ASPECTS OF SWITCH REFERENCE IN MARUBO, A PANOAN LANGUAGE OF WESTERN AMAZONIA by C. SEAN SMITH Bachelor of Arts in Biblical Studies, Moody Bible Institute, 2009 Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of MASTER OF ARTS IN LINGUISTICS in the FACULTY OF GRADUATE STUDIES TRINITY WESTERN UNIVERSITY MAY 2021 © C. Sean Smith, 2021 i Abstract This thesis is a description of the switch reference system in Marubo (ISO 693-3 [mzr]), an underdescribed Panoan language of western Amazonia. Marubo has a fascinating switch reference system which carries a high functional load in the grammar. This complex system, comprised of nine markers, tracks referents across clauses, often displaying sensitivity to an argument’s grammatical role; temporal and logical relations are also encoded as extended functions of the markers. Of particular typological interest is the cross-referencing of O arguments with other S, A, or O arguments. Switch reference markers may occur in clause chains where they target either adjacent or non-adjacent clauses. In certain cases, the standard order of clauses may be reversed, often producing a reading which elaborates on the preceding information. In addition, non-coreferential clauses may be interposed in clause chains for brief alternations of topic. Lastly, areas which deserve more study are presented, such as the flexible use of switch reference to mark discontinuities related to time, weather and events. All data and analysis come from four years of immersion-based fieldwork by the author, with abundant examples from a variety of Marubo discourse genres. ii Acknowledgements The completion of this thesis has only been possible through the help, encouragement, prayers and support of many. First, I would like to thank the Marubo people for their hospitality, friendship and patience as they shared their language and ways of life with me and my family. From running after yawa, to eating in the shovo, to trying to play the ako, I have learned so much together with you all. To all of my language teachers–Saĩpapa (João Marques Cruz), Nomanawa (Arnildo Marques Cruz), Võchĩpa (Paulo Nascimento), Võpa (Sílvio Dionísio), Votẽpa (Samuel Isaac), Tota (Altanir Oliveira), Venãpa (Robson Dionísio Doles), Rarẽewa (Madalena de Oliveira), Panã (Sidney Brasil Dionísio), Vama (Daniel), and Vinã (Ronaldo do Nascimento Domingos)–I am so grateful for the fascinating stories and careful descriptions of how to use your language. You all have made learning truly enjoyable. I am also indebted to all of the kakáyas, who after inviting me, gave their verbal and written permission for me to live and work among them. Yorarasĩní, roaka akĩ ea mã vesovaĩvaĩchná. Askásho matomẽ mekíai iinã mato ẽ yoãshoávere. Thanks to SIL International for allowing me to devote several years to language and culture acquisition and for granting time to pursue further education. Thanks, Dave Eberhard, for being an encouraging language coach and guiding me through the Growing Participator Approach. I’m grateful to Tom Payne for talking me through one of my switch reference challenges. Thanks, Josh Smolders, for our virtual writing sessions/linguistic chats and formatting tips. To the mapping team–John Amdahl, Cartographer, Joe Bryant, Project Manager, and Matt Benjamin, Lead Cartographer–thank you very much for your work in putting together the fine-looking map used in this thesis. 1 Besides the many teachers and professors that have instructed me in various subjects over the years, I’m especially thankful for Steve Clark, who taught my first linguistics class (phonetics). To my thesis committee: Sean Allison, thanks for making yourself available to talk, for your willingness to read countless drafts, for the quick and detailed feedback, and for guiding me through this whole process. Steve Nicolle, I am grateful for your insightful comments and suggestions which have resulted in a much clearer presentation. Pilar Valenzuela, thank you for 1 The map in this publication includes the use of geodata from ©Esri. You can see a list of specific data sources here: Esri Data & Maps: https://www.esri.com/content/dam/esrisites/en- us/media/legal/redistribution-rights/redistrights-104.pdf. The map also includes the use of geodata from UNEP-WCMC and IUCN (2015), Protected Planet: Indigenous areas Brazil: The World Database on Protected Areas (WDPA)/OECM Database [On-line], May 2015, Cambridge, UK: UNEP-WCMC and IUCN. Available at: www.protectedplanet.net. The map is used by permission, and redistribution is not permitted. iii stepping up at such short notice, for your thorough review and expert input; this thesis greatly benefitted from your experience in Panoan languages. While I am thankful for each member’s feedback, any errors are, of course, my own responsibility. Roberto Zariquiey, thanks for our virtual conversation on switch reference. Thanks to my colleagues, Paul and Sheryl Rich, for welcoming us, helping us in so many ways, and for answering many questions about the Marubo people and language. To Jevon and Danica Rich, thanks for being such generous and caring neighbors. I’ll never forget how you let us use your house during our first session (a huge boost for language learning) and for the times you cared for us when we were sick with malaria. Obrigado, Palmira Geralda da Silva, por nos receber tão bem e por nos levar para visitar quase todas as aldeias quando chegamos. Te apreciamos. Thanks to MNTB for all of the logistical support, including Izaquiel Melgarejo who regularly checked up on us and often sent us much needed items, and to Joel Rich, who organized countless flights for us. Thanks also to all the pilots who flew us safely to and from the village. Obrigado, Everaldo e Ana Brum, pelo carinho e pelas várias maneiras que nos apoiaram. Special thanks to Facundo Nardone for passing on data from Katukina. To my best friend, Diego Senn, who left us three years ago, I dedicate this thesis in your memory. To Dad and Mom, thanks for your love and support, and for instilling a love for learning from a young age. To my kids, thanks for frequently interrupting my writing/research sessions with your creative drawings, interesting questions or comments, or calling me to come eat–I love you guys so much! Melissa, I couldn’t have done this without you. Thanks for your encouragement, your selfless and caring attitude, for reminding me to take snack breaks, and for all of our grammar conversations that definitely enriched this thesis. This journey, among all our adventures, has been so much better with you beside me, my love. Finally, thanks to ὁ διδάσκαλος ‘the Teacher’, the ultimate giver of all wisdom and understanding. iv TABLE OF CONTENTS CHAPTER 1: INTRODUCTION .............................................................................................................. 1 CHAPTER 2: BACKGROUND ................................................................................................................. 3 2.1 MARUBO PEOPLE .................................................................................................................................... 3 2.2 MARUBO LANGUAGE OVERVIEW ........................................................................................................... 5 2.3 PREVIOUS MARUBO LINGUISTIC PUBLICATIONS.................................................................................. 9 2.4 FIELDWORK AND APPROACH ................................................................................................................. 9 CHAPTER 3: WHAT IS SWITCH REFERENCE? .............................................................................. 11 3.1 MAJOR VIEWS ABOUT THE NATURE AND FUNCTION OF SWITCH REFERENCE ................................. 11 3.2 TYPOLOGICAL OVERVIEW AND TERMINOLOGY ................................................................................. 13 CHAPTER 4: SWITCH REFERENCE FORMS IN MARUBO .......................................................... 17 4.1 OVERVIEW ............................................................................................................................................ 17 4.2 -VAI ‘S/A>S.PREV’ .................................................................................................................................. 19 4.3 -VAIKI/VAIKĨ ‘S/A>A.PREV’.................................................................................................................... 21 4.4 -I ‘S/A>S.SIM’ ......................................................................................................................................... 22 4.5 -KI/KĨ ‘S/A>A.SIM’.................................................................................................................................. 25 4.6 -SHO/SHÕ ‘S/A>S/A’ ............................................................................................................................... 27 4.7 -INÃ ‘S/A>S/A’ ........................................................................................................................................ 31 4.8 -A ‘O>S/A/O’ ........................................................................................................................................... 34 4.9 -MAĨNÕ ‘DS/A’ ........................................................................................................................................ 38 4.10 -TXINĨ ‘DS/A’ ........................................................................................................................................ 42 4.11 PANOAN SWITCH REFERENCE COGNATES......................................................................................... 45 CHAPTER 5: TYPES OF SWITCH REFERENCE CONSTRUCTIONS .......................................... 53 5.1 SWITCH REFERENCE IN NON-DEFAULT ORDERED CLAUSES .............................................................. 53 5.2 THE SCOPE AND TARGET OF SWITCH REFERENCE CLAUSES ............................................................. 58 5.3 INTERPOSED NON-COREFERENTIAL CLAUSES .................................................................................... 70 5.4 EXCEPTIONS TO SWITCH REFERENCE AND TOPICS FOR FURTHER RESEARCH................................. 75 CHAPTER 6: CONCLUSION ................................................................................................................. 81 REFERENCES .......................................................................................................................................... 84 APPENDICES ............................................................................................................................................ 87 APPENDIX A: EVEN/ODD SYLLABLE ALLOMORPHY ................................................................................. 87 APPENDIX B: KAMÃNE PIKẼNKEÃ YOÃ (THE JAGUAR’S ALMOST EATING STORY) .................................. 93 APPENDIX C: PANI AKÁ (TO MAKE A HAMMOCK) .................................................................................... 99 v LIST OF FIGURES Figure 1 Map of the Vale do Javari reservation including several Marubo villages ...................... 4 Figure 2 Marubo village on the upper Ituí (photo taken by J.D., 2008) ......................................... 5 Figure 3 Crafting a palm-fiber hammock (photo taken by Melissa Smith, 2016) ...................... 105 vi LIST OF TABLES Table 1 Inventory of Marubo consonants ....................................................................................... 8 Table 2 Inventory of Marubo vowels.............................................................................................. 8 Table 3 Marubo switch reference markers.................................................................................... 17 Table 4 Panoan switch reference markers indicating subsequent events or purpose ................... 47 vii Abbreviations > Ø 1 2 3 A ABS ADJZ ADVS ANA ANTIMIR ASS AUG BEN CAUS CHAR COM COMPL COND DEM DET DETR DIM DIR DIST DISTR DM DS/A DUR EMPH ERG EXH EXP FOC FRUST FUT HAB HOD IDEO IMP IMPOS INSTR INT INTJ INTS INTR IRR ITER lit. LOC MAL is coreferential with zero (null) 1st person 2nd person 3rd person agent-like argument of transitive clause absolutive adjectivizer adversative anaphoric antimirative assertive augmentative benefactive applicative causative “characterizer” comitative completive aspect conditional demonstrative determiner detransitivizer diminuative directional distal distributive developmental marker different S or A arguments durative aspect emphatic ergative exhaustive expanding focus frustrative future habitual aspect hodiernal (today’s past) ideophone imperative impossibility instrumental intent interjection intensifier intransitive irrealis iterative aspect literally locative malefactive applicative viii MCM1 MCM2 MCM3 NEG NMLZ O OBL PA PAR PL POSS PREV PROSP PROX PST1 PST2 PST3 REP REPL Q S SG SIM SIML TOP VER VOC main clause marker 1 main clause marker 2 main clause marker 3 negation nominalizer patient-like argument of transitive clause oblique argument participant agreement parallel plural possessive previous event prospective aspect proximal recent past distant past far distant past repetitive repletetive question particle single argument of intransitive clause singular simultaneous event similitive topicalizer veridical vocative 1 Chapter 1: Introduction Switch reference was first identified by Jacobsen (1967) in the Hokan-Coahuiltecan language, although its canonical realization was later defined by Haiman & Munro (1983: ix) as “an inflectional category of the verb, which indicates whether or not its subject is identical with the subject of some other verb”. Since its establishment as a cross-linguistic category, switch reference systems have been demonstrated to be more complex and often difficult to categorize precisely. For example, while such systems track coreferential or noncoreferential relations between clauses, they may also include the marking of other functions such as temporal and logical information. Nevertheless, most theories of switch reference agree that referential tracking across clauses always forms part of its function (Gijn 2016; Roberts 2017; Stirling 2006). In Marubo (ISO 639-3 [mzr]), an underdescribed Panoan language of western Amazonia, switch reference system plays a major role in the grammar. This system tracks referents across clauses, often displaying a sensitivity to an argument’s particular syntactic role; furthermore, its forms may also encode temporal and logical relations. While previous linguistic investigation includes a grammar sketch (Kennell 1978), works on phonology (Costa 1992; Costa 2000) and a study of alignment (Costa 1998), there has not yet been a description of switch reference in this language. Thus, a major contribution of this thesis is to describe, for the first time, the form and function of this phenomenon in Marubo. Beginning in chapter 2, I present background information, briefly introducing the Marubo people, their physical context, and some of their subsistence activities. I also share basic sociolinguistic information relevant to the vitality of their language, list possible genetic classifications, summarize some of the major linguistic features, and name major publications 2 made about the Marubo language. Finally, I provide details about fieldwork and the approach used in this thesis. In chapter 3, I illustrate the canonical use of switch reference and discuss some of the major theoretical views on the topic. To provide perspective, I explain some of the typological features of switch reference, its geographic distribution and the terminology used to describe clause types. I conclude this chapter by mentioning some characteristics of switch reference in Panoan languages. Chapter 4 is the central, descriptive unit of this thesis and features each of the nine switch reference forms in Marubo. For each form, I offer numerous examples along with some crossreferences to additional material in the appendices. When relevant, I also make explicit the possible comparative links between Marubo and other Panoan languages. For example, I conclude the chapter by noting how two cognate forms do not function as switch reference markers in Marubo. In chapter 5, I build upon the previous chapter by explaining some of complexities within Marubo switch reference constructions. These include: a discussion of structures that do not follow the default order, clause chaining and the scope of switch reference, and the unusual interposition of non-coreferential clauses. Additionally, I document some of the exceptions to the switch reference system and the need for further research in these as well as other areas. Chapter 6 concludes this thesis and summarizes the Marubo switch reference system. 3 Chapter 2: Background 2.1 Marubo people The Marubo people live in the Vale do Javari reservation located in the westernmost parts of the Brazilian Amazon basin. While they recognize the designation “Marubo”, they call themselves Yora, which may also refer to the general category of ‘human’ or ‘person’. Similarly, they refer to their language as Yorã vana ‘Yora’s talk’ or ‘people’s talk’. According to a census taken in 2014 by the Secretaria Especial de Saúde Indígena [Special Secretariat of Indigenous Health], their total population is approximately 2000 people (ISA 2018). The Marubo are believed to be an amalgamation of several Panoan groups (Werlang 2001), possibly having some connections to the Panoan Kulinas of São Olivença (Fleck 2007). While they used to be an interfluvial people, in the last 70 years they have split off into two major groups which currently live in various villages on the upper and middle Curuçá and Ituí rivers, tributaries of the Javari and Itaquaí respectively (see Figure 1 below). Several other indigenous groups are located in the same reservation, including: the Korubo, Kulina, Matis, and Matses (Panoan), the Kanamari and Tsohom Dyapá (Katukinan), and some isolated groups whose languages are unknown. Today, the Marubo do not interact regularly with other groups apart from contact at occasional meetings pertaining to health, education, or politics. Such meetings often occur in the city of Atalaia do Norte, which may take between two and six days to reach, depending upon the river level and village of departure. Towards the southwestern end of their reservation lies the city of Cruzeiro do Sul, where they might encounter other Panoan speaking people, such as the Katukina, or Huni Kuin (Kaxinawá). 4 Figure 1 Map of the Vale do Javari reservation including several Marubo villages The Marubo practice slash-and-burn farming, growing an impressive assortment of crops, including different species of manioc, bananas, corn, and squash as well as inhame, cará, papaya, and sugar cane. This diet is supplemented with seasonal palm fruits such as peach palm, açaí, buriti (Mauritia flexuosa), and Oenocarpus palm species (patauá and bacaba). While some chickens and pigs are raised for consumption, the Marubo still rely strongly upon hunting and fishing for their meat. The Marubo village (illustrated in Figure 2) varies greatly in size and population and may contain a single longhouse (called shovo) or several. Often each shovo contains several nuclear family groups, averaging about 20 people in all. The shovo is surrounded by smaller houses on 5 stilts called a tapo, which typically are used to store personal belongings.2 Villages also usually have a school, a health post (containing a radio) and a soccer field. Figure 2 Marubo village on the upper Ituí (photo taken by J.D., 2008) 2.2 Marubo language overview Researchers differ in their classification of the Marubo language, with studies such as Loos (1999) placing it in the Capanawa subgroup, and Fleck (2013) grouping it as Mainline branch, Nawa group, Marubo subgroup. Additionally, Valenzuela & Guillaume (2017) group Marubo as Central-Southern Branch, Marubo-Katukina subgroup and the Ethnologue (Eberhard, Simons & 2 For a detailed study of the Marubo shovo, see Melatti & Melatti (1986). Several other dissertations have been published about Marubo in the field of anthropology, including an ethnography on healing rituals (Melatti 1985), a study of the saiti myth-chants (Werlang 2001), a study of “verbal-arts” (Cesarino 2008), and an ethnohistory (Welper 2009). For a more complete list of works related to Marubo history and culture, refer to Anonby & Holbrook (2010). 6 Fennig 2020) classifies it as Mainline, Shipibo (subgroup). Despite the relatively long period of time that groups on the Curuçá and Ituí rivers have been separated, to my knowledge, there are no significant dialectical differences. In terms of language vitality, at present, Marubo seems fairly stable. Children who are born in and live on the reservation learn Marubo as their first language. Most men eventually learn to communicate at a basic level in Portuguese, while most women understand some Portuguese but do not speak it. This however, is changing to some degree with the increase in youth who move to the cities of Atalaia do Norte or Cruzeiro do Sul to enroll in primary schools. Based on a more objective evaluation known as the “Expanded Graded Intergenerational Disruption Scale” (EGIDS),3 Marubo falls into the level known as “Developing”. This level signifies that “the language is in vigorous use, with literature in a standardized form being used by some though this is not yet widespread or sustainable” (Eberhard, Simons & Fennig 2020). Certainly, the availability and use of literature in Marubo is due to the creation of an alphabet and robust literacy program by Missão Novas Tribos do Brasil in the early 1970s.4 For example, many users under the age of 50 are literate at a basic level and occasionally make signs or send written notes; remarkably, when in the city, almost all use written Marubo for social media and text messaging. Turning now to grammatical features, the order of constituents in Marubo is AOV/SV5 in unmarked clauses; often, arguments are covert. Morphologically, Marubo is mostly agglutinative and has a fairly synthetic verbal complex as well as forms that are highly syncretic; that is, it is 3 This scale was developed by Lewis & Simons (2010), based upon Fishman (1991), and provides 13 levels which used to evaluate language endangerment and vitality. 4 Currently, almost every major village has a government-funded bilingual school which is staffed by one or more Marubo teachers. 5 A=most agent-like argument of transitive clause; O=most patient-like argument of a transitive clause; V=verb; S=single argument of an intransitive clause. 7 not uncommon to find affixes which are identical in form but have up to three (or more) separate functions.6 Dependent marking is the main strategy used to show agreement, though some head marking is present as well. In terms of alignment, ergativity is displayed by case marking, although there are splits in this system. Other features include associated motion, transitivity agreement,7 as well as a complex switch reference system typical of Panoan languages (Valenzuela 2003). Observe examples (1) and (2) below, a simple intransitive and transitive clause respectively, which illustrate one of the alignment patterns:8 (1) Tama-Ø nachi-katsa Tama-ABS bathe-FUT ‘Tama will bathe’ (2) Tamã txona-Ø pi-katsai Tama\ERG9 woolly.monkey-ABS eat-FUT ‘Tama will eat woolly monkey’ (cf. absolutive Tama) In (1), the proper noun Tama (S argument) is unmarked for case. In (2), while the O argument txona ‘woolly monkey’ is also unmarked, Tama is marked with the ergative nasal suprafix, marking it as the A argument.10 The Marubo data throughout this paper is represented in the practical orthography which is currently used in all the bilingual schools; this is based on a phonological analysis carried out by Kennell (1978). I have adapted the orthography in some minor cases in order to represent word breaks more consistently (clitics written together with their hosts rather than written 6 For example, on nouns, nasalization of the final syllable can mark either ergative, locative, instrumental, or possessive cases depending in which context it is used. 7 Transitivity agreement refers to the use of distinct suffixes, auxiliaries, or pro-verbs depending upon the transitivity value of the corresponding verb (Valenzuela 2003). 8 There are several splits depending on how the verb is inflected. In contrast to (2) above, in transitive constructions which take the impossible modal, habitual aspect, or negation markers, the A and O are both (null)-marked (see Costa 1998 for examples). The motivation for this split seems to be one of realis/irrealis, as all three of these markers lean towards the irrealis side of the spectrum, but more analysis is needed to confirm this theory. 9 Note that, following the Leipzig Glossing Rules, I mark what is known as “morphophonological changes” (in this case, the nasal ergative suprafix), with a backslash and then provide the uninflected form in parentheses following the free translation. 10 Throughout the rest of this paper, I will not mark S or O arguments with the null symbol, since they are the default. 8 separately) and to fully represent the underlying forms of affixes (e.g. where elision at morphological boundaries results in a shorter surface form). The Marubo consonants number 14 and may be found in Table 1 below: Table 1 Inventory of Marubo consonants Bilabial Alveolar Plosives Nasals p Retroflex Palatal t m Velar k n ɾ Flaps Fricatives PostAlveolar β Affricates s ʃ ͡ts ͡tʃ Approximants ʂ j w (Note that orthographically, these are written as: p, t, k, m, n, r [ɾ], v [β], s, ch [ʃ], sh [ʂ], ts [t͡s], tx [t͡ʃ], y [j], and w) There are four oral vowels as well as their four nasalized counterparts, which are shown in the table below: Table 2 Inventory of Marubo vowels Front Central High i, ĩ Mid ɨ, ɨ ̃ Low a, ã Back o, õ (Orthographically, these are represented as: a, i, e [ɨ], o)11 In general, primary stress falls on the first syllable of the word, but nasalized syllables typically attract stress. Any variation to this basic pattern is indicated by an acute accent, which is also used to mark secondary stress. 11 Note that the symbol [ɨ] (high central) is used to represent the mid central vowel. Although acoustic analysis points to [ɘ] (the close-mid central unrounded vowel) as its true quality, I have chosen to represent this sound with [ɨ] in keeping with Panoan tradition. To be clear, this is also not the “schwa” symbol [ə]. 9 2.3 Previous Marubo linguistic publications The following is a list of the major linguistic descriptions of the Marubo language, a full bibliography of which can be found in Fleck (2013). Kennell (1978) is the earliest phonological and grammatical sketch, which is presented in Tagmemic Theory. Switch reference is not explicitly identified, but a few of the forms are described as maintaining the same referent as well as encoding temporal functions. Costa (1992) is a thesis on the stress system as well as some description of case marking. Costa (1998) is an article on alignment which describes (split) ergativity and split intransitivity. Costa (2000) is a dissertation on the Marubo phonology based on several formal frameworks including Optimality Theory. As already noted in the introduction, Marubo is an understudied language, so no full grammatical description exists. Thus, almost all of the glosses and analyses presented are based on my own research. However, some of my work related to phonology is compatible with Kennell (1978) and some analyses of morphosyntax agree with Costa (1998). 2.4 Fieldwork and approach Between 2015-2019, my wife Melissa and I spent a total of 25 months in various Marubo villages under the auspices of SIL International.12 During this time, I devoted many hours to language learning, both one-on-one settings with teachers and in large groups. The data in this study is based upon a collection of different sources including: a ~2000 word database in FieldWorks, a corpus of over 50 transcribed texts, and phrases and short stories elicited in language acquisition. Genres of texts/elicited data include: narrative (folktales, historical accounts, recounting stories from wordless picture books), hortatory (e.g. instruction on food 12 SIL also has a working partnership with Missão Novas Tribos do Brasil in this location. 10 taboos during pregnancy), procedural (e.g. how to hunt an armadillo) and a mixture of genres in dialogue (radio conversations between individuals). Audio recordings were produced from over 15 different speakers from several different villages on both the Ituí and Curuçá rivers.13 These included both male and female speakers, whose ages range from 25 to 80. The total time for completed transcribed texts is over two hours (138 minutes). As to its approach, this thesis is descriptive in nature and mostly proceeds from form to function. Along with many other recent linguistic descriptions, I base my work in what has come to be known as “basic linguistic theory”. In contrast to formal theories which are often deductive, basic linguistic theory is inductive and is concerned about language based on its own internal organization. At the same time, the theory is continually being shaped and expanded by new descriptions of languages. Consequently, while the treatment of switch reference in this paper is primarily descriptive and illustrates its function within the Marubo language, it also may contribute to typological work on the subject (Dixon 2010). In fact, Dryer (2006: 229), when discussing linguistic description within basic linguistic theory, makes the case that it “provide[s] the major source of data for theoretical work in typology. In that sense, descriptive work in basic linguistic theory is always of theoretical significance”. With this in mind, I have included over 75 examples illustrating the function of switch reference in Marubo. Since examples are sometimes difficult to understand on their own, I have attempted to guide the reader by supplying the basic context. As a secondary purpose, I have endeavored to choose examples that are interesting and that allow the reader to catch a glimpse into the world of the Marubo people. 13 Thanks is due to my friend and colleague Jevon Rich, who graciously shared several of his recordings with me from the Curuçá river villages and one from the Ituí river. 11 Chapter 3: What is switch reference? 3.1 Major views about the nature and function of switch reference As mentioned in the introduction, canonical switch reference is a verbal category specifying whether an argument of one verb is the same as or different than an argument of another verb. To illustrate from Mojave, observe example (3) where the A arguments of both clauses are the same; in (4) however, the A arguments of each clause are different, the first being “I” and the second being “Judy”. (3) pap ˈ-akchoor-k ˈ-salyii-k potato 1-peel-same 1-fry-tns ‘After I peeled the potatoes, I fried them’ (4) ˈinyech pap ˈ-akchoor-m Judy-ch salyii-k I potato 1-peel-diff Judy-sub fry-tns ‘After I peeled the potatoes, Judy fried them’ (Munro 1979: 144)14 This phenomenon can be further demonstrated in Usan, a language of Papua New Guinea. In (5), the A argument, ye ‘I’ of the verb su ‘cut’, is referentially the same as the S argument of the verb isomei ‘I.went.down’; this is marked by the same subject marker -ab. In (6), the A and S arguments are not referentially the same, as the first is ‘I’, while the second is ‘it.went.down’ referring to the tree; this change in subject is marked by the different subject marker -ine. (5) Ye nam su-ab isomei I tree cut-SS I.went.down ‘I cut the tree and went down’ (6) Ye nam su-ine isorei I tree cut-DS it.went.down ‘I cut the tree down’ (Reesink 1983: 217-218) Since the establishment of switch reference (SR) as a cross-linguistic category, switch reference systems have been demonstrated to be more complex and often difficult to categorize precisely. 14 Examples from Munro (1979) are presented verbatim, as there is no provision of gloss abbreviations. All other examples follow the Leipzig Glossing Rules. 12 As one author insightfully asserts, “The reason for this ‘evasiveness’ of SR has to do with the fact that, while linguists do not seem to have problems spotting SR systems (as prototypical instances of SR are easy to identify), the various SR systems vary greatly in their lessprototypical behaviours” (de Souza 2016: 55). Understandably, because of the complexity, there are differing views on the nature and function of switch reference, though there is significant agreement that switch reference is primarily a reference tracking device. Haiman & Munro (1983) for example, characterize switch reference as a syntactic device whose function is to avoid ambiguity of reference.15 They argue that there should be no languages in which switch reference tracks first and second person but not third, as the former are easily determined from context, while the latter is often unclear. They also argue that subject most often refers to syntactic subject rather than semantic or pragmatic subject (agent or topic). While somewhat different in their approaches, both Givón (1983) and Stirling (1993) argue that switch reference is a broader discourse device, not only tracking reference but marking topic and event (dis)continuities.16 For example Stirling (1993) identifies instances where different subject markers are used which do not indicate a switch in reference, but rather a discontinuity in agentivity value (in impersonal constructions), time, location, difference in modality (realized and non-realized events), and unexpected change in course of events. It is important to note, however, that both authors concede that reference tracking is at the heart of switch reference systems. 15 In his recent typological study on switch reference, Roberts (2017: 566) holds an almost identical view of its function, stating that it is “primarily a syntactic method of tracking participants in discourse”. He acknowledges cases where switch reference does seem to mark events or situations as same or different, however he maintains that these are extended functions. 16 Givón (1983) argues that the canonical definition of switch reference is too narrow and needs to be defined functionally rather than structurally. 13 Perhaps one of the most diverging views of switch reference is found within a Chomskyan Binding Theory framework. Finer (1985) argues that switch reference does not have a pragmatic function of tracking reference, but is purely a localized syntactic operation, that is, only between adjacent clauses. Furthermore, he claims that switch reference is found only in subordinate, not coordinate clause constructions. However, this theory has shown to be inadequate by more recent typological studies of switch reference (Gijn 2016; Roberts 2017; Stirling 2006) since certain languages possess switch reference markers that do not coreference adjacent clauses but sometimes skip clauses; also, some languages (mainly in Papua New Guinea) do use switch reference markers in coordinate as well as subordinate constructions.17 While the purpose of this thesis is not meant to establish the validity of one theory over another, establishing a definition of switch reference is an important first step in my analysis. Thus, following many others who study switch reference, I recognize the “canonical” definition given by Haiman & Munro (1983) and describe the realization in Marubo as mostly following it, with several exceptions. This is because in Marubo, while switch reference primarily concerns referential tracking, it often includes other functions as well. 3.2 Typological overview and terminology Switch reference is found in several parts of the world, including western and central Australia, Papua New Guinea and Oceania, western North America and western South America; there are 17 It is important to note that the distinction between coordination and subordination is not as clear as often held. For example, Haspelmath (2007) points out that the coordinate structure constraint (proposed in generative syntax) is often used as a formal way of determining coordination vs. dependency (essentially subordination). However, he argues that “only semantically based notions can be applied cross-linguistically – formal criteria are generally too language-particular (for instance, not all languages have extraction constructions that would show the effect of the coordinate structure constraint)” (Haspelmath 2007: 6). Helpfully, Haspelmath (2007) describes coordinate structures as demonstrating symmetry (although there are exceptions) between units, while dependency structures demonstrate asymmetry, with one unit being a head and the other a dependent. In general, Marubo complex clause constructions demonstrate dependency (or subordination). 14 also some cases reported in Africa and northern Eurasia. Many consider switch reference to be an areal phenomenon, as linguistic areas with unrelated groups share the feature (Gijn 2016). As mentioned in the canonical definition above, switch reference is an inflectional category that is usually marked on the verb (although in some cases, it can appear as free form clitics). Typologically, it is found in head-marking languages which do not have an elaborate case or gender system; often, these languages allow for the omission of core verbal arguments (Stirling 2006). Usually, both same subject and different subject markers are found, with some languages possessing several forms for each marker. However, some languages contain only one marker, with the same subject form commonly being the unmarked one (Roberts 2017). There is quite a variety of terminology used to describe switch reference clause types. The clause that contains the switch reference marker has been called the marked, medial, or nonfinal clause while the clause being cross-referenced has been called the reference, matrix, controlling, main or final clause, among others. In this thesis, I adopt the terms marked and matrix clause following the descriptions of other Panoan languages (Fleck 2003; Valenzuela 2003; Zariquiey 2011). The marked clause is always the one that contains a switch reference marker and is subordinate to another clause. The matrix clause, however, may be a subordinate or main (independent) clause. That is, there are certain constructions where a marked clause is subordinate to a following controlling clause (not the main clause), which is in turn also marked and subordinate to another controlling clause (see §5.2). In these cases, I specify whether the matrix clause is a main clause or another marked clause. Cross-linguistically, the marked clause verb lacks TAM (tense, aspect, modality) markers; the matrix clause, on the other hand, is often but not always the final clause and is not marked for switch reference but contains TAM information. The relationship between the 15 marked and matrix clause is often described as one of subordination (Stirling 2006). In addition, (Roberts 1988) gives compelling evidence for switch reference being used in coordinate clauses, citing not only Amele (seen perhaps by some as an exception), but other Papuan languages as well. In Panoan languages, switch reference is complex18 and usually involves a large set of markers specifying which core argument is being coreferenced. For example, there is often a form that coreferences a matrix clause’s S argument and a distinct form that coreferences a matrix clause’s A argument. Some Panoan languages even coreference O arguments. This orientation towards a particular argument finds it origins in a system known as “participant agreement (PA)”. As Valenzuela (2005: 259) describes, “participant agreement (PA) can be considered the typologically most salient feature of Panoan grammar. It refers to the use of a distinct inflectional morphology on adjuncts, in correlation with the syntactic function of the participant they are predicated of”. As is demonstrated in Valenzuela (2003), this system functions at both the intraclausal and interclausal levels, thus many Panoan languages contain switch reference morphemes which are similar or identical both in form and function (their orientation towards either S, A, or O arguments) to that of participant agreement morphemes marked on non-clausal adjuncts. The status of intraclausal participant agreement markers in Marubo is not yet entirely clear and needs future study; however, cognate forms are highlighted either in the body or in a footnote. Besides tracking participants, Panoan switch reference systems also indicate temporal or logical relations between clauses. That is, forms also may encode whether an event in one clause 18 For example, Matses (Fleck 2003) and Kashibo-Kakataibo (Zariquiey 2011) have both been described as having systems with ~20 switch reference markers. 16 occurs previously, simultaneously or subsequently to an event in another clause. Logical relations such as reason, result or purpose may also be encoded. In terms of constituent order, most Panoan languages are AOV/SV and mark switch reference as a suffix on verbs which usually do not contain TAM information. Often, marked clauses precede adjacent matrix clauses, although there are cases where they follow the matrix clause. In some cases, marked clause arguments may coreference a non-adjacent clause. Finally, it is not uncommon to find chaining constructions, where there is a relatively long series of marked clauses followed by a final matrix clause (Loos 1999; Fleck 2003; Valenzuela 2003; Sparing-Chávez 2007; Zariquiey 2011). 17 Chapter 4: Switch reference forms in Marubo 4.1 Overview In §3.2, Panoan switch reference systems were described as being complex and carrying a high functional load at the discourse level, not only marking participant reference across clauses, but also indicating temporal and logical relations. Similarly, the Marubo system is complex, possessing nine markers which have various different functions; this system is presented below in Table 3. Table 3 Marubo switch reference markers SR marker Marked clause Matrix clause Temporal/logical function19 -vai S/A = S Previous -vaiki/vaikĩ S/A = A Previous -i S/A = S Simultaneous/(Manner/Purpose) -ki/kĩ S/A = A Simultaneous/(Manner) -sho/shõ S/A = S/A (Previous/Simultaneous/Manner) -inã S/A = S/A (Result/Purpose) -a O = S/A/O (Previous/Simultaneous) -maĩnõ S/A ≠ S/A (Previous/Simultaneous) -txinĩ S/A ≠ S/A (Previous/Simultaneous) As can be observed, the marked clause and matrix clause columns list the syntactic function of the argument being coreferenced (whether S, A, or O) while the temporal/logical function column lists the relative order of events or the logical relation between propositions (e.g. previous indicates that the event in the marked clause occurs before the event in the matrix clause).20 19 Only the primary function which can be considered a part of the form itself is listed without parentheses. Those which serve as a secondary function, or whose function can only be derived from context are listed in parentheses. 20 As already mentioned in §3.1, the dividing line between dependency (subordination) and coordination is sometimes fuzzy. However, in Marubo switch reference constructions, there is an obvious asymmetry between 18 As is apparent, several of the switch reference markers demonstrate variation in form. These allomorphs seem to be conditioned by the number of syllables (or possibly moras) that precede them within the domain of the word, a phenomenon similar to one identified in ShipiboKonibo by Lauriault (1948). While more research is needed in order to determine all of the factors which influence this variation, there is a clearly identifiable pattern. Consider, for example, the forms presented below with the first line presented phonetically to illustrate syllable boundaries: [ˈɾɨ.ɾa.ki] rera-ki ‘chop-S/A>A.SIM’ ‘chopping’ (8) [ˈni.ka.ki] nika-ki ‘jump-S/A>A.SIM’ ‘jumping’ (9) [pi.ˈkĩ] pi-kĩ ‘eat-S/A>A.SIM’ ‘eating’ (10) [ka.ˈkĩ] ka-kĩ ‘go-S/A>A.SIM’ ‘going’ (11) [ˈɾɨ.ɾa.ma.kĩ] rera-ma-kĩ ‘chop-CAUS-S/A>A.SIM’ ‘making chop’ (12) [ˈni.ka.ma.kĩ] nika-ma-kĩ ‘jump-CAUS-S/A>A.SIM’ ‘making jump’ (7) In (7) and (8), the form -ki ‘S/A>A.SIM’ is found in environments where even number of syllables (two) precede it. Conversely, in (9) - (12), the corresponding form -kĩ ‘S/A>A.SIM’ is found when there is an odd number of syllables (one or three) preceding it. Note that the transitivity of the verb is not a conditioning factor as is the case in other parts of Marubo grammar. clauses which indicate dependency. In general, clauses that are marked by TAM and can be uttered on their own are considered independent or main clauses. Clauses that lack TAM marking and cannot be uttered on their own without a supporting clause are considered dependent or subordinate. 19 This pattern of allomorphy is widespread and can be observed not only with switch reference markers, but also in a number of motion-related, valency changing, and TAM suffixes (see Appendix A). Since most verb roots are disyllabic, the majority of examples throughout this paper contain forms which follow the even syllable (or mora) pattern; however, I do refer to examples in the appendix which contain odd syllable (or mora) forms. In the following sections, I present each of the markers listed in Table 3, giving several examples to illustrate their different functions. The symbol > is adopted in the glosses to show the particular arguments being cross referenced between the marked clause and the matrix clause (following Zariquiey 2011). For example, in a form such as -ki ‘S/A>A.SIM’, the S or A argument of the marked clause would be coreferential with the A argument of the matrix clause. For clarity, I often make use of brackets in order to highlight the particular clauses being analyzed. Finally, since participants are rarely overt, I often list them in parentheses before the marked clause, even though they may have been introduced much earlier. 4.2 -vai ‘S/A>S.PREV’ The first marker presented is -vai ‘S/A>S.PREV’, which can be illustrated by its usage in (13): (13) (ea take…) [ano nõti atxi-vai=se]MARKED [wakari 1SG sibling there boat catch-S/A>S.PREV=DM21 downriver nõ=vo-tã-vai, waa Nelio Moraes ni-á-namã]MATRIX 1PL=go:PL-GO.DO.THERE-PST1 INTJ Nelio Moraes be-NMLZ22-LOC ‘I and brother…after catching a boat there, we went downriver, uhh, where Nelio Moraes lives’ The form =se, while also used to mark focus of assertion, seems to be repurposed as a “developmental marker, [indicating] that the material so marked represents a new development in the story or argument, as far as the author’s purpose is concerned” (Dooley & Levinsohn 2001: 48). This has been observed in both Marubo narrative and procedural texts. 22 The mopheme -a has many functions in Marubo. Besides functioning as a nominalizer, it may also function as a switch reference marker (§4.8), and a hodiernal tense marker. When it is inflected on verbs which function as main clauses, I have glossed it as “MCM3” which belongs to a set of three markers needing more analysis; these I have termed “main clause markers”. 21 20 This story is a personal account about going to get a cow from a Brazilian river community outside of the Marubo reservation. In the first clause, the participants are the narrator and his brother (mentioned in a previous clause), and are the A argument of atxi ‘catch’. This verb is marked with the -vai ‘S/A>S.PREV’ switch reference marker, indicating that the participants in the following clause will be the same, and specifically, that they will refer to the S argument of the verb. As can be observed, the following verb vo ‘go:PL’ is indeed intransitive and thus only takes one argument (S). In addition to specifying coreferentiality, this suffix also indicates that the event in the first clause occurs before the event in the second clause, as is the case here (first they ‘caught’ the boat and then ‘went’ downriver). In the majority of examples that I have collected, -vai ‘S/A>S.PREV’ is marked on transitive verbs, however, it does occur on intransitive verbs as well (coreferencing S with S): (14) (noke…) [paké-vai=se]MARKED [nõ=ve-vai]MATRIX 1PL fall-S/A>S.PREV=DM 1PL=come:PL-PST1 ‘we…after going down (into the boat), we came’ (lit. ‘after falling down…’)23 This example comes from a peccary hunt which occurred several days before it was recorded. The narrator includes himself and several family members as participants (marked throughout the narrative as first person plural clitics). The marked clause verb is intransitive paké ‘fall’ and assumes a previously mentioned referent noke ‘1PL’ as its S argument (even though it is not overt in the marked clause). The matrix clause verb ve ‘come:PL’ is also intransitive, thus it is clear that -vai is used here to mark S arguments as referentially identical. Temporally, the marked clause event ‘going down’ happens before ‘come’, rather than simultaneously or subsequently. The nonliteral use of the verb paké ‘fall’ as ‘go down’ is quite common and usefully describes the descent from the high river bank (where the hunt occurred) into the boat. Also, note that while the form -vai as a switch reference marker is identical to the form for the past tense marker, the meanings are not synchronically related. 23 21 4.3 -vaiki/vaikĩ ‘S/A>A.PREV’ The form, -vaiki/vaikĩ ‘S/A>A.PREV’,24 corresponds to the form found in the previous section; but rather than marking that the S/A is coreferential with the S argument in the matrix clause, it marks that the S/A is coreferential with the A argument. Compare the following example, where (15) coreferences the marked S with the matrix A: (15) (txasho... ẽ=papã…) [richki-vaiki=se]MARKED [ano-sho=se deer 1SG.POSS=father\ERG hit-S/A>A.PREV=DM DEM:ANA-PA=DM pote-i]MATRIX gut-MCM1 ‘deer…my father…after hitting it, from there (he) gutted (it)’ (cf. absolutive papa) In this recounting of a childhood hunting story, the narrator describes how he scared his father when shooting a hidden deer (the father was not prepared for the loud gunshot). After his father finishes off the wounded deer, he field dresses it. As can be observed here, the marked clause contains the transitive verb richki ‘hit’ whose A argument is papã ‘father.ERG’ (from a previous clause) and whose O argument is txasho ‘deer’ (also from a previous clause). The A argument in the marked clause refers to the same person as the A argument in the matrix clause, the father, who ‘guts’ the deer. In addition, ‘hit’ occurs sequentially before ‘gut’, showing that -vaiki also encodes that events in the marked clause occur before events in the matrix clause. In the same way, this form also may be marked on intransitive verbs, coreferencing S with A. This is exemplified in (16): (16) [oo-vaiki]MARKED [ã=nĩkã-a]MATRIX… (yora oni-pa=tõ) INTJ-S/A>A.PREV 3SG=hear-O>S/A/O man brave-AUG=ERG ‘after making the noise oo, (he) hearing…very brave man’ In this narrative, when a jaguar attacks a couple, the husband flees back to his village, leaving his wife and child to face the jaguar alone. When a “very brave man” (only formally introduced at 24 See Appendix C for examples of the allomorph -vaikĩ. 22 the end), hears about the attack, he immediately sets out to find the woman and child, listening for any sign of them. This man (while not overt in the text) is the S of the marked clause, with the interjection oo serving as an intransitive verb.25 The A argument of the matrix clause is referentially also the man, who stops to ‘hear’ any noises that might be made by the woman. Note that the action of calling out oo occurs before ‘hearing’, as also indicated by the switch reference form. 4.4 -i ‘S/A>S.SIM’ In §4.2, we saw that the switch reference marker -vai ‘S/A>S.PREV’ encodes coreferentiality between clausal arguments (S/A>S) and encodes a temporal difference between clausal events (previous). The form -i ‘S/A>S.SIM’ is similar in function, signaling that either the S or A argument in the marked clause will be the same as the S in the matrix clause; but instead of marking a temporal difference between two clauses, -i marks a temporal overlap.26 Consider (17): (17) (Mĩshõ...) korek korek korek korek Mĩshõ IDEO:climb IDEO:climb IDEO:climb IDEO:climb [nee-iná-i]MARKED [ka-í]MATRIX be.on-UPWARD-S/A>S.SIM go-MCM1 ‘Mĩshõ…korek (climbing noise), korek, korek, korek, climbing, (he) went’ In this excerpt, a jungle trickster named Mĩshõ supposedly wants to help a man (who is already high in a tree) get nane ‘jenipapo fruit’. Here, Mĩshõ is the S argument of the marked clause and the S argument of the matrix clause, indicated by the switch reference suffix -i ‘S/A>S.SIM’; 25 It is common for Marubo verbs to be derived by simply adding inflectional morphology onto the root or formative as in the case above. 26 The form -i has an identical function in other sister languages, marking coreferential relations with the matrix S argument. Additionally, in some Panoan languages (such as Shipibo-Konibo), -i may mark intraclausal participant agreement, also indicating a semantic orientation towards the S argument (Valenzuela 2003). 23 furthermore, this form also encodes that the events of ‘climbing’ and ‘going’ happen at the same time.27 While the majority of these markers are inflected on intransitive verbs, they also may occur on transitive verbs, indicating that the A argument of the marked clause is the same as the S argument of the matrix clause: (18) (Mĩshõ…) *rera-ki,28 [rera-i]MARKED [vana-i]MATRIX … Mĩshõ chop-S/A>A.SIM chop-S/A>S.SIM talk-S/A>S.SIM ‘Mĩshõ…while chopping the tree, (he) was saying’… In this example, Mĩshõ becomes angry, and begins chopping down the tree where the man was getting jenipapo fruit. In the marked clause, the A argument of rera ‘chop’ is Mĩshõ, who is also the S argument of the matrix clause (vana ‘talk’). Temporally, these actions overlap, as Mĩshõ chops while talking, not before or after.29 In (17) and (18), the marker’s primary function is one which encodes coreferential relations (oriented towards the matrix S argument) and an overlap in time. However, in (17), it is possible to understand a manner reading as well, describing how Mĩshõ goes (‘climbing’). When the associated motion suffix (-iná ‘UPWARD’) is applied to the stative verb nee ‘be on’, it While the translation ‘climb’ of the form nee ‘be.on’ may seem odd, it belongs to a small class of verbs (including tsaó ‘be.sitting’, ni ‘be.standing’, raká ‘be.laying’) which are stative and describe the particular position of the S argument. For example, in other contexts, the same word could be used to describe the position of a mosquito on one’s body, whether hanging vertically from one’s leg or horizontally on one’s foot; the same would apply to writing, which is “on” a page. Once an associated motion suffix is applied to these verbs however, they become dynamic, with the semantics expressing translational motion, that is, a change of location (see Guillaume 2016). So, once the verb nee ‘be.on’, is inflected with this particular associated motion suffix, the meaning becomes ‘climb’ (literally something like ‘be on, going in an upward direction’). 28 Note that the narrator makes a brief performance error; this will be mentioned in the following section. 29 It is important to mention that the overlap encoded by -i could be translated in several different ways. As Thompson, Longacre & Hwang (2007: 379-380) point out, overlapping constructions are “coterminous”, that is, they start and stop at roughly the same time. Subtypes of this include: “punctiliar-continuous”, “continuouspunctiliar” or “punctiliar-punctiliar”. While the free translation of (18) doesn’t exactly fit any of these three types, (as the matrix clause is also a marked clause and is translated as continuous), it could just as easily have been translated as “while chopping the tree, (he) said…(continuous-punctiliar)”. Additionally, this could be rendered as ‘he chopped the tree while saying’ (punctiliar-continuous). While any of these are possible translations (including the reversed order of clauses), Marubo does not distinguish between these types in the same way English does, since aspect is not inflected on the marked clause verb. Thus, my translation follows the Marubo order of verbs (when it makes sense), and renders the marked clause as continuous (since this seems to one of the clearest ways to indicate dependency/subordination in English). 27 24 expresses translational motion. Examining the matrix clause, we can see that it also contains a verb encoding translational motion (ka ‘go’), Thus, one generalization that may be made is that manner readings are found in constructions where the marked and main verbs express translational motion. This is also illustrated in (19): (19) (txasho... ẽ=papã…) [papi-varã-i]MARKED deer 1SG.POSS=father\ERG carry.on.back-INWARD:PL/TR-S/A>S.SIM [o-yãtá]MATRIX come:SG-PST3 ‘deer…my father… (he) came carrying (it) on his back’ (cf. absolutive papa) In this example, the primary meaning still holds true. The A argument of the marked clause is ‘father’ (found in a previous clause) which is also referentially the S argument of the matrix clause; also, the verbs papi ‘carry’ and o ‘come:SG’ occur simultaneously. But, it can also be understood that this describes the manner in which the father ‘came’, that is, ‘carrying’. Here, the marked verb is inflected with the associated motion suffix -varã ‘INWARD:PL/TR’ (making it translational), and the matrix verb o ‘come:SG’ is a verb encoding translational motion. When we look back at (18), however, where the marked verb rera ‘chop’ is a motion event (but not translational motion), and the matrix clause verb vana ‘talk’ is not a motion event, a manner reading is not possible; here it is only simultaneous action that is encoded. In addition to these two meanings (simultaneous and manner), in even more restricted environments, -i may also indicate purpose. Consider (20) and (21), for example, which are constructions that contain motion verbs in the matrix clause: (20) [Nãkẽpa oĩ-i]MARKED [ka-katsai]MATRIX Nãkẽpa see-S/A>S.SIM go:SG-FUT ‘(I) will go seeing Nãkẽpa (go, in order to see)’ (21) [mĩ=poké-i]MARKED [o-a, Tamá]MATRIX 2SG=visit-S/A>S.SIM come:SG-HOD Tama\VOC ‘Tama, do you come visiting? (come in order to visit)’ (cf. absolutive Tama) 25 Beginning with (20), the arguments in both the marked and matrix clause (A and S) refer to the same person, that is, the speaker who is advising his brother via short-wave radio that he will visit Nãkẽpa. Here the relationship between these clauses is one of purpose, ‘go in order to see’. The same is true of (21), where Tama is the argument of both the marked and matrix clause; here, the relationship is also one of purpose, ‘come in order to visit’. Although a simultaneous reading of these examples is still understood (‘will go, seeing Nãkẽpa’; ‘come visiting’), the semantics of purpose are unquestionable. While used quite frequently in everyday speech for various different functions (‘went to get bananas’, ‘went to fish’, ‘came to ask’, ‘came to drink water’ etc.), these purpose constructions generally occur with the following motion verbs in the matrix clause: ka ‘go:SG’, vo ‘go:PL’, o ‘come:SG’, ve ‘come:PL’.30 In contrast, even though (17) & (19) contain ka ‘go:SG’ and o ‘come:SG’ respectively, purpose readings are incompatible (?(he) went to climb’; ‘?(he) came to carry’). Thus, considering all of this evidence together, while context may allow for a manner or purpose reading, simultaneity is the core function of the -i switch reference marker; that is, the former two should be considered extended meanings, while the latter is the primary meaning. 4.5 -ki/kĩ ‘S/A>A.SIM’ In the previous section, I demonstrated that -i marks coreferentiality between S/A and S arguments in the marked and matrix clause, respectively. The -ki/kĩ switch reference marker is essentially its counterpart, coreferencing marked S/A arguments with matrix A arguments, while 30 In some ways, these constructions share many properties of serial verb constructions (SVCs), one of which is that they almost seem to conceptualize a single event; furthermore, the historical development of SVCs is often traced from verbs of motion. However, the major difference is that these SR constructions overtly mark dependency (subordination) and their intonational properties are not quite “the same as those of a monoverbal clause” (Aikhenvald 2006: 1). 26 also indicating an overlap in time.31 This is evidenced in (18) above, where the narrator makes a small performance error, using -ki ‘on rera ‘chop’, although he quickly corrects himself with -i since the matrix argument is S and not A. In the following example, from a text which begins with a monkey hunt, -ki is marked on a transitive verb, indicating that the A in the marked clause is coreferential with the A in the matrix clause: (22) (txona…) [wetsã txivã-paké-ki]MARKED woolly.monkey another\ERG chase-DOWNWARD-S/A>A.SIM [tekõ-i]MATRIX shoot.with.arrow-MCM1 ‘woolly monkey…another, while chasing downhill, shot (it)’ (cf. absolutive wetsa) As can be observed, the A argument of the marked clause is wetsã ‘another\ERG’, which refers to one of the hunters in the group who is chasing after woolly monkeys. Following this comes the verb txivã ‘chase’, which is marked by -ki ‘S/A>A.SIM’ indicating that the following referent (in the matrix clause) will refer to the same A argument in the marked clause. In the matrix clause, this is indeed the case, as the verb tekõ ‘shoot with arrow’ has an A argument, which is also referentially wetsã ‘another\ERG’. Futhermore, -ki also marks that the events of the marked and matrix clause have a temporal overlap, as the ‘chasing’ and the ‘shooting’ happen simultaneously, not sequentially. Since -ki may also indicate that the S (of the marked clause) is coreferential with the A (of the matrix clause), note the following example where it is marked on an intransitive verb: (23) (kamã-ne, ã=aĩ…) [nika-ki]MARKED jaguar-ERG 3SG.POSS=wife jump-S/A>A.SIM [isõ-ne voí-vaĩ-i]MATRIX urine-INSTR splash-OUTWARD:PL/TR-MCM1 ‘jaguar, wife…while jumping, (it) splashed (them) with urine’ 31 As with -i, -kĩ is also found in other Panoan languages, marking coreferential relations with the matrix A argument and also indicating a semantic orientation towards the A argument at the intraclausal level (Valenzuela 2003). In Marubo however, the “default” form is -ki, with -kĩ alternating based upon preceding odd-syllable (or mora) patterns. See Appendix C for examples of the allomorph -kĩ. 27 In this narrative, when a jaguar attacks a couple, the cowardly husband runs away, leaving the wife alone with her small child; thankfully, the wife quickly finds cover in between two trees. Meanwhile, however, the jaguar paces around them, jumping in the air and urinating on them, in order to tire them out. In the marked clause in this example, the jaguar is the S argument, and in the matrix clause, it is referentially the A argument. Additionally, the verbs nika ‘jump’ and voí ‘splash’ occur simultaneously. Similar to -i, -ki/kĩ may also express the manner in which an action is performed; this is illustrated in (24): (24) [shavá wetsã-rivi mato oĩ-ó-nõ]MATRIX [poké-sho-kĩ]MARKED day another\LOC-FOC:CONTRASTIVE 2SG see-REP-INT visit-BEN-S/A>A.SIM ‘on another day (I) want to see you again, visiting (you)’ (cf. absolutive wetsa) In this example, the speaker is taking his leave, expressing to his hosts that he plans to visit them again in the future. Here, the marked clause occurs after the matrix clause,32 coreferencing A with A (I), and expressing simultaneity. While this is the primary meaning, a manner reading is also possible, describing the way in which the speaker will ‘see’ his hosts, that is, ‘visiting’.33 Although this form -ki/kĩ is essentially the counterpart to -i ‘S/A>S.SIM’, I have not found any occurences of purpose as an extended meaning in these constructions. 4.6 -sho/shõ ‘S/A>S/A’ In contrast to the forms described above, -sho/shõ seems to be one of the most flexible, covering a wide range of functions. While marking that arguments are coreferential, -sho/shõ does not specify a particular argument in the matrix clause. In other words, the A argument in the marked 32 A further discussion of clause order (where the marked clause follows the matrix clause) can be found in §5.1. Although it might be tempting to interpret this whole construction as purpose, this is only because the verb oĩ ‘see’ is inflected with the intent marker -nõ, giving it the meaning ‘want to see’. However, when looking at the marked and matrix clauses together, a purpose reading is not possible (‘?visiting in order to want to see you’; ‘?want to see you in order to visit’). 33 28 clause may be coreferential with S or A in the matrix clause or the S argument in the marked clause may be coreferential with S or A in the matrix clause.34 Another notable difference is that -sho/shõ does not have one specific temporal or logical function associated with it. Unlike -vai or -vaiki/vaikĩ which mark temporal succession and -i or -ki/kĩ which mark temporal overlap (as well as other secondary logical functions), -sho/shõ sometimes takes a reading of temporal succession or overlap, and sometimes indicates manner; however, compared to other forms, it is less clear which of these is the default. For this reason, I have chosen not to include its temporal or logical function in the gloss as with other forms. In the following personal story of moving to a new village, the speaker uses the form -sho, marking that either the S or A arguments of the marked clause are the same as either the S or A arguments of the matrix clause: (25) [ẽ=ketsaa=võ ea mera-sho]MARKED [ea atxi-ke-mtá]MATRIX 1SG.POSS=brother.in.law=PL\ERG 1SG find-S/A>S/A 1SG catch-COMPL-PST3 ‘when my brother-in-laws found me, (they) caught me’ (cf. absolutive =vo) In the marked clause, the A argument is ‘brothers-in-law’ which is coreferential with the A argument in the matrix clause. While the matrix clause contains no overt A argument, it is contextually determined as 3PL, since there must be some A to act on the O argument (the absolutive form ea ‘1SG’). Considering other possible semantics, -sho seems to express temporal succession or overlap, which is best translated in English as “when”. While the temporal values described in §4.2-4.5 are clearly either previous or simultaneous events, here, the events of “finding” and “catching” seem to be simultaneous, but it is possible that some time has elapsed between the events (succession). This form could also simply be glossed as ‘same-subject’, but in keeping with the rest of the paradigm, which specifies the particular argument being cross-referenced (as well as with other Panoan works, which include this cognate), I have decided to gloss it as ‘S/A>S/A’. 34 29 In (26), rather than coreferencing A arguments, -sho marks that the S argument of the (first) marked clause is coreferential with the A argument of the matrix clause. (26) eã [vevo-kaĩ-sho]MARKED [wanĩ vi-shõ,]MARKED 1SG\ERG be.in.front-OUTWARD:SG/INTR-S/A>S/A peach.palm get-S/A>S/A [waka a-shõ]MARKED [mato maná-ta-nõ]MATRIX drink make-S/A>S/A 2PL wait.for-GO.DO.THERE-INT ‘“When I go out in front, (and) get peach palm, (and) make drink, (I) want to wait for you (out there)”’ (cf. absolutive ea) This portion of narrative contains the direct speech of a man speaking to his fellow villagers; in it, there are 3 marked clauses which are coreferential with the matrix clause. The first marked clause does not contain an argument, but referentially is eã ‘1SG\ERG’, and is coreferential with the A in the matrix clause (note that it is ergative because it is governed by the matrix clause, not because there are two arguments in the first marked clause). As in (25), the argument in the matrix clause is covert but identified by context, since mato ‘you’ (the O argument) is being acted upon by someone (the man). The temporal succession between these marked clauses and matrix clause is clear in this case. The first event, “go out in front”, happens before “wait for”. In the second marked clause (coreferencing A>A), “get” also occurs before “wait for”. Finally, in the third marked clause (coreferencing A>A), “make” also occurs before “wait for”.35 In the two examples above, we saw that it was the A argument that was cross-referenced in the matrix clause. In the following two examples, it is the S that is cross-referenced: 35 Note the iconic relationship between the order of events and the syntax in which they occur. 30 (27) (yora=rasĩ)… [tipi sana-ni-sho=se]MARKED person=PL gun raise-UPWARD-S/A>S/A=DM [para i-vaĩ-vai=vo]MATRIX fall:PL do.INTR-OUTWARD:PL/TR-PST1=PL ‘the people…with their guns raised upwards, fell 36 outwards (into the water)’ OR ‘the people…while raising their guns upwards, fell outwards (into the water)’ In this excerpt from a hunting narrative, the A argument of the first clause, “people” also refers to the S argument of the matrix clause, which is covert. Here -sho also describes the manner in which the “falling” occurred, that is “with guns raised upwards”. While a temporal reading (simultaneous) is possible, speakers tended to emphasize how the matrix verb’s action was performed, not the temporal aspect. In (28), which cross-references S with S, we can see that this is also the case: (28) [aa=ro peshé teké wetsã ronó-iná-sho]MARKED 37 3SG=FOC:PAR shelter section another\LOC be.hanging-UPWARD-S/A>S/A [ã=osha-a]MATRIX 3SG=sleep-MCM3 ‘(she) was hanging (in her hammock) at another section of the shelter, sleeping’ (cf. absolutive wetsa) In this story of a woman who gets eaten by a jaguar, both the marked and the matrix clause arguments are S arguments (the woman), as indicated by -sho. It is also evident that the relationship between the two clauses is one of manner, describing how the woman sleeps (in a hanging position). Considering each of the examples above, it is apparent that -sho/shõ indicates that arguments across clauses are coreferential; however, it does not specify which particular 36 The intransitive form i (or ik) and its transitive counterpart a (or ak) vary greatly in meaning and function depending on their contexts. For example, they may occur as speech verbs (‘say’ vs. ‘tell’), or as auxiliaries, or as pro-verbs (replacing semantic content of either intransitive or transitive verb with ‘do’). The transitive form a (or ak) may also mean, ‘get’, ‘make’, ‘drink’ or ‘copulate with’. When these forms function as auxiliaries (where they don’t actually contribute the meaning ‘do’), I gloss them as AUX:TR or AUX:INTR and when they function as a pro-verb or take the meaning ‘do’ , I gloss them as do:TR and do:INTR. 37 Since Marubo seems to have at least four subtypes of focus markers, I have glossed them following work by Dik (1981) and Watters (1979). 31 argument (whether S or A) that it is cross-referencing.38 Secondly, unlike other forms which have a primary function (with secondary functions possible), it isn’t possible to assign a primary temporal or other semantic function to -sho/shõ, as there is too much variation. In general, verbs which describe a position or posture and are marked with -sho/shõ indicate the manner in which the matrix verb was performed. With other verbs, -sho/shõ can either take a previous or simultaneous temporal reading, but much depends upon the context of the utterance. 4.7 -inã ‘S/A>S/A’ Another switch reference marker that indicates coreferentiality is -inã ‘S/A>S/A’. This form may reference either S or A arguments in the marked clause with either S or A arguments in the matrix clause. It may occur in contexts where events occur simultaneously or where one event occurs slightly before the next one; this temporal aspect however, does not seem to be marked by the form itself. In fact, this seems to be one of the main differences between -inã and -sho/shõ. While -sho/shõ always appears to take a temporal or manner reading, -inã, is found in constructions with the logical relations of result or purpose.39 Regarding its distribution, -inã largely occurs with the verb root i ‘say’, marking direct speech or thought, but it may occur with other verbs as well. Observe the following example, where the S argument in the marked clause is coreferential with the S argument of the matrix clause: 38 It is important to mention briefly that in several other Panoan languages, the cognates for -sho/shõ denote that arguments are coreferential, but specifically, that the matrix clause referent is A ([Amahuaca] Sparing-Chávez 2012; [Kashibo-Kakataibo] Zariquiey 2011; [Matses] Fleck 2003; [Shipibo-Konibo] Valenzuela 2003; [Yaminawa] Eakin 1991). Furthermore, there is a corresponding form (usually [-aʂ] or [-ʂ]) which specifies the matrix clause referent as S (Valenzuela 2003: 885). As already described above, this distinction (or corresponding form) is not found in Marubo. 39 As described in (Dixon 2009), some languages may use the same marker for both result and purpose. In Marubo, it seems that the marker itself does not indicate these logical relations, but rather an analytic reading of the clauses in which it occurs results in a result/purpose interpretation. 32 (29) OR [ka-kĩ…]MARKED [shawe ẽ=mera-nõ]MAIN40 [i-inã]MARKED go:SG-S/A>A.SIM tortoise 1SG=find-INT say-S/A>S/A [patá~patá-kawã-i]MATRIX ITER~gaze-ACROSS:SG/INTR-MCM1 ‘As (he) was going along… “I want to find a tortoise,” saying this, (he) looked back and forth’ ‘he looked back and forth in order to find a tortoise’ The individual in this narrative is a man who goes out hunting. As may be noted, the storyline transitions from third person in the marked clause, to first person in the main clause then back to third person in the last marked and matrix clauses. The second clause records the man’s thoughts “I want to find a tortoise”. In the following clause, the form i ‘say’ is used to mark that the preceding clause is direct speech or thought; this form is inflected with -inã, marking that the referent in this clause will be coreferential with the referent in the following (matrix) clause. This is indeed the case, as it is the man who “says” (thinks to himself) and who “looks around” for the tortoise.41 The form -inã also may cross-reference S arguments in the marked clause with A arguments in the matrix clause, as shown in (30): (30) [kamóshe iná a-vaikĩ]MARKED [nakõti-nĩ torõ-ki]MATRIX… bushmaster domesticate do.TR-S/A>A.PREV clay.pot-LOC be.coiled\CAUS-MCM2 [raká~raká-s-tso Shekioisakará42]MAIN [i-inã]MARKED [vepẽ-ní-ki]MATRIX ITER~be.laying-DUR-IMP Shekioisakara\VOC say-S/A>S/A open-UPWARD-MCM2 ‘After domesticating a bushmaster (snake), (he) coiled (it) inside a clay pot… “keep laying there Shekioisakara” so saying this, (he) opened up (the clay pot)’ (cf. absolutive Shekioisakara; verb root toro) In this short story, a man raises a bushmaster snake, which he names Shekioisakara. When the man prepares to leave for a party, he tells the snake to stay put; this is found preceding the Here, I use the label “main”, since this clause is not “marked” with a switch reference marker, nor is it a “matrix” clause for another clause. 41 For more discussion on the alternative free translation expressing purpose, refer to §4.11. 42 Remarkably, the (compound) name given to this snake seems to be a preserved form of older speech. One language teacher explained that this is composed of the words sheki ‘corn’ + oi ‘grill’ + sakara ‘?’. While the first word is still in use, the second form seems to have a more commonly used counterpart, shoi ‘grill’. Finally, it is unclear if or how the last word should be broken down, however, my language teacher explained that it is equivalent to the modern word keská ‘like’. Thus, this name is based upon the appearance of the snake, the pattern of which is “like grilled corn”. 40 33 speech verb i ‘say’. This speech verb is marked with -inã, indicating that the referent in the following clause is the same. Here, the referent (the man), is an A argument of the matrix clause verb vepẽ ‘open’. Now, consider (31) where -inã cross references an A argument with a following S argument: (31) (teté… yoini nacha-vaiki=se…) [a-nõ õpo hawk animal bite-S/A>A.PREV=DM DEM:ANA-OBL clothes a-katsi-inã]MARKED [mekí=vo]MATRIX make-PROSP-S/A>S/A be.happy=PL ‘hawk…after (it) bit the animal…wanting to make clothes with it, (they) were happy’ This text is based off of the recounting of an episode from the documentary series by BBC, “Human Planet”. The characters in this narrative are a boy, his father, their community and a domesticated hawk. The hawk kills small animals and brings them to the boy and his father. Despite not being introduced earlier, the characters in the marked clause refer to the community, not the boy or his father (this is known from the context of the film and from the plural marker in the matrix clause). The verb a ‘make’ is marked with -inã indicating that the following referent (the community) is the same; here, the matrix clause verb is mekí ‘be.happy’, (cross referencing an A argument with an S argument). Finally, an example of cross referencing A with A is given in (32): (32) ã=a-ská-s-maĩnõ=se [oĩtá roé raká-a 3SG= DEM:ANA-SIML-DUR-DS/A=DM INTJ ax be.laying-NMLZ oĩ-inã=se]MARKED [nini-varã-i]MATRIX see-S/A>S/A=DM pull-INWARD:PL/TR-MCM1 ‘after (he)i did this, (he)j seeing the ax which was laying, snatched (it)!’ (lit. pull towards himself) Here, the argument of the marked clause is a man, who finds the talking ax, which belongs to the jungle trickster Mĩshõ. When the man “sees” the ax, he “snatches” it, as indicated in the matrix 34 clause by the verb nini. Thus, it is the A of the marked clause which is cross referenced with the A argument of the matrix clause. It is important to briefly note the contexts where a result reading is found. For example, in (30), the result of speaking to the snake led the man to “open up” the clay pot. In (31), the translation “wanting to make clothes with it, (they) were happy” can also mean “they were about to make clothes with it, so (they) were happy”. Example (32) also demonstrates this, as it may be understood that the result of “seeing” the ax laying there, led the man to “snatch” it. To summarize, in most constructions where -inã is marked on the clause indicating the “lead-up”, the “natural consequence” (result) is found in the matrix clause (Dixon 2009). 4.8 -a ‘O>S/A/O’ One of the more extraordinary Marubo switch reference forms is -a. Rather than crossreferencing an S or A argument in the marked clause, this form cross-references the O argument of the marked clause with either the S, A, or O argument of the matrix clause. As its gloss denotes, -a ‘O>S/A/O’ is restricted to being marked on transitive verbs. This consistent with all examples that I have gathered to date, as there are none which are marked on an intransitive verb. In terms of other functions, it tends to also indicate that the event in the marked clause occurs immediately before the event in the matrix clause (in English, this is best translated as “when”); however, it can also mark that the events occurred simultaneously. For this reason, this secondary function is not included in the gloss itself. Consider (33): 35 (33) (ẽ=papa… yama-mtá). [Eirunepé-namã atõ 1SG.POSS=father be.nothing-PST3 Eirunepé-LOC 3PL\ERG ewe-vaĩ-a,]MARKED [a-no-sho43 vopi-mtá]MATRIX get-OUTWARD:PL/TR-O>S/A/O DEM:ANA-LOC-PA die-PST3 ‘my dad…died (became nothing). When they took (him) to Eirunepé, from there, (he) died.’ (cf. absolutive ato) In this personal description, the narrator describes his father’s death that occurred in the distant city of Eirunepé. In the marked clause, an unnamed party, “they”, transport the father, who is the O argument. This argument is coreferential with the S argument of the matrix clause, the father, who dies. As already mentioned above, the temporal value can either describe a previous or simultaneous event. In this case, the event of “taking” in the marked clause occurs slightly before that of “dying” in the matrix clause. While (33) cross-references the O with the S argument, (34) demonstrates that the O is coreferential with the A: (34) (aĩvo…vake…) [ve-kĩ] [atõ oĩ-a]MARKED [naa woman…child come:PL-S/A>A.SIM 3PL\ERG see-O>S/A/O DEM:PROX txona a-vaikĩ]MARKED [masõ-ni~ní-i]MATRIX woolly.monkey get-S/A>A.PREV be.gathered\CAUS-ITER~UPWARD-MCM1 ‘woman…child…coming, when they saw (them)i, (they)i were gathering up the monkeys, after (they)i had gotten them’ (cf. absolutive ato; verb root maso) In this narrative, a woman takes her child and leaves her husband behind, searching for another group of hunters. As she and her child “come” towards the other group, they “see” this group, the O argument of the first marked clause. When examining the matrix clause, it is apparent that “they” (this group) is also the A argument, since “they” are the ones gathering up the monkeys. Temporally, the events in both clauses occur simultaneously. Finally, -a also may coreference O arguments with other O arguments, as illustrated in (35): Note that this form, while identical to the switch reference form -sho ‘S/A>S/A’, is syntactically distinct, marking intraclausal participant agreement. Certainly, more investigation is needed, but the form and function of -sho ‘PA’ are similar enough to that of Shipibo-Konibo (and a few other Panoan languages) to preliminarily gloss it as such (in this case, it is oriented towards the S argument and functions as an ablative). 43 36 (35) [yawíchi atxi-vaiki=se] [ẽ=vi-varã-a]MARKED armadillo catch-S/A>A.PREV=DM 1SG=get-INWARD:PL/TR-O>S/A/O [nawã ea viãkã-vai]MATRIX white.man\ERG 1SG get\MAL-PST1 ‘when I brought the armadillo, after catching (it), the white man took (it) from me (to my detriment)’ (cf. verb root vi44; absolutive nawa) As can be observed from the marked clause, the verb is transitive, with ẽ= ‘1SG’ as the A argument and yawíchi ‘armadillo’ as the O argument. In the matrix clause, the same transitive verb is used, but here, it is marked with the malefactive applicative suffix, adding one argument. So, in the matrix clause, the A is nawã ‘white.man\ERG’, the (patient) O is “armadillo” (covert) and the (beneficiary) O is ea ‘1SG’.45 Obviously, the “white man” does not take ea ‘1SG’, but rather the armadillo, which although covert, is referentially the same as the O in the marked clause. This object-to-object coreferentiality can be further illustrated in (36), an example with only one object in the matrix clause: (36) [Tamã ãwẽ shovo ene-a=se,]MARKED [nakáshe-nẽ pi-chĩa]MATRIX Tama\ERG 3SG.POSS house leave-O>S/A/O=DM termite-ERG eat-PST2 ‘When Tama left his house, the termites ate (it)’ (cf. absolutive Tama) In the marked clause, shovo ‘house’ is the O, while in the matrix clause, the O is also referentially “house”, the participant which undergoes eating by the “termites”. Considering the examples above, it might be tempting to analyze -a ‘O>S/A/O’ as simply switching to a different referent, rather than cross-referencing O with S, A or O. However, when attempting to elicit examples, such as (37) which introduce a completely new participant, these were rejected by native speakers: The malefactive applicative suffix is –(Ṽ)kã; that is, it nasalizes the verb root’s final vowel and adds kã. However, likely due to prosodic factors, when the malefactive follows monosyllabic verb roots, the vowel a is epenthesized, and this is what gets nasalized, not the monosyllabic verb’s root vowel. 45 In constructions which contain two objects, Marubo marks both of them in the same way (with null-marking). The patient and theme are only distinguished by context, as word order is also flexible. 44 37 (37) *[Tamã richki-ti ãwẽ õtxi-a,]MARKED Tama\ERG hit-INSTR.NMLZ 3SG throw-O>S/A/O [ã=aĩ txakar ik-í]MATRIX 3SG.POSS=wife txakar do.INTR-HOD ‘Tama, when he threw the machete, his wife yelled’ (cf. absolutive Tama) (lit. made the noise txakar) This example above became grammatically acceptable only when the different-subject marker was used, as in (38), or, when the object of the marked clause was included in the matrix clause, as in (39). (38) [Tamã richki-ti ãwẽ õtxi-maĩnõ,]MARKED Tama\ERG hit-NOM 3SG throw-DS/A [ã=aĩ txakar ik-í]MATRIX 3SG.POSS=wife txakar do.INTR-HOD ‘Tama, when he threw his machete, his wife yelled’ (cf. absolutive Tama) (lit. made the noise txakar) Here, the example is identical to that of (37) except that instead of being marked with -a ‘O>S/A/O’, the verb õtxi ‘throw’ is marked with the -maĩnõ suffix which indicates a different subject (see §4.9). Conversely, in (39), -a ‘O>S/A/O’ is used, but it is acceptable only when the matrix clause contains the O argument of the marked clause, either referentially as the S, the A or the O argument. In this case, the O of the marked clause, “machete” is the S argument of the matrix clause: (39) [Tamã richki-ti ãwẽ õtxi-a,]MARKED [wakã Tama\ERG hit-INSTR.NMLZ 3SG throw-O>S/A/O river\LOC paké-kaĩ-vai]MATRIX fall\INTR-OUTWARD:SG/INTR-PST1 ‘Tama, when he threw the machete, (it) fell into the river’ (cf. absolutive Tama; verb root pake) 38 Therefore, examining all the evidence, it seems best to analyze the marker -a, as one which coreferences O with S, A or O arguments;46 this is consistent with two other Panoan languages, Capanawa and Shipibo-Konibo, which use cognate markers that function in the same way.47 4.9 -maĩnõ ‘DS/A’ Up to this point, we have seen switch reference forms which demonstrate coreferentiality between a combination of different arguments. In contrast, the form -maĩnõ is used when either the S or A of the marked clause is not coreferential with the S or A of the matrix clause.48 It is essentially the opposite of -sho/shõ ‘S/A>S/A’. Temporally, -maĩnõ can be used to describe events which happen simultaneously, or one event before the other (marked clause before matrix). Since this value is also determined by context, it is not included in the gloss. Consider (40), where the S argument of the marked clause is not the same as the S argument of the matrix clause: (40) [ẽ=vene nokó-karã-maĩnõ]MARKED [ea wai i-vai]MATRIX 1SG.POSS=husband reach\DETR-INWARD:SG/INTR-DS/A 1SG wai do.INTR-PST1 ‘When my husband arrived, I cried’ (cf. verb root noko) (lit. did the noise wai) This narrative describes a husband’s return home to his wife, after a long trip. In the marked clause, it is the husband, the S argument, who is arriving. Note that the verb is a detransitivized form (an antipassive); had it been transitive, the wife would have been “reaching” the husband, (the O argument), which is certainly not the case in this account. In the matrix clause, it is no longer the husband, who is the S argument, but rather the wife, who “cries” when he arrives. 46 Interestingly, the idea that O arguments may be cross-referenced violates one of the principles put forth in Finer (1985: 53), which states: “switch-reference involves subjects only”; additionally, object coreferentiality falls outside of “canonical” switch reference as defined by Haiman & Munro (1983). 47 For example, -a in Shipibo-Konibo “indicates that the object of the dependent clause which codes a previous event is coreferential with the subject (S/A) argument of its matrix clause” (Valenzuela 2003: 424) See also (Loos 1999) for Capanawa. Note that this marker in Shipibo-Konibo and Capanawa coreference O with S/A, but not with O, as in Marubo. 48 Note that this means the marked S is different from the matrix S or the A, or the marked A is different from the matrix S or A, that is S/A≠S/A. I avoid the use of DS (“different subject”) since “subject” is not a very useful notion in Marubo. 39 Temporally, the event in the marked clause occurs just slightly before the event in the matrix clause. Now, observe the following example, where the S is different than the A argument: (41) K: mĩ=vicha, ã=epã rayóse=nã 2SG.POSS=younger.brother 3SG.POSS=paternal.uncle\POSS father.in.law=TOP a-no ni-á=ra DEM:ANA-LOC be.standing-HOD=Q ‘your younger brother, his uncle’s father-in-law, is he there?’ (cf. absolutive epa) W: manãri ka-vai=vere… upriver go:SG-PST1=VER ‘(he really) went upriver’49 K: Ase, [ã=o-maĩnõ,]MARKED [ea tana-ma-shõ-tso]MATRIX… INTJ 3SG=come:SG-DS/A 1SG understand-CAUS-BEN-IMP e=rí iwi nisa-ma-katsi ik-í rama=ma ẽ=tapo 1SG=DIR tree saw-CAUS-PROS AUX:INTR-HOD now=NEG 2SG.POSS=hut a-chĩa tsaó-a… make-PST2 be.sitting-HOD ‘Ok, when he comes, (you) tell (him) (about the boards) for me (lit. “cause him to understand”)…I myself want to make someone saw trees. My house which I made ages ago sits’. This short-wave radio dialogue concerns “K”, talking with “W” about the possibility of one of “W’s” relatives sawing boards for “K”. In the first line, “K” asks “W” if this relative is there in the community. “W” informs “K” (in the second line) that this relative went upriver. In the third line, “K” responds, telling “W” to inform this relative on his behalf (about the sawing). In this line, the marked clause contains an intransitive verb, with “he” still referencing the relative. This intransitive verb is marked with the -maĩnõ suffix, indicating that the following S or A will not be this relative. As we can see, the matrix clause contains the transitive verb tana ‘understand’, The marker -vere behaves very similar to “veridical markers” described in Payne (1997: 224), which “indicate the degree of commitment the speaker makes as to the truth of the assertion”. 49 40 with “you” (referring to “W”) as the A argument. Coincidentally, one of the O arguments (there are actually three), refers to the relative, as he is the causee.50 The -maĩnõ suffix may also mark that the A argument of the marked clause is different than the following matrix clause arguments; here, the A is different than the S argument: (42) (wanĩ…) [ã=vakẽ shoka-maĩnõ]MARKED [aa=ro peshé peach.palm 3SG.POSS=child\ERG peel-DS/A 3SG=FOC:PAR shelter teké wetsã, ronó-iná-sho]MARKED [ã=osha-a]MATRIX section another\LOC hang-UPWARD-S/A>S/A 3SG=sleep-MCM3 ‘peach palm…while her child was peeling (them), (she) was hanging in her hammock at another section of the shelter, sleeping’ (cf. absolutive vake; wetsa) In this example, which contains a partial repetition of (28), the A argument of the marked clause is “child”. After -maĩnõ is applied to the verb shoka ‘peel’, we can observe that in the following clauses, both verbs are intransitive, taking the S argument “she”; this no longer refers to the child, but to her mother, who “sleeps”. Notice that in this case, the events of “peeling” and “sleeping” are simultaneous; this demonstrates that the temporal value of this morpheme is somewhat flexible, since in previous examples, we saw that it indicated that events in the marked clause took place slightly before events in the matrix clause. As already observed, -maĩnõ may also mark that A arguments are not coreferential with following A arguments; this is illustrated in (43): (43) (awá...) ã=ivo nacha-ó-i, [a-maĩnõ]MARKED tapir 3SG.POSS=owner bite-REP-MCM1 do.TR-DS/A [awá=ro yama-ma-chĩ=vo]MATRIX tapir=FOC:PAR be.nothing-CAUS-PST2=PL ‘tapir…bit its owner. When (it) did this, they killed the tapir (lit. ‘make nothing’)’ 50 It is not uncommon for only one of the arguments to appear overtly in multiple object constructions. In this case, the transitive verb tana ‘understand’ receives the causative and benefactive markers, increasing its valency by two. The beneficiary ea ‘1SG’ is the only overt argument; the causee “him” and the theme “sawing boards” are left to be inferred by context. 41 This is a non-fictitious account of a pet tapir which bites its owner, and then is killed because of this. In the first clause, the tapir (covert, but inferred from previous context), bites the owner. The second clause, marked with -maĩnõ, is a tail-head linkage construction,51 where the verb a ‘do’ refers back to the verb nacha ‘bite’. Thus, the A argument of a ‘do’ still refers to the tapir. In the following clause, however, the referents switch, and it is the owner (and others from his community) that kill the tapir. It is important to note that, unlike several other Panoan languages, Marubo does not have a marker which encodes that S or A arguments in the marked clause are coreferential with O arguments in the matrix clause;52 when such arguments do occur in both the marked and matrix clause, -maĩnõ is used, even though it does not specifically encode such a function. In (43), for example, the A in the marked clause (the tapir) happens to occur as an O in the matrix clause. However, as already observed in examples (40) and (41), where -maĩnõ is marked on intransitive verbs, or (42), where the matrix clause verb is intransitive, marking A as coreferential with O is not the function of this morpheme. The same is the case when an S argument in the marked clause occurs as an O in the matrix clause: (44) [Atalaia-namã ã=ka-maĩnõ]MARKED [yora wetsã teka-vai]MATRIX Atalaia-LOC 3SG=go:SG-DS/A person another\ERG shoot.with.gun-PST1 ‘when hei went to Atalaia, someone shot (himi)’ (cf. absolutive wetsa) In the marked clause, the S argument is ã= ‘3SG’, which also occurs in the matrix clause as “he”, a covert O argument. But, when observing how -maĩnõ functions throughout the rest of the corpus, it is clear that it does not actually encode S with O coreferentiality. Rather, here, it is 51 See Thompson, Longacre & Hwang (2007) for more information about tail-head linkage constructions. For example, Capanawa (Loos 1999), Kakataibo (Zariquiey 2011), and Matses (Fleck 2003) all possess switch reference markers which have this function. 52 42 merely coincidental and -maĩnõ is actually encoding that the S in the marked clause is not coreferential with the A in the matrix clause.53 4.10 -txinĩ ‘DS/A’ Another suffix which marks non-coreferentiality is -txinĩ. This marker is unusual in that it occurs after the marking of tense, most often following the recent past tense marker -vai ‘PST1’. Initially, I did not include this form in the list of switch reference markers because of the limited number of examples.54 However, after discovering an almost identical form functioning as a switch reference marker in the Panoan language of Katukina, I began to recall its use in conversation, and decided to investigate further through elicitation.55 After careful testing, I suggest that this form should tentatively be considered a switch reference marker. Recall that many Marubo morphemes are identical in form, but often have significantly different functions. This is the case with -txinĩ, as one form functions as a noun denoting a season or period of time, while the other marks different referents. Consider (45), where it is found as a noun in a verbless clause construction: (45) aĩvo kamã-ne pi-á=ro… aa=ro wanĩ txinĩ woman jaguar-ERG eat-MCM3=FOC:PAR DEM:ANA=FOC:PAR peach.palm season ‘the (story) of the jaguar eating the woman is…this was during peach palm season’ This introduction to a narrative is typical and contains a statement of the title and information about the setting. The first clause is incomplete, but would typically contain a proximal demonstrative (for example, see Appendix 3, line 1). The narrator starts a new thought, using a 53 Interestingly, a similar form from Kashibo-Kakataibo, -mainun is described in Zariquiey (2011: 589); this marks “simultaneous events that do not share either subjects or objects”. Also Lord (2016: 91) describes the Yora suffix -aino as marking “that the subject of that clause is not coreferential with the subject of the governing clause…the event of the governing clause can be interpreted as taking place sometime during or immediately after the event of the marked verb”. 54 My corpus only contains two examples from the same text, a radio conversation. 55 In Kennell (1998: 80), the form -va’i txinĩ is listed as marking that “os sujeitos do verbo subordinado e do verbo principal são diferentes, e o verbo principal pode ser transitivo ou intransitivo, tal como o verbo subordinado (the subjects of the subordinate verb and of the main verb are different, and the main verb may be transitive or intransitive, just as with the subordinate verb)”. 43 verbless clause construction with two juxtaposed elements, one containing the anaphoric demonstrative aa and the other, a noun phrase. Here the head of the noun phrase is txinĩ ‘season’ with wanĩ ‘peach palm’ acting as another nominal modifier, describing the type of season. This is commonly used to describe other times of year such as rainy and dry seasons, as found in (46): (46) Q: awe=tiã ka-chĩ=ra, Tama=nã? what=TEMPORAL go:SG-PST2=Q Tama=TOP ‘when did (he) leave, you know, Tama? (lit. ‘what time did he leave’) A: oi txinĩ ka-chĩa56 rain season go:SG-PST2 ‘(he) left in rainy season’ In the response to the question above, txinĩ is used as a noun, describing the time when Tama left. It is modified by oi ‘rain’ denoting the kind of season. In contrast to this, while the switch reference marker -txinĩ also functions temporally, it must always mark a change in S or A arguments from the previous clause. Consider this example from a radio dialogue between a mother and daughter: (47) [otxĩ [mia ẽ=oĩ-i ka-katsai older.brother\ERG 2SG 1SG=see-S/A>S.SIM go:SG-FUT a-no=ka=sho=se ea maná-tso] ea a-vai-txinĩ]MARKED DEM:ANA-LOC=FOC:EXH=PA=FOC:ASS 1SG wait-IMP 1SG tell-PST1-DS/A [otxi-nĩ ea kaa=ma i-vai]MATRIX older.brother-COM 1SG go:SG=NEG AUX:INTR-PST1 ‘when older brother told me, “I’m going to see you, wait for me right there”, I didn’t go with older brother’ At this point in the conversation, the mother recounts what her brother had told her. The marked clause is complex as it contains more than just the verb a ‘tell’ and its arguments; it also contains the direct speech of the brother, presented as two independent clauses.57 However, for the Note that while normally -chĩa ‘PST2’ is in verb final slot, it may take the shortened form -chĩ when preceding other verb final morphemes, such as =ra ‘Q’. 57 Much more research is needed in the area of direct speech and whether or not a construction like this constitutes a complement clause or simply a clause which is juxtaposed with the speech report verb. Dixon (2010: 307), when describing direct speech states, “in some languages, the direct speech functions as O argument for a transitive verb 56 44 purposes of illustration, we will only focus on the arguments of a ‘tell’, which is marked by -txinĩ ‘DS/A’. As may be observed, the A argument is otxi ‘brother’ and the O argument is ea ‘1SG’. In the matrix clause, however, we find the intransitive verb ka ‘go:SG’ which takes the S argument ea ‘1SG’. This no longer refers to the brother, but to the mother. Interestingly, much like -maĩnõ, the temporal aspect of this marker is not specified, but is understood from context, sometimes as a previous event and sometimes as a simultaneous event; it is best translated as “when”. Of even more interest is the verbal slot in which this marker occurs. In all other cases, switch reference marked clauses are not marked for tense and thus the switch reference marker either occurs directly after the verb root or after suffixes marking valency-change, aspect, or associated motion. In this case, however, the form -txinĩ occurs after the recent past tense marker -vai. Though -txinĩ is most often found following the speech report markers -a ‘tell’ and -i ‘say’, it may occur on other verbs as well, as is illustrated in (48): (48) [Maronal=namã ẽ=ka-tã-chná-txinĩ]MARKED [yora=rasĩ-ni Maronal=LOC 1SG=go:SG-GO.DO.THERE-PST2-DS/A person=PL-ERG raké-sho-chĩ=vo]MATRIX be.afraid-BEN-PST2=PL ‘when I went out to the Maronal (village), the people feared me’ In this elicited example, the marked clause contains the intransitive verb ka ‘go:SG’, which is marked by -txinĩ. The S argument is the pronominal clitic ẽ= ‘1SG’. Turning to the matrix clause, one can see its S argument, yora ‘person’, is noncoreferential with ẽ= ‘1SG’. As already noted, the temporal value is not specified, but in this case the “going” seems to occur just before “being afraid”. As described above, initially, I did not consider this to be a switch reference marker, of speaking…in others, the direct speech is in apposition to the framing verb and does not function as a syntactic argument for it”. In Marubo, the latter is much more likely, since the speech marker which follows this clause is a transitive verb a ‘tell’. Since this verb only permits two arguments, both of which already occur overtly, the speech clause cannot act as another O argument to the verb a; this would require the addition of a valency-increasing suffix. 45 since it seemed to simply describe a period of time or season. However, when I attempted to elicit clauses containing -txinĩ with coreferential arguments, speakers rejected them as ungrammatical. Observe (49), for example: (49) *[Maronal=namã ẽ=ka-tã-chná-txinĩ]MARKED [yora=rasĩ Maronal=LOC 1SG=go:SG-GO.DO.THERE-PST2-DS/A person=PL ẽ=raké-sho-chĩa]MATRIX 1SG=be.afraid-BEN-PST2 ‘when I went out to the Maronal (village), I feared the people’ This example is identical to (48), except here, the A and O arguments of the matrix clause are switched. Since the S of the marked clause is coreferential with the A of the matrix clause, -txinĩ may not be used. Considering the examples above, -txinĩ ‘DS/A’ has a significantly different function than that of txinĩ ‘season’, marking a change in reference from either S or A arguments. Although -txinĩ is semantically similar to the other non-coreferential form, -maĩnõ ‘DS/A’, structurally, -txinĩ is unique (among all of the switch reference forms) in that it occurs following tense markers. 4.11 Panoan switch reference cognates As described in the sections above, Marubo switch reference markers are complex and carry a high functional load. Interestingly, the markers found in §4.2-4.10 are similar in form (and some in function) to those of other Panoan languages.58 In this section, I describe two Marubo markers which are quite similar in form but which do not function as switch reference markers. This 58 As noted in preceding sections, four of these morphemes (marked phonetically), -i, -kĩ, -(a)ʂ, and -ʂõ, form a part of the participant agreement system in many Panoan languages. The markers -i and -(a)ʂ are oriented towards matrix clause S arguments, while -kĩ and -ʂõ are oriented toward matrix clause A arguments (Valenzuela 2003). Marubo has lost the distinction between -(a)ʂ and -ʂõ at the interclausal level, with -ʂo/ʂõ taking over the function of -(a)ʂ, but more research is needed to determine if there is still any distinction at the intraclausal level. Moreover, it is likely that -vai is historically derived by the fusion of the past tense marker (-vai) plus (-i), with the subsequent loss of -i. This would explain its orientation towards a matrix S argument. Similarly, -vaiki/vaikĩ likely arose by the fusion of the past tense marker (-vai) plus (-kĩ), which explains its orientation towards a matrix A argument (special thanks to Pilar Valenzuela for pointing this out). 46 description is included for comparative purposes, but more importantly, to demonstrate that they indeed can be ruled out as switch reference markers in Marubo. Several other Panoan languages possess between one and three markers which form a part of the switch reference system: -non or -nun, -nush or -nux and -noxon or -nuxun. The former signals that referents are nonidentical while the latter two (with differences based on tracking S and A arguments, respectively) signal identical referents across clauses.59 All three markers also express that the event in the marked clause takes place after the event in the main clause; along with this, it is often used to express purpose. In the following table, observe the forms and their various functions in nine Panoan languages: 59 Note that in Matses and Matis, -nun marks same, rather than different referents. 47 Table 4 Panoan switch reference markers indicating subsequent events or purpose Language Form Reference Temporal/logical function function Amahuaca (Sparing-Chávez 2012) Capanawa (Loos 1999) -non S/A≠S/A Subsequent/Purpose -non S/A≠S/A Subsequent -noʂon S/A=S/A Subsequent S/A≠S/A Subsequent S/A=S Subsequent/(Purpose) -nuxun S/A=A Subsequent/(Purpose) -nõ S/A≠S/A Simultaneous?60 -nun S/A=A Subsequent/Purpose -nuʂ S/A=S Subsequent/Purpose -nun S/A=(S)/A Subsequent/Purpose -nush S/A=S Subsequent/Purpose -non S/A≠S/A Subsequent/(Purpose) -nox S/A=S Subsequent/(Purpose) -noxon S/A=A Subsequent/(Purpose) -nun Kashibo-Kakataibo (Zariquiey 2011) -nux Katukina (Kennell 1998) Matis (Ferreira 2005) Matses (Fleck 2003) Shipibo-Konibo (Valenzuela 2003) Yaminawa (Eakin 1991) -nõ S/A≠S/A Subsequent/Purpose Yora (Lord 2016) -nõ S/A≠S/A Subsequent/Purpose Although this is not an exhaustive list, the similarity between many of the forms and their functions is apparent. In Marubo, two of these cognates exist, but function quite differently. For example, -nõ occurs in main clauses (which is not characteristic of Marubo switch reference markers) and is most often used to mark intent: (50) [Ase, kamã-ne pi-á mato yoã-sho-nõ]MAIN [aa INTJ jaguar-POSS eat-NMLZ 2PL tell-BEN-INT DEM:ANA nĩkã-tso-ma vake=rasĩ-ní]MAIN listen-IMP-PL child=PL-VOC ‘Ok, I want to tell you (the story of) the jaguar’s eating. Listen to it, children’. 60 While the two examples provided in (Kennell 1998) indicate a simultaneous reading, its temporal function is not explicitly described. 48 (51) [“Mãa ase, mia veso-sho-nõ]MAIN [a-no=se INTJ INTJ 2SG watch-BEN-INT DEM:ANA-LOC=FOC:ASS tsaõ-Ø”]MAIN be.sitting\CAUS-IMP ‘Ok, yeah, I want to watch (it) for you. Put it down there’ (cf. verb root tsao) (lit. “cause it to sit”) In (50), the introduction to a tale about jaguars, there are two main clauses. In the first, the verb yoã ‘tell’ is marked with the intent marker -nõ, communicating that the narrator wants or intends to tell the story to the children. In (51), this is also the case; in the first main clause, -nõ is marked on the verb veso ‘watch’ and expresses the speaker’s assurance to the hearer that he wishes to comply with the hearer’s request for him to watch the item for him. As described to me by my language teachers, this marker is typically used when one actually will follow through with one’s intention, as opposed to other markers which express a weaker desire, or a suggestion that it would be helpful or good to perform some action. Furthermore, this marker only seems to denote the intentions of the first person, either singular or plural, that is, it cannot be used to talk about a second or third person’s intent. Interestingly, in narratives which are told in the third person, clauses which contain -nõ often are used in direct speech or thought before transitioning back to the third person; this was described above in (29), repeated here for convenience: (52) [ka-kĩ…]MARKED [shawe ẽ=mera-nõ]MAIN [i-inã]MARKED go:SG-S/A>A.SIM tortoise 1SG=find-INT say-S/A>S/A [patá~patá-kawã-i]MATRIX ITER~gaze-ACROSS:SG/INTR-MCM1 ‘As (he) was going along, “I want to find a tortoise,” so saying this, (he) looked back and forth’ OR ‘He looked back and forth in order to find a tortoise’ As noted, the narrator transitions from third person in the marked clause to first person in the main clause and then back to third person in the matrix clause. The second clause, whose verb is marked by -nõ, represents the speaker’s thoughts (his intent to find a tortoise). Since the man’s thoughts are then followed by the speech marker i, which is marked with -inã ‘S/A>S/A’, we 49 know that the argument in the following clause also refers to the man. Structurally, -nõ is marked on a verb in a main clause, which is reflected in the first free translation. On its own, -nõ does not encode purpose or a subsequent event. However, when interpreted together with the following speech verb and matrix clause, a purposive meaning is understood. In other words, purpose is not expressed morphologically through this marker, but analytically. 61 Initially, in certain constructions, I had analyzed -nõ as a switch reference marker. For example, when one verb marked with -nõ was directly followed by another verb, there was virtually no pause between the two suggesting clause dependency; this is exemplified in (53) and (54):62 (53) Tama ẽ=tana-ma-vai Tamá, yawa Tama 1SG=understand-CAUS-PST1 Tama\VOC white.lipped.peccary a-nõ kaõ-Ø get-INT come.by:PL-IMP ‘I told Tama, “Tama, come let us get white-lipped peccary”’ OR ‘We want to get white-lipped peccary. Let’s come on.’ (cf. absolutive Tama) (54) (Mĩshõ…) mia tokẽ a-shõ-nõ Mĩshõ 2SG armadillo get-BEN-INT kawã-Ø Shoĩ Wirõmé come.by:SG-IMP Shoĩ Wirõme\VOC ‘Mĩshõ…“I want to get for you an armadillo. Come on, Shoĩ Wirõme!”’ OR ‘“Come let us get an armadillo for you, Shoĩ Wirõme”’ (cf. absolutive Shoĩ Wirõme) In both examples, prosodically, there is no pause between the two verbs. This initially led me to believe that the first clause was dependent upon the second, and that -nõ was marking that the S or A argument of the marked clause was coreferential with the S argument of the matrix clause; furthermore, I interpreted it to also be expressing purpose. Observe that this hypothesis seems to 61 The analytic expression of purpose by using speech report constructions is not exceptional. For example, Overall (2009: 175) reports that in Aguaruna (Chicham), “a speech report accompanied by a subordinate form of the verb tu ‘say’ is a widely used grammatical and stylistic device…and is obligatory in different-subject purpose clauses and complementation strategies”. 62 Note that in these examples, I do not demarcate clauses with brackets because I am still not completely certain whether the verbs in question belong to two separate clauses or could be considered a serial verb construction. 50 be valid in (53), where the referents of the verbs a ‘get’ and kaõ ‘come.by:PL’ are both first person singular. However, in the nearly identical construction in (54), the A argument of the verb a ‘get’ is Mĩshõ, whereas the S of the verb kawã ‘pass.by:SG’ is the son Shoĩ Wirõme (the second free translation was accepted but not preferred). Therefore, since -nõ does not demonstrate a consistent pattern of marking coreferentiality or non-coreferentiality and since it clearly marks intent in main clauses, it is unnecessary to analyze it as having a completely different meaning (as a switch reference marker) only when it occurs in a limited set of constructions. Nevertheless, it it still unclear whether these types of constructions are simply two separate main clauses spoken without a pause or if they form a type of serial verb construction; since this is beyond the scope of this thesis, it will remain for future analysis. Another marker, -noshõ/nosho, is also similar to switch reference forms of other Panoan languages (initially, I analyzed it as an SR marker). Admittedly, compared to other suffixes, this one is the least understood, in part because it occurs the fewest times in my database. In spite of this, several generalizations may be made about the marker -noshõ/nosho: it often occurs in direct speech, it does not consistently mark coreferentiality or non-coreferentiality, but rather, modality expressing ability or prediction, and when juxtaposed with other clauses, it sometimes may indicate purpose. Consider (55): (55) [verone=vo=rasĩ… [weta-ti aya-rá] ea atõ=a-maĩnõ,]MARKED male.youth=PL=PL63 fish-INSTR.NMLZ exist-Q 1SG 3PL=tell-DS/A [shavá=ma oĩ-noshõ] [ato ẽ=ak-a]MATRIX today=NEG see-ABILITY 3PL 1SG=tell-HOD ‘when the teenagers asked me “is there fishing hooks,” I told them “you can/may see (them) tomorrow”’ 63 More investigation is needed in understanding the semantic differences between the clitics =rasĩ and =vo, which both indicate plurality and may both be used on the same host. Usually =vo, when used on nouns, identifies them as belonging to a group while =rasĩ simply specifies that there is more than one (but see (90) for an exception). A similar enclitic =bo in Matses denotes a “homogenous category” (Fleck 2003: 272). 51 In this dialogue, the speaker explains that when a group of youth comes asking for fishing hooks, he tells them that they “may see” the next day. The verb oĩ that is marked with -noshõ occurs in a clause containing the speaker’s direct speech and is juxtaposed with the matrix clause “I told them”. Here, the meaning expressed by -noshõ is the ability or permission that is given by the speaker to the group of youth. Now, observe the following example, where -noshõ marks prediction: (56) [ẽ=gol a-katsai] [oĩ-noshõ] 1SG=goal make-FUT see-PREDICTION ‘I will make a goal. (You) shall see.’ OR ‘I will make a goal (so) that you shall see’ Here, the A in the first clause is the clitic ẽ= ‘I’ and the A of the following clause with the verb marked by -noshõ is “you”. Here, -noshõ, rather than marking ability or permission, indicates prediction, that the addressee “shall see”. Additionally, a purpose reading may be understood from the juxtaposition of the clauses (“so that you shall see”). In contrast to (56), where the referents are different in both clauses, the following example demonstrates that -noshõ may also occur in clauses which have identical referents: (57) (atõ pani txiwa-sho-a…) [yoá kãtxi-ke-ta-kĩ]MATRIX 3PL hammock connect-BEN-MCM3 clay.pot gather-COMPL-GO.DO.THERE-MCM2 [tarok ik-í nõ=nĩkã-noshõ] [i-inã]MARKED tarok do.INTR-NMLZ 1PL=hear-ABILITY say-S/A>S/A ‘they tied up their hammocks for themselves…(they) went and gathered the clay pots (in a circle), saying “We will be able to hear the clattering sound” OR ‘they went and gathered the clay pots so that they would be able to hear the clattering sound’ (lit. make the noise tarok) In this tale, two boys become afraid that an older man (the poisoner) will poison them during the night while they sleep. Because of this, they gather clay pots around their hammocks to alert in case the poisoner comes. Here, the referents of the clause in which -noshõ occurs are the same as all the other clauses. Although the referent of the direct speech clause is first person plural, 52 referentially, it is the same as the A of the matrix clause and the same as the S of the marked clause, which are both third person plural. Considering all three examples, -noshõ does not behave like a switch reference marker, as there is no regular pattern of coreferentiality or non-coreferentiality. Futhermore, it occurs in main clauses; this is markedly different from any of the other switch reference markers, which all occur in dependent clauses. Finally, -noshõ seems to fall into the category of modality, marking ability, permission or prediction. 53 Chapter 5: Types of switch reference constructions In the previous sections, I identified the Marubo switch reference forms and described their various functions. As was noted, the main function of these markers seems to be to track reference as well as to mark some temporal and logical relations between clauses. In these next sections, I focus on various aspects of switch reference constructions. For example, in §5.1, I describe how marked clauses may follow matrix clauses, and what the function of this nondefault order is. In §5.2, I discuss the scope and target of switch reference markers in clause chaining. §5.3 contains a description of embedded non-coreferential clauses, without an explicit marker. Finally, §5.4 discusses the exceptions to the switch reference system as well as areas for further research. 5.1 Switch reference in non-default ordered clauses Up until this point, all examples of switch reference constructions, except (24) and (57), have followed a predictable order. The marked clause, (containing the switch reference morpheme) precedes the matrix clause. This is consistent with the generalization made by Haiman & Munro (1983), where they note the relationship between the position of the switch reference marker and the order of the clauses: if the SR form is a prefix, the marked clause follows the matrix clause and if it is a suffix, the marked clause precedes the matrix clause. However, they also observe that in some languages, clauses may be freely ordered, while in others, the meaning changes depending on the ordering; in Quechua for example, marked clauses which indicate a temporal relationship precede the matrix clause while marked clauses of purpose follow the matrix clause. In Marubo, clauses may take a non-default order, where the matrix clause occurs first and is then followed by the marked clause. This may be observed with each of the switch reference markers except for -a ‘O>S/A/O’. 54 Consider (58) where the second marked clause containing -vai ‘S/A>S.PREV’ occurs after the matrix clause: (58) txaĩ aa=tishõ=ka cross.cousin\VOC\EMOTIVE64 DEM.ANA=DISTANCE=FOC:EXH õtxĩkã-ai-kĩ ik-í ik-í [a-ská-vai=se]MARKED1 throw\MAL-HOD-EMOTIVE say-MCM1 say-HOD DEM:ANA-SIML-S/A>S.PREV=DM [ka-ai-tĩ,]MATRIX [roé vi-tima-vaí]MARKED2 go:SG-HOD-ANTIMIR ax get-FRUST-S/A>S.PREV ‘“cross cousin, you threw it that far (to my detriment)?”, he said, they say. After it happened like this, (he) leaves, you know, after unsuccessfully getting the ax’ (cf. absolutive txai; verb root õtxi) Here, the jungle trickster Mĩshõ laments that his magical ax has been thrown away. His direct speech is followed by two speech report markers ik ‘say’ (the second indicating that this is what “they say”), which conclude the sentence. The next clause begins with the anaphoric demonstrative used to mark tail-head linkage; this refers back to the first speech marker ik ‘say’.65 The referent in the marked clause (Mĩshõ) is coreferential with the following matrix clause referent which describes his leaving. Following this is another marked clause, containing -vai ‘S/A>S.PREV’, indicating that the A in this clause (Mĩshõ) is coreferential with the S of the matrix clause; this also marks the event of “unsuccessfully getting his ax” as occurring previous to his “leaving”. After verifying with speakers, I was assured that the semantics are not affected by this change in clause order. However, the displaced clause does seem to provide clarificational information, as without it, the hearer might guess but not be certain that Mĩshõ was unsuccessful in retrieving his ax. 64 Note that this noun is modified by two suprasegmentals. The absolutive form is txai, while the vocative form is txaí (however, only the nasal symbol is represented to avoid stacking diacritics). While the vocative is indicated by a movement in stress to the final syllable, this form also contains an “emotive nasal” which is often added clause finally (or as a full morpheme -kĩ) to indicate the speaker’s emotions, which are generally negative and correspond with anger, despair, fear or laziness. For the description of a similar phenomenon in Kakataibo, see (Zariquiey 2018). 65 While the phenomenon of two adjacent speech markers requires more research, it seems clear that the narrator is not making a switch in reference from Mĩshõ’s speech to that of another party (“they”); consequently, the tail-head linkage construction refers back to Mĩshõ’s speech. 55 A similar instance of clarification is found in (59), where the coreferential marker -ki is used in a marked clause following the matrix clause: (59) a-ská-maĩnõ Marubo=taíse [i-inã]MARKED [teske DET:ANA-SIML-DS/A Marubo=PROBABLE say-S/A>S/A tie ak-a-ya=vo=rasĩ-ni noke chinã-ní-ki]MATRIX do.TR-NMLZ-CHAR.NMLZ=PL=PL-ERG 3PL think-UPWARD-MCM2 [noke ravã-ki]MARKED 1PL assume-S/A>A.SIM ‘After this happened, “(It was) probably the Marubo” so saying this, the soldiers went to us upriver, assuming (it) was us’ (lit. ‘the ones who do tying’)66 In this historical account, when a rubber-tapper’s child is kidnapped, Brazilian soldiers wrongly assume that the Marubo were responsible and go upriver to talk with them. Here there are two marked clauses which are governed by the matrix clause. The first clause contains the marker -inã ‘S/A>S/A’ and precedes the matrix clause. The S argument of the marked clause is coreferential with the A of the matrix clause and a result reading is understood; that is, as a result of thinking, the soldiers went. The second marked clause contains -ki ‘S/A>A.SIM’ and follows the matrix clause. Here it is attached to the transitive verb ravã ‘assume’, which almost always refers to a wrongly made assumption. Although covert, the A is clearly the soldiers, which is coreferential with the A in the matrix clause. What is significant however, is that this clause is somewhat of an afterthought and doesn’t seem essential to the utterance. Looking back at the direct speech, the soldiers already say (or think) that the Marubos were probably the ones who kidnapped the child. However, the narrator reiterates this information in the postposed clause, noting that they also assume (wrongly) that it was the Marubos (noke ‘1PL’). In the following example, this same pattern is seen with -inã, where the marked speech clause follows the matrix clause rather than preceding it as usual: The form -ya derives nouns from verbs, with the meaning “one who does X”. Following Fleck (2003), who describes a similar form in Matses, I have labeled it a “characterizer nominalizer”. 66 56 (60) [ẽ=papa-mẽ notsi-kawã-i]MARKED67 [ka-yãtá]MATRIX 1SG=father-OBL despair-ACROSS:SG/INTR-S/A>S.SIM go:SG-PST3 [nawa=namã ni-nõ]MAIN [i-inã]MARKED2 white.man=LOC be.standing-INT say-S/A>S/A ‘Beginning to despair about my father, I went to live where the white man lives. (lit. ‘I went, saying “I want to live in the white-man-place”’) The matrix clause describes the narrator’s actions of leaving when his father died. But rather than end the thought here, the narrator adds in a main clause that is followed by a marked clause containing -inã; this indicates that the S (or A) of this clause is coreferential with the S (or A) of the matrix clause. The marked clause (and the main clause directly before it) supplies necessary information (his thoughts), but it is somewhat less important than his “leaving”. That is, the narrator seems to emphasize his leaving rather than his volition to live in the place of the white man. Clauses containing the non-coreferential switch reference marker -maĩnõ may also occur following the matrix clause, as may be observed in (61): (61) [kamã jaguar nõ=yama-ma-chĩa]MATRIX 1PL=be.nothing-CAUS-PST2 [noke 1PL ãwẽ nacha-katsi-maĩnõ]MARKED 3SG bite-PROS-DS/A a-tí=se DEM:ANA-QUANTITY=FOC:ASS ‘We killed the jaguar, when it was about to bite us. That’s all.’ (lit. ‘that much’) In this example, the narrator concludes the story by repeating that he and his villagers had killed the jaguar. This statement is made in the matrix clause, but directly following this is a marked clause containing -maĩnõ, which indicates a different S or A argument than in the matrix clause. While the A argument of the matrix clause is nõ= ‘1PL’, referring to the group of villagers, the A argument of nacha ‘bite’ is ãwẽ ‘3SG’, which refers to the jaguar. Although it is already well known throughout the context of the narrative that the jaguar was attempting to bite the villagers 67 When attached to emotion or time verbs, the associated motion form -kawã and its counterpart -vawã/vã may have the extended function of marking inchoative aspect. See also (75) and Appendix 2, line 22. 57 and that they eventually killed the jaguar, the narrator includes both of these bits of information again. The difference in this order versus the default order seems to be more of an emphasis upon the event of “killing” rather than the jaguar’s “biting”.68 Furthermore, this reversal of order with -maĩnõ is quite interesting compared to nondefault ordered clauses which demonstrate coreferentiality, such as (58) - (60). In each of these examples, the listener hears the matrix clause then the marked clause which points back again to the matrix clause; however, compared to (61), there seems to be less mental processing as the S or A arguments remain the same. This is typologically consistent, as usually the same-subject marker is the unmarked choice (Roberts 2017; Stirling 2006). However, in (61), the matrix clause transitions from the referent as nõ= ‘1PL’ directly to ãwẽ ‘3SG’ (the jaguar) without any encoding; this is only signaled by -maĩnõ at the end of the clause. Although the pronoun and pronominal clitic in the marked clause (as well as context) make it clear to the listener that the referent is no longer nõ= ‘1PL’, the hearer must still reach the end of the marked clause before confirming that the identity of the marked clause referent is different than the identity of the matrix clause.69 Considering all of the non-default ordered constructions, there does not seem to be any case where ordering affects the underlying proposition, as in some other languages with switch reference systems. The ordering does however seem to relate to information structure, that is, the “speaker’s hypotheses about the hearer’s mental states…as they are reflected in grammatical structure” (Lambrecht 1994: 3). In one sense, the non-default ordering in Marubo clauses corresponds to what has typically been called “tails”. Tails are rightward-dislocated elements 68 The intonational contour of the second clause is falling, with significantly less stress than the first clause; this may also point to its status as more of elaborative or clarificational information. 69 A similar, but different type of construction will be treated more fully in §5.3. 58 which are usually considered afterthoughts meant to clarify or modify the predicate (Dik 1978). While formally, Marubo marked clauses which are found after the matrix clause, should not be understood to correspond to elements within a clause (as tails typically do), their function is almost identical to that of rightward-dislocated elements. In each of the examples presented above, the marked clause which follows the matrix clause contains clarificational information of which the hearer is most likely already aware; this information seems optional in some cases (59), and somewhat more necessary in others (58). In conclusion, these types of constructions occur quite pervasively and seem to relate to the speaker’s presentation in relation to the hearer’s assumptions. 5.2 The scope and target of switch reference clauses Another complex aspect of the Marubo switch reference system relates to the level of control (scope) a clause possesses and whether or not it targets (refers to) another adjacent clause. In the previous chapter, the majority of examples contain one marked clause (which is a subordinate clause) that is adjacent to one matrix clause (which is a main clause). In the previous examples, the target of the marked clause is the matrix clause; inversely stated, the matrix clause has scope over the marked clause. In clause chains, however, this is not always the case. Some clause chains, for example, are composed of not just one, but several marked (subordinate) clauses which do not target their following adjacent clause, but skip over them and target a matrix clause, which is also a main clause. In other cases, a marked (subordinate) clause targets another adjacent marked (subordinate) clause, which eventually targets a matrix (main) clause. In these cases, the marked (subordinate) clause also functions as a matrix clause; it has scope over another marked clause, even though it is not a main clause. Since matrix clauses are not necessarily main clauses, I have included the additional label “main” clause when necessary in 59 the following examples. This specifies that it is not subordinate to any other clause. Additionally, I have numbered marked and matrix clauses for ease of reference.70 Recall that out of the nine suffixes marking switch reference, seven mark coreferential relations, while only two mark non-coreferential relations. Dividing these further, four markers display sensitivity to the matrix clause argument’s grammatical role, that is, whether it is an S, or an A. With this in mind, the most straightforward way of determining the scope of the matrix clause is by observing which switch reference marker is used in the marked clause. For example, if the switch reference form marks coreferential relations with an A argument, the matrix clause to which it refers must contain a transitive verb.71 Observe, for example, the following, where one matrix clause has scope over three marked clauses: (62) (yora oni-pa=tõ…) [aa kamã rete-vaiki=se]MARKED1 person fierce-AUG=ERG 3SG jaguar spear-S/A>A.PREV=DM [veso-ke-ki=se]MARKED2 oĩtá [aĩvo ã=vake turn-COMPL-S/A>A.SIM=DM INTJ woman 3SG.POSS=child viã-ke-vaiki=se]MARKED3 [neõ tere-i]MAIN(MATRIX1-3) get\MAL-COMPL-S/A>A.PREV=DM stream.leaves\INSTR clean-MCM1 ‘the very fierce man…after spearing the jaguar, turning72 around, after taking the child from the woman, (he) cleaned (the child) with neo leaves’ (cf. verb root vi; absolutive neo) In this example, the participants include a fierce man, a woman, her child, and a jaguar. When examining each of the three marked clauses, we find that each one is marked with either -vaiki ‘S/A>A.PREV’ or -ki ‘S/A>A.SIM’ indicating that each argument in the marked clause must corefer 70 In the case of multiple marked and matrix clauses, I have numbered the matrix clause with the corresponding marked clause(s) over which it has scope. For example, MATRIX1-3 would indicate that this clause has scope over the first, second and third marked clauses. Where an analysis of the scope is unclear, I do not include such numbering. 71 While -a cross references O with S, A, or O in the matrix clause, note that there is not a Marubo switch reference marker which is exclusively oriented towards an O in the matrix clause. If such a morpheme existed in Marubo, it would not be possible to distinguish between A and O arguments by simply observing the matrix verb’s transitivity value. 72 There are two homophonous verb roots with the form veso: a transitive verb meaning ‘watch; care for’ and an intransitive verb meaning ‘awaken’. The form translated as ‘turn’ above may take the causative suffix (resulting in vesõ) and is likely another sense of ‘awaken’, something akin to ‘reverse’ or ‘change course’, since waking up involves a change in state from sleeping. 60 with an A argument. In the second marked clause, however, we find the intransitive verb veso ‘turn’, which clearly cannot be a controlling clause, as there is no marker which cross-references an S argument in the matrix clause. Had the first clause containing the verb rete ‘spear’ been marked by -vai ‘S/A>S.PREV’, then it would suggest that the controlling clause would be one containing an intransitive verb, and not any other clause (since they all contain transitive verbs). However, since rete is not marked by -vai, its controlling clause must be another clause which contains a transitive verb. Since the main clause contains the transitive verb tere ‘clean’ and since each of the marked clauses contain A-oriented markers, we may conclude that this main clause is also the matrix clause for each of the preceding marked clauses. This long-distance scope may be further illustrated in (63), where the main clause contains an intransitive verb and is preceded by three marked clauses:73 (63) [a-ská-vai=se]MARKED1 [yora=rasĩ atõ õpo DEM:ANA-SIML-S/A>S.PREV=DM person=PL 3PL\POSS clothes kocha-vai]MARKED2 [tipi sana-ni-sho=se]MARKED3 remove-S/A>S.PREV gun raise-UPWARD-S/A>S/A=DM [para i-vaĩ-vai=vo]MAIN(MATRIX1-3) fall:PL do.INTR-OUTWARD:PL/TR-PST1=PL ‘after this, after the people took off their clothes, with their guns raised upwards, (they) fell outwards (into the water)’ (cf. absolutive ato) The first marked clause contains the anaphoric demonstrative a, which is used as tail-head linkage, signalling that the final verb of the previous sentence is intransitive. This form is marked by -vai ‘S/A>S.PREV’ which indicates that the matrix clause must contain an S argument, and thus be intransitive; the second marked clause also contains this form, inflected on the transitive verb kocha ‘remove’. In the third marked clause, the verb sana ‘raise’ is marked by -sho ‘S/A>S/A’ indicating that its matrix clause may contain either an S or A argument. Therefore, since the final clause is the only one which contains an intransitive verb, it is clear that this clause must have 73 This example contains a partial repetition from (27). 61 scope over the first and second marked clauses. Although the third clause’s target is unclear (since it cross-references an S/A with an S/A), the semantics and chronological order of the clauses suggest that the final clause is its controlling clause. In other words, the people did not take their clothes off with their guns raised, rather, they fell (into the water) with their guns raised. In the two examples above, we saw that each marked clause was controlled by one final clause; this was primarily determined by observing the switch reference marker used (whether oriented towards a matrix S or A argument) and its surrounding verbs. In contrast, Marubo clause chains may also demonstrate local scope in which one clause depends upon another adjacent clause, which in turn depends upon another, and so on. Consider (64), for example, where there is a chain of three marked clauses which each targets the following adjacent clause: (64) ãwẽ i-maĩnõ 3SG say-DS/A richki-∅ hit-IMP a-tima-vaikĩ tell-FRUST-S/A>A.PREV ẽ=ene-sho-a 1SG=leave-BEN-O>S/A/O [tekõ-ki]MARKED1 [kane-kõ-vaĩ-ki]MARKED2(MATRIX) shoot.with.arrow-S/A>A.SIM miss-DISTR-OUTWARD:PL/TR-S/A>A.SIM [txivã-ke-vaĩ-i]MARKED3(MATRIX) [ka-i]MAIN(MATRIX) chase-COMPL-OUTWARD:PL/TR-S/A>S.SIM go:SG-MCM1 ‘after he said this, after I told (him) in vain “hit (it)”, when I let (it) go for him, shooting, missing, chasing (it), he went’ This example is quite complex as it contains several different kinds of switch reference constructions as well as direct speech embedded within direct speech.74 The participants involve a woman who is reporting the events which happpened moments before to another group of villagers. Within her story, she explains how she urged her husband in vain to hit the monkey that she was holding. Then, she describes the event of releasing the monkey and the ensuing actions of her husband. These events are presented in three marked clauses which each target the Note that the target of a ‘tell’ is the following adjacent clause. This type of strategy will also be discussed below in (68). 74 62 following adjacent clause; the argument of each clause refers to the husband. In the first marked clause, the verb tekõ ‘shoot with arrow’ is marked with -ki ‘S/A>A.SIM’ which indicates that the following argument will be an A. Looking ahead at the final verb ka ‘go:SG’, we can see that it is intransitive and thus cannot be the target of this first marked clause. However, the second marked clause contains the transitive verb kane ‘miss’, which logically is the target of the first marked clause. This second marked clause also contains the suffix -ki which points to the third marked clause, which contains the transitive verb txivã ‘chase’. The third marked clause, on the other hand, is inflected with the suffix -i ‘S/A>S.SIM’, indicating that its corresponding argument must be an S. Considering that the other two marked clauses do not contain intransitive verbs, the target of the third marked clause must be the final matrix clause, which is also a main clause. Now, it is possible, but less likely, that only one of the marked clauses is the matrix clause of the other two. For example, the third marked clause contains the transitive verb txivã ‘chase’ which could serve as the matrix clause for both of the preceding marked clauses, as they both contain verbs inflected with -ki ‘S/A>A.SIM’. This matrix clause in turn could target the (final) main clause. However, observe what happens in the following (nearly identical) elicited sentence where the second marked clause containing a transitive verb is substituted with an intransitive verb: 63 (65) [teka-kõ-vaĩ-i]MARKED1 [eré-ki]MARKED2(MATRIX) shoot.with.gun-DISTR-OUTWARD:PL/TR-S/A>S.SIM run-S/A>A.SIM [txivã-ke-vaĩ-i]MARKED3(MATRIX) [ka-i]MAIN(MATRIX) chase-COMPL-OUTWARD:PL/TR-S/A>S.SIM go:SG-MCM1 ‘shooting, running, chasing (it), he went’ In the first clause, the verb is teka ‘shoot with gun’ and is marked with -i ‘S/A>S.SIM’. This requires that its matrix clause contains an intransitive verb. Considering the three other clauses, the only two that contain intransitive verbs are the second marked clause or the final main clause. Either of these could be analyzed as the controlling clause for the first marked clause, but it seems more likely that it is the second marked clause. This is because the second marked clause contains -ki ‘S/A>A.SIM’, which requires its matrix clause to contain a transitive verb. Since the only transitive verb following the second marked clause is found in the third marked clause, it is almost certain that the third marked clause is a matrix clause for the second.75 Thus, in this chain, each marked clause targets the following adjacent clause.76 This strategy where a marked clause also operates as a matrix clause is illustrated again in (66): (66) vake veso-maĩnõ [oĩ-ma-vaikĩ=se]MARKED1 [oshã-vaiki=se]MARKED2(MATRIX) child wake-DS/A see-CAUS-S/A>A.PREV=EMPH sleep\CAUS-S/A>A.PREV=EMPH [rakã-vai=se]MARKED3(MATRIX) [nachi-i]MAIN(MATRIX) be.laying\CAUS-S/A>S.PREV=EMPH bathe-MCM1 ‘when the child wakes up, after (one) shows (it), (one) puts (it) to sleep, then (one) lays (it) down, then (one) takes a bath’ (cf. verb roots osha; raká) In this description of customs related to newborns, the narrator describes the actions of the mother after the baby wakes up early in the morning. The mother first takes time to show the baby to the other villagers. The first marked clause contains oĩ ‘see’ and is marked by -vaikĩ 75 The first marked clause also contains a transitive verb, however, it is highly unlikely for the second marked clause to refer backwards rather than forwards; this non-default ordering where the marked clause occurs after the matrix clause is used for clarificational purposes, which does not seem to be the case here. 76 If one strictly considers the transitivity of the markers used in each clause, it would be possible to suggest that the first marked clause targets the main clause, skipping over the other two while the second marked clause targets the third and the third targets the main clause. This would be a combination of both types of scope, however, I have not yet encountered any strong evidence for this type of construction. 64 ‘S/A>A.PREV’ which requires the corresponding argument to be an A. Since the final main clause does not contain an A, the most likely target of the first marked clause is the following adjacent marked clause. The verb in this clause is also marked by -vaiki which in turn corresponds with the third marked clause. Finally, the third marked clause contains the marker -vai ‘S/A>S.PREV’, which requires the corresponding matrix clause to contain an S argument. Examining the final clause, which is a main clause, one can see that it contains an intransitive verb nachi ‘bathe’, thus is must be the target of the third marked clause.77 As already mentioned above, notions relating to targets and the scope of the matrix clause can usually be determined by testing which core argument is targeted in the matrix clause. However, markers such as -sho/shõ ‘S/A>S/A’ which occur in clause chains do not definitively show which clause is being targeted, as they do not specify whether or not the corresponding argument is an S or an A: (67) a-ská-s-maĩnõ [totó tachi-karã-o-i] DEM.ANA-SIML-DUR-DS/A doctor appear-INWARD:SG/INTR-REP-S/A>S.SIM vana-i [ã=keshkã-ke-sho]MARKED1 [raká-sho]MARKED2 talk-S/A>S.SIM 3SG.POSS=side-COMPL-S/A>S/A be.laying-S/A>S/A [mĩ=tapóshe-namã atxi-sho]MARKED3 [toro-iná-sho]MARKED4 2SG.POSS=ankle-LOC hold-S/A>S/A curl-UPWARD-S/A>S/A [raká-∅]MAIN(MATRIX) ea a-vai be.laying-IMP 1SG tell-PST1 ‘then, the doctor, appearing, saying, “being on your side while laying, holding your ankle, while curling up, lay down” (he) told me’ In this personal report, a woman explains how the doctor told her to lay in a certain position as she is about to deliver her baby. Each of the bolded clauses above contain marked clauses which are part of the doctor’s direct speech. While the verbs within this clause chain are both transitive 77 It is possible that the first and second marked clauses are embedded within the third clause, and that the third clause is embedded within the main clause. This would require that the SR marker of the first clause skip over the second clause. However, as already stated above, there does not seem to be strong evidence for this, as eliciting similar examples with different transitivity values results in constructions where each adjacent marked clause also serves as a matrix clause. 65 (e.g. atxi ‘hold’) and intransitive (e.g. toro ‘curl’), there is no way to be certain as to the scope and/or target of each one. Had another marker, such as -vai ‘S/A>S.PREV’ or -ki ‘S/A>A.SIM’ been used, there would be evidence one way or another, but since -sho cross-references either an S or A argument, it remains unclear which kind of chaining construction this is. In the above examples, we have examined clause chains which mark coreferentiality between S/A and S/A arguments. These constructions contain matrix clauses which have either local or long-distance scope. However, clauses which contain either -a ‘O>S/A/O’, -maĩnõ ‘DS/A’, or -txinĩ ‘DS/A’ behave somewhat uniquely and deserve their own treatment. Remember that -a always occurs on transitive verbs, marking that an O argument is coreferential with an S, A, or O argument in its matrix clause. So, similar to constructions with -sho, strictly observing the transitivity of the following adjacent clause cannot definitively prove whether or not it is the matrix clause. However, clauses marked with -a must always serve as the matrix clause to the immediately preceding clause; this is illustrated in (68): (68) (Mĩshõ...) [tsaó-iná-ki]MARKED1 [ã=nĩkã-a]MARKED2(MATRIX) Mĩshõ be.sitting-UPWARD-S/A>A.SIM 3SG=hear-O>S/A/O [ã=vake tachi-kaĩ-ki]MAIN(MATRIX) 3SG.POSS=child appear-OUTWARD:SG/INTR-MCM278 ‘Mĩshõ…sitting up, when he heard (him), his child appeared’ In this part of the tale about Mĩshõ teaching his son to hunt, Mĩshõ is sleeping while his son is chasing a tapir. In the first marked clause, the verb tsaó ‘be sitting’ is inflected with -ki, requiring that the coreferential argument be an A. This corresponding A is found in the second marked clause, which contains the transitive verb nĩkã ‘hear’, still referring to the actions of Mĩshõ. The matrix clause for the first marked clause must be this second marked clause; it cannot be the 78 Note that the main clause marker -ki is identical in form (but completely different in function) to the switch reference marker -ki. In fact, -i ‘MCM1’ and -a ‘MCM3’ also resemble switch reference markers, but function as main clause markers. 66 main clause (containing tachi ‘appear’), as it contains an S (and also no longer refers to Mĩshõ).79 This is a pattern that occurs without exception in -a marked clauses, that is, one never finds the preceding clause marked with either -i ‘S/A>S.SIM’ or -vai ‘S/A>S.PREV’, since -a marked clauses always contain an A argument. Thus, it may be stated that -a marked clauses alway serve as the matrix clause for the preceding marked clause(s). Stated differently, marked clauses which precede -a marked clauses do not target other non-adjacent clauses. Clause chains which contain the non-coreferential markers -maĩnõ ‘DS/A’ and -txinĩ ‘DS/A’ are similar to -a marked clauses in that they also always serve as the matrix clause for preceding marked clauses. This is not surprising since everything following the clause on which they are marked is a different referent: (69) (txasho…) [ã=raká-paké-i]MARKED1 [ka-maĩnõ=se]MARKED2(MATRIX) deer 3SG=be.laying-DOWNWARD-S/A>S.SIM go:SG-DS/A=EMPH [richki-mtá ẽ=papã]MAIN(MATRIX) hit-PST3 1SG=father\ERG ‘deer…when it went to lay down, my father hit (it)’ (cf. absolutive papa) In (69), there is a chain of three clauses, two marked clauses and one main clause. The referent of the first two clauses is txasho ‘deer’, while the referent in the final clause is papa ‘father’. The target of the second marked clause is the final main clause. However, the clause that is marked with -maĩnõ, also serves as a matrix clause to the first marked clause, since the first marked clause cannot skip the clause marked with -maĩnõ and refer to another clause. This is always the case in constructions containing -maĩnõ; that is, the preceding marked clause always targets -maĩnõ marked clauses. One exception to this however, is a special construction which is presented in §5.3. 79 Also see Appendix C: Pani aká (To make a hammock) line 7, which contains a verb marked by -ki in the clause preceding the -a marked clause and an intransitive verb following the -a marked clause. Had these two clauses surrounding the -a marked clause been coreferential, -i or -vai would have been used. 67 While clauses marked with -maĩnõ ‘DS/A’ or -txinĩ ‘DS/A’ always serve as matrix clauses for a preceding clause or clauses, it is unclear whether they target the following adjacent clause or a final clause. This is because these forms are not exclusively oriented towards a particular matrix clause argument. Consider (70), which contains a chain with five clauses: (70) [motore rechni-kaĩ-s-maĩnõ=se]MARKED1 [aa yorã motor fade-OUTWARD:SG/INTR-DUR-DS/A=EMPH DEM:ANA person\ERG tachi-karã-ki]MARKED2 [ã=ewa kayã-vaiki]MARKED3 appear-INWARD:SG/INTR-S/A>A.SIM 3SG.POSS=mother flee\CAUS-S/A>A.PREV [vake atxi-ki]MARKED4 [vi-vaĩ-ki]MAIN(MATRIX) child catch-S/A>A.SIM get-OUTWARD:PL/TR-MCM2 ‘when the motor (sound) faded away, that person, appearing, after chasing away the mother, grabbing the child, took (him)’ (cf. absolutive yora) The first clause contains the verb rechni ‘fade’ and is marked by -maĩnõ; this refers to the sound of the motor. Each of the clauses following this refers to a man who kidnaps a child. Since -maĩnõ does not specify whether the following different referent will be an S or A argument, it cannot be determined which clause it is targeting by observing the transitivity value of the following clauses. It is also unclear which clauses are targeted in the following chain of coreferential clauses since they all contain transitive verbs and are all marked with -ki ‘S/A>A.SIM’; that is, they could each be targeting an adjacent clause or all targeting the final clause. Although the kind of scope (whether local or long-distance) is unclear from this example, a nearly identical elicited example provides confirmation that (70) is most likely long-distance scope: 68 (71) [motore rechni-kaĩ-s-maĩnõ=se]MARKED1 [aa yorã motor fade-OUTWARD:SG/INTR-DUR-DS/A=EMPH DEM:ANA person\ERG tachi-karã-ki]MARKED2 [nika-ki]MARKED3 [ã=ewa appear-INWARD:SG/INTR-S/A>A.SIM jump-S/A>A.SIM 3SG.POSS=mother kayã-vaiki]MARKED4 [vake atxi-ki]MAIN(MATRIX2-4) [a-ská flee\CAUS-S/A>A.PREV child catch-MCM2 DEM:ANA-SIML a-vaikĩ]MARKED1 [vi-vaĩ-ki]MAIN(MATRIX) do.TR-S/A>A.PREV get-OUTWARD:PL/TR-MCM2 ‘when the motor (sound) faded away, that person, appearing, jumping, after chasing away the mother, grabbed the child. After doing this, (he) took (him)’. (cf. absolutive yora) The only difference between this example and (70) is the addition of a clause containing an intransitive verb, changing atxi ‘catch’ from a marked clause to a main clause, and the addition of a tail-head linkage construction to begin a new sentence. In the third marked clause, the intransitive verb nika ‘jump’ directly follows the second marked clause containing tachi ‘appear’. Since this verb tachi retains the same marker, -ki, as in (70), this suggests that the scope is indeed long-distance; had tachi been targeting the clause with nika, it would have been marked with -i, marking that the following argument is an S. Even though this long-distance scope is present in the chain of coreferential clauses, it does not necessarily apply to the first marked clause containing -maĩnõ; this clause could still target the second marked clause. Finally, it is worth noting, that under no circumstances does one find a clause immediately following a -maĩnõ marked clause that does not contain a different referent (whether S or A); that is, there cannot be, for example, a following adjacent clause which contains the same referent as the clause marked with -maĩnõ.80 80 A possible exception to this, however, was one example I recently found in a composed text (which has not been verified with other speakers), where -maĩnõ serves to switch topics (with two different O arguments across clauses) but keeps the same A arguments. However, throughout the rest of my corpus, I don’t have any other examples of -maĩnõ functioning in this way. 69 In this section, I have demonstrated that Marubo switch reference chains may target adjacent clauses (local scope) or non-adjacent clauses (long-distance scope).81 Besides considering the logical sequence of events, the simplest way of determining this is to observe which matrix argument is being targeted (whether S or A) and find its corresponding verb, whether intransitive or transitive. Logically, this is only possible in chains which contain, -vai ‘S/A>S.PREV’, -vaiki/vaikĩ ‘S/A>A.PREV’, -i ‘S/A>S.SIM’, and -ki/kĩ ‘S/A>S.SIM’ as other markers allow for either S/A (or O) as a corresponding referent. Additionally, the markers -a ‘O>S/A/O’, -maĩnõ ‘DS/A’, and -txinĩ ‘DS/A’ always serve as the matrix clause to their preceding clause(s). While I am not completely certain of all of the factors influencing these two types of constructions, there does seem to be a pattern that emerges in coreferential chains. The major difference when comparing (62), (63), and (71) with (64) – (66) is the structure of the marked clauses. In the former three, there are new or different O arguments in the clause chain, while in the latter, the O arguments are all the same. This is not to say that clause chains do not contain intransitive verbs, as there are examples of intransitive verbs in both types of chains.82 What seems to be occuring, is that in clause chains that contain several transitive verbs with the same object, (regardless of the occurrence or not of an intransitive verb or its position relative to other clauses), each marked clause targets the following adjacent clause.83 On the other hand, in clause chains that contain more than one object (and thus are overt), marked clauses all target the main clause occurring at the end of the sentence. 81 Possibly, there may be a combination of both of these, where a marked clause precedes a clause marked with -a, -maĩnõ, or -txinĩ, which is then followed by a chain of clauses all targeting a main clause. This type of construction would illustrate both local and long-distance scope; unfortunately, I do not have any relevant examples. 82 In chains with intransitive verbs, such as (74), it is a bit more difficult to determine. More testing is needed, as there are much fewer natural examples containing chains with multiple intransitive verbs. 83 Another difference worth pointing out is that the latter three examples contain no overt core arguments; however, this is probably not a conditioning factor as there are other examples in my corpus of local scope constructions where the O is overt. 70 5.3 Interposed non-coreferential clauses The following construction, while relating to clause chaining and scope, is specific to clauses containing -maĩnõ ‘DS/A’, and thus is treated in this section on its own. In this type of construction, a clause marked with -maĩnõ ‘DS/A’ is inserted or “interposed” between coreferential clauses. Additionally, this construction involves what is known as clause skipping, where one clause is ignored and skipped over by another (Gijn 2016). Examine the structure of (72), a fairly simple chain involving three clauses: (72) (ã=)… [peshé mashte-vaĩ-vaiki]MARKED1 [atõ=seté-maĩnõ]MARKED2 3SG= shelter finish-OUTWARD:PL/TR-S/A>A.PREV 3PL=be.sitting:PL-DS/A [ato richki-ke~ke-ó-ki]MATRIX 3PL hit-ITER~COMPL-REP-MCM2 ‘he…after finishing the hut, when they were sitting down, (he) hit them again, over and over’ In this narrative, there are three characters, a husband and his two wives.84 The context is their journey back from a party, where the husband becomes angry and beats his wives. After making a temporary hut for the night, he beats them again (over and over). In the first clause, we see that the referent is the husband, who finishes building the hut. The verb mashte ‘finish’ is marked with -vaiki, indicating that the matrix clause referent will be an A; this A is found in the final clause containing the verb richki ‘hit’. Examining the second marked clause however, we find a clause which does not refer to the husband, but to the two wives, who are sitting down. This clause is marked by -maĩnõ, indicating a switch to a different S or A (back to the husband). Now, one could analyze this is as a clause chain with long-distance scope, since the first marked clause doesn’t target the adjacent clause, but rather, the final clause. However, this is significantly different, as long-distance scope chains involve a chain of coreferential clauses. Here, a clause 84 The husband is not ever introduced with a full noun phrase; he is only referred to with the pronominal clitic and as “having two wives”. 71 which contains a different referent appears directly after the first marked clause and remarkably, is not preceded by a marker indicating a switch in referents. That is, typically, when switching to another referent, one must mark the preceding clause with either -maĩnõ ‘DS/A’ or -txinĩ ‘DS/A’, otherwise one assumes that the clause continues to maintain coreferential relations. In this case, however, the clause marked with -maĩnõ seems to be added in, or interposed between the two coreferential clauses. While perhaps not essential to the narrator’s point, the function of this interposed clause seems to be that of adding explanatory or clarificational information about the other referent. This construction allows the speaker to do this without using two noncoreferential markers in a row. The following example is fairly similar, but contains a chain of four clauses, with the interposed clause occuring just before the matrix clause: (73) (yama-ti-ya)… (michpo)… (Shenipavo)… be.nothing-INSTR.NMLZ-CHAR.NMLZ ashes Shenipavo [vi-varã-i]MARKED1 [o-shõ]MARKED2 [ã=osha-maĩnõ]MARKED3 get-INWARD:PL/TR-S/A>S.SIM come:SG-S/A>S/A 3SG=sleep-DS/A [a-nõ a-kĩ]MATRIX DEM.ANA-OBL do.TR-MCM2 ‘the poisoner…ashes…Shenipavo…bringing (it), coming, while (hei) slept, with this, (hej) poisoned (himi)’ (lit. ‘the one who makes nothing’; ‘did him’) Throughout this narrative, an old man is trying to poison a boy called Shenipavo. The first, second, and matrix clauses refer to the poisoner, with the verbs vi ‘get’, o ‘come:SG’, and a ‘do’ respectively. In the first clause, the covert O refers to michpo, the “ashes” containing the poison, while in the matrix clause, the covert O refers to Shenipavo, who is poisoned. However, between the second marked clause and the matrix clause, a clause containing osha ‘sleep’ is marked with -maĩnõ, indicating that the following clause will contain a different S or A. In the context of the narrative, Shenipavo is the one sleeping, so it must refer to him. Interestingly though, there is 72 no marking that indicates a change in S or A between the second marked clause (referring to the poisoner) and the third marked clause (referring to Shenipavo). The second marked clause skips over the third marked clause and targets the matrix clause. This again exemplifies how clauses containing -maĩnõ may be inserted between coreferential clauses to add explanatory information without using -maĩnõ twice. To further illustrate how pervasive this construction is, examine (74), where -maĩnõ marked clauses are inserted three times in this portion of discourse: (74) [ã=ka-maĩnõ]MARKED1 [ã=aĩ txipo-sho wanĩ paĩ-i]MAIN(MATRIX1) 3SG=go:SG-DS/A 3SG.POSS=wife behind-PA peach.palm cook.in.pot-MCM1 [a-ská a-vaikĩ]MARKED2 [wanĩ naí-maĩnõ]MARKED3 DEM:ANA-SIML do.TR-S/A>A.PREV peach.palm cook-DS/A [reo-paké~paké-i]MAIN(MATRIX2) [a-vaí]MARKED4 [ã=vakẽ dump-ITER~DOWNWARD-MCM1 do.TR-S/A>S.PREV 3SG.POSS=child\ERG shoka-maĩnõ]MARKED5 [aa=ro peshé teké wetsã peel-DS/A 3SG=FOC:PAR shelter section another\LOC ronó-iná-sho]MARKED6 [ã=osha-a]MAIN(MATRIX4,6) [ã=osha-maĩnõ]MARKED7 be.hanging-UPWARD-S/A>S/A 3SG=sleep-MCM3 3SG=sleep-DS/A [ã=vakẽ shoká ak-í]MARKED8 [tsaó-sho]MARKED9 [oĩ-a=se]MARKED10 3SG=child\ERG peel do-S/A>S.SIM be.sitting-S/A>S/A see-O>S/A/O=EMPH [yosí-karã-ki=se]MARKED11 [panĩ know-INWARD:SG/INTR-S/A>A.SIM= EMPH hammock\LOC ã=ronó-s-maĩnõ=se]MARKED12 [tok nacha-a]MAIN(MATRIX11) 3SG=be.hanging-DUR-DS/A=EMPH IDEO:bite bite-MCM3 ‘after (the man) left, his wife, from behind, cooked the peach palm in a pot. After doing like this, when the peach palm became cooked, (the wife) dumped (it) out. After doing this, while the child was peeling (it), (the wife) was hanging (up in her hammock) at another section of the shelter, sleeping. While (the wife) was sleeping, when the child saw (the jaguar), while sitting, while peeling, when (it) came knowingly (sneakily),85 while (the wife) was hanging in the hammock, (it) tok, bit (the wife)’. (cf. absolutive vake; wetsa) 85 When inflected with the associated motion suffix -karã, denoting an inward motion, this verb takes the meaning ‘come sneakily’ literally, ‘come knowingly’. 73 Here, there are four participants, a man, his wife, their child, and a jaguar. In the first marked clause, the man leaves to go hunting. In the main clause, following the -maĩnõ marker, we find the verb paĩ ‘cook in a pot’ whose A refers to the wife. In the second marked clause, the tailhead linkage construction aská avaikĩ ‘after doing like this’ is used to refer back to the wife’s cooking the peach palm fruits. Since a ‘do’ is marked with -vaikĩ ‘S/A>A.PREV’, the matrix clause must contain an A. However, the following adjacent clause (the third marked clause) does not contain an A, but rather a completely different referent, wanĩ ‘peach palm’. The A is only found in the matrix clause containing reo ‘dump’. Thus, this skips over the interposed -maĩnõ clause, which briefly refers to the cooking of the peach palm and then switches back to the mother as an A argument. The fourth marked clause also contains a tail-head linkage construction, with the pro-verb a referring back to reo ‘dump’. This pro-verb is marked by -vaí ‘S/A>S.PREV’ indicating a following S argument which is only found in the sixth marked clause containing ronó ‘be hanging’ (or seventh marked clause containing osha ‘sleep’).86 Thus, it skips over the fifth marked clause, which is another interposed clause. The verb in this interposed clause, shoka ‘peel’ is marked with -maĩnõ and takes vake ‘child’ as its A argument. Similar to the previous interposed clause, the change in referent (in this case from the wife to the child) is not marked with a switch reference marker; however, since the A is marked as a full noun-phrase, there is no doubt as to the identity of the referent in this clause. Skipping ahead, the seventh marked clause is another tail-head linkage construction, with the repetition of the verb osha ‘sleep’, still referring to the wife. This verb is marked with -maĩnõ indicating a change in referent. In the following clause, (the eighth marked clause) the referent is again the child who is the A of this clause, the S of the ninth marked clause, and the A of the 86 Since both clauses contain intransitive verbs, it is unclear as to which is the target for the fourth marked clause. 74 tenth marked clause. Since the tenth marked clause contains -a ‘O>S/A/O’, we know that the O of this argument will appear in the following clause as an S, A or O. The verb of the eleventh marked clause is intransitive, yosí ‘come knowingly’, thus the argument is an S, however, the identity of this argument is still unknown; context and the following clauses help determine this. Note that yosí ‘come knowingly’ is marked with -ki ‘S/A>A.SIM’ which indicates that the matrix clause must contain an A. This A is not found in the following clause (it contains the verb ronó ‘hang’), but only in the final clause containing nacha ‘bite’. Thus, the eleventh marked clause skips over the twelfth marked clause, targeting the final matrix clause. The twelfth marked clause is an interposed clause containing -maĩnõ. In this clause containing ronó ‘hanging’, the referent appears as a pronominal clitic ã= ‘3SG’, which could either refer to the child or the wife. Since two previous clauses mention the wife sleeping, and since this clause also contains pani ‘hammock’, it is undoubtedly the wife to which this refers. But what of the eleventh and final clauses’ referents? While there is no overt argument revealing who or what is performing actions in these clauses, three details make clear that it is a jaguar. First, the title of the narrative given in the introduction is, aĩvo kamãne piá, which translates as ‘the jaguar’s eating of the woman’. Second, the verb nacha ‘bite’ that is used in the final clause suggests a non-human participant, since usually humans don’t bite other humans. Finally, in a later part of the narrative, the child tells her father that a jaguar ate her mother. In summary, the abrupt insertion of the clause containing ronó ‘hanging’ functions to confirm the setting where the jaguar bit the wife, reemphasizing her location in the hammock. There are a few characteristics that define interposed clauses containing -maĩnõ. First, these clauses involve a change in reference from one argument to another and then back to the first argument; the first change is made without the marking of a non-coreferential switch 75 reference marker. However, there is always a full noun, pronoun or pronominal clitic which identifies the referent either explicitly or by context. Second, clause chains may not be interposed; rather, only one clause is allowed to appear in this type of construction.87 Third, the function of interposed clauses allows a topic to be briefly introduced (or reintroduced) without using two consecutive -maĩnõ markers. This brief appearance of the topic supplies clarifying information without significantly interrupting the storyline. That is, the topic which precedes the interposed clause is maintained, and skips over the topic in the interposed clause. In one sense, interposed clauses contain a much less permanent topic than standard constructions involving -maĩnõ, as usually in the latter, the new topic continues throughout several clauses. Overall, this strategy involving interposed clauses allows the narrator to make brief clarifications while still maintaining continuity across clauses. 5.4 Exceptions to switch reference and topics for further research The Marubo switch reference system is quite efficient, tracking referents across clauses and signaling temporal or logical relations. However, exceptions exist within every system. This section is concerned with some of these exceptions as well as constructions requiring further exploration. In normal usage, switch reference markers indicate whether (particular) arguments are the same or different. Usually these arguments are animate (e.g. “Mĩshõ” or “woolly monkey”) but they occasionally may be inanimate (e.g. “machete” or “peach palm”). One non-canonical use of Marubo switch reference markers, rather than marking a referential change, is to mark a change in time, weather, and possibly, events. This use of switch reference follows the structure of 87 While it may be possible for clause chains to appear in this type of construction, I do not have any examples of this in my corpus. 76 interposed clauses (§5.3) and most often occurs directly after tail-head linkage constructions. Examine (75), for example: (75) …a-kĩ [a-ská a-vaikĩ]MARKED1 tell-MCM2 DEM:ANA-SIML do.TR-S/A>A.PREV [yãta-kawã-maĩnõ]MARKED2 [pani twiwa-ki]MATRIX be.night-ACROSS:SG/INTR-DS/A hammock connect-MCM2 ‘…(he) said. Doing like this, as it became night, (he) tied up the hammock’ (lit. ‘connected the hammock’) Here, the referent in the marked and matrix clause is a boy who after speaking to his friend, ties up his hammock. The first marked clause, a tail-head linkage construction, targets the matrix clause, skipping over the second marked clause. This interposed clause does not contain information about another animate or inanimate referent, but rather marks a change from late afternoon to night. Now, consider the next example, which marks a change in weather: (76) …petxi=ri rakã=vo-ki [atõ a-ská ak-a=se]MARKED1 behind=DIR be.laying\CAUS=PL-MCM2 3PL\ERG DEM:ANA-SIML do-O>S/A/O=EMPH [oi i-pá-maĩnõ]MARKED2 [sheré-kaĩ~kaĩ-i…]MARKED3 rain do.INTR-DOWNWARD-DS/A slide\DETR-ITER~OUTWARD:SG/INTR-S/A>S.SIM ‘they made (him) lay behind (them). Doing like this to (him), when the rain came down, (he) sliding over and over…’ (cf. absolutive ato; verb root shere) Here, the two participants are a group of villagers and a “lazy boy”. The verb in the first marked clause is inflected with -a, signaling that the O of this clause (the boy) will appear as an S, A or O in the corresponding clause, which is the third marked clause. The second marked clause, however, which is interposed, does not refer to either of the participants in the immediate context. Rather, it describes a change in weather to that of rain. Similarly, (77) seems to mark a change in event: 77 (77) (nõ…) …txasho pasa-i [a-ská a-vaikĩ=se]MARKED1 1PL beef make.stew-MCM1 DEM:ANA-SIML do.TR-S/A>A.PREV=EMPH [aa DEM:ANA mashté-kaĩ-maĩnõ=se]MARKED2 [nõ=panã finish\DETR-OUTWARD:SG/INTR-DS/A=EMPH 1PL=açaí motsa-i]MATRIX make.drink-MCM1 ‘(we)…made the beef into stew. Doing like this, when (that) was finished, we made açaí into drink’ The only participant in this example is nõ= ‘1PL’, referring to a group making food. This referent is found in the previous context as well as in the first marked clause and the matrix clause. The second marked clause is skipped over, as it does not contain the same referent. This skipped clause appears to mark a change in events, from describing the process of making stew to that of making açaí drink. While it could be argued that aa ‘DEM:ANA’ refers to the stew itself being cooked (thus making it an inanimate referent), the less specific verb mashte ‘finish’ seems to point to the previous event as a whole.88 Although marking these types of discontinuities is a rather flexible use of switch reference, this is not uncommon in the world’s languages, as already noted in §3.1. More research is needed to determine whether marking a change in events is actually a secondary function of switch reference in Marubo, however marking a change in time or weather is certainly widespread.89 Another exception to the switch reference system is often found in adjacent speech clauses containing two different referents. While it is often the case (e.g. (55)) that speech markers such as a ‘tell’ or i ‘say’ are marked with -maĩnõ to signal that the following clause 88 When verifying this with a native speaker, he confirmed that the demonstrative pronoun refers to the whole activity of making stew. For another similar example, see Appendix C, line 34 and line 38. 89 In general, clauses marking such types of discontinuities directly follow tail-head linkage constructions as in (75)(77), however this is not always the case. 78 contains a different referent who is speaking, sometimes the different referent clauses are simply juxtaposed without an explicit marker: (78) …[ẽ=tekõ-nõ]MAIN1 [ene-∅]MAIN2 [ik-i]MAIN3 [Yai, mia 1SG=shoot.with.arrow-INT leave-IMP say-MCM1 INTJ 2SG aká=ma i-katsai]MAIN4 get=NEG AUX:INTR-FUT ‘“…I want to shoot (it). Leave (it),” (he) said. “Ahhh, you won’t get (it).”’ In this example, the referents are a husband and his wife, who argue over how to kill a monkey. The first two clauses contain the direct speech of the husband and the third clause is the speech marker, still referring to him. In the fourth clause however, the direct speech is no longer that of the husband, but of his wife. Rather than marking this with switch reference, the narrator signals this switch simply by changing his intonation. Context, of course, is also a factor as the use of the first person proclitic ẽ= in the first clause and the second person pronoun mia in the fourth clause hint at different referents.90 While this exceptional use of switch reference is found in speech clauses with two referents, it is worth investigating further into whether it is possible in speech clauses with more than two referents. Finally, one other area needing more research is that of referential overlap, and what is considered a same or different referent. In many languages, switch reference markers show unusual behavior in part-whole relationships or when there are singular and plural referents which make up the same group. In some cases, the choice of which marker to use is based upon how the speaker chooses to frame the events; in others, this is determined by whether the marked or matrix clause contains the inclusive argument (Stirling 2006). Many of the Marubo examples presented previously are straightforward. When a single participant performs the same action across different clauses, coreferential markers are used; 90 For another switch in reference without an explicit marker, see Appendix B, lines 12 and 13. 79 when two or more participants perform different actions across different clauses, a noncoreferential marker is used. However, in situations where there is referential overlap, there is some irregularity. Consider this example containing a part-whole relationship: (79) [posto=namã ã=nasẽ-paké-a=se]MARKED1 [posto vook stomach=LOC 3SG=slice.front-DOWNWARD-O>S/A/O=EMPH stomach IDEO:spill poko okó-kaĩ-i=se]MARKED2 [vopi-i]MATRIX intestines spill\DETR-OUTWARD:SG/INTR-S/A>S.SIM=EMPH die-MCM1 ‘when he cut his front on his stomach, the stomach vok, the intestines spilling out, he died’ (cf. verb root oko) In this tale, a man loses his senses and begins eating himself. The first marked clause refers to the man and contains the verb nasẽ ‘slice front’.91 This is marked with -a, signalling that the O of this clause, the general frontal area, will occur in the corresponding clause as an S, A or O. However, in the S of the second marked clause, it is the stomach and intestines which spill out. Apparently, since the stomach and intestines are a part of the general frontal area, they appear to be coreferential arguments. Additionally, there is overlap in the second marked clause and matrix clause. The second marked clause contains the detransitivized verb oko ‘spill’ which is marked with -i. This signals that the corresponding clause will contain a coreferential S argument. While the matrix clause does contain an S, (the argument of vopi ‘die’) it actually refers to the man, not the stomach or intestines. Normally, a non-coreferential marker would be used, but since these are also in a part-whole relationship, they are considered coreferential. While this is just one type of referential overlap, there are a few other parameters that must also be explored. For example, are inanimate objects in a part-whole relationship (e.g. a bow and its string) treated the same way? How are individuals within a group treated, whether One widespread phenomenon in Panoan languages is what has been termed “body part prefixes…monosyllabic forms, representing mostly body parts, which are phonologically attached to the front of verbs, adjectives, and nouns” (Fleck 2006: 59) While quite productive in other Panoan languages, Marubo forms, such as nasẽ ‘slice front’ appear to have become lexicalized and are not very productive. However, it is still usually possible to trace the historical development of such forms. For example, the formative na refers to the front area of the body or other objects, while seno (currently still in use) means ‘slice’. 91 80 animate or inanimate? For example, in the following example, a coreferential marker is used, despite the difference in marking of number: (80) […ẽ=rera-nõ]MAIN [i-inã]MARKED [nõ=tea-vaĩ=mẽkĩ…]MATRIX 1SG=chop-INT say-S/A>S/A 1PL=block-OUTWARD:PL/TR=ADVS … ‘“I want to chop (it)” so saying this, we blocked it, but…’ OR ‘…we blocked it in order that I could chop it, but…’ The main clause containing direct speech and the marked clause with the speech marker i ‘say’ refer to a man, who wants to chop a jaguar. This speech marker i is inflected with the marker -inã ‘S/A>S/A’, signalling its referent is the same as in the matrix clause. In the matrix clause however, the pronominal clitic nõ ‘1PL’ refers not just to the man, but also his fellow villagers. Thus, perhaps related actions which are carried out by members of a group are considered to be coreferential. Relatedly, it is important to identify whether or not there is a further distinction between humans and non-humans; for example, Zariquiey (2011: 561) points out that in Kakataibo, arguments such as “a peccary and its herd” are treated as noncoreferential. In any case, more data and testing is needed to confirm these questions. 81 Chapter 6: Conclusion When describing the concept of switch reference, Haiman (1983: 105) has rightfully noted that it is “weird”, explaining that “…it seems to violate very general iconic tendencies whereby categories which define properties of nouns are expressed by nominal affixes, while conversely, categories which define properties of verbs are expressed by affixes on the verb”. Indeed, switch reference is peculiar and also relatively uncommon in languages around the world. In fact, initially, while focusing my attention primarily upon Marubo language acquisition, I completely overlooked the significance of this system, misinterpreting most of the forms as temporal markers. The extensive use of switch reference, however, is unmistakable and may be found in all genres of Marubo discourse.92 Of course, quantitative studies are needed to determine the frequency of use in each genre, but even in daily conversation, they are used quite regularly. The system itself is relatively complex, and as noted in chapter 4, is comprised of nine forms, almost all of which have a shared etymological origin with forms in other Panoan languages. Two exceptions are the forms -nõ and -nosho/noshõ which do not function as switch reference markers in Marubo. Over half of the Marubo forms track a specific argument across clauses; of these, the most cross-linguistically unusual (although common in Panoan), is -a, which tracks the object of the marked clause as coreferential with the S, A, or O of the matrix clause. In addition, many of the forms fulfill an extended function of marking temporal and logical relations between clauses. As demonstrated in chapter 5, the ordering of switch reference clauses may be varied. While usually the matrix clause follows the marked clause, in some cases, the order is reversed, 92 These forms are even found in the genre of myth-chants known as sai iti. 82 with the marked clause typically offering clarificational information. Also described in this chapter was the scope and target of switch reference in clause chains. While some constructions target an adjacent clause which also may be a marked clause, others target the final main clause. As already noted by other researchers,93 this is quite extraordinary, considering the number of clauses that may come between the marked and final main clause. As may be observed from examples throughout this thesis, Marubo possesses a minimalist structure in its clauses, often preferring little or no encoding of arguments. While ambiguity is often avoided by considering the broader context, the switch reference system certainly carries a heavy functional load in the tracking of these arguments. This is especially seen in the discussion of interposed non-coreferential clauses in §5.3. For example, in (74), -maĩnõ ‘DS/A’ is used to juggle the tracking of three participants, with one (the jaguar) being introduced by a switch in reference. Remarkably though, none of the clauses referring to the jaguar contain an overt argument. In this case, it is only the switch reference marker combined with the semantics of the verb that define the participant. Finally, as described in §5.4, there are some exceptions to my analysis of the switch reference system that deserve further examination. One of these is the use of switch reference markers to mark a discontinuity in time, weather, or events, rather than tracking a different referent. Interestingly, this usage of switch reference appears as interposed clauses which often follows tail-head linkage constructions. Another notable exception is the alternation of referents in speech clauses. While normally speech clauses are followed by speech markers, which in turn are marked by switch reference, sometimes, speech clauses produced by different participants are For example, Zariquiey (2016: 489) ponders the processing involved, pointing out that the “system described here requires the speaker to anticipate the transitivity of the main predicate of the sentence, since many of the forms presented here are specifically oriented towards the A or the S of the controlling clause. How is this possible, considering that the main verb may be preceded by a long chain of other dependent ones”? 93 83 juxtaposed and only differentiated by the narrator’s intonation or context. A third exception needing more research is the parameters which condition referential overlap. For example, when individuals are part of a group, what conditions whether the individual’s actions are viewed as the same or different than the group’s? In addition to these questions, I am sure that others related to switch reference will arise and warrant further analysis and description. In this thesis, I have tried to present a description of the switch reference system that details its complexities yet also demonstrates its broad usage throughout discourse. Throughout much of the analysis and writing, I was also challenged to understand other less familiar parts of the Marubo grammar. Some of these other areas will probably need reanalysis, yet the importance of documenting this data is significant, as a full grammar does not yet exist. Overall, I hope that this work will benefit others learning about Marubo, those in the broader context of Panoan studies, and typologists seeking to comprehend the exciting and complex world of switch reference. 84 References Aikhenvald, Alexandra Y. 2006. Serial verb constructions in typological perspective. In Alexandra Y. Aikhenvald & R. M. W. Dixon (eds.), Serial verb constructions: A crosslinguistic typology (Explorations in Linguistic Typology), 1–68. Oxford: Oxford University Press. Anonby, Stan & David J. Holbrook. 2010. A survey of the languages of the Javari River Valley, Brazil (SIL Electronic Survey Reports 2010–003). Dallas: SIL International. https://www.sil.org/resources/archives/9091 (15 August, 2020). Cesarino, Pedro de Niemeyer. 2008. Oniska: A poética da morte e do mundo entre os Marubo da Amazônia ocidental. Rio de Janeiro: Universidade Federal do Rio de Janeiro dissertation. Costa, Raquel R.G. 1992. Padrões rítmicos e marcação de caso em Marubo (Pano). Rio de Janeiro: Universidade Federal do Rio de Janeiro thesis. Costa, Raquel R.G. 1998. Aspects of Ergativity in Marubo (Panoan). Journal of Amazonian Linguistics 1(2). 50–105. Costa, Raquel R.G. 2000. Aspectos da fonologia Marubo (Pano). Rio de Janeiro: Universidade Federal do Rio de Janeiro dissertation. Dik, Simon C. 1978. Functional grammar. Amsterdam: North-Holland. Dik, Simon C. 1981. On the typology of focus phenomena. In Teun Hoekstra, Harry van der Hulst & Michael Moortgat (eds.), Perspectives on functional grammar, 41–74. Dordrecht: Foris Publications. Dixon, R. M. W. 2009. The semantics of clause linking in typological perspective. In R. M. W. Dixon & Alexandra Y. Aikhenvald (eds.), The semantics of clause linking : A crosslinguistic typology (Explorations in Linguistic Typology), 1–55. Oxford: Oxford University Press. Dixon, R. M. W. 2010. Basic linguistic theory (Volume 1: Methodology). Oxford: Oxford University Press. Dooley, Robert A. & Stephen H. Levinsohn. 2001. Analyzing discourse: A manual of basic concepts. Dallas: SIL International. Dryer, Matthew S. 2006. Descriptive theories, explanatory theories, and basic linguistic theory. In Felix Ameka, Alan Dench & Nicholas Evans (eds.), Catching language: The standing challenge of grammar writing, 207–234. Berlin: Mouton de Gruyter. Eakin, Lucille. 1991. Lecciones para el aprendizaje del idioma Yaminahua. (Documento de Trabajo 22). Lima: SIL International. Eberhard, David M., Gary F. Simons & Charles D. Fennig. 2020. Ethnologue: Languages of the World. https://www.ethnologue.com/ (13 May, 2020). Ferreira, Rogério Vicente. 2005. Língua Matis (Pano): Uma descrição gramatical. Campinas: Universidade Estadual de Campinas dissertation. Finer, Daniel L. 1985. The syntax of switch-reference. Linguistic Inquiry 16(1). 35–55. Fishman, Joshua A. 1991. Reversing language shift (Multilingual Matters 76). Clevedon: Multilingual Matters Ltd. Fleck, David W. 2003. A grammar of Matses. Houston: Rice University dissertation. Fleck, David W. 2006. Body‐Part prefixes in Matses: Derivation or noun incorporation? International Journal of American Linguistics 72(1). 59–96. Fleck, David W. 2007. Did the Kulinas become the Marubos? A linguistic and ethnohistorical investigation. Tipití: Journal of the Society for the Anthropology of Lowland South America 5(2). 137–207. Fleck, David W. 2013. Panoan languages and linguistics (Anthropological Papers of the American Museum of Natural History 99). New York: American Museum of Natural History. Gijn, Rik van. 2016. Switch reference: An overview. In Rik van Gijn & Jeremy Hammond (eds.), Switch reference 2.0 (Typological Studies in Language 114), 1–53. Amsterdam: John Benjamins. Givón, Talmy. 1983. Topic continuity in discourse: The functional domain of switch-reference. In John Haiman & Pamela Munro (eds.), Switch reference and universal grammar (Typological Studies in Language 114), 51–82. Amsterdam: John Benjamins. 85 Haiman, John. 1983. On some origins of switch reference marking. In John Haiman & Pamela Munro (eds.), Switch reference and universal grammar (Typological Studies in Language 2), 105–128. Amsterdam: John Benjamins. Haiman, John & Pamela Munro. 1983. Introduction. In John Haiman & Pamela Munro (eds.), Switch reference and universal grammar (Typological Studies in Language 2), iv–xv. Amsterdam: John Benjamins. Haspelmath, Martin. 2007. Coordination. In Timothy Shopen (ed.), Language typology and syntactic description (Volume 2: Complex Constructions), 1–51. Cambridge: Cambridge University Press. ISA. 2018. Povos indígenas no Brasil: Marubo. https://pib.socioambiental.org/en/Povo:Marubo (13 May, 2020). Jacobsen, William H. 1967. Switch-reference in Hokan-Coahuiltecan. In Dell Hymes & William E. Bittle (eds.), Studies in southwestern ethnolinguistics: Meaning and history in the languages of the American southwest, 238–263. The Hague: Mouton. Kennell, Gerald Jr. 1978. Descrição da gramática e da fonêmica da língua Marubo. Unpublished manuscript. Kennell, Gerald Jr. 1998. Katukina lessons and answers. Unpublished manuscript. Lambrecht, Knud. 1994. Information structure and sentence form: Topic, focus, and the mental representations of discourse referents. Cambridge: Cambridge University Press. Lauriault, James. 1948. Alternate-Mora timing in Shipibo. International Journal of American Linguistics 14(1). 22–24. Lewis, M. Paul & Gary F. Simons. 2010. Assessing endangerment: Expanding Fishman’s GIDS. Revue roumaine de linguistique LV(2). 103–120. Loos, Eugene E. 1999. Pano. In R. M. W. Dixon & Alexandra Y. Aikhenvald (eds.), The Amazonian languages, 227–250. Cambridge: Cambridge University Press. Lord, Mary Ann. 2016. Some features of Yora grammar (Documento de Trabajo 33). Lima: Instituto Lingüístico de Verano. Melatti, Delvair Montagner. 1985. O mundo dos espíritos: Estudo etnográfico dos ritos de cura Marubo. Brasília: Universidade de Brasília dissertation. Melatti, Delvair Montagner & Julio Cesar Melatti. 1986. A maloca Marubo: A organização de espaço. Revista de Antropologia 29. 41–55. Munro, Pamela. 1979. On the syntactic status of switch-reference clauses. In Pamela Munro (ed.), Studies of switch-reference (UCLA Papers in Syntax 8), 144–166. Los Angeles: UCLA. Overall, Simon. 2009. The semantics of clause linking in Aguaruna. In R. M. W. Dixon & Alexandra Y. Aikhenvald (eds.), The semantics of clause linking : A cross-linguistic typology (Explorations in Linguistic Typology), 167–192. Oxford: Oxford University Press. Payne, Thomas E. 1997. Describing morphosyntax: A guide for field linguists. Cambridge: Cambridge University Press. Reesink, Ger P. 1983. Switch reference and topicality hierarchies. Studies in Language 7(2). 215–246. Roberts, John. 1988. Amele switch-reference and the theory of grammar. Linguistic Inquiry 19(1). 45–63. Roberts, John. 2017. Switch reference. In Alexandra Y. Aikhenvald & R. M. W. Dixon (eds.), The Cambridge handbook of linguistic typology, 538–573. Cambridge: Cambridge University Press. Souza, Hilário de. 2016. Some non-canonical switch reference systems and the fundamental functions of switch reference. In Rik van Gijn & Jeremy Hammond (eds.), Switch reference 2.0 (Typological Studies in Language 114), 55–92. Amsterdam: John Benjamins. Sparing-Chávez, Margarethe W. 2012. Aspects of Amahuaca grammar: An endangered language of the Amazon basin (SIL E-Books 51). Lima: SIL International. Stirling, Lesley. 1993. Switch-reference and discourse representation. Cambridge: Cambridge University Press. Stirling, Lesley. 2006. Switch-reference. Encyclopedia of language and linguistics. Amsterdam: Elsevier. 86 Thompson, Sandra A., Robert E. Longacre & Shin Ja J. Hwang. 2007. Adverbial clauses. In Timothy Shopen (ed.), Language typology and syntactic description (Volume 2: Complex Constructions), 237–300. Cambridge: Cambridge University Press. Valenzuela, Pilar M. 2003. Transitivity in Shipibo-Konibo grammar. Eugene: University of Oregon dissertation. Valenzuela, Pilar M. 2005. Participant agreement in Panoan. In Nikolaus Himmelmann & Eva Schultze-Berndt (eds.), Secondary predication and adverbial modification: The typology of depictives, 259–298. Oxford: Oxford University Press. Valenzuela, Pilar M. & Antoine Guillaume. 2017. Estudios sincrónicos y diacrónicos sobre lenguas Pano y Takana: Una introducción. In Antoine Guillaume & Pilar M. Valenzuela (eds.), Estudios sincrónicos y diacrónicos sobre lenguas Pano y Takana (Amerindia 39:1), 1–52. Paris: Association d’ethnolinguistique amérindienne. Watters, John. 1979. Focus in Aghem: a study of its formal correlates and typology. In Aghem grammatical structure: With special reference to noun classes, tense-aspect and focus marking., 137–197. Los Angeles: University of Southern California. Welper, Elena Monteiro. 2009. O mundo de João Tuxaua: (Trans)formação do povo Marubo. Rio de Janeiro: Universidade Federal do Rio de Janeiro. Werlang, Guilherme. 2001. Emerging peoples: Marubo myth-chants. St Andrews: University of St Andrews dissertation. Zariquiey, Roberto D. 2011. A grammar of Kashibo-Kakataibo. Bundoora: La Trobe University dissertation. Zariquiey, Roberto D. 2016. Target, embedding and switch-reference constructions in Kakataibo (Panoan, Peru). In Rik van Gijn & Jeremy Hammond (eds.), Switch reference 2.0 (Typological Studies in Language 114). Amsterdam: John Benjamins. Zariquiey, Roberto D. 2018. The encoding of emotions in Kakataibo (Panoan): Morphological markers and prosodic patterns. In Maïa Ponsonnet & Marine Vuillermet (eds.), Morphology and emotions across the world’s languages (Studies in Language 42:1), 182–201. Amsterdam: John Benjamins. 87 Appendices Appendix A: Even/odd syllable allomorphy In §4.1, I briefly described the pattern of allomorphy in switch reference markers, noting that this seems to be based on whether the number of syllables (or moras) preceding the affix is even or odd. This allomorphy is widespread, occurring not only in switch reference markers, but also in other verbal affixes such as valency-changing operations, associated-motion morphemes, and tense and aspect markers. As also noted in §4.1, most Marubo verb roots are disyllabic, thus forms following disyllabic or other even-numbered roots will occur much more often than forms following odd-numbered roots. This “default” form, which occurs with even-numbered syllables (or moras), is listed second in each heading below (in bold), while the form that occurs with oddnumbered syllables (or moras) is listed first. Note that this list is not exhaustive. Benefactive applicative -sho/shõ (81) [a.ˈʂõ.βai] (1 syllable preceding) nawã ora=ka noke a-shõ-vai white.man\ERG far=ADJZ 1PL make-BEN-PST1 ‘the white man made (it) expensive for us’ (cf. absolutive nawa) (82) [ˈβɨ.so.ʂo.βai] (2 syllables preceding) ewã=ro vake=rasĩ ea veso-sho-vai mother\ERG=FOC:PAR child=PL 1SG watch-BEN-PST1 ‘mother watched the children for me’ (cf. absolutive ewa) Associated-motion: DOWNWARD -paké/pá (83) [vi.ˈpai] (1 syllable preceding) mani vi-pá-ai banana get-DOWNWARD-HOD ‘(she) got down bananas’ 88 (84) [ts.a.ˈo.pa.kɨ.ai] (2 syllables preceding) Tama maĩ tsaó-paké-ai Tama ground\LOC sit-DOWNWARD-HOD ‘Tama sat on the ground’ (cf. absolutive mai) Associated-motion: COME.DO.HERE -wa/wã (85) [a.ˈwãi] (1 syllable preceding) rama=ro ma=ea a-wã-ai now=FOC:PAR already=1SG do.TR-COME.DO.HERE-HOD ‘now (he) already came and did (poisoned) me’ (86) [ˈɾiʃ.ki.wa.βai] (2 syllables preceding) Tamã shavá=ma neno shovo naki rono Tama\ERG today=NEG here roundhouse middle snake richki-wa-vai hit-COME.DO.HERE-PST1 ‘Yesterday, Tama came here and hit a snake inside the roundhouse’ (cf. absolutive Tama) Associated-motion: GO.DO.THERE -ta/tã (87) [ɨ.̃ pi.ˈtã.βai] (1 syllable preceding) e-rí ka-shõ ẽ=pi-tã-vai94 1SG-DIR go-S/A>S/A 1SG=eat-GO.DO.THERE-PST1 ‘“I myself going, I went and ate (there)’ (88) [ˈɨ.wɨ.ta.yã.ta] (2 syllables preceding) chinã-vaĩ-sho ea ewe-ta-yãtá think-OUTWARD:PL/TR-SS 1SG get-GO.DO.THERE-PST3 ‘(this person) going out, got me (at another place)’ (lit. ‘this person thinking out’) Habitual Aspect -mtsãwã/miska (89) [a.ˈmis.ka] (1 syllable preceding) wetsã a-ská a-míska another\ERG DEM:ANA-SIML do.TR-HAB ‘another typically does like this’ (cf. absolutive wetsa) Even though two syllables precede this form, the proclitic ẽ= is most likely extrametrical, since examples without the proclitic (e.g. pi-tã-∅ ‘eat-GO.DO.THERE-IMP’) retain the same form. See (95),(97), (104) and (105) for further examples of proclitics behaving as extrametrical. 94 89 (90) [βa.ˈtʃĩm.tsã.wã] (2 syllables preceding) nokẽ yõsha=rasĩ noke vatxĩ-mtsãwã, 1PL\POSS old.woman=PL 1PL scold-HAB ‘our old women regularly scold us’ (91) [ˈtɨ.ka.ʂo.mis.ka] (3 syllables preceding) papa yoini noke teka-sho-míska father animal 1PL shoot.with.gun-BEN-HAB ‘Our father regularly shoots animals for us’ (92) [tsa.ˈo.pa.kɨm.tsã.wã] (4 syllables preceding) noke tsaó-pake-mtsãwã, maĩ 1PL be.sitting-DOWNWARD-HAB ground\LOC ‘We typically sit down on the ground’ (cf. absolutive mai) Remote past -chná/chĩa (93) [ka.ˈʃĩ.a] (1 syllable preceding) vopi=ma ka-chĩa awá=ro die=NEG go:SG-PST2 tapir=FOC:PAR ‘(It) didn’t die. The tapir went.’ (94) [ˈβo.piʃ.na] (2 syllables preceding) a-ská-i vopi-chná DEM:ANA-SIML-S/A>S.SIM die-PST2 ‘(he) died (being) like that’ (95) [nõ.ˈya.ma.ma.ʃĩ.a] (3 syllables preceding) kamã ani=ka nõ=yama-ma-chĩa jaguar be.large=ADJZ 1PL=be.nothing-CAUS-PST2 ‘We killed a large jaguar’ (96) [ˈno.ko.pa.kɨʃ.na] (4 syllables preceding) ne-no nokó-pake-chná DEM:PROX-LOC reach\DETR-DOWNWARD-PST2 ‘(they) arrived down here (from upriver)’ (cf. verb root noko) 90 Distant past -mtá/yãtá (97) [ɨ.̃ ˈo.yã.ta] (1 syllable preceding) ẽ=o-yãtá ne-no 1SG=come:SG-PST3 DEM:PROX-LOC ‘I came here’ (98) [ˈno.kom.ta.vo] (2 syllables preceding) vo-i noke noko-mtá=vo go:PL-S/A>S.SIM 1PL reach-PST3=PL ‘going, they reached us’ Future -katsa/katsai (99) [ˈβi.ka.tsa.i] (1 syllables preceding) ã=vake vi-katsai 3SG.POSS=child get-FUT ‘(he) will get his child’ (100) [ma.ˈna.ka.tsa] (2 syllables preceding) vai naki maná-katsa trail middle wait.for-FUT ‘(we) will wait for (you) in the middle of the trail’ (101) [ˈta.na.ma.ka.tsa.i] (3 syllables preceding) mia tana-ma-katsai 2SG understand-CAUS-FUT ‘(I) will tell you’ (lit. ‘cause you to understand’) (102) [i.ˈo.ka.ɾã.ka.tsa] (4 syllables preceding) shavá=ma ne-no i-ó-karã-katsa today=NEG DEM:PROX-LOC do.INTR-REP-INWARD:SG/INTR-FUT ‘tomorrow I will come back here again’ 91 Examples with more than 4 syllables preceding the affix (103) [ˈno.ko.βa.ĩ.ta.yã.ta] (5 syllables preceding) ka-kĩ noke noko-vaĩ-tá-yãtá go:SG-S/A>A.SIM 1PL reach-OUTWARD:PL/TR-GO.DO.THERE-PST3 ‘going, (he) reached us out there’ (104) [nõ.ˈʂo.ko.βa.ĩ.βa.ĩʃ.na] (6 syllables preceding) a-no nõ=shoko-vaĩ~vaĩ-chná DEM:ANA-LOC 1PL=be.standing:PL-ITER~OUTWARD:PL/TR-PST3 ‘we used to live out there’ Exceptions to the pattern (105) [mĩ.pi.ˈã.kãʃ.na] (-chná ‘PST2’ usually follows even-numbered syllables) vake ea mĩ=piãkã-chná child 1SG 2SG=eat\MAL-PST2 ‘you ate (my) child (to my detriment)’ (cf. verb root pi) (106) [õ.ˈtʃĩ.kãʃ.na] (-chná ‘PST2’ usually follows even-numbered syllables) ea roé õtxĩkã-chná 1SG ax throw\MAL-PST2 ‘you threw (my) ax (to my detriment)’ (cf. verb root õtxi) (107) [ɾ.a.õ.kãm.ta] (-mtá ‘PST3’ usually follows even-numbered syllables) ẽ=vake=rasĩ ano-sho Kanamari-kã raõ-kã-mtá 1SG=child=PL DEM:ANA-PA Kanamari-ERG treat-MAL-PST1 ‘from there, the Kanamari people, cursed my children’ (cf.verb root raõ) (108) [ˈa.tʃi.kɨm.ta] (-mtá ‘PST3’ usually follows odd-numbered syllables) mera-sho ea atxi-ke-mtá find-S/A>S/A 1SG catch-COMPL-PST3 ‘finding (me), (they) caught me’ (109) [kɨ.ˈpɨ.̃ kɨ.ʂo] (-sho ‘S/A>S/A’ usually follows odd-numbered syllables) yama-ti txené kepẽ-ke-sho be.nothing-INSTR.NMLZ pouch open-COMPL-S/A>S/A ‘with the poison pouch open,’ 92 Observations • Applicatives and associated motion suffixes occur in the slot right after the verb root (compared to other suffixes); thus, it is possible, but less likely for them to have more than two intervening syllables (or moras). • Proclitics do not seem to be counted as a syllable (or mora) • There are a few exceptions to this pattern, where affixes which typically occur after an even number of syllables occur after an odd number of syllables and vice-versa. These often occur following the completive aspect marker -ke and the malefactive applicative -(Ṽ)kã. • It is probable that stress is a secondary factor conditioning this allomorphy. • While there are many other verbal suffixes with some similar charateristics (i.e. monosyllabic and/or containing nasalized vowels), it seems that even-odd allomorphy does not occur with those other suffixes (e.g. -i ‘S/A>S.SIM’, -a ‘O>S/A/O’, or -nĩ ‘COM’ don’t have allomorphs). 93 Appendix B: Kamãne pikẽnkeã yoã (The jaguar’s almost eating story) This narrative is about a man, his wife, and their child, who go out to recover a peccary he had killed and left floating in a stream.95 When they come across the peccary, it is being eaten by a jaguar which turns and attacks them. The husband flees, leaving the wife to defend herself and her child from the jaguar, which urinates on them in order to tire them out. When the husband arrives back at the village, he tells everyone that he saw his wife being eaten. One “fierce man” however, takes a spear and tapir skin, which serves as a shield and goes out to find the woman and child. He eventually locates them and slays the jaguar. After this, he washes them off with nearby plants and takes them back to the village. The elders chant for the woman and child, but the child dies. After the woman recovers, the villagers do not give her back to her husband, but to the “fierce man”. 1 Aa yoã-ya=vo=taíse nĩkã-tso tell-COMPL=PL=PROBABLE hear-IMP ‘They have probably told this one before, listen’ DEM:ANA 2 yawa nonõ-vai pote-nõ i-inã ã=aĩ white.lipped.peccary float\CAUS-PST1 gut-INT say-S/A>S/A 3SG.POSS=wife ewe-vaĩ-a, ã=aĩ-nĩ ka-a ik-í get-OUTWARD:PL/TR-MCM3 3SG.POSS=wife-COM go:SG-MCM3 say-MCM1 ‘(he) took his wife to gut a white-lipped peccary that (they) had floated, he went with his wife, they say’ (cf. verb root nonó) (lit. ‘the white-lipped peccary that (he) had floated, “I want to gut (it)” saying this’…) 3 ka-kĩ ãwẽ oĩ-a, yawa nonõ-vai=vo go:SG-S/A>A.SIM 3SG see-O>S/A/O white.lipped.peccary float\CAUS-PST1=PL nini-ni-sho pi-á sotã-i raká-a, drag-UPWARD-S/A>S/A eat-MCM3 wait.for-S/A>S.SIM be.laying:SG-MCM3 ‘going along, when they saw (it), it was eating the white-lipped peccary they had floated, laying in wait, after having dragged it up’ (cf. verb root nonó) 95 When hunting white-lipped peccary, the Marubo often kill more than they are able to carry back in one trip. Because of this, they often float the peccary in a nearby stream and tie it down with a vine, which preserves the meat. 94 4 aa DEM:ANA a-ská-s-maĩnõ=se kamã-ne yawa DEM:ANA-SIML-DUR-DS/A=DM jaguar-ERG white.lipped.peccary piã-i raká-sho=se vi-varã-ki eat\MAL-S/A>S.SIM be.laying-S/A>S/A=DM get-INWARD:PL/TR-MCM2 ‘When this happened, the jaguar, while laying down, eating the white-lipped peccary (to their detriment), came running’96 (cf. verb root pi) 5 ãwẽ a-ská-maĩnõ tekõ-ti roa=ka=mẽkĩ=ro=tsẽ 3SG DEM:ANA-SIML-DS/A shoot.with.arrow-IRR good=ADJZ=ADVS=FOC:PAR=NEXT97 veso-ke nash kaya-kawã-katsa turn-COMPL IDEO:run flee:SG-ACROSS:SG/INTR-FUT ã=aĩ ene-ai-tĩ 3SG.POSS=wife leave-HOD-ANTIMIR ‘(it) doing this, it would have been good to shoot it, however he turned around, nash, he will run away, he is leaving his wife, you know’ 6 a-ská ak-a ã=aĩ vake toá-ya DEM:ANA-SIML do.TR-O>S/A/O 3SG.POSS=wife child be.spread-CHAR.NMLZ ik-i say-HOD ‘Doing like this to (her), “the wife was carrying her child in a sling” they say (lit. ‘wife was a child-slinger’) 7 ii INTJ ea kamã-ne ak-a o=rí-∅ 1SG jaguar-ERG get-HOD come:SG=DIR-IMP ea kamã-ne ak-a o=rí-∅ ik-í 1SG jaguar-ERG get-HOD come:SG=DIR-IMP say-MCM1 ‘“A jaguar is getting me, come here, a jaguar is getting me, come here!”, (she) said. 8 Naa, ã=vake ninĩ-nõ i-inã ãwẽ DEM.PROX 3SG.POSS=child drag\MAL-INT say-S/A>S/A 3SG nika-kaĩ-maĩnõ ne=ki=ri=se vake jump-OUTWARD:SG/INTR-DS/A DEM.PROX=LOC=DIR=FOC:ASS child wachí-ké-i, deny-COMPL-MCM1 ‘“(I) want to drag away her child (to her detriment)” (the jaguar) thinking this, when (it) jumped, she denied the child.’ (cf. verb root nini) The apparent mismatch between the lexeme vi ‘get’ and its free translation “came running” is explained by understanding this as a figure of speech. This non-literal understanding is also found in other texts, when used in the context of jaguars. 97 The form -tsẽ may be another type of discourse marker which advances the storyline, similar to -se described earlier. 96 95 9 kamã-ne isõ-ne voi-vaĩ-i jaguar-ERG urine-INSTR splash-OUTWARD:PL/TR-MCM1 ‘The jaguar splashed (them) with urine’ 10 o-ri=se i-o-kawã-ó-ki DEM.DIST-DIR=FOC:ASS do.INTR-REP-ACROSS:SG/INTR-REP-S/A>A.SIM ã=vake-mẽ wachí-ké-s-maĩnõ=se, kamã-ne isõ-ne 1SG.POSS=child-OBL deny-COMPL-DUR-DS/A=DM jaguar-ERG urine-INSTR voi-vaĩ-i splash-OUTWARD:SG/INTR-MCM1 ‘(It) going over there again, when (she) denied (the jaguar) her child, the jaguar splashed (them) with urine. 11 A-ská=tsẽ ã=shavá=namã {tachi-kaĩ-ki DEM.ANA-SIML=NEXT 3SG.POSS=village=LOC appear-OUTWARD:SG/INTR-S/A>A.SIM ã=nĩkã-a mã} tachi-kaĩ-i 3SG=listen-O>S/A/O INTJ appear-OUTWARD:SG/INTR-MCM1 ‘Happening like this, {when he appeared out at his village, oh} 98 he appeared. 12 ii INTJ ẽ=aĩ ea kamã-ne piã-ai ik-í ẽ=aĩ ea 1SG=wife 1SG jaguar-ERG eat\MAL-HOD say-MCM1 1SG=wife 1SG kamã-ne piã-ai jaguar-ERG eat\MAL-HOD ‘“A jaguar ate my wife (to my detriment)!” he said. “A jaguar ate my wife (to my detriment)!” (cf. verb root pi) 13 ãwẽ i-maĩnõ yora oni-pa ii ma=ak-á=ra 3SG say-DS/A person fierce-AUG INTJ already=get-NMLZ=Q mĩ=oĩ-varã-ai 2SG=see-INWARD:PL/TR-HOD ‘when he said this, a very fierce man (said) “When you came, did you already see (its) getting of her?”’ 14 ii ma=ak-á oĩ-varã-ai already=get-NMLZ see-INWARD:PL/TR-HOD ‘“When I came, I saw (its) getting of her!”’ INTJ 98 The use of curly brackets indicates a false start, tangent, or filler word used by the narrator. 96 15 Veso-ke-ki=se, vitxi vi-i, awá vitxi pichtama, turn-COMPL-S/A>A.SIM=DM skin get-MCM1 tapir skin huge naa-tió=taíse, kenã shavá-tió=taíse, aa DEM:PROX-WIDTH=PROBABLE sitting.logs space-WIDTH=PROBABLE DEM:ANA vi-i, paka-visi mespo-ta-ni get-MCM1 spear-CLEARLY hold.in.hand-GO.DO.THERE-MCM1 eré~eré-i eré~eré-i ITER~run-MCM1 ITER~run-MCM1 ‘turning around, he got the skin. The tapir skin was huge, probably this wide, probably the width of the space (between) the sitting logs. He got that one. He obviously went and grabbed his spear. He ran and ran. He ran and ran.’ 16 ka-kĩ ãwẽ nĩkã-a, ii, kená-i ik-i ooo, ooo, go:SG-S/A>A.SIM 3SG hear-O>S/A/O INTJ call-MCM1 say-HOD INTJ INTJ kamã-ne ea pi-a oĩ-rí, kamã-ne ea pi-a oĩ-rí, jaguar-ERG 1SG eat-HOD see-DIR jaguar-ERG 1SG eat-HOD see-DIR kená-i. call-MCM1 ‘going, when he heard her, (she) called out, they say. “Ooo, ooo, a jaguar is eating me, look here, a jaguar is eating me, look here,” she called out. 17 ii, INTJ mia oĩkanãke99 i-inã, ka-kĩ ãwẽ oĩ-a, aĩvo 2SG see say-S/A>S/A go:SG-S/A>A.SIM 3SG see-O>S/A/O woman ma=yome-ya, ã=machít a-vévakĩ~vakĩ-a already=be.tired-COMPL 3SG=jump do.TR-ITER~APART-MCM3 kane-ma-vakĩ~vakĩ-a. miss-CAUS-ITER~APART-MCM3 ‘You’ll see” saying this, going along, when he saw (her), the woman was already tired, she was jumping back and forth, making it miss back and forth’ The form oĩkanãke “you’ll see” or “you’ll get what’s coming” is synchronically unsegmentable and may either be an interjection or a form of older speech that has been preserved in this narrative. The only recognizable part is the verb root oĩ ‘see’. 99 97 18 a-ská-s-maĩnõ=se ka-kĩ=se oĩtá, vitxi DEM:ANA-SIML-DUR-DS/A=DM go:SG-S/A>A.SIM=DM INTJ skin nitxĩ-paké-i. ãwẽ ak-á=ivo, ãwẽ vitxi kene stand.up-DOWNWARD-MCM1 3SG do:TR-O>S/A/O=DET 3SG skin pattern mera-varã-i=se, aa chinã-varã-i find-INWARD:PL/TR-S/A>S.SIM=DM 3SG think-INWARD:PL/TR-S/A>S.SIM o-a-tĩ come:SG-HOD-ANTIMIR ‘(she) doing like this, (he) going along, oĩtá, stood the skin up. When he did that one, when (it)i was finding the skin pattern, coming toward (itj), iti is coming, you know.1 19 a-ská ak-a ã=vitxi ãwẽ ne-ská a-kĩ, DEM:ANA-SIML do.TR-O>S/A 3SG.POSS=skin 3SG DEM:PROX-SIML do.TR-S/A>A.SIM mai tavá-sho, ãwẽ vepẽ-ní-s-maĩnõ=se, ground beside-PA 3SG open-UPWARD-DUR-DS/A=DM veso-ke-ki=se oĩtá, pakã nash pakã turn-COMPL-S/A>A.SIM=EMPH INTJ spear\INSTR IDEO:spear spear\INSTR rete-ki spear-MCM2 ‘when (it) did like this, his skin, doing like this, from beside the ground, when (it) opened (it) up, turning around, oĩtá, he went nash with his spear, with his spear he speared it’ (cf. absolutive paka) 20 a-ská a-vaikĩ kamã tenã-vaiki aĩvo, vake aká, DEM:ANA-SIML do.TR-S/A>A.PREV jaguar kill-S/A>A.PREV woman child and neõ tere-vaĩ-ki, neõ stream.leaves\INSTR wipe-OUTWARD:PL/TR-MCM2 stream.leaves\INSTR tere-vaĩ-i võkíchi-nĩ shaki-vaĩ-i. wipe-OUTWARD:PL/TR-MCM1 plant.species-INSTR rub-OUTWARD:PL/TR-MCM1 a-vaikĩ ewe-varã-ki do.TR-S/A>A.PREV get-INWARD:PL/TR-MCM2 ‘after doing like this, after killing the jaguar, (he) wiped the woman and child with stream leaves. He wiped the child with stream leaves. He rubbed (them) with võkíchi leaves. After doing this, (he) brought (them)’ (cf. absolutive neo; võkíchi) 98 21 ã=raká-ti-nĩ noko-varã-ki, vake atõ 3SG.POSS=be.laying-INSTR.NMLZ-LOC reach-INWARD:PL/TR-S/A>A.SIM child 3PL\ERG ichná ak-á. ã=ewa-nĩ-shõ=se, atõ ichná bad do.TR-MCM2 3SG.POSS=mother-COM-PA=EMPH 3PL\ERG bad ak-á. ichná~ichná=chta yama-i, kamã-ne isõ-ne do.TR-MCM2 INTS~bad=DIM be.nothing-MCM1 jaguar-ERG urine-INSTR voi-a-tõsho ichná~ichná=chta yama-i, splash-O>S/A/O-BECAUSE INTS~bad=DIM be.nothing-MCM1 ‘reaching his dwelling, they chanted for the child. From with his mother, they chanted (for him). He died of seizures, because the jaguar had splashed him with urine. He died of seizures’ (cf. absolutive ato) (lit. ‘do bad’ and ‘really bad’)100 22 a-ská-maĩnõ ãwẽ ewa=ro, roa DEM:ANA-SIML-DS/A 3SG.POSS mother=FOC:PAR good aka-vã-a do.TR-ACROSS:PL\TR-MCM2 ‘happening like this, (its) mother, (they) made (her) well’ (lit. ‘do good’) 23 roa a-shõ=se=tsẽ, ãwẽ vene inã-ó=ma, ã=kamã good do.TR-S/A>S/A=DM=NEXT 3SG.POSS husband give-REP=NEG 3SG=jaguar rete-ya=tõ=se viã-ké-ai-tĩ. spear-CHAR.NMLZ=ERG=FOC:ASS get\MAL-COMPL-HOD-ANTIMIR ‘after making (her) well, (they) didn’t give her back to her husband, the spearing one took her (to his detriment)’ (cf. verb root vi) 24 a-ská a-kĩ yoã=vo nikã-paóa, kamã-ne pi-á=rasĩ. DEM:ANA-SIML do.TR-S/A>A.SIM story=PL hear-PST3.HAB jaguar-ERG eat-NMLZ=PL ‘doing like this, I used to hear the stories, the jaguar eating ones’ 100 The free translations, ‘chanted’ and ‘seizures’ are idiomatic expressions. 99 Appendix C: Pani aká (To make a hammock) This is a procedural account of how to make a traditional hammock from the pani palm (bactris setosa). The narrator describes the challenging conditions in which the palm shaft is extracted, as the pani palm is covered by long thorns and often there are fairly dangerous creatures. Once the shaft is extracted and trimmed, it is beaten into small strands. These are then dried, then hand rolled into string. Then, two poles are put into the ground (corresponding to the preferred length of the hammock), and the string is wrapped around these poles. Once the desired width is reached, the same string is used to weave supports, which run perpendicular to the string wrapped around the poles. Finally, ropes are made with which the hammock is tied up. 1 pani ak-á=ro ne-ská hammock make-NMLZ=FOC:PAR DEM:PROX-SIML ‘to make a hammock is like this’ 2 yomã=ka nokẽ pani ak-á=ro yomã=ka hard=ADJZ 3PL\POSS hammock make-NMLZ=FOC:PAR hard=ADJZ ‘it’s hard, to make our hammock is hard’ 3 ne-ská ak-á DEM:PROX-SIML make-NMLZ ‘to make (it) is like this’ 4 txitá-ki=ro pani mera-i ka-a begin-S/A>A.SIM=FOC:PAR pani.palm find-S/A>S.SIM go-MCM3 ‘beginning, (you) go find a pani palm’ 5 wai sheni matxi mera-i ka-a garden old on.top find-S/A>S.SIM go-MCM3 ‘(you) go find it at the top of an old garden’ 100 6 niĩ aya=se, wai sheni-namã aya=se aká, jungle\LOC exist=FOC:ASS garden old-LOC exist=FOC:ASS and pani shãko pani.palm shaft ‘(you) can find them in the jungle and also in old gardens, pani palm shafts’ 7 a-ská-maĩnõ vi-nõ i-inã ka-kĩ mĩ=oĩ-a DEM:ANA-SIML-DS/A get-INT say-S/A>S/A go-S/A>A.SIM 2SG=see-O>S/A/O {ichná-namã mocha-yãwã-namã nii-namã ichná=ka chinã-tso} bad-LOC thorn-REPL-LOC jungle-LOC bad=ADJZ think-IMP a-ská-sho ni~ni-kawã-ki mĩ=oĩ-a DEM:ANA-SIML-S/A>S/A ITER~be.standing-ACROSS:SG/INTR-S/A>A.SIM 2SG=see-O>S/A/O wetsa another ni-i stand-MCM1 i-o-kaĩ-ki do-REP-OUTWARD:SG/INTR-S/A>A.SIM mĩ=oĩ-a 2SG=see-O>S/A/O wetsa ni-i aká another stand-MCM1 and ‘this happening, going to get (it), you see (it), {in the bad, the very thorny101 place, the jungle is bad you know}, so this happening, walking around, you see (it), another one stands. And going out again you see (it), another one stands’ 8 a-ská=mẽkĩ mocha-yãwã=ka raké=ka {raké} DEM:ANA-SIML=ADVS thorn-REPL=ADJZ fear=ADJZ fear yomãna-ki vi-á-rivi pani shãko struggle-S/A>A.SIM get-MCM3-FOC:CONTRASTIVE pani.palm shaft ‘But, it is thorny, scary, {scare}…struggling, (you) get pani palm shafts’ 9 a-ská-sho ãwẽ txishkẽ ivo=ro {rera-sho vi-i} DEM:ANA-SIML-RESULT 3SG.POSS short.tree DET=FOC:PAR chop-S/A>S/A get-MCM1 rera-i chop-MCM1 ‘So, its short tree one, {cutting you get (it)}…(you) cut (it)’ 10 a-ská a-vaikĩ mocha-yãwã=namã ãwẽ DEM:ANA-SIML do.TR-S/A>A.PREV thorn-REPL=LOC 3SG.POSS pei peré-i leaf chop.floppy.branch-MCM1 ‘after doing that, in the thorny place, (you) chop its branch’ The morpheme -yãwã derives nouns into adjectives (similar to -y in English in the word “bumpy”) and has the meaning “full of”, which is why I have termed it a “repletetive” (thanks goes to Josh Smolders for suggesting this term). Interestingly, it is allowed to co-occur with another adjectivalizer -ka, without any obvious change in meaning. 101 101 11 a-ská a-vaikĩ shate-vaiki vi-i DEM:ANA-SIML do.TR-S/A>A.PREV cut.off-S/A>A.PREV get-MCM1 ‘after doing that, after cutting (it) off, (you) get it’ 12 {ãwẽ,}102 no=rí mai richki-ki toá-ma-shõ vi-i 3SG 1PL=DIR ground hit-S/A>A.SIM be.spread-CAUS-S/A>S/A get-MCM1 ‘{Umm}, we ourselves, hitting the ground, making (it) spread out, get (it) 13 a-ská a-vaikĩ ãwẽ txichkẽ=ma DEM:ANA-SIML do-S/A>A.PREV 3SG.POSS short.tree=NEG maĩ=se tsaó-a ivo=ro, aa=ro weti ground\LOC=FOC:ASS be.sitting-NMLZ DET=FOC:PAR PROX=FOC:PAR hooked.pole a-shõ testa ak-í make-S/A>s/A cut do.TR-MCM1 ‘after doing this, its non-short one, the one that sits on the ground, (you) make a hooked pole and cut it (cf. absolutive mai) 14 a-ská a-kĩ {testa a-shõ} weti a-shõ DEM:ANA-SIML do-S/A>A.SIM cut TR-S/A>S/A hooked.pole make-S/A>S/A tova-varã-sho vi-i, shate-i a-ská ak-á break-INWARD:PL/TR-S/A>S/A get-MCM1 cut-MCM1 DEM:ANA-SIML do-MCM3 ‘Doing this, {after cutting}, after making a hooked pole, breaking it (you) get it, you cut it, (you) do like this. 15 a-ská a-vaikĩ=se shãko sirĩ-i DEM:ANA-SIML do-S/A>A.PREV=DM shaft strip.leaves-MCM1 ‘after doing this, (you) strip the shaft’ 16 a-ská a-vaikĩ=se masõ-ní-i DEM:ANA-SIML do-S/A>A.PREV=DM be.gathered\CAUS-UPWARD-MCM1 ‘after doing this, (you) gather them up’ (cf. verb root maso) 17 a-ská a-vaikĩ=se ã=revo sirĩ-vaikí=se DEM:ANA-SIML do-S/A>A.PREV=DM 3SG.POSS=end strip.leaf-S/A>A.PREV=DM a-nõ, nesha-i DEM:ANA-OBL tie=MCM1 ‘after doing this, after stripping its ends of leaves, with that, (you) tie (it)’ 18 a-ská a-vaikĩ=se vi-varã-a DEM:ANA-SIML do-S/A>A.PREV=DM get-INWARD:PL/TR-MCM3 ya-varã-tĩpá=sevi mosha-yãwã carry.on.shoulder-INWARD:PL/TR-IMPOS=FOC:EXP thorn-REPL ‘after doing this, (you) bring (it), you also can’t carry it on your shoulders, (it) is thorny’ 102 Occasionally, Marubo pronouns are repurposed as filler words, as is the case here. 102 19 a-ská-vaiki kesí-a DEM:ANA-SIML-S/A>A.PREV pull.out.thorns-MCM3 ‘After this happens, (one) pulls out the thorns’ 20 kesí-vaiki=se tachpi-ó-i pull.out.thorns-S/A>A.PREV=DM open.end-REP-MCM1 ‘after pulling out the thorns, (you) also open the ends’ 21 a-ská-vaiki=se seno-i DEM:ANA-SIML-S/A>A.PREV=DM cut-MCM1 ‘after this happens, (you) also cut it, one cuts it’ 22 a-ská a-vaikĩ=se seno-vaiki=se DEM:ANA-SIML do-S/A>A.PREV=DM cut-S/A>A.PREV=DM a-no-sho=se shawa-a DEM:ANA-LOC-PA=DM scrape-MCM3 ‘after doing this, after cutting (it), from there, (you) scrape (it)’ 23 shawa-vaiki=se voshó-i scrape-S/A>A.PREV=DM pull.out.end-MCM1 ‘after scraping (it), (you) pull out the ends’ 24 a-ská a-vaikĩ=se txaka-i DEM:ANA-SIML do-S/A>A.PREV=DM beat-MCM1 ‘after doing this, (you) beat (it)’ 25 a-ská a-vaikĩ=se varĩ ak-í DEM:ANA-SIML do-S/A>A.PREV=DM sun\LOC do-MCM1 ‘after doing this, (you) do it under the sun’ (cf. absolutive vari) 26 ã=eshtá-a maná-vaiki=se maya-i 3SG.POSS=dry-NMLZ wait-S/A>A.PREV=DM make.rope-MCM1 ‘after waiting for its drying, (you) make rope’ 27 maya-vaiki=se {mĩ=oĩ-a} ani=ma oĩ-inã=ro make.rope-S/A>A.PREV=DM 2SG=see-O>S/A/O be.big=not see-S/A>S/A=COND wetsa mĩ=a-vai=tõ=sevi a-kĩ vi-ó-a another 2SG=get-PST1=OBL=FOC:EXP get-S/A>A.SIM get-REP-MCM3 ‘after making rope, {when you see (it)}, when you see that (it) is small, another, doing the same thing that you also did before, (you) get another’ 28 mĩ=a-vai=tõ=sevi a-kĩ wetsa mĩ=mera-ó-a 2SG=get-PST1=OBL=FOC:EXP get-S/A>A.SIM another 2SG=find-REP-MCM3 ‘doing the same thing that you did before, you find another’ 103 29 wetsa mera-sho mĩ=vi-ó-a another find-S/A>S/A 2SG=get-REP-MCM3 ‘finding another, you get it again’ 30 mĩ=a-ská ak-a ãwẽ ãtsa-i dez=namã oito=namã 2SG=DEM:ANA-SIML do-O>S/A/O 3SG be.many-MCM1 ten=LOC eight=LOC wetsã petá=ka keyá=ka ak-a-ya=tõ=ro another\ERG wide=ADJZ tall=ADJZ make-NMLZ-CHAR.NMLZ=ERG=FOC:PAR quinze=namã i-míska, a-ská vi-i fifteen=LOC be-HAB DEM:ANA-SIML get-MCM1 ‘when (you) do (it) like this, (it) becomes lots, ten, eight, another, the wide and long maker, there can be 15, like that, (you) get (it)’ (cf. absolutive wetsa) 31 a-ská a-vaikĩ petá=ka, õpo pani-tió=ka DEM:ANA-SIML do-S/A>A.PREV wide=ADJZ cloth hammock-WIDTH=FOC:EXH ẽ=a-nõ i-inã ak-á 1SG=do-INT say-S/A>S/A make-MCM3 ‘after doing this, “I want to make a wide one, just as wide as a cloth hammock” saying this, (you) make (it)’ 32 a-ská a-vaikĩ wetsã a-ská a-míska, DEM:ANA-SIML do-S/A>A.PREV another\ERG DEM:ANA-SIML make-HAB ãtsa=ka vi-kĩ be.many=ADJZ get-S/A>A.SIM ‘after doing this, others do this, when getting lots’ 33 a-ská a-vaikĩ mĩ=mashte-a DEM:ANA-SIML do-S/A>A.PREV 2SG=finish-MCM3 ‘after doing this, you finish (it)’ 34 ã=mashté-maĩnõ=ro a-ská tore-ó-i 3SG=finish\DETR-DS/A=COND DEM:ANA-SIML unroll-REP-MCM1 ‘If (it) is finished, (it is) like this: (you) unroll (it)’ 35 mĩ=sevi-chná mĩ=shoko a-chĩa ivo tore-ó-i 2SG=coil-PST2 2SG=be.gathered do-PST2 DET unroll-REP-MCM1 ‘the (one you) had coiled, the one (you) had gathered, (you) unroll it 104 36 a-ská a-vaikĩ txĩti DEM:ANA-SIML do-S/A>A.PREV stick nitxĩ-vakĩ-vaiki=se {õpi-a} kake-i stand.up-APART-S/A>A.PREV=DM wrap.around-MCM3 wrap.around.stick-MCM1 ‘after doing this, after (you) stand two sticks up, {(you) wrap (it) around}, (you) wrap it around the sticks’ 37 a-ská a-vaikĩ mĩ=keko-a DEM:ANA-SIML do-S/A>A.PREV 2SG=weave-MCM3 ‘after doing this, (you) weave (it)’ 38 mĩ=a-ská ak-á ã=mashté-kaĩ-maĩnõ=ro 2SG= DEM:ANA-SIML do-MCM3 3SG=finish\DETR-OUTWARD:SG/INTR-DS/A=COND a-ská DEM:ANA-SIML ‘You do (it) like this, if it is finished, (it) is like this’ (cf. verb root mashte) 39 a-no ã=respí ak-í DEM:ANA-LOC 3SG.POSS=hammock.rope make-MCM1 ‘there, (you) make its ropes’ 40 a-ská a-vaikĩ mĩ=narĩ-vakĩ-i DEM:ANA-SIML do-S/A>A.PREV 2SG=put.through.APART-MCM1 ‘after doing this, you put (the rope) through (the ends)’ 41 a-ská a-vaikĩ mĩ=txiwa-a DEM:ANA-SIML do-S/A>A.PREV 2SG=connect-MCM3 ‘after doing this, you tie (it) up’ (lit. connect it) 42 a-no=ro ã=mashtékaĩ-a DEM:ANA-LOC=FOC:PAR 3SG=finish\DETR-OUTWARD:SG/INTR-MCM3 ‘there, it is finished’ 43 a-ská ak-á pani ak-á=ro DEM:ANA-SIML do-MCM3 hammock make-NMLZ=FOC:PAR ‘(You) do like like this, making a hammock’ 44 yomã=ka, mosha raké=ka, rono-yãwã, nivo-yãwã, {manipei be.hard=ADJZ thorn be.afraid=ADJZ snake-REPL scorpion-REPL leaf.species mĩ=teve-vaĩ-a, a-ská=sevi vina-yãwã} 2SG=put.head.under-OUTWARD:PL/TR-NMLZ DEM:ANA-SIML=FOC:EXP wasp-REPL aká yomãna-ki vi-á=rivi pani=ro and struggle-S/A>A.SIM get-MCM1=FOC:CONTRASTIVE hammock=FOC:PAR ‘it is hard, there are scary thorns, there are lots of snakes, there are lots of scorpions, {and the putting of your head under a leaf, it is also like this, there are lots of wasps}, one struggling, gets hammocks’ 105 45 yomã=ka, nokẽ mevĩ me-a, be.hard=ADJZ 1PL\POSS hand\INSTR touch-MCM3 mapõ chinã-sho ak-á=ma, head\INSTR think-S/A>s/A do-NMLZ=NEG ‘it is hard, with our hands (we) touch, it is not with your head thinking while doing’ (cf. absolutive noke; mevi) 46 a-ská=mẽkĩ yomã=ka, pani ak-á DEM:ANA-SIML=ADVS be.hard=ADJZ hammock do-MCM3 ‘But to make a hammock is hard’ Figure 3 Crafting a palm-fiber hammock (photo taken by Melissa Smith, 2016)