roles, but requires reference to a
more articulated semantic event structure.
207
208
Chapter 10
The Semantics of Split Intransitivity in
Rotokas
In Chapter 8 and Chapter 9, valency in Rotokas was described and found to be an imperfect
predictor of verbal inection. Derived verb stems show a fairly consistent pattern of verbal
inection: with the exception of reexives/reciprocals, derived monovalent verb stems take
inection while derived bivalent verb stems take inection. However, monovalent verb roots
(underived by denition) are split between two classes: those that take inection and those
that take inection. Since verbal inection is not predictable on the basis of valency alone,
it remains to be seen whether it can be predicted on semantic grounds. The semantic basis
of Rotokas split intransitivity is discussed in 10.1 and the semantic roles associated with the
various grammatical roles found in Rotokas are examined in 10.2. In the following chapter,
these results are situated within a broader typological context and the wider implications of the
split intransitivity found in Rotokas are discussed.
10.1 Semantic Classes of Split Intransitivity
There are four broadly-dened semantic classes where the distinction between and mono-
valent verb roots emerges more clearly: verbs of motion (10.1.1), verbs of bodily process
(10.1.2), verbs of sound emission (10.1.3), and (for want of a better label) verbs of inferred
causation (10.1.4). Each is discussed in turn.
10.1.1 Motion
Verbs of motion are verbs that lexicalize a motion eventi.e., a situation containing move-
ment or the maintenance of a stationary location (Talmy, 1985). Talmy (1975, 1985, 2007)
distinguishes between various components in the semantics of motion, listed in (682).
(682) gure a moving or conceptually movable object whose path or site is at issue
209
ground a reference-frame, or a reference object stationary within a reference-frame,
with respect to which the Figures path or site is characterized (Talmy, 2007:71)
source the start point of the moved object
path the path followed or site occupied by the Figure object with respect to the Ground
object (Talmy, 2007:70)
goal the end point of the moved object
manner the particular way in which the movement occurs
cause the cause of the motion event (a human actor, an event, or simply gravity)
These various components of a motion event can be illustrated with a few sentences from
Rotokas. In (683) through (685), three different motion events are described that involve rolling
objects. In (683) and (684), the motion event is encoded as an intransitive verb. In (683), the
downward trajectory (the path) is inferred from context and the manner is lexically specied by
the verb; in (683), the downward trajectory is explicitly specied by the main verb while the
manner is specied by the dependent verb. In (685), the motion event is encoded as a transitive
verb, where the cause plays the role of subject and the moved object plays the role of object.
(683) pukui-a
mountain-SG.N
iava
POST
peri-piro-i
roll-RES-PRES
opita
coconut
isi
CL
uva
and
Tate
Tate
upo-voi
hit-PRES
kukue
head
iava
POST
The coconut has rolled down from the mountain and hit Tate on the head.
(684) oira-to
man-SG.N
periko-pa-oro
roll-CONT-DEP.SIM
pukui-a
mountain-SG.N
iava
POST
kove-ro-e
fall-3SG.M
-IP
uvare
because
rera
PPRO.3.SG.M
rita-i-voi
shoot-3PL
-PRES
The man fell rolling down the mountain because they shot him.
(685) vuruko-a
log-SG.N
peri-re-vo
roll-3SG.M
-PRES
Iteirea
Iteirea
eisi
LOC
vaesi-a
mountain-SG.N
Iteirea rolled the log down the mountain.
As can be seen in the contrast between (683) and (684), verbs of motion in Rotokas are not
uniform in their verb classication. While some monovalent verb roots denoting motion events
take inection, others take inection. For example, the verb root ava go is an verb, as
illustrated in (686), while the verb root voka walk is a verb, as illustrated in (687).
(686) Riopeiri
Riopeiri
kakae
child
vure
FP
voka-pie-pa-oro
walk-CAUS-CONT-DEP.SIM
ava-ro-e
go-3SG.M
-IP
raiva=ia
road=LOC
Riopeiri went along the road walking the children.
(687) Jon
Jon
kovuru-vira
cross-ADV
voka-pa-re-voi
walk-CONT-3SG.M
-IP
raiva=ia
road=LOC
John walked across the road.
210
What appears to distinguish motion of verbs with agreement from motion verbs with
agreement is that the former lexicalize manner of motion. The verbs of motion that show
agreement are more schematic verbs of motion whereas the verbs of motion that show
agreement are more semantically restricted, having a manner component. This emerges fairly
clearly from the list of monovalent verbs of motion provided in Table 10.1.
Class Class
ava go
iipa ascend
ira go rst, precede
kare return
koata enter
kosi go out, exit
kove fall, drop
urio come
vara descend
varu go up
vore return
aata swim
gosigosi limp
ikau run, speed
kapere swim on surface
papa y
paru ow, move, go, run
raurau sway back and forth
roko enter jungle
tou be, live, reside
viku go to garden
voka walk
vusi rush out, erupt
Table 10.1: Monovalent Verb Roots of Motion in Rotokas
Although many of the motion verbs that take inection typically occur with human sub-
jects, they are not subcategorized as such and do occur with inanimate subjects, as illustrated in
(688) and (689).
(688) uva
and
riro-to
big-SG.M
kiuvu
wind
voea=re
PRO.3.PL=ALL
vusi-re-voi
emerge-3SG.M
-PRES
ovusia
while
sipiro-pa-a-i
play-CONT-3PL
-PRES
A big wind is rushing out on them while they play.

(689) Ivitu
Ivitu
paru-pa-e-veira
ow-CONT-3SG.F
-HAB.ANIM
vara-pie-vira
lower-CAUS-ADV
eisi=re
LOC=ALL
avakava
ocean
Ivitu [a river near Togarao] runs down to the ocean.
10.1.2 Bodily Process
Verbs describing bodily processes (e.g., coughing, sneezing, breathing, defecating, etc.) have
been singled out in many discussions of split intransitivitye.g., the discussion of verbs of bod-
ily function and process in Merlan (1985:350) or the discussion of processes whose domain
is an animate body in Rosen (1984:64).
211
In Rotokas, the distinction between and inection cuts across the class of bodily process
verbs. Some verbs describing bodily processes show inection while others show inection.
For example, the verb roots vavau breathe, voevoe belch, burp, and eavi ooze pus are ,
as illustrated by (690) through (692).
(690) oearo-vu
PPRO.3.PL-ALT
vuri
bad
geesi-vira
smell-ADV
vavau-pa-a-veira
breathe-CONT-3PL
-HAB
Some people have bad breath (literally: breathe in a bad-smelling manner).
(691) oira-to
man-SG.M
voevoe-pa-ro-i
belch-CONT-3SG.M
-PRES
The man is belching.

(692) sipareo
nger
vii
PRO.2.SG
iava
POST
eavi-pa-o-i
ooze pus-CONT-3SG.F
-PRES
Your nger is oozing pus.

However, other verbs of bodily process show agreement. For example, the verb roots
ritoko defecate (pig), puu fart, and opoko defecate (generic) show agreement, as illus-
trated in (693) through (694).
(693) ragai
PRO.1.SG
rera-aro
PPRO.3.SG.M-POSS
koie-to
pig-SG.M
ritoko-pa-re-vora
defecate-CONT-3SG.M
-DP
evoa
there
My pig defecated over the there. [Firchow (1984)]
(694) Seseva
Seseva
riro-vira
big-ADV
puu-e-vo
fart-3SG.F
-IP
uva
and
oira
PPRO.3.SG.F
agesi-i-vo
laugh-3PL
-IP
oira-ra
man-HUM.PL
Seseva ripped a big fart and people laughed at her.
(695) aako-va
mother-SG.F
kakae-to
child-SG.M
iava
POST
takato
dangling turd
vera-pa-e-voi
remove-CONT-3SG.F
-PRES
uvare
because
opoko-re-vo
defecate-3SG.M
-IP
The mother wiped a dangling turd from the child because he defecated.
There is at least one verb whose classication is unclear. It shows agreement in an exam-
ple sentence found in Firchow (1984), provided in (696), but agreement in (697). It is unclear
whether this represents a mistake, a change in classication, or inter-speaker variability; how-
ever, it does not repesent the norm, which is for a verb to be assigned uniquely to one of the two
classes of subject agreement.
(696) repoo-pa-ro-i
hiccup-CONT-3SG.M
-PRES
He is hiccupping. [Firchow (1984)]

212
(697) Karevaute
Karevaute
repoo-re-voi
hiccup-3SG.M
-PRES
Karevaute is hiccuping.
A list of verbs of bodily process and their classication as or is provided in Table 10.2.
Class Class
asige sneeze
eavi ooze pus
kokoisi sweat, perspire
kuuri grunt, huff and puff
repoo hiccup
revasi bleed
vagoto cough
vavau breathe
voevoe belch
opoko defecate (generic)
eeko defecate (human)
pigo defecate (chicken)
puu fart
ritoko defecate (pig)
tugisi defecate (dog)
tupi defecate (rat or insect)
voakou excrete (urine or feces)
viviko urinate
Table 10.2: Bodily Process Verbs in Rotokas
Table 10.2 shows that verbs of bodily process are not uniform with respect to verb classi-
cation in Rotokas. The split appears to between what might be labelled verbs of excretion,
which uniformly show inection, and all other verbs describing bodily processes, which show
inection. The verbs of excretion in fact represent a hierarchy of lexical hyponymy (Cruse,
1986), as shown in (698). Elicitation work with informants reveals that the verbs specic to
a particular type of animal are not truly subcategorized for the animal in question, but rather
reect the type of feces typically produced by such an animal. It is therefore possible to use one
of the animal-excretion verbs with a human actor, but the result is perceived to be humorous,
since it attributes an unusual state of affairs to a human being.
(698) voakou excrete
viviko urinate opo defecate
eeko
human
tugisi
dog
ritoko
pig
tupi
rat or insect
It is tempting to account for the difference between the two classes of verbs of bodily process
in terms of the notion of control, which frequently gures into discussion of split intransitivity
213
(as we will see later in Chapter 10.3). According to this view, the semantic difference between
these two classes is that one class consists of bodily processes that are, from the conventional-
ized viewpoint of the Rotokas cultural worldview, at least in principle controllable (urinating,
defecating, and farting), while the other consists of bodily process that are not (bleeding, sweat-
ing, coughing).
10.1.3 Sound Emission
There is another semantic eld that is found within the class of intransitives, and these are
verbs of sound emission (Snell-Hornby, 1983; Levin and Hovav, 1995a). Verbs of sound emis-
sion are verbs whose primary meaning involves the emission of sound (e.g., creak, groan, or
rumble in English). They can be distinguished from speech act verbs, whose primary meaning
revolves around a communicative act, which typically involves sound as the medium, but not
necessarily. For example, in (699) and (700), there is no involvement of sound as the medium
of communication in the use of the speech act verb root tavi tell.
(699) roo
DEM.PROX.SG.M
iava
POST
ito-to
banana-SG.M
vao
DEM.SG.N
guru-va
leaf-SG.F
vevei
yellow
oa
RPRO.3.SG.N
ragai
PRO.1.SG
tavi-pa-i
tell-CONT-3PL
oiso
COMP
kopi-pa-i
die-CONT-PRES
vo-guru-va
SPEC-leaf-SG.F
The yellow leaf of this banana tells me that the leaf is dying.
(700) vuuta
time
keke-pa-to
look-DERIV-SG.M
igei
PRO.1.PL.EXCL
tavi-pa-re-veira
tell-CONT-3SG.M
-HAB
oiso
COMP
voki-ei
day-PRES
o
or
ravire
sun
vuuta
time
o
or
avi-ei
light-PRES
The clock tells us that it is night or daytime or morning.

Verbs of sound emission in Rotokas are split between and agreement. While some verbs
of sound emission show agreement (e.g., era sing in a high pitch), others show agreement
(e.g., pupi sing and dance with wind instruments).
(701) riako-ra
woman-
karapi-vira
high pitched-ADV
era-pa-a-veira
sing-CONT-3PL
-HAB
era-ara
song-PL.N
rutu=ia
very=LOC
vo-voki-ro
SPEC-day-
rutu=ia
very=LOC
The women sing all of the songs in a high pitch every day.
(702) Voitari
Voitari
oisioa
always
riro-va=va
big-SG.F=COM
avurara
large axe
pupi-pa-re-ve
sing-CONT-3SG.M
-SUB
Voitari always sings with a big axe.
214
Table 10.3 provides a listing of various verbs of sound emission, classied according to
whether they show or agreement.
Class Class
era sing
geuru snarl and spit
giigiirau groan
karapi sing soprano
koi high pitched sound
kokovae sing
koova sing
kovokovo play Jews Harp
ogaaga whisper, talk quietly
oive shout, yodel, yell
ruvaku low pitch, bass
siiguru drum, beat drum
sirava hiss
utave blow Tritons trumpet
vepu yell
vikuta whistle
viokeke/viokoko whistle
aka open the mouth, shout
gau cry, weep
gipugipu whimper
guruko make noise
kakupie shout, yodel
kapuu dumb, mute
koikoi groan with pain
kokoroku crow
kukuuku make footfall
pegu bark
pupi play bamboo pipes
vekaveka gasp
vauvau make loud chopping noise
Table 10.3: Sound Emission Verbs in Rotokas
As observed in Levin et al. (1997), verbs of sound emission have not received much attention
in the typological literature, and they are largely absent from discussions of split intransitivity.
As a result, there are few obvious candidates for semantic differentiation based on the treatment
of this semantic eld cross-linguistically. However, it is noteworthy that a large number of the
verb roots that take inection denote discrete events with a well-dened end point (bounded,
telic) (Comrie, 1985a; Chung and Timberlake, 1985). For example, although there are verb
roots in both classes whose meaning involve the playing of an instrument, the meanings differ
in subtle but important respects. Whereas the verb root kovokovo play Jews harp and pupi
play bamboo pipes both denote the playing of an instrument, the former refers to the general
activity whereas the latter refers to a performance in the context of a traditional song and dance
performance (known as a singsing kaur in Tok Pisin).
(703) Uriora
Uriora
vata
CL
koova-pa-i
sing-CONT-PRES
pupi-pa-oro
singsing-CONT-DEP.SIM
The people of Uriora are singing while performing a singsing.
215
10.1.4 Inferred Causation
There are a number of monovalent verbs that do not t neatly within the semantic domains
discussed in the previous sections. For example, the verbs aviavi light up, exemplied in
(704), or sipokoro sprout, exemplied in (705) (which lacks subject agreement due to the fact
that its subject is neuter but can be identied as from the TAM marking).
(704) aviavi-re-voi
light up-3SG.M
-PRES
parakau-oro
spread-DEP.SIM
uva
and
oira-ra
man-HUM.PL
ora-sita-a-i
RR-startle-3PL
-PRES
vaasia-vira
strong-ADV
The lightning is lighting up the sky and people are startled.
(705) kukara
corn
takura-aro
seed-POSS
sipokoro-voi
sprout-PRES
The corn seed is sprouting.

The semantic commonality of the remaining monovalent verb roots that show agreement
is not as easily pinned down. These verbs denote events that can be conceptualized as arising
from inherent properties of the entity participating in the event, such as lightning ashing, seeds
growing, re producing smoke, etc. These verb roots are listed below in Table 10.4.
Verb Root Meaning
koke make rain
kovo work, garden
parakau light up, spread across an expanse
pika splash
raraka become light
sikere streak of light, start to shine, dawn
sipokoro sprout through surface
sipukao sprout
sipusipu grow, shoot up
sirusiru be shiny
siruvau be good-looking, nice appearance
sisikore shine, gleam, glisten
ukauka swish around, splash around
Table 10.4: Monovalent Verbs of Internal Causation
This semantic class is not as widely recognized in discussion of split intransitivity as verbs
of motion or bodily processes, but it is an interesting class, sinceunlike the other semantic
216
classes discussed in the previous sectionthe verb roots in it do not denote events that neces-
sarily involve an agent (e.g., aviavi light up brightly in (704)), and in some cases preclude
one (e.g., sipokoro sprout in (705)).
DeLancey (1985) provides a discussion of these event types and their treatment in languages
with split intransitivity in which he observes that some types of predicates can be conceptualized
as events which cause discrete results, much like transitive predicates.
We can also easily accomodate the ambiguity of the sparkle class of predicates,
which according to Rosens observation sometimes predicate agentivity of their
argument in spite of the fact that volition is not only absent but impossible, for here
too there is an aspect of the eventthe sparkle, in the case of that predicatewhich
can be conceptualized as simply the perceptual manifestation of the event denoted
by the predicate or as a distinct product of the event of sparkling. (DeLancey,
1985:9)
This provides one way of understanding the agentivity of verbs such as those listed in
Table 10.4. However, a number of the event types discussed in DeLancey (1985) do not show
inection in Rotokas. For example, predicates such as bleed, vomit, and sweat do not
show inection in Rotokas, as already observed in 10.1.2.
10.1.5 Conclusion
In the previous sections, the various monovalent verb roots that show agreement were grouped
into a number of semantic elds. These semantic elds are not exhaustive, in the sense that there
are a few monovalent verb roots that show agreement which do not obviously t into any of
these semantic eldse.g., the verb root roru be happy, exemplied in (706) and (707).
(706) uva
and
roru-a-voi
be happy-1SG
-PRES
rutu
very
uvare
because
vii
PRO.2.SG
ragai
PRO.1.SG
tauva-ri
help-2SG
I am glad, because you helped me. [Firchow (1984)]

(707) roru-pa-i-vo
be happy-3PL
-IP
uvare
because
wiri-a
win-SG.N
aue=ia
CONN=LOC
vorivoro
volleyball
They were happy because of a win at volleyball.
Likewise, the verb root tou to be, exist also fails to t neatly into the previously discussed
semantic elds. Although it could be construed as a verb of motion, to the extent that it denotes
a lack of motion, there is no manner component, and its meaning is quite general, in many cases
effectively serving as a copula, as in (708) or (709).
1
1
Comparative research on the East Papuan languages most closely related to Rotokas might shed light on the
grammaticalization of tou. It seems likely that it once had a more dynamic meaning but has undergone semantic
bleaching but retained its former classication due to its high frequency.
217
(708) Tutue
Tutue
pukui
mountain
kaepie-vira
high-ADV
tou-pa-i-voi
be-CONT-3PL
-PRES
Mount Balbi is up high.

(709) riako-va
woman-SG.F
pugu-pa-vira
pregnant-DERIV-ADV
tou-pa-e-voi
be-CONT-3SG.F
-PRES
uvare
because
kakae-to
child-SG.M
oira
PRO.3.SG.F
kovu-aro
belly-POSS
sovara=ia
inside=ENC
tou-pa-re
be-CONT-3SG.M
The woman is pregnant because a child is inside of her belly.

Given the diversity of event types denoted by the monovalent verbs that show inection
in Rotokas, it is difcult to extract a single semantic parameter that is common to all of them.
While there is a striking family resemblance (Wittgenstein, 1953/2001) among the various
monovalent verb roots that show agreement, it is difcult to articulate a set of necessary and
sufcient conditions, and there is a risk of lapsing into ad hoc explanations in order to provide a
semantic basis for a formal difference that could very well be arbitrary. In the following section,
a number of thematic roles will be examined in order to assess the extent to which they are able
to predict the previously described patterns of verbal inection.
10.2 Thematic Roles
Since many theories of split intransitivity make reference to semantic notions (such as agentivity
or affectedness), a proper evaluation of them with respect to Rotokas requires a more detailed
examination of the semantic roles associated with verbs in Rotokas.
Andrews (2007) observes that a distinction is typically found in languages, and to vary-
ing degrees insisted upon by linguistic theory, between two types of case: semantic case and
grammatical case. The distinction is sometimes characterized as the difference between core
and oblique grammatical functions (Andrews, 2007:154): One set of cases, commonly called
syntactic, structural, or direct cases, mark the core functions, another, commonly called
semantic cases, mark the oblique functions. Andrews (2007:154) also observes that the dis-
tinction between the two boils down to semantic generality: NPs with syntactic cases tend to
express a wide range of semantic functions and to be targetted by rules sensitive to grammatical
function, while NPs with semantic cases tend not to have these properties.
The idea that the distinction between semantic case and grammatical case is one of semantic
generality is made explicit in Role and Reference Grammar (RRG) (Van Valin Jr. and LaPolla,
1997; Van Valin Jr. and Wilkins, 1996; Van Valin Jr., 2005), where semantic roles are treated
according to three levels of generality:
Verb-specic semantic roles Semantic roles that are specic to a particular verb (e.g., killer,
hearer, smasher, etc.).
218
Thematic relations Semantic roles that generalize over verb-specic roles (e.g., Agent, Pa-
tient, etc.).
Macro-roles Semantic roles that generalize over thematic relations (e.g., Actor and Under-
goer).
The way in which verb-specic semantic roles might be grouped together into thematic rela-
tions and thematic relations in turn grouped together into macro-roles is illustrated for Rotokas
in Figure 10.5.
Giver
Runner
Killer
Speaker
Dancer
Thinker
Believer
Knower
Presumer
Hearer
Smeller
Feeler
Taster
Liker
Lover
Hater
Seen
Heard
Liked
Given to
Sent to
Handed to
Located
Moved
Given
Broken
Destroyed
Killed
Agent
Cognizer
Perceiver
Emoter
Experiencer
Stimulus
Recipient
Theme
Patient
Actor
Undergoer
Subject
Object
Oblique
Table 10.5: From Verb-Specic Semantic Roles to Grammatical Relations: Adapted for Rotokas
from Van Valin Jr. and Wilkins (1996) and Van Valin Jr. (2005)
219
In the following section, a handful of the more commonly discussed semantic roles and their
encoding in Rotokas, as well as its effect on verbal inection, will be discussed.
10.2.1 Agent
The thematic role of agent has played a very central role in grammatical theory. Since there
have been many different conceptions of agenthood in the literature, it pays to pin down a
bit more precisely what is meant by the term. The prototypical agent is human, volitional,
and intentional (DeLancey, 1985; Frawley, 1992), and performs an action that brings about an
immediate, observable change of state. For this reason, verbs such as break or kill are typically
used for the purposes of illustration. In Rotokas, there are at least three verbs that would be
translated as kill: upo strike, ght, as in (710); the causative verb kopiipie kill (literally:
make die), as in (711); and tagoro assassinate, kill in secret, as in (712).
(710) Tapi
Tapi
araoko-to
brother-SG.M
eaka-re-va
hand over-3SG.M
-RP
viuru-pa-irara
enemy-DERIV-HUM.PL
vavaea-ro=ia
hand-PL.N=LOC
oisio ra
COMP
rera
PPRO.3.SG.M
upo-i-ve
hit-3PL
-SUB
Tapi put his brother in the hands of the enemy and they killed him.
(711) Kokota
Kokota
sora-to
sorcerer-SG.M
ira
RPRO.3.SG.M
oira-ra
people
kopii-pie-pa-re-veira
die-CAUS-CONT-3SG.M
-HAB
Kokota is a sorcerer who kills people.
(712) ora-upo-pa-oro
RR-ght-CONT-DEP.SIM
ira-vu
RPRO.3.SG.M-ALT
tagoro-i-vo
kill-3PL
-IP
oa iava
therefore
koopi-ro-e
die-3SG.M
-IP
While ghting, he killed one man and thats why he died.

Van Valin Jr. and Wilkins (1996) observe that a verb such as kill is not necessarily agentive
to the extent that it does not require that the agent act intentionally, whereas a verb such as
murder does, as illustrated by (713) and (714).
(713) a. Larry accidentally killed the deer.
b. * Larry accidentally murdered the deer.
(714) a. The falling tree killed the camper.
b. * The falling tree murdered the camper.
220
This observation has relevance for Rotokas. Although both kopiipie and tagoro can be trans-
lated as kill, the latter appears to require intentionality, as discussions with native speakers
reveal. Although the verb stem upo is translateable as hit in some cases and kill in others,
its semantics are primarily concerned with the manner of action (striking, hitting) rather than
its outcome. In (715), the verb stem kopiipie kill serves as the independent verb while upo
ght, strike plays the role of dependent verb and species the manner in which the killing
takes place.
(715) Tomas
Tomas
Jon
Jon
kopii-pie-re-vo
die-CAUS-3SG.M
-IP
rera
PPRO.3.SG.M
upo-oro
hit-DEP.SIM
Tomas killed John by hitting him.
Languages vary in the extent to which departures from the prototypical transitive situation
require different grammatical treatment. For example, natural forces depart from the proto-
typical transitive situation to the extent that changes of state caused by them do not involve a
volitional human agent. Accordingly, they cannot play the role of subject in a transitive verb in
some languages, such as the Papuan language Usan (Reesink, 1984), as illustrated in (716).
(716) a. munon
man
eng
the
nam
tree
s-orei
cut-3SG.FP
The man cut a tree.
b. * moon
wind
aib
big
nam
tree
boat-erei
break-3SG.FP
A strong wind broke the tree. (Reesink, 1984:131)
In Rotokas, neither animacy nor volitionality are necessary conditions for subjecthood, as
both prototypical and non-prototypical agents are able to serve as subjects of bivalent verb roots,
as illustrated by the non-prototypical subjects in (717) through (719).
(717) uuvau-va
tuberculosis-SG.F
Rara
Rara
kopii-pie-e-va
die-CAUS-3SG.F
-RP
Tuberculosis killed Rara.

(718) riro
big
kou
CLASS
toru
wave
kou
CLASS
opuruva
canoe
gasigasi-voi
break.RDP-PRES
A big wave is breaking the canoe.

(719) kiuvu
wind
erako-va
tree-SG.F
rukeruke-re-voi
shake.RDP-3SG.M
-PRES
uvare
because
riro-vira
big-ADV
kae-ro-i
blow-3SG.M
-PRES
The wind is shaking the tree because it is really blowing.

221
In addition to the roles of Agent and Instrument, some authors have also postulated a role
of Cause or Reason, which differs from the thematic role of Agent to the extent that it is not
necessarily human, volitional, or intentional and its involvement in the situation is less direct
(Frawley, 1992). In Rotokas, Cause or Reason typically takes the form of an oblique argument
marked by the postposition iava, as illustrated in (720) and (721).
(720) vuri-vira
bad-ADV
kovu-to
stomach-SG.M
siovo-a
feel-1SG
aue
CONN
iava
POST
atu
ying fox
siupu
soup
oa
RPRO.3.SG.N
aio-a-vo
eat-1SG
-IP
I feel bad in the stomach from the ying fox soup that I ate.
(721) oira-to
man-SG.M
ora-karekare-pa-ro-i
RR-scratch-CONT-3SG.M
-PRES
veruveru
grille
iava
POST
The man is scratching himself because of the grille (skin disease). [Firchow (1984)]
There is some exibility in the grammatical realization of Cause or Reason, and it is similar
to that of natural forces to the extent that it can also serve as the subject of a bivalent clause, as
in (722) or (723).
(722) kapu-a
sore-SG.N
Pita
Peter
upia-pie-pa-i-voi
pain-CAUS-CONT-3PL
-PRES
uva
and
gau-pa-re-voi
cry-CONT-3SG.M
-PRES
The sore is causing Peter pain and he is crying.

(723) sitoka-irao-pa-ra-i
intense pain-INTEN-CONT-1SG
-PRES
kapu-a
sore-SG.N
iava
POST
Im in intense pain from the sore.
The postulation of a thematic role of Agent and the identication of it with particular verb
stems does not provide a good account of the distribution of verbal inection, either in bare verb
roots or derived verb stems. Although the derivation of a bivalent verb stem from a monovalent
verb root involves the introduction of a causer role which could be characterized as Agent,
and bivalent verb stems uniformly take inection, the direct identication of an agentive role
with inection is problematic, given that the Agent role is not uniformly associated with
inection in at least three cases. First, as seen in the previous section on the semantic basis
of split intransitivity, the monovalent verb roots associated with inection are not uniformly
associated with an agent role. Second, reexive/reciprocal constructions involve an Agent but
nevertheless show inection, as illustrated in (724).
(724) riako-rirei
woman-
ora-upo-ere-i-e
RR-ght-3DL.F-EPEN-IP
oira-toa=pa
man-SG.M=BEN
uva
and
vairei
PRO.DL.F
rite-pie-i-voi
stop-CAUS-3PL
-PRES
oira-ra
man-HUM.PL
The two women fought because of the man, and the men stopped them.
222
Finally, noun incorporation reveals that the presence of an Agent role does not by itself
determine the form of verb inection, since subject agreement for the actor role takes the form
of inection in a simple transitive clause but inection in a transitive clause with noun
incorporation. For example, both (725) and (726) involve an animate, voltional actor, but (725)
shows inection whereas (726) show inection.
(725) Agiosi
Agiosi
aako-va=re
mother-SG.F=ALL
tara-pa-e-vo
look for-CONT-3SG.F
-IP
Agiosi is looking for (his) mother.

(726) eto
re
tara-pa-ro-e
look for-CONT-3SG.M
-IP
Siape
Siape
oisio ra
COMP
eto kasi-ro
re-PL.CL
Siape is looking for re in order to make a re.
In the case of both reexive/reciprocal construction and noun incorporation, the properties
of the undergoer (patient/theme) are relevant to the alternation between and inection,
suggesting that it is a more likely locus of explanation for the alternation between and
inection.
10.2.2 Theme/Patient
The roles of theme and patient have been dened inconsistently in the literature. Although
the two terms are frequently used interchangeabley, some authors distinguish between them on
the basis of animacy: patients are human whereas themes are either non-human (animals) or
inanimate (Andrews, 2007:140). The issue is not entirely terminological, since there is real
debate concerning the nature of the theme/patient in a prototypical transitive situation (Naess,
2007). In English, for example, a wide variety of semantic roles are associated with objecthood,
as demonstrated by the various example sentences in (727) (Levin, 1999).
(727) The engineer cracked the bridge. [patient]
The engineer destroyed the bridge. [patient/consumed object]
The engineer painted the bridge. [incremental theme]
The engineer moved the bridge. [theme]
The engineer built the bridge. [effected object/factitive]
The engineer washed the bridge. [location/surface]
The engineer hit the bridge. [location]
The engineer crossed the bridge. [path]
The engineer reached the bridge. [goal]
The engineer left the bridge. [source]
The engineer saw the bridge. [stimulus/object of perception]
223
The engineer hated the bridge. [stimulus/targt or object of emotion]
Although it may be worthwhile to distinguish between theme and patient, the distinction
does not appear to be particularly relevant as far as the surface coding properties of Rotokas are
concerned. Verbs that involve a change-of-state in an affected object behave similarly to verbs
that do not, and objecthood encompasses a number of different semantic roles, as illustrated by
(728) through (733).
(728) erako
rewood
turu
CLASS
pero-re-vo
split-3SG.M
-IP
Rausiere
Rausiere
torara=ia
axe=LOC
Rausiere split the rewood with an axe. [patient]
(729) Kavato
Kavato
aakova=re
mother=ALL
kasipu-pa-oro
angry-CONT-DEP.SIM
itoo
banana
kovo
garden
teki-re-vo
destroy-3SG.M
-IP
Kavato, angry with his mother, destroyed the banana garden. [patient/consumed object]
(730) Leo
Leo
kepa
house
kopuasi-pie-re-vo
restore-CAUS-3SG.M
-IP
va
PPRO.3.SG.N
kipu-oro
paint-DEP.SIM
uva
and
vearo
good
keke-pa-i
look-CONT-PRES
Leo restored his house by painting it, and it looks good. [incremental theme]
(731) Ravi
Ravi
kepa
house
pau-re-voi
build-3SG.M
-PRES
vo-avukarei=pa
SPEC-couple=BEN
ra
and
va=ia
PPRO.3.SG=LOC
uusi-pa-si
sleep-CONT-3DL.M
Ravi is building a house for the couple so that they sleep in it. [effective object/factitive]
(732) varo-ara
clothing-PL.N
sisiu-ve-vo
wash-1DL-IP
vokipaua
morning
We washed the clothes in the morning. [location/surface]
(733) topu-a
hole-SG.N
keke-pa-a-voi
look at-CONT-1SG
-PRES
kepa
house
siovara=ia
inside=LOC
Im looking at a hole inside of the house. [stimulus/object of perception]
Although a number of potentially distinct semantic roles are found in (728) through (733),
there are a few that are systematically absentnamely, source, location, and goal. In Rotokas,
these spatial roles are realized as oblique arguments rather than direct objects. There is one
apparent exception, and this is the construction illustrated in (734) and (735), where an inalien-
able possessor is encoded obliquely and an inalienably possessed body part (which could be
construed as a location) serves as a core argument.
224
(734) kaakau
dog
iava
POST
porike
tail
ua
CL
toe-re-vo
cut-3SG.M
-IP
Poro
Paul
uvare
because
kookotu
chicken
kaviru-e-vo
steal-3SG.F
-IP
rera
PPRO.3.SG.M
oira-aro
PPRO.3.SG.F-POSS
Paul cut the tail of the dog because he bit his chicken.
(735) keetaa
jaw
oirato
man
iava
POST
gasi-i-vo
break-3PL
-IP
ora-upo-pa-oro
RR-ght-CONT-DEP.SIM
vokiaro
night
They broke the mans jaw while ghting at night.
However, examples such as (736) and (737) show that it is not the location encoded as
theme in this construction, but rather the affected part of an inalienable possessor, the oblique
argument of the postposition iava.
(736) Vepo
Vepo
koie
pig
iava
POST
arevuo-to
tongue-SG.M
ori-re-vo
cook-3SG.M
-IP
Vepo cooked pig tongue.

(737) Tovariri
Tovariri
votokara
car
iava
POST
taea
tire
goru-pie-re-voi
strong-CAUS-3SG.M
-PRES
Tovariri strengthened the tire of the car.

The more general nature of this construction, and its existence outside of the context of a
transitive clauses, is further illustrated in (738) and (739), where the subject of a monovalent
verb is an inalienably possessed body part: the monovalent verb kapua have sore has the body
part noun gisipo mouth as subject in (738) while the monovalent verb kata be exhausted
has the body part noun vara ua body as subject in (739)
(738) gisipo
mouth
ragai
PPRO.1.SG
iava
POST
kapua-o-i
have sore-3SG.F
-PRES
uvare
because
tavute
mango
isi
CL
aio-a-voi
eat-1SG
-PRES
kopu-pa
unripe-DERIV
isi
CL
My mouth is sore because I ate a red mango.
(739) ragai
PPRO.1.SG
iava
POST
vara
body
ua
CL
kata-piro
exhaust-COMPL
uvare
because
riro
big
kaekae-vira
long-ADV
voka-a-vo
walk-1SG
-IP
My skin was exhausted because I walked a long way.

10.2.3 Experiencer
The term experiencer is used to describe a number of semantic roles relating to predicates of
thought, belief, perception, and emotion. In Rotokas, the experiencer is systematically encoded
225
as subject; however, the stimulus is encoded as direct object for some verbs and as an oblique
argument for others.
In verbs of perception, the subject corresponds to the experiencer and the direct object to
the stimulus, as illustrated for the verb root vura look, see in (740) and (741) and the verb
root siovo feel in (742) and (743) (see 9.1.1.2).
(740) ora-ruvu-ro-e
RR-jump-3SG.M
-IP
uvare
because
rakoru
snake
vura-re-vo
see-3SG.M
-IP
He jumped because he saw the snake.

(741) kokopuo-vira
distant-ADV
rutu
very
Tokii
Mt Bagana
vura-pa-a-voi
look at-CONT-1SG
-PRES
Im looking at Mt. Bagana from afar.

(742) uteo-va
cold-SG.F
siovo-a-vo
feel-1SG
-IP
vokiaro
night
usii-pa-oro
sleep-CONT-DEP.SIM
I felt cold sleeping at night.
(743) araiva-vira
clear-ADV
rutu
very
vii
PPRO.2.SG
uvu-pa-a-voi
hear-CONT-1SG
-PRES
ovusia
while
reo-pa-u
talk-CONT-2SG
I hear you clearly (g., understand) when you talk.

Verbs of perception show a different valency pattern than other psychological verbs (psych-
verbs). In verbs of perception, the stimulus plays the role of object while in other pysch-verbs
the stimulus plays the role of oblique argument. For example, the verb roots ruipa desire, want
and vasiare dislike encode the stimulus as an oblique argument marked by the postpositional
enclitic =pa, as in (744) and (745).
(744) oari=pa
DEM.DIST.SG.F=BEN
ruipa-pa-ra-i
desire-CONT-1SG
-PRES
riako-va
woman-SG.F
I like that woman.
(745) riako-va
woman-SG.F
oira-to=pa
man-SG.M=BEN
vasiare-pa-o-e
dislike-CONT-3SG.F
-IP
oa iava
hence
oira
PPRO.3.SG.F
toe-re-vo
cut-3SG.M
-IP
The woman disliked the man and thats why he cut her.
The verb root kasipu angry also encodes the stimulus as an oblique argument, but it selects
for the postpositional enclitic =re, as illustrated in (746).
(746) Rupi
Rupi
ravuru-vira
clustered-ADV
rutu
very
pirati
peanut
pau-o-e
plant-3SG.F
-IP
uva
and
aako-va
mother-SG.F
oira=re
PPRO.3.SG.F=ALL
kasipu-o-i
angry-3SG.F
-PRES
Rupi peanut-planted in heaps and her mother was angry with her.
226
10.2.4 Source and Goal
The thematic roles of source and goal are rooted in the semantics of verbs of motion. Loosely,
the source and goal can be dened as the start and end points, respectively, of a motion event.
(The notion of source and goal have been extended in the literature to encompass roles that
are not anchored to the semantics of motion events, but here we will adhere to a more strict
interpretation of the terms.)
Source and goal in Rotokas are oblique arguments (see 5.1.3). The most common marking
takes the form of postpositional enclitics. The enclitic =iava is used to case mark sources, as
illustrated by (747) and (748).
(747) Ibu
I.
iava
POST
aapaapau
visit
rovo-ro-epa
precede-3SG.M
-RP
He came rst from Ibu to visit.

(748) Rarasori
Robinson
Rotokas
Rotokas
reo
language
porepore-pie
turn.RDP-CAUS
oa
RPRO.3.SG.N
urio-ro-era
come-3SG.M
-DP
Averika
America
iava
POST
Robinson came from America to translate the Rotokas language.
The enclitic =iare is used to mark goals of varying sorts, as illustrated in (749) through
(751).
(749) ragai
PPRO.1.SG
rugo-pa-ra-e
think-CONT-1SG
-IP
oisio
COMP
voki-pa-vira
tomorrow
ava-pa-ra-i
go-CONT-1SG
-PRES
Togarao
Togarao
iare
POST
I thought that Id go to Togarao tomorrow.
(750) utupakou-visivi
soon-ADV
urio-pa-ro-i
come-3SG.M
-PRES
Rake
Rake
visii
PPRO.2.PL
iare
POST
Rake is coming to you (pl.) soon.
(751) Tesi
Tesi
vaio
ANIM.DL
ora
and
Sira
Sira
ava-pa-ere-i-ei
go-CONT-3DL.F-EPEN-PRES
toara
market
iare
POST
vovokio
today
Tesi and Sira are going to the market today.
Locations, sources, and goals also co-occur with the particle eisi, as in (752) or (753).
(752) Riopeiri
Riopeiri
arao-rei
brother-DL.M
ora
and
Vaviata
Vaviata
ava-si-e
go-3DL.M-IP
eisi
LOC
Buka
Buka
The brothers Riopeiri and Vaviata went to Buka.
227
(753) aakovatorei
parents
ava-si-e
go-3DL.M-IP
eisi
LOC
Wakunai
Wakunai
uu-sia
meet-DEP.SEQ
The parents went to Wakunai to meet (at market).
The particle eisi differs from the postpositional enclitics that mark source and goal in at least
two respects. First, the particle eisi precedes the goal, as already illustrated in (752) and (753).
Second, the particle eisi is an unbound (free) form, given that it also occurs alone, as illustrated
in (754).
(754) aruvea
yesterday
ava-ro-e
go-3SG.M
-IP
aite-to
father-SG.M
eisi
LOC
vara-vira
down-ADV
aue
CONN
tara-sia
nd-DEP.SIM
atari
sh
Dad went down yesterday to nd sh.
The particle eisi is in some cases marked by the postpositional enclitics used to mark source
and goal. In other words, if an enclitic occurs, it takes as its host the particle eisi rather than the
noun that plays the role of location, source, or goal, as illustrated in (755) and (756).
(755) asia-pa-ra-i
disinclined-CONT-1SG
-PRES
utu-arapa
follow-DEP.NEG
eisi-re
LOC=ALL
kovo-a
garden-SG.N
I dont want to come along to the garden.
(756) ora-tuguru-ra-e
RR-bump-1SG
-IP
aveke=ia
aveke=LOC
vavo
there
kare-pa-oro
return-CONT-DEP.SIM
eisi=va
LOC=COM
Togarao=ia
Togarao=LOC
vokiaro
yesterday
I bumped my leg on a stone there while returning from Togarao last night.
10.2.5 Recipient/Addressee
One use of the enclitic =pa can be described as benefactive, in the sense that it marks a NP as
being the recipient of an action. It is unclear at present whether its use is licensed by particular
verbs or whether it can appear anywhere that it is semantically felictious.
(757) Matari
Matari
uraura-re-vora
photograph-3SG.M
-DP
Rarasori
Robinson
uva
and
rera=pa
PRO.3.SG.M=BEN
va
PRO.3.SG.N
vate-re-vora
give-3SG.M
-DP
Robinson photographed Matan and gave it to him.

Sometimes lumped together with the role of recipient, the addressee is also encoded as an
oblique, although it does not occur with the enclitic =pa but rather with the enclitic =re, as
illustrated for three different verbs of speaking: pura say, tavi tell, and reo speak.
228
(758) apeisi
what
pura-u-e
say-2SG
-IP
Raki
Raki
ragai=re
PPRO.1.SG=ALL
Raki, what did you say to me?
(759) ragai
PPRO.1.SG
tare-raga-pa-a-vo
wait for-only-CONT-1SG
-IP
aako=re
mother=ALL
raivaro
road
uvare
because
ragai
PPRO.1.SG
tavi-e-vora
tell-3SG.F
-DP
I waited in vain for my mother on the road because she told me.
(760) Riki
Riki
tavavaia-pie-i-voi
frustrated-CAUS-3PL
-PRES
rera=re
PPRO.3.SG.M
reoreo-pa-oro
speak.RDP-CONT-DEP.SIM
They caused Ricky to be frustrated while talking to him.
10.2.6 Conclusion
In this section (10.2), a number of traditionally recognized semantic roles were examined in an
attempt to determine the extent to which they are able to predict the occurence of and verbal
inection. Although the notion of a thematic role is somewhat useful, to the extent that it allows
higher-level generalizations about grammatical roles, it fails to account for the distribution of
and verbal inection. For example, while the notion of a thematic role may help explain
why a Perceiver associated with a monovalent verb takes inection while a Feeler associated
with a monovalent verb takes inection, it does not exlain why an agent takes inection
for subject agreement when associated with some verbs of motion (e.g., ava to go) but
inection for others verbs of motion (e.g., voka to walk).
The same conclusion has been reached in the study of split intransitivity in other languages.
For example, in her discussion of split intransitivity in Dutch, Zaenen (1988:332) observes that
notions like theme and agent are not primitive terms, and it is not reasonable to
expect that empirical studies of natural language will ever lead to a universal def-
inition. But in practice there is the temptation to assume that they provide a basis
for crosslinguistic comparison of the meaning of lexical items. As our discussion
indicates, their use is in fact likely to lead to confusion.
The inadequacy of thematic roles in accounts of split intransitivity is part of a long-standing
skepticism in the theoretical literature concerning the explanatory adequacy of thematic roles
(Dowty and Ladusaw, 1988; Dowty, 1989, 1990; van Voorst, 1988; Jackendoff, 1988), which
goes beyond the scope of this work (see Levin and Hovav (2006) for a survey). A few short-
comings of the approach merit discussion in the context of Rotokas.
As already seen in 10.2.2 from the discussion of the thematic roles of Theme or Patient,
there are issues concerning how thematic roles are dened and what is the appropriate grain-
size (Levin and Hovav, 2006:38-41). For example, the thematic roles of Agent and Patient
229
are present in most inventories but they are not uniformly dened. Some authors conate the
notions of theme (inanimate undergoer) and patient (animate/human undergoer) while others
distinguish the two. The distinction will be relevant for some languages sensitive to the animacy
of the undergoer but not for others. Similarly, some authors carefully distinguish a volitional,
human causer from a natural force. In Rotokas, the granularity of the roles of subject and object
are not relevant for assignment of the subject role (i.e., human actors and natural forces serve
equally well as subjects), but for other languages (e.g., Usan) more ne-grained distinction will
be necessary.
There is long-standing recognition that adherence to the strict one-to-one mapping between
thematic roles and grammatical arguments is problematic, since there are many cases where an
argument appears to play more than one thematic role in a clause (Yip et al., 1987). This dif-
culty has been avoided in some proposals by positing more ne-grained roles that essentially
involve a combination of rolese.g., affected agent, which combines the role of Agent with a
property typically associated with the role of Patient (Naess, 2007). A proliferation of thematic
roles weakens the explanatory power of the theory and suggests a fundamental problem with
the approach.
10.3 Split Intransitivity from a Theoretical Perspective
Split intransitivity is generally dened as a phenomenon where intransitive verbs are heteroge-
nous with respect to their grammatical behavior, typically such that one subclass of intransitive
subjects behaves like transitive subjects while another subclass of intransitive subjects behaves
like transitive objects. Using this fairly broad denition of the term, split intransitivity en-
compasses a number of phenomena described using different terminology in the literature, such
as split ergativity (Dixon, 1979), case marking splits (Tsunoda, 1981), active-inactive
(Danziger, 1996), or active-stative alignment (Mithun, 1991).
Using the grammatical primitives of S, A, and O (Dixon, 1979; Andrews, 1985; Dixon,
1994), the various possibilities for the alignment of grammatical roles can be represented dia-
grammatically as in Table 10.6 (see 7.3.2).
Nominative-Accusative Ergative-Absolutive Tripartite Split Intransitivity
A O
S
A O
S
A O
S
A O
S
Table 10.6: Alignment Possibilities for S, A, and O
230
Dixon (1994) draws a useful distinction between two types of split-intransitivity, split-S
systems and uid-S systems:
Languages that distinguish between S
a
and S
o
, as subtypes of S, are of two kinds.
The rst kind are like ergative and accusative languages in having syntactically
based marking of core constituents [...] Each verb is assigned a set syntactic frame,
with case marking or cross-referencing always being done in the same way, ir-
respective of the semantics of a particular instance of use. We call such a system
split-S. The second kind employs syntactically based marking for transitive verbs,
but employs semantically based marking [...] just for intransitive verbs an intran-
sitive subject can be marked as S
a
(i.e., like A) or as S
o
(like O) depending on the
semantics of a particular instance of use. We can call this a uid-S system.
The difference between these two systems has to do with the degree to which the alignment
of S with either A or O is exible. In a split-S system, the class of intransitive verbs is split
between the two subclassesi.e., a particular intransitive verb is either of the type S
A
or of the
type S
O
. In a uid-S system, however, there is uidity of assignment to the two classes. A
particular intransitive verb can be assigned to either one of the two classes, depending upon the
semantics of individual tokens.
Fluid-S systems appear to be more rare, and clear-cut instances of them are few in number.
They are found in Acehnese (Durie, 1985, 1987), Eastern Pomo (McLendon, 1978), and Tsova-
Tush (Holisky, 1987). Split-S systems are far more common. Dixon (1994:75) observes that,
Careful study of the grammars of split-S languages shows that they do work in terms of a
unitary S category with this being subdivided, for certain grammatical purposes, into S
a
and
S
o
. In Acehnese, it has been argued that grammatical relations of S, A, and O do not exist
(Durie, 1985). Rather, there are simply two semantic categories, Agent and Patient. Dixon
(1994) argues that it is nevertheless still possible to posit a grammatical relation of subject:
It may be that for Acehnese the only viable denition of subject is [Duries] Actor
(the concatenation of A and S
a
, in my terms) which is in fact dened grammatically,
in terms of its cross-referencing properties, but it is a grammatical category with a
relatively simple and unusually consistent semantic characterisation.
Although the approach advocated in Dixon (1979, 1994) is attractive from a purely descrip-
tive standpoint, it leaves a number of important theoretical issues unresolved. First, the syntactic
status of S, and any subclasses of it, is left unclear. Although Dixon argues for a unitary S that
is split into subclasses in some languages (S
a
and S
o
), it is also possible to analyze the phe-
nomenon in terms of two distinct categories that are unied in most languages but distinguished
in others (unergative and unaccusative). Second, Dixons account is largely unconstrained as
far as the semantic motivation of S
a
and S
o
is concerned. Although Dixon focuses on the se-
mantic notion of control, other parameters have been proposed in the literature (e.g., aspect
231
in Van Valin Jr. (1990)), and it is worth considering whether there are universal constraints on
the relevant parameters and their interaction. In the literature on split intransitivity, these two is-
sues have received considerable attention, with theoretical proposals essentially falling into two
camps: those that argue in favor of a semantic account of split intransitivity and those that deny
any such basis can be found and urge a purely syntactic account of the phenomenon. These two
approaches will be contrasted in the following sections and the signicance of Rotokas in this
debate will be considered.
10.3.1 Syntactic Accounts of Split Intransitivity
The formulation of the Unaccusative Hypothesis (UH) in Perlmutter (1978) has motivated a
great deal of theoretical interest in split intransitivity. It embodies two claims. First, intransitive
verbs fall into two classes: unaccusative and unergative. The single argument of an unaccusative
verb is an underlying direct object and displays many of the same syntactic properties. Second,
the distinction is syntactically represented but semantically motivated: unergativity correlates
with agentivity and unaccusativity with patienthood.
2
For example, in Italian, verbs take one of two auxiliary forms: either avere have or essere
be. Transitive verbs occur with avere, as in (761), while derived intransitives occur with
essere, as in (762) and (763).
(761) Mario
Mario
ha
has
difeso
defended
Luigi
Luigi
Mario defended Luigi [Rosen (1984:43)]
(762) Mario
Mario
si
RR
` e
is
difeso
defended
Mario defended himself. [Rosen (1984:44)]
(763) Mario
Mario
si
RR
` e
is
concesso
conceded
un
a
momento
moment
di
of
riposo
rest
Mario allowed himself a moments rest. [Rosen (1984:44)]
2
This hypothesis was originally couched within the framework of Relational Grammar (RG) and was meant
to account for the fact that languages differ with respect to the ability of intransitive verbs to form impersonal
passives by appealing to initial grammatical relations. According to the hypothesis, intransitive verbs were split
according to their underlying derivational source: one class of intransitives, the unergatives, were proposed to have
an initial 1, while another class of intransitives, the unaccusatives, were proposed to have an initial 2. Although
RG is no longer at the center of theoretical attention and has few practitioners, the split between two classes of
intransitives has become widely recognized and continues to the source of theoretical interest since a variety of
grammatical phenomena have been found that recognize the distinction and the basic insight has been adapted to
other grammatical frameworks, particularly multistratal theories that posit a distinction between an underlying and
surface representation (deep versus surface).
232
Underived intransitive verbs are split into two classes: some intransitive verbs occur with
avere, as in (764), while others occur with essere, as in (765).
(764) Mario
Mario
ha
has
esagerato
exaggerated
Mario exaggerated. [Rosen (1984:44)]
(765) La
the
pressione
pressure
` e
is
aumentata
increased
The pressure increased. [Rosen (1984:44)]
Since its original formulation, a constant thread in the literature on the UH is debate over the
extent to which the phenomenon is amenable to a purely semantic explanation. Rosen (1984)
examines split intransitivity within the framework of Relational Grammar (RG) and explicitly
rejects a purely semantic account as part of a wider claim concerning the need for grammatical
relationals in syntactic representation. More specically, she claims that there is no regular
homomorphism between semantic representation and initial GRs [Grammatical Relations], that
one cannot equate these two concepts, and that frameworks which do are necessarily inade-
quate (p. 38-39). This claim is assessed by examining, and ultimately rejecting, two specic
hypotheses concerning the relationship between syntax and semantics in particular languages
and cross-linguistically.
The rst hypothesis is labelled the Little Alignment Hypothesis (LAH) in Rosen (1984:53).
It is little to the extent that it applies only to individual languages and therefore represents a
weaker claim that does not presuppose that whatever semantic basis found in a particular lan-
guages generalizes more widely. It is provided in (766).
(766) For any one predicate in any one language, there is a xed mapping which aligns each
semantic role with an initial GR. The alignment remains invariant for all clauses with that
predicate.
The second hypothesis is labelled the Universal Alignment Hypothesis (UAH) in Rosen
(1984:40), since it represents a stronger claim that goes beyond particular languages to make a
cross-linguistic claim of putative universality. It is provided in (767).
(767) There exists some set of universal principles on the basis of which, given the semantic
representation of a clause, one can predict which initial GR each nominal bears.
In order to evaluate (766) and (767), Rosen (1984) examines a handful of languages: Choctaw,
Dutch, Italian, Lakhota, Sanskrit, and Turkish. Rosen (1984) points out two problems for the
UAH that are raised by these languages.
First, Rosen (1984) claims that even individual languages are not internally consistent with
respect to the alignment of semantic roles and accusativity, undermining the claim for language-
specic alignment in (766). For example, Rosen (1984:53) cites the following pair of sentences
233
in Italian, observing what is the typical patternnamely, that when the subject of a transitive
verb corresponds to the subject of an intransitive verb, the same auxiliary is selected, as in
shown in (768), and that when the object of a transitive verb corresponds to the subject of an
intransitive verb, different auxiliaries are selected, as shown in (769).
(768) a. Il
the
pubblico
audience
ha
has
schiato
booed
il
the
tenore.
tenor
The audience booed the tenor.
b. Il
the
pubblico
audience
ha
has
schiato
booed
The audience booed.
(769) a. Bertini
Bertini
ha
has
calato
lowered
il
the
sipario.
curtain
Bertini lowered the curtain.
b. Il
the
sipario
curtain
e
is
calato.
lowered
The curtain fell.
But Rosen (1984:53) observes that this pattern is not consistently maintained in Italian.
Other verbs show the opposite pattern: when the subject of a transitive verb corresponds to
the subject of an intransitive verb, different auxiliaries are selected, as in shown in (768), and
when the object of a transitive verb corresponds to the subject of an intransitive verb, the same
auxiliary is selected, as shown in (769).
(770) a. Aldo
Aldo
ha
has
fuggito
ed
ogni
all
tentazione.
temptation
Aldo ed all temptation.
b. Aldo
Aldo
e
is
fuggito.
ed
Aldo ed.
(771) a. Bertini
Bertini
ha
has
deviato
deected
il
the
colpo.
blow
Bertini deected the blow.
b. Il
the
colpo
blow
ha
has
deviato.
deected
The blow went awry.
234
Second, verbs that are supercially similar in meaning show different behavior across lan-
guages, undermining the claim for universal alignment. To illustrate this point, Rosen (1984:61)
observes that die is unergative in Choctaw but unaccusative in Italian, as shown in (772), and
sweat is unaccusative in Choctaw and unergative in Italian, as shown in (773).
(772) a. illi-li-tok
die-1-POST
kiyo
not
I did not die. [Choctaw]
b. non
not
sono
be+1.SG
morto
died
I did not die. [Italian]
(773) a. sa-laksha
1-sweat
I sweated. [Choctaw]
b. ho
1.SG
sudato
sweated
I sweated. [Italian]
The arguments marshalled by Rosen (1984) against a semantic account of split intransitivity
are valid, but they address a relatively simplistic semantic analysis that is couched in terms of
xed thematic roles (Agent, Patient). However, as already seen in 10.2, the analysis of verbal
semantics in terms of thematic roles is deeply awed, and more ne-grained analyses of event
semantics have evolved in response to the limitations of such an approach, as pointed out in Van
Valin Jr. (1990:253):
When semantic theories of split intransitivity are discussed by proponents of the UH
[Unaccusative Hypothesis], they are normally characterized as simplistic thematic-
relations-based accounts, e.g., the subject of class-S
A
verbs is always an agent,
while that of class-S
O
verbs is always a theme/patient. The semantic variation in
the split intransitivity argues strongly against any theory of these phenomena based
entirely on thematic relations, since, for example, some of the subjects of class-
S
O
verb in Italian are clearly agentive [...] while none of the class-S
O
subjects in
Acehnese are. This is a signicant point, because the arguments in Rosen 1984
regarding the impossibility of an adequate semantic characterization of split intran-
sitivity are directed against a very simplistic thematic-relations analysis [...]
It is possible to provide alternative semantic accounts that do not suffer from these problems,
and, if these theories provide insights or empirical generalizations that the UH fails to capture,
they are to be preferred. In the following section, the semantic accounts of split intransitivity
will be reviewed and evaluated.
235
10.3.2 Semantic Accounts of Split Intransitivity
One of the earliest crosslinguistic surveys of the semantic basis of split intransitivity is that of
Merlan (1985), which examines a handful of languages and draws a number of broad conclu-
sions (Merlan, 1985:350):
smaller class restricted to animates The specialized intransitive lexical subclass will con-
tain (with few or no exceptions) verbs requiring animate subject; the distributionally un-
marked intransitive class(es) will not be unitarily speciable as to animacy of the subject.
verbs of bodily function and process The specialized intransitive lexical class will contain
some verbs of bodily function and process...
subjective inection associated with agentivity If the specialized intransitive class requir-
ing animate subject is coded by subjective inectional forms, verbs in the class will be
composed partly, perhaps principally, of verbs in which the semantic relation of NP to
verb is agentive to neutral...
objective inection associated with patienthood If a language marks the specialized intran-
sitive class requiring animate subject by means of object pronominals, the verbs contained
within it will be principally of a kind to which the subject has a netural to patientive rela-
tion.
verbs of physical sensation and perception have objective tendencies Some verbs of phys-
ical sensation and perception are likely to be within the objective class.. (p. 350-351)
Finally, Merlan (1985) makes a fairly strong claim concerning the expected alignment of
semantic and grammatical roles:
No languages will be found in which the restricted class is objectively inecting
and the verbs in it are primarily of the kind in which the subject bears an agentive
relation to the verb. Nor will the reverse situation be found, in which a smaller class
of subjectively inecting intransitive contains verbs for which the semantic relation
of the subject to the verb is primarily patientive. (p. 350)
This claim is formulated in such a way that it begins to address some of the objections
to a semantic account of split intransitivity raised in Rosen (1984), since it does not assert
a direct relationship between semantic roles and intransitive verb classes, but rather places a
markedness constraint on the relationship, such that particular alignments are more natural than
others. However, the vagueness of the terms agentive or patientive makes it very difcult to
judge the extent to which these generalizations hold true. In other words, before a semantically-
oriented theory of split intransitivity can be provided, a more explicit account of the semantics
of agent and patient (among other categories) must be worked out. Other authors who have
examined split intransitivity cross-linguistically have spelled out more explicitly the semantic
236
features involved in these systems, going beyond the loose characterization of particular classes
agent-like and patient-like in Merlan (1985). One particularly clear account that illustrates
a few of the issues involved and sets the stage for a discussion of more explicit theoretical
accounts of the phenomenon is Mithun (1999).
Mithun (1999) discusses split intransitivity in three languages of the Americasnamely,
Guarani (colloquial), Lakhota, and Central Pomo. The patterns of case-marking found in these
languages resembles the intransitivity split found in Italian. In Lakhota, for example, the subject
of some intransitive verbs display the same person agreement as the subject of transitive verbs
(e.g., the verb meaning to jump, as in (774)) while the subject of some other intransitive verbs
display the same agreement as the object of transitive verbs (e.g., the verb meaning to be sick,
as in (775)) (Mithun, 1991:514).
(774) a. wa-ps ca
1.SG-jump
I jumped.
b. wa-kt ekte
1.SG-kill
Ill kill him.
(775) a. ma-k
h
u ze
1.SG-sick
Im sick.
b. ma-kt ekte
1.SG-kill
Hell kill me.
Mithun (1999) proposes that the verbs in the languages she discusses can be divided into
classes on the basis of their specication for a number of semantic features, listed below:
Event This features refers to the aspectual classication of a predicate, following the widely
recognized distinction between events and states discussed by Vendler (1967). The rel-
evance of aspect to split intransitivity is widely recognized in the literaturee.g., the
discussion of aspect and punctuality as parameters of semantic transitivity in Hopper and
Thompson (1980) (see 7.3.1).
P/E/I The notion of agency is characterized by Mithun in terms of the entity understood to be
the performer, effector, and/or instigator of an action. The notions of performance, effect,
and instigation are largely taken for granted and are not spelled out explicitly.
237
Control Another aspect of agency is the notion of control. The notion of control is also brought
up in Dixon (1994), who denes control in terms of the semantic role that is most rele-
vant to the success of the activity: the semantic role of a verb which is most relevant to
the success of the activity (if human: which could initiate or control the activity) is linked
to A function; and that role which is most saliently affected by the action is linked to O
relation (Dixon, 1994:29)
Affectedness The last feature refers to the affectedness of the intransitive actor. This feature
has been a recurrent theme in the literature of transitivity, and is considered by some to be
the sine qua non of semantic transitivity. Unlike the other features, which are orthogonal
to one another, this feature is applied only to stative verbs by Mithun (1999).
The combination of these features identies a number of verb classes, which are listed with
their feature analysis in Table 10.7.
Class Illustrative Verbs Event P/E/I Control Affected
a jump, go, run + + + n.a.
b hiccough, sneeze, vomit + + - n.a.
c fall, die, slip + - - n.a.
d reside, be prudent, be patient - + + n.a.
e be tall, be strong, be righthanded - - - -
f be sick, be tired, be cold - - - +
Source: Mithun (1991:524)
Table 10.7: Analysis of Verb Classes By Semantic Features
Although Mithun (1999) does not provide labels for the various congurations assumed in
her analysis, the implicitly recognized verb classes might be characterized as follows:
motion jump, go, run
bodily process hiccough, sneeze, vomit
uncontrolled event fall, die, slip
controlled state reside, be prudent, be patient
inherent property be tall, be strong, be righthanded
affected state be sick, be tired, be cold
In the case of Guarani and Lakhota, she concludes that a single semantic parameter governs
the split: eventhood for Guarani and P/E/I for Lakhota. In the case of Central Pomo, however,
she concludes that there are two parameters at play: Control and Affectedness, with Affected-
ness being relevant only where Control is absent, as shown in (776).
238
(776) Control
+
S
A
Affectedness
S
A
+
S
O
There are a fewpoints to make concerning Mithuns analysis. First, the features discussed by
Mithun are not completely independent of one another, especially performance/effect/instigation
and control, which are two facets of a broader notion of agency. This may explain why Mithun
does not discuss all of the logically possible combinations of these features. For example,
Mithun does not discuss two types of non-event predicates predicted by her features: plus PEI
and minus Control vs. minus PEI and plus Control. Second, it is unclear how these features
interrelate. The feature of Affectendess is invoked only in the analysis of Central Pomo, but is
ignored for the other languages. Finally, Mithuns classication of verbal predicates in terms
of events versus non-events is fairly coarse, and most studies of event structure posit more ne-
grained distinctionse.g., the four-way classication of activities, accomplishments, achieve-
ments, and states in the classic study of Vendler (1957).
Some of the parameters identied in Mithun (1999) are relevant to Rotokas (e.g., control
is arguably relevant to the verbs of bodily process), but there are nevertheless splits found that
do not t into her classicatory scheme. For example, verb roots denoting motion events are
split according to their specication for manner, but this parameter is not found in the inven-
tory discussed by Mithun. While the orientations towards more ne-grained lexical semantic
analysis has merit, an account is needed that addresses some of the specic shortcomings while
preserving the spirit of its intent.
One theory of split intransitivity that shares the orientation towards lexical semantics found
in Mithun (1999) but provides a more sophisticated predicate decomposition is found in Role
and Reference Grammar (RRG) (Foley and Van Valin Jr., 1984; Van Valin Jr., 1984, 1987,
1990; Van Valin Jr. and LaPolla, 1997; Van Valin Jr., 2005). Van Valin Jr. (1990) argues that
the various phenomena which the Unaccusative Hypothesis (UH) is meant to explain are better
understood in semantic (rather than syntactic) terms:
According to the UH there are two types of intransitive verbs, and in both theories
the differences between them are characterized in purely syntactic terms: in one
type the surface object is also the underlying subject, and in the other the surface
subject is the underlying direct object. (Van Valin Jr., 1990:221)
RRG postulates a direct linking between semantic and syntactic representations, analyzing
the former in terms of a lexical semantic theory that involves predicate decomposition in the
239
form of logical structure, following Dowty (1979). Thematic roles are generalizations over
logical structure (LS), dened in terms of logical operators, as shown in Figure 10.8.
3
I. STATE VERBS
A. Locative be-at(x,y) x = locative, y = theme
B. Nonlocational
1. State or condition predicate(x, (y)) x = patient
2. Perception see(x, (y)) x = experiencer, y = theme
3. Cognition believe(x, (y)) x = experiencer, y = theme
4. Possession have(x, (y)) x = locative, y = theme
5. Attributive/Identicational be(x, (y)) x = locative, y = theme
II. ACTIVITY VERBS
A. Uncontrolled predicate(x, (y)) x = effector (y = locative)
B. Controlled DO [predicate (x, (y))] x = agent (y = locative)
Table 10.8: RRG Denitions of Thematic Roles
One aspect of this style of analysis is that it provides a formal mechanism for capturing
the derivational relationship between aspectual classes. As Van Valin Jr. (1990:225) points out,
the activity and accomplishment readings of a verb can be coerced through event type shifting
rules, as in (777), obviating the need to list more than once in the lexicon a verb that admits
both readings.
(777) Activity [motion, creation, consumption] Accomplishment: given an activity LS [
. . .
predicate . . .], add CAUSE [

BECOME predicate . . .] to form a CAUSE accom-
plishment LS
A semantically based account of split intransitivity also explains phenomena that are un-
motivated within a purely syntactic account. In Italian, for example, the verb correre run
behaves both unaccusatively and unergatively, but this variable behavior of the verb reects
two different construals of its semantics, as either an activity or an accomplishment. The verb
behaves unaccusatively (i.e., takes the auxiliary e) when it has an activity reading, but behaves
behaves unergatively (i.e., takes the auxiliary avere) when it has an accomplishment reading, as
illustrated in the contrast between (778a) and (778b) (Van Valin Jr., 1990:237).
(778) a. Luisa
Luisa
ha
has
corso
run
nel
in.the
parco
park
per/*in
for/in
un
an
ora.
hour
Luisa ran in the park for/*in an hour.
3
Van Valin Jr. and LaPolla (1997) provides an even more ne-grained taxonomy than that shown in Figure 10.8.
Since the specic details of the theory are not the primary concern here, only the original formulation is presented.
Those interested in a detailed mapping of the original taxonomy and the current formulation are referred to Van
Valin Jr. (2005:45).
240
b. Luisa
Luisa
e
has
corso
run
nel
in.the
parco
park
per/in
for/in
un
an
ora.
hour
Luisa ran in the park for/in an hour.
Rosen (1984) treats this alternation as idiosyncratic behavior that undermines the semantic
basis for the intransitivity split in Italian, but Van Valin Jr. (1990) shows that it is well-motivated
within a semantic account, and consistent with the behavior of split intransitivity systems in
other languages. But to what extent does this style of analysis work for Rotokas? As was
observed in 10.1.3, Aktionsart appears to be relevant to the split between and inection
among verbs of sound emission. It is less clearly relevant in the case of the other semantic elds
in which the distionction is operative, such as the verbs of inferred causation.
One of the issues that consistently emerges in the theoretical literature on split intransitivity
is whether the split between the two classes of intransitives boils down to a single parameter.
While Van Valin Jr. (1990) shows that split intransitivity is not as unsystematic as Rosen (1984)
contends, the RRG analysis still faces some difculty in providing a well-motivated analysis for
splits that are motivated by multiple factors.
One step in the direction of such an account is that of Zaenen (1993) (Zaenen, 1988; Bresnan
and Zaenen, 1990), which is couched within the framework of LFG (Bresnan and Kaplan,
1982; Bresnan, 2001; Falk, 2001). Zaenen (1993) analyzes Dutch auxiliary selection in terms
of intrinsic argument classication (Levin, 1987; Bresnan and Kanerva, 1988). According to
the theory of instrinsic argument classication, grammatical roles are analyzed in terms of two
features: r and o. The former is shorthand for restrictedness while the latter is shorthand
for object. The fourway classication resulting from the interaction of these two features is
mapped to the grammatical roles of LFG as shown in (779).
(779)
SUBJ r o
OBJ r +o
OBJ
+r +o
OBL +r o
Zaenen (1993) accounts for auxiliary selection in Dutch by anchoring it to the feature of r
with the selection principle provided in (780).
(780) When an intrinsically r marked participant is realized as subject, the auxiliary is zijn.
In order to determine which participant is intrinsicially r marked, Zaenen (1993:150) pro-
poses the basic principle provided in (781).
(781) a. If a participant has more patient properties than agent properties, it is marked r.
b. If a participant has more agent properties than patient properties, it is marked o.
241
In addition to the basic principle, Zaenen (1993:150) notes that the two ancillary assump-
tions provided in (782) are required.
(782) a. If a participant has an equal number of agent and patient properties, it is marked r.
b. If the sole participant of a verb has neither agent nor patient properties it is marked
o.
For the purposes of determining agent and patient properties, Zaenen (1993:147) follows
Dowty and Ladusaw (1988) in attributing the properties listed in (783) with agenthood and the
properties listed in (784) with patienthood.
(783) a. volition
b. sentience (and/or) perception
c. causes event
d. movement
(784) a. change of state
b. incremental theme
c. causally affected by the event
d. stationary (relative to movement of proto-agent)
e. referent may not exist independent of action of verb, or may not exist at all
A detailed assessment of Zaenen (1993) against the facts of Dutch goes beyond the scope of
this discussion, but it is worth pointing out two weakenesses of the account that have relevance
for Rotokas. First, as Zaenen (1993) acknowledges, the list of agenthood and patienthood
factors in (783) and (784) is not exhaustive and it is unclear whether they are in fact the correct
list for Dutch. Second, the algorithm used in the assignment of intrinsic argument classication
relies on a simple tally of the number of factors that accrue to an argument, which assumes
that all factors are equally weighted. However, some factors appear to be more important than
others, not only in Dutch, but also cross-linguistically.
These considerations have led some authors to posit a hierarchy of factors. For example,
Foley (2005) examines split intransitivity in a number of languages in the Austronesian fam-
ily: Acehnese, Tolai, three Maluku languages (Dobel, Larike, and Taba), and the Philippine
languages Kimaragang and Tagalog. To account for the fact that the unergative/unaccusative
division varies across these languages, Foley (2005) proposes a hierarchy of accessibility to
macro-roles, such that unergativity is associated with the verb classes at the top of the hierarchy
and unaccusativity with the verb classes at the bottom of the hierarchy.
242
A O
Actor volitional performer
causing an event or change of state
sentience
movement
stationary
causally affected
incremental theme
Undergoer undergoing a change of state
Table 10.9: Revised Macro-Role Hierarchy from Foley (2005)
The hierarchy of accessibility in Foley (2005) is similar to the Auxiliary Selection Hierarchy
proposed in Sorace (2004).
CONTROLLED PROCESS (NON-MOTIONAL) Selects HAVE (least variation)
CONTROLLED PROCESS (MOTIONAL)
UNCONTROLLED PROCESS
EXISTENCE OF STATE
CONTINUATION OF A PRE-EXISTING STATE
CHANGE OF STATE
CHANGE OF LOCATION Selects BE (least variation)
Table 10.10: The Auxiliary Selection Hierarchy
The hierarchies in Table 10.9 and Table 10.10 provide a means of capturing the alignment
of semantic semantic parameters with syntactic congurations that take into account the fact
that some features are more important than others but it does not provide a mechanism for
accomodating the fact that some factors may be in conict with one another, and languages
may differ in the way that they are ranked in importance. For example, in Rotokas, it was
observed that a number of verb roots denoting processes that produce a result show inection
and a potential explanation for their agentivity is provided by DeLancey (1985). However, verbs
of bodily process that product a visible result, but which are not controllable, such as bleed
or sweat, show inection. Controllability in this case appears to win over inferred
causation (if this is the right characterization), suggesting that the factors relevant to verb
classication in Rotokas should be ranked accordingly.
10.3.3 Conclusion
In this section, the theoretical literature on split intransitivity was reviewed. Two major ap-
proaches to the phenomenon were discussed: the syntactic analysis, which treats split intran-
243
sitivity as a purely syntactic phenomenon, and the semantic analysis, which treats split intran-
sitivity as a semantic phenomenon. While the syntactic analysis typically takes the form of
Unaccusative Hypothesis, the semantic approach takes many forms, usually consisting of some
type of mapping from lexical semantics to morphosyntax via linking rules.
The semantic account of split intransitivity has two distinct advantages in accounting for
the facts of split intransitivity in Rotokas. First, by positing a mapping from lexical semantics
to syntactic behavior, it provides a motiviation for the semantic clustering of monovalent
verb roots observed in 10.1. Although the various semantic elds where the contrast is found
cannot be easily characterized in terms of thematic roles or a single semantic parameter (such as
telicity), they do show some coherence and reect a number of the semantic factors identied in
the literature. Second, the facts of Rotokas are consistent with Van Valin Jr. (1990)s claim that
split-intransitive phenomena provide no evidence of analyzing the subject of class-S
o
verbs as
underlying syntactic object. There is no evidence in Rotokas of any object-like properties for
monovalent verb roots, since there are no known syntactic processes in Rotokas distinguishing
monovalent verbs with inection from those with inection (see, for example, 9).
244
Chapter 11
Conclusion
In the previous chapters of the second part of this thesis, the nature of verbal inection in
Rotokas was systematically described. A preliminary hypothesis was put forward concerning
the relationship between the two forms of verbal inection found in Rotokas and grammatical
roles. Although the evidence from valency-changing derivations generally supported the view
that verbal inection is predictable on the basis of valency, the behavior of bare verb roots
revealed a more complicated picture, due to the existence of split intransitivity. The semantic
motivations of split intransitivity were examined and a partially semantically motivated system
was described, which was sensitive to some of the semantic factors described in the typological
literature. The split between those verb roots that show agreement and those that show
agreement resembles those described for other languages but the similarity found is more of a
family resemblance (Wittgenstein, 1953/2001) than a systematic cross-linguistic parameter.
In 11.1, some directions for future research on Rotokas are spelled out. In 11.2, the theoretical
implications of Rotokas are drawn out.
11.1 Directions for Future Research
There are a fewdirections that future research on the nature of verbal inection in Rotokas might
take: more detailed analysis of tense/aspect/mood (11.1.1); a systematic study of the behavior
of loan verbs (11.1.2); and comparative evidence from other dialects of Rotokas and/or other
languages in the Rotokas family (11.1.3). Each will be discussed in turn.
11.1.1 Tense/Aspect/Mood
Earlier it was concluded that there was no evidence of a single parameter governing whether a
verb stemshows or agreement. It is important to bear in mind that absence of evidence is not
evidence of absence. A detailed investigation of the meaning of the various tense/aspect/mood
distinctions found in the language remains to be done. Firchow (1987) provides little more than
245
an inventory of forms and here the formal properties and basic meaning of those forms are laid
out in 5.2.2.7, but a detailed analysis of their interaction with verb classes (i.e., an inventory of
Aktionsart types) may shed some light on the proper analysis of the two classes of intransitive
verbs.
The split between and inection in the monovalent verb roots denoting sound emission
events remains unexplained, but it is telling that many of the verb roots that show inection
denote event types that can be construed as being bounded, since verbs of achievement (Vendler,
1967) are associated with unergativity in the cross-linguistic literature. Some additional support
for this idea comes from the behavior of noun incorporation. The exception to the rule that bi-
valent verb stems show inection is noun incorporation, as described in 9.2.2. Although the
generalization previously made was that incorporated nouns are non-specic/non-referential, it
is equally true to say that noun incorporation describes non-telic, unbounded eventsthat is,
activities in the Vendlerian classication. If the difference between these two classes proves
to be aspectual in nature, it would provide additional evidence in favor of a systematic seman-
tic basis for split intransitivity in Rotokas as part of a wider generalization concerning verbal
inection.
Unfortunately, the explanation of the distinction between and inection in terms of
Aktionsart and/or aspect remains speculative since my own elicitation work with native speakers
of Rotokas failed to yield an unequivocal test for telicity in the language, such as the distinction
between the prepositions in and for in prepositional phrases (e.g., John breathed the poisioned
air for less than ve minutes versus John ate the hot dog in less than ve minutes). Future
work on the language will have to delve more deeply into the nature of Aktionsart and aspect in
Rotokas.
11.1.2 Loan Verbs
One lingering question concerning the various semantic classes identied in 10.1 is the extent
to which it reects a productive semantic system. Closer examination of the behavior of loan
verbs may help illuminate the issue. Many verbs from Tok Pisin and English are being borrowed
into Rotokas and their classication as or should provide some insight into the nature of the
system. Generally speaking, verb stems borrowed into Rotokas from Tok Pisin show the form
of inection expected given their syntactic behavior. For example, the verb stem iusi to use
(from usim) takes a direct object and shows inection, as illustrated in (785).
(785) ragai
PRO.1.SG
opo
taro
guru-va
leaf-SG.F
iusi-a-voi
use-1SG
-PRES
aue
CONN
ruu-sia
cover-DEP.SEQ
arua
vegetable
tai
CLF
I use taro leaves in order to cover vegetables.
The majority of the verbs borrowed into Rotokas from Tok Pisin are monovalent and show
agreemente.g., rotu attend church, as in (786), or sikuru attend school, as in (787).
246
(786) ragai
PRO.1.SG
Rieko
Rieko
ruvara=ia
near=LOC
pau-pa-ra-e
sit-CONT-1SG
-IP
ovusia
while
rotu-pa-i-e
church-CONT-1PL.EXCL-IP
eisi
LOC
rotu-pa
church-DERIV
kepa
house
I was sitting down next to Rieko while we prayed in church.
(787) vieiasia-to
illiterate-SG.M
Viviere
Viviere
uvare
because
viapau
NEG
sikuru-ro-epa
school-3SG.M
-RP
Viviere is illiterate because he didnt go to school.

Unfortunately, very few of the verb stems borrowed into Rotokas from Tok Pisin fall into
the various semantic classes identied in 10.1. However, the author recalls hearing the English
verbs bounce and ring used as verb stems, as in (788) and (789).
(788) vaunsi-re-voi
bounce-3SG.M
-PRES
It is bouncing.
(789) terepori
telephone
rigi-pa-re-voi
ring-CONT-3SG.M
-PRES
The telephone is ringing.

The occurence of a verb of motion-cum-manner and a verb of sound emission with inec-
tion suggests that the semantic subclasses discussed in 10.1 represent a productive system, but
there are other borrowed verb roots that are not as well behaved. For example, Firchow (1984)
cites a borrowed monovalent verb root politiki to play politics that shows agreement, as in
(790).
(790) politiki-pa-re-voi
politics-CONT-3SG.M
-PRES
He is in politics.
11.1.3 Comparative Evidence
Another line of evidence that may shed light on the nature of the Central Rotokas system of
verbal agreement is comparative analysis of other dialects of Rotokas (Aita, etc.) or other
languages in the Rotokas family. Some preliminary investigation of the Aita dialect was ini-
tiated by the author during his last trip to Bougainville, but this work is in its infancy, having
only established the basic phonological inventory of the Aita dialect (Robinson, 2006). The
only language in the Rotokas family that has been described in print is Konua (also known as
247
Rapoisi), thanks to the vocabulary and grammatical notes of M uller (1954).
1
It would appear
that the distinction between and verbal inection is also found in Rapoisi, judging from the
contrast between the inection of the verbs in (791): the verb root aba go shows one form of
inection while the verb root uri dig shows another (M uller, 1954:???).
(791) a. aba-ra-ea
go-1SG
-PRES
I am going.
b. uri-a-ba
dig-1SG
-PRES
I am digging.
The contrast is also seen in the contrast between verb roots and their causative counterparts.
For example, the causative verb stem uhipie kill shows a different form of inection than the
verb root uhi die, from which it is derived, as shown in (792).
(792) a. uhi-ea
die-PRES
He is dead.
b. uhi-pie-a-ba
die-CAUS-1SG
-PRES
I am killing.
The correspondences between Rotokas and Konua are fairly clear-cut: the form of the rst
person singular is identical in the two languages (-ra for agreement versus -a for agreement)
and the realis present is -ei () or -voi () in Rotokas and -ea () or -vo () in Konua.
More than one form of verbal agreement can be found for a particular verb root in Konua,
and the difference appears to be attributable to valency. In other words, Konua also possesses
labile verb roots (see 9.1.1 for a discussion of labile verbs in Rotokas). For example, M uller
(1954:73/107) contains the verb root sisio wash with both agreement and agreement with
a note indicating that the agreement is associated with transitive usage.
(793) a. sisio-a-ba
wash-1SG
-PRES
I wash.
b. sisio-ra-ea
wash-1SG
-PRES
I wash myself; I am washed.

1
The data in M uller (1954) must be interpreted with caution due to M ullers lack of formal training in linguis-
tics. SIL missionaries linguists working on the language note that M uller (1954) overlooks the allophony between
[n], [l], and [r] and between [b] and [m] (Steve Blewett, pc).
248
Without more information concerning the valency of these verb forms, the data in M uller
(1954) cannot shed a great deal of light on the diachronic origins of the Rotokas system. How-
ever, the existing data suggests that the other dialects of Rotokas and the other languages in the
Rotokas family are likely to possess this distinction, and the differences between the various
systems may shed light on the semantic and syntactic parameters involved.
11.2 Typological Implications of Rotokas
Before delving into some of the theoretical issues raised by the facts described here for Rotokas,
it is worth summarizing the various construction types associated with and inection. The
construction types associated with and inection are listed in Table 11.1.
monovalent verb roots
ambivalent verb roots with a single core argument
monovalent verb stems derived with -ora
monovalent verb stems derived with -piro or -viro
bivalent verb roots with an incorporated object
monovalent verb roots with an incorporated oblique argument
bivalent verb roots
ambivalent verb roots with two core arguments
verbs of perception with a single core argument
bivalent verb stems derived with -pie
Table 11.1: Grammatical Phenomena Associated with and Inection
Is there a single parameter which can account for the split between those constructions that
show inection and those that show inection? The answer would appear to be negative.
The generalization that all verbs with two core argument show inection is complicated by
noun incorporation, which does not show the demotion of a core argument to oblique status
but rather involves some form of tighter integration between the verb root and the incorporated
argument. Furthermore, noun incorporation is not restricted to bivalent verbs but also occurs
with monovalent verbs that take oblique arguments.
From a typological perspective, the form of split intransitivity is somewhat novel in a few
respects. First, to the extent that an account of the assignment of intransitive verb roots to the
two classes of verbal inection can be provided in terms of lexical semantics, it does not appeal
to the usual suspectsi.e., the semantic features typically discussed in the literature on split
intransitivity. Second, there is no identication between one of the two types of subjects and
direct objects. There is no verbal agreement for direct objects in Rotokas and direct objects
show very different patterns of constituent order than subjects, regardless of whether they are
associated with or agreement. The nature of split intransitivity in Rotokas is of typological
249
interest since it undermines the Unaccusative Hypothesis while simultaneously challenging the
various theories concerning the parameters of its semantic basis.
Does Rotokas have grammatical relations? Verbs always take their agreement features from
a single core argument and it therefore makes sense to posit a grammatical relation of subject.
What other grammatical relations need to be posited for Rotokas? Although monovalent verb
roots show differing patterns of verbal agreement some show agreement by default while
others show there are no other syntactic consequences of this division. The evidence from
previous chapters shows that split intransitivity is only skin deep in the sense that it is a mor-
phological phenomenon without deep syntactic implicationsi.e., no major syntactic processes
have been identied which reect the distinction. This is not the case in some split-S languages,
as Dixon (1994) observes. In the Northern Athapaskan language Slave, for example, causatives
can be derived from S
o
but not from S
a
; passives can be derived from S
a
but not S
o
; and noun
incorporation operates on O or S
a
but not on S
o
. However, in Rotokas, causatives can be derived
from either or monovalents (cf. 9.1.2); there is no passive, but the resultative construction
is not limited to a particular valency class (cf. 9.2.3; and noun incorporation operates on direct
objects and oblique arguments of both and verbs (cf. 9.2.2).
Predictions concerning which form of agreement is taken by a particular verb involves fea-
tures of S and O but not of A. Although it is possible to chacterize this as some form of syn-
tactic ergativity, there seems little need to postulate ergative grammatical relations. Rather, a
more modular, monostratal theory of grammar involving interacting constraints is better able
to handle the multiple factors that determine the form of verbal agreement in Rotokas. This is
consistent with the reappraisal of the phenomenon of syntactic ergativity in languages where it
has been argued to exist, such as the Mayan family. Stiebel (2006) analyzes the agent focus con-
struction in the Mayan family using an optimality-theoretic analysis of agent focus (following
Aissen (1999b,a)) and concludes that there is no need to posit distinct grammatical relations,
such as ergative and absolutive, in the syntax of Mayan languages:
The analysis of agent focus presented in this paper also indicates that syntactic
ergativity in Mayan an interpretation of the data that might be invoked by the
separate treatment of the transitive subject in focus, questions and relativization
is just an epiphenomenon of conicting constraints and does not result from a
distinct syntactic representation.
The form of verbal agreement is not always semantically motivated, as was seen earlier
with the aspectual verbs rovo start and opesi nish (see 6.3.2.2), which take their form of
agreement from the bare verb stem with which they co-occur. Although it could be argued that it
is the semantics of the bare verb stem that determines the form of agreement, this phenomenon
argues in favor of some form of syntactic representation for feature sharing, where the feature
in question has one of two possible values: or .
The classication of a verb stem as or is not a property of a verb root by itself, since a
given verb root can show more than one type of inection (as shown by the labile verb roots).
It does, however, appear to be a lexical property, judging from a number of consideration.
250
First, there are a number of verb stems that obligatorily occur with the reexive/reciprocal
marker (see 9.2.1) and cannot occur alonee.g., oravurevure shake, churn, illustrated in
(794). The fact that these stems cannot be derived from their corresponding verb roots means
that they must be considered stand-alone entries in the lexicon.
(794) avaka-va
salt-SG.F
ora-vurevure-pa-o-i
RR-move.RDP-CONT-3SG.F
-PRES
riro-toa=ia
big-SG.M=LOC
kiuvu
wind
The ocean is churning from the big wind, it isnt still.
There also appear to be a few idiosyncratic cases of causative verb stems that do not have
an identiable root and even a few that are monovalentfor example, the verb kakupie shout,
which appears to contain the causative sufx -pie but does not show the properties normally
associated with such derived stemsi.e., it lacks a corresponding root and is monovalent, as
(795) oira-to
man-SG.M
kakupie-pa-re-vo
shout-CONT-3SG.M
-RP
uvavu=va
somewhere=ABL
A man is shouting from somewhere.
(796) Virepa
Virepa
riro-vira
big-ADV
kakupie-re-vo
shout-3SG.M
-IP
ovusia
while
evao-va
tree-SG.F
kove
fall
uvare
because
va
PRO.3.SG.N
toe-re-vo
cut-3SG.M
-IP
koora-toa=va
possum-SG.M=ABL
Virepa shouted loudly when the tree fell because he cut it with a possum (on it).
For example, although there is no verb root uruuru fail to notice, even though there is a
reexive/reciprocal verb stem orauruuru fail to notice, illustrated in (797), and a causative
verb stems uruurupie distract (make fail to notice), illustrated in (798).
(797) ora-uruuru-pa-ra-i
RR-fail to notice-CONT-1SG
-PRES
osia
as
oira-ra
man-HUM.PL
ragai=ia
PRO.1.SG
pute-oro
pass-DEP.SIM
kare-a-i
return-3PL
-PRES
I didnt notice when the men passed by me on their way back.

(798) Rapeasi
Rapeasi
uruuru-pie-pa-re-vo
fail to notice-CAUS-CONT-3SG.M
-IP
Kori
Kori
ovusia
while
Siopi
Siopi
urio-ro-e
come-3SG.M
-IP
Rapeasi
Rapeasi
toe-sia
cut-DEP.SEQ
Kori distracted Rapeasi while Siopi was coming to cut Rapeasi.
251
A similar pattern is observed for other verb roots, such as the hypothetical verb root ruvu
startle, there are nevertheless two verb stems which appear to be derived from itnamely,
a reexive stem oraruvu to be startled, illustrated in (799), and a causative stem ruvupie
startle, illustrated in (800).
(799) ora-ruvu-ro-epa
RR-startled-3SG.M
-IP
Ropi
Ropi
uvare
because
rera
PRO3.SG.M
sita-pie-re-va
surprised-CAUS-3SG.M
-RP
Rausirea
Rausirea
Ropi was startled because Rausirea surprised him.
(800) Sitae
Sitae
ruvu-pie-re-vo
startled-CAUS-3SG.M
-IP
Koka
Koka
ovusia
while
kapu-a
sore-SG.N
iava
POST
upia-pa-o-e
feel pain-CONT-3SG.F
-IP
Koka startled Sitae while he was in pain from a sore.

Verb roots of this sort are in some sense the mirror image of the labile verb roots discussed
in 9.1.1. These verb roots can only function with derivational morphology while labile verbs
can only function without derivational morphology. The existence of these verb roots is impor-
tant, because it underscores the lexical nature of derivational processes. In other words, reex-
ive/reciprocal and causative verb stems cannot be treated as a purely syntactic derivation that
relates two types of sentences, since there is not a base form from which the reexive/reciprocal
or the causative could be derived in the case of some stems, such as the previously-mentioned
verb stems kakupie or orauruuru. Consequently, the two forms of verbal inection cannot be
viewed simply as reexes of a particular syntactic conguration. Similar arguments could be
made on the basis of intransitive verb stems that occur in the reexive/reciprocal or resultative
construction discussed in 9.2.1.
There has been a great deal of debate concerning the proper analysis of split intransitivity,
and whether it is primarily a semantic or syntactic phenomenon. According to the analysis of
Rotokas advocated here, this is a false dichotomy, in the sense that is not really either, since split
intransitivity occurs at the intersection of syntax and semantics, and while both are necessary
elements of a complete account, neither is sufcient.
11.3 Conclusion
To conclude, this thesis has focussed on tracking down the syntactic and semantic correlates of
two distinctive patterns of inection (reected in both subject agreement and TAM marking).
Although the search has been ultimately inconclusiverequiring a high degree of arbitrary stip-
ulation in the verbal lexiconthe analysis of the languages morphosyntactic complexities has
served to lay out fundamental aspects of this relatively undescribed Papuan language.
252
Appendix A
An Inventory of Verb Classes
An electronic lexical database was developed by the author during his eldwork in Bougainville.
This database was based on a pre-existing Shoebox dictionary of Rotokas that was originally
developed and released to the public by Irwin Firchow and made available by the Summer
Instititute of linguistics (Firchow, 1973, 1984). All of the verbs in the dictionary are listed
below according to their valency (in angle brackets), argument type (in square brackets), and
classication with respect to the distinction between and (in vertical brackets).
This list was automatically generated from a Shoebox dictionary using a Python script writ-
ten by the author. The script takes advantage of a Shoebox class library developed by the
author, which has been incorporated into the Natural Language Toolkit for Python (Loper and
Bird, 2002, 2004)see www.nltk.org for the latest version.
A.1 1 [SUB] (Total: 385)
Total: 385
aaoaao become grandparents
aapaapau unfamiliar, visit
aasi decorate with beads
aau blinded by light
agara quiet, unasserting, calm
agasi be full
ageagesi laugh
agesi laugh
aguvi clean or worn clear
aio eat
aioaio snack
aiva easy, simple
aku salty
akuta open mouth wide
api embarrassed
apopoi difcult, expensive, hard to get
araiva easy, simple
arakasi deserted, vacant
ararave wilt, weaken
arasi skillfully, carefully
areo recover
arii be ashamed
arikisi curdled
aritaru delay, linger, hesitate
253
arua ???
asige sneeze
asikauru rust
asiriko dirty, soiled
asisoe numb, sickly, paralyzed
asitaisi march in line
atara huddle together, sleep together
atari sh
atario hunt
atoro weak, disinterested, lazy, disabled
ava go
avavaia frustrated, confused
aveave cross, fussy
aveaveo soreness in groin
averu tissue-like, thin
avi sunset
avuka age
eaka relax, be tranquil
eavi ooze pus
eoro suspend like fruit
erako collect rewood
erakusi persisting
erao wake up, arise from sleep
ereere walk acrosssomething suspended or
lying on the ground (e.g., log)
ereviako walk along, step along
eru stink, spoil, decay
eteku short
gae run
gai aghast, shocked
gapu naked
garagarako excited, startled, anxious, shake
gare small, little
gariava ???
garigariava ???
garo loose, slack
garogaro loose
garutu slow, dilly-dally
gaurirao slippery, smooth, polished
gauru migrate, uproot, leave home
gausisi smooth
gauvioro easy, slow
gavata rot, putrify
gavogavoto loosened, slack
gesi smell, taste
giigiirau groan
gisi drown, ll up with
gitagita tough
goagoara boiling
goegoe slack, loose
gogoura pass on responsibility, go on com-
pletely, leave behind
gorogoro boil, broil
gorotu soft, pithy
goru strong, tight, rm, hard
gotogoto hung up
govuto gray, muddy
gue lean
guvuguvurio bubble up, effervesce, splash
iipa go up, go on top
ira go ahead, go rst
iruviro quarantine
itako sour
itoroko stiff
kaa gag
kaakasi hot
kaava feint with bow and arrow
kaeviro lift off, take off
kaie make trash, create a mess
kaipori perky, alert
kaitutu resolute, steadfast, tight
kaki cracked open, split open
kapeaa insubstantial, imsy, unstable
kapoo poor, destitute
kapua have sores
karapi sing high pitched
karavisi angry, upset
karavuru get dusty
kare return
karekare itch
254
karekare return
kareke appear, happen, come to be
karivai have an appetite
kasi start a re, make a re
kasikasi cross, angry, difcult, diligent
kasirao hot
katukatu rot away, ake off, unfastened
kauo jump
kauokauo jump up and down
kavau be born
kavee cool off in a shaded spot
kaviru steal, rob
kavori collect craysh or lobster
kavu left behind, left over
keekee chipped, shattered
keke look
kekeputu nearly, almost
keopa taste good
kerau stiff, rigormortis, rigid
keru harden like bone
kerui thin, bony, skinny
keruria persistent, stubborn, determined
kesi limp
kevaita kid, joke, jest
kevoisi persistent, determined
kii short of, lacking
kiire play tag
kirava ???
kiru have sore near mouth
kirukiru crisp
koata enter
koeta mature, grow, ripen
koi high pitched sound
koie act like a pig
kokoisi sweat, perspire
kokopeko unconscious, in a stupor
kokoro crazy, insane, foolish, stupid
kokoruu insect-infested
kokosi itch, sting
kokovae sing
kooe swing on something
kookooia mourn, singsing-cry
kookoopeko faint
kooroo have hampered speech, be hoarse
koova sing
kopii die, very ill
kopuasi restored, rejuvenated
kora vent anger or frustration on an object
korara spin top in play
korau clear, unobstructed
kosi go out, exit, come out
kosikosi come out, exit
kosiviro go out, exit
koto hang
koukouo laugh heartily
kova grow, mature
kovasi pregnant
kovata thrilled, happy
kove fall, drop
kovekove drip repeatedly
kovokovo play Jews Harp
kue reproduce, bear fruit
kukauviro deteriorate
kupukupu excited, anxious
kurokuro arthritic, paralyzed
kusii cool off
kuukuuvu lie, deceive
kuuvaki quiet
kuuvu lie, deceive
kuvau alone
kuvoro burned out, extinguished
lotu worship, attend church
oe throw up
ogaaga whisper, talk quietly, be sly
ogoe be hungry
oirao true, valid, real
okoee crab-hunt, collect crabs
okote collect crabs
oku miss out
255
opaopara disoriented, lost
opesiko vanish, disappear
opokavu put belt around the stomach to end
hunger pangs
oreore tired of doing somthing, dislike doing
something
ori cook
otara recline, lean back
oveove revealed, uncovered, exposed
ovoi nish
ovoio be.last
papeo obedient
papu extinguish, die out, without
paro wander?
pau sit
pekapekara line up
pekoe uneasy, restless, impatient
pepe sleep
perepere roll
periko roll, fall
peru ripe
pesipesiko fade away, disappear, lost
piaopa stubborn, not open to suggestions
pigoga infected
pikarata explode
piru slip, slippery
pogata burst open
pokapoka lazy, unenthusiastic
poko explode, erupt
pokopoko explode repeatedly
popote whiten, turn white
pore turn
porete recovered
poro wet, damp
posige snort, half-sneeze
posiposi dry
pou arrive
pouka lean, inclined
povuvau dull, blunt
pugu busy, occupied
pupukai dirty from dust
pupuraki perspire, sweat
puupuru darkened
raaka dry up
ragegeta dried out, dessicated
rageragete weakened
ragorea slump, wilt, nod, doze
raipi clear
raka dry, reef
rakote die down, near completion
rao drain
rarakeo light weight
rare be ashes
rasirasi satised, content
rasivauru ???
raverave weaken, tire
rearea rest, relax
reasi be disinclined, tired of, dislike
regeri play, insincere
regore bent, crooked
rerei make mischief, play pranks, play
around, goof around
riariao sweat
riga spread, scattered
siee slack, loose
sieru rainwashed, soaked, drenched
siiguru drum, beat drum
siiroi stop activity, quiet, silent
siitako be troubled
sikasika spread, disperse
sikopa nearly half full
siku wallow in mud
sipari comb
sipei salty, sting
sipiro play
sirakoisi sit and worry or sorrow
sirao pity, feel sorry for, care for
sirorova foggy, hazy
sisigarue clean
sisisa shine brilliantly, glory
256
sisiu bathe, wash
sitoka intense pain
tagugu cloudy, overcast, uncertain
takau tired, disappointed
takoto shout
taku bow over, bend over
takutaku low to the ground
taoro fat, obese
taovi thick
tapetuta criss-crossed
tapurisi unconscious, sleep soundly
taraigegea stubborn, not open to sugges-
tions
tarao divine sickness
tariata scorched
tariri wander about, run around
taritarikoi go in circles
taruko pregnant
taruu continuously falling
tasi wear shoes
tauai far off
tava expose hidden deeds, dry in the sun
tavatavari disperse
tugura complete journey, arrive at an in-
tended point
tupe next in line, follow suit
tupetupereo in pairs
tupitupi wet, moist, damp
turi exceed limits
tutuvagi nightfall, become night, dark
tuukau stiff, rigid
tuutuusi shake, quiver
tuuvu brave, steady, resolute
tuuvuu swell
tuvituvito sore
tuvutuvuke frequent
ugoro cold
uguro soggy, placid
upe wear Upe
urio come
uririko scared stiff, stiff with fright
ururupa shut the eyes
urusi dream
utave blow Tritons trumpet
uteo cold, cool
uturoo walk hesitantly, toddle along, walk
like toddler
uu meet together, gather
uuge slack, loose
uukaio drink
uureo sour, bitter
uusi sleep
uvagi deaf
uviro cross over
uviru cooked completely
uvui be able
uvuru meet, gather, assemble
vaagi pit cook, steam bake
vagapa fall a great distance
vagevage race, compete
vakuvaku scorn, doubt, scoff
vara come down, descend
variri pray, petition
varivarike hasten
varu go up, ascend, loose
varu nd meat
varuvaru healthy, vigorous
vasava cover over, grow new skin
vasi ???
vasivasi important, outstanding
vatasioko unsettled, discontent
vatatopo be ready, be careful
vatau hide
vatukoro coagulate, thicken
vavarai wild, undomesticated
vavata heavy
vavau breathe
vavauko talk in ones sleep
vavavu bitter taste
vavio dodge, avoid
257
vavorii shut eyes
vearo good, ne, well
vegovego picnic in the jungle
veke become gel, be sticky, become paste
vepu yell
vera leave
verevereko roll
veri worthless
vesi portion out a bit at a time, space out ac-
tivity
veve completely ripe
vevei fully ripe
viae clear, innocent
viaka clear
vieiasia illiterate
vieviei enumerate, count several objects
vigovigo hot, hot
viiaka empty, void of, vacant
vioro ripe
viovoko become adolescent
virakoi orphaned
virata wild, untamed after once being tamed
virikasi very hot, difcult
virivari protecting, shielding, averting dan-
ger
virivirio think of ones self only, self centered
thinking
viroo return as inevitable consequence
viru move
vitavoko hard
viuru ght
viuviu straight, unpretentious
voevoe belch, burp
vogeta draw in stomach, have empty stom-
ach
vogete ecstatic, joyful, smile
vogisi saturated
voki become night, get up
voosi blind
voovoosi settle out of a liquid, solidify
vore return, come back, go back
voruvoru wrinkled
vovosi settle out of a liquid
vovueo unsalty, tasteless
vuato clear out
vuavua cool
vuivui dirty
vuri bad, inferior, spoiled, wrong
vuro out-of-it, stupied, drunk
vutuko round, panlike
vuvui transparent
vuvure blow
vuvutau vaporize, steam, smoke
A.2 1 [SUB] (Total: 66)
Total: 66
aata swim
aka open the mouth, shout
asigo speak Rotokas
eeko defecate
gau cry, weep
gaugau cry
gipugipu whimper
gosigosi limp
guruko make noise
ikaikau run
ikau run, hurry, speed
kapere swim with part of the body out of the
water
kapuu dumb, not speaking
koikoi groan with pain
koke make rain
258
kokoroku crow
kovo work
kupare smoke, produce smoke
opoko defecate, eliminate
pai confused, difcult, stuck
papa y
parakau light up, spread across an expanse
paru ow, move, go, run
pegu bark
pegupegu bark
pigo defecate (chicken)
pika splash
puapuata splash over
pupi play bamboo pipes
puu break wind
raraka become light
raurau sway back and forth
reku genuect, kneel, fold over
rekureku kneel repeatedly, dance bending the
knees deeply
rere descend
rigorigo stroll, roam
ritoko defecate (pig)
roge thirst for
roko go into, penetrate
roru happy, glad, pleased
ruu stop
sikere streak of light, start to shine, dawn
sipokoro sprout through surface
sipukao sprout
sipusipu grow, shoot up
sirusiru shiny
siruvau good-looking, nice appearance
sisikore shine, gleam, glisten
taaripa circle, spin
tori run away, ee
tou be, stay
tugisi defecate (dog)
tupi defecate (rat or insect)
ukauka swish around, splash around
uuko get water
vauvau make noise, make a ruckus
vekaveka gasp, breath heavily
veu be angry
viku go to garden
visiko play
viviko urinate, piss, pee
voakou eliminate feces or urine
voka walk, scan, glance through
vusivusi burst forth, erupt, break out
vusivusivi appear, come out
A.3 2 [SUB, OBL] (Total: 55)
Total: 55
agigio [pa] respect
aite [re] father
apota [pa] poor, lacking
arikoko pay respect, honor by avoidance
asia dislike, without
aukue [re] show off
avekata easy, simple
avivike [pa] mark as important, pay respect
towards, honor
ega [ia] rejoice, feel pleasure
era sing
geuru [re] snarl and spit
isiva [re] back up, reverse, reject, turn back
on, turn back towards
kasipu angry, cross, pissed off
kaureo contradict, disagree, stubbornly
against, rebellious
kausiopa stubborn, unrelenting, concerned,
259
anxious
kavorou covet, keep something intended for
another, intercept
keera call for, beckon to, signal for meeting
keri [va] make enemies with, reject friend-
ship
korukoru [re] block, obstruct, hinder, de-
ter
oive [va] shout, yodel, yell
oove menace, frighten with gestures, chal-
lenge with gestures
oto [va] fornicate, commit adultery, rape
ovaovari [re] forget something recently
thought of, remember but not for long
ovau [re] forget
pako break, raze, tear down
paupau race, compete
pesi [re] forget
poreo [va] commit incest
rate [va] stare at
reoreo [re] converse, discuss
riata [ia] disclose, reveal hidden message,
boast about somehing
riu [re] irritate, pester
rui spit out
ruipa [pa] like, want, desire
sirava [re] hiss
siririko [re] peek through opening
sisivare inspect, examine intently, search
taea [pa] deceive, deny, accuse
takato [re] argue
tarai understand
tarataraa [pa] embarrassed for lack of
something
tavitavi [pa] tell
upia [va] in pain, sick
uugaa [va] kiss someone
vagu [re] proud
vari [re] feint an action with a spear or axe,
threaten
vasiare dislike
vavagisi [ia] difcult, confused
viiroo [pa] repulsive
viki [ia] toss out, throw away, lose
vikuta [re] whistle with the lips, tongue, or
teeth
viokeke whistle with pursed lips
voki become night
vootu vote for, elect
voroko [va] disobedient
A.4 2 [SUB, OBL] (Total: 35)
Total: 35
aivaro meet with, go directly to
apo miss out on something, come up short
of
atu too much, overow
aveavero [ia] incite to anger
iru delouse
kapekape embrace, grip with arms not meet-
ing
kavikavi [re] combine, work together
kaviko love intensely
kokee [re] peek through a blind or crack
koroto [re] meet together
kuara [va] yell at
kuga [ia] bump into, nudge
oruo [ia] diligent
pae appropriate anothers possession, iden-
tify
pitu [ia] hold, alight
raavaa [re] ready, meet
siga [ia] open
260
siki [re] moon, expose bare ass to
taagau [ia] step over something, jump over,
pass over
tagau [ia] jump over
tagava salute, shield the eyes with hand
tara look for, search for, seek
taratara [ia] unable to recognize, uncom-
prehending
tare [re] await in vain
tasiasi [ia] stomp on, step on repeatedly
tauo [pa] offer in ceremony
tue harvest, pick a leaf crop, wait
uvisi [ia] grip tightly, hold onto rmly
vato [pa] honor
vaute decorate with owers, feathers, etc.
veku [va] bark
vikiviki [??? ia] toss several things
vorevore repeat
vura look at, gaze upon
vusi burst forth, erupt, break out
A.5 2 [SUB, O OBL] (Total: 5)
Total: 5
piiro point towards
rekesi explain clearly, recognize truth of
vate [pa] give
vatevate [pa] exchange
virutu [re] squeeze out a liquid, extract
A.6 2 [SUB, O] (Total: 482)
Total: 482
aato answer
aaviito purify, remove altogether
aavito cure
agaru complain
agesi laugh at
agiagi greet, welcome back, be reconciled
agoagoto atter
aio eat
aioaio snack
airerei safeguard, protect
ake ask
akoro charm with powder or with package of
powerful objects
aku salted
apeapei claw at
apei scratch
apui dig a ditch
are request, ask for, call for
areii organize
arirao harvest food
aroviaku cool anger, pacify, persuade
aruo weed garden
arupa fertilize, cause growth in garden
asita apply putty
asivuru collect melons or cucumbers
ate weigh, scale
ateate weigh, scale
ato harvest from tree by cutting or picking
atoato wipe away, smear on
auau quiet someone
aue ignore
avaavaeo sort out
261
avaavao act simultaneously, anticipating
avaisi spice food, avor vegetables with aro-
matic plants or herbs
aveverau release from ones responsibility
unintentionally
avokori another kind, different, not recog-
nized
avu bite, remove
avui make hole in the nose, pierce septum in
nose
avuru swarm, swarm on, be attracted to
eaka hand over, give up, concede
earova give generously without thought of
reward
eerii prompt someone to do something, urge
someone to action
egaega compliment, be enthusiastic
epa peel, husk
eri dig, chisel
erieri dig, chisel
erii persuade
eriikasi push
eto build a re
evei recognize
gaa wedge, pry
gaari hoe, heap up ground
gae follow
gaegaere drift
gagari plane wood
gagarike scratch, rake with claw
gago skin
garigari scrape
garu shave head with bladed instrument
gasi break, penalize, break the law, con-
demn
gasigasi break into parts, splinter
gatagata chew
gatao extract out juice, suck out juice
gatu pour out, overow into another vessel
gavagava soften, cook to soft texture
gaveru drop, lose grip
gavi wipe off, rub
gerigeri knaw on, bite lightly
getegete spoil something, ruin something
getu break
gevo secure something
goagoa ???
gogi loot, take spoils
gogu overlook, miss seeing
gope drop
gopori tickle
gopu break, take out of proper place, dislo-
cate
gore bow down, bend down
gori turn aside from, separate from
goro dislodge, pry loose, take out
gota catch
govugovu clean out, purge
gua shake penis
guagua masturbate, jerk off
gugi twist
gugiugi twist repeatedly, screw or unscrew
gugura gather in a heap, bunch together
guiguisi spray out
guru meet, heap up, assemble
guruguru gather in a heap, bunch together
guvaguva cool off
guvi come out of hiding, reveal something,
expose
iia shoot
ipu dam up
iraira stretched out in front
ireire shoo out of the way, warn of impending
danger
iruuta mess up, disorder, make untidy
ito struggle with, pull back and forth, grap-
ple
iusi use
ivia investigate, scout out, test
ivu pull
262
kaa strangle
kaapisi pinch together, grip with pincers
kaareko scour, clean by scraping
kae carry
kakapu place in sling for purpose of carry-
ing
kakavu scoop up with the hands
kaki crack open, split open
kakiaki crack open, fracture
kaku split open
kakuaku break into pieces with instrument
kapa eat after fasting
kapara roast without pan or container
kaparu short of, missing
kapatau augment, add to, cap up, supple-
ment
kapeaa imsy
kapo join together, clamp together, fasten
on coverstrips, put cover strips on house or
wall
kapokapo fasten.cover.strips
kapokapora carry between two peoples
shoulders
karakarao take without permission
karata deal out, divide up, apportion
kareo penetrate, pierce through
kari rip, tear
karikari tear, shred
karo spoon out a liquid
karokaropo deal out, distribute, send
karopo portion out, divide up
karu open, unlock, untie, unhook
karukaru open
karutu divide up, portion out
kasi burn
kata exhaust
katuara scour
kavakavau reproduce, bear many children
kavau give birth
kave whisper, reduce the strength or heat of
something
kaveruko hold in arms
kavikaviru steal
kaviru steal, rob
kavo scavenge, pick up, collect
kavokavo perform sorcery, work black
magic
kavu leave behind
kavusi spit forcefully towards mark, spit out
kee shatter, fracture, chip
keke look at
kepi fracture, break
keravisi plough under, turn soil over
kerete turn around
kerikerisi evaluate, judge carefully
kerisi discern, evaluate, judge talk or situa-
tion well
ketaka notch out, make groove
ketu break.off, break off a piece
kevaita kid, joke, jest
kiki kick
kikira mix meat and greens
kikitausi tear off with teeth
kio attract attention by touching, tapping, or
scratching
kipe cut grass with a sickle
kipu paint, smear on surface
kipukipu rub on, smear on, massage
kiri rip open, tear open
kiro write
kirokiro write
kitukitu scrub clothes
kiu put in, insert
koa bark, skin, peel
koakoa bark a tree, remove the skin
koara put together
koe spoon out a solid
koekoe spoon out
kogo cut, chop
263
koka agree
koki chisel out, chip away
koko pour, serve, dish out, portion out
kokovu shave head
koku break off at base, snap off at base
kopa swallow, gulp down, ingest
kopakopa swallow quickly, qulp down
kopikopi baptize, sprinkle
korita carve, carve, dissect, cut up
koroviri braid, plait, twist together
kosikosi cut off sago palm leaves
kosipa ???
kotu bite
kotukotu gnash teeth, grind teeth together
kou lay egg, defecate
koukouo laugh heartily at
kove fell
kovokovo fence off, surround
kovovo fence, protect
kukiuki shake something, rattle something
kuku spoonfeed
kupekupe fan
kuri scrape, scratch, gnashing, gritting
kurikasi urge along, prod along
kurikuri scratch repeatedly
kuru strip off branches
kururu crumble something
kuva work sorcery, do black magic
kuvu ll up, put inside bamboo, clothe
kuvukuvu ll up, stamp the ground
oapa carry
oe vomit, sea sick
oga follow behind
ogo conceal, hide
oku miss, miss out on
oovaau track
opari lose
opesi end, nish
opi intercept, interrupt, cut across, shortcut
orere look intently, size something up, stare
at
ori cook
oriori scrape, scratch
oriorisi suspect, distrust
oriru store away, keep, save
orito decorate
orivo name, label
oru trim down, shave away
otu sharpen to a point
ou get, take, receive
ove pour out
ovuovu try
paipai blocked, obstructed, stymied
pako pull down
paku net
pao open something
papu extinguish, put out
parasire exchange places
pare remove from net
paripari split in half
pariparikou alternating, exchange repeat-
edly
pau plant, build
peara open
pege break into pieces
pegepege break open repeatedly
peka turn over, ip over, reveal, turn page
peo push, shove, heave
peopeo pump
pera shove, kick out of the way, motion
aside
perapera kick repaeatedly aside, shove out of
the way
pero slice into planks, split apart
peto overturn, pour
petopeto rock to and fro
pia prune, trim off
pigi twist, squeeze, wring out
piiuu rape
264
pikipiki blind with light, dazzle with light
pikopiko whip
piku break, have tip broken off, nod the head
pio smear white substance from Pioto hot
spring on something
piopio discuss, argue
pire allow to be harmed
piro mislead, divert, cause someone to err
piruiripa wash
piruko forgive, restore, clean
pisikui tie in knot, knot something
pisipisikui tie in knot, knot something
poera reveal, expose
poo test out
pooke bend taut
porapora space apart
porepore mix, steer
poroporo break up into pieces
poroporoko wind along, follow winding
path
potu break off
puaka ???
puepue weed
pui sweep
puko reply
puku swell up, hump
pukupukui mound up, hill up
pupiupi puff, blow
pura make, do, create
puraka spy out, survey
purepure fan something
puri lay on side, press down(?)
putepute cut up, chop up
raerae test, try out
ragi whip, beat, thrash, whack
ragui care for animals
raku cover over
rakuraku cover over
rakurakuo pile up trees or posts
rao drain
rapasi notch out with axe or knife in tree
raravio loose grip on something
rata heat up, sear, singe
rataa trick
ratarataa trick, deceive
rau grab, hug, hold
ravaa ready something, prepare, meet
ravarava attempt, try
raviravisi dodge, elude, go around, bypass
ravoko hold onto
ravu restrain, hold back
ravutu le something
reesi mark, measure
reesireesi warn
rego bend
rekareka break apart, crack into pieces
reko preserve, repair, correct
rereo smoke food
resiresi warn
retu cut into sections, section off by cutting
returetu cut into sections, section off by cut-
ting
rigariga erase
rigato write, print, type
riri covet, envy
roe place above
roi have sex with, screw, fuck
roo cut
rooka portion out, dole out, share
roorookaa divide into (two?) parts
rugurugu heap together, gather together
ruu cover, enclose, enwrap, envelop
ruvaru medicate, give medicine
sie wipe nose, move something away
sigi deate, reduce size of, release pressure
in
sigu take away and destroy, expel
siguri miss the mark
sigusigu shoo away
sigusiguva join together, weld together
265
siguva join together, mix, join after, splice,
add onto
sii cut hair, groom hair
sika separate, divide
sikuro aid child or disabled person to walk,
support physically
siopai not recognized, be unfamiliar with
siopore explain to, give counsel to, enlighten,
elucidate for
siovo feel, touch, sense
sipa tear
sipo send
sipoko up-end, turn upside down
sira cover over, weight down
siresire make a anged edge
siruru give blessing, charm someone, deco-
rate with charms
sisi pacify, change a persons mood or atti-
tude
sisiputa shake head or hands
sisiputapa shake
sisiu wash, wash off
situe look, watch, observe
sivesive peel, strip off
sora work sorcery on
sosope standby with disinterest, avoid
sosovo taste, sample something, feel some-
thing
taasi put together, complete something
taava judge
taavo anticipate, watch with anticipation
taavore help, assist
taavoto shoot accurately
taga mark off, stake out
tage insert, put inside
tagi be responsible for, care for
tagoro secretly do something, conceal from,
secretly kill
take build walls
taketake cause trouble, rape
taki hold, pin, hold steady, hold down
takitaki fasten together
takou cover up, package, cook in an enclo-
sure
tapa hit, slap, crucify
taparako slap, punch in anger
tapo fasten together, join together
taporo conceal talk
tapotapoko persist, stick to it
tarauru polish
tari surround, encircle
tariko encircle, surround
tario chase, pursue
tarita smash, mash, mince, grind, chew
taritariko go in circles
taroro jack up, pry
taruru atten out, smash at
tavario exchange, change places
tavo wall up with sago palm leaves
tavore help, forgive
tavuru cover up
tesiko polish
toaera give food as engagement invitation
toe cut, chop, slice
toetoe chop or cut repeatedly
toga spear, shoot with a spear
toitoi shake, tap, pluck
toko cut, break
tokotoko cut, prune
too punch, hit with hand or st
tooguu ring-bark a tree to kill it
toova bury
tosi cut with blade
tova bury
tovi kid, jest, belittle
tovitovi restore to value, repair
tovo put, place, position
tovotovo distribute
tovutovu erode away, dig out
tukituki break into pieces
266
tupa close, lock
turo beckon to, go with, carry away
turu string up, sew up
tutu carry on the back
tuuke fasten, lock, nail
tuutuuko repay
uga force through
ugo fasten, close
upo strike, ght, murder
ura chew (betel nut)
uraura picture, photograph, lm
uriri frighten, scare
uririo make tingle, cause prickly sensation
urouro exemplify, copy, repeat, reiterate
ururau hide from, secret away
uto shield, hide from view
utuvaiko surpass, pass by, beat
uvere join together, mix together
uvu hear, smell
vaagi pit cook, steam bake
vaagore persuade, trick
vaavaavu embitter
vaere hoe, turn over soil
vagevage ???
vago slack, loosen, open
vagogo scout, spy on, scout out
vagore stop
vaisi name, call, label
vaivaisi name things
vaki mistake for something else, fail to rec-
ognize
varia fasten with a noose, trap with a noose
varoova care for, be responsible for
varovaro pursue, follow in pursuit
vatako mix together, join together
vatatopo ready, prepare
vatau hide
vatavata go around, by pass
vatavatau hide several things
vatave join together, include together with
vea lick
veavea lick
veeku disregard talk
veepo shove aside, move out of the way
veera line up, form a line
veeto slash through
veeveera line up, put in rows, form a line
vega cut one side
vera remove
veravera change, get rid of
verete move to one side, shove aside
veriverisi wander, make the rounds, on the
move
veruveru scale sh, make circular marks
veta gnaw on
veu stain
viaviatarau clear
viei count, read
vieviei enumerate, count
vigu loosen, slacken
viioo mimic, imitate, copy
viivii strip away
viko fold, bend over, roll up
viou cut away, clean, sweep
viovioe exemplify
virako bless, do good supernaturally
virava ???
viri twist
viriviriko twist something
viroviro entwine, wrap
viruviru move back and forth, retreat, make
go back
visi poke, jab, hit
visirako whip, strike with object
visivisi beat a slit gong, tap
vitu excrete, urinate, defecate
vivi underestimate, be short of, slurp with the
mouth
vogo roll up something at
vogovogo crumple, wad up, knead
267
voreri oscillate, go back and forth, vascilate,
repeat
vori cost, pay, buy
voro roll up, wind up
voroo hunt with dogs
vovovo warn, caution
vuravura scan, gaze, watch
vurita divide exactly in half
vuroko stone, throwrocks at, throwstones at
vuruko section off
vuta taste
vutuo carry on shoulders
vututu go altogether
vuvure blow
268
Appendix B
A Finite State Transducer for Rotokas
Morphology
B.1 Overview
The author has developed a nite state transducer (FST) for Rotokas morphology using the XFST
program (Beesley and Karttunen, 2003). A nite state transducer (FST) is a nite state machine
with two tapes: an input tape and an output tape. An FST transduces (i.e., translates) the
contents of its input tape to its output tape, by accepting a string on its input tape and generating
another string on its output tape. It may do so nondeterministically, potentially producing more
than one output for each input string. A transducer may also produce no output for a given input
string, in which case it is said to reject the input. We will refer to the input tape as the upper
side and the output tape as the lower side. The lower side of the Rotokas FST contains the word
forms of the language while the upper side provides one or more morphological analyses of
the corresponding word form. A few examples of the analysis provided for unambiguous word
forms are provided in (801).
(801) fst[1]: up oirato
oira+Noun+Sg+Masc
fst[1]: up uusiparoi
uusi+Verb+Alpha+Cont+3rd+Masc+Sg+Real+Pres
fst[1]: up upoparevoi
upo+Verb+Beta+Cont+3rd+Masc+Sg+Real+Pres
Using the rst example for purposes of illustration, the FST takes the inected noun oirato
as input and return a single morphological analysis as output. The morphological analysis
identies the root, oira, and provides three tags: +Noun, +Sg and +Masc, which serve to
identify the part of speech and its inection, which in the case of a noun consists of its number
and gender.
269
When the lower side contains more than one morphological analysis, the word form on
the upper side can be considered ambiguous. In cases of ambiguity, there will be more than
one morphological analysis corresponding to a given word form, as illustrated by the form
riakova, which could be analyzed either as the singular form of the noun riako woman or as
the classier riako together with the comitative enclitic =va.
(802) fst[1]: up riakova
riako+Cl+Enc
riako+Noun+Sg+Fem
In the practical orthography, a hyphen marks a clitic boundary and can therefore be used to
rule out the analysis where va is analyzed as nominal inection for the singular feminine.
(803) fst[1]: up riako-va
riako+Cl+Enc
Note, however, that the lack of a hyphen does not rule out an enclitic analysis. This is an
intentional design feature of the FST. Since native-speaker consultants do not consistently indi-
cate clitic boundaries, the Rotokas morphology FST would misanalyze words that lack proper
formatting of the enclitic. To avoid this type of systematic misreading, the FST is permissive.
The nite state transducer for Rotokas morphology provides a testable model whose cover-
age can be quantitatively assessed (see Karttunen (2006) for a plea in favor of the formalization
and computational implementation of linguistic theory). To assess coverage, the example sen-
tences in the Shoebox dictionary described in Appendix A were tokenized (broken up into
individual words) and analyzed by the FST. The results are summarized in Table B.1, where the
number of word forms recognized by the FST are tabulated. Two different counts are provided:
one for the number of word forms recognized regardless of whether they occur multiple times
(tokens) and another for the number of unique word forms recognized (types).
Tokens Recognized 45,590 96%
Unrecognized 1593 3%
Total 47,183 100%
Types Recognized 14,006 90%
Unrecognized 1527 9%
Total 15,533 100%
Table B.1: Coverage for Shoebox Dictionary Example Sentences
270
B.2 Source Code
The full source code for the Rotokas morphology FST is relatively short and is provided in full
below:
# ---------------------------------------------------------
# Author: Stuart Robinson
# Date: 5 July 2008
# Desc: Script that creates a finite state transducer for
# Rotokas morphology
# ---------------------------------------------------------
set flag-is-epsilon ON
source english.char.defs.infile
# ---------------------------------------------------------
# Misc.
# ---------------------------------------------------------
define Particle [ @txt "lex-particles.txt" "+Part" : 0 ] ;
define Sep [ 0:[ {-} | {=} ] ] ;
define Consonant [v|p|t|s|r|k|g|m|n|l|w] ;
define Vowel [a|e|i|o|u] ;
define SyllableLt (Consonant) Vowel ;
define SyllableHvy (Consonant) Vowel Vowel ;
define FootDegen [ SyllableLt ] ;
define FootFull [ SyllableHvy | SyllableLt SyllableLt ] ;
define Foot [ FootDegen | FootFull ] ;
define Flag "@U.CLASS.ALPHA@" | "@U.CLASS.BETA@";
271
define MarkFoot FootFull @-> "[" "[" ... "]" "" 2 "]" ||
.#. Flag _ ?
*
"+Redup" ;
define Cleanup "+Redup" -> 0;
set flag-is-epsilon off
# ---------------------------------------------------------
# Verb Morphophonemics
# ---------------------------------------------------------
define Rule1 {io} -> {i} || {-} _ {-e} ;
define Rule2 {vio} -> {vi} || {-} _ {-e} ;
define Rule3 {ei} -> {i} || {-} [ {a} | {o} ] _ ;
define Rule4 {e-e} -> {eie} || _ ([ {pa} | {ra} | {i} ]) .#. ;
define Rule5 {o-e} (->) {o} || [ {-vir} | {-pir} ] _
[ {pa} | {ra} ({o}) ] .#. ;
define Rule6 {o-e} (->) {o} || {-ira} _
[ {pa} | {ra} ({o}) ] .#. ;
define Rule7 {o-e} (->) {o} || {-} _
[ {pa} | {ra} ({o}) ] .#. ;
define Rule8 {o-oro} (->) {ouoro} || _ .#. ;
define Morphophonemics [ Rule1 .o. Rule2 .o. Rule3
.o. Rule4 .o. Rule5 .o. Rule6
.o. Rule7 .o. Rule8 ] ;
# ---------------------------------------------------------
# Verb
# ---------------------------------------------------------
272
define AVerbStem [ "@U.CLASS.ALPHA@"
@txt "lex-alpha-verbs.txt"
[ "+Verb" "+Alpha" ] : 0 ] ;
define BVerbStem [ "@U.CLASS.BETA@"
@txt "lex-beta-verbs.txt"
[ "+Verb" "+Beta" ] : 0 ] ;
define ABVerbStem [ AVerbStem | BVerbStem ] ;
define VerbStemRedup ABVerbStem ( "+Redup" ) ;
regex VerbStemRedup .o. MarkFoot .o. Cleanup;
set retokenize off
compile-replace lower
define VerbStem
define VerbSfxModifier [
"+Emph" : {irao}
| "+Emph" : {vasi}
| "+Delim" : {raga}
] ;
define VerbSfxContinuous [
"+Cont" : {pa}
] ;
define VerbSfxCompletive [
[ "+Compl" "+Inanim" ] : {piro}
| [ "+Compl" "+Anim" ] : {viro}
] ;
define VerbSfxPersonNeutral [
[ "+3rd" "+PL" ] : {ta}
| [ "+1st" "+PL" "+Incl" ] : {vio}
| [ "+3rd" "+Dl" "+Masc" ] : {si}
| [ "+3rd" "+Dl" "+Fem" ] : {ere}
273
| [ "+1st" "+Dl" ] : {ve}
| [ "+1st" "+Excl" "+Pl" ] : {io}
| [ "+1st" "+Incl" "+Pl" ] : {vi}
] ;
define VerbSfxPersonA [ "@U.CLASS.ALPHA@" [
[ "+3rd" "+Masc" "+Sg" ] : {ro}
| [ "+3rd" "+Pl" ] : {a}
| [ "+1st" "+Sg" ] : {ra}
| [ "+2nd" "+Sg" ] : {u}
| [ "+3rd" "+Sg" "+Fem" ] : {o} ]
] ;
define VerbSfxPersonB [ "@U.CLASS.BETA@" [
[ "+3rd" "+Masc" "+Sg" ] : {re}
| [ "+3rd" "+Sg" "+Fem" ] : {e}
| [ "+3rd" "+Pl" ] : {i}
| [ "+1st" "+Sg" ] : {a}
| [ "+2nd" "+Sg" ] : {ri} ]
] ;
define VerbSfxPerson [
VerbSfxPersonNeutral
| VerbSfxPersonA
| VerbSfxPersonB
] ;
define VerbSfxIrr [
[ "+Irr" "+DF" "+1st" "+Dl" ] : {vearea}
| [ "+Irr" "+DF" ] : {verea}
| [ "+Irr" "+Dl" "+Anim" "+Hab" "+1st" ] : {veaira}
| [ "+Irr" "+Hab" "+Anim" ] : {aira}
| [ "+Irr" "+Hab" "+Anim" ] : {veira}
| [ "+Irr" "+Hab" "+Inanim" ] : {peira}
| [ "+Irr" "+NF" "+1st" "+Dl" ] : {veare}
| [ "+Irr" "+NF" "+Inanim" ] : {pere}
| [ "+Irr" "+NF" "+Anim" ] : {vere}
| [ "+Irr" "+Sub" "+Inanim" ] : {pe}
| [ "+Irr" "+Sub" "+Inanim" ] : {pi}
| [ "+Irr" "+Sub" "+Anim" ] : {ve}
274
] ;
define VerbSfxRealA [ "@U.CLASS.ALPHA@" [
[ "+Real" "+DP" ] : {era}
| [ "+Real" "+IP" ] : {e}
| [ "+Real" "+Pres" ] : {ei}
| [ "+Real" "+Pres" ] : {i}
| [ "+Real" "+RP" ] : {epa}
| [ "+Real" "+NP" ] : {erao} ]
] ;
define VerbSfxRealB [ "@U.CLASS.BETA@" [
[ "+Real" "+IP" ] : {vo}
| [ "+Real" "+NP" ] : {vorao}
| [ "+Real" "+DP" ] : {vora}
| [ "+Real" "+Pres" ] : {voi}
| [ "+Real" "+RP" ] : {va} ]
| [ "+???" ] : {voiva}
] ;
define VerbSfxReal [ VerbSfxRealA | VerbSfxRealB ] ;
define VerbSfxTenseMood [ VerbSfxIrr | VerbSfxReal ] ;
define VerbSfxDep [
[ [ "+Dl" "+Sub" ] : {pe} # Whats going on here?
| [ "+Dep" "+Sim" ] : {oro}
| [ "+Dep" "+Dis" ] : {arapa}
| [ "+Dep" "+Purp" ] : {sia} ]
] ;
define VerbSfxIndep [ ( 0:{-} VerbSfxPerson )
( 0:{-} VerbSfxCompletive )
( 0:{-} VerbSfxTenseMood ) ] ;
define VerbSfxDepIndep [ VerbSfxDep | VerbSfxIndep ] ;
define VBase [ VerbStem
( 0:{-} VerbSfxCompletive )
( 0:{-} VerbSfxModifier )
275
( 0:{-} VerbSfxContinuous )
( 0:{-} VerbSfxDepIndep ) ] ;
set flag-is-epsilon on
define VerbsHyphens [ VBase .o. Morphophonemics ] ;
set flag-is-epsilon off
define Verbs [ VerbsHyphens .o. [ {-} -> 0 || ?
*
_ ?
*
] ] ;
# ---------------------------------------------------------
# Noun
# ---------------------------------------------------------
define NounGeneric [ @txt "lex-nouns.txt" "+Noun" : 0 ] ;
define NounRoot1 [ @txt "lex-nouns-1.txt" "+Noun" : 0 ] ;
define NounRoot1M [ @txt "lex-nouns-1-m.txt" "+Noun" : 0 ] ;
define NounRoot1F [ @txt "lex-nouns-1-f.txt" "+Noun" : 0 ] ;
define ProperN [ @txt "lex-proper-nouns.txt" "+Prop" : 0 ] ;
define Class [ @txt "lex-classifiers.txt" "+Class" : 0] ;
define BarePro [ @txt "lex-pronouns.txt" "+Pro" : 0 ] ;
define ReflPfx [ "@U.CLASS.REFL@" 0 : {ora} (Sep) ] ;
276
define ReflTag [ "@U.CLASS.REFL@" "+RR" : 0 ] ;
define SpecPfx [ "@U.CLASS.SPEC@" 0 : [ {o} | {vo} ] ] ;
define SpecTag [ "@U.CLASS.SPEC@" "+Spec" : 0 ] ;
define NPfx [ ReflPfx | SpecPfx ] ;
define NumMasc [ [ "+Sg" "+Masc" ] : {to} ({a})
| [ "+Dl" "+Masc" ] : {toarei}
| [ "+Pl" "+Masc" ] : {irara} ] ;
define NumFem [ [ "+Sg" "+Fem" ] : {va}
| [ "+Dl" "+Fem" ] : {rirei} ] ;
define NumGenCl1a [ NumMasc | NumFem ] ;
define NumGenCl1b [ [ "+Sg" "+Masc" ] : {to} ({a})
| [ "+Pl" "+Masc" ] : {irara}
| [ "+Sg" "+Fem" ] : {va}
| [ "+Dl" "+Fem" ] : {rirei}
| [ "+Pl" ] : {vure} ] ;
define NumGenCl1c [ [ "+Sg" "+Masc" ] : {to} ({a})
| [ "+Pl" "+Masc" ] : {irara}
| [ "+Sg" "+Fem" ] : {va}
| [ "+Dl" "+Fem" ] : {rirei}
| [ "+Pl" ] : {ra} ] ;
define NumGenCl2 [ [ "+Sg" "+Masc" ] : {to} ({a})
| [ "+Pl" "+Masc" ] : {irara}
| [ "+Sg" "+Fem" ] : {va}
| [ "+Dl" "+Fem" ] : {rirei} ] ;
define NumGenCl3 [ [ "+Sg" "+Neuter" ] : {a}
| [ "+Dl" "+Neuter" ] : ({a}) {rei}
| [ "+Pl" "+Neuter" ] : {ara} ] ;
277
define NumGenCl4 [ [ "+Sg" "+Masc" ] : {to} ({a})
| [ "+Pl" ] : {ara} ] ;
define NumGenCl5 [ [ "+Sg" "+Fem" ] : {va}
| [ "+Dl" "+Fem" ] : {rirei}
| [ "+Pl" ] : {ara} ] ;
define DerivNRoot [ VerbStem | ProperN ] ( 0 : {pa} ) ;
define DerivNStem DerivNRoot ( NumGenCl1a | NumGenCl3 ) ;
define NounStem [ NounGeneric
| DerivNStem
| NounRoot1 ( NumGenCl1a | NumGenCl3 )
| NounRoot1M (NumMasc)
| NounRoot1F (NumFem)
| {kakae} (NumGenCl1b)
| [ {oira} | {riako} ] (NumGenCl1c)
| NounRoot2 (NumGenCl2)
| NounRoot5 (NumGenCl5) ] ;
define ClassSfxNum [ [ "+Dl" ] : {rei}
| [ "+Pl" ] : {ro} ] ;
define ClassStem [ Class (ClassSfxNum) ] ;
define NSfxPoss [ "+Poss" : {aro} ];
define NSfxDim [ "+Dim" : {vi} ] ;
define NSfxAlt [ "+Alt" : {vu} ] ;
define NSfxIndef [ "+Indef" : {vai} ] ;
define NSfxTopic [ "+Topic" : {a} ] ;
278
define NSfxNonGender (NSfxPoss) (NSfxDim) (NSfxAlt) (NSfxIndef) ;
define NSfx1 [ {o} | NSfxNonGender ] ;
define NSfx2 [ {o} | ReflTag | NSfxNonGender SpecTag ] ;
define Nominal [ NounStem | DerivNStem | ClassStem ] ;
define NounBase [ Nominal (NSfx1)
| NPfx Nominal NSfx2 ] ;
define ClassBase [ Class (ClassSfxNum) (NSfxPoss) ] ;
define RelMarker [ "+Enc" : {ia}
| "+Enc" : {re}
| "+Enc" : {pa}
| "+Enc" : {va} ] ;
define Num [ %0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 ]+
"+Num" : 0 ;
define NounEnclitic (Sep) [ NSfxTopic | RelMarker ] ;
define Nouns [ [ Num | NounBase ] (NounEnclitic) ] ;
# ---------------------------------------------------------
# Pronouns
# ---------------------------------------------------------
define SubjSfx [ "+Subj" : {pe} ] ;
define ProSfx [ "+???" : {i}
| ReflTag
| NSfxIndef
| SubjSfx | NSfxPoss | (NSfxDim) (NSfxAlt) ] ;
define ProBase [ (ReflPfx) BarePro (ProSfx) ] ;
define Pronouns [ ( ProPfx (Sep) ) ProBase (NounEnclitic) ] ;
279
# ---------------------------------------------------------
# Adverb
# ---------------------------------------------------------
define AdvStem [ @txt "lex-adverbs.txt" "+Adv" : 0 ] ;
define AdvSuffixDeriv [ "+Deriv" : {pa} ] ;
define AdvSuffix [ "+Adv" : [ {vira} | {visivi} ] ] ;
define AdvBase [ NounStem | VerbStem ] ;
define DerivAdvs [ AdvBase (AdvSuffixDeriv) AdvSuffix ] ;
define Adverbs [ AdvStem | DerivAdvs ] ;
# ---------------------------------------------------------
# Create a single FST that is case-insensitive
# ---------------------------------------------------------
define Word [ Particle | Verbs | Nouns | Adverbs | Pronouns ] ;
regex [ Word .o. [ $(Upcase) ]
*
] ;
save stack rotokas-morphology.fst ;
280
Appendix C
Sample Texts
This appendix provides two sample texts in Rotokas. These are retellings of a traditional folk
tale describing the origins of the red leaves of the Terminalia catappa tree, a large tropical tree
in the Family Combretaceae. This tree is known as Talis or Talisa in Tok Pisin, and goes by
a variety of names in English: Java almond, Indian almond, Bengal almond, Singapore almond,
Malabar almond, Tropical almond, Sea almond, or Umbrella tree. Although the two tellings of
the story differ in various details, they share the same basic plot, which concerns a sacred taro
that is mistakenly harvested by two girls. After being castigated by their parents, the girls are
so lled with shame and sorrow that they leave home for the coast, where they are eaten by a
shark and their blood permanently stains the tree.
C.1 Matevu, Version 1
This version of the story was published in Rotokas in Firchow (1974a). A synopsis of the story
is provided in English, but without line-by-line glossing or translation. (No author information
is provided but David Akoitai is a likely source, given that he served as a consultant and co-
author for a great deal of Firchows work on Rotokascf. Firchow and Akoitai (1974).) During
my rst eldwork trip to Bougainville, native speaker consultants translated this story into Tok
Pisin. It was then entered it into a Toolbox database, given interlinear glossing, and translated
into English.
Firchow (1974a:109) claims that this folk tale and the associated song originate from the
neighboring Autronesian language, Teop. Firchow (1974a) does not provide the basis for this
claim, and admits parenthetically that the meaning of the lyrics is unknown: Only the names of
the taros, Matevu and Siraveru are recognized in the words of this song. The other words remain
unknown. Although consultation with Ulrike Mosel (a Teop specialist) and Ruth Spriggs (a
native speaker of the language) has conrmed that the song is known in the Teop region, they
deny that the lyrics are in Teop. While the story itself is in Rotokas, the provenience of the
associated song is therefore an open question. There is evidence of a good amount of lexical
281
borrowing between Rotokas and Teop and in some cases the direction of borrowing appears
to be into Rotokas from Teop. For example, the Rotokas word okaoto taro appears to be
a borrowing of the Teop word kaoto taro along with its associated article o (Schwartz and
Mosel, 2006).
1
(1) aue
CONN
vao-ia
PRO.DEM.PROX.3.SG.N-LOC
siposipo-a
story-SG.N
o-avuka-arei-vu
SPEC-old.person-DL.N-ALT
iava
ABL
This is a story about a married couple.
Dispela stori em bilong tupela marit.
(2) oire
okay
vo-avuka-arei
SING-old.person-DL.N
aiterei-ia
PRO.PER.3.DL.M-LOC
oisioa
always
tou-pa-si
be-CONT-3DL.M
o-urui-vu-ia
SPEC-village-ALT-LOC
Okay, these two, they were always in one village.
Dispela tupela marit i save stap long wanpela ples.
(3) uva
so
voa
here
tou-pa-oro
be-CONT-DEP.SIM
o-voki-vu-ia
SPEC-day-ALT-LOC
ogoe-a-epa
hungry-3PL
-RP
They were hungry in this place one day.

Na taim ol i stap long dispela ples wanpela de ol i bin hangre.
(4) uva
so
ovii-rirei
child-DL.F
oaesi
PRO.POSS.3.DL.M
aru-pa-si-va
order-CONT-3DL.M-RP
The two of them ordered their two (female) children.

Na tupela i bin salim tupela pikinini meri bilong tupela.
(5) oisio
like
pura-si-epa
say-3DL.M-RP
The two of them said,

Tupela i tok olsem,
(6) ava-ere
go-2DL.F
opo
taro
kuio
round
tate-sia
extract-DEP.SEQ
You two go dig up a taro plant.
Yutupela go kamautim wanpela taro.
(7) vo-kuio
SING-round
vaisi-aro
name-POSS
Vatevu
name
The name of this taro plant is Matevu.
Name bilong dispela taro, Matevu.
1
Shoffner (1976:291) also records the Teop word kaoto for Terminalia catappa.
282
(8) uva
so
o-kuio-rei-ia
SPEC-round-DL.CL-LOC
vo-taru
SING-bone
vearovira
good
va
PPRO.3.SG.N
rovo-pa-a-voi
start-CONT-1SG
-PRES
???
???
(9) vairei
PPRO.2/3.DL.F
tavi-si-va
tell-3DL.M-RP
The two of them told the two of them,

Tupela i tokim tupela,
(10) ava-ere
go-2DL.F
opo-a-vai
taro-SG.N-INDEF
ou-sia
get-DEP.SEQ
vegei-pa
PPRO.1.DL.EXCL-BEN
The two of you go get some taro for us.
Yutupela go kisim wanpela taro bilong mitupela.
(11) uva
so
oavu
another
oa
RPRO.3.SG.N
vatatopo-pa-ere-vere
ready-CONT-2DL.F-NF
And look out for something else.
Na wanpela samting bai yutupela i lukaut long em.
(12) teapi
lest
Vatevu
name
kuio
round
ou-ere-vere
get-2DL.F-NF
You cant get the Matevu taro.
No ken kisim taro Matevu.
(13) ari
but
Siraveru
name
kuio
round
ou-ere-vere
get-2DL.F-NF
But you two will get Siraveru.
Tasol yutupela kisim Siraveru.
(14) oire
okay
vaiterei
PPRO.2.DL.M
reo-aro
talk-POSS
uvu-ere-va
hear-3DL.F-RP
oa
RPRO.3.SG.N
iava
ABL
viku-ere-va
go.to.garden-3DL.F-RP
Okay, the two of them heard his talk and went to the garden.
Orait, tupela i bin harim tok bilong tupela na tupela i go long gaden.
(15) uva
so
ava-ere-i-epa
go-3DL.F-EPEN-RP
The two of them went.

Na tupela i bin go.
283
(16) uva
so
ora-reo-pa-ere-i-epa
RR-talk-CONT-3DL.F-EPEN-RP
And the two of them said to one another,

Na tupela i bin toktok.
(17) vo-kuio-re
SING-round-ALL
vegei
PPRO.1.DL.EXCL
aru-si-vo
order-3DL.M-IP
The two of them told us about the taro.

Tupela i salim mitupela long kisim dispela taro.
(18) uva
so
viapau
NEG
oisio
like
vo-kuio
SING-round
ou-ere-va
get-3DL.F-RP
Siraveru
name
kuio
round
And they didnt get this taro, the Siraveru taro,
Na tupela i no bin kisim dispela taro Siraveru,
(19) ari
but
Vatevu
name
kuio
round
ou-ere-va
get-3DL.F-RP
but the two of them got the Matevu taro.

tasol tupela i kisim Matevu,
(20) vo-kuio
SING-round
oa-pa
RPRO.3.SG.N-BEN
vairei
PPRO.2/3.DL.F
vatatopo-pie-raga-si-va
ready-CAUS-only-3DL.M-RP
the taro about which they [the parents] told them [the daughers] about.
dispela taro we tupela i bin tok lukaut nating long tupela.
(21) teapi
lest
Vatevu
name
kuio
round
ou-pa-ere-vere
get-CONT-3DL.F-NF
You two musnt get the Matevu taro.
Yutupela no ken kisim Matevu,
(22) ari
but
Siraveru
name
kuio
round
ou-ere-ve
get-3DL.F-SUB
but you two should get the Siraveru taro.
tasol bai yutupela kisim Siraveru.
(23) ovoi-ei
nish-PRES
Done.
Em inap.
(24) voa-vi-va
here-DIM-ABL
kare-ere-i-epa
return-3DL.F-EPEN-RP
vo-kuio-va
SING-round-ABL
ato-ia-re
harvest-LOC-ALL
From here the two of them return with the taro to the village.
284
Long dispela taim tasol tupela i bin karim dispela taro i go wantaim long ples.
2
(25) uva
so
vo-kuio-va
SING-round-ABL
koata-ere-i-epa
enter-3DL.F-EPEN-RP
And the two of them went inside with the taro.

Na tupela i bin go insait wantaim.
(26) oire
okay
aite-toarei
father-
vo-kuio
SING-round
evei-si-va
recognize-3DL.M-RP
Vatevu
name
kuio
round
Okay, the two parents recognized the taro, the Matevu taro.
Orait, tupela papa mama i bin luksave long dispela taro Matevu.
(27) uva
so
oisio
like
pura-si-epa
say-3DL.M-RP
And they said,

Na tupela i bin tok olsem,
(28) viapau
NEG
oisio
like
vo-kuio-re
SING-round-ALL
vei
PPRO.2.DL
tavi-ve-vo
tell-1DL-IP
We didnt tell you two about this taro.

Mitupela i no bin tokim yutupela long dispela taro.
(29) ari
but
Siraveru
name
kuio-re
round-ALL
vei
PPRO.2.DL
tavi-ve-vo
tell-1DL-IP
No, we told you two about the Siraveru taro.

Tasol mitupela tokim yutupela long Siraveru.
(30) ari
but
vuri-a
wrong-SG.N
pura-ere
make-2DL.F
But you two did a bad thing.
Tasol yutpela i wokim rong.
(31) uva
so
vairei
PPRO.2/3.DL.F
kopii-pie-si-va
die-CAUS-3DL.M-RP
And they killed the two of them.

???
3
2
In the original, vokuiova is followed by vokuio. This inexplicable repetition is treated as a typsetting error
here.
3
Something is missing in this sentence, since it is clear from the rest of the story that the parents did not in fact
kill the children.
285
(32) oire
okay
voa-va
here-ABL
uusi-a-epa
sleep-3PL
-RP
Okay, they slept.

Orait, ol i slip.
(33) uva
so
rirovira
big-time
sirao-pa-ere-i-epa
pity-CONT-3DL.F-EPEN-RP
The two of them were very sorry.

Na tupela i bin sori tru.
(34) apeisi-vai
how-INDEF
pie-ve
do-1DL
What do we do?
Bai mitupela mekim wanem?
(35) uva
so
vegei-vi-pa
PPRO.1.DL.EXCL-DIM-BEN
riro-a
big-SG.N
vate-si
give-3DL.M
sirao-a
pity-SG.N
Vatevu
name
kuio-pa
round-BEN
vegei
PPRO.1.DL.EXCL
upo-pa-oro
hit-CONT-DEP.SIM
The two of them will make us very sorry for the Matevu taro by beating us.
Na tupela i givim bikpela sori long mitupela long taro Matevu.
(36) viku-si-va
go.to.garden-3DL.M-RP
vairei
PPRO.2/3.DL.F
arova
without
voo
here
uvare
because
avi-epa
sunset-RP
The two of them went to the garden without them at dawn.

Na tupela i lusim tupela i go long gaden taim i tulait.
(37) oa
RPRO.3.SG.N
iava
ABL
sirao-pa-oro
pity-CONT-DEP.SIM
tou-pa-ere-ve
be-CONT-3DL.F-SUB
arakasi-aro
deserted-POSS
erava-ia
song-LOC
koova-pa-oro
sing-CONT-DEP.SIM
vo-kuio-rei
SING-round-DL.CL
va
PPRO.3.SG.N
vaisi-pa-oro
call-CONT-DEP.SIM
opo
taro
kuio-rei
round-DL.CL
???
Long dispela tupela i bin stap na sori long ples
(38) oarea-ia
RPRO.3.DL.N-LOC
era-pa-oro
sing-CONT-DEP.SIM
oisio
like
pura-ere-i-epa
say-3DL.F-3PL
-RP
Singing about it, they said,

Tupela i bin singsing long dispela tupela na tok olsem,
286
(39) Vatevu
name
kuio-pa
round-BEN
vei
PRO.PER.2.DL
upo-re
hit-3SG.M
aite
father
vaiterei
PRO.PER.2.DL.M
ora
and
aako
mother
Father will hit you for the Matevu taro, father and mother.
Long Matevu papa i paitim mitupela wantaim mama.
(40) ari
but
aue
CONN
Siraveru
name
kuio
round
ou-ve-vo-ri
get-1DL-IP
-2SG
oa-re
RPRO.3.SG.N-ALL
vegei
PPRO.1.DL.EXCL
tavi-raga-re-vo
tell-only-3SG.M
-IP
But we should have gotten the Siraveru taro which they told us about.
Tasol mitupela i mas bin kisim Siraveru em i bin tokim mitupela long em.
4
(41) uva
so
varei-ia
DEM.MED.DL.N-LOC
koova-pa-oro
sing-CONT-DEP.SIM
korovo
oil
pura-ere-va
make-3DL.F-RP
And they made oil singing about the two of them.

Tupela i singsing long tupela singsing na mekim oil bilong kokonas.
(42) reasi-pa-ei
be.disinclined-CONT-PRES
ra
COMP
voo
here
raga
only
tou-pa-oro
be-CONT-DEP.SIM
ora-sirao-pie-pa-ve
RR-pity-CAUS-CONT-1DL
Its no good for us to be here making each other feel bad.
I no gutpela long mitupela stap tasol long hia na mekim mitupela yet sori.
(43) ari
but
vearo-pa-ei
good-CONT-PRES
ra
COMP
tauai-vai-re
distant-INDEF-ALL
ava-ve
go-1DL
And it is good if we go far away.
tasol em i gutpela sapos mitupela i go longwe.
(44) oire
okay
korovo
oil
ovi
liquid
pura-ere-va
make-3DL.F-RP
va
PPRO.3.SG.N
ovoi-ere-voi-va
nish-3DL.F-PRES
-RP
orapura-ere-i-epa
appear-3DL.F-3PL
-RP
The two of them made coconut oil and nished putting it on each other.
Orait, tupela i bin wokim oil bilong kokonas na taim tupela i redim pinis, tupela i bin
putim long skin bilong tupela yet.
(45) uva
so
oravasie-ere-i-epa
leave-3DL.F-EPEN-RP
oira-ia
PPRO.3.SG.F-LOC
era-pa-oro
sing-CONT-DEP.SIM
erava
song
And the two of them left singing a song.
Na tupela i stat wakabaut na singim dispela singsing.
4
The nal sufx -ri on the verb ou is unrecognized.
287
(46) uva
so
uva-vu-va
so-ALT-ABL
avu-to
grandparent-SG.M
vairei
PPRO.2/3.DL.F
uvu-re-va
hear-3SG.M
-RP
osia
as
oira-ia
PPRO.3.SG.F-LOC
era-pa-ere-i-epa
sing-CONT-3DL.F-EPEN-RP
And their grandfather heard the two of them there as they sang it (the song).
Long narapela hap bubu man i bin harim tupela.
5
(47) era-pa-oro
sing-CONT-DEP.SIM
ava-pa-ere-i-epa
go-CONT-3DL.F-EPEN-RP
The two of them sang as they went.

Tupela i singsing i go.
(48) uva
so
vairei
PPRO.2/3.DL.F
iare
towards
vusi-re-va
erupt-3SG.M
-RP
avu-rirei
grandchild-DL.F
oisio
like
He appeared to his two granddaughters,
Na em i bin go autsait long tupela bubu meri bilong em,
(49) ovu
where
iare
towards
ava-pa-ere-i-ei
Where are you two of you going?

Yutupela i go we?
(50) ovuvaia
No!
Nowhere.
Nogat hap.
(51) ari
but
vegei
PPRO.1.DL.EXCL
upo-si-vo
hit-3DL.M-IP
aite
father
vaio
DL.ANIM
But our parents hit us.
Tasol tupela papa i paitim mitupela.
(52) uva
so
riro-vira
big-ADV
sirao-pa-oro
pity-CONT-DEP.SIM
ava-pa-ve-i-ei
go-CONT-1DL-EPEN-PRES
We are going feeling very sorry.

Na mitupela i sori tru na mitupela i go.
(53) uva
so
sirao
pity
isi
round
raga
only
uutu-ro-epa
follow-3SG.M
-RP
vairei
PPRO.2/3.DL.F
sirova
behind
And this sorry is following behind us.
Na long dispela bikpela sori tasol na em i bin bihainim tupela i go.
5
In the original, the verb form provided was uvareva; however, the verb root uva does not exist. This is treated
as a typo and corrected here.
288
(54) oire
okay
voka
walk
kata
exhaust
pura-re-va
make-3SG.M
-RP
He made an exhausting walk.

Orait, em bin wakabaut na em i bin skin i dai.
6
(55) uva
so
vairei
PPRO.2/3.DL.F
sirova
behind
uutu-pa-ro-epa
follow-CONT-3SG.M
-RP
And he followed behind them.

Na em i bin bihainim tupela i go.
7
(56) viapau
NEG
oisio
like
uvui-pa-ro-epa
be.able-CONT-3SG.M
-RP
oisio
like
ra
COMP
voka-pa-re-ve
walk-CONT-3SG.M
-SUB
He wasnt able to walk.
Nogat em i no bin inap olsem bai em i wakabaut.
(57) oa
RPRO.3.SG.N
iava
ABL
rera
PPRO.3.SG.M
kapokaporo-ere-va
grip-3DL.F-RP
voa
here
raiva-ro
road-PL.CL
Because of this they held him on the road.
Long dispela tupela i bin holim em long saitsait na go long rot.
(58) uva
so
avaka-va
ocean-SG.F
iare
towards
vusi-ere-va
erupt-3DL.F-RP
rera-va
PPRO.3.SG.M-ABL
oira
PPRO.3.SG.F
raga-ia
only-LOC
kova-pa-oro
grow-CONT-DEP.SIM
The two of them arrived at the ocean with him singing just this
Na tupela i kamap long nambis wantaim em, na singim dispela singsing.
(59) osia
as
rera-vi
PPRO.3.SG.M-DIM
kopii-ro-epa
die-3SG.M
-RP
vo-rogara
SING-sand
ua
CLASS
as the poor one died on the beach.
long taim trangu i bin dai long arere long nambis.
(60) uva
so
rera-va
PPRO.3.SG.M-ABL
ava-pa-ere-i-epa
The two of them went with him,

Na tupela i bin go wantaim daiman karim em tasol.
(61) vo-kopii
SING-die
raga-ia
only-LOC
kae-raga-pa-oro
carry-only-CONT-DEP.SIM
ava-pa-ere-i-epa
they went just carrying the dead man.

None
6
The form kata appears to function as a noun or classier here, but this usage is unattested elsewhere.
7
The verb root uutu is spelled as utu in the original; however, its initial vowel is long.
289
(62) uva
so
reasi-oro
be.disinclined-DEP.SIM
uva-vu
so-ALT
rera
PPRO.3.SG.M
tova-ere-va
bury-3DL.F-RP
vairei
PPRO.2/3.DL.F
raga
only
ava-oro
go-DEP.SIM
And the two of them buried him and went.
Na tupela i bin les na planim em long wanpela hap na tupela tasol i bin go.
(63) uva
so
gau-pa-oro
cry-CONT-DEP.SIM
ava-pa-ere-i-epa
And they cried as they went.

Na tupela i krai i go.
(64) aako
mother
vaio-vi
DL.ANIM-DIM
vegei
PPRO.1.DL.EXCL
viki-si-vo
throw.away-3DL.M-IP
voraro-re
around-ALL
Our parents threw us away around here.
Tupela mama ol i troim mitupela nabaut long hia.
(65) ovoi-ei
nish-PRES
Done.
Em inap.
(66) oire
okay
voa-va
here-ABL
keke-ere-va
look.at-3DL.F-RP
uva
so
oisioa
always
vuri-to
bad-SG.M
tou-pa-re-ve
be-CONT-3SG.M
-SUB
roo
DEM.PROX.SG.M
ira
RPRO.3.SG.M
oisioa
always
oira
PPRO.3.SG.F
aio-pa-ro
eat-CONT-3SG.M
Okay, from there the two of them looked and a bad man who always ate people was
there.
Orait, tupela i bin lukim hap dispela man nogut i bin save stap em i save kaikai ol man.
8
(67) uva
so
voa-va
here-ABL
vo-pouka
SING-lean
keke-ere-va
look.at-3DL.F-RP
evao
tree
pouka
lean
va
PPRO.3.SG.N
vaisi-aro
name-POSS
okaoto
talis
pouka
lean
And then the two of them saw a bent-over tree, the name of which was talisa.
Na bihain tupela lukim dispela diwai em i krungut, nem bilong dispela em talisia.
(68) oire
okay
vo-rao-ia
SING-branch-LOC
ava-ere-i-epa
go-3DL.F-EPEN-RP
Okay, the went on this branch.

Orait, tupela i go antap long dispela han diwai.
8
The original text contains a typo: voavo instead of voava.
290
(69) vo-rao-ia
SING-branch-LOC
ava-ere-i-epa
go-3DL.F-EPEN-RP
pouka
lean
rao
branch
They went on this branch, the leaning branch.
Tupela i go antap long dispela han diwai krungut.
(70) uva
so
voa-va
here-ABL
era-pa-oro
sing-CONT-DEP.SIM
vo-rao
SING-branch
ivara-ia
on-LOC
tou-pa-ere-va
be-CONT-3DL.F-RP
And they were on top of the branch singing.

Na tupela i singsing taim tupela i stap antap long dispela han.
(71) uva
so
urio-ro-epa
come-3SG.M
-RP
rera
PPRO.3.SG.M
roo
DEM.PROX.SG.M
ira
RPRO.3.SG.M
oisioa
always
oirara
people
aio-pa-re-ve
eat-CONT-3SG.M
-SUB
rera
PPRO.3.SG.M
vaisi-aro
name-POSS
vakuvaku
scoff
And the one who came, the one who was always eating people, his name was Vakuvaku.
Na dispela man ya i bin kam em i save kaikai ol man.
9
(72) rera
PPRO.3.SG.M
vaisi-aro
name-POSS
vakuvaku
scoff
tugarato
spirit
The spirit, his name was Vakuvaku.
Nem bilong masalai, em Vakuvaku.
(73) uva
so
rera
PPRO.3.SG.M
tue-ere-va
wait-3DL.F-RP
roo-ia
DEM.PROX.SG.M-LOC
ira
RPRO.3.SG.M
urio-ro-ei
come-3SG.M
-PRES
The two of them waited for him to come.

Na tupela i lukim em, em i kam nau.
(74) uva
so
orapiopio-pa-ere-i-epa
argue-CONT-3DL.F-EPEN-RP
And the two of them talked,

Na tupela i wok long toktok long tupela yet.
(75) tepa
Hey!
vii
PPRO.2.SG
ira
RPRO.3.SG.M
rovo-u
start-2SG
Hey, you go ahead rst.

Goan yu go pas nau.
9
The word vakuvaku also means cynic or skeptic. It is unclear whether the two meanings are somehow
related.
291
(76) uva
so
oisio
like
pura-pa-o-epa
say-CONT-3SG.F
-RP
And she said,

Na wanpela meri i tok olsem,
(77) oari
DEM.DIST.SG.F
oisio
like
vii
PPRO.2.SG
What about you?
Na olsem wanem yu.
(78) oire
okay
iria-vu
RPRO.3.SG.F-ALT
topogovira
recklessly
oraviki-o-epa
jump-3SG.F
-RP
Okay, one reckessly jumped.

Orait, wanpela i bin kalap i go stret.
(79) teapi
lest
vorevira
backward
sirao-ve
pity-1DL
aite
father
vaio-re
DL.ANIM-ALL
ra
COMP
vorevira
backward
vore-ve
return-1DL
Lest we feel sorrow for our parents and go back.
Nogut mitupela sori bek long tupela papa na bai mitupela i go bek.
(80) uva
so
oira
PPRO.3.SG.F
kopa-oro
swallow-DEP.SIM
revasiva
blood
oe-re-va
vomit-3SG.M
-RP
vorevira
backward
okaoto
talis
kavusi-sia
spit.out-DEP.SEQ
And when he [the shark] swallowed the blood, he threw it back up on the talisa tree.
Na taim em i daunim meri, em trautim blut i spet i go bek long talisia.
(81) uva
so
ovoio-pa-va
be.last-DERIV-SG.F
uutu-o-epa
follow-3SG.F
-RP
And the last woman followed.

Na laspela meri i bin go bihain.
(82) uva
so
oisio
like
ita
again
pie-re-va
do-3SG.M
-RP
oira
PPRO.3.SG.F
kopa-oro
swallow-DEP.SIM
rera
PPRO.3.SG.M
kavusi-re-va
spit.out-3SG.M
-RP
vao
DEM.PROX.3.SG.N
oa
RPRO.3.SG.N
iava
ABL
He did it again, he swallowed her and spat it out.
Na em i bin mekim olsem gen, em i bin daunim meri na spetim em.
(83) ovoi-ei
nish-PRES
Done.
Em inap.
292
(84) rovi-ro-epa
mix-3SG.M
-RP
okaoto
talis
vosia
when
veve-pe
ripe-SUB
vo-guruva
SING-leaf
ra
COMP
oira
PRO.PER.3.SG.F
pura-ve
make-1DL
revasiva
blood
oo
DEM.PROX.SG.F
iria
PRO.REL.3.SG.F
pura-ere-va
make-3DL.F-
voo
here
vorevira
backward
vairei
PRO.PER.2/3.DL.F
kavusi-oro
spit.out-DEP.SIM
???
The talisa tree mixes when its leaves ripen and the blood that was spit out makes it red.
10
(85) oire
okay
eva
DEM.MED.SG.N
oira
PPRO.3.SG.F
opesipie-aro-ia
nish-POSS-LOC
vo-siposipo
SING-story
Okay, that is the end of the story.
Orait, pinis bilong dispela meri em pinis bilong dispela stori,
(86) opesipie-aro-ia
nish-POSS-LOC
aue
CONN
iava
ABL
oo
DEM.PROX.SG.F
erava
song
vo-siposipo
SING-story
iava
ABL
vairei
PPRO.2/3.DL.F
iava
ABL
evairei
DEM.MED.DL.F
aireia-pa
DEM.MED.DL.F-BEN
sirao-a
pity-SG.N
vate-si-va
give-3DL.M-RP
aite-toarei
father-
The end of the song, the song about the two of them when both parents gave sorrow to
the two girls.
Dispela singsing em i long stori ya long tupela taim tupela papa i bin givim sori long
tupela meri.
C.2 Matevu, Version 2
This version of the folk tale was recorded in the village of Togarao in 2003 and then transcribed
by Timothy Taureviri and translated into Tok Pisin by Sera Mon before being entered into
Shoebox and translated into English by the author. The narrator of the story is Caleb Karuru
(shown in Figure 1.2), an older speaker of Rotokas who also worked with Irwin Firchow.
(1) oire
okay
erao-pie-pa
two-CAUS-DERIV
siposipo-a
story-SG.N
vao
DEM.PROX.3.SG.N
oa
RPRO.3.SG.N
pura-pa-a-voi
make-CONT-1SG
-PRES
Okay, this is the second story I want to tell.

Orait, em numba tu stori mi laik wokim.
(2) oavao-vu
family-ALT
iava
ABL
oisoa
always
tou-pa-i-ve
be-CONT-3PL
-SUB
Its about a family that existed.
Long wanpela pamili i bin save i stap.
10
This sentence was overlooked by consultants when the text was translated into Tok Pisin.
293
(3) o-avuka-rei-vu-ia
SPEC-age-DL.CL-ALT-LOC
va
PPRO.3.SG.N
aiterei-ia
PPRO.3.DL.M-LOC
oisoa
always
tavauru-rirei
teenage.girl-DL.F
tapo
also
oisio
like
tou-pa-si
be-CONT-3DL.M
A couple with two young girls.
Long tupela marit ol i bin save stap wantaim tupela yangpela pikinini meri.
(4) ovii-rirei
child-DL.F
Two daughters.
Tupela pikinini meri.
(5) oire
okay
kovoa-ia
garden-LOC
opo
taro
kovo
garden
rera
PPRO.3.SG.M
vo-kovo-aro
SING-garden-POSS
eva
DEM.MED.SG.N
None
Orait, long dispela gaden bilong em.
(6) vo-aao
SING-family
vo-kovo-aro
SING-garden-POSS
opo
taro
kovo
garden
raga
only
pura-pa-i-ve
make-CONT-3PL
-SUB
opo
taro
kovo
garden
raga
only
pura-pa-i-ve
make-CONT-3PL
-SUB
This family, they just worked the taro garden.
Dispela pamili em wok bilong ol long wokim gaden taro tasol.
(7) viapau
NEG
oisio
like
oavuavu-vai
something-INDEF
ari
but
va
PPRO.3.SG.N
raga
only
opo
taro
There wasnt anything else, just taro.
Nogat narapela samting, tasol em taro tasol.
(8) oire
okay
vosia
when
vo-kovo
SING-garden
siovara-ia
inside-LOC
vo-kuio-rei
SING-round-DL.CL
tou-pa-i-ve
be-CONT-3PL
-SUB
Okay, inside of this garden, there were two taro.
Orait, na insait long dispela gaden tupela taro i bin save i stap.
(9) virapie
transfer
kuio-rei-vi
round-DL.CL-DIM
oarea
RPRO.3.DL.N
oisoa
always
vaisi-pa-i-ve
call-CONT-3PL
-SUB
oisio
like
Vatevu
name
ora
and
Siraveru
name
These two taro, they called them Vatevu and Siraveru.
Dispela tupela taro hia ol i save kolim olsem Vatevu wantaim Siraveru
294
(10) evo
DEM.N
kuio-rei
round-DL.CL
oarea
RPRO.3.DL.N
pau-re-va
build-3SG.M
-RP
These two taro that he planted.

Dispela tupela taro em i bin planim.
(11) oire
okay
oisoa
always
tou-pa-i-ve
be-CONT-3PL
-SUB
Okay, they were there.
Orait, ol i bin save stap.
(12) uva
so
riro-epa
grow up-RP
vo-opo
SING-taro
kovo
garden
siovara-ia
inside-LOC
They grew big inside of the taro garden.
Na tupela taro i bin kamap bikpela insait long gaden taro.
(13) uva
so
o-voki-vu-ia
SPEC-day-ALT-LOC
vairei
PPRO.3.DL.F
tavi-pa-si-va
tell-CONT-3DL.M-RP
Okay, one day the two of them talked to the two of them,
Orait, na tupela i bin tokim tupela pikinini bilong tupela olsem,
(14) ai
hey
kovo-sia
work-DEP.SEQ
ava-pa-ere-i-ei
opo
taro
kovo-ia
garden-LOC
kovo-sia
work-DEP.SEQ
ava-pa-ere-i-ei
opo
taro
kovo-ia
garden-LOC
Hey, you two go work in the taro garden, you two go work in the taro garden.
Bai yutupela go wok long gaden taro, bai yutupela go wok long garden taro.
(15) oire
okay
ava-ere-i-epa
go-3DL.F-3PL
-RP
ava-ere-i-epa
go-3DL.F-EPEN-RP
ava-ere-i-epa
go-3DL.F-EPEN-RP
Okay, the two of them went, they went, they went.

Orait, tupela i bin go, tupela i bin go, tupela i bin go.
(16) oisoa
always
oisiopie-pa-ere
pretend-CONT-3DL.F
voki-ara
day-PL.N
rutu-ia
very-LOC
oisoa
always
oisiopie-pa-ere
pretend-CONT-3DL.F
voki-ara
day-PL.N
rutu-ia
very-LOC
kovo-pa
garden-BEN
The two of them always did this, they always did this for the garden.
Na tupela i bin save mekim olsem olgeta taim, [???]
11
(17) voki-ara
day-PL.N
rutu-ia
very-LOC
kovo-pa-sia
work-CONT-DEP.SEQ
ava-pa-ere
go-CONT-3DL.F
Every day the two of them went to work.
Olgeta dei tupela i bin save go wok.
11
Not sure about the analysis of last word (kovopa). Is it really a noun?
295
(18) uva
so
opo
taro
kuio
round
tate-ere-va
extract-3DL.F-RP
And the two of them removed (dug up) a taro.

Na tupela i bin kamautim wanpela taro.
(19) vo-kuio-rei
SING-round-DL.CL
iava
ABL
virapie
transfer
kuio-rei
round-DL.CL
rera
PPRO.3.SG.M
varei-aro
DEM.MED.DL.N-POSS
oarea
RPRO.3.DL.N
oisoa
always
virapievira
transfered-like
toki-pa-re-ve
look.after-CONT-3SG.M
-SUB
rera
PPRO.3.SG.M
???
Long dispela tupela taro bilong em em i bin save lukautim narakain.
(20) oire
okay
vosia
if
varei-va
DEM.MED.DL.N-ABL
kare-ere-i-epa
return-3DL.F-EPEN-
When the two of them returned with these two (taro),
Orait, na taim tupela i bin karim i go,
(21) uva
so
varei
DEM.MED.DL.N
evei-re-va
recognize-3SG.M
-RP
rera
PPRO.3.SG.M
aite-to
father-SG.M
their father recognized the two (taro).
Olsem na papa bilong em i bin luksave long tupela taro.
(22) ai
hey
vairei-o
PPRO.3.DL.F-?
apeisi
how
oisio
like
ragavira
just
keke-pa-ei
look-CONT-PRES
vo-kuio-ia
SING-round-LOC
Hey, why do these two taro look this way?
Eh, olsem wanem na dispela tupela taro i luk olsem?
(23) oisio
like
osia
as
vo-kuio-rei-o
SING-round-DL.CL-?
oarea
RPRO.3.DL.N
iava
POST
vei
PPRO.2.DL
tavi-pa-a-veira
tell-CONT-1SG
-HAB.ANIM
These are the two taro that I am always telling you about.
Em olsem dispela tupela taro mi bin save tokim yutupela long em.
(24) aure
Yes
evoa
DEIC.MED
vairei-re
PPRO.3.DL.F-ALL
reo-pa-si-epa
talk-CONT-3DL.M-RP
Yes, the two of them told them,

Tupela i bin tokim tupela.
(25) vuri-a
wrong-SG.N
pura-ere-voi
make-3DL.F-PRES
rutu
very
the two of you did very bad.
Yutupela i wokim pasin nogut.
296
(26) uva
so
opo-a
taro-SG.N
tate-ere-voi
extract-3DL.F-PRES
virapie
transfer
kuio
round
rutu
very
vao-ia
DEM.PROX.3.SG.N-LOC
The two of them took out this taro that was truly different.
Na yutupela i kamautim dispela taro em i narakain tru. [Not sure if the last word is
properly analyzed (check transcription).]
(27) oire
okay
vairei-re
PPRO.3.DL.F-ALL
kasipu-si-epa
angry-3DL.M-RP
vaiterei
PPRO.2.DL.M
rutu
very
Okay, the two of them (the parents) were really angry with the two of them (the
children).
Orait, tupela wantaim i bin krosim tupela.
(28) uva
so
ritu-pa-oro
ashamed-CONT-DEP.SIM
uusi-sia
sleep-DEP.SEQ
koata-ere-i-epa
enter-3DL.F-EPEN-RP
So two of them went inside to sleep in shame.

Olsem na tupela i bin kros na tupela i bin go insait
(29) viapau
NEG
aio-ere-va
eat-3DL.F-RP
The two of them didnt eat.

taim ol i no kaikai.
(30) ari
but
uusi-raga-sia
sleep-only-DEP.SEQ
koata-ere-i-epa
enter-3DL.F-3PL
-RP
But the two of them went inside and just slept.

Em tupela i bin go insait na slip nating.
(31) ora-reo-pa-ere-i-epa
RR-talk-CONT-3DL.F-EPEN-RP
The two of them talked,

Na tupela i bin toktok,
(32) apeisi
how
ragavira
only
pie-pa-ve-voi
do-CONT-1DL-PRES
What will we do?

Bai mitupela i mekim wanem?
(33) ee
hey
raga
only
ava-pa-ve-i-ei
Should the two of us go?

Bai mitupela i go?
297
(34) ava-pa-ve-i-ei
rara
later
ritu-pa-oro
disgusted-CONT-DEP.SIM
Were embarassed and well go.
Bai mitupela i kros na i go.
(35) oire
okay
uusi-ere-epa
sleep-3DL.F-RP
uusi-ere-epa
sleep-3DL.F-RP
uusi-ere-epa
sleep-3DL.F-RP
The two of them slept, the two of them slept, the two of them slept.
Orait, tupela i bin slip, tupela i bin slip, tupela i bin slip.
(36) uva
so
voari
before
rutu
very
vokipakou
morning
rutu
very
tore-ere-i-epa
stand-3DL.F-EPEN-RP
In the early morning the two of them got up.

Olsem na long moning tru tupela i bin kirap.
(37) oravasike-ere-i-ei
leave-3DL.F-EPEN-PRES
The two of them left.

Tupela i bin kirap i go.
(38) vasike-ere-i-ei
leave-3DL.F-EPEN-PRES
voka-pa-oro
walk-CONT-DEP.SIM
The two of them left on foot.
Tupela i bin kirap i go wokabaut.
(39) ava-ere-i-ei
go-3DL.F-EPEN-PRES
The two of them went,

Tupela i bin go,
(40) ava-ere-i-ei
go-3DL.F-EPEN-PRES
the two of them went.

tupela i bin go
(41) ai
hey
aite
father
vaio
DL.ANIM
vegei-re
PPRO.1.DL.EXCL-ALL
reo-pa-si-e
talk-CONT-3DL.M-IP
opo
taro
kuio-rei-pa
round-DL.CL-BEN
Vatevu
name
kuio-rei
round-DL.CL
ora
and
Siraveru
name
kuio
round
Hey, our parents talked to the two of us about the two taro, Vatevu and Siraveru.
Ae, tupela papa i krosim mitupela long tupela taro, Vatevu wantaim Siraveru.
298
(42) oire
okay
iria-vu
RPRO.3.SG.F-ALT
koova-va
song-SG.F
iria-ia
RPRO.3.SG.F-LOC
kovo-pa-oro
work-CONT-DEP.SIM
voka-pa-ere-va
walk-CONT-3DL.F-RP
raiva-ro
road-PL.CL
Okay, this song they sang as they walked on the road.
Orait, wanpela singsing tupela i bin singim taim tupela i wakabaut i go long rot.
(43) iria-ia
RPRO.3.SG.F-LOC
oisoa
always
koova-pa-a-ve
sing-CONT-3PL
-SUB
voari
before
tuariri
before
vo-siposipo
SING-story
pura-pa-oro
say-CONT-DEP.SIM
They always sang this long ago telling this story.
Dispela singsing em ol i bin save singim bipo taim ol i wokim dispela stori.
(44) uva
so
oira-ia
PPRO.3.SG.F-LOC
koova-pa-ere-i-epa
oisio
like
ragavira
just
So the two of them sang this song just like this,
Na tupela i bin singim dispela singsing olsem,
(45) SONG
(46) oire
okay
ava-pa-ere-i-epa
ava-pa-ere-i-epa
ora-sirao-pie-pa-oro
RR-pity-CAUS-CONT-DEP.SIM
The two of them went, feeling sorry for themselves.
Orait, tupela i bin go, tupela i bin go, na mekim sori tupela yet.
(47) gau-pa-oro
cry-CONT-DEP.SIM
ava-pa-ere-i-epa
vo-raiva-ro
SING-road-PL.CL
The two of them went crying along the road.
Tupela i bin krai i go long rot.
(48) osia
as
vairei
PPRO.3.DL.F
vore-raga-pa-oro
return-only-CONT-DEP.SIM
uutu-pa-ro-epa
follow-CONT-3SG.M
-RP
He was tired following them.

Olsem na em i bin tait long pasim tupela.
(49) vairei
PPRO.3.DL.F
vore-raga-pa-oro
return-only-CONT-DEP.SIM
uutu-pa-ro-epa
follow-CONT-3SG.M
-RP
osia
as
viapau
NEG
rutu
very
He was tired of following the two of them.
Em i bin tait long pasim tupela tasol nogat tru.
299
(50) uva
so
vairei
PPRO.3.DL.F
vuripie-si-va
ruin-3DL.M-RP
rutu
very
vo-avuka-rei
SING-age-DL.CL
vairei-re
PPRO.3.DL.F-ALL
kasipu-pa-oro
angry-CONT-DEP.SIM
The two of them harmed the two of them when they got angry.
Na tupela i bin bagarapim tupela taim tupela i krosim ol.
(51) ava-pa-ere-i-epa
ava-pa-ere-i-epa
avakava-re
ocean-ALL
tara-pa-oro
seek-CONT-DEP.SIM
ava-pa-ere-i-epa
The two of them went, they went to the ocean.

Tupela i bin go, tupela i bin go, tupela i bin go bilong painim solwara.
(52) avakava-re
ocean-ALL
tara-pa-oro
seek-CONT-DEP.SIM
ava-pa-ere-i-epa
The two of them went to nd the ocean.

Tupela i bin go bilong painim solwara.
(53) oira-ia
PPRO.3.SG.F-LOC
koova-pa-oro
sing-CONT-DEP.SIM
ra
COMP
uva-vi-vu
so-DIM-ALT
ita
again
oira
PPRO.3.SG.F
kaepie-re
raise-3SG.M
ita
again
The two of them sang and wherever they went, they two sang again.
Tupela i bin singsing na go wanem hap tupela i kamap bai tupela singim gen.
(54) SONG
(55) ava-pa-ere-i-epa
ava-pa-ere-i-epa
The two of them went, the two of them went.

Tupela i bin go,
(56) pukui-ia
hill-LOC
tugura-ere-i-epa
arrive-3DL.F-EPEN-RP
The two arrived on the hill.

Tupela i bin go kamap antap long maunten.
(57) atoi
village
vura-ere-va
look.at-3DL.F-RP
They looked at the village.

Na tupela lukluk i go bek long ples.
300
(58) ato-ia
harvest-LOC
vura-oro
look at-DEP.SIM
avau-ere-va
say goodbye-3DL.F-RP
voa
here
The two of them looked at their village and said good-bye.
Tupela i bin lukluk i go bek long ples na tok gutbai.
(59) ora-putepie-ere-i-epa
RR-overtake-3DL.F-3PL
-RP
The two of them went over (the mountain).

Na tupela bin kalap i go long hapsait.
(60) ava-ere-i-epa
go-3DL.F-EPEN-RP
ava-ere-i-epa
go-3DL.F-EPEN-RP
ava-ere-i-epa
go-3DL.F-EPEN-RP
ava-ere-i-epa
go-3DL.F-EPEN-RP
The two of them went, they went, they went, and they went.
Tupela i bin go, tupela i bin go, tupela i bin go, tupela i bin go.
(61) koova-ere-i-epa
sing-3DL.F-EPEN-RP
ita
again
The two of them sang again.
Tupela i bin singsing gen.
(62) SONG
(63) oire
okay
pou-ere-viro-epa
arrive-3DL.F-COMPL-RP
vo-avaka-va-ia
SING-ocean-SG.F-LOC
Okay, the two of them came to the ocean.
Orait, tupela i bin kamap long solwara.
(64) oire
okay
oavu-va
another-SG.F
oa
RPRO.3.SG.N
vaisi-pa-i-veira
call-CONT-3PL
-HAB.ANIM
oisio
like
okaoto-va
talis-SG.F
One tree, they call okaoto.
Orait, wanpela diwai ol i save kalim olsem talisa.
(65) o-pouka-ia
SPEC-lean-LOC
ereere-ere-i-epa
walk.across-3DL.F-EPEN-RP
The two of them walked up on the leaning (tree).

Tupela i bin kalap.
(66) ava-ere-i-epa
go-3DL.F-EPEN-RP
voa
here
pau-pa-ere-i-epa
sit-CONT-3DL.F-EPEN-RP
voa
here
pau-pa-oro
sit-CONT-DEP.SIM
koova-pa-ere-i-epa
The two of them went, they sat down, and they sang.
Tupela i bin go na tupela i bin sindaun. Tupela i bin sindaun na singsing.
301
(67) SONG
(68) pau-pa-ere-i-epa
sit-CONT-3DL.F-EPEN-RP
tue-pa-oro
wait-CONT-DEP.SIM
osia
as
riro-to
big-SG.M
siaka
shark
urio-ro-epa
come-3SG.M
-RP
urio-ro-epa
come-3SG.M
-RP
The two of them sat down and waited when one big shark came.
Tupela i bin sindaun na wait taim wanpela bikpela sak i bin kam.
(69) oire
okay
okaoto-va
talis-SG.F
reroaro
underneath
viri-pa-re-va
twist-CONT-3SG.M
-RP
Okay, he circled under the tree.

Orait, na em i bin raun undanit long talisa.
(70) viri-pa-re-va
twist-CONT-3SG.M
-RP
He went around
Em i bin raun
(71) vairei
PPRO.3.DL.F
gesi-re-va
smell-3SG.M
-RP
and smelled them.

taim em i smelim tupela.
(72) viri-pa-re-va
twist-CONT-3SG.M
-RP
He went around.
Em i raun.
(73) oire
okay
avaio-pa-va
rst-born-DERIV-SG.F
isiva
turn.back.on
oari
DEM.DIST.SG.F
tavi-pa-e-va
tell-CONT-3SG.F
-RP
kikoo-pa-va
second-born-DERIV-SG.F
Okay, the big sister told the little sister.
Orait, bikpela sista bilong em i bin tokim liklik sista bilong em.
(74) oraviki
jump
rovo-pa-u-ei
start-CONT-2SG
-PRES
vii
PPRO.2.SG
You jump off rst.
Bai yu kalap pastaim.
302
(75) oire
okay
ovoio-pa-vira
be.last-DERIV-ADV
koova-ere-i-epa
sing-3DL.F-EPEN-RP
oira-ia
PPRO.3.SG.F-LOC
koova-ere-i-epa
sing-3DL.F-EPEN-RP
Okay, for the last time the two of them sang this song.
Orait, na laspela taim tupela i bin singim dispela singsing.
(76) voa
here
oraviki-o-epa
jump-3SG.F
-RP
osia
as
siaka
shark
ira
RPRO.3.SG.M
oira
PPRO.3.SG.F
kopa-re-va
swallow-3SG.M
-RP
osia
as
revasiva
blood
iria
RPRO.3.SG.F
vorevira
backward
kae-o-viro-epa
carry-3SG.F
-COMPL-RP
okaoto-va
talis-SG.F
iare
towards
guruva-ro
leaf-PL.CL
iare
towards
Here the two of them lept as the shark eats her while her blood goes goes back onto the
leaves of the tree.
Long hap em i bin kalap na sak i bin daunim em taim blut i bin kalap i go antap long lip
bilong talisa.
(77) oire
okay
oira
PPRO.3.SG.F
aio-re-voi
eat-3SG.M
-PRES
oira
PPRO.3.SG.F
aio-re-voi
eat-3SG.M
-PRES
Okay, he ate her, he ate her.

Orait, em i kaikai em, em i kaikai em.
(78) oo
DEM.PROX.SG.F
avaio-pa-va
rst born-DERIV-SG.F
iria
RPRO.3.SG.F
tou-pa-e-va
be-CONT-3SG.F
-RP
This last one remained.

Dispela laspela i bin stap.
(79) vosia
when
oira
PPRO.3.SG.F
opesi-re-va
nish-3SG.M
-RP
uva
so
viri-pa-re-va
twist-CONT-3SG.M
-RP
voari
before
When he nished her, he twisted around again.
Taim em i pinis kaikai em na em i bin wok long raun.
(80) uva
so
tarai-o-epa
understand-3SG.F
-RP
oisio
like
opesi-o-e
nish-3SG.F
-IP
oraviki-o-ei
jump-3SG.F
-PRES
She knew that her sister was nished and jumped.

Na em i bin save olsem em i pinisim susa bilong em na em i bin kalap.
(81) oire
okay
eira
DEM.MED.SG.F
ita
again
revasi-aro
bleed-POSS
voari-re
before-ALL
okaoto
talis
guruva
leaf
iare
towards
kae-o-viro-ei
carry-3SG.F
-COMPL-PRES
Okay, the blood of this girl was carried back on top of the leaf of the tree again.
Orait, blut bilong dispela narapela meri em i go antap gen long lip bilong talisa.
303
(82) oire
okay
oisio
like
oisoa
always
va
PPRO.3.SG.N
aue-pa-i-ve
ignore-CONT-3PL
-SUB
Okay, so they would always think this way.
Orait, na ol i bin save tingting olsem,
(83) va
PPRO.3.SG.N
eva
DEM.MED.SG.N
siposipo-a
story-SG.N
opesi-aro
end-POSS
Thats the end of the story.
Em i pinis bilong stori.
(84) oire
okay
voa-va
here-ABL
reo-pa-ra-ei
talk-CONT-1SG
-PRES
aue
CONN
iava
ABL
okaoto-a-i
talis-?-
oisio
like
osia
as
pura-pa-ve
make-CONT-SUB
evairei
DEM.MED.DL.F
revasi-aro-a
bleed-POSS-
evairei
DEM.MED.DL.F
okaoto-a-ia
talis-SG.N-LOC
voto-ere-i-epa
stuck-3DL.F-EPEN-RP
Okay, Ill talk about the talisa ???.

Orait, mi laik toktok long talisa em olsem blut bilong tupela meri i bin pas long ol lip.
(85) oire
okay
vosia
if
okaoto
talis
keke-pa-ri
look.at-CONT-2SG
osia
as
veve-pa-ei
ripe-CONT-PRES
revasivira
bloody
Okay, if you look at this tree as it ripens and turns red, its the blood of two women.
Orait, sapos yu lukim talisa taim lip bilong em i red, em blut bilong tupela meri.
(86) oire
okay
eisi-vira
like.this-ADV
raga
only
osia
as
opesi-ei
nish-PRES
Thats how it ends.

Orait, em i pinis olsem.
(87) osia
as
vo-siposipo
SING-story
reo
talk
pura-a-voi
make-1SG
-PRES
As I work this story,

Olsem mi wokim dispela tupela stori.
(88) ragai
PPRO.1.SG
Caleb
name
Karuru
name
Im Caleb Karuru.
Mi, Caleb Karuru.
304
Bibliography
Aikhenvald, Alexandra Y. 2000. Classiers: A typology of noun classication devices. Studies
in Typology and Linguistic Theory. Oxford: Oxford University Press.
Aissen, Judith. 1999a. Agent focus and inverse in Tzotzil. Language 75:451485.
Aissen, Judith. 1999b. Markedness and subject choice in optimality theory. Natural Language
and Linguistic Theory 17:673711. Reprinted in Optimality-Theoretic Syntax, edited by
Geraldine Legendre, Jame Grimshaw, and Sten Vikner. Cambridge: MIT Press, 61-96.
Allen, Jerry, and Conrad Hurd. 1963. Languages of the Bougainville District. Ukarumpa:
Summer Institute of Linguistics.
Anderson, Stephen R. 1985. Typological distinctions in word formation. In Language Typology
and Syntactic Description, Volume 3: Grammatical Categories and the Lexicon, ed. Timothy
Shopen, 356. Cambridge: Cambridge University Press.
Anderson, Stephen R., and Edward L. Keenan. 1985. Deixis. In Language typology and syn-
tactic description, volume iii: Grammatical categories and the lexicon, ed. Timothy Shopen,
259308. Cambridge: Cambridge University Press.
Andrews, Avery. 1985. The major functions of the noun phrase. In Language Typology and
Syntactic Description, Volume I: Clause Structure, ed. Timothy Shopen, 62154. Cambridge:
Cambridge University Press.
Andrews, Avery. 2007. The major functions of the noun phrase. In Language Typology and Syn-
tactic Description, Volume I: Clause Structure, ed. Timothy Shopen, 132223. Cambridge:
Anward, Jan, Edith Moravcsik, and Leon Stassen. 1997. Parts of speech: A challenge for
typology. Linguistc Typology 1:167184.
Bauman, J.J. 1975. Pronouns and pronominal morphology in Tibeto-Burman. Doctoral Disser-
tation, University of California at Berkeley.
305
Beesley, Kenneth R., and Lauri Karttunen. 2003. Finite state morphology. Stanford: CSLI.
Bhat, D.N.S. 1991. Grammatical relations: the evidence against their necessity and universal-
ity. London/New York: Routledge.
Blevins, Juliette. 1995. The syllable in phonological theory. In The handbook of phonological
theory, ed. John Goldsmith, 206244. London: Basil Blackwell.
Bohnemeyer, J urgen, Melissa Bowerman, and Penelope Brown. 2001. Cut and break video
clips. In Field manual volume 2001, ed. Stephen C. Levinson and Nick J. Eneld. Nijmegen:
Max Planck Institute for Psycholinguistics.
de Bougainville, Louis Antoine. 1772. A voyage round the world. performed by ordrer of his
most Christian majesty in the years 1768, 1767, 1768 and 1769. London: J. Nourse and T.
Davies.
Bresnan, J., and J. Kanerva. 1988. Locative inversion in Chichewa: A case study of factorization
in grammar. Linguistic Inquiry 20:150.
Bresnan, Joan. 2001. Lexical-functional syntax. Oxford: Blackwell Publishers.
Bresnan, Joan, and Ronald M. Kaplan. 1982. Introduction: grammars as mental representations
of language. In The mental representation of grammatical relations, ed. Joan Bresnan, xvii
lii. Cambridge, MA: The MIT Press.
Bresnan, Joan, and Annie Zaenen. 1990. Deep unaccusativity in LFG. In Grammatical rela-
tions: A cross-theoretical perspective, ed. K. Dziwirek, P. Farrell, and E. M. Bikandi, 4557.
Stanford: CSLI Publications.
Bugenhagen, Robert D. 1993. The semantics of irrealis in Austronesian languages of Papua
New Guinea, a cross-linguistic study. In Topics in Austronesian linguistics, ed. Ger P.
Reesink, Semaian 11, 139. Leiden, The Netherlands: Vakgroep Talen en Culturen van
Zuidoost-Azi e en Oceani e, Rijksuniversiteit te Leiden.
Butt, Miriam, Tracy Holloway King, Mara-Eugenia Ni no, and Fr ed erique Segond. 1999. A
grammar writers cookbook. Stanford, California: CSLI.
Chung, Sandra, and Alan Timberlake. 1985. Tense, aspect, and mood. In Language Typology
and Syntactic Description, Volume III: Grammatical Categories and the Lexicon, ed. Timothy
Shopen, 202258. Cambridge: Cambridge University Press.
Comrie, Bernard. 1975. Causatives and universal grammar. In Transactions of the philological
society 1974, ed. M. Shibatani, 132. Oxford: Basil Blackwell.
306
Comrie, Bernard. 1976. The syntax of causative constructions: cross-language similarities
and divergences. In The grammar of causative constructions, ed. M. Shibatani, Syntax and
Semantics, Volume 6, 261312. New York: Academic Press.
Comrie, Bernard, ed. 1985a. Aspect. Cambridge: Cambridge University Press.
Comrie, Bernard. 1985b. Causative verb formation and other verb-deriving morphology. In
Language typology and syntactic description, volume 3: Grammatical categories and the
lexicon, ed. Timothy Shopen, 309347. Cambridge: Cambridge University Press.
Comrie, Bernard. 1989. Language Universals and Linguistic Typology. Chicago: University of
Chicago Press, second edition.
Croft, William. 2006. Word classes, parts of speech, and syntactic argumentation. Linguistic
Typology 9:431441.
Cruse, D.A. 1986. Lexical semantics. Cambridge: Cambridge University Press.
Cysouw, Michael. 2003. The paradigmatic structure of person marking. Oxford: Oxford
University Press.
Danziger, Eve. 1996. Split intransitivity and active-inactive patterning in Mopan Maya. Inter-
national Journal of American Linguistics 6:379414.
Davis, Anthony R. 2001. Linking by types in the hierarchical lexicon. Stanford: CSLI.
DeLancey, Scott. 1985. Agentivity in syntax. In Papers from the Parasession on Causatives
and Agentivity, ed. W.H. Eilfort, P.D. Kroeber, and K.L. Peterson, volume 21 of Chicago
Linguistics Society, 112.
Dixon, R. M. W. 1977. A grammar of Yidiny. Cambridge Studies in Lingusitics, 19. Cambridge:
Dixon, R. M. W. 1979. Ergativity. Language 55:59138.
Dixon, R. M. W. 1994. Ergativity. Cambridge: Cambridge University Press.
Dixon, R. M. W. 2004. Adjective classes: A cross-linguistic typology. Oxford: Oxford Univer-
sity Press.
Dixon, R.M.W. 1982. Where have all the adjectives gone?. Berlin: Mouton Publishers.
Dixon, R.M.W. 2000. A typology of causatives: form, syntax, and meaning. In Changing
valency, ed. R.M.W. Dixon and Alexandra Y. Aikhenvald, 3083. Cambridge: Cambridge
University Press.
307
Dixon, R.M.W., and Alexandra Y. Aikhenvald, ed. 2000. Changing valence: Case studies in
transitivity. Cambridge: Cambridge University Press.
Dorney, Sean. 1998. The Sandline Affair. Sydney: ABC Enterprises.
Dowty, David R. 1979. Word meaning and Montague grammar. New York: Springer.
Dowty, David R. 1989. On the semantic content of the notion thematic role. In Properties,
types, and meaning, ed. G. Chierchia, Barbara Partee, and R. Turner, 69129. Dordrecht: D.
Reidel.
Dowty, David R. 1990. Thematic proto-roles, argument selection, and lexical semantic defaults.
Language 67:547619.
Dowty, David R., and William Ladusaw. 1988. Toward a nongrammatical account of thematic
roles, thematic relations. In Syntax and semantics 21, ed. W. Wilkins, 6274. San Diego:
Academic Press.
Dryer, Matthew S. 1997. Are grammatical relations universal? In Essays on language function
and language type: Dedicated to Talmy Giv on, ed. J. Bybee and S.A. Haiman, 115143.
Amsterdam: John Benjamins.
Dunn, Michael, Ger Reesink, and Angela Terrill. 2002. The East Papuan languages: A prelim-
inary typological appraisal. Oceanic Linguistics 41:2862.
Dunn, Michael, and Angela Terrill. submitted. Assessing the lexical evidence for the Central
Solomons family (Papuan) using the Oswalt Monte Carlo Test. Diachronica .
Dunn, Michael, Angela Terrill, Ger Reesink, Robert A. Foley, and Stephen C. Levinson.
2005. Structural phylogenetics and the reconstruction of ancient language history. Science
309:20722075.
Durie, M. 1988. Preferred argument structure in an active language: Arguments against the
category intransitive subject. Lingua 74:125.
Durie, Mark. 1985. A grammar of Acehnese, on the basis of a dialect of North Aceh. Dordrecht:
Foris.
Durie, Mark. 1987. Grammatical relations in Acehnese. Studies in Language 11:365399.
Evans, Nicholas. 2000. Word classes in the worlds languages. In Morphology: a handbook on
inection and word formation, ed. Geert Booij, Christian Lehmann, and Joachim Mugdan,
708732. Berlin: Mouton de Gruyter.
308
Evans, Nicholas, Stephen C. Levinson, N.J. Eneld, Alice Gaby, and Asifa Majid. 2004. Re-
ciprocal constructions and situation type. In Field manual volume 9, ed. Asifa Majid, 2530.
Nijmegen: Max Planck Institute for Psycholinguistics.
Evans, Nicholas, and Toshiki Osada. 2005. Mundari: the myth of a language without word
classes. Linguistic Typology 9:351390.
Evans, Nicholas R., Alice Gaby, and Rachel Nordlinger. 2007. Valency mismatches and the
coding of reciprocity in australian languages. Linguistic Typology 11:543599.
Falk, Yehuda N. 2001. Lexical-Functional grammar: An introduction to parallel constraint-
based syntax. Stanford, CA: CSLI Publications.
Farr, J., and C. Farr. 1975. Some features of Korafe morphology. In Studies in languages of
Central and South-East Papua, ed. T.E. Dutton, Pacic Linguistics C 29, 731770. Canberra:
Australian National University.
Ferguson, Charles A. 1966. Assumptions about nasals. In Universals of language, ed. Green-
berg, Joseph H., 5360. Cambridge, MA: MIT Press.
Fillmore, Charles. 1970. The grammar of Hitting and Breaking. In Readings in English trans-
formational grammar, ed. R. Jacbos and P. Rosenbaum, 120133. Waltham: Ginn and Co.
Firchow, Irwin. 1973. A Vocabulary of Rotokas. Ukarumpa: Summer Institute of Linguistics.
Firchow, Irwin. 1984. Shoebox dictionary. Electronic les on CD.
Firchow, Irwin B. 1971. Rotokas referentials. Kivung 4:175186.
Firchow, Irwin B., ed. 1974a. Rotokas songs. Ukarumpa: Summer Institute of Linguistics.
Firchow, Irwin B., ed. 1974b. Rotokasipairara pitupituaro / Rotokas customs. Ukarumpa:
Firchow, Irwin B. 1977. Some functions of Rotokas referentials. In Proceedings of the S.I.L.
Consultants Seminar, Ukarumpa, 1976, ed. Richard Loving and David Thomas, number 20
in Workpapers in Papua New Guinea Languages, 13342. Ukarumpa: Summer Institute of
Linguistics.
Firchow, Irwin B. 1987. Form and function of Rotokas words. Language and Linguistics in
Melanesia 15:5111.
Firchow, Irwin B. n.d. Papua New Guinea. Ukarumpa: Summer Institute of Linguistics.
309
Firchow, Irwin B., and David Akoitai, ed. 1974. Rotokas reo vo siposiporoaro vo toupai /
Rotokas stories. Ukarumpa: Summer Institute of Linguistics.
Firchow, Irwin B., and Jacqueline Firchow. 1969. An abbreviated phoneme inventory. Anthro-
pological Linguistics 11:271276.
Foley, William, and Robert Van Valin Jr. 1984. Functional syntax and universal grammar.
Cambridge: Cambridge University Press.
Foley, William A. 2005. Semantic parameters and the unaccusative split in the Austronesian
languages. Studies in Language 29:385430.
Frawley, William. 1992. Linguistic semantics. Hillsdale, New Jersey: LEA.
Gailey, Harry A. 1991. Bougainville 1943-1945: The forgotten campaign. Lexington, Ken-
tucky: The University Press of Kentucky.
Giv on, Talmy. 1984. Direct object and dative shifting: Semantic and pragmatic case. In Objects:
Towards a theory of grammatical relations, ed. Frans Plank, 151182. New York: Academic
Press.
Greenberg, Joseph H. 1971. The Indo-Pacic hypothesis. In Current trends in linguistics vol 8:
Linguistics in oceania, ed. T. A. Sebeok, 807871. The Hague: Mouton and Co.
Grifn, Helga M. 2005. Sources on pre-mining Bougainville. In Bougainville before the conict,
ed. Anthony J. Regan and Helga M. Grifn, 199221. Canberra: Pandanus Books.
Grinevald, Colette. 2000. A morphosyntactic typology of classiers. In Systems of nominal
classication, ed. Gunther Senft, 5092. Cambridge: Cambridge University Press.
Haiman, John, ed. 1985. Natural syntax: Iconicity or erosion. Cambridge: Cambridge Univer-
sity Press.
Haspelmath, Martin. 2000. Coordination. In Language typology and syntactic description,
volume ii: Complex constructions, ed. Timothy Shopen, 151. Cambridge: Cambridge Uni-
versity Press.
Hayes, Bruce. 1995. Metrical stress theory: principles and case studies. Chicago: University
of Chicago.
von Heusinger, Klaus. 2002. Specicity and deniteness in sentence and discourse structure.
Journal of Semantics 19:245274.
310
Holisky, Dee A. 1987. The case of the intransitive subject in Tsova-Tush (Batsbi). Lingua
71:103132.
Hopper, Paul J., and Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Lan-
guage 56:251299.
Hostetler, Roman, and Carolyn Hostetler. 1975. A tentative description of Tinputz phonol-
ogy. In Phonologies of ve Austronesian languages, ed. Richard Loving, 544. Ukarumpa:
Hurd, Conrad. 1977. Nasioi projectives. Oceanic Linguistics 16:11178.
Hurd, Conrad, and P. Hurd. 1970a. Nasioi language course. Port Moresby: Department of
Information and Extension Services.
Hurd, Conrad, and Phyllis W. Hurd. 1970b. Nasioi verbs. Oceanic Linguistics 9:3778.
Jackendoff, Ray. 1988. The status of thematic relations in linguistic theory. Linguistic Inquiry
18:369411.
Karttunen, Lauri. 2006. The insufciency of paper-and-pencil linguistics: The case of nnish
prosody. In Intelligent linguistic architectures, ed. Miriam Butt, Mary Dalrymple, and
Tracy Holloway King, 287300. Stanford: CSLI.
Kitching, Ian J., Peter L. Forey, Christopher J. Humphries, and David M. Williams. 1998.
Cladistics: Theory and practice of parsimony analysis. Oxford: Oxford University Press.
Kittil a, Seppo. 2002. Remarks on the basic transitive sentence. Language Sciences 24:107130.
Kituai, August Ibrum K. 1998. My gun, my brother: the world of the Papua New Guinea
colonial police, 1920-1960. Pacic Islands Monograph Series 15. Honolulu: University of
Hawaii Press.
K onig, Ekkehard, and Peter Siemund. 2007. Speech act distinctions in grammar. In Language
typology and syntactic description, volume i: Clause structure, ed. Timothy Shopen, 276
324. Cambridge: Cambridge University Press.
Lakoff, George. 1977. Linguistic gestalts. In Papers from the 13th regional meeting of the
Chicago Linguistic Society, 236287. Chicago: Chicago Linguistic Society.
Langendoen, D. Terence. 1992. Symmetric relations. In The joy of grammar: A festschrift in
honor of James D. McCawley, ed. Diane Brentari, Gary Larson, and Lynn MacLeod, 199
212. Amsterdam: John Benjamins.
311
Laycock, D.C. 1970. Eliciting basic vocabulary in New Guinea. In Pacic studies in honour
of Arthur Capell, ed. S.A. Wurm and D.C. Laycock, C-13, 11271176. Canberra: Pacic
Linguistics.
Laycock, D.C. 2003. A dictionary of Buin, a language of Bougainville. Canberra: Pacic
Linguistics.
Lazard, Gilbert. 2003. Transitivity revisited as an example of a more strict approach in typo-
logical research. Folia Linguistica XXXVI:141190.
Lehmann, Christian. 1983. Directions for interlinear morphemic translations. Folia Linguistica
16:193224.
Levin, B., G. Song, and B.T.S. Atkins. 1997. Making sense of corpus data: A case study of
verbs of sound. International Journal of Corpus Linguistics 2:2364.
Levin, Beth. 1999. Objecthood: an event structure perspective. In Chicago Linguistic Society
35, ed. Sabrina J. Billings, John P. Boyle, and Aaron M. Grifth, 223247. Chicago: Chicago
Linguistic Society.
Levin, Beth, and Malka Rappaport Hovav, ed. 1995a. Unaccusativity at the syntax-lexical
semantic interface. Cambridge, MA: MIT Press.
Levin, Beth, and Malka Rappaport Hovav. 1995b. Unaccusativity: At the syntax-lexical seman-
tics interface. Cambridge, MA: MIT Press.
Levin, Beth, and Malka Rappaport Hovav. 2006. Argument realization. Research Surveys in
Linguistics. Cambridge: Cambridge University Press.
Levin, Lori. 1987. Toward a linking theory of relation changing rules in LFG. Technical Report
No. CSLI-87-115. Stanford: CSLI Publications.
Lichtenberk, Frantisek. 1985. Multiple uses of reciprocal constructions. Australian Journal of
Linguistics 5:1941.
Litteral, Robert. 2001. Language development in Papua New Guinea. Radical Pedagogy 21.
Loper, Edward, and Steven Bird. 2002. NLTK: The Natural Language Toolkit. In Proceedings
of the ACL Workshop on Effective Tools and Methodologies for Teaching Natural Language
Processing and Computational Linguistics, Philadelphia, July 2002, Association for Compu-
tational Linguistics. URL http://arxiv.org/abs/cs/0205028.
312
Loper, Edward, and Steven Bird. 2004. NLTK: The Natural Language Toolkit. In Proceed-
ings of the ACL Demonstration Session, Barcelona, July 2004. URL http://www.ldc.
upenn.edu/sb/home/papers/nltk.pdf.
Maddieson, Ian. 1984. Sound patterns of language. Cambridge: Cambridge University Press.
McClure, William. 1990. A lexical semantic explanation for unaccusative mismatches. In
Grammatical relations: a cross-theoretical perspective, ed. K. Dziwirek, P. Farrell, and
E. Mejias-Bikandi, 305318. Stanford: CSLI Publications.
McElhanon, K.A. 1987. Basic vocabulary as a tool for classifying Papuan languages. In A
world of language, ed. D.C. Laycock and W. Winter, 424433. Canberra: Pacic Linguistics.
McKoon, Gail, and Talke Macfarland. 2000. Externally and internally caused change of state
verbs. Language 76:833858.
McLendon, Sally. 1978. Ergativity, case, and transitivity in Eastern Pomo. International Journal
of American Linguistics 44:19.
Merlan, Francesca. 1985. Split intransitivity: functional oppositions in intransitive inection. In
Grammar inside and outside the clause, 324362. Cambridge: Cambridge University Press.
Michaelis, Laura. 2001. Exclamative constructions. In Language typology and language uni-
versals, ed. Martin Haspelmath, Ekkehard K onig, Wulf Oesterreicher, and Wolfgang Raible,
10381058. Berlin: Mouton de Gruyter.
Mithun, Marianne. 1984. The evolution of noun incorporation. Language 60:847897.
Mithun, Marianne. 1986. On the nature of noun incorporation. Language 62:3237.
Mithun, Marianne. 1991. Active/agentive case marking and its motivation. Language 67:510
546.
Mithun, Marianne. 1999. What are S, A, and O? Studies in Language 23:579606.
Mosel, Ulrike. 1991. Towards a typology of valency. In Das sprachliche verhalten von sachver-
halten. participation, ed. H. Seiler and W. Premper, 240251. T ubingen: Narr. S.
Mosel, Ulrike, and Jessika Reinig. 2000. Valence changing clitics and incorporated prepositions
in Teop (Oceanic, Bougainville). In Proceedings of AFLA VII, ed. Marian Klamer, 133140.
Amsterdam: Dept. of Linguistics, Vrije Universiteit.
313
Mosel, Ulrike, and Ruth Spriggs. 1999a. Gender in Teop. In Gender in grammar and cognition,
ed. B. Unterbeck and M. Rissanen, Trends in Linguistics, Studies and Monographs 124.
Mouton de Gruyter.
Mosel, Ulrike, and Ruth Spriggs. 1999b. Negation in Teop. In Negation in oceanic languages,
ed. E. Hovdhaugen and U. Mosel, 4556. Munchen: Lincom.
M uller, Fr A. 1954. Grammatical notes on the Konua language. Micro-Bibliotheca Anthropos
12.
Nss,

Ashild. 2006. Case semantics and the agent-patient opposition. In Case, valency and
transitivity, ed. Leonid Kulikov, Andrej Malchukov, and Peter de Swart, Studies in Language
Companion Series 77, 309327. Amsterdam/Philadelphia: John Benjamins.
Naess, Ashild. 2007. Prototypical transitivity. Typological Studies in Language 72. Stanford:
John Benjamins.
Nedjalkov, Vladimir, and Sergej Je. Jaxontov. 1988. The typology of resultative constructions.
In Typology of resultative constructions, ed. Vladimir Nedjalkov, 362. Amsterdam: John
Benjamins.
Nelson, Hank. 2005. Bougainville in World War II. In Bougainville before the conict, 168198.
Canberra: Pandanus Books.
Noonan, Michael. 1985. Complementation. In Language Typology and Syntactic Description,
Vol I: Clause Structure, ed. Timothy Shopen, 42140. Cambridge: Cambridge University
Press.
Oates, W., and L. Oates. 1964. Gugu-Yalanji linguistic and anthropological data. In Occasional
papers in aboriginal studies, volume 2, 117. Canberra: Australian Institute of Aboriginal
Studies.
Onishi, Masa. 1994. A grammar of Motuna (Bougainville, Papua New Guinea). Canberra:
Onishi, Masa. 2002. An annotated word list of the Motuna language. Tokyo: Research Institute
for Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies.
Palmer, F.R. 2001. Mood and modality. Cambridge: Cambridge University Press, second
edition.
Payne, Thomas E. 1997. Describing morphosyntax: a guide for eld linguists. Cambridge:
314
Perlmutter, David. 1978. Impersonal passives and the unaccusative hypothesis. Papers from the
Annual Meeting of the Berkeley Linguistics Society 4:157189.
Pesetsky, David, ed. 1995. Zero syntax. Cambridge, MA: MIT Press.
Reesink, Gerard. 1984. Structures and their functions in Usan, a Papuan language of Papua
New Guinea. Doctoral Dissertation, University of Amsterdam.
Regan, Anthony J., and Helga M. Grifn, ed. 2005. Bougainville before the conict. Canberra:
Pandanus Books.
de Reuse, W. J. 1964. Noun incorporation. In Concise encyclopedia of grammatical categories,
ed. Keith Brown and Keith Miller, 252258. Amsterdam: Elsevier.
Richard, E.L. 1975. Sentence structure of Guhu-Samane. In Studies in languages of Central and
South-East Papua, ed. T.E. Dutton, Pacic Linguistics C 29, 771816. Canberra: Australian
National University.
Robinson, Stuart. 2006. The phoneme inventory of the Aita dialect of Rotokas. Oceanic Lin-
guistics 45:206209.
Robinson, Stuart, Greg Aumann, and Steven Bird. 2007. Managing eldwork data with toolbox
and the natural language toolkit. Language Documentation and Conservation 1:4457. URL
http://nflrc.hawaii.edu/ldc/June2007/robinson/robinson.html.
Robinson, Stuart, and Sera Mon. 2006. Rotokas elementary school reader.
Rosen, Carol G. 1984. The interface between semantic roles and initial grammatical relations.
In Studies in relational grammar 2, ed. David Perlmutter and Carol Rosen, 3875. Chicago:
The University of Chicago Press.
Rosen, S.T. 1989. Two types of noun incorporation: A lexical analysis. Language 65:294317.
Ross, Malcolm. 2001. Contact-induced change in Oceanic languages in North-West Melanesia.
In Areal diffusion and genetic inheritance, ed. Alexandra Y. Aikhenvald and R.M.W. Dixon,
134166. Oxford: Oxford University Press.
Sack, Peter. 2005. German colonial rule in the Northern Solomons. In Bougainville before the
conict, 77107. Canberra: Pandanus Books.
Sadock, Jerrold, and Arnold M. Zwicky. 1985. Speech act distinctions in syntax. In Language
typology and syntactic description, ed. Timothy Shopen, 155196. Cambridge: Cambridge
University Press.
315
Sadock, Jerry M. 1986. Some notes on noun incorporation. Language 62:1931.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing
phylogenetic trees. Molecular Biology and Evolution 4:406425.
Sapir, Edward. 1911. The problem of noun incorporation in American languages. American
Anthropologist 13:250282.
Sasse, Hans-J urgen. 1993. Das nomeneine universelle kategorie? Sprachtypologie und Uni-
versalienforschung STUF 46:161236.
Schachter, Paul. 1985. Parts-of-speech systems. In Language Typology and Syntactic De-
scription, Volume I: Clause Structure, ed. Timothy Shopen, 361. Cambridge: Cambridge
University Press.
Schwartz, Marcia L., and Ulrike Mosel. 2006. Teop-English lexical database. Kiel: Institut fur
Allgemeine Sprachwissenschaft.
Shibatani, Masayoshi. 1976. The grammar of causative constructions, volume 6 of Syntax and
Semantics. New York: Academic Press.
Shoffner, R.K. 1976. The economy and cultural ecology of Teop: an analysis of the shing,
gardening, and cash cropping systems in a Melanesian society. Doctoral Dissertation.
Sinclair, James Patrick. 1981. Kiap: Australias patrol ofcers in Papua New Guinea. Sydney:
Pacic Publications.
Snell-Hornby, M. 1983. Verb descriptivity in German and English. Heidelberg: Carl Winter.
Somers, H.L. 1987. Valency and case in computational linguistics. Edinburgh Information
Technology Series 3. Edinburgh: Edinburgh University Press.
Song, Jae Jung. 1996. Causatives and causation: A universal-typological perspective. London
and New York: Longman.
Sorace, A. 2004. Gradience at the lexicon-syntax interface: evidence from auxiliary selec-
tion and implications for unaccusativity. In The unaccusativity puzzle: explorations of the
syntax-lexicon interface, ed. A. Alexiandou, E. Anagnostopoulou, and M. Everaert, 243268.
Oxford: Oxford University Press.
Stassen, Leon. 1997. Intransitive predication. Oxford University Press: Oxford.
Stiebel, Barbara. 2006. Agent focus in Mayan languages. Natural Language and Linguistic
Theory 24:501570.
316
Talmy, Len. 1975. Semantics and syntax of motion. In Syntax and semantics 4, ed. Timothy
Shopen, 181238. New York: Academic Press.
Talmy, Len. 1985. Lexicalization patterns: Semantic structure in lexical forms. In Language
Typology and Syntactic Description, Volume III: Grammatical Categories and the Lexicon,
ed. Timothy Shopen, 57149. Cambridge: Cambridge University Press.
Talmy, Len. 2007. Lexical typologies. In Language Typology and Syntactic Description, Volume
III: Grammatical Categories and the Lexicon, ed. Timothy Shopen, 66168. Cambridge:
Terrill, Angela. 2002. Systems of nominal classication in East Papuan languages. Oceanic
Linguistics 41:6388.
Tesni` ere, Lucien. 1959.

El ements de syntaxe structurale. Paris: Klincksieck.
The New Testament. 1982. Airepa vao reo ovoi (The New Testament). South Holland, Illinois,
U.S.A.: World Home Bible League.
The Old Testament. 1993. Siposipoara vao iava tuariripa reo ovoi (old testament stories).
South Holland, Illinois, U.S.A.: World Home Bible League.
Thomason, Sarah G., and Daniel L. Everett. 2001. Pronoun borrowing. In Proceedings of the
annual meeting of the Berkeley Linguistics Society, volume 27, 301316.
Tryon, Darell. 2005. The languages of Bougainville. In Bougainville before the conict, ed.
Anthony J. Regan and Helga M. Grifn, 3146. Canberra: Pandanus Books.
Tsunoda, Tasaku. 1981. Split case-marking patterns in verb-types and tense/aspect/mood. Lin-
guitics 19:389438.
Van Valin Jr., Robert D. 1984. A typology of syntactic relations in clause linkage. In Pro-
ceedings of the 10th annual meeting of the Berkeley Linguistics Society, ed. C. Brugman and
M. Macaulay, 542558. Berkeley: University of California at Berkeley.
Van Valin Jr., Robert D. 1987. The unaccusative hypothesis vs. lexical semantics: Syntactic vs.
semantic approaches to verb classication. NELS 17:641661.
Van Valin Jr., Robert D. 1990. Semantic parameters of split intransitivity. Language 66:221
260.
Van Valin Jr., Robert D. 2005. Exploring the syntax-semantics interface. Cambridge: Cam-
bridge University Press.
317
Van Valin Jr., Robert D., and Randy LaPolla. 1997. Syntax: Form and function. Cambridge:
Van Valin Jr., Robert D., and David P. Wilkins. 1996. The case for effector: Case roles,
agents, and agency revisited. In Grammatical constructions: their form and meaning, ed.
M. Shibatani and S. Thompson, 289322. Oxford: Oxford University Press.
Vendler, Z. 1957. Verbs and time. Philosophical Review 97121.
Vendler, Zeno. 1967. Linguistics in philosophy. Ithaca: Cornell University Press.
Vernon, Don. 2005. The Panguna mine. In Bougainville before the conict, ed. Anthony J.
Regan and Helga M. Grifn, 258273. Canberra: Pandanus Books.
Vogel, Petra M., and Bernard Comrie, ed. 2000. Approaches to the typology of word classes.
Berlin: Mouton de Gruyter.
van Voorst, Jan. 1988. Thematic roles are not semantic roles. Revue queebecoise de linguis-
tique 17:245260.
Waiko, John Dademo. 1993. A short history of Papua New Guinea. Melbourne: Oxford Uni-
versity Press.
White, J.P., and J.F. OConnell. 1982. A prehistory of Australia, New Guinea and Sahul. Aca-
demic Press.
Wickler, S., and MatthewSpriggs. 1988. Pleistocene human occupation of the Solomon Islands,
Melanesia. Antiquity 62:703706.
Wittgenstein, Ludwig. 1953/2001. Philosophical investigations. Oxford: Blackwell Publishing.
Wurm, S. A. 1972. The classication of Papuan languages and its problems. Linguistic Com-
munications 6:118178.
Wurm, S. A. 1975a. The East Papuan Phylum in general. In New Guinea area languages and
language study, Volume 1: Papuan languages and the New Guinea linguistic scene, ed. S. A.
Wurm, number 38 in C, 783804. Canberra: Pacic Linguistics.
Wurm, S.A. 1975b. Eastern Central Trans-New Guinea Phylum languages. In Papuan lan-
guages and the New Guinea linguistic scene, Pacic linguistics C 38, 461526. Canberra:
Wurm, Stephen, and S. Hattori. 1981. Language atlas of the Pacic Area. Canberra: Australian
Academy of the Humanities.
318
Yip, Moira, Joan Maling, and Ray Jackendoff. 1987. Case in tiers. Language 63:217250.
Zaenen, Annie. 1988. Unaccusative verbs in Dutch and the syntax-semantics interface. Stan-
ford: CSLI/SRI International.
Zaenen, Annie. 1993. Unaccusativity in Dutch: Integrating syntax and lexical semantics. In
Semantics and the lexicon, ed. James Pustejovsky, 129161. London: Kluwer.
Zwicky, Arnold M. 1985. Clitics and particles. Language 61:283305.
Zwicky, Arnold M., and Geoffrey K. Pullum. 1983. Cliticization vs. inection: English nt.
Language 59:502513.
319
Index of Bound Morphology
=a, 86, 91, 92, 149
=ia, 30, 34, 35, 3841, 4349, 53, 5658,
6365, 67, 6972, 74, 75, 8186,
88, 90, 92, 9598, 103108, 110
112, 116, 120122, 124127, 129,
133, 134, 136, 137, 139141, 143,
144, 148, 151, 152, 157, 163168,
170174, 176, 179, 186, 188191,
194, 196, 198, 199, 201, 204206,
210, 214, 217, 218, 220, 224, 228,
247, 251
=pa, 35, 40, 50, 54, 56, 62, 71, 73, 74, 80,
82, 87, 89, 91, 104, 109, 124, 132
134, 136, 137, 142, 152, 154, 157,
166, 170173, 175, 176, 178, 180,
181, 201, 202, 222, 224, 226, 228
=re, 43, 44, 46, 50, 51, 56, 60, 65, 66, 68,
69, 7274, 82, 87, 91, 95, 99, 105,
109, 120, 133, 140, 143, 162, 164,
170175, 196, 198, 202204, 211,
223, 224, 226, 229
=va, 43, 44, 47, 52, 61, 65, 68, 69, 71, 72,
74, 82, 83, 90, 91, 97, 124, 125,
127, 135, 149, 164, 165, 169174,
176, 177, 191, 205, 214, 228, 251
ora-, 35, 45, 46, 49, 57, 61, 6567, 70, 74,
80, 81, 88, 93, 94, 96, 101, 104,
105, 108, 110, 112, 116, 120, 124,
125, 127129, 133, 137, 173, 174,
187, 193198, 205, 216, 220, 222,
225, 226, 228, 251, 252
o-, 65, 72, 82, 8587, 96, 97, 105, 136
s-, 221
vo-, 30, 35, 44, 46, 49, 51, 52, 55, 69, 75,
81, 8486, 90, 92, 95, 96, 107, 110
112, 120, 122124, 126, 127, 141,
150, 151, 172, 179, 196, 199, 201,
203, 205, 214, 224
-, 53, 104, 149, 150, 152
-arapa, 99, 228
-ara, 35, 41, 44, 46, 47, 49, 52, 53, 57, 59,
71, 75, 80, 82, 85, 87, 92, 102, 104,
105, 107109, 111114, 122, 124
126, 129, 135, 136, 140, 141, 150,
151, 165, 188, 194, 202, 203, 205,
214, 224
-arei, 49, 88, 144
-aro, 35, 43, 45, 5659, 61, 62, 64, 69, 71
75, 82, 84, 86, 88, 90, 92, 97, 98,
103, 104, 110, 120, 122124, 126,
133, 136, 142, 149151, 163, 172,
174, 175, 196, 197, 199, 201, 204,
212, 216, 218, 225
-a, 30, 34, 41, 44, 45, 4750, 52, 5559, 61
65, 67, 7075, 8183, 8592, 95,
96, 99, 101106, 108, 110, 112
116, 120127, 132137, 139144,
148, 149, 152, 153, 157, 162, 165,
166, 168, 170172, 174, 176179,
188, 191196, 199, 201203, 205,
320
210212, 214, 216, 217, 222, 224
226, 228, 229, 246, 248, 251, 252
-ba, 248
-ea, 248
-ei, 35, 40, 46, 49, 52, 54, 56, 5965, 67,
68, 70, 8183, 85, 87, 89, 90, 92
94, 9699, 101, 103105, 108, 109,
112, 114, 115, 122, 125, 127129,
133, 135, 137140, 142, 148, 149,
151, 152, 157, 162, 167, 168, 171,
173175, 186, 192, 196, 197, 199
202, 204206, 211, 212, 214216,
221, 224227, 251
-epa, 38, 43, 47, 56, 61, 69, 74, 84, 88, 91,
95, 106, 109, 114116, 122, 124,
125, 137, 139, 140, 168, 169, 171
173, 175, 186, 189, 201, 227, 247,
252
-erao, 96, 105, 108, 122, 173
-era, 57, 61, 69, 96, 105, 112, 140, 164, 227
-erei, 221
-ere, 44, 101, 115, 116, 125, 127, 136, 149,
222, 227
-e, 35, 37, 40, 4245, 47, 5052, 58, 60, 61,
64, 65, 67, 68, 70, 8183, 86, 88,
90, 92, 94, 95, 99, 104, 105, 109,
111, 112, 114116, 126129, 135
137, 139, 141144, 149, 152, 162
169, 171, 172, 175, 177179, 181,
186, 187, 190, 192, 198201, 203
205, 210212, 218, 220223, 225
229, 247, 251, 252
-ia, 71, 75, 135
-io, 38, 52, 67, 68, 74, 94, 101, 109, 120,
149
-irao, 56, 89, 9396, 176, 205, 222
-irara, 43, 44, 57, 71, 75, 82, 90, 91, 108,
110, 121, 125, 132, 138, 144, 172,
176, 186, 194, 220
-iva, 47, 103
-i, 35, 39, 40, 4246, 48, 49, 5153, 55, 56,
58, 59, 62, 6572, 74, 75, 8588,
90, 91, 9498, 101, 104109, 112,
115, 116, 120124, 126, 127, 134,
137142, 144, 149154, 157, 164,
165, 168, 172, 173, 177, 179, 186,
190, 192197, 202204, 210, 212,
214, 217, 218, 220, 222, 225, 227
229, 247
-oro, 35, 41, 43, 4547, 51, 54, 61, 63, 85,
9294, 9698, 101, 105, 116, 120,
125, 127, 128, 138140, 144, 149,
164, 167, 169, 173, 175, 194, 204,
205, 210, 215, 216, 220, 221, 224
226, 228, 229, 251
-o, 42, 44, 46, 47, 51, 52, 60, 66, 68, 70, 71,
83, 94, 95, 97, 99, 103, 105, 107,
112, 114, 115, 129, 133, 135, 140,
143, 144, 149, 164, 166, 167, 173,
179, 186, 192, 198, 200, 201, 204
206, 212, 225, 226, 251, 252
-pa, 30, 34, 35, 3841, 4349, 5254, 56
75, 8188, 90101, 104112, 114
116, 120125, 127144, 148154,
157, 162180, 186, 188, 192205,
210212, 214, 215, 217, 218, 220,
222229, 247, 251, 252
-peira, 49, 95, 111, 112
-pere, 110
-pe, 40, 41, 53, 107, 110
-pie, 35, 37, 40, 4345, 49, 57, 61, 63, 64,
74, 81, 82, 85, 90, 93, 94, 96, 97,
106, 107, 110, 112, 116, 125, 126,
134, 137140, 143, 144, 152, 163,
165, 169, 173, 179, 188193, 204,
210, 211, 220222, 224, 225, 227,
321
229, 248, 251, 252
-piro, 88, 97, 98, 109, 122, 203, 204, 210,
225
-pi, 122
-raga, 49, 58, 61, 85, 94, 96, 128, 229
-ra, 35, 40, 44, 49, 50, 53, 55, 61, 63, 6570,
75, 81, 85, 89, 90, 9699, 101, 102,
104106, 108, 115, 120, 123, 125,
129, 136, 138, 139, 142, 148, 153,
154, 157, 162, 164166, 168, 173,
175, 176, 189, 196, 197, 202, 203,
212, 214, 216, 220, 222, 226228,
247, 248, 251
-rei, 49, 52, 85, 90, 128, 227
-re, 34, 35, 38, 40, 43, 4547, 4953, 55
59, 6367, 70, 71, 7375, 8190,
93, 94, 96, 97, 101, 103112, 115,
120, 122124, 129132, 134138,
140, 142144, 152, 157, 164166,
169, 170, 172, 176, 178182, 187
192, 196, 198, 201, 203206, 210
214, 216, 218, 220222, 224226,
228, 247, 251, 252
-rirei, 43, 44, 47, 115, 137, 222
-ri, 30, 46, 49, 59, 66, 80, 84, 86, 8991,
101, 102, 105, 107, 120, 121, 123,
133, 140142, 144, 169, 173, 179,
180, 188, 217
-ro, 35, 38, 43, 4547, 52, 53, 56, 60, 61, 64,
65, 6770, 74, 8287, 9197, 101,
104, 105, 108, 109, 111, 114, 120,
124, 125, 135, 137, 139144, 150,
152, 163, 165, 167169, 171173,
175, 177, 186, 187, 193195, 198,
199, 201, 205, 210, 212, 214, 220
223, 226228, 247, 251, 252
-sia, 34, 44, 49, 52, 53, 56, 64, 67, 69, 82
84, 87, 91, 94, 98, 107, 111, 114,
122, 125, 127, 136139, 141, 148
150, 162, 165, 166, 173, 188, 191,
199, 203205, 228, 246, 251
-si, 43, 54, 81, 88, 99101, 110, 127, 128,
149, 151, 174, 175, 196198, 224,
227, 228
-ta, 57, 74, 101, 114, 122, 165, 203
-toarei, 43, 44, 49, 95, 149
-toa, 40, 69, 70, 72, 73, 84, 85, 89, 90, 95,
106, 111, 120, 144, 166, 168, 173,
176, 180, 188, 191, 192, 222, 251
-to, 35, 38, 4346, 49, 55, 57, 61, 63, 64, 70,
74, 8184, 87, 9092, 9497, 103,
105, 106, 110112, 116, 120, 122
124, 129132, 135144, 150, 163
167, 169, 171, 178, 181, 186, 188,
190192, 196, 201, 205, 210212,
214, 218, 220, 222, 225, 226, 228,
247, 251
-u, 35, 56, 57, 66, 7173, 75, 83, 84, 89, 91,
101, 102, 104, 108, 111, 112, 124,
133, 157, 164, 174, 175, 199, 226,
229
-vai, 40, 49, 65, 71, 73, 75, 80, 84, 8792,
98, 100, 108, 114, 120, 134, 135,
137, 144, 162, 180
-va, 35, 37, 38, 4247, 50, 52, 53, 5658,
6163, 65, 66, 68, 73, 74, 82, 83,
8689, 9294, 98, 102104, 106
112, 116, 120123, 126, 129132,
134, 136, 138, 140142, 144, 151,
154, 163, 165, 166, 171175, 178,
181, 186, 188192, 196, 198, 200
203, 205, 212, 214, 218, 220, 221,
223, 226, 246, 251, 252
-veaira, 112
-vearea, 110
-veare, 54, 110
322
-veira, 41, 43, 46, 47, 52, 5658, 64, 68, 71,
85, 86, 88, 92, 98, 105, 111, 112,
120, 125, 132, 141, 150, 152, 157,
163, 168, 169, 172, 173, 176, 180,
188, 192, 202, 211, 212, 214, 220
-verea, 91, 110
-vere, 30, 43, 56, 71, 80, 86, 98, 109, 110,
135, 136, 138, 140
-ve, 40, 43, 44, 49, 50, 52, 54, 57, 59, 73,
75, 76, 86, 87, 89, 91, 98, 101, 104,
105, 107109, 135137, 140144,
157, 163, 166, 171, 176, 177, 198,
203, 214, 220, 224
-vio, 82, 114
-vira, 30, 35, 38, 41, 42, 46, 49, 51, 54, 63,
65, 66, 68, 69, 82, 83, 86, 88, 90,
92, 9496, 99, 104, 105, 108111,
122, 129, 136, 137, 139142, 144,
148, 149, 151153, 165, 167, 171,
177180, 186, 188, 189, 197, 198,
201, 202, 204, 205, 210212, 214,
216, 218, 221, 222, 225228, 251
-viro, 68, 82, 97, 98, 105, 115, 142, 204
206
-visivi, 179, 227
-vi, 44, 59, 73, 82, 85, 88, 95, 98, 101, 114,
125, 141, 143, 148, 152
-voiva, 103
-voi, 34, 35, 39, 46, 5053, 55, 58, 59, 62,
66, 67, 71, 75, 81, 83, 87, 8991,
94, 95, 99, 104, 112, 116, 120, 122,
124, 127, 130, 131, 134, 136, 138
140, 142, 144, 148153, 163, 164,
169171, 174, 178, 179, 186, 188,
190, 193, 196, 197, 201, 202, 204
206, 210213, 216218, 221, 222,
224226, 229, 246, 247
-vorao, 45, 52, 53, 57, 105, 143, 152
-vora, 45, 52, 57, 105, 138, 212, 228, 229
-vo, 35, 38, 42, 44, 45, 47, 50, 51, 53, 55,
58, 59, 63, 67, 68, 73, 74, 81, 85,
86, 90, 9497, 105, 109, 112, 120,
122, 123, 126, 127, 129, 130, 132,
134, 136, 142144, 162, 164166,
168, 170, 172, 175, 177182, 187,
188, 190192, 200, 201, 203, 204,
210, 212, 217, 220226, 229, 251,
252
-vu, 44, 47, 72, 82, 8588, 96, 97, 105, 135,
136, 212, 220
323
Samenvatting
Gespleten intransitiviteit in het Rotokas, een Papoea taal uit Bougainville
Het doel van dit proefschrift is tweevouding. Ten eerste geeft het een vrij uitvoerige beschri-
jving van de grammatica van het Rotokas, een Papoea taal (non-Austronesisch) die gesproken
wordt in Bougainville, Papoea Nieuw Guinea. Er bestaan al eerdere grammaticale beschrijvin-
gen van het Rotokas (zie 2.2.1 voor een compleet overzicht), maar dit zijn voornamelijk kleine
publicaties die soms moeilijk te volgen zijn, en een overzicht ontbreekt. Ten tweede richt dit
werk zich op een speciek onderdeel van de grammatica van het Rotokas dat problematisch is
voor grammaticale theorie: werkwoordsvervoeging. Daarbij gaat het speciek om het bestaan
van twee elkaar uitsluitende vervoegingsklassen voor congruentie met het onderwerp en mark-
ering van tijd en aspect. Verschillende aspecten van de morfosyntaxis van het Rotokas zullen
onderzocht worden en uiteindelijk zal geconcludeerd worden dat het Rotokas vanuit een typol-
ogisch standpunt gezien een interessante vorm van gespleten intransitiviteit heeft. De aard van
gespleten intransitiviteit in het Rotokas heeft implicaties voor theorien over gespleten intran-
sitiviteit in specieke zin en voor theorien over transitiviteit, valentie, en de interface tussen
semantiek en syntaxis in het algemeen.
Deel I
Hoofdstuk 1 introduceert de doelen en de opbouw van het proefschrift en geeft achtergrond-
informatie met betrekking tot het veldwerk dat de auteur in Bougainville verricht heeft tussen
2003 en 2005.
Hoofdstuk 2 geeft achtergrondinformatie met betrekking tot de Rotokas taal en haar sprek-
ers. In 2.1 wordt de recente geschiedenis van Bougainville en de diepere geschiedenis van
de regio besproken. 2.2 geeft belangrijke achtergrondinformatie over het Rotokas: eerdere
beschrijvingen van de taal, informatie over de sprekers, een overzicht van dialectologische vari-
atie, een overzicht van de talen die gesproken worden in Bougainville, en een samenvatting van
wat bekend is over de genetische relaties van deze talen.
Hoofdstuk 3 geeft een overzicht van de fonologie van het Rotokas, die typologisch gezien
ongebruikelijk is vanwege de kleine foneeminventaris. 3.1 beschrijft de foneeminventaris van
de taal en 3.2 beschrijft wat bekend is over de suprasegmentele fonologie van de taal.
325
Hoofdstuk 4 bekijkt de woordklassen die gevonden zijn in de taal. Er wordt een onderscheid
gemaakt tussen wortels, stammen en woorden. De volgende woordklassen worden onderschei-
den: zelfstandig naamwoorden, maatwoorden, voornaamwoorden, werkwoorden, bijvoegelijk
naamwoorden, bijwoorden, achterzetsels, vraagwoorden, voegwoorden en exclamatieven.
Hoofdstuk 5 geeft een overzicht van de vrij uitgebreide morfologie van het Rotokas. Er
wordt eerst gekeken naar de morfologie van het zelfstandig naamwoord en daarna naar die van
het werkwoord. Ook reduplicatie en morfofonemische regels worden beschreven.
Hoofdstuk 6 zich op syntaxis. Eerst wordt de naamwoordgroep besproken en daarna de
gehele zin. Zowel de syntaxis binnen zinsdelen (6.2) als de syntaxis tussen zinsdelen (6.3)
wordt behandeld. In 6.2 komen de basiswoordvolgorde, de verplaatsing van O, vraagzinnen
en negatie aan bod. In 6.3.1 worden complementatie, verbale constituenten en grotere syntac-
tische eenheden (zinnen met voegwoorden) besproken.
Deel II
In hoofdstuk 7 wordt de probleemstelling geformuleerd. Deze wordt ingegeven door de twee
verschillende klassen van werkwoordsvervoeging in het Rotokas, die zullen worden aangeduid
met en . 7.1 geeft een uitvoerige beschrijving van het formele onderscheid en een duideli-
jke lijst met kenmerken voor de herkenning van deze twee klassen. In 7.2 wordt het prob-
leem beschreven en een tentatieve hypothese voor een oplossing geformuleerd. In de komende
hoofdstukken wordt deze hypothese nader uitgewerkt door in te gaan op valentie, valentieveran-
derende derivaties en de semantiek van de twee klassen.
In hoofdstuk 8 wordt het karakter van valentie in het Rotokas bestudeerd. Er zijn twee
typen valentie in het Rotokas: monovalente werkwoordswortels (intransitief) met e en argu-
ment en bivalente werkwoordswortels (transitief) met twee of mogelijk drie kernargumenten.
Bivalente werkwoordswortels vervoegen altijd volgens het patroon. Monovalente werkwo-
ordswortels echter, vallen in twee groepen wat betreft de vorm van hun vervoeging: de meeste
vervoegen volgens het patroon, maar sommige vertonen vervoeging.
Hoofdstuk 9 geeft een overzicht van derivaties die de valentie van werkwoordswortels ver-
meerderen of verminderen. Valentievermeerderende derivaties worden besproken in 9.1 en
valentieverminderende derivaties worden besproken in 9.2. Deze derivaties zijn niet gevoelig
voor het onderscheid tussen en monovalente werkwoordswortels en geven dus geen aanlei-
ding om een onderliggend syntactisch verschil tussen deze twee typen te veronderstellen. De
derivaties geven wel extra bewijs voor een nauw verband tussen valentie en werkwoordsver-
voeging, aangezien een afnemende valentie in verband staat met vervoeging en een toene-
mende valentie met vervoeging.
In hoofdstuk 10 wordt de semantische basis van gespleten intransitiviteit in het Rotokas
besproken. De semantische rollen die met de verschillende grammaticale rollen samenhangen
worden bestudeerd in 10.2. In 10.3 worden de resultaten van deze studie in een bredere,
typologische context geplaatst en worden verder strekkende implicaties van de gespleten in-
326
transitiviteit in het Rotokas besproken.
Samenvattend: dit proefschrift draagt bij aan het debat over de juiste analyse van gespleten
intransitiviteit, en of dit in de eerste plaats een semantisch of een syntactisch verschijnsel is.
De tweedeling tussen semantiek en syntaxis is misleidend in het licht van de analyse van het
Rotokas waarvoor hier gepleit wordt. Gespleten intransitiviteit bevindt zich op de grens tussen
syntaxis en semantiek; beide zijn noodzakelijk voor een complete verklaring. Er lijkt geen
enkelvoudig principe te zijn dat de vervoeging van werkwoorden in het Rotokas aanstuurt en
de verklaring die in dit proefschrift wordt gegeven brengt veel willekeurige stipulaties in het
werkwoordslexicon met zich mee. Hoewel er daarom geen sterke conclusies getrokken kun-
nen worden, geeft deze analyse ons wel meer inzicht in de morfosyntactische complexiteit van
het Rotokas. Bovendien worden fundamentele aspecten van de grammatica van een relatief
onbeschreven Papoea taal blootgelegd.
Dit proefschrift bevat drie bijlagen. De eerste bijlage is een gedetailleerde lijst van werk-
woordsstammen, afkomstig uit een elektronische database van het Rotokas lexicon ontwikkeld
door de auteur. De tweede bijlage is een formele implemetatie van de morfologische analyse
van het Rotokas binnen het framework van nite state morfology, waarvoor de PARC sofware
toolkit is gebruikt. De derde bijlage bevat twee voorbeeldteksten met interlineaire glossen en
vertalingen in het Engels en het Tok Pisin. Het betreft twee verschillende versies van een tra-
ditioneel volksverhaal, de n gedocumenteerd door Irwin Firchow en de ander door de auteur
zelf.
327
Curriculum Vitae
Stuart Robinson received a liberal arts education at Reed College, where he received a B.S.
in Linguistics in 1996. He continued his studies in linguistics at the Australian National Uni-
versity, where he received a M.A. in Linguistics in 1999. He was offered a Ph.D. scholarship at
the Max Planck Institute for Psycholinguistics in Nijmegen, The Netherlands, where he resided
intermittently from 2003 and 2007 between eldwork trips to Bougainville. After leaving the
MPI, he joined Powerset, a startup in San Francisco that specialized in natural language web
search. The company was acquired by Microsoft in 2008 and was incorporated into Microsofts
search engine, Bing.
MPI Series in Psycholinguistics
1. The electrophysiology of speaking: Investigations on the time course of semantic, syn-
tactic, and phonological processing. Miranda van Turennout
2. The role of the syllable in speech production: Evidence from lexical statistics, met-
alinguistics, masked priming, and electromagnetic midsagittal articulography. Niels O.
Schiller
3. Lexical access in the production of ellipsis and pronouns. Bernadette M. Schmitt
4. The open-/closed-class distinction in spoken-word recognition. Alette Haveman
5. The acquisition of phonetic categories in young infants: A self-organising articial neural
network approach. Kay Behnke
6. Gesture and speech production. Jan-Peter de Ruiter
7. Comparative intonational phonology: English and German. Esther Grabe
8. Finiteness in adult and child German. Ingeborg Lasser
9. Language input for word discovery. Joost van de Weijer
10. Inherent complement verbs revisited: Towards an understanding of argument structure in
Ewe. James Essegbey
11. Producing past and plural inections. Dirk Janssen
12. Valence and transitivity in Saliba: An Oceanic language of Papua New Guinea. Anna
Margetts
13. From speech to words. Arie van der Lugt
14. Simple and complex verbs in Jaminjung: A study of event categorisation in an Australian
language. Eva Schultze-Berndt
15. Interpreting indenites: An experimental study of childrens language comprehension.
Irene Krmer
16. Language-specic listening: The case of phonetic sequences. Andrea Weber
17. Moving eyes and naming objects. Femke van der Meulen
18. Analogy in morphology: The selection of linking elements in Dutch compounds. Andrea
Krott
19. Morphology in speech comprehension. Kerstin Mauth
20. Morphological families in the mental lexicon. Nivja H. de Jong
21. Fixed expressions and the production of idioms. Simone A. Sprenger
22. The grammatical coding of postural semantics in Goemai (a West Chadic language of
Nigeria). Birgit Hellwig
23. Paradigmatic structures in morphological processing: Computational and cross-linguistic
experimental studies. Fermn Moscoso del Prado Martn
24. Contextual inuences on spoken-word processing: An electrophysiological approach.
Danille van den Brink
25. Perceptual relevance of prevoicing in Dutch. Petra M. van Alphen
26. Syllables in speech production: Effects of syllable preparation and syllable frequency.
Joana Cholin
27. Producing complex spoken numerals for time and space. Marjolein Meeuwissen
28. Morphology in auditory lexical processing: Sensitivity to ne phonetic detail and insen-
sitivity to sufx reduction. Rachl J. J. K. Kemps
29. At the same time...: The expression of simultaneity in learner varieties. Barbara Schmied-
tov
30. A grammar of Jalonke argument structure. Friederike Lpke
31. Agrammatic comprehension: An electrophysiological approach. Marlies Wassenaar
32. The structure and use of shape-based noun classes in Miraa (North West Amazon). Frank
Seifart
33. Prosodically-conditioned detail in the recognition of spoken words. Anne Pier Salverda
34. Phonetic and lexical processing in a second language. Mirjam Broersma
35. Retrieving semantic and syntactic word properties. Oliver Mller
36. Lexically-guided perceptual learning in speech processing. Frank Eisner
37. Sensitivity to detailed acoustic information in word recognition. Keren B. Shatzman
38. The relationship between spoken word production and comprehension. Rebecca zdemir
39. Disuency: Interrupting speech and gesture. Mandana Seyfeddinipur
40. The acquisition of phonological structure: Distinguishing contrastive fromnon-contrastive
variation. Christiane Dietrich
41. Cognitive cladistics and the relativity of spatial cognition. Daniel B.M. Haun
42. The acquisition of auditory categories. Martijn Goudbeek
43. Afx reduction in spoken Dutch. Mark Pluymaekers
44. Continuous-speech segmentation at the beginning of language acquisition: Electrophysi-
ological evidence. Valesca Kooijman
45. Space and iconicity in German Sign Language (DGS). Pamela Perniss
46. On the production of morphologically complex words with special attention to effects of
frequency. Heidrun Bien
47. Crosslinguistic inuence in rst and second languages: Convergence in speech and ges-
ture. Amanda Brown
48. The acquisition of verb compounding in Mandarin Chinese. Jidong Chen
49. Phoneme inventories and patterns of speech sound perception. Anita Wagner
50. Lexical processing of morphologically complex words: An information-theoretical per-
spective. Victor Kuperman
51. A grammar of Savosavo, a Papuan language of the Solomon Islands. Claudia Wegener
52. Prososdic structure in speech production and perception. Claudia Kuzla
53. The acquisition of niteness by Turkish learners of German and Turkish learners of
French: Investigating knowledge of forms and functions in production and comprehen-
sion. Sarah Schimke
54. Studies on intonation and information structure in child and adult German. Laura de
Ruiter
55. Processing the ne temporal structure of spoken words. Eva Reinisch
56. Semantics and (ir)regular inection in morphological processing. Wieke Tabak
57. Processing strongly reduced forms in casual speech. Susanne Brouwer
58. Ambiguous pronoun resolution in L1 and L2 German and Dutch. Miriam Ellert
59. Lexical interactions in non-native speech comprehension: Evidence fromelectro-encephalography,
eye-tracking, and functional magnetic resonance imaging. Ian FitzPatrick
60. Processing casual speech in native and non-native language. Annelie Tuinman

The Language of Rotokas, Papua New GUINEA

Uploaded by

Copyright:

Available Formats

The Language of Rotokas, Papua New GUINEA

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Language of Rotokas, Papua New GUINEA

Uploaded by

Copyright:

Available Formats

Split Intransitivity in Rotokas,

a Papuan Language of Bougainville

Perairi broke the spear because he speared himself with it.

Raviata distributed food to everyone.

Riopeiri, you think too much.

She killed the pig.

Jesus told Peter so that he would go catch many sh.

They placed the stone insecurely and it fell down.

The two of them ordered their two daughters. [C.1]

Father is whipping the boy because he was disobedient.

The woman is nursing the child.

When you were little, you planted this coconut tree.

I am eating a really strong Singapore fruit.

Now Im marrying you. [Firchow and Akoitai (1974:71)]

The tree that is in Rarova, I got roots from it.

Youre big thieves, you stole the sh.

Your knife, it rusted.

Your auntie, she is weeding the sweet potato garden.

Thats their house that youre looking at.

Joseph is gathering all of his tools.

Thats our house which were going into.

I wasnt going to church in the past.

Okay, theyre smiling at each other and shook hands. [RR-Ata:49]

I am jumping with joy.

Kokora is building a new house.

Sadly, you killed my mother. [Firchow and Akoitai (1974:80)]

Yesterday we looked at frogs.

The bananas are very heavy.

I sat down for a long time and now Im numb.

I returned to the village last because they left me.

The boys armpits had sores.

My leg swole up because I bumped it.

It rains every day.

Kepi, why are you racing with me?

How much money do you have on you?

Why did he hit you?

Afternoon, in-law, where are you going?

Who took my axe?

We saw the marks on his body from where he cut himself.

Ill wait for you while you go ahead.

Hey, wait, woman, settle down! [Firchow and Akoitai (1974:71)]

The man is watching the women as they bathe.

Water that is muddy is bad, as you cant drink from it.

You cant see a man in the dark.

The other four little logs are on the clothes. [CB:LR]

She gave birth to (more) children as the others were walking.

There is not something that is bad (i.e., there is nothing wrong).

I smell something that stinks.

Clear everything lest you leave one behind.

Skip Firchow was an American who came to translate Rotokas.

He killed him. [Caleb, Another Togarao Story]

I only got two little chickens.

I really was just going days ago. [(Firchow, 1987:16)]

One day I went to Ibu while they fought.

I jumped up and down truly happy.

Were going to collect rewood.

The ying fox sleeps hanging.

Shes killing us two! [Firchow and Akoitai (1974:32)]