Waimiri Atroari grammar: Some phonological, morphological,

and syntactic aspects

Item type text; Dissertation-Reproduction (electronic)

Authors Bruno, Ana Carla

Publisher The University of Arizona.

Rights Copyright is held by the author. Digital access to this

material is made possible by the University Libraries,
University of Arizona. Further transmission, reproduction
or presentation (such as public display or performance) of
protected items is prohibited except with permission of the
author.

Downloaded 30-Nov-2017 22:34:45

Link to item http://hdl.handle.net/10150/280345

 WAIMIRI ATROARI GRAMMAR:

SOME PHONOLOGICAL, MORPHOLOGICAL, AND SYNTACTIC ASPECTS

by

Ana Carla Bruno

Copyright Ana Carla Bruno

A Dissertation Submitted to the Faculty of the

JOINT DEGREE IN ANTHROPOLOGY AND LINGUISTICS

In Partial Fulfillment of the Requirements

For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

2003
 UMI Number: 3106972

Copyright 2004 by
Bruno, Ana Caria Dos Santos

All rights reserved.

UMI
UMI Microform 3106972
Copyright 2004 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company

300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
 2

THE UNIVERSITY OF ARIZONA @

GRADUATE COLLEGE

As members of the Final Examination Committee, we certify that we have

read the dissertation prepared by Ana Car la Bruno

entitled Waimiri Atroari Grammar: some phonological,

morphological and syntactic aspects.

and recommend that it be accepted as fulfilling the dissertation

requirement for the Degree of Doctor of Philosophy

jane Hill

erry Langend

g. 2^03
Ofelia Zep Date

Date

Date

Final approval and acceptance of this dissertation is contingent upon

the candidate's submission of the final copy of the dissertation to the
Graduate College.

I hereby certify that I have read this dissertation prepared under my

direction and recommend that it be accepted as fulfilling the dissertation
requirement.

^ Ajt) 3
Dissertation DirectorWill/Terry' Lange^idoen Date
 3

STATEMENT BY THE AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an

advanced degree at The University of Arizona and is deposited in the University Library
to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission,
provided that accurate acknowledgment of source is made. Requests for permission for
extended quotation from or reproduction of this manuscript in whole or in part may be
granted by the copyright holder.

SIGNED:
 4

ACKNOWLEDGMENTS

Many people have carried this project forward to its completion. I am grateful for
the non-academic and academic support from those here and in Brazil.
My first THANKS go to the Waimiri Atroari people who shared their language
and culture with me. Thanks Warakaxi, Dauna, Ewepe, Damixiri, and Tuwadja for telling
me stories, repeating verbal paradigms, and patiently repeating what I could not hear, and
explaining things that I did not understand.
I would like to thank all whom I met in the Waimiri Atroari Program and who
made my stay in the field unforgettable and my research possible, particularly Parazinho,
Paulino Dantas, Goreti, Kissia Daniela, Trissiana Santos, Sara Moraes, Katia Nunes,
Mara Santos, Carla Yamane, Edith Lacerda, Antonio Carlos, Carmen do Vale, Marcilio
Dias and Porfirio Carvalho.
I also would like to say thanks to all my teachers and classmates from The
University of Arizona. Principally, thank you Heidi and Tania not only for sharing your
knowledge with me, but also for your friendship and support during the difficult
moments.
Thank you, Du, Gessi, and Sergiofrom the Museu Emilio Goeldifor helping
me with your ideas, friendship and support. Du, I will never forget when we shared the
apartment in Belem (laughing and fighting about Linguistics, dancing reggae, listening to
brega, redoing many times drafts of our projects and reports to Dermy, and eating 'unha
de caranguejo' because we did not have money to buy other stuff).
Thanks to my Brazilian friends in Tucson, Alzira and Daniel, Claudio and Sandra,
Lu and Marcal, Ana e Guto for your friendship, psychological and emotional support. I
am very grateful to you for entertaining my husband and taking care of my daughters
when I needed to have peace to study.
Thanks to Aldir, Stellinha, Carla, and Odileiz for helping me with your friendship.
Thank you to my mother who taught me to be strong even in the worst times.
Thank you, God, for giving me strength, serenity, and patience during the most
difficult moments of my life.
And finally THANK you Adair Palacio (my first advisor in Linguistics), Denny
Moore, Ofelia Zepeda, Dr. Langendoen, and Dr. Hill for being the most supportive
advisors that I could ever have hoped for through my graduate career and personal life. I
hope that one day I will be able to do the same good job for someone else as you have
done for me.

The fieldwork was largely funded and backed by the Programa Waimiri Atroari, which
provided transport, food, and accommodation.

This work was financially supported by CNPq scholarship process 200290/97-5.

 5

DEDICATION

To the Waimiri Atroari people

and

To Aldevan, Kaina, and Wina

and

To my mother and sister

 6

TABLE OF CONTENTS

LIST OF FIGURES AND TABLES 9

ABSTRACT 10

CHAPTER 1- INTRODUCTION 12

1. Waimiri Atroari Language and Culture 12

1.1 The people 12
1.2 Historical Background 12
1.3 Cultural and Linguistics Aspects 16
1.4 Linguistics and Anthropological Studies 19
1.5 The Waimiri Atroari and Formal Education 20
1.6 Types of Data 29
1.7 Organization of this dissertation 30

CHAPTER 2 - PHONOLOGY 32

2. Phonology 32
2.1 Segments 32
2.1 Consonants 33
2.1.1 Stops 33
2.1.2 Palatals 34
2.1.3 Fricatives 35
2.1.4 Nasals 36
2.1.5 Glottal 37
2.1.6 Liquid 37
2.1.7 Approximant 38
2.1.8 Contrast 38
2.2 Vowels 39
2.2.1 High vowels 40
2.2.2 Mid vowels 40
2.2.3 Low vowels 41
2.2.4 Vowel Length 41
2.2.5 Contrast 41
2.3 Stress 42
2.4 Syllable Structure 43
2.5 Morphonological Process 45
2.5.1 Vowel replacement 45
2.5.2 Vowel deletion 46
2.5.3 Vowel harmony 47
2.5.4 Reduplication 47
2.6 Theorethical Discussion 48
2.6.1 Syllable structure 48
 7

TABLE OF CONTENTS (CONTINUED)

2.6.1.1 Tableaux section 50

2.6.2 Reduplication 55
2.6.2.1 Tableaux section 60

CHAPTER 3 - MORPHOLOGY AND LEXICON 67

3. Morphology and Lexicon 67

3.1 Waimiri Atroari Parts of Speech 67
3.1.1 Nouns 68
3.1.1.1 Possession 69
3.1.1.1.2 Relational Morpheme 71
3.1.1.2 Derivational Morphemes 73
3.1.1.3 Pronouns 76
3.1.1.4 The Non-third person pronouns 77
3.1.1.5 Third-person pronouns 80
3.1.2 Verbs 84
3.1.2.1Tense/Aspect suffixes 87
3.1.2.2 Mood 90
3.1.2.2.1 Imperatives 90
3.1.2.2.2 The negation suffix 92
3.1.2.3 Interrogative clitic 92
3.1.2.3.1 Interrogative forms 93
3.1.2.4 Causative forms 99
3.1.2.4.1 Construction with the 'made' reading 100
3.1.2.4.2 Construction with the 'let reading' 102
3.1.2.5 Miscellaneous Suffixes 103
3.1.2.5.1 Desiderative suffix 103
3.1.2.5.2 Instrumental Nominalizer 104
3.1.3 Adjectives 104
3.1.3.1 List of adjectives 105
3.1.4 Adverbs 109
3.1.5 Postpositions 110
3.1.6 Particles 113
3.1.6.1 The Evidential ka 114
3.1.6.2 The ram particle 115
3.1.6.3 Negation particle 115
3.1.6.4 The conjuction particle many 115
3.2 Case Marking in Waimiri Atroari 116
3.2.1 Background 116
3.2.2 Gildea's classification of morphosyntatic properties of the verbal system 118
3.2.2.1 The ditransitive construction 124
3.2.3 The agentivity particle 127
 8

TABLE OF CONTENTS (CONTINUED)

3.2.4 Final remarks 131

CHAPTER 4 - SYNTAX 133

4. Phrase structure, clauses, and word order in Waimiri Atroari 133

4.1 Phrase structure in Waimiri Atroari 134
4.1.1 Noun phrases 135
4.1.1.1 Noun phrases containing adjectives 136
4.1.1.2 Noun phrases containing quantifier words 137
4.1.1.3 Noun phrases containing numerals 140
4.1.2 Verb phrases 142
4.1.3 Postpositional phrases 143
4.2 Waimiri Atroari clausal order 145
4.3 Topicalization 149
4.3.1 OVS order (movement of the whole VP) 149
4.3.2 OSV order (movement of the object) 151
4.4 Adverbial subordinated clauses 153
4.4.1 Temporal subordinated clauses 154
4.4.2 Locative subordinated clauses 155
4.5 Nominalized clauses 157

ABBREVIATIONS 159

APPENDICES 160
A1 Tyiyry ikaa 160
A2 Kyrywaky weriri ikaa 161
A3 lanana ikaa 163
A4 Kaapy tahkome karykapa 168
A5 Meie ikaa 172
A6 Wyty ikaa 173
B Kara mo pyky wenpatypy 175
C More verbal paradigms 176
D List of animals and plants found in this sketch with their scientific names 178

REFERENCES 180
 9

LIST OF ILLUSTRATIONS AND TABLES

Figure 1 Waimiri Atroari area map, from Do Vale (2002) 13

Table 1 Some dialectal variation 18

Table 1.1 Schwade's orthography 21
Table 1.2 The difference between Schwade's and Hill's orthography 22
Table 1.3 Meeting to discuss the orthography and linguistic aspects 23
Table 1.4 Differences among the old orthographies and the utilized today 26
Table 1.5 Current orthography utilized in the literacy project 27

Table 2 Waimiri Atroari consonant distinctive segment 32

Table 2.1 Some minimal pairs focusing on consonants 39
Table 2.2 Waimiri Atroari vowel distinctive segments 39
Table 2.3 Minimal pairs and analogous pairs focusing on vowels 42
Table 2.4 The distribution of the phonemes in the syllable 43

Table 3.1 Nouns and Verbs Characteristics 68

Table 3.2 Lexical differences in men and women's speech 68
Table 3.3 Some examples of Waimiri Atroari nouns 69
Table 3.4 Possessive Markers 70
Table 3.5 Possessor-possessed paradigms 70
Table 3.6 Waimiri Atroari pronominal system 77
Table 3.7 Structure of Waimiri Atroari verbs 85
Table 3.8 Illustration of verbal morphology 86
Table 3.9 Illustration of Tense/Aspect suffixes 86
Table 3.10 Person Marking clitics and prefixes 87
Table 3.11 Person marking examples 88
Table 3.12 Activities versus non-activities 121
Table 3.13 Verb -ini- 'see' 123
Table 3.14 Verb -ry- 'give' 126

Table 4.1 Frequency of occurrence of each clausal order 147

 10

ABSTRACT

The Waimiri Atroari people, who call themselves kinja 'people' and whose

language belongs to the Carib family, live today in an area in the northern part of the

State of Amazonas and in the southern part of the State of Roraima. Like many other

languages of the Carib family, Waimiri Atroari is a chronically underdescribed language.

There are few linguistics studies about Waimiri Atroari, most of them being phonological

sketches (Hill and Hill 1985; and Lacerda 1991, 1996).

Taking this situation into consideration, this dissertation intends to describe some

phonological, morphological, and syntactic aspects of the Waimiri Atroari grammar.

First, in the introductory chapter I provide some information about their language and

culture, and I discuss their experience with formal education. Second, I describe the

segmental phonology and analyze the syllable structure and reduplication process under

Optimality Theory. Next, I present the word classes and a description of their

morphology. Then, I investigate the system of case marking. Finally, in syntax I analyze

the phrase structure and the word order under the framework of X-bar theory. The

appendices contain a set of verbal paradigms and a collection of texts.

 11

When the data-base is incomplete, as it always is in real science, waiting for

"enough" data means waiting forever. Although I still have to understand and explain

many aspects of the Waimiri Atroari language, I did not want to wait forever to write this

grammatical sketch.
 12

CHAPTER 1INTRODUCTION

1. Waimiri Atroari Language and Culture

1.1 The People

The Waimiri Atroari people, who call themselves kinja 'people' and whose

language belongs to the Carib family, live today in an area in the northern part of the

State of Amazonas and in the southern part of the State of Roraima (see Figure 1 below).

They form 19 villages, located along the Camanau/Curiau, Alalaii, Jauaperi, and Rio

Branquinho rivers. The total population is 970 individuals (PWA, Waimiri Atroari

Program, April 2003).

1.2 Historical Background

The Waimiri Atroari people have suffered a long history of violent and

disrespectful invasion in their territory, related to variation in prices of forest products in

the international market, such as the hides of forest animals, wood, Brazilian nuts, and

latex. According to Barbosa Rodrigues, the first contacts between the Waimiri Atroari

(whom he called Crichands) and non-indigenous Brazilians date from the 1P''' century in

the lower Rio Negro River (Rodrigues 1885:10). These contacts were not welcomed by

the Waimiri Atroari people. In the early 1900s, Alipio Bandeira^who worked for the

Indian Protection Servicetried to establish new contacts with the Waimiri Atroari

people, but he was not successful, either.

Until the latter half of the 20'^ century, invasions undertaken by local populations

and those organized by local governments were sporadic. However, toward the end of
 13

the 1960s, the Federal government began massive occupation of the Waimiri Atroari

territory through the establishment of regional development projects. Figure 1 shows the

land of the Waimiri Atroari people.

Mapa da terra Waimiri Atroari

Vai tiara
Bd.'l Ui-qt;} ^ y /
/y A Kisiwc
/ .A V
J,.,. J ' \ f
\
A^ anahkwa'
ANAHKWA ' ! 1 '
! A ; /
i<. .^AIKON .
'r"% a"" a\ Rorat-J^a
WvrI ALALAU
a- f
-"WfiyslnUij
. A-J i " '"Will.UkJ r-illuiH -
' \

Anuui-'iigs MVNAWA ' /\ Legeiida

> A AW.ia
'iKARATYMA-,,1i
\t*., karl-bSYNAa\\
. \ ., 2 e^ I ilglomerados
,
' ^; . I / \\. I Gon^etido OS gi'upos
/ np'^>^4 Vaipar^ locfjisaftns
^MAUMA..^^ARD(I

LOCALIZAgAO DA TERRA WA NO MAPA DO BRASIL

Terra
VVriimifi Al'oiiri"--.. Sf? c

AM
B R AS:( L

Figure 1Waimiri Atroari Land^ (do Vale 2002)

In the early 1970s, the BR-174 highway linking Manaus (Amazonas State) to Boa

Vista (Roraima State) was pushed through the middle of the Waimiri Atroari territory

' In this map, each triangle is a village. This big line crossing circle 01 and circle 02 is the BR-174 road that
links Manaus (Amazonas State) to Boa Vista (Roraima State). As mentioned in Section 1.3, each circle is
analyzed as a speech community (Gumperz 1962)
(Bruno 1999). The construction of this road marked, for the Waimiri Atroari people, "the

beginning of a period of escalating contact with the outside world." (Milleken and Miller

1992:07).

According to Porfirio Carvalho (personal communication, 1998), in 1974 the

Waimiri Atroari population was 1,500. Because of the construction of the road and

resulting innumerable health problems brought to the community, in 1986 the population

was reduced to 374. Moreover, in 1974, a project to identify all mineral resources of

Amazonia (Radam-Brasil) was established. Then, the Parapanema mine company, based

on the results of this project, solicited permission to mine in the Pitinga River. The

solicitation was approved with FUNAFs (National Foundation of the Indiansthe

Brazilian Bureau of Amerindian Affairs) endorsement. Legally, this was not possible,

because the land it was located on was within the bounds of the Waimiri Atroari

reservation. Since 1981, this mining company has extracted minerals in that region.

Another historic event central to the current situation of the Waimiri Atroari

community is the construction of the Balbina hydroelectric plant. In the 1980s, one

hydroelectric plant was built to supply the city of Manaus with power, flooding part of

the Waimiri Atroari reservation. Consequently, two villages were relocated to different

locales, causing political problems to the extracted groups (Silva 1993). Moreover, as a

result of this construction, in 1987, a contract was established between FUNAI and the

Eletronorte (the company responsible for providing electricity in the northern part of

Brazil). This contract stated that the company would be responsible for providing

education, health, and environmental consultation to the Waimiri Atroari villages for the
next 25 years, in return for being given permission to build a hydroelectric plant on the

Waimiri Atroari reservation. As a result, the Programa Waimiri Atroari (The Waimiri

Atroari Program) was created, and FUNAI was designated to administer the money

provided by Eletronorte.

The Waimiri Atroari Program strives to minimize the direct and indirect impact of

the changes in the ecosystem of the Waimiri Atroari reservation. It has as a principal

goal to prepare the Waimiri Atroari people to become self-sufficient through helping

them to develop the necessary skills to manage their economic and environmental

resources and to make political decisions important for their cultural survival. Many

Brazilian anthropologists do not agree with the political actions developed by the

Waimiri Atroari Program. They classify the actions of the program as paternalistic and

oppressive. Baines (1992, 1994, 1997, 1999), an anthropologist who did fieldwork with

the Waimiri Atroari people from 1982 to 1985, classifies the indigenism developed by the

program as "business indigenism- indigenismo empresariaT'. He claims that the program

utilizes an ecological discourse of "environmental worries" and intensive marketing

campaigns as a model of an "alternative indigenousness," capable of providing services

to the indigenous communities and at the same time establishing large projects in

indigenous territories.

Baines (1994:27) claims that in spite of the image of the Waimiri Atroari people

constructed by the media, in fact "it makes up the asymmetric relations established by an

indigenist administration that excludes the natives of the decisive process of planning and

evaluation, but incorporates them in its politic marketing." I agree that the attitudes and
behaviors of some of the Waimiri Atroari today reflect the interethnic experiences

explored by Baines (1991). However, the situation in the Waimiri Atroari area today is

very different. Baines did fieldwork in this community when the area was administered

by the FAWA (Waimiri Atroari Attraction Front 1970-1987). FAWA was constituted by

the old group of FUNAI employees, who utilized the traditional policies which

encouraged the dependence of indigenous people on State resources and patronage; this is

a different historical and political situation than the current one. Since the time Baines

was expelled from the area (1984), he has been unable to come back. Having done

fieldwork more recently (1995-1998, 2000), I observed that the Waimiri Atroari people

have been active in the decision-making process of planning and evaluating the program,

and in choosing what they think is best for them.

1.3 Cultural and Linguistic Aspects

Among ethnographic studies regarding the Carib family, it is possible to

distinguish three Carib "types:" 1) Northwest Amazon groups; 2) a second group located

basically at the "Guiana area;" and 3) a third group found in the Upper Xingu Basin.

Using the criteria for classification established in Carib Speaking Indian: Culture,

Society, and Language (Basso 1977), the Waimiri Atroari people are best identified as

members of the second Carib group.

They live in single round houses, forming a communal family household that

constitutes a political group as well as a kin group (the whole village lives in a single

round house). The men move in with wife's natal family. The villages are usually small,

averaging about 20-40 kinja. For sustenance, they depend on bitter manioc, which is
utilized to make manioc flour with which to prepare large tortillas on a ceramic griddle.

Unlike other peoples of this second Carib group, the Waimiri Atroari people do not

prepare a manioc-fermented beverage; moreover, they practice shamanism without the

use of hallucinogenic drugs. The shamans not only have the power to act as healers, but

also they name children and have knowledge of the sacred and divine.

The Waimiri Atroari people still live essentially according to their traditional

ways. They inhabit remote villages and subsist by slash and bum horticulture, hunting,

fishing, and collecting forest fruits. In their gardens, they plant mainly bitter manioc (to

make manioc flour and tortillas), banana, sugar cane, pineapple, and different species of

sweet potato. The villages are located close to small and big rivers, each village has

political and economic autonomy, and there is no centralized power. The distribution of

work among men and women is distinct, but sometimes they can work together. Men

hunt, prepare the gardens in order to plant together with women, and make basketry and

artifacts for fishing and hunting, such as arrows and bows. The women take care of the

domestic space, but they also collect fruits and can fish. They make hammocks and the

aruma^ bracelets.

Like the majority of Carib groups of that region, the Waimiri Atroari group has a

strong tendency toward local group endogamy, uxorilocality, and small local groups.

According to Silva (1995:212), they present a system of relational categories of the

Dravidian type, based on the distinction between "consanguineal group" and "affinal

ties", from genealogical level, age, sex, and lineage. He also explains that

^ Bracelets make with a kind of native liana (vine) called aruma {Ischnosiphon arouma.Marantaceae).
 18

The ideal marriage for a man is with a weriky 'cousin' category that
includes an ambilateral cross-cousin. It is possible for other kinds of
marriage, such as of a man with his pasky 'sister's daughter', or his
imekyky 'brother's daughter' ... The single men can have sexual access to
their brother's wives... According to the Waimiri Atroari's perception of
conception, a child will have as many fathers as possible depending on the
number of sexual partners that his/her mother had in the time of
conception. As a result, a person can have more than one apapa
'biological father.' Silva 1995:216
Linguistically, using Gildea's classification (1998), which is based on

morphosyntactic properties of each verbal system, the Waimiri Atroari language is in the

set I system (nominative or inverse/split-S). When compared to other Amazonian groups,

the Waimiri Atroari language has not suffered dramatic changes. All members of the

community (children, adults, and elders) speak the language. Generally, only the male

leaders of the villages and the native teachers speak Portuguese, and occasionally some of

the women understand, but full fluency in Portuguese is lacking.

As mentioned by do Vale (2002:17), today the Waimiri Atroari people constitute

a unique ethnic group. They speak a language which presents some phonetic and lexical

variation, as shown in Table 1.

Table 1Some dialectal variation

Waimiri Atroari
(circles 2 and 3 in Figure 1) (circle 1 in Figure 1)
Clay wonji wudji
Jaguar tymeri temere
Type of leaf sopia ia sypia ia
Sun woi wyie
I escaped, ran away aa wytymaky aa wytarybaky
I came back aa warampia aa warempia
 19

In the Waimiri Atroari area (Figure 1 above), it is possible to identify three speech

communities. A speech community is "a social group that shares a set of verbal signs"

(Gumperz 1962). The first speech community can be observed in the region of Alalau

River; all the villages in this area share the same verbal signs in speaking the Atroari

dialect. The second speech community is constituted by what I call the Samauma group

(the Mynawa, Karebi Syna, Pardo, and Arine villages), and the third is composed of the

Curiaii, Cacau, and Mare. The second and third speech communities both speak the

Waimiri dialect. Silva (1993:44) noted that among this aglomerado 'cluster', as he calls

it, the linguistic unity among the groups is mentioned more often than the disparities.

Interestingly, each speech community makes fun of "the way of speaking" of the others.

1.4 Linguistic and Anthropological Studies

Until recently, little linguistic work had been done to document this language

family. Although today there is an increase in linguistic studies related to the Carib

languages (Meira 1999, Gildea 1998, Campetela 1997, Pacheco 1997, Hall 1988, Hill and

Hill 1985, Tavares 1993, Payne 1990, Derbyshire 1985, 1999, and Hoff 1986), these

studies are restricted to linguistic aspects, such as phonology, morphology, and syntax,

and do not address the interaction of language and culture.

Like the majority of languages in the Carib family, there are few linguistic studies

of the Waimiri Atroari language, most of them being phonological sketches (Hill and Hill

1985, Lacerda 1991, Bruno 1995, 1999, 2000, and 2002). Taking this situation into

consideration, this grammatical sketch provides a first attempt to describe different

aspects of the Waimiri Atroari language, including phonological, morphological and

syntactic features.

Since Basso's edited collection (1977), few ethnographic studies have been done.

Studies relating to the languages and cultures of the Carib family are limited (Basso

1985, 1987, 1995; Francheto 1986; Kloos 1971; Baines 1991; and Mattel Muller 1991).

Recently, it is possible to find reduced but decent anthropological descriptions relating to

the Waimiri Atroari people (Baines 1999, 1997, 1994, 1993, 1992, 1991; Bandeira 1926;

Barbosa Rodrigues 1885; Carvalho 1982; Monte 1992; Milleken et al 1992; Silva 1993,

1995; Do Vale 1994, 2002; and Espindola 1995). Baines's work explores the interethnic

experience among the FUNAJ employees and the Waimiri Atroari people; Silva's work

discusses the kinship system, and Do Vale (2002) has a very interesting ethnographic

description of the ceremonial system.

1.5 The Waimiri Atroari and Formal Education

Until recently, the Waimiri Atroari language was not written. In 1985, one village

had contact with a national group of Catholic missionaries known as CIMI (Indigenous

Missionary Council). During this time, a missionary coupleEgydio and Dorothy

Schwadeestablished a phonological proposal and developed an alphabet, based on their

linguistic studies. They taught about twenty village men to read. Later, the couple was

refused access by FUNAI, but the village people wanted to continue to study.
 21

Table 1.1Schwade's orthography

Phoneme Orthographic Symbol

/p/ P
/t/ t
/tj/ tx
/k/ k
/b/ b
/d/ d
/d3/ dj
/m/ m
/n/ n
/n/ n
/s/ s
X
/J/
/h/ h
/r/ r
/w/ w
/y/ y to represent the glide
/i/ i
Id e
N i
/a/ a
/u/ u
/o/ 0
They did not register long vowels.

In 1986, another missionary couple from MEVA (Evangelical Mission of the

Amazonian)Joseph and Tamara Hill^who had worked and done linguistic research

with the Wai-Wai people (another Carib group), tried to continue to work in the same

village. They also did a preliminary description of the phonology and established another

orthography that diverged from the Schwade's orthography in the following aspects (see

Table 1.2 below). Later on, this couple was expelled as well.
 22

Table 1.2The difference between Schwade's and Hill's orthography

Schwade's orthography Hill's orthography

/JT/ ii n
/y/ y to represent the glide J
/i/ i y
They did not register long vowels They registered long vowels as VV

In 1988, with the contract established between FUNAI and Eletronorte, the

Waimiri Atroari people decided that any educational experience from now on would be

realized for all the villages (no more restricting education to one village). In 1987, Marcio

Silva, a researcher from Museu NacionalRio de Janeiro, started to do fieldwork in the

Camanau/Curiaii area. He wanted to collect data about the kinship system, and as a

condition for his research the Waimiri Atroari asked him to teach them to read and write.

Silva utilized Schwade's orthography in his classes. From 1988 to 1993, FUNAI hired

non-native teachers to go to the villages to teach the Waimiri Atroari people their

language, Portuguese, and mathematics.

In May 1988, Joao das Letras, responsible for the educational sub-program,

analyzed the different orthographic systems and chose particular aspects of each of them.

In 1989, the linguist Ruth Monserrat started to provide linguistic advice to the project.

Under her supervision and with the help of Edith Lacerda (the teacher) and Edilberto

Fonseca (the coordinator of educational sub-program), the educational group (the non-

native teacher together with the linguist) started to have meetings to discuss linguistic

aspects to help them in the literacy project.

^ Lacerda, having a natural talent, developed by herself some linguistic research describing many aspects of
the Waimiri Atroari language. I learned a lot from her.
 23

Table 1.3Meetings to discuss the orthography and linguistic aspects

November 1989 March 1990 November 1990

Do not register long vowels.

Represent the phoneme /y/,

registered as /j/, by the letter
/i/. As a result, maja
become maia 'knife.'

Eliminate the letter /w/ used The glottal marked before

after or before /u/. Paruwe the vowel would be
become Parue 'proper marked after the vowel.
name.' mar'e become mare' (sp.
of bird).
Do not use closed syllables; Syllable reduction would be It is possible to have CV,
instead of am.pa write selected only for some VC, CCV, CVC, and
amypa 'other.' words; as a result, amypa CCVC.
would be ampa.

The problem with this system was that the Waimiri Atroari did not attend the

discussions. Although the non-native teachers discussed the issues with Waimiri Atroari

students in the villages, in the meetings with the linguist no kinja were there. Moreover,

Monserrat was very busy giving linguistic advice to different indigenous groups.

Consequently, she did not develop any profound linguistic study of the language. This

system generated a climate of uncertainty among the community in relation to the way of

writing the language. Besides, the non-native teachers did not stay very long in the

villages. Then, every time that a village got a new teacher, they needed time to leam the

language and adapt to the village life style (no electricity, no running water, and two

months in the jungle in order to get eight days of vacation in the city).'^

'' My first contact with the Waimiri Atroari people happened in 1991. At that time, I also was a non-native
teacher. I worked with them until 1993. At this time, I did not have any intention of doing linguistic
 In April 1992, the non-native teacher of Mynawa village became the new

educational coordinator of the project. Together with the community, she decided to not

change anything in the orthography until someone could provide a close and permanent

linguistic study. In 1994, the Waimiri Atroari community decided to invest in forming

native teachers inasmuch as the non-native teachers had problems with the Waimiri

Atroari language and did not stay very long in the villages. At the end of 1994, I

proposed to the community to do linguistic research with them. Tired of linguistic and

anthropological research, they only accepted me because I had lived with them before.

In 1995, the educational sub-program and the community decided that some

modifications to the orthography would be made. At that time, people from different

villages used different orthographies; even in the same groups, people wrote sometimes

with a different orthography than the one they were exposed to. The Waimiri Atroari

teachers wanted to produce literacy material, but they were worried that these materials

would not take into consideration dialectal differences or would not work at all because

of the different ways that the language was being written.

Then the community, the educational group and I decided to rethink the

orthography. I went to different villages and asked many questions, such as what they

thought would change or not change in the orthography; and, from the different

orthographies, what aspects we could continue to use. I did not want to make drastic

changes because many kinja already knew how to write. Later on, the educational sub-

research. However, I had the opportunity to work in a village that had the tradition of producing a lot of
mythological stories and different kinds of texts, continuing the work established by the teacher, Carlos
Augusto Queiroz, who had needed to leave the village because of health problems.
program organized a big meeting on the reservation where I and 35 Waimiri Atroari

people selected by the community discussed and made the necessary changes, such as

covering the dialectal differences, marking the long vowels because the non-native

teachers could not hear them; the kinja described things were missing when writing some

words. Non-native people did not hear the differences between kymy 'bacaba berry' and

kymyy 'hand', for example, and the orthography needed to change to record this

distinction.

I believe that any orthography should represent the sounds of the language in

question, not only being phonologically well based, but also taking into consideration the

speakers' beliefs and perspectives. An idea of what is going on in the language and how

the sounds can be represented must be given by linguists, explaining the consequences,

but only the speakers must decide. The idea that indigenous people 'are not able to

understand linguistics' needs to be discouraged (see in the appendix material elaborated

by the kinja teachers about phonology). Besides, language is not only an instrument of

communication, language carries symbolic value that conditions social, political, and

economic spheres.
 26

Table 1.4Differences between the old orthographies and the one utilized today

Schwade Hill and Joao das Letras and Ruth Current

Silva Monserrat Orthography
/ji/- [n] /ji/-[n] /ji/ [nj] /y - [nj]
To represent To represent the glide [j]
the glide [y]
m - [i] M - [y] /i/ - [y] /i/ - [y]
Did not Did register They began marking long Long vowels are
register long long vowels. vowels, but later on decided not registered.
vowels. to mark them because the non-
native teachers could not hear
them.
The other problem was related to In this current
syllable structure: words that orthography, we just
typically have a CV pattern were made some
sometimes written with a CVC adjustments that took
pattern. For example, [ma.ry.ba] into consideration the
and [te.ty.ky] became [mar.ba] perceptions and
and [tet.ky]. requests of the kinja.
The words with a CV
pattern are still
marked in that way,
having slow speech
as the referent.
We did not change
any symbols from the
previous orthography.

In fact, I would have preferred to use [y] to represent the glide sound, but the

community decided to continue denoting it as in Hill's orthography. Today, it is

registered as [i]. If they used [j] to register the glide, it could make things more confusing

because the letter [j] in Portuguese is used in a different way. Because the Waimiri

Atroari schools are bilingual, I think we need to keep this in mind when elaborating

orthographies.
 27

Table 1.5Current orthography, utilized in the literacy project

Phoneme Orthographic Symbol

/p/ P
/t/ t
/tl/ tx
/k/ k
/b/ b
/d/ d
/dj/ dj
/m/ m
/n/ n
/-P/ nj
/s/ s
X
/J/
/h/ h
M r
/w/ w
III to represent the glide
/i/ i
Id e
/i/ y
/a/ a
/u/ u
lol o
Long vowels are marked as VV aa, yy, ee, oo

The new role and figure of being a teacher brought some changes and problems to

the routine of the community. One of the problems was the power struggle among the

traditional leaders and the native teachers regarding changing the notion of leadership.

The first changes in the concept of leadership were situated during the opening

(construction) of the highway. According to Baines (1991), the leadership conception in

the Waimiri Atroari culture was changed by FUNAI's intervention in the area. He

explains that in order to establish contact with the community, FUNAI "created" a system
of capitaes 'captains' (military view) for each village. The capitaes were selected by their

ability (in that time extremely limited) to speak Portuguese, lack of resistance to the

construction of the road, and easy manipulability. To sustain this manipulative tactic,

FUNAI "offered" some privileges to the capitaes, such as food, clothes, radios, lanterns,

and batteries, and allowed them to live with FUNAI employees.

As a consequence of this new role'captains'the traditional system of

authority and leadership was modified. In order to sustain this change, not only the

FUNAI employees, but also the new leadersthe captainsexplained that the elders (the

traditional leaders) died with the contact; therefore, they needed new leaders. However,

Baines mentions that not all of the elders died; in reality, the "elders that survived were

deprived of their authority by FUNAFs workers and the Waimiri Atroari captains"

(Baines 1991:282).

The other problem is related to how the Waimiri Atroari teachers can reconcile

both types of knowledge; the problem that they are having is that they do not yet know

how to find the time for both forms of knowledge. Usually, they say that;

It is hard to find time to hunt, fish, leam how to make basketry, learn

traditional songs, and at the same time to prepare classes, study, and attend

meetings and conferences. Group of Waimiri Atroari teachers. 1998

Today, there are two native teachers in each of the seventeen schools. The school

classes average from five to fifteen students. Generally, the women have classes separate

from the men, but there are a few exceptions, where men and women study together.

Students range in age from eight to forty. Currently, the younger children are not
formally enrolled to attend on a regular basis (however, they are welcomed at the schools

to do some activities). The curriculum includes ethnoscience and ethnomathematics

(taking the Waimiri Atroari conceptions and knowledge of the world to study Science

and Math), Geography, Waimiri Atroari language, and Portuguese.

Due to the fact that there is currently 100% use of the language, the Waimiri

Atroari community has not developed a concern to sustain language use. On the other

hand, the Waimiri Atroari teachers are very concerned about the elaboration of material

for use in the school setting to assure use and retention of the language and culture.

Therefore, in 1997, the educational program and I decided to improve the list of words

organized by Lacerda and Queiroz (1991), and create an illustrated dictionary that could

be used as literacy material.

1.6 Types of Data

This description is based on data collected through fieldwork with native speakers

of the language. Waimiri Atroari consultants of four different villages and two speech

communities ranging in age from fifteen to fifty years helped me to obtain specific

information on phonological, morphological, and syntactic aspects of their language.

Three consultants provided intensive consultation, such as WarakaxiJose Maria,

EwepeMarcelo, and DamixiriRenato. The elicitation was done in three stages. The

first one occurred when I was a teacher in the Alalau and Paryry villages in 1991. As I

did not have the intention of conducting professional linguistic research, I just collected

some verbal paradigms and mythological stories with Dauna (Shaman and storyteller of

the Alalau village). In the second stage (1995-1998), I collected more focused
 30

morphological and syntactic structures in order to understand parts of speech, case

marking, causatives, and word order. Natural texts in the form of stories and m34:hs were

tape recorded and then transcribed with the assistance of the Waimiri Atroari consultants.

Discussion of the grammaticality of some materials was divided into categories; for

example, (1) acceptable grammatically, (2) understandable, and judgments of different

consultants from different villages were also checked.

The third stage was in 2000, when I concentrated my attention on reduplication

and syllable structure. I also utilized this time to make some revisions to the dictionary

material.

1.7 Organization of this dissertation

The present work is organized as follows. Chapter 1 discusses some historical and

cultural aspects of the Waimiri Atroari people. It also provides a discussion of the

Waimiri Atroari orthographic system. Chapter 2 presents a structural description of the

segmental units in Waimiri Atroari and analyzes some aspects such as syllable structure

and reduplication using Optimality Theory.

Chapter 3 introduces the morphological units, defining and describing the

morphology of each lexical class. It also provides an analysis of case marking in this

language. Chapter 4 describes the syntax of phrases, clauses, and simple and some

complex sentences. It analyzes the possible types of word order in this language and

provides a brief discussion of topicalization.

The description follows a notational convention:

brackets ([]) - 'surface or phonetic representation';

 31

slashes (//) - 'phonemic representation';

colons (:) - vowel length;

dots (.) - syllable boundaries;

(=) - clitic boundaries

dashes (-) - morpheme boundaries

(?) - interrogative symbol in the glosses line means that I do not know the

meaning;

(*) - incorrect or non-attested form

In the grammatical sketch, interlinear glosses follow a three-line system: original

utterance in the first line, glosses and morphemic analysis in the second line, and a free

translation in the third line.

 32

CHAPTER 2 - PHONOLOGY

2. Phonology

This chapter is organized as follows. Sections 2.1 and 2.2 present a structural

description of the segmental units in Waimiri Atroari. Section 2.3 analyzes the stress

pattern in this language. Section 2.4 provides a description of the syllable structure.

Section 2.5 explores some morphophonological processes such as vowel replacement,

vowel deletion, vowel harmony, and reduplication. Finally, section 2.6 analyzes some

aspects such as syllable structure and reduplication using Optimality Theory.

The Waimiri Atroari phonemic inventory includes seventeen consonant phonemes

and twelve vowel phonemes, seven short and five long. The consonant phonemes are /p/,

Pol, /t/, /d/, /k/, /?/, /s/, /J/, /h/, /tj"/, /d3/, /m/, /n/, /jV, /r/, /w/, /j/, and the vowel phonemes

are III, Id, Id, /a/, /i/, /u/, /o/, /e:/, /a:/, /e;/, /i:/, /o:/.

2.1 Segments

Table 2.0 and Table 2.2 show all 29 distinctive segments of Waimiri Atroari

represented together with the orthographic symbols used in the literacy project (see

discussion in section 1.5) and throughout this grammar sketch.

Table 2.0Waimiri Atroari consonant distinctive segments

Bilabial Apico- Palatal Dorso-Velar Glottal
Alveolar
Occlusive pb td tl(tx) d3(dj) k ?C)

Fricative s I(x) h

Nasal m n Ji(nj)

Liquid r

Approximant w j(i)
 33

2.1 Consonants

2.1.1 Stops

Phonemes /p/, III, fk! are voiceless occlusives at bilabial, apico-alveolar, and

dorso-velar points of articulation. They can occur initially and intervocalically. Phonemes

/b/ and /d/ are voiced occlusives at bilabial and apico-alveolar points of articulation. They

can only occur in the beginning of the word and intervocalically. Waimiri Atroari does

not allow stops in coda position or at the end of the word.^ Note that /p/ and /k/ appear in

cluster /pr/, /kr/, discussed in 2.1.6.

(l)/p/ {2)1X1

py.ru.wa [pi.ru.wa]'arrow' te.ty.ky [te.ti.ki]'Brazilian nut'

pa.pa [pa.pa] 'father' ty.ru.wa [ti.ru.wa] 'ceramic pan'
pa.tu.wa [pa.tu.wa] 'pataua berry' wa.ty [wa.ti] 'fire'
wa.ra.pi.nji [wa.ra.pi.jii]'proper name' ma.ty.ty [ma.ti.ti]'kind of basket'
py.ry.ry [pi.ri.ri] 'frog' ta.ra.ra [ta.ra.ra] 'thunder'

(3)/k/

ka.nu.wa [ka.nu.wa] 'canoe'

sam.ka [sam.ka] 'hammock'
xi.ri.ki.ki [fi.ri.ki.ki] 'Sp of parakeet'
kaa.pa [ka:.pa] 'plantation field'

(4) /b/ (5) /d/

bi.ky [bi.ki] 'son' (female speech) ma.da.na [ma.da.na] 'lier'

ma.ba [ma.ba] 'macaw' de.he,de.de [de.he.de.he] 'cough'
ba.ky.ma [ba.ki.ma] 'salt' my.dy [mi.di] 'house'
ka.ba.ha [ka.ba.ha]'Sp. of armadillo' da.nja [da.jia]'cicada'

' The only exception is for the negation particle kap.

 34

Sometimes /p/ occurs in free variation with the voiced stop /b/ and the nasal /m/.

Like the phoneme /p/, the phoneme /b/ also occurs in variation with the nasal /m/.

Although the examples below show this free variation, /p/, /b/, and /m/ are distinctive

segments as shown in Table 2.1, which shows minimal or near minimal pairs.

(6) [p]~[b] (7) [p]~[m]

kampa~kamba 'Let's go' kampa~kama 'Let's go'

akybyhy~akymyhy 'porridge' panana-manana 'bumblebee'

(8) [b]~[m]

babyka~bamyka 'wide'
benry-menry 'design, drawing'

The examples in (6) and (7) show that Waimiri Atroari does have nasal

assimilation: for example, impa 'then, after.' However, it is also possible to find

examples where nasal assimilation does not occur, such as nytynpa 'he/she went' and

sakenpa 'angry'. The cases where nasal assimilation does not occur involve a morpheme

boundary, for example ny-tyn-pa or saken-pa.

2.1.2 Palatals

The phonemes /tx/, and /dj/ are respectively voiceless and voiced phonemic

occlusive affricates. In Waimiri Atroari, the sequence of these palatalized segments is

perceived as a single unit. They can occur initially and intervocalically.

(9) /tx/ (10)/dj/

txa.my.ry [tja.mi.ri] 'elder, old' dja.ky.ny [d3a.ki.ni] 'proper name'

txi.dji [tji.d3i] 'type of basket' wu.dji [wu.d3i] 'clay'
i.txi [i.tji] 'forest' ka.dji.wi [ka.d3i.wi] 'worm'
 35

Although /dj/ occurs more frequently before /i/ than /d/, this does not mean that

/d/+/i/ always results in the palatalization of /d/. See the examples in (11)-(13) where /d/

remains before /i/.

(11) di.mi.na ' proper name'

(12) dia [di.ja] 'species of tree'
(13) di.re.he 'scale'

2.1.3 Fricatives

The phonemes /s/, /x/, and /h/ are voiceless fricatives at apico-alveolar,

alveopalatal, and glottal points of articulation, respectively. They can occur in the

beginning and in the middle of the word. In this group, /s/ and /h/ can appear in word-

medial coda position, never at the end of the word.

(14) /s/ (15)/x/

se.he [se.he] 'tall' xi.ba [[i.ba] 'fish'

py.kwa.se [pi.kwa.se] 'to shoot arrows' a.wa.xi[a.wa.Ji] 'sugarcane'
ias.ka [jas.ka] 'relatives' xeri Lfe.ri] 'skate fish'

(16)/h/

hi.ri [hi.ri]'Cajun fruit'

tah.ko.me [tah.ko.me] 'elder man'
tyhnaka [tih.na.ka] 'over, above, on'
wa.ha [wa.ha] 'many, much'

In some words, it is possible to find free variation among [x]~[h], [s]~[h], and

[s]~[x], such as baxinja~bahinja 'child, small,' anahkwa~anaskwa 'native fruit,' and

syhy-xyhy 'body hair'.

Interestingly, Waimiri Atroari presents phonological peculiarities uncommon in

the Carib family, the most striking of these being the occurrence of a consonantal system
 36

which includes five segments produced with friction (three fricatives, /x/, fhl, and /si, and

two affricates, /tx/ and /dj/). The existence of both the glottal fricative /h/ and the glottal

stop /'/ is a rare phenomenon in the family. According to Gildea (1995), these consonants

originated from a historical process of syllabic reduction: in some languages, this process

results in a stop; in other languages, the result is a Ixicative. In Waimiri Atroari, though,

both outcomes occur.

2.1.4 Nasals

The phonemes /ml, /n/ and /nj/ are nasals produced at bilabial, apico-alveolar,

palatal points of articulation, respectively. They can occur in the beginning and middle of

the word. The phonemes im/ and /n/ can occur in the coda position. However, they do not

appear at the end of the word. The only exceptions are found with the second position

^ Phonological inventory of other Carib languages

Waiwai Tiriyo Hixkaryana Waimiri/A Ikpeng
Conso p($) t k p t k p t k p t k ? p t k
nants b d b d b g
m n ji m n m n ji m n ji m n 11,
s h m sJ h tJ
sJ h r sJ h
tJ w y d3tj
dldy) tJ 1
r r r r
r r w y
w y w y w y
Vowels
i i u i i u i u i no u i i u
e 0 e 9 0 e o(o) e(:) o(:) e 0
a a a e(:) a
a(:)
 37

particle ram and the word takrehen 'delay, linger.'^ Although the phoneme /nj/ usually

appears following and preceding /i/, it is also found in other environments.

(17)/m/ (18)/n/
a.ba.ma [a.ba.ma] 'blind' na.na [na.na] 'pineapple'
ma.ma [ma.ma]'mother' ia.na.na [ja.na.na]'mythological pers'
sam.ka [sam.ka] 'hanrniock' ny.tyn.pa [ni.tin.pa] 'she/he went'

(19)/nj/

nja.wa [pa.wa] 'rain'

xi.nja Lfi-Jia] 'flute'
wo'.nji [wo?.jii] 'clay'

2.1.5 Glottal

The occlusive glottal /'/ occurs between vowels and between a vowel and a

consonant. When a glottal appears in an intervocalic envirormient, the two vowels are

always identical. It can appear in the middle of the word, and in the end of the word as

the coda of a final syllable.

(20) /V'V/ (21) /V'C/

ta.be'e [ta.be?e] 'capybara' no'.sa [no?.sa]'she/he sleeps'

a'a [a?a]'1+3 PRO' wo'.nji [wo?.jii] 'clay'

(22) /CV'/ (23) /CCV'/

i.re' [i.re?] 'how' a.kra' [a.kra?] 'Sp. offish'

2.1.6 Liquid

The phoneme /r/ is the only liquid segment in this language. It is a flap with some

lateral release. In Waimiri Atroari, with the exception of the second particle position ram.

^ In the text 'kaapy tahkome karykepa' in the appendix, it is possible to see that the underlying
representation of takrehen is takrehene; therefore, this coda at the end of the word is not common.
 38

there are no words beginning with /r/. Moreover, /r/ can form clusters /kr/ and /pr/ in

CCV syllables.

(24)
wa.ra.ra [wa.ra.ra] 'turtle' te.me.re [te.me.re] 'jaguar'
py.ry.ry [pi.ri.ri]'frog' pa.na.ry [pa.na.ri]'long time ago'
ma.kry.kry [ma.kri.kri] 'pernilong' kra.txi.ni [kra.tji.ni] 'proper name
we.ri [we.ri] 'woman' me.pry [me.pri] 'tapir'

2.1.7 Approximant

Phonemes /w/ and /y/ are a bilabial glide and a palatal glide, respectively. Hill

(1985) calls them ambivalent segments (consonantal or vocalic). In this grammatical

sketch, /w/ and /j/^ are interpreted as a consonant in initial position of the syllable, but in

the nucleus the same segments are interpreted as the vowels /u/ and I'll.

(25) /w/ (26) /j/

kwa.ta [kwa.ta] 'spider monkey' ia.na.na [ja.na.na] 'mythological person'

ka.wa [ka.wa] 'menstruation' ma.ba.ia [ma.ba.ja] 'papaya'
wo.ky [wo.ki] 'banana' iee [jee] 'tooth'

2.1.8 Contrast

This section provides examples that show contrast among the consonant

phonemes in minimal and analogous set.

^ Today, in the orthography used in the literacy project, the community does not use [y] to represent the
glide. For the glide, they use jj] as established by Hill and Hill (1986). Now /i/ is used both in onset and
nuclear position, such as hi.ri 'cajun fruit' and ma.ba.ia [ma.ba.ya] 'papaya.' In order to address the
community request and not do radical changes, we used the symbol [y] for the /i/.
 39

Table 2.1Some minimal and near-minimal pairs focussing on consonants

/p/ peri door

/b/ beri feather crown
/p/ papa father
/m/ mama mother
/b/ bixi peel, skin
/m/ mixi a kind of fruit from a palm
/b/ bahinja small, children
/w/ wahinja proper name
/p/ pyty wife
/w/ wyty meat
/t/ waty fire
/d/ sady crab
/d/ sadada shrimp
/r/ tarara thunder
/x/ xeri skate (species of fish)
/s/ seri proper name
/tx/ itxi jungle
/dj/ txidji kind of basket
Iml mama mother
InJ nana pineapple
InJ sana termite
/nj/ sanja manioc flour
/nj/ njawa rain
/j/ jawa come here!

2.2 Vowels

Table 2.2Waimiri Atroari vowel distinctive segments

Front Central Back

High i i(y) i:(yy) u

Mid e e:(ee) o;(oo)

Low e(e) a a; (aa)

 40

2.2.1 High Vowels

From this group of high vowels, only the /i/ high front vowel and the /i/ high

central unrounded vowel can occur in the beginning of the word as a V syllable. On the

other hand, only the high central unrounded /i/ can be lengthened. The high round vowel

IvJ has a very restricted distribution. It only occurs preceding or following bilabials.

(27) III (28) /i/

i.txi [i.tji] 'forest' ky.my [kimi] 'bacaba berry'

hi.ri [hi.ri] 'cajun fruit' py.ry.ry [pi.ri.ri] 'frog'
wi.we [wi.wi] 'wood' sy.na [si.na] 'water'
i.na.xi.xi [i.na.j"i.Xi] 'bat' y.ry.se [i.ri.se] 'to do'

(29) IvJ

su.we.ri [su.we.ri] 'deer' (animal)

ty.ru.wa [ti.ru.wa] 'ceramic pan'
wu.se [wu.se] 'to kill'
wu.re [wu.re] 'fan'
hu.myn.ta.py.pia [hu.min.ta.pi.pi.ja] 'We made him bleed'
hu.pa.kah.py.pia [hu.pa.kah.pi.pi.ja] 'We made him wake up'

2.1.5.2 Mid Vowels

In this group, both vowels can be lengthened. The middle, front unround vowel Id

ranges from closed to open: /e/ occurs in closed syllables, and is glottalized or lengthened

in an open syllable followed by a consonant. On the other hand, the middle back rounded

vowel /o/ does not allow for this possibility.

(30) /e/ (31)/e/

me.re.py [me.re.pi] 'palm fruit' iake [ya.ke] 'alligator'

we.ri [we.ri] 'woman' ma.re'e [ma.re?e] 'guam'(bird)
 41

yee [ye:] 'tooth' ke.pa [ke.pa] 'she/he said'

be.he [be.he] 'ceremonial slash' pa.ru.we [pa.ru.we] 'proper name'

(32) /o/ and /o;/

tah.ko.me [tah.ko.me] 'elder man, antique man'

wo.ky [wo.ki] 'banana'
poo.po [po-po] 'moth'

2.2.3 Low vowels

In Waimiri Atroari, /a /is a low, central, open unrounded segment.

(33) /a/ and /a:/

a.ma.na [a.ma.na] 'Amazonian dolphin'

a'A [a?a] '1+3PRO'
kaa.pa [ka;pa] 'plantation garden'
kwa.da [kwa.da] 'bad, ugly'

2.2.4 Vowel length

In Waimiri Atroari, vowel length occurs only in open syllables. There are no

sequences of non-identical vowels. As demonstrated below in (34), long vowels are

considered phonemes contrasting with short vowels; they are found among nouns, verbs,

possessives, and object prefixes marking 1' and 2"^* person.

(34)

i.myy [i.mii] 'his/her hand' woo.ky [wo:.ki] 'climb!'

kymy [ki.mi] 'bacaba berry' wo.ky [wo.ki] 'banana'
aa.ma.ma [a:ma.ma] 'my mother' aa.bi.ky [ai.bi.ki] 'my son'(female)
a.ma.ma [a.ma.ma] 'your mother a.bi.ky [a.bi.ki] 'your son'(female)

2.2.5 Contrast

This section provides examples that show contrast among the vowel phonemes in

a minimal analogous set.

 42

Table 2.3 Some minimal and near-minimal pairs focussing on vowels

/o/ woky 'banana'

/u/ wuky 'species of bird'
IvJ wuky 'specie of bird'
N wi'ky 'a fruit'
/o/ noosa 'she/he climbs'
/y/ nysa 'she/he goes'
/o/ woky 'banana'
/o:/ wooky 'climb!'
/y/ kymy 'bacaba berry'
lyj kymyy 'hand'
lyl wyty 'meat'
Id wety 'feces'
lyl iky 'species of ant'
laJ ikaa 'history'
/a/ kapy 'anus'
/a:/ kaapy 'sky'
/e/ iee [yee] 'tooth'
Izl ie [ye] 'no!'
/e:/ peepe 'paper, butterfly'
/o:/ poopo 'moth'

2.3 Stress

In Waimiri Atroari, stress is not a contrastive feature. In bisyllabic words, stress

falls on the final position syllable. Words with more than two syllables may possibly

have primary and secondary stress. The length of the vowel influences the stress pattern.

Syllables with long vowels and glottalized vowels are stressed.

(1) kaa.pa 'garden' (2) ka.pa 'she/he talked'

(3) my.dy 'house' (4) vm.re 'fan'

(5) ta.be'e 'capybara' (6) ty.ry.ny 'any small bird'

There are a lot of questions that I was not able to answer in relation to the stress

system in Waimiri Atroari. I need to do additional research.

 43

2.4 Syllable Structure in Waimiri Atroari

In Waimiri Atroari, each syllable has a nucleus constituted by a vocalic phoneme.

The syllable can also include a consonant element in onset position and another in final

position as a coda. Onsetless syllables (V, VV, VC) only occur at the begirming of the

word. Any consonant may appear in syllable-initial position, but only nasals /m/ /n/,

fricatives /s/, and glottals /V /h/ may appear in coda position. When CC.C occurs, the

second C can only be either the liquid M or the approximant /w/.

Table 2.4The distribution of the phonemes in the syllables

Onset Coda
Left edge of the intervocalically Left edge and Right
word middle of word edge of
the word
/p/ + + - -

/b/ + + - -

/t/ + + - -

/d/ + + - -

/k/ + + - -

/'/ + +
/h/ + + + -

/s/ + + + -

/x/ + + - -

/tx/ + + - -

/dj/ + + - -

/m/ + + + -

/n/ + + + -

/nj/ + + - -

/r/ - + - -

/w/ + + - -

/y/ + + -

In Waimiri Atroari, eight different syllable patterns are observed: V, VV, VC,

CV, CVV, CVC, CCV, and CCVC.

 44

(35) CV syllables occur word-initially, word-medially, and word-finally.

wa.ra.ra 'turtle' ma.ty.ty. 'kind of basketry' sy. na 'water'

cv. cv.cv cv.cv.cv cv.cv

nia.ba.ia 'papaya' ma.ba 'macaw' ta.ra.ra 'thunder'

cv.cv.cv cv.cv cv.cv.cv

py.ru. wa 'arrow' ka.nu.wa 'canoe' te.ty.ky'Brazilian nut'

cv.cv.cv cv.cv.cv cv.cv.cv

V, VV, and VC syllables occur only at the leftmost edge of the word. In the V

syllables, all the vowels can fill this slot, except the high, back unrounded /u/. On the

other hand, VV syllables only appear as the prefix that marks 1' possessive and first

object.

(36) /V/

a.ma.na 'Amazonian dolphin' i.txi 'forest, jungle' i.myy 'somebody's hand'

v.cv.cv v.cv v.cvv

e.ty.pa 'pain, sickness' e.py.ry.ry 'flower' a.ma.kra 'species of fish'

v.cv.cv v.cv.cv.cv v.cv.ccv

(37) N N !

aa.my.dy 'my house' aa.sam.ka 'my hammock' aa.py.ty 'my wife'

vv.cv.cv w.cvc.cv vv.cv.cv

(38) INa

ixn.ka 'if im.pa 'then, after' am.pa 'other, another'

vc.cv vc.cv vc.cv

CVC, CCV, and CVV syllables, like CV syllables, can occur in different

environments. CVC and CCV syllables can occur initially, medially, and finally;

however, CVV syllables do not occur in medial position.

 45

(39) /CVC/

tah.ko.me 'elders' ny.tyn.pa 'he/she went' no'.sa 'she/he sleeps'

cvc.cv.cv cv.cvc.cv cvc.cv

sam.ka 'hammock' ias.ka 'relatives' pas.ky 'niece'

cvc.cv cvc.cv cvc.cv

i.re' 'how' ram 'second position particle'kap 'negation particle'

v.cvc cvc cvc

(40) /CCV/

kwa.da 'ugly, bad' sa.kra 'white' ma.kry.kry 'mosquito'

ccv.cv cv.ccv cv.ccv.ccv

(41)/CVV/

kaa.pa 'field' ky.myy 'our hands' hee.ia [he:ya] 'I drink'

cw.cv cv.cvv cw.cv

CCVC syllables occur only in medial position, as a result of morphophonological

conditions. In Waimiri Atroari, there is a morpheme hkypahpa 'after something' that

appears at the end of the verb. Then, when syllabifying the /h/, it will fill the coda

position.

(42) /CCVC/

ny.bi.xi.kwah.pa 'after being injured' ki.ri.kwah.ky.pa 'after being burned'

cv.cv.cv.ccvc.cv cv.cv.ccvc.cv.cv

w^.ka.prym.pa 'wood that serves as support for the house'

cv.cv.ccvc.cv

2.5 Morphophonological Processes

2.5.1 Vowel Replacement

2.5.1.1 In Waimiri Atroari, word-final vowels /e/, /y/, and III are replaced by /a/ when

immediately followed by the second position particle.

 46

(43)

weri ram [wera ram nysa itxi taka]

woman 2PART woman-2PART go jungle AL
'The woman goes to the jungle"

wykyry ram [wykyra ram]

man 2PART

2.5.1.2 In interrogative phrases, the final vowel of the last word is usually replaced by a

glottalized /eV.

(44)

temere m-itxiky-piany. [temere m-itxiky-pian=e']

jaguar 2A-shoot arrows-REC.P jaguar 2A-shoot arrows-REC.P-INT
"You shot arrows at the jaguar" "Did you shoot the jaguar?"

2.5.2 Vowel Deletion

In texts and also in normal speech, between two morphological words, the word-

final vowels of the first word delete, and almost always after [h], one of the consonants

that is allowed in the coda position.

(45)

[maraha taka]>-[marah taka] 'to the old field'

[waha paky]->[wahpaky] 'many'

[waha pary]->[wahpary] 'many'

[behe ke]->[behke] 'with slash'

Moreover, when adding the suffixes valuative -e 'me and the devaluative -erne the
last vowel of the word is deleted.
wykyry 'man'^ wykyr-eme
weri 'woman'> wer-eme
b a h i n j a 'c h i l d b a h i n j - e m e
 47

2.5.3 Vowel harmony

2.5.3.1 When the phoneme /y/ precedes Ci, /y/ is replaced by /i/.

(46)

ny-txi-piany [ni-txi-piany] 'she/he went'

3-gO-REC.PAST

2.5.3.2 Vowel harmony also occurs between the verb stem and the suffix marking

tense/aspect

-synehk- disappear -eka- 'defecate'

wy-synehka-pa 'I disappeared' w-ek^-pia

wy-synehk^-pianx 'I disappeared'
wy-synehka-tape 'I will disappear'

2.5.4 Reduplication

In Waimiri Atroari, reduplication is applied to verbs, generally for the semantic

meaning of repetition (to do many times) or continuation (to keep doing). Waimiri

Atroari has a case of partial reduplication: the reduplicant copies part of a segment of the

verb stem. However, I observed that reduplication is conditioned by the number of

moras. The reduplicant has to be bimoraic.

Before describing the process of reduplication in Waimiri Atroari, it is important

to give the possible verb stem shape in this language, inasmuch as the reduplicant copies

part of the verb stem.

(47) Verb Stem Shape Example

a) V -y- 'go'
b) VV -ee- 'drink'
c) VC -ym- 'fall, drop'
d) VCV -ini- 'see'
e) VCVCV -irima- 'rest'
 48

f) vcvcvcv -emenyky- 'forget'

g) CV -ry- 'give'
h) CCVCV -niety- 'write'
i) CVCCV -wenta- 'throw up'
j) cvcvcv -iakyby- 'make porridge'
k) CVCVCCV -synehky- 'disappear'

When the process of reduplication in Waimiri Atroari occurs, the process of

reduplication applies to the left edge of the verb stem.

2.6 Theoretical Discussion

2.6.1 Syllable Structure

Under the OT approach, Waimiri Atroari violates different syllabic constraints

(Prince & Smolensky 1993). For example, the patterns V, VC, and VV violate the onset

constraint by which syllables prefer to begin with consonants. However, these patterns

only occur at the left edge of the word. Although in this language the only VV syllable is

found in the prefix that marks 1^' possessive or 1' object, I will consider it to be a syllable

because this pattern undergoes the same constraints established for VC and V syllables.

Another interesting aspect related to this pattern is the sequence of VV and V syllables;

for example, when i.myy ' somebody's hand' gets the possessive marker, it will be

syllabified aa.i.myy 'my hand.' The two syllables do not violate left-edge alignment.

Moreover, as Kager (1999:91) mentions, the syllable ftinctions in the demarcation of

morpheme edges.

On the other hand, although Waimiri Atroari allows syllables without onsets in

the leftmost syllable of the word, this is not true for the reduplicant, which is unmarked

relative to the language as a whole. Waimiri Atroari has a typical case of emergence of

the unmarked structure (McCarthy & Prince 1994) when selecting the shape of the
reduplicant (RED); as a result RED must have an onset, thus it must be unmarked (Bruno

2000).

Moreover, we can observe that the CCV and CCVC syllables also violate another

constraint related to onset: * Complex Onset. Above in the syllable structure section, I

explained that CCVC syllables occur only in word-medial position. One of the reasons

for this must be related to morphological conditions. In Waimiri Atroari there is a

morpheme -hkypa~hpa 'after something,' which appears at the end of the verb. As a result,

when syllabifying, the /h/ will be the coda, contributing to the structure of the syllable.

On the other hand, the patterns VC, CVC, and CCVC violate the constraint, which states

that codas are not allowed. In Waimiri Atroari, codas are allowed; however, they can

only be nasal, fricative, or glottal and cannot occur at the end of the word, as was

explained above in the section on the phonological inventory.

According to the observations established above, these constraints were selected

to account for the data set in section 2.4: Align onsetless syllable left, *Complex Onset,

Coda condition (Nasal/Fricative/Glottal), * Coda/Align/Right, DEP-IO, MAX-IO to cover

Parse and Fill as suggested by Kager (1999), and MAX-BR to cover the emergence of the

unmarked syllable in the reduplicant case. I decided to use the align-onset left instead of

just onset because Kager (1999:110) has a very interesting discussion about the "naive

notion that syllables must have onset except word-initially." He argues that redundancy

can be avoided by using the alignment constraint. Below I provide the definition of each

constraint.

Align onsetless SYL/Left - Syllables without onset must be aligned to the left
edge of the word;
 50

*Complex Onset - Onset are simple;

Coda Condition (Nasal/Fricative/Glottal) - Codas can only be nasals, fricative, or
glottal;
* Coda/Align/Right - Codas cannot occur at the rightmost edge of the word;
DEP I-O - Every element in the output has a correspondent in the input;
MAX I-O - Every element in the input has a correspondent in the output;
MAX B-R - Every element in the base must have a correspondent in the
reduplicant.

2.6.1.1 Tableaux Section

In this section, I provide some tableaux to illustrate the generalizations and

constraints established above.
Tableau 1

Candidate MAX I-O DEP I-O Align *Coda/Align/

Onsetless/Syl/Left
/amana/ Right
'Amazonian
dolphin'
a) am.a. na *!

b)'a.ma.na *!

*
c) a.man *!

d) a.ma.na

Waimiri Atroari does not allow insertion or deletion of segments; consequently,

candidates (b) and (c) violate DEP I-O and MAX I-O, respectively. Candidate (a) does

not violate any faithfulness constraints; however, it does violate the constraint align

onsetless/Syl/Left. Moreover, candidate (a) violates a cross-linguistic observation in

which syllabifications which create onsets are favored over syllabifications which create

codas. Therefore, candidate (d) is the optimal candidate.

 51

Tableau 2

Candidate MAX DEP I-O Align *Complex

I-O onset
/makryry/ Onsetless/Syl/Left
'mosquito'
a) ma.ky.ry.kry *! *

**
b) ma.kry.kry

c) ma.kry.y.kry *! **
*!

As Waimiri Atroari allows syllables with complex onsets, this constraint will

always be violated. Candidates (a) and (c) violate the constraint DEP I-O; however,

candidate (c) is the worst candidate because it violates the align onsetless/Syl/Left

constraint, as well. As a result, candidate (b) is the winner because it only violates

*complex onset.

Tableau 3

Candidate MAX DEP I-O Align *Cotnplex

/amakra/ 'species of I-O Onsetless/Syl/Left onset
fish'
a) am.a. kra *! *

b) a.ma.kra

c) a.ma ky.ra *!

As demonstrated in Tableau 1, in Tableau 3 candidate (a) cannot be the optimal

candidate because it not only violates the alignment constraint, it also contradicts the

generalization in which onset syllabication is preferred over coda syllabification.

Candidate (c) is the worst candidate because it violates the faithfulness constraint, DEP

lO. Therefore, candidate (b) is the wirmer.

 52

Tableau 4
Candidate MAX DEP Coda Align *Complex
/nybixikwahpa/ I-O I-O Condition Onsetless/Syl/ onset
'after injured' (Nasal/ Left
Fricative/
Glottal)
*
a) nyb.i.xi.kwah.pa *!

b) nyb.xi.kwah.pa *! *!

c) ny.bi.xi.kwah.pa

In Tableau 4, the worst candidate is candidate (b) because it violates the

faithfulness constraint, MAX I-O, and the coda condition, as well. Candidate (a) does not

violate the undominated constraints, but it does violate the coda condition and the align

onsetless/Syl/left constraint. Consequently, candidate (c) is the winner.

Tableau 5

Candidate MAX I-O DEP Coda Condition Align

/no'sa/ I-O (Nasal/ Onsetless/Syl/Left
'He/she sleeps' Fricative/Glottal)
*
a) no'.o.sa *!

b) no.sa *!

c) no'.sa

In Tableau 5, all candidates satisfy the coda condition constraint. Candidates (a)

and (b) cannot be the optimal candidates because they violate the faithfulness constraints

MAX 10 and DEP 10. However, candidate (a) is the worst candidate because it violates

the alignment constraint. Therefore, candidate (c) is the winner.

 53

Tableau 6

Candidate MAX DEP Coda Align MAX BR

/RED-n-arvma-pa/ I-O I-O Condition Onsetless
'they come back many Nas/glottal/ Syl/Left
times' Fricative
a) a.ry.-na.ry.ma.pa *!

b) ar. v.-na.ry.ma.pa * *

c) ma.rv.-na.ry.ma.pa *! - * -

*
d) na.ry-na.ry.ma.pa

Because in Waimiri Atroari, the leftmost edge of the word can have an onsetless

syllable, we would assume that the reduplicant shape in Tableau 6 could have an

onsetless structure, inasmuch as it was aligned to the left edge. However, in this

language, RED behaves differently: it must have an onset. To account for the emergence

of this unmarked syllable, I ranked the MAX-BR constraint below the markedness

constraint, as suggested by McCarthy and Prince (1994). Candidate (c) has an onset but

an extraneous segment is epenthesized, violating DEP-IO. Candidate (d) obeys the

constraint MAX-BR and does not violate the undominated faithfulness constraints,

winning the competition.

 54

Tableau 7

Candidate MAX DEP Coda Align MAX BR

/RED-w-oo-sa I-O I-O Condition Onsetless
'I climb many times' Nas/glottal/ Syl/Left
Fricative
a) oo-wo.o.sa *!

b) ow-woo.sa *! *
SSlSiJIIIMiSiliiS
c) kwoo-woo.sa *!
WMMSSiMMUM
d) woo-woo.sa

In tableau 7, candidate (a) is out of the competition because it violates the

alignment constraint and the MAX-BR constraint, established to account for the

reduplicant case. Candidate (b) cannot compete either because it violates the markedness

constraint, coda condition. Candidate (d) violates one of the undominated constraints,

DEP-IO. As a result, candidate (d) is the optimal candidate because it does not violate the

faithfulness constraints, and obeys the emergence of the unmarked structure, where the

reduplicant must have an onset.

In this language, well-formedness is not a central issue inasmuch as we cannot

delete (deletion is not permitted) or insert (epenthesis is not allowed either) segments to

avoid an ill-formed syllable. Therefore, I assume that faithfulness constraints prevent any

attempt to create 'perfect syllables.' Therefore, MAX-IO and DEP-IO are ranked above

markedness constraints. However, this is not true in the case of the reduplicant structure

(being unmarked relative to the language as a whole) inasmuch as it requires an onset.

 55

2.6.2 Reduplication

Reduplication in this language is a prefixing process. In Waimiri Atroari, the

reduplicant must be bimoraic. It copies the first two moras of the left edge of the verb

stem and includes the prefix (marking person) if the verb stem starts with a vowel. This

inclusion of the prefix is not relevant for the bimoraic pattern inasmuch as the simple

consonantal onsets are prosodically irrelevant. In no case was the temporal suffix taken

into account. Therefore, I consider the base only the verb stem as demonstrated in

examples (48) and (49) below.

(48) (49)

[ r n y r y jRmyrykas -ky [w e n ] R w e n t a B-pa

Unlike Tiriyo (see examples 50 and 51), another Cariban language described by

Meira (1999:98), which lengthens the vowel to substitute for the coda, Waimiri Atroari

does not fail to include the coda of the first syllable in the reduplicated segment.

(50) w-enpa-e wec-wenpae T am teaching'

(51) n-ahta-n n^a-nahtan 'It is budding'

The coda is counted as the second mora; therefore, we must assume that the

number of moras determines the shape of the reduplicant. This modification has not only

made the analysis more elegant, but it has also unified the pattern.

The reduplication process in Waimiri Atroari avoids an onsetless RED segment.

Therefore, when the verb stem begins with a vowel, the prefix marking person has to

count as an onset. For example, in the verb stem -aryma-, if we have an input as n- aryma

-pa, it is not possible to reduplicate it as *ary-narymapa 'They come back many times or
 56

repeatedly.' The prefix must become part of the reduplicant; as a result, the reduplication

occurs as narv-narymapa.

(52)
Problem: Onsetless RED Solution: Prefix as part of RED

*ary-narymapa nary-narymapa

However, this example creates two problems. The first one is related to the

alignment of the reduplicant. In Waimiri Atroari, RED has to be aligned to the left edge

of the verb stem n-aryma-pa; however, in this case RED was aligned before the prefix n-

that marks person in this language, so as a result we have nary(RED)-n-aryma (Stem)-

pa. The second problem is related to the need for an onset. Why does the prefix need to

serve as the onset? Mester (1988:202) observed that in Chumash, a Southern California

language, if the stem starts with a vowel, the reduplicant would include any consonant

immediately preceding the stem because the RED would mainly have the invariant CVC

shape, which he called overapplication of onset formation in reduplicant structure.

However, this is not the case in Waimiri Atroari because the only condition required for

the reduplicant is that it must be bimoraic.

As a result, the only coherent explanation for this case must be based on the

emergence of the unmarked approach (McCarthy & Prince 1994). This approach claims

that unmarked elements (syllables, affixes, and structures) are cross-linguistically

preferred more over marked elements (Archangeli 2000; Kager 1999). Although Waimiri

Atroari allows onsetless syllables, I must assume that the structure of the reduplicant is

unmarked relative to the language as a whole, so it must have an onset. However, this
 57

only happens in the cases of onsetless stems because this does not work for such marked

syllables as CVC and CVV.

According to McCarthy and Prince (1997), reduplication is a matter of identity:

"the reduplicant copies the base." However, they claim that perfect identity cannot

always be attained. Under this assumption, it is necessary to utilize the notion of

correspondence, which is a relation between two structures, such as base reduplicant (B-

R) or input and output (I-O). Correspondence theory assumes that given two strings Si

and S2, correspondence is a relation 9? from the elements of Si to those of S2.

Consequently, it is the existence of a correspondence relation that makes an appropriate

environment for reduplication.

In my first attempt (Bruno 2000) to analyze the phenomenon of reduplication in

Waimiri Atroari, I claimed that the weight of the syllable determined the shape of the

reduplicant. As a result, Waimiri Atroari would have two patterns of reduplication (one

for light syllables and the other for complex syllables);

(53) When the first two syllables are light, the reduplicant copies the two initial
syllables of the verb stem CVCV (disyllabic reduplicant) and prefixes this material to the
base.
a)/iakyby-ia/ iaky-iakvby-ia 'Make porridge many times.'
cv.cv.cv-cv cv.cv
b) /myryka-ky/ myry-myryka-kv 'Mix well many times.'
cv.cv.cv-cv cv.cy
c) /n-aryme-pa/ nary-naryme-pa 'They came back many times.'
cv.cv.cv-cv cv.cv

(54) When the first syllable is heavy, it copies the initial heavy syllable sequence of
the verb stem (CVC CVV, VV, and CCV) and prefixes this material to the base.
 58

a) /nu-wenta-pa/ nu-wen-wenta-pa 'They threw up many times.'

cv-cvc.cv-cv cvc
b) /ny-mynta-pa/ nv-myn-mvnta-pa 'They keep bleeding.'
cv-cvc.cv-cv cvc
c) /jaa-pa/ iaa-iaa-pa 'Took many times.'
cvv-cv cvv
d) /w-oo-ky/ woo-woo-ky 'Climb, go up repeatedly.'
cvv-cv cvv
e) /njeti-ki/ nie-niety-ky 'They keep writing or write many times.'
ccv.cv-cv CCV

However, the problem with this analysis was that I needed to classify the CCV

syllable traditionally considered as a light syllable as heavy because it was behaving

according to the heavy pattern. According to Hayes (1989, 1995), the distinction between

light and heavy syllables is based on moraic structure. He claims that a heavy syllable

contains two moras; however, a light syllable contains only one mora. On the other hand,

Hayes (1989:224) explains that moraic theory posits language-specific prosodic

structures, which vary according to a language's criterion of syllable weight. As a result,

the moraic structure of languages can vary (McCarthy & Prince 1986). In the

phonological inventory of Waimiri Atroari, I classify the palatal glide phoneme /j/ as an

approximant consonant. Consequently, in the examples (53a) iakv-iakybv-ia 'Make

porridge many times' and (54c) iaa-iaa-pa 'Took many times,' if the phoneme /j/ is not

taken as a consonant, I would have a trimoraic RED. Therefore, to solve this problem I

have decided to factor the complex onset into the mora calculation. However, as this is

the only example that I have at this moment, I consider this aspect a remaining issue to be
 59

studied. Thus, disregarding my previous syllable-weight analysis, I decided to analyze the

reduplication phenomenon based on the number of moras.^

In analyzing Waimiri Atroari reduplication under this notion of moras, in the

domain of base-reduplicant identity, the completeness of mapping is partial. The

reduplicant is a prefixing process and it normally preserves the linear order of elements of

the base, and the copied segments in the base and reduplicant are identical in features.

Based on these assumptions, some correspondence constraints and well-formedness

constraints were selected for explaining the reduplication phenomenon in Waimiri

Atroari under the correspondence approach.

(55) Well-formedness constraints

(1) RED=Bimoraic - The reduplicant has to be bimoraic.

(2) ONSET/RED - Reduplicants must have an onset.

(3) Align RED L, Stem, L - The reduplicant has to be aligned with the left edge
of the stem.

(56) Correspondence constraints

(a) Linearity - SI is consistent with the precedence structure of S2 and vice-versa.
(b) DEP B-R - Every element in the reduplicant has a correspondent in the base.

(c) MAX B-R - Every element of the base has a correspondent in the reduplicant.

(d) IDENT B-R - Reduplicant correspondent of a base [aF] segment is also [aF].

As demonstrated in the analysis, a special property of this language is related to

the emergence of the unmarked: 1) the onset is required for the reduplicant, but not for

the language as a whole.

' Originally, Keiichiro Suzuki and Tania Granadillo suggested explaining my data based on the notion of
the foot; however, after reviewing my fieldwork notes and my first analysis, I decided to use moras.
 60

2.6.2.1 Tableaux Section

In this section, I give some tableaux to illustrate the generalizations and

constraints established above.

Tableau 8

Input:/RED+ia2k3y4b5y6 b -ia RED=BI Linearity IDENT DEP MAX

Make porridge-asp MORAIC B-R B-R B-R
'Make porridge many times' (F)
****
a. iia2R-iia2k3y4b5y6B -ia *!

**
b. iia2k3e4R-iia2k3y4b5y6B -ia *!

* *
C.iia2k3b5y4 R.iia2k3y4b5y6 b -ia *!
** * ****
d. iia2k3k4y5R-iia2k3y4b5yB -ia *!

****
e. iia2k3y4 R-iia2k3y4b5y6B -ia

In tableau 8, the worst candidates are candidates (a), (c), and (d) because they

violate the undominated constraint RED=Bimoraic in the Waimiri Atroari reduplication

process. Candidate (c) is ruled out because it not only violates the undominated

RED=Bimoraic constraint, it also violates the correspondence constraint Linearity.

Although candidate (b) could possibly be a well-formed candidate because it does not

violate the RED=Bimoraic constraint, it is worse than candidate (e) because it violates the

constraint IDENT B-R (F). Therefore, the optimal candidate is candidate (e) because it

only violates the constraint MAX B-R. Inasmuch as Waimiri Atroari has partial

reduplication, the constraint MAX B-R will always be violated. As result, the ranking of

constraints must follow the hierarchy given below:

(57) RED=BimoraicLinearity, IDENT B-R, DEP B-RMAX B-R.

 61

Tableau 9
Input: RED-t-myryka-ky RED=BI IDENT B-R(F) DEP B-R MAX B-R
mix-IMP MORAIC
'Mix well many times'
***
a. myrka-myryka b -ky *!

b. myre-myryka b -ky *! **

He*
c.myry-myrykaB -ky

d. myryy-myryka B-ky *! *! *

In Tableau 9, as in Tableau 8, the optimal candidate must be the candidate that

does not violate the RED=Bimoraic constraint; therefore, candidate (c) wins. Candidate

(b) could be the winning candidate; however, it violates the IDENT B-R constraint and as

observed in Tableau 8, the change of features between the reduplicant and the base is not

allowed in the reduplication process. Candidate (a) is bad because it violates the

undominated constraint. Candidate (d) not only violates RED=Bimoraic, it also violates

the DEP B-R constraint. Consequently, the ranking of constraints is:

RED=BimoraicIDENT B-R, DEP B-RMAX B-R.

Tableau 10

Input: RED-t-iaa-pa RED=Bl IDENT DEP B-R MAX B-R

Take-tense MORAIC B-R(F)
'Take many times'
*
a. iaan-iaa b -pa *! :

b. iae-iaas-pa *!

*c.iaa-iaa b -pa

In Tableau 10, the whole verb stem is copied; therefore, MAX B-R is not violated

for any of the three candidates. Candidate (a) is the worst candidate because it violates

not only the Bimoraic constraint, but also the DEP B-R constraint. Candidate (b) could
 62

have been considered as an optimal candidate; however, it violates the IDENT B-R

constraint. Therefore, the optimal candidate is candidate (c).

Tableau 11

Input:/RED-nu-wenta b -pa RED=BI Align, IDENT DEP MAX

3S-throw up-tense Morale RED,L B-R(F) B-R B-R
'They threw up many times' Stem,L
a. nu-wenta'-wentaB -pa *! *

b. nu-win-wenta b -pa **
*!

c. nu-weny-wenta b -pa *

**
d. nu-wen-wenta b -pa

e. wen-nu-wentaB -pa **

In Tableau 11, candidate (a) is the worst candidate because it violates not only the

well-formedness constraint RED=Bimoraic, but also the B-R constraint DEP B-R.

Candidates (b) and (c) do not violate the undominated constraint, but they do violate two

important B-R constraints, IDENT B-R, and DEP B-R. Candidate (e) does not violate the

undominated constraint; however, it does violate the Align RED, L, Stem, L constraint.

Therefore, the optimal candidate is candidate (d). The ranking is:

RED=Bimoraic Align RED,L, Stem,L IDENT B-R, DEP B-RMAX B-R.

 63

Tableau 12

Input: RED+ny-tahkwa b -ky RED=BI Align IDENT DEP MAX

3S-jump- MORAIC RED,L B-R(F) B-R B-R
'Repeatedly, he jumped' Stem,L
a. ny-tahkwa-tahkwaB -ky *!

He**
b. ny-tah-tahkwa B -ky

c. ny-tan-tahkwa B -ky *! ***

d. ny-taah-tahkwa b -ky *! * **

e. taah-ny- tahkwa b -ky *! ** * **

Although candidate (a) in Tableau 12 has fewer violations than candidates (c), (d),

and (e), it cannot be an optimal candidate because it does violate the Bimoraic condition

established for the RED structure. Candidate (c) does not violate Red=Bimoraic, but it

does violate IDENT B-R (F). Candidates (d) and (e) are completely out because they not

only violate the undominated constraint, they also violate the DEP B-R constraint.

However, in this case candidate (e) is the worst candidate because it violates the

alignment constraint, as well. Candidate (b) is the optimal candidate because it does not

violate the RED=Bimoraic condition. As a result, the ranking established is;

RED=BimoraicAlign RED, L, Stem, LIDENT B-R, DEP B-RMAX B-R.

 64

Tableau 13

Input: RED+ny-kynky b -pia RED=BI Align IDENT DEP MAX

3S-break-tense MORAIC RED,L B-R(F) B-R B-R
"repeatedly, it break" Steni,L
a. ny-kynky-kynky b -pia *!

b. ny-kyn-kynky b -pia **

c. ny-ken-kynky b -pia *! **

d. kyn-ny- kynky b -pia ** 1 **

Although candidate (d) does not violate the bimoraic condition in Tableau 13, it

cannot be the optimal candidate because it violates the alignment constraint in which the

reduplicant has to be aligned with the left edge of the verb stem. Candidate (a) violates

the undominated constraint and candidate (c) violates the IDENT B-R constraint. As a

result, candidate (b) is the optimal candidate because it only violates MAX B-R.

Tableau 14

Input: RED+n-aryme B-pa RED/ RED= Align, IDENT DEP MAX

3S-come back-tense ONS BIMO RED, B-R B-R B-R
'They came back many times' RAIC L,
Stem,
L

a. ary-naryme b -pa *! * **

b. na-naryme b -pa *!
v. * ****

*
c. nyry'-naryme b -pa *! *

i , ,1 **
d. nary-naryme b -pa

In Tableau 14, it was necessary to include other well-formedness constraints, in

addition to the RED/ONSET constraint. Although the syllable structure in Waimiri

Atroari allows syllables without onsets, the structure of the reduplicant must be
 65

unmarked: it has to have an onset. As a result, when the verb stem starts with a vowel, the

prefix is copied, as well. Consequently, candidate (a) is not an optimal candidate because

it violates the well-formedness constraint (Kager, 1999) and the emergence of the

unmarked form required for the RED. Therefore, candidate (d) is the optimal candidate.

Candidate (b) could be an optimal candidate, but it violates RED=Bimoraic. Candidate

(c) is the worst candidate because it violates three important constraints. In this tableau,

the violation of the alignment constraint seems to be irrelevant inasmuch as the

RED/ONSET constraint caimot be violated.

Tableau 15

Input: RED+w-oo B -se RED/ RED= Align IDENT MAX B-R

pref-climb-suff ONSET BIMO RED,L B-R
'climb repeatedly' RAIC Stem,L

a. oo-w-oo B -se *! *

b. woo'-w-oo B - se *! *

c. oa-w-oo B -se *! * *

* **
d. WOO -W-OOB -se

In Tableau 14 as in Tableau 15, the RED/ONSET requirement has to be

undominated; consequently candidates (a) and (c) are completely out of the competition

because they do not obey the requirement. Candidate (b) cannot be the optimal candidate

because it violates the RED=Bimoraic constraint. Although candidate (d) violates the

alignment constraint, it is the optimal candidate because it does not violate the

RED/ONSET and the RED=Bimoraic constraints.

 In Waimiri Atroari, the number of moras seems to define the shape of the

reduplicant. As a result, the reduplicant is prosodically determined; it must be bimoraic.

The template is not just segmental (or skeletal as assumed by Marantz 1982, Mester

1988), but it is a prosodic entity as demonstrated in the Waimiri Atraori case. This

analysis shows that the well-formedness constraints are above the Base-Reduplicant

constraints, following this ranking; well-formedness contraints (RED/ONSET

RED=BimoraicAlign RED, L, Stem, L)Base-Reduplicant constraints (IDENT B-R,

DEP B-RMAX B-R). The superiority of the well-formedness constraints reinforces the

analysis that Waimiri Atroari has a typical case of emergence of the unmarked structure

when selecting the shape of the reduplicant; consequently, RED must have an onset.

According to the data collected thus far, reduplication examples with V and

CVCVC verb stem have not yet been found. However, it can be predicted that in a verb

with V stem shape, the reduplicant would copy the prefix to avoid the onset constraint

and lengthen the vowel to satisfy the bimoraic condition (as explained for Tiriyo in

section 2.5.2). For example, n-y-sa, 'he goes' would reduplicate as nyy-ny-sa, obeying

the bimoraic condition. On the other hand, in the case of a CVCVC verb stem, the coda

of the second syllable would be deleted to avoid violating the RED-bimoraic constraint.

For example, ny-synehky-pia, 'he disappeared' would reduplicate as ny-syne-synehky-pia.

In this case, one more violation would account for MAX B-R, without breaking the

established ranking between the well-formedness and the base-reduplicant constraints.

 CHAPTER 3 - MORPHOLOGY AND LEXICON

3 Morphology and Lexicon

Languages differ in their inventories of parts of speech. Determination of the parts

of speech for a given language is based on a consideration of several factors, such as

semantic, morphological, and syntactic criteria. According to Meira (2003), few studies

of the Carib languages discuss and define criteria for word class (Hoff 1968 on Carib;

Derbyshire 1985, 1979 on Hixikaryana; and Meira 1999 on Tiriyo).

Therefore, this chapter is organized as follows: Section 3.1 provides a discussion

of the criteria that were used to define the parts of speech in Waimiri Atroari and to

describe each lexical class. Section 3.2 analyzes case marking in this language.

3.1 Waimiri Atroari Parts of Speech

Waimiri Atroari has a complex and rich morphology typical of the Carib family.

It presents both prefixes and suffixes. The prefixes mark person and the suffixes mark

tense-aspect-mood, causativization, and a variety of meaning-changing derivations. On

morphological and syntactic grounds, Waimiri Atroari stems can be divided into six

lexical classes: (1) nouns including pronouns; (2) adjectives; (3) verbs; (4) adverbs; (5)

postpositions; and (6) particles.

As already noted for other Carib languages, in Waimiri Atroari the nouns and the

verbs are unproblematic and are easy to identify (see Table 3.1 below). However, the

remaining classes, such as adjectives, adverbs, and postpositions, are much more difficult

to characterize.
 68

Table 3.1Noun and Verb Characteristics

Nouns Verbs
Can take inflectional affixes Can take a large set of inflectional
marking person (possession); affixes, such as prefixes marking
Can take derivational affixes such person (A, S, 0), and suffixes
as verbalizers and adverbalizers; marking tense, aspect, mood;
Occur as the head of a simple Noun Can take a set of specific
Phrase; derivational affixes used to form
Occur together with a transitive nouns {-typy) and adverbs {-esa,
verb to form a VP; -pesa)-.
Can occupy core argument position
in a clause, such as A, S, 0.

3.1.1 Nouns

The classic definition of nouns argues that nouns are said to represent 'persons,

places, and things;' that is, nouns give names to, or identify, objects. In Waimiri Atroari,

the nouns take inflectional affixes to indicate possession. Nouns also take derivational

affixes like the verbalizer -ta, the absentive -my, the devaluative -erne, and the valuative

-e'me (this will be discussed below). The nouns in this language do not inflect for

gender, number, or degree. In certain cases, gender differentiation is realized by means of

different words for men and women, through male and female speech (principally on

kinship terms).

Table 3.2Lexical differences in men and women's speech

Female speaking Male speaking

'son' biky myryky

'niece' sister's daughter imekyky pasky

 69

Nouns occupy core argument positions in a clause, such as A, S, and O. They

occur as the head of a simple noun phrase. Together with transitive verbs, nouns form

verb phrases (see discussion in chapter 4). The noun can be formed by one root or by two

roots by means of a compounding process.

(1)

emyry ba kyse dykry

penis seed/egg leg joint
'testicles' 'knee'

Table 3,3Some examples of Waimiri Atroari nouns

wyty 'meat, food' weri 'women' wyie 'sun' xiba 'fish'

sanja 'manioc flour' wykyry 'men' nenuwe 'moon' kwata 'spider monkey'
woky 'banana' bahinja 'child' tarara 'thunder' kyrywy 'snake'
syna 'water' txamyry 'elder' kaapy 'sky, heaven' warara 'turtle'

3.1.1.1 Possession

In Waimiri Atroari, there is alienable possession and inalienable possession.

Certain nouns, such as body parts and kinship terms, must have a possessor. There are

some nouns which are never possessed and so never take person markers (See Table 3.4

below), such as nature elements (sun, moon, star). Interestingly, some nouns that could be

possessed, for example, dog, bird, banana, or fish, are possessed with a generic

expression for pet ieky and food wyty. For example, aa-ieky can mean 'my dog'. I also

can say aa-iehy naminja 'my pet dog'. Possessor-possessed order is obligatory, and the

possessed item, not the possessor, is morphologically marked. In Waimiri Atroari, a

relational morpheme i- occurs on the possessed item to link two contiguous elements and

form a phrase. (See discussion in the section 3.1.1.1.2 - Relational morpheme).

Table 3.4Possessive Markers

my aa=
your a= ... -ty, a=
his/her,their kyy-
our (1+2) ky- ... -ty
our (1+3) a'-
Reflexive ty- (3'^'^ person)

Table 3.5Possessor-possessed paradigms

samka 'hammock' mydy'house' kaapa 'plantation
garden'
M^samka 'my hammock' aa=mydy 'my house' aa=kaapa 'my garden'
a=samka-ty 'your hammock' a=mydy-ty 'your house' a=kaapa-tv 'your
kw-samka 'his hammock' kyy-mydy 'his/her house kyy-kapa 'his garden'
ky-samka-ty 'our hammock ky-mydy-ty 'our house' ky-kaapa-ty 'our garden'
a'=samka 'our hammock' a'=mvdv 'our house a^=kaapa 'our garden
ty-sUXLka 'his own hammock' ty-mydy 'his own house' ty-kaapa 'his own garden'

pana 'ears' eba 'eyes' nata 'noose'

aa=pana 'my ear' ^=i-eba 'my eyes' M=i-nata 'my noose'
a=pana-ty 'your ear' a-i-eba-ty 'your eyes' a=i-nata 'your noose'
kyy-pana 'his/her ear' kyy-ieba 'his/her eyes' kyy-nata 'his/her 'noose'
al=pana 'our ears' a^=i-eba 'our eyes' ^=nata 'our noose'
ty-pana 'his own ear' ty-nata 'his own noose'

biky 'son' mama 'mother' iaska 'relatives'

aa=biky 'my son' aa=mama 'my mother' aa=iaska 'my relatives'
a=biky 'your son' a=mama 'your mother' a=iaska 'your relatives'
kyy=biky 'her son' kyy=mama'his, her mother kyv=iaska 'his/her relatives'
a'=bikv 'our son' a'=mama 'our mother' a'-=iaska 'our relatives'

Some examples sentences with possessive prefixes follow:

(2) aa-=eky ram ietypy-pa na.
IPOS-pet 2PART sick-EMPH COP
'My pet is sick.'

(3) kyy=pyta ram abemyh-pa.

3POS-mouth 2PART swelling-EMPH
'His/her mouth is swelling.'
 71

(4) aa-=awo txi-pia itxi taka wyty ipo-se.

iPOS-father-in-law go-IM.P jungle AL meat look for-in order to
'My father-in-law went to look for meat in the jungle'

(5) aa=se 'my feet'

a-se-ty 'your feet'
kw=se-ty 'his/her feet'
^-se-ty 'our feet' (1+2)
a^=se-ty 'our feet' (1+3)

(6) mykyky ty-se kinj-e

3PRO 3REFLX-feet wash-xM
'He is washing his own foot.'

(7) Dauna ram aa=se karyky-pia.

Dauna 2PART IPOS-feet step on-LM.P
'Daiina stepped on my foot.'

(8) Dauna kvv=se karyky-pia

Dauna 3POS-feet step on-LM.P
'Dauna stepped on his feet.'

(9) Kynetxiri ram tabe'e i-se pyky-piany pyruwa ke.

Kynetxiri 2PART capybara REL-feet shoot-REC.P arrow INSTR
'Kynetxiri shot an arrow at the capybara's foot.'

(10) Kaina ram txi-pia n-aram-piany ty-mydy taka.

Kaina 2PART go-IM.P 3S-come back-REC.P 3REFLX-house AL
'Kaina came back to her own house.'

3.1.1.1.2 Relational Morphemes

In Waimiri Atroari, a number of vowel-initial noun and verb stems take a 'linking

prefix' i- when immediately preceded by their determiners (that is, the possessor, with

nouns, and the object, with transitive verbs). Besides its occurrence with transitive verb

stems, such as akyna 'to sweep' (11), this prefix generally occurs with obligatorily

possessed nouns (body-part and kinship terms, etc.), such as eba 'eye' (12):
 72

(11) bahinjan-itxi-pia mydy i-akyna-se

child 3-go-LM.P house REL-sweep-in.order.to
'The child went to sweep the house.'

(12) a. Ewepe i-eba b. a=i-eba

Ewepe REL-eye 2=REL-eye
' Ewepe 'seye' 'yovireye'

On the other hand, consonant-initial stems, such as pana 'ear' and xiky 'to cut'

(13), do not present any linking prefix under these circumstances:

(13) Kynetxiri ram Irie pana xiky-pia maia ke

Kynetxiri 2PART Irie ear cut-LM.P knife INSTR
'Kynetxiri cut Trie's ear with a knife.'

Similar morphological devices are also found in other Carib languages, such as

Hixkaryana (14). Although Derbyshire (1985:200) describes the i- prefix of Hixkaryana

as a 3rd person marker, his

*
analysis is probably not totally accurate, since, as in Waimiri

Atroari, the prefix i- can also co-occur with 2"'' and 1' plural inclusive possessors:

o-y-owani 'your chest', amnay-owani 'our (EXCL) chests.'

Hixkaryana (Derbyshire 1985:5, reinterpreted)

(14) a. Haname y-awo-ru b. 0-awo-ru

Haname REL-uncle-POS 3-uncle-POS
'Haname's uncle' 'his/her uncle'

Following a well-established tradition in South American linguistics, I term i- a

relational prefix. Although rarely described as such for Carib languages (where they

have been traditionally analyzed as 3'^'' person markers), relational prefixes are very

common in languages of the Tupi and Macro-Je stocks, a fact that has been pointed out as

evidence for the genetic relationship between Carib and those two language groupings

(Rodrigues 1994).
 73

Besides providing further evidence for the existence of relational prefixes in the

Carib family, the consequences of such an analysis muddies the question of how to

distinguish pronominal prefixes from clitics in Waimiri Atroari (and maybe other Carib

languages as well). The hypothesis to be investigated in future work is that both noun

and verb stems in Waimiri Atroari present only one slot for prefixation. This slot can be

occupied either by a personal prefix or by the relational prefix i-. Since personal prefixes

and the relational prefix cannot co-occur, the presence or absence of a relational prefix

would provide a straightforward criterion to determine whether a given pronominal

morpheme preceding a noun or a verb stem is a prefix or a clitic. This distinction would

explain the differences in morphological behavior between the 1' person plural inclusive

morpheme k(y)- (15a, 16a) and the 3'^'' person morpheme kyy= (15b, 16b), for example.

(15) a. k-eba b. kyy=i-eba

1+2-eye 3=REL-eye
'our eyes' 'his/her eyes'

(16) a. ky-pana b. kyy=pana

1+2-ear 3=ear
'our ears' 'his/her ear'

This analysis is also strongly corroborated by syntactic evidence. Apparently,

only the elements here analyzed as clitics seem to have argument status when attached to

a verb, while 'true' prefixes seem to be mere agreement markers.

3.1.1.2 Derivational Morphemes

The nouns also occur with 3 different classes of suffixes: the verbalizers -to, a

suffix that indicates absentive -my, and a group of suffixes that I called valuative -e 'me

and devaluative -eme suffixes.

 74

(17) Verbalizer -to

Nouns like wena 'vomit' and myny 'blood' appear suffixed with -ta form verbs;

for example, wen-ta 'to vomit,' and myn-ta 'to bleed.'

wen-ta-pa 'I vomit' wy-myn-ta-pa 'I bleed'

my-wen-ta-pa 'you vomit' my-myn-ta-pa 'you bleed'

(18) Absentive -my

In Waimiri Atroari, some nouns can take a suffix that indicates absence,

deprivation, and lack.

ety-my 'nameless/without name' njydy-my 'without house'

iee-my 'without teeth iaska-my 'without relatives'
iyhia-my 'without hair' ameky-my 'without daughter'
eba-my 'without eyes/blind' myny-my 'without blood'
kiawa-my 'without menstruation' emy-my 'without penis'

(a) pyruwa ram nata-my.

arrow 2PART point-without
'The arrow is without an arrowhead.'

Interestingly, I tried to put this suffix in words like xiwia 'beautiful' to see if we

could have xiwia-my 'without beauty' or tykomia 'cold' then tykomia-my 'without cold',

and etypa 'sick', then etypa-my 'without sickness'. But they are not allowed. This could

be because those words can take the suffix -pa 'emphatic' that only adjectives can take.

(19) Valuative -e 'me and Devaluative -erne

In Waimiri Atroari, nouns (animate or inanimate), pronouns, and adjectives can

take these two different morphemes. The devaluative -eme can indicate that a person or

an animal is dead or sick, and that an object is no good anymore, old, or no longer usable.

On the other hand, the valuative -e 'me, when used for persons and animals, indicates that

they are alive; when used with objects, it indicates that they are still good to use.
 75

These morphemes can occtar with subjects of intransitive and transitive verbs,

example (22), and objects of transitive verbs (23). I haven't seen a situation in which both

of them can occur in the same sentence. This may be because only one is necessary

inasmuch as they are not required to disambiguate the sentence. Moreover, -e 'me and -

erne are not obligatory, and the valuative occurs less often than the devaluative.

(20) aa ram naminja ya k-eme iika h-ini-pia.

Ipro 2PART dog AGT.PART 3PRO-DEV bite IS-see-IM.P
'I saw the dog bite him.'

(21) wykyr-eme wu-piany ram tymere.

man-DEV kill-REC.P 2PARt jaguar
'The jaguar killed the man.'

(22) wykyr-eme ny-damem-pa.

man-DEV 3S-die-REM.P
'The man died.'

(23) Pana Mateus imeses-eme wu-piany wiwe ke.

yesterday Mateus bat-DEV kill-REC.P wood INSTR
'Yesterday Mateus killed the bat with the wood.'

(25) byby ram tykoxinja-p-eme njaminj-eme.

This 2PART dirty-EMPH-DEV dog-DEV
'This dirty dog.'

(26) aa ram mepr-eme i-webyry hy-kyty-pia.

IPRO 2RAM tapir-DEV REL-belly lA-cutiM.P
'I cut the tapir's belly.'

(27) aa wo'nj-e'me h-aminjaky-piany a=wenpa-typa tre'me tyruwa kapry pyky.^"

IPRO clay-VAL IS-permit/let-REC.P 20-leam-? PART pan make how
'I permitted you to dabble in the clay to learn how to make a ceramic pan.'

(28) aa-pap-e'me sehe.

IPos-father-VAL tall
'My father is tall.' (He is alive)

In this sentence, the valuative morpheme is marked in the word for clay, to indicate that this is a good
clay to make a ceramic pan.
 76

(29) a'-itxiri ram tah-e'me na.

l+3P0S-land 2part big-VAL COP
'Our land/territory is big.'

(30) apia iakaha tah-e'me n=e'

what drawing big-VAL COP-INT
'What is the big drawing?'

3.1.1.3 Pronouns

In this grammatical sketch, pronouns are treated as a subclass of nouns. The

characteristics used to describe them are the existence of animacy and deictic distinction.

In Waimiri Atroari, pronouns can occur in subject and object position, and like nouns,

they can also take the valuative and devaluative morphemes. The Waimiri Atroari

pronominal system includes demonstratives, interrogatives, and personal pronouns.

Waimiri Atroari has the typical Carib system for pronouns: first person (1),

second person (2), first person dual inclusive (1+2), first person exclusive (1+3), and the

third person pronouns, including the demonstratives (proximal, medial, and distal). The

third-person pronouns in Waimiri Atroari are sensitive to aspects such as visibility and

proximity. The terms proximal, medial, and distal are used to label the demonstrative

categories representing the degree of proximity (closeness) to the reference point, which

is often the speaker's location.

 77

Table 3.6Waimiri Atroari Pronominal System

ist J awy, aa, kara~kra

2"'' you amyry-amyra.

1+2 we incl kyky

1+3 we excl a'a

3rd

anaphoric (he, she, they, it) mykyky, mykyka'a, ka, iry

proximal (h)anji, kanji, anjinji, byby, by
medial myry
distal mo'o, mymo', myky

3.1.1.4 The Non-Third Person Pronouns

Personal pronouns identify people and things on the basis of their relationship to

the discourse in which they are referred to, such as their conversational role. For

example, the pronouns awy, kara, aa T are used by anyone who takes the role of speaker

in a discourse. Note that the first-person pronoun has three different forms. The form

kara T', with its alternate kra, occurs when somebody is answering a question or

emphasizing that he did the action, or that he wants something. It is the only pronoun that

can occur in OSV order, separating the verb phrase (see discussion on word order in

chapter 4). The forms aa and awy T are used without any specific distinction (see

below).

(31) kaapa i-akytx-iany kra aa=i-aska many.

Garden REL-cut-X/A IPRO iPOS-REL-relatives and/too/with
T was preparing the plantation garden with my relatives.'
 78

(32) wyty ipo-se kra w-y-sa.

meat look for-in order to IPRO IS-go-I/A
'I am going to look for meat.'

(33) meprypyny kara h-yn-iany.

Tapir meat IPRO LA-eat-T/A
'I was eating tapir meat.'

(34) araky kra wy-tyty-pia.

Today IPRO IS-come-LM.P
'I came today.'

(35) saken-pa kara w-ia.

Angry-EMPH IPRO 1-COP
'I am angry.'

(36) wasypy-pa kra w-ia.

Hungry-Emph IPRO 1-COP
'I am hungry.'

Note that examples (31) to (36) are answering a question.

(37) aminjaky awy w-y-sapa wiwe by-se.

Tomorrow IPRO IS-go-X/A wood cut-in order to
'Tomorrow I will go to cut wood.'

(38) awy wa' h-ipinj-e.

IPRO rope IS-pull-I/A
T am pulling the rope.'

(39) awy kyrywy h-ini-pia.

IPRO snake IS-see-IM.P
'I saw the snake.'

(40) awy wy-tam-pia.

IPRO 1 S-cry-IM.P
'I cried.'

Observe that the free independent pronouns aa '1 pro' and a'a '1+3 pro' are

identical to the person prefixes for possessed nominals. The 1+3 possessive mark loses

the vowel after the glottal (see Table 3.4 above).

 79

(41) aa ram w-aram-piany aa=mydy taka.

IPRO 2PART 1S-come back-REC.P IPOS-house AL
'I came back to my house.'

(42) aa ram naminja x-ipotx-ia wiwi ke.

IPRO 2PART dog lA-kill-T/A stick INSTR
'I killed the dog with the wooden stick.'

(43) aa xiba hy-myryk-e kyky-pesa.

IPRO fish lA-fish-T/A night-attime
'I fish at nighttime.'

(44) aa w-y-sa kaapa taka awaxi kyty-se.

IPRO IS-go-x/A garden AL sugarcane cut-in order to
'I am going to the plantation garden to cut sugar cane.'

As illustrated in Table 3.6 above, the other non-third-person pronouns do not have

different forms from the first person singular. The pronoun amyry (a) 'you' is used to

refer to anyone who is being spoken to in a discourse. First-person plural in Waimiri

Atroari makes a distinction between inclusive and exclusive. First-person plural inclusive

kyky includes the person addressed, meaning something like 'weyou and I'. The first

person plural exclusive a'a excludes the person or persons addressed, meaning 'we, that

is, I and others, but not you.' The examples below illustrate sentences with second-

person, 1+2, and 1+3 pronouns.

(45) amyra mykyka m-ary-py-pia mykyka ini-se.

2PRO 3PRO 2A-order-CAUS-IM.P 3PRO see-in order to
'You ordered him to see him.'

(46) amyry m-om-pia syna kaka

2PRO 2S-dive-IM.P water Loc
'You dove into the water.'

(47) araky ram kyky h-y-sa xiba myryka-se

today 2PART 1+2PRO l+2S-go-T/A fish fish-in order to
'Today we go fishing.'
 80

(48) kyky hu-wa-pa.

1+2PRO l+2S-dance-REM.P
'We danced.'

(49) a'a n-ikin-pia syna ke paxa taka.

1+3PR0 l+3S-wash-IM.P water INSTR bowl LOC
'We washed with water in the bowl.'

(50) a'a n-itxi-piany kaapa taka.

1+3PRO l+3S-go-REC.P garden AL
'We went to the plantation garden.'

3.1.1.5 Third-Person Pronouns

In Waimiri Atroari, as for other languages in the Carib family, the third-person

pronouns are sensitive to features such of proximity, visibility, and animacy. The

anaphoric class has three kinds of third-person pronouns; mykyky 'he, she, it, they, who

are close, proximal attending the conversation;' mykyka'a, ka 'he, she, it, they, who are

not close, but whom we can see, distal;' and iry 'he, she, they, it, who are not close or

observing the conversation, obviative.' In mythological histories, iry is clearly more

utilized (see examples below).

(51) mykyky ty-se kinj-e

3PRO 3REFLX-foot wash-T/A
'He is washing his own foot.'

(52) mykyky topy i-mapysa

3PRO stone REL-throw-x/A
'He is throwing stones.'

(53) mykyka'a ram arema.

3PRO 2PART singer
'He is a singer' also translated as 'That kinja is a singer.'

(54) mykyka'a ram n-yma-pa.

3PRO 2PART 3S-fall-REM.p
'He fell.'
 81

(55) ka ram aa=ini-pia.

3PRO 2PART 10-see-iM.p
'He saw me.'

(56) ka ram ka ini-huwa na.

3 PRO 2PART 3PRO see-NEG COP
'He does not see him.'

(57) iry n-itxi-pia mixi i-amek-se.

3PRO 3S-go-LM.P buriti fruit REL-take-in order to
'She went to take buriti fruit.'

(58) iry n-aryma-pa te'xy n-eeni-pa.

3PRO 3S-come back-REM.p DESID 3S-stay-REM.P
'He did not want to come back.'

Both mykyky and ka can occur in the subject and object position, but in my data

ka seems to be preferred when occupying an object position. On the other hand, I did not

find many examples with iry occurring in the object position.

The demonstrative or deictic forms are marked for proximity and animacy. In the

proximal group, the (h)anji, kanji, anjinji forms mean 'this, these,' but in some situations,

they can be translated as 'here.' In this group, only byby, by is used for animate objects.

It means 'this,' but in some examples it means 'he/she' (see examples below).

(59) anja ram akyrypa i-akyda-typa.

This 2PART trash REL-clean-used to
'This is used to clean trash' also can be translated as 'This is a broom.'

(60) anji ram waiepa.

This 2PART jamaxi (kind of basketry utilized to carry wood and food)
'This is a jamaxi.'

(61) apia hanji?

What this
'What is this?'
 82

(62) aa=samka ram kanji.

lPOS=hammock 2PART this
'This is my hammock.'

(63) aa^iaska i-akaha kanji

lpos=relatives rel-pictures this
'This is a picture of my relatives.'

(64) pip-ky anjinji i-etaty

look for-IMP here/this REL-name
'Look here for the names!'

(65) byby ram aa=myryka.

This 2PART lPOS=son
'This is my son.'

(66) bypa i-eka by karyka-e' ?

whose REL-pet this chicken-iNT
'Whose is this chicken?'

(67) byby maryba ka-tape.

3PRO song sing-REM.F
'He will sing.'

The medial form myry also means 'this,' but like the proximal forms anji, kanji it

can only be used with inanimate objects.

(68) wyty ka myry.

meat EVID this
'This is meat.'

Examples (65), (66), and (67) illustrate examples with animate entities; therefore,

the pronoun byby/by is selected. Unlike other Carib languages, which have an elaborate

system to describe animacy, Waimiri Atroari seems to lack it.

In the distal group, the forms mo 'o and mymo' mean 'that, those, there.' They are

used with inanimate objects. The other distal form, myky, also means 'that,' but it only

occurs in examples with animate objects.

 83

(69) mymo marehe ram abremyhsa.

that sieve 2PART round
'That sieve is round.'

(70) mymo' ram aa=samka.

that 2PART lPOS=hammock
'That is my hammock.'

(71) bypa kanuwa mo-e'?

whose canoe that-iNT
'Whose is that canoe?'

(72) myky ram tabe'a.

that 2PART capybara
'That is a capybara.'

(73) myky ram mare'a.

that 2PART jacu (Sp. of bird)
'That is a jacu.'

An interesting aspect related to the form mo 'o is that when translated as 'there,' it

seems that in Waimiri Atroari there are two kinds of 'there.' The first is a 'there' that

gives an idea of a permanent state or remaining there; in this context the morpheme ipy is

utilized. The second 'there' has a temporary notion; in this case the form mo 'o is used

(see examples (76) and(77),

(74) ka ipy xiba naky karara syna ka naka.

EVID there fish COP karara water/river EVID COP
'There, in the Karara River, there are many fish.'

(75) Mawa kynynora kyty-pa iakypa n-yma-pa itxi iary

mythological Sp. of leaf cut-REM.P then 3S-fall-REM.P jungle tree
entity

tyhnaka araky ipy na.

on today there COP
'Mawa cut the kynynora leaf, then he fell on the tree and (remained) there until
today.'
 84

(76) mo'o ka samka.

There EVID hammock
'There, it is the hammock.'

(77) a'a txi-piany mo'o ase mydy taka.

1+3PRO go-REC.P there new house AL
'We went there to the new village.'

3.1.2 Verbs

Verbs are words which signify actions, events, or temporary states in relation to

beings and things in the world. They are the core, the binding element in most sentences

in any language. Verbs can serve as head of verb phrase, predicates of clauses, and they

code events in text. In this section, I discuss primarily the word-structure (morphology)

of verbs. I discuss the various grammatical affixes (prefixes and suffixes) that can be

attached to the verbal word.

The structure of the verb in Waimiri Atroari is basically prefix-stem-suffix. Of all lexical

classes, the verb is the richest in morphological possibilities. It can take a large set of

different grammatical markers indicating person, tense-aspect-mood, negation, and

causativization, as well as a specific derivational suffix used to form nouns -typy.

 85

Table 3.7Structure of Waimiri Atroari verbs

Case marking ROOT Verbalizer Causative

Nominalizer Negation TAM
Imperative tense Interrogative
Desiderative aspect- clitic
Clitic Prefix mood
suffixes

see aa= -ini- -py -pia

lo see -CAU IM.P

h- -ini- -py -pia

LA see -CAUS IM.P
m- -ini- -pi -e'
2A see IM.P INT

vomit aa= -wen- -ta -py -pia

lo vomit VERBL CAUS IM.P

hu- -wen- -ta -py -pia

LA vomit VERBL CAUS IM.P

Table 3.8 below illustrates the morphological possibilities with some examples of

the verbs 'vomit' and 'see.'

 86

Table 3.8Illustration of verbal morphology

Vomit See
1. -wen-ta 1. -ini-
vomit-VERBL see

2. hu-wen-ta-pa 2. h-ini-pia
1S-VOmit-VERBL-REM.P lA-see-lM.P
'I vomited' 'I saw'

3. wen-ta-ha 3. ni-huwa
vomit-VERBL-NEG see-NEG
'not vomit' 'not see'

4. wen-ta-ky 4. ni-ky
vomit-VERBL-iMP see-iMP
'Vomit!' 'Look!'

5. hu-wen-tah-py-pia 5. h-ini-py-pia
1A-vomit-VERBL-CAUS-IM.P lA-see-CAUS-iM.p
'I made him vomit.' 'I made him see.'

6. aa=wen-tah-py-pia 6. aa=ini-py-pia
10-VOmit-VERBL-CAUS-IM.P 10-see-CAUS-iM.P
'He made me vomit.' 'He made me see.'

Verb stems in this language can have different shapes, including V, VV, VC,

VCV, CV, CVC, and CVCVCCV. Tense/aspect markers distinguish non-past, remote,

recent, and imminent past, and future; and a modal suffix marks the manner of the action.

3.9 Illustration of Tense-Aspect Suffixes

Past Non-Past Future
REM.P -pa -i -e, -sa IMM.F -te ~ -txe
REC.P -piany -pysa 'interative' REM.F -tape ~ -txape
IM.P -pia -sapa
Non-specified -ky
 87

There are three subclasses of verbs in Waimiri Atroari: transitive, intransitive, and

copular. The transitive verbs are traditionally defined as the group of verbs which

subcategorize for a direct object. Transitive verb stems can take both A and O prefixes

(see Table 3.10 below). The intransitive verbs are often referred to as the group of verbs

which do not subcategorize for a direct object. Here, I use the term more narrowly, to

mean verbs that have only a subject, not an object. Intransitive verbs can only take one

set of person markers. Intransitive verbs change valence when taking causative suffixes;

as a result, they become transitive verbs taking direct objects. Transitive verb stems make

implicit reference to two participants, A and O, while intransitive verbs denote only one

participant, S.

Table 3.10Person-marking clitics and prefixes

Subjects Objects Possessives

Intransitive Transitive
1' sing. w-/wy-/wu- h-/hy-/hu- aa= aa=
2nd m-/my-/mu- m-/my-/mu- a=,k-/ky-/ku- a=
3rd n-/ny-/nu- n-/ny-/nu- kyy=/ty=(reflx)
1+2 (incl.) h-/hy-/hu- h-/hy-/hu- k-/ky-/ku k-/ky-
1+3 (excl) n-/ny-/nu- n-/ny-/nu- a'= a'=

The forms w-, m-, n-, and h- occurs with verbal stems beginning with vowels, wy-

, my-, ny, hy- are used with verbal stems beginning with consonants. The forms, wu-, mu-,

nu, and hu- only appear with verbal stems beginning with bilabials.^' Paradigms of

transitive and intransitive verb forms with the prefixes marking person follow (more

paradigms are provided in the appendix C).

wy-, my-, ny-, and hy- C

W", m-, -,and h- V
wu-,mu-, nu-, hu- ^bilabials
 88

Table 3.11Person-marking examples

Transitive Intransitive

hy-myryky-piany T fished' wy-rymy-tape 'I will die'

my-myryky-piany 'you fished' my-rymy-tape 'you will die'
ny-myryky-piany 'she/he fished' ny-rymy-tape 'she/he will die'
hy-myryky-piany 'we fished' hy-rymy-tape 'we wiil die'
ny-myryky-piany 'we fished' ny-rymy-tape 'we will die'

h-ee-ia 'I drink' w-yma-pa 'I fell'

m-ee-ia 'you drink' m-yma-pa 'you fell'
n-ee-ia 'she/he drinks' n-yma-pa 'she/he fell'
h-ee-ia 'we drink' h-yma-pa 'we fell'
n-ee-a 'we drink' n-yma-pa 'we fell'

h-ape'-pia 'I embraced' w-ineh-tape 'I will dream'

m-ape'-pia 'you embraced' m-ineh-tape 'you will dream'
n-ape'-pia 'she/he embraced' n-ineh-tape 'she/he will dream'
h-ape'-pia 'we embraced' h-ineh-tape 'we will dream'
n-ape'-pia 'we embraced' n-ineh-tape 'we will die'

3.1.2.1 Tense/Aspect suffixes

The most important and indispensable elements attached to the verbal stem are the

tense-aspect suffixes. They always appear finally in the verb construction. They serve to

indicate a wide range of functions and meanings around the time value of the

action/event/state described by the sentence.

In general, 'present time' is considered immediate, while an event long ago is

considered remote. However, in Waimiri Atroari, it is possible to make events in the past

seem more immediate or recent; therefore, in this language there are suffixes that indicate

remote past -pa, recent past -piany (events that could be occurred in the same day,but

also in the day before), immediate past -pia (an event that just happened), and a non-
 89

specified past -ky. The suffix -pa is most coraumonly used in narratives and in stories

about remote events in the past, things that are finished and done with.

Formally, the future means talking about an event/action/state that has not yet

begun, but is said to begin sometime in the future. The Waimiri Atroari language divides

the future into imminent -teZ-txe and remote -tapeZ-txape. In this grammatical sketch, the

suffixes -e, -ia, -sa, -sapa, -pysa (this last suffix indicates iterative) are classified as

tense/aspect markers that denote a non-past notion. The difference among them is not

clear. It requires futher research. Waimiri Atroari has other suffixes that denote aspects,

but they will be not discussed in this grammatical sketch. Paradigms of intransitive and

transitive verb forms with the tense/aspect markers follow.

(78) remote past -pa

wy-synehka-pa m-ini-Jja
IS-disappear-REM.P 2A-see-rem.p
'I disappeared.' 'You saw it.'

(79) recent past -piany

wy-sjoiehky-piany m-ini-piany
IS-disappear-REC.P 2A-see-REC.P
'I disappeared.' 'You saw it.'

(80) immediate past -pia

wy-synehky-pia m-ini-pia
1S-disappear-IM.P 2A-see-IM.P
'I disappeared.' 'You saw it.'

(81) non-past (-e, -ia, -sa)

(a) aa ram wy-mynt-e apiemeiaky aa=wokyty-pia

IPRO 2PART IS-bleed-T/A because 10-cut-iM.P
'I am bleeding because I cut myself
 90

(b) aa w-y-sa kaapa taka awaxi kyty-se

IPRO IS-go-T/A garden AL sugarcane cut-in order to
'I am going to the garden plantation to cut sugar cane.'

(c) arawata ram ie'ry i-ee-ia

Sp. of monkey 2PART fruit REL-drink-T/A
'The guariba monkey is drinking the fruit.'

Apparently, there are no distinction among them, examples (a), (b), and (c) above

shows that an action is happening.

(82) future {-tetxe/-tapetxape)

(a) aa ram wy-synehka-tape

IPRO 2PART 1S-disappear-REM.F
'I will disappear.'

(b) k-eme ram ny-ryma-tape txamyry-pesa

3PR0-DEV 2PART 3S-die-REM.F elder-intime
'He will die in old age.'

(c) aa h-eni-te
Ipro lA-see-iMM.F
'I will see it.'

3.1.2.2 Mood

Speech aspects in human languages may be used for many purposes, but the three

main ones seem to be: to give information (declarative sentences), to ask for information

(interrogative sentences), and to command action (imperative sentences). In this section, I

discuss the imperative suffix and the negative suffix.

3.1.2.2.1 Imperatives

The purpose of the imperative sentence/clause form is for the speaker to get the

hearer to act. This can be done in several ways, such as subtle and indirect, and direct.

Here, I discuss just the direct form. There are three types of imperative suffixes in
 91

Waimiri Atroari: -ky, -hne, -kwapy. The suffixes -hne and -kwapy have a more restricted

distribution; they occur principally with the verbs speak, hear, and look. However, they

are found with other verbs, as in (84) below.

(83) ni-ky ni-kwapy

See-IMP see-iMP
'Look!' 'Look!'

(84) wen-ta-hne
Vomit-VERBL-IMP
'vomit!'

(85) ka-ky ka-kwapy ka-hne

speak-iMP speak-iMP speak-iMP
'Speak!' 'speak!' 'Speak!'

(86) kyta-ky
yell-iMP
'Yell!'

(87) ee-ky
drink-iMP
'Drink!'

(88) kyrywy ni-ky

snake see-iMP
'Look at the snake!'

(89) suwa ram ky-pia matuwe ia ana kar-eme ika-ky

Suwa 2PART talk-IM.P Matuwe AGT.P Ana kar-DEV bite-iMP
'Suwa ordered Matuwe to bite Ana Kare'me.'

(90) taha iakaha ehry-ky

big drawing paint-iMP
'Paint the big drawing!'

In the negative imperative form, the form used to command people not to do

something, which involves the negative suffix, the verb remains in its bare form, but now

the imperative is marked by the particle kwe'ky.

 92

(91) ni-huwa kwe'ky

see-NEG IMP
'Do not look!'

(92) wen-ta-ha kwe'ky

Vomit-VERBL-NEG IMP
'Do not vomit!'

(93) kyta-ha kwe'ky

yell-NEG IMP
'Do not yell!'

3.1.2.2.2 The Negation Suffix

In Waimiri Atroari, two negation suffixes occur with the verb form: -huwa, and

-ha (V-stem/_a). They are always positioned after the verb root.

(94) awy we-piany dauna se kaka-ha

IPRO ICOP-REC.P Dauna leg scratch-NEG
'I did not scratch Dauna's leg.'

(95) warypa kynka-ha kwe'ky

bow break-NEG IMP
'Don't break the bow!'

(96) aa ram karyka iny-huwa we'xi w-ia

IPRO 2PART chicken eat-NEG DESID 1-cop
'I do not want to eat chicken.'

(97) aa ram wy-tyrym-piany tymeri aa-iape-huwa

IPRO 2PART IS-scape-REC.P jaguar 10-take-NEG
'I escaped so the jaguar wouldn't get me.'

3.1.2.3 The Interrogative Clitic

As shown in table 3.7 in the template for verbs, the interrogative clitic is attached

to the tense/aspect suffix slot. Therefore, I decided to discuss it in this section.

Questions are a request for information, when the speaker misses some

information, and thinks that the hearer knows it. There are two general types of
 93

information questions: (a) yes/no questions in which the speaker wants the hearer to say

whether it is true (yes) or false (no); and (b) Wh- questions in which the speaker knows

most of the information, but is missing one elementas a result, the speaker asks the

hearer to identify that missing element, for example the subject, direct or indirect object,

time, place, or reason.

3.1.2.3.1 Interrogative Forms

Interrogative forms are used when asking questions regarding who, what, whose,

how, where. In Waimiri Atroari there are two forms bypa 'who, whose' and apia 'what'.

The first form is used with animate entities and the second form is used with inanimate

entities.

(98) bypa ha k=e'

who ? 3PRO-INT
'Who is he/she?'

(99) bypa ha b=e' Dauna=e'

who ? this-INT Dauna-INT
'Who is Dauna?'

(100) bypa ieka by naminj=e'

whose pet this dog-INT
'Whose is this dog?'

(101) apia ha iria ka typohinj=e'

what ? make 3PRO like/delicious-INT
'What does she like to make?'

(102) apia ha m-ipipix=e'

what ? 2A-look for-INT
'What are you looking for?'

(103) Ira' a=iakyna n=e'

how 2POS-brother cop-INT
'How is your brother?'
 94

(104) epe ky marym-eme my-myryky-pia=e'

Where this piranha-DEV 2A-fish-IM.P-INT
'Where did you fish piranha?'

(105) apie'me iake saken-pa m-i=e'

why angry-EMPH 2-COP-LNT
'Why are you angry?'

In Waimiri Atroari, even with a Wh- word occurring in the first position, we have

the interrogative clitic =e' in the last word of the sentence. However, sometimes the

interrogative clitic does not appear. In the examples below, I illustrate where the

interrogative clitic occurs and what it attaches to.

Cases Where the Interrogative Clitic Does Not Appear

(106) m-ii-pia' kyry-heme

2A-kill-IMP cobra-DEV
'Did you kill the snake?'

(107) kjoy-huwa tahsa iry

cobra-? big 3PRO
'Was the snake big?'

(108) wahpa a-iwapo-txany

much 2S-dance-T/A
'Did you dance a lot?'

(109) waha ka ipy marymah na

many ? there piranha COP
'Are there many piranha?'

The cases where the interrogative clitic sometimes does not appear seem to be
related to questions of the Yes/No type. However, that optionality does not occur with
Wh- questions.
Interrogative clitics attach to verbs
(110) Temeh-eme m-itxiky-pian=e'
Jaguar-DEV 2A-shoot-REC.P-RNT
'Did you shoot the jaguar with an arrow?'
(111) Epe ka my-pyny-p=e'
where ? 2A-take-REM.P-INT
'Where did you take it?'

(112) Epe ky marym-eme my-myryky-pian=e'

where ? pirana-DEV 2A-fish-REC.P-INT
'Where did you fish pirana?'

(113) Apie'me iake' ipyna xiba myryka m-y-sa-p=e'

why there fish fish 2A-go-REM.P-LNT
'Why did you go there to fish?'

(114) Ira kak=e'

what talk-INT
'What is he saying?'

(115) Apiaiakypa myda taka m-aryma-tap=e'

when house AL 2s-come back-REM.F-INT
'When will you come back home?'

(116) Apiapy ka kwat-eme wu-pian=e'

where 3PRO monkey-DEV kill-REC.P-INT
'Where did he kill the black spider monkey?'

(117) Ira iake kyry-heme my-ry-pi=e'

what snake-DEV 2A-do-LM.P-LNT
'What did you do with the snake?'

(118) Apia ha m-ipipix=e'

what ? 2A-look for-INT
'What are you looking for?'

(119) Epe iake my-taky-pi=e'

where 2A-lose-IM.P-INT
'Where did you lose it?'

(120) Apie'me iake k-eme i-myy-py ny-mynyt=e'

why 3PR0-DEV REL-hand-? 3s-bleed-LNT
'Why is his hand bleeding?'

(121) M-yty-pi=e'
2s-understand-IM.P-INT
'Did you understand?'
 96

To assure that this is a clitic, not just a verb suffix, I demonstrate that it can occurs

with all the word classes in this language. In verbs, the last vowel of the suffix that marks

tense/aspect is replaced by the interrogative clitic: for example, -piany (recent past)

becomes -pian^^e', -pa (remote past) becomes -p=e', -pia (immediate past) becomes -

pi=e\ and -e (t/a) becomes =e' (glottalized).

Interrogative clitic with nouns and pronouns

(122) Bypa iekaby karyka=e'

whose pet this chicken-lNT
'Whose is this chicken?'
In the example (122) above, you can note that when the interrogative suffix is

added to the word karyka 'chicken,' the last vowel of the word is not deleted.

(123) Bypa iekaby naminj=e'

whose pet this dog-iNT
'Whose is this dog?'

(124) Apiaiaky iake katy iky-piany naminja^^e'

when pet ? bite-REC.P dog-iNT
'When did the dog bite the child?'

On the other hand, examples (123) and (124) show the two possibilities. In (123),

the last vowel of the word naminja 'dog' is deleted, but in (124), the interrogative suffix

is attached to the word.

(125) Bypa ieka by waiama=e'

whose pet this tortoise-INT
'Whose is this tortoise?'

(126) Bypa ha be' Dauna=e'

who ? this Dauna-INT
'Who is Dauna?'
(127) Bypa ha b=e'
who ? this-INT
'Who is this?'

(128) Bypa ha k=e'

who ? SPRO-INT
'Who is he?'

Interrogative clitic with adverbs

(129) Bypa ha itxa ta m-inipe-txane pana=e'

what ? jungle LOC 2A-ver-T/A yesterday-INT
'What did you see in the jungle yesterday?'

Interrogative clitic with adjectives

(130) Bypa mydy-hy tahs=e'

whose house-? big-INT
'Whose big house is this?'

(131) Apia wiwe bahinj=e'

what wood stick small-INT
'What is the small wood stick?'

Interrogative clitic with locatives

(132) Bypa ha mymo'o tahsa samka tyhn=e'

who ? that big hammock LOC-INT
'Who is in that big hammock?'

Interrogative clitic with the copula

(133) Epe ky wer-e'me n=e'

where ? woman-VAL cop-INT
'Where are the women?'

(134) Epe ky wykyre-he'me n=e'

where ? man-VAL coP-INT
'Where are the men?'

(135) Sakenpa m-ia

Angry 2-coP
'Are you angry?'
 98

(136) Sakenpa m-i=e'

angry 2-COP-INT
'Are you angry?'

(137) Wasypa m-i=e'

hungry 2-COP-INT
'Are you hungry?'

(138) Wasypa m-ia

Hungry 2-cop
'Are you hungry?'

Examples (135) to (138) show the flexibility of the interrogative suffix with

Yes/No questions. As you can see above in examples (110) to (121), the clitic occurs in

the verb. However, it can occur in the last word of the sentence, such as nouns and

pronouns (examples 122-128), adverbs (example 129), adjectives (130 and 131),

locatives (example 132), and copulas (examples 133-138). Phonologically, sometimes the

last vowel of the verb or noun is replaced by the clitic =e' (examples 110-121) and 124);

on the other hand, note that sometimes there is no replacement of the last vowel. In that

case, the clitic =e' is added at the end of word (examples 124-126^ and 129. The second

case does not occur with verbs.

Interrogative clitic with disjunctive sentences

(139) ny-kwatympotxa kyn=e' ny-proky-pia kyn

3s-run DISJ.PART-INT 3s-run down-LM.P DISJ.PART
'Did he run or "kick one's heels"?'

(140) maryba takypa kyn txi-pi=e' ny-proky-pia kyn

Party ? DISJ.PART go-IM.p-INT 3S-stay-IM.p DISJ.PART
'Did he go to the party or did he stay?'
 99

(141) ny-damem-piany kyn=e' nyry kyn=e'

3S-die-REC.P DISJ.PART-INT alive DISJ.PART-INT
'Did he die or is he alive?'

Observe that in this kind of sentence, the interrogative clitic could occur in

different situations. For example, in (139), it occurred in the first particle that marks

disjunction, but in the example (140), it occurred in the verb; however in the last example

(141), the interrogative clitic occurred not only in the first particle, but also it appeared in

the last disjunctive particle. This shows that the interrogative clitic always appears as the

last element of the sentence. Previously, I had considered the interrogative clitic to be a

suffix. However, two factors, distribution (it can occur in different parts of speech) and

location (specific place/rightmost word in the sentence) made me change the

classification. Utilizing Anderson's view (1992), the interrogative clitic in Waimiri

Atroari seems to behave as a 'special clitic', which is located with a phrase that

constitutes its scope, and it may appear at least initially, finally, or post-finally. In the

case of the Waimiri Atroari language, the interrogative clitic occurs in the final position.

Anderson says that clitics appear to be somewhat freer than affixes. They are much less

sensitive to the properties of words (Anderson 1992).

3.1.2.4 Causative Forms

In Waimiri Atroari, I observed two kinds of causative construction. For example,

there is one in which causativization is morphologically marked with the suffix -py,

generally with the 'made' reading, having the semantic meaning 'make somebody do

something' or 'cause something without resistance.' The other kind has the 'let' reading,

where the morpheme -py does not occur.

 100

3.1.2.4.1 Construction with the 'made' reading

In this type of construction, Waimiri Atroari allows two possible structures: one

structure in which the causative morpheme appears on the lexicalized made/cause verb,

as observed in examples (142) to (145); and the other structure in which this lexicalized

verb does not appear, as demonstrated in examples (146) to (156) below.

(142) Aa Kaina h-ary-py-pia kyrywu ini-se.

IPRO Kaina Is-tell-CAUS-IM.P snake see-in order to
'I told/dictated to Kaina to see the snake.'

(143) Aa Mypyny h-ary-py-pia mepiri ini-se itxi tanyme

IPRO Mypynyls-tell-CAUS-IM.P tapir see-in order to jungle AL
'I told/dictated to Mypyny to see the tapir that came from the jungle.'

(144) Paruwe aa-iry-py-pia woky yry-ky Marta inaka.

Parawe lo-tell-CAUS-IM.P banana give-IMP Marta DAT
'Paruwe told/dictated to me to give the banana to Marta.'

(145) Amyra mykyka m-ary-py-pia mykyka ini-se.

2PRO 3PRO 2s-told/dictated-LM.P 3PRO see-in order to
'You told him to see him.'

(146) Amyra ram ka mu-wen-tah-py-piya

2PRO 2PART 3PRO 2A-threw UP-VERBL-CAUS-IM.P
'You made him throw up.'

(147) ka ram a-wen-tah-py-pia.

3PRO 2PART 2o-threw up-VERBL-CAUS-IM.p
'She/he made you throw up.'

(148) kyka ram ka hu-myny-tah-py-pia.

1+2PR0 2PART 3PR0 l+2A-bleed-VERBL-CAUS-IM.P
'We made him bleed.'

(149) Ka k-yeepitxah-py-pia.
3PRO l+2o-laugh-CAUS-IM.p
'She/he made us laugh.'
 101

(150) Ka ram a-irima-py-piany

3PRO 2PART 2o-rest-CAUS-REC.P
'He made you rest.'

(151) Ka ram aa=kytah-py-pia

3PRO 2PART LO-shout-CAUS-IM.P
'He made me shout.'

(152) Ka ram aa=ima-py-pia syna kaka.

3PRO 2PART lo-jump-CAUS-IM.p water LOG
'He made me jump in the water.'

(153) Kyka Joanico hy-ma-py-piya syna kaka.

1+2PRO Joanico l+2A-jump-GAUS-IM.P water LOG
'We made Joanico jump in the water.'

(154) Aa ram ka hu-pakah-py-pa.

IPRO 2PART 3PRO Is-wake UP-GAUS-REM.P
'I made him wake up.'

(155) Aa ram Kaina ia kyrywu h-ini-py-pia.

IPRO 2PART Kaina AGT.PART snake lA-see-CAUS-IM.P
'I made Kaina see the snake.'

(156) mykyky jacira n-ahpa-py-pia.

3PRO Jacira 3A-laugh-caus-IM.P
'He made Jacira laugh.'

In the cases described above, we can note that when an intransitive verb, such as

shout, rest, wake up, laugh, jump, throw up, and bleed, takes the causative, it changes its

valence, behaving as a transitive verb with the following structure: V[Intr +Caus [A O]].

Moreover, it was shown in example (155) that optionally when we have a CAUSEE, it is

followed by the agentivity particle (Meira 1999, Tavares 1995).

Interestingly, in (142) Aa Kaina haripipiya[ PRO kiriwu inise], we can observe a

kind of control structure: Kaina can control the subject PRO of the complement. As a
 102

result, the old subject Kaina becomes an indirect object. However, as in Japanese, in

Waimiri Atroari the causers (as initiators of the events) are generated in the SPEC of the

event Phrase (Harley 1995). To illustrate this assumption, I decided to use the tree Event

Phrase suggested by Harley (1995).

(157)
a) EyentP b) EyentP

NP Event' Aa Event'
A A
Cause EventP Kaina EventP

(NP) Event' ause Event'

/X /X
Cause/Happen VP PRO VP

NP V Kyrywu inise

In this structure, Aa is the CAUSER, having the highest position in the tree, or

being generated in the first Event Phrase; Kaina is the CAUSEE; the lexicalized

cause/made verb was put in the second Event Phrase; and finally the VP kiriwu inise is

what the causee has to do.

3.1.2.4.2 The Construction with the 'let' reading

One of the differences between the 'make' and 'let' causative is that in the 'let'

causative construction people are not forced to do something; as a result, this does not

imply an order and an accomplishment. Therefore, in these cases we can observe a

particle tre 'me that is used when we permit or order somebody to do something, but we

do not know if the person will do it. Levin (2000) argues that "causative and

accomplishment are independent notions." Consequently, in (158), it is not necessarily

 103

true that the person will learn how to make a ceramic pan or if in (159), the person will

leave to hunt.

(158) Aa wo'nj-e'me h-aminjaky-piany a-wenpa-typah tre'me tyruwa kapry pyky.

IPRO clay-VAL LA-permit/let-REC.P 2o-leam-? PART pan make how
'I permitted you to/let you dabble in the clay to learn how to make a ceramic pan.'

(159) Aa ka m-injaky-piany wyty ipy-na tre'me.

IPRO ka 2o-permit/let-REC.P meat look for-? PART
'I permitted you to/let you leave to hunt.'

The other difference between the two types of constructions is that in the 'let'

reading construction, we do not have the causative morpheme -pi. On the other hand, I

cannot assume that the particle tre 'me is particular to the 'let' reading causative because

we can have it as in the example below:

(160) Aak-aa-piany maryba taka a-iwapy-try pyky a-wenpa-typa tre'me.

lPR0 20-take-REC.P party/song AL 2o-sing-? how 2o-leam-? PART
'I took you to the party for you to learn how to sing.'

In this sentence, as in the sentences in (158) and (159), it is not necessarily true

that the person will learn to sing.

3.1.2.5 Miscellaneous Suffixes

3.1.2.5.1 Desiderative suffix

In Waimiri Atroari, constructions that denote desire are marked with the suffixes

-xy ~ -sy, as in the examples below.

(161) aa ram wen-ta-sy w-ia

IPRO 2PART vomit-VERBL-DESID 1-COP
'I want to vomit.'

(162) syna iee-sy m-ia

water drink-DESiD 2-CoP
'Do you want to drink water?'
 104

(163) syna iee-sy kra w-ia

water drink-DESiD IPRO 1-COP
T want to drink water.'

(164) ka ram tymere wu-sy na

3PRO 2PART jaguar kill-DESiD COP
'He wants to kill the jaguar.'

(165) k-e'me aa ia tymera wu-sy na

3PRO-VAL IPRO AGT.PART jaguar kill-DESiD COP
'He wants me to kill the jaguar.'

3.1.2.5.2 Instrumental nominahzer

In this language, there is a suffix -typy that can be attached to transitive or

intransitive verb stems, resulting in a noun. This noun denotes an instrument used for the

event described by the verb stem. For example:

(166) wykyry akyrypy i-akyny-typy i-wak-e

man trash REL-sweep-LNSTR.N REL-hold-T/A
'The man is holding the broom.'

(167) awaxi daxkia-typy ram aa=i-akyny minja

sugarcane squeeze-LNSTR.N 2PART IPOS-REL-brother thing/property
'The machine that squeezes sugar cane is my brother's.'

(168) k-ara mo pyky wenpa-typy

l+2pos-language sound how study-ESFST.N
'Phonology,' also translated as 'what is used to study sounds of our language.'

3.1.3 Adjectives

Adjectives are, most commonly, words that designate properties, qualities, or

states that are relatively stable over time. In the Waimiri Atroari language, adjectives are

free morphemes formed by only one root. They usually precede the noun. They are the

only word class that can take the suffix -pa 'emphatic.'

(169) a) xiwia b) xiwia-pa (170) a) kwada b) kwada-pa

 105

beautiful beautiful-EMPH ugly Ugly-EMPH

'really beautiful' 'very ugly'

(171) aa=i-akyna ram tamkwa tyska iyhia.

IPOS-REL-brother 2PART short straight hair
'My brother is short and has straight hair.'

(172) tapyryma iyhia,

black hair
'The hair is black.'

(173) ak-eme na tykoxinja-pa

pestle-DEV COP dirty-EMPH
'The pestle is dirty.'

(174) aa=papa ram sehsa tydapra

Ipos-father 2PART tall fat
'My father is tall and fat.'

Adjectives in this language can occur in subject and object position, as arguments.

(175) sehsa ram temere wu-pia

tall 2PART jaguar kill-LM.P
'The tall killed the jaguar.'

(176) temeraram sehs-eme wu-pia

jaguar 2PART tall-DEV kill-IM.P
'The jaguar killed the tall.'

The lexical category 'adjective' in Waimiri Atroari is not as clear as it is in

English or Portuguese. As is evident in the list below, in Waimiri Atroari some adjectives

can behave as a noun or have a nominal meaning.

3.1.3.1 List of Adjectives

Abemyhpa puffy
Aberymyhy-abemyhy round
Anykyxi thin
Ase new
Awinihe-awinini, awinjehe, awenjahky one, alone
Ba'xiri-be'xiri small
Bahinja-baxinja small, little, child
Bakyma sour, salt
Byryryma twisted
Djapyma straight, correct
Etypa hot
Inama weak
Kareme, karany good, beautiful
Kyby short for animals and plants
Kybyma bitter
Kykyryhpa burned
Mady-madyma cripple
Maiwu~maiy fat, thick
Masara lazy, apprentice
Mixopy long, lengthy
Nerimy brave, courageous
Nyryny alive
Panapy hardworking
Panaxi headstrong
Pine short
Pitymy single
Sakra white
Sakyna angry
Saweny-sewuna light
Sehe tall
Ta'kwa short for person
Taha big, large, leader
Tamxa soft, lazy
Tapany~typany~tipyna hard
Tapyryma black, dark
Temyna dry
Texiba sad
Teximy bad, not good taste
Tirika strong, potent
Trewine~trewuna agile, fast
Tuwaka, happy, content
Tuwera ripe
Txamyry old, elder
Txanpa pregnant
Txika sharp
Txipikia ashamed
Txitymy single
Tybyska smooth
Tykomia cold
Tykoxinja dirty
Typyra stink, bad smell
 107

Waina thin for person

Woxypia heavy
Xiwia good, beautiful, yellow, red

The following sentence examples illustrate the difference in meaning for the

words that can be classified as adjectives and nouns:

(177) bahinja-^^baxinja small, little, child

(a) Tuwadja wehma ram bahinj-eme Damixiri wehp-eme kapy-ma.

Tuwadja arrow 2PART small-DEV Damixiri arrow-DEV not alike
'Tuwadja's arrow is smaller than Damixiri's arrow.'

(b) Dauna bahinja xiba myryky-piany araky.

Dauna few fish fish-REC.P today
'Dauna caught few fish today.'

(c) bahinja n-itxi-pia mydy i-akyna-se.

children 3S-go-IM.P house rel-sweep-IN ORDER TO
'The children went to sweep the house.'

(178) taha big, large, leader

(a) Paryry kaapa ram taha Xeri Kaapa wata'.

Paryry field/garden 2PART big Xeri field/garden as...as/like
'Paryry's field is as big as Xeri's field.'

(b) pana a'a n-itxi-piany taha kanuwa ta warara hi pipe-se

Yesterday 1+3PRO l+3s-go-REC.P big canoe LOG turtle egg look for-in
'Yesterday, we went in the big canoe to look for turtle eggs.'

(c) Wame Kwene taha n-idjaka-pa Pyraty wu-pa

Wame warrior leader 3A-kill-REM.P Pyraty kill-REM.P

Piawuna many Kiawuna many

Piawuna and Kiawuna and
'Wame, the warrior leader, killed Pyraty, Piawuna, and Kiwauna too.'

This symbolis indication of free variation; we can fmd both forms. It is not necessarily related to
dialectal difference.
 108

(179) awinini~awinihe~awinjehe one, alone

(a) awinihe petxi ka-ky ampa la.

one wild pig talk-PAST other to
'One petxi talked to the other.'

(b) awinih-pa ka kinja txi-pia itxi taka.

alone-EMPH EVID person gO-IM.P jungle AL
'The kinja went to the jungle alone.'

(180) xiwia good, beautiful, yellow, red

(a) myda ram xiwia.

house 2PART beautiful
'The house is beautiful.'

(b) kyrywy xiwia ram mixopa.

snake red 2PARt lengthy
'The long red snake.'

(181) txamyry old, elder

(a) aa=mydy be'me asko ia ky amohbeh ram txamyrym-pa.

Ipos-house DEV ubim leaves ? made of 2PART OM-EMPH
'My house made of ubim leaves is old.'

(b) ky txamyra ram ebam-pa.

this elder 2PART blind-EMPH
'This elder is blind.'

Moreover, the words bahinja and txamyry can take the suffix -pesa esa that

means 'in time.' Two examples are bahinja-pesa 'in time of child-childhood' and

txamyry-pesa 'in time of elders, old age' (see examples 182 and 183 below). However,

the words taha, awinini, and xiwia cannot take this suffix. Therefore, I cannot say *xiwia-

pesa 'in time of beauty' or *awini-pesa 'in time of loneliness'.

(182) ianana myryky mepri soh-kwe-pa bahinj-esa

lanana son tapir fiir-remove-REM.P child-in time
'Ianana's son in childhood removed tapir's hair.'
 109

(183) k-eme ram ny-ryma-tape txamjry-pesa

3PR0-DEV 2PART 3s-die-REM.F elder-intime
'He will die in old age.'

In addition, I would like to discuss the semantic features of other adjectives

provided in the list, for example, the word tamxa 'soft, lazy'. Originally, this word only

meant soft. However, it seems that because of non-native influence, this word has

acquired a new meaning, 'lazy'. Today, the Waimiri Atroari use the word tamxa to

indicate a person who is not hardworking and is lazy. Not only does it mean a quality of

material, but it is now also a characteristic of a person. Finally, the other word in the list

that has a very curious semantic feature is the word xiwia. This word can mean good,

beautiful, yellow, or red (warm colors). Interestingly, the Waimiri Atroari people love

anything yellow and red. Therefore, it makes sense that xiwia also means beautiful.

3.1.4 Adverbs

Traditionally, adverbs are defined as a class of words which modify verbs,

adjectives, and clauses, typically expressing notions of time, location, maimer, degree,

and circumstances. Adverbs do not inflect for person or tense/aspect. They cannot be the

possessor nor the possessed in possessive constructions. They cannot occupy subject or

object positions. Another characteristic that makes adverbs different from other word

classes is its mobility in the clause. Syntactically, adverbs function as adjxmcts in any

type of clause.

Waimiri Atroari has many words fitting into this category. For example, time

adverbs are most commonly independent words referring to the time when the events/

actions occurred: araky 'today, now,' pana 'yesterday, long ago,' aminjaky 'tomorrow,'
 110

kyky 'at night,' and kokyny 'early.' Location adverbs illustrate where the action/event

takes place: mie 'far away,' kypy 'near, close,' ipyna 'there,' iky 'under.' A few adverbs,

such kokyny 'early' and waha 'many,' can take the emphatic suffix -pa that occurs in

adjectives. For example, it is possible to say kokyny-pa 'very early.' The following

examples illustrate some sentences with adverbs.

(184) wera ram minja pitx-e araky

woman 2PART manioc peel-T/A now/today
'The women are peeling the manioc now.'

(185) aa w-y-sapa mamyhkypa arakypahky ipyna

IPRO Is-go-T/A tomorrow again there
'I will go there tomorrow again.'

(186) danja n-aryn-e waha itxi ta

cicada 3S-talk-T/A many/a lot jungle LOC
'The cicada makes a lot of noise in the jungle.'

(187) a'a n-itxi-piany mie wyty i-po-se

1+3PR0 l+3s-go-REC.P faraway meat REL-look for-in order to
'Far away, we went to hunt.'

(188) tahkome wapy n-oo-sapa kamakaxi taka xirikiki

elder many 3s-climb-T/A Sp.oftree AL parekeet

baka-paiky
kill-T/A
'Many elders climbed in the kamakaxi trees in order to kill the parakeets.'

(189) topy iky ta ikere'e i-akaha ehry-ky

stone under LOC alligator REL-drawing paint-IMP
'Paint the alligator that is under the stone!'

3.1.5 Postpositions

Traditionally, postpositions are claimed to occur after NPs to form postpositional

phrases. They also appear as part of PPs as adjuncts in non-copular clauses. Unlike other

languages of the family, such as Tiriyo, Macuxi, and Carib, in Waimiri Atroari, except
 Ill

for the dative, the words classified as postpositions do not take person-marking prefixes.

Therefore, in this grammatical sketch I call the postpositions that do not take the person-

marking morphology of postpositional particles: Instrumental, Locative, Ablative, and

Allative.

(190) Instrumental - This posposition is used to relate an instrument to an event.

(a) kynetxiri ram irie pana xiky-pia maia ke

Kynetxiri 2PARt Irie ear cut-iM.p knife INSTR
'Kynetxiri cut Irie's ear with a knife.'

(b) pana mateus imeses-eme wu-piany wiwe ke

yesterday Mateus bat-DEV kill-REC.P wood INSTR
'Yesterday, Mateus killed a bat with a wooden stick.'

(c) impa warakypa i-apremy peri i-kysa-pa wiwe ke

then lucky REL-owner door REL-measure-REM.P wood INSTR
'Then the lucky person measured the size of the door of lanana's house with a
piece of wood.'

(d) a'a mepri wu-piany makuwa ke

1+3PR0 tapir kill-REC.P rifle INSTR
'We killed the tapir with a rifle.'

(e) iakypa a'a n-ikin-pia syna ke paxa taka

after that 1+3PRO 1+3S-wash-LM.P water INSTR bowl AL
'After that we washed it with water and put it in the bowl.'

(191) Locatives - The main use of this postposition is to indicate where the action/event

is taking place or to indicate the position (inside, on, above) of the event.

(a) itxa ta ram waha akenbehe na

jungle LOC 2PART many armadillo COP
'In the jungle, there are many armadillos.'

(b) wera ram kaapa ta na

women 2PART garden LOC COP
'The women are in the garden.'
 112

(c) bahinja akra' myryk-e myda ka

children acara fish fish-T/A small river LOC
'The children fish acara in the small river.'

(d) ka aa-ima-py-pia syna kaka

3PRO lo-jump-CAUS-IM.p water LOC
'They were made to jump in the water.'

(e) impa a'a n-ike'ia-pa meie impa-ry axinjaty tyhnaka

then 1+3PRO l+3A-toast-REM.P bread then-? stone gridle on
'Then we toasted the manioc flat bread on the stone gridle.'

(f) kyky-pesa a'a n-arma-pa samka tyhnaka

night-attime 1+3PRO l+3S-come back-REM.P hammock on
'At night, we returned to the hammocks.'

Observe that all the locatives begin with the phonemes /k/ or III. The locative ta

seems to be used only on surfaces or places that do not have water. On the other hand, the

locatives ka and kaka in my data-base only occur with situations that involve water.

(192) Directionals - The directional postpositions indicate the motion toward their

objects or indicate the goal. Here, I divided them into two types; the allative taka and the

ablative tany.

(a) weri n-itxi-pia itxi taka ie'ry i-apykwa-se

woman 1+3S-go-IM.P jungle AL fruit REL-collect-in order to
'The womem went to the jungle to collect fruits.'

(b) impa ianana ny-bia-pa ty-mydy taka

then ianana 3S-come-REM.P 3REFLX-house AL
'Then Ianana came into his own house.'

(c) maryba taka kara w-itxi-pia

party AL IPRO IS-go-iM.P
'I went to the party.'

(d) apiaiake xeri tany m-o-pi-e'?

when xeri ABL 2S-come-LM.P-LMP
'When did you come from the xeri village?'
 113

(e) kyky ram h-irima-tape maryba tany

1+2PR0 2PART l+2S-rest-REM.FUT party ABL

t-aryma-hkypa
3REFLX-come back-after that
'We will rest after we return from the party.'

(f) iawara mydy tany ka ky

iawara village ABL 3PRO ?
'He is from lawara's village.'

Note that the directional postpositions also begin with /t/.

(193) Dative - This postposition is used to indicate the indirect object or the

beneficiary. As mentioned at the beginning of this section, the dative postposition is the

only one that takes the same affixes that mark case and possession in Waimiri Atroari

(see more detailed discussion in section 3.2 Case marking).

(a) byba ram xiba i-ry-pia aa=inaka

3 PRO 2PART fish REL-give-IM.P 1-DAT
'He gave fish to me.'

(b) byba xiba i-ry-pia a=inaka

3PRO fish REL-give-IM.P 2-DAT
'he gave fish to you.'

3.1.6 Particles

Particles in Waimiri Atroari can be identified by their lack of any morphological

possibilities, either inflectional or derivational. Syntactically and semantically, in

Waimiri Atroari the particles can be divided into various subclasses. For example, the

second-position particle ram is defined by its position after the first syntactic constituent.

Subordinate particles such as impa 'then,' imka 'when, if,'and the lexical negative

particle constitutes by themselves utterance , for example, ie, iahe 'no,' wan 'no,' kap,
 114

kapy 'neg,' ie'xeme 'all.' The agentivity particle that marks the most agentive person in

the event is discussed below in the case marking section.

In this section, I illustrate examples of the evidential particle ka, the second-

position particle ram, the negation particle kap, kapy, and the conjunction particle many

'and, too, with.'

3.1.6.1 The evidential ka

This particle is used to demonstrate and express that the participants of the

event/actions have evidence for what they saying. They do not have any doubt about the

statements. See examples below:

(194) kwata ka n-yty-e kw-e na

spider monkey EVID 3S-yell-T/A say-T/ACOP
'The spider monkey is yelling, I am saying.'

(195) iakere ka by
alligator EVID this
'This is an alligator.'

(196) a=imeka-ty ka bixuwe-me by-pia

2pos-daughter-pos EVID bowl-DEV break-LM.P
'Your daughter broke the bowl.'

(197) iry ka aa=kaka-ky

3PRO EVID lo-scratch-PAST
'He scratched me.'

Like the ram particle, this particle only occurs after the first constituent of the

utterance. These are characteristics observed by Hoff (1985) for evidentials in the Carib

language.
 115

3.1.6.2 The ram particle

In Waimiri Atroari, the particle ram can be utilized to identify constituents. Here,

in this grammatical sketch, I will not discuss this function; rather, I will illustrate its

occurrence. As you can see in the examples below, ram can appear after NP, VP, and PP

structures.

(198) damixiri mydy ram xiwia

damixiri house 2PART beautiful
'Damixiri's house is beautiiful.'

(199) anjinji samka ram tahsa mixopa karana

this hammock 2PART big lengthy good
'This hammock is big, long, and good.'

(200) wykyr-eme wu-piany ram tymere

man-DEV kill-REC.P 2?ART jaguar
'The jaguar killed the man.'

(201) itxa ta ram waha akenbehe na

jungle LOG 2PART many armadillo COP
'In the forest there are many armadillos.'

3.1.6.3 Negation particle

In this section, I discuss non-verbal negation. These are negation sentences that

are not marked in the verb, but rather marked by particles. In Waimiri Atroari, kap ~ kapy

~ kapa and \van ate the particles that indicate negation. These particles are usually used

to negate existence.

(202) aiana ram wyty kapa

sp.bird 2PART meat/food NEG
'Anu-preto is not food.'

(203) wyty wan naminja

meat NEG dog
'Don't eat the food, dog!'
 116

(204) iakypa ny-tyn-pa kinja many kapy te'xy n-eeni-pa

then 3S-go-REM.P people with NEG DESID 3S-stay-REM.P
'Then she went off, she did not want to stay with the people.'

(205) kaapa ta ram minja wan na

garden LOG 2PART manioc NEG COP
'There isn't manioc in the garden'

(206) kanuwa ta ram wan epikiri na

canoe LOG 2PART NEG oar GOP
'In the canoe there is not an oar.'

3.1.6.4 The conjunction particle many

This particle is used to indicate inclusion, corresponding to English 'and'

Portuguese 'too, also, and.' It also has a comitative meaning (with).

(207) pinapa tete tahkome pyna mepri ny-tyta-pa pakia many

close ? elder place tapir 3S-come-REM.P wild pig and
'The tapir and the wild pig came close to the elder's place.'

(208) aa=mama ram typohinji xibamyryky many tyruwa ka-sa

Ipos-mother 2PARt like fish fish too ceramic pan make-T/A
'My mother likes to fish and to make ceramic pans.'

(209) kaapa i-akytx-iany kra aa=i-aska many

garden rel-cut-t/a Ipro Ipos-relative with
'I was preparing the plantation garden with my relatives.'

(210) impa nysakome iaxinjaty-rypy i-kynk-epa mepry se-py

then eldery woman stone gridle-? REL-break-T/A tapir foot-?

tyruwa many
ceramic pan too
'The tapir's foot touched the stone griddle and broke, the ceramic pan too

3.2 Case Marking in Waimiri Atroari

This section describes and analyzes case marking in Waimiri Atroari. This

analysis explores whether Waimiri Atroari has the typical inverse split-S system
 117

demonstrated by Gildea (1998) for 19 languages in this linguistic family, or whether

Waimiri Atroari presents a typical nominative/accusative system. Moreover, it discusses

whether in this language the feature of ergativity occurs.

The organization of this section is as follows. Section 3.2.1 gives some

background on the Waimiri Atroari case marking system. Section 3.2.2 provides a

description of the data and a discussion of Gildea's classification (1998) about

morphosyntactic properties of the verbal system from a group of languages in the Cariban

family. Moreover, this section provides some examples of the ditransitive construction. In

section 3.2.3, a discussion about the agentivity particle ia is provided in order to verify if

it can be utilized as an ergative. Final remarks are developed in section 3.2.4.

3.2.1 Background

I am using Blake's definition of case (2000:12) in which case in its most central

manifestation is a system of marking dependent nouns for the type of relationship that

they bear to their head. However, it is not the only grammatical mechanism for marking

head-modifier relation. One important type of alternative is the principle of marking the

head rather than any dependent. As it is analyzed, this is the case of Waimiri Atroari. By

this 1 mean that in this language, nouns do not get any affixes that can be identified as

case marking (nominative/accusative or ergative/absolutive).

For Waimiri Atroari transitive verbs, both A and O marking prefixes occur (see

Table 1 in Chapter 1). Their distribution is conditioned by a person hierarchy that ranks

non-third persons (first person '1,' second person '2,' and first person dual '1+2') as

higher than third persons (speaker/hearer>non-speaker/hearer). Basically, whenever a

 118

participant (either A or O) is non-third person, it is marked on the verb with the

corresponding A or O prefix. If both participants are third persons, it is possible to get an

n-prefix or a zero-prefix (see discussion below). The higher person will be marked

overtly on the verb (1, 2, 1+2, l+3>3).

When first person, second person, and first person plural inclusive act on the third

person, the subject prefixes will be chosen, agreeing with the first or second person.

1A30 Aa ram ka h-ini-pia.

IPRO 2PART 3PRO lA-see-iM.P
'I saw him.'

2A30 Amyraram ka m-ini-pia.

2PRO 2PART 3PRO ^-see-iM.p
'You saw him.'

2A30 Amyra ram aa-papa ia tymere wo m-ini-pa.

2PRO 2PART IPOS-father AGT.PART jaguar kill 2A-see-REM.P
'You saw my father kill the jaguar.'

However, when third person acts on first and second person, the object prefixes

will be selected, still agreeing with the first and second person.

3A10 Ka ram aa=ini-piva.

3PRO 2PART lo-see-IM.p
'He saw me.'

3A20 Ka ram a-ini-piya.

3pro 2part 20-see-imd.past
'He saw you.'

3.2.2 Gildea's classification of morphosyntatic properties of the verbal system

According to Gildea (1998), the Cariban family presents seven different

independent clause verbal systems, such as Set I (inverse/split-S), Full Set II (ergative).
 119

Partial Set II (ergative). Progressive (nominative), De-ergative (nominative), t-V-ce

(ergative), and t-V-ce-mi (nominative). He says that

"the seven systems are identified by means of six distinct, but interrelated,

morphosyntactic properties: forms and patterns of verbal personal prefixes

and suffixes, verbal tense-aspect-modality (TAM) suffixes, nominal case-

marking patterns, word order restrictions, existence and agreement

patterns of auxiliaries, and forms and morphological placement of the

collective number markers." (Gildea 1998:15).

Gildea claims that the nominative languages have retained the original system and

that ergativity is an innovation, resulting from the reanalysis of old nominalizations as

main verbs.

Taking Gildea's classification, Waimiri Atroari is classified in the 'Set I System

(nominative or inverse/split-S)'. According to him this set shares some characteristics,

such as personal prefix set and the collective number suffixes; word order is generally

nominative, in that the OV unit is clear; there are no auxiliaries and A and O nominals are

not case-marked. Gildea (1998:59) says that in all Set 1 systems for which we have

evidence, the A may occur either preceding or following the OV unit. As with Carib of

Surinam (Hoff 1978), in Waimiri Atroari (Bruno 2001) the more neutral order (in terms

of both frequency and markedness) is preverbal; AOV. Moreover, personal prefixes in

the Set I system identify the subject of an intransitive verb (S) and both the subject (A)

and (O) of a transitive verb. In addition, this set has the inverse prefixes (all those in

which 3A acts on SAP-Speech Act Participants '1,' '2,' and '1+2' O) that are identical to
 120

the subset of the intransitive subject prefixes that he calls So; the direct prefixes (those in

which SAP acts on 30) are most similar to the Sa subset of the intransitive prefixes.

Although I agree with Gildea's classification in many aspects and I can see that

Waimiri Atroari has some of the characteristics of the Set I system, I still have to verify

more accurately whether Waimiri Atroari follows this inverse split-S pattern. Van Valin

(1990), using the Role and Reference Grammar (RRG) fi"amework, proposed a semantic

analysis for the intransitive verb classes of Italian and Georgian, according to which

'activity' intransitive verbs belong to the Sa category while 'non-activity' ones are in the

So category. Meira (1995) observed that the intransitive verb case marking in some

Cariban languages seems to follow Van Valin's assumption, dividing the verb class into

ones that denote activities versus those which do not denote activities.

Particularly, I need to look with more detail into this possibility for Waimiri

Atroari. The data that I have do not clarify the distinction between intransitive verbs that

take different first-person singular prefixes depending on semantic content (activity

versus non-activity). As can be seen from the examples in table 3.12, the distinction

between verbs denoting activities versus non-activities does not seem to work for

Waimiri Atroari. For example, the first person in the paradigm with the verb 'shout' takes

the prefix fi-om the object set. Here, we have two problems: this is supposed to take an A

prefix instead of an O set because it is classified in the Sa set; the other problem is related

to the fact that we expected that the second person in this paradigm would take the prefix

from the O set, as well. Therefore, before I claim that Waimiri Atroari has the same

Inverse Split-S system demonstrated by Gildea (1998) for the Set I, I need to collect more
 121

paradigms with intransitive verbs. On the other hand, according to Dixon (1979:83), there

are languages of the Split-S-marking where the two intransitive subclasses do not have as

good a semantic basis, such as Mandan and Guarani. This could be an explanation for the

situation in Waimiri Atroari. In this language, the tense/aspect device seems to not

influence the inverse system in the examples below.

Table 3.12^Activities versus non-activities

Activities - Sa Non-activities - So
Go '-y-' Bleed'^ '-myn-ta'
w-y-sa 'I go' wu-myn-ta-pa 'I bled'
m-y-sa 'you go' mu-myn-ta-pa 'you bled'
n-y-sa 'she/he goes' nu-myn-ta-pa 'she/he bled'
h-y-sa 'we go (1+2)' hu-myn-ta-pa 'we bled (1+2)'
n-y-sa 'we go (1+3) nu-myn-ta-pa 'we bled (1+3)

Shout '-kyta-' Know '-py-'

aa=kyta-ky 'I shouted' hy-py-sa 'I know'
my-kyta-ky 'you shouted' my-py-sa 'you know'
ny-kyta-ky 'she/he shouted' ny-py-sa 'she/he knows'
hy-kyta-ky 'we shouted' (1+2) hy-py-sa 'we know' (1+2)
ny-kyta-ky 'we shouted' (1+3) ny-py-sa 'we know (1+3)

Jump '-tahkwa-' Die '-rymy-'

wy-tahkwa-ky 'I jumped' wy-rymy-tape 'I will die'
my-thakwa-ky 'you jumped.' my-rymy-tape 'you will die'
ny-tahkwa-ky 'she/he jumped' ny-rymy-tape 'she/he will die'
hy-tahkwa-ky 'we jumped (1+2)' hy-rymy-tape 'we will die' (1+2)
ny-tahkwa-ky 'we jumped (1+3) ny-rymy-tape 'we will die' (1+3)

Descend '-ooty-' Dream '-inin-'

w-ooty-piany 'I descended' w-ininy-pa 'I dreamed'

m-ooty-piany 'you descended' m-ininy-pa 'you dreamed'
n-ooty-piany 'she/he descended' n-ininy-pa 'she/he dreamed'
h-ooty-piany 'we descended'(1+2) h-ininy-pa 'we dreamed' (1+2)
n-ooty-piany 'we descended'(l+3) n-ininy-pa we dreamed' (1+3)
There are two sets of first person singular subject prefixes; w-/wy-Avu- and h-/hy-

/hu-. The first set is used with strictly intransitive verbs such as rymy 'die', y 'go', tahkwa

The word for blood is myny, in order to get the verb 'bleed,' we add the verbalizer -ta.
 122

'jump', and mynta 'bleed'. The second set is used with transitive verbs with personal

objects such as ini 'see' and aape' 'push', and transitive verbs with impersonal objects

such as py 'know', ee 'drink', and wenta 'vomit'. (The Classification of certain verbs

such as mynta 'bleed' (intransitive) and wenta 'vomit' (transitive) is not explained.) This

distinction is not made for any other person and number combination.

In addition, at least one intransitive verb kyta 'shout' takes the first person

singular object clitic prefix aa=, for which I also have no explanation.

The examples below illustrate more clearly the hierarchy discussed above and

demonstrate how case is marked. It is observed that usually head-marking languages

mark the subject's grammatical relation on the verb, but they can mark the object as well

in some languages. This happens in Waimiri Atroari when the objects are first and second

person.
 123

Table 3.13Verb -ini- 'see'

1A30 Aa ram ka h-ini-pia.
IPRO 2PART 3PRO 1A-see-IM.P
'I saw him.'
2A30 Amyra ram ka m-ini-pia.
2PRO 2PART 3PRO 2A-see-lM.P
'You saw him.'
3A30 Mykyka ram ka 0-ini-pia.
3PRO 2PART 3PRO 0-see-IM.P
'She/he saw him/her.'
1+2A30 Kyka ram ka h-ini-pia.
H-2PRO 2PART 3PRO l+2A-see-LM.P
'We saw him.'
3A10 Ka ram aa=ini-pia.
3PRO2PART lo-see-iM.p
'She/he saw me.'
3A20 Ka ram a=ini-pia.
3PRO2PART 2o-see-IM.P
' She/he saw you.'
3A1+30 Ka ram a^=ini-pia.
3PRO2PART l+30-see-LM.P
'She/he saw us.'
3A1+20 Iry k-ini-pe-sy na.
3PRO l+2o-see-?-DESID COP
'She/he wants to see us.'
1A20 Aa ram k-ini-pia.
Ipro 2part 2o-see-im.p
'I saw you.'
2A10 Amyra ram aa=ini-pia
2PRO 2PART lo-see-lM.P
or
Amyra aa=k-ini-pia
2PRO lo-2A-see-]M.P
'You saw me.'

We can observe that the full pronouns of first, second, plural inclusive and

exclusive do not occur when they behave as objects. Preferentially, the free

speaker/hearer pronouns are subjects only. This demonstrates that the first-person plural

inclusive and exclusive are also above the third person in the hierarchy.
 124

When in Waimiri Atroari a third subject acts on a third object (3S30), it is not

totally clear whether the zero marks NOM or ACC; therefore, in the table of prefixes, I

hesitated to put a zero there. Usually, it is observed cross-linguistically that if the case-

marking paradigm includes a zero, it will normally be the NOM, but zero ACC is also

attested. In addition, if A aligns with S, then it will have to be O that is marked. Waimiri

Atroari seems to behave like that.

3.2.2.1 The Ditransitive Construction

According to Wunderlich and Lakamper (2001:378), morphological case appears

in two variants: as structural or semantic case.

Structural case (such as accusative, ergative, dative, genitive, and

sometimes also partitive) reflects the ranking of arguments, while

semantic case (such as, for example, instrumental, comitative, locative,

directional) encodes a semantic relation between DP and the governing

head. Structural case is typical of complements ... Structural case forms a

closed system, with dative as the most highly marked case, accusative

(genitive) and ergative as less marked.

They observed that is possible that semantic case and structural case compete with

each other, principally "if the number of arguments to be realized exceeds the number of

distinct structural cases. This happens with ditransitive verbs if the language has only one

object case available, and with causativized ditransitives in general because no languages

have three different structural object cases" (2001:379). As we can see in example (211),

part of their assumption seems to be confirmed, but here what is marked is the ACC.
 125

(211) Pamwe aa=iry-py-pia woky iry-ky Marta inaka.

ParuwelO-tell-CAUS-lM.P banana give-iMP Marta DAT
'Paruwe told/dictated to me to give the banana to Marta.'

(212) amyra ram woky my-ry-pia Kaina inaka.

2PRO 2PART banana 2A-give-lM.P Kaina DAT
'You gave the banana to Kaina.'

(213) ka katyba i-aa-pa t-akyna inaka

3PRO calendar REL-take-REM.P SREFLX-brother DAT

t-aska i-epa-se.
SREFLX-relatives REL-invite-in order to
'He took the ceremonial calendar to his brother to invite his relatives.'

On the other hand, example (211) is problematic because the two arguments of

the verb 'tell' are not both NPs. It shows one NP and one clausal argument, and clause

does not show case. Unfortunately, I could not find more examples of ditransitive verbs

to see what happens in this case. Example (213) is interesting because as we are dealing

with 3 A and 30, it seems that it does not get any marking.

In Waimiri Atroari ditransitive constructions, as demonstrated in Table 3.16

below, I confront a problem: that of assuming that the recipient must become the direct

object as in Lummi (Jelinek 1983). When the indirect object is a first, second, or first

plural exclusive, it is the dative that receives the prefixes instead of the verb. The verb in

this situation is marked with the relational prefix. In Waimiri Atroari, the dative is the

only postposition that gets the prefixes that mark person. This indicates a separate

predicate status for the DAT marker.

Suggestion given by Dr. Dooley and Dr. Hill (2003).

 126

Table 3.14Verb -ry- 'give'

1A30 Aa ram xiba hv-rv-pia ka inaka.
IPRO 2PART fish 1 A-give-iM.p 3PRO DAT
'I gave fish to him/her.'
2A30 Amyra ram xiba mv-ry-pia mykyka inaka.
2PRO 2PART fish 2A-give-IM.P 3PRO DAT
'You gave fish for him.'
or
Amyra ram ka inaka xiba mv-ry-pia.
2PRO 2PART3PRO DAT fish 2S-give-IM.P

*Observe that the dative can occur in the end of the sentence, but also in other
environments.
3A30 Mykyka xiba i-ry-pia mykyka inaka.
3PRO fish REL-give-iM.p 3PRO DAT
'She/he gave fish for him'
1+2A30 Kyka ram ka inaka xiba hv-rv-pia.
1+2PRO 2PART 3PRO DAT fish l+2A-give-LM.P
'We gave fish for him.'
or
Kyka xiba h^-ry-pia mykyka inaka.
1+2PRO fish l+2A-give-IM.P 3PRO DAT

3A10 Byba xiba i-ry-pia ^-inaka.

3PRO FISH REL-give-IM.P 1-DAT
'She/he gave fish to me.'
or
Ka ram ^-inaka xiba i-ry-pia.
3PRO 2PART 1-DAT fish REL-give-iM.p

3A20 Ka ram a-inaka xiba i-ry-pia.

3PRO 2PART 2-DAT fish REL-give-IM.P
' She/he gave fish to you.'
or
Byba xiba i-ry-pia a-inaka.
3PRO fish REL-give-IM.P 2-DAT

3A1+30 Ka ram ^-inaka xiba i-ry-pia.

3PRO2PART 1+3-DAT fish REL-give-IM.P
'She/he gave fish for us,'

1A20 Aa ram xiba a-inaka hy-rv-pia.

IPRO 2PART fish 2-DAT lA-give-IM.p
'I gave fish for you.'
2A10 Amyra ram aa-inaka xiba my-ry-pia.
2PRO 2PART 1-DAT fish 2A-give-IM.p
'You gave fish for me.'
 Interestingly in Table 3.16, we can note that in the last two examples, 1A20 and

2A10, the dative is marked with the object marker, but the verbs are marked with the

subject marker. Although we have observed these new cases, we can see that the

hierarchy still works. Moreover, the examples in Table 3.16 demonstrate that some

nominal expressions such as xiba 'fish' and the person who receives the fish do not need

to have a fixed order. The dative also does not have a fixed order; this made me rethink

the idea that the dative could be a core argument. Other aspects observed are that Waimiri

Atroari requires an external argument acting as subject; we cannot omit any of the

subjects given in the examples above.

Jelinek (1984), in her article 'Empty categories, case, and configurationality,'

explains that there is a distinction between pronominal clitics and nominal expressions

where independent pronouns are included. According to her, nominal expressions can be

considered adjuncts with non-argumental fijnction, in this sense nominal expressions, as

adjuncts do not need to have fixed order. I wonder whether this is true for Waimiri

Atroari inasmuch as we can note in the examples that the subject has a fixed order and

has an argumental fiinction. This aspect will be considered as a remaining issue.

3.2.3 The Agentivity Particle

Interestingly, Waimiri Atroari has a particle ia that tells which argument is the

participant that performs the action in the sentence. In the examples below, we can

observe that this particle occurs in different kinds of sentences and it is not necessarily

related to the causative construction. It can appear when it is necessary to identify the
 128

CAUSEE as the agent of the structure (see example 217), but it is not obligatory. As we

can observe, it also appears after nouns or pronouns.

(214) Ipaikypa naminja ia t-iika-hkypa wykyr-eme ny-tam-pia,

after dog AGT.PART REFLX-bite-after man-DEV 3s-cry-IM.P
'After the dog bit the man, he cried.'

(215) naminja ia aa=ika-hkypa kara wy-tam-pia.

dog AGT.PART lo-bite-after IPRO Is-cry-LM.P
'When the dog bit me, I cried.'

(216) naminja ia a-ika-hkypa my-tam-pia.

dog AGT.PART 2o-bite-after 2s-cry-im.p
'When the dog bit you, you cried.'

(217) aa ia aa=ika-hkypa ka my-tam-pia.

Ipro agt.part lo-bite-after EVID 2s-cry-LM.P
'When I bit you, you cried.'

(218) k-eme ia aa=ika-hkypa ka k-eme ny-tam-pia.

3PRO-DEV AGT.PART lo-bitc-after EVID 3PRO-DEV 3S-cry-IM.P
'He cried when he bit me.'

(219) aa ram Kaina ia kyrywu h-ini-py-pia.

IPRO 2PART Kaina AGT.PART snake IS-see-CAUS-LM.P
'I made Kaina see the snake.'

(220) ka ram aa-mama ia sanja iahkwa ini-piany.

3PRO 2PART IPOS-mother AGT.PART manioc flour make see-REC.P
'He saw my mother make manioc flour.'

(221) aa ram wyty pys-any itxi ta kyrywu huwa ia aa=iika-paiky.

IPRO 2PART meat look for-T/A jungle LOG snake ? AGT.PART lo-bite-after
'I was hunting when the snake bit me.'

(222) k-e'me aa ia tymera wu-sy na

3PRO-VAL IPRO AGT.PART jaguar kill-DESID COP
'He wants me to kill the jaguar.'

(223) daniel wyty bykwa-py-pia jacira ia roseli wytyme

Daniel food prepare-CAUS-LM.P Jacira AGT.PART Roseli food
'Daniel made Jacira prepare food for Roseli.'
 129

According to Gildea (1998) and Meira (1999), in the Carib family there is a

postposition ya (in the Proto-Carib w+ya) that marks different kinds of participants;

directionals, datives, causees, ergative, and agent markers. Meira (1999:512) explains

that the various kinds of participants that ya marks do seems to show some 'common

semantic threads'they are all human or sentient. In Tiriyo, the language described by

Meira, we can see in example (226) below that the distinction is not always clear.

(224) pahko ya wit-te-e.

1 father DIR lSa-go:PRS.LPF-CTY
'I am going to my father's (house, village).'

(225) maja wi-ri-po ii-ya.

knife lA-make-CAUS:PRS.PRP 3-Causee
'I had him make a knife.'

(226) pireu w-ekarama-po Asehpe ya Simetu ya.

arrow lA-give-CAUS:PRS.PRF Asehpe CAUSEE/DAT? Simetu Causee/Dat?
'I made Asehpe give the arrow to Simetu' ~ 'I made Simetu give the arrow to
Asehpe.'

Gildea (1998:121) claims that "the Cariban system of nominalization is ergatively

organized, with the sole genitive relationship to the nominalized verb being claimed by

the verb's notional absolutive argument (S and O), and the notional ergative argument

(A) being therefore forced into oblique status." According to him, in the northern Full Set

II dominant languages, such as Makuxi and Kapon, the goal/dative function of this

morpheme has been lost, leaving only the ergative agent-marking fianction (see examples

227 and 228).

Makuxi-(227) [t-ekkari aret+'ka-sa'-t+u-ya] yai aw-enna'po-'p+.

3.Refl-food finish-Nomlzr-3Refl-ERG at 3-retum-TAM
'When he finished his food, he returned.' or
(lit. 'at the finishing of his food by himself he returned').
 130

Kapon-(228) [makonaiama y-akwar-ri ota-0] eyne-pu 0-ia

god Rel-spirit-Psn descend-Nmlzr see-TAM 3-ERG
'He saw the spirit of God descending.'

As we can observe in the Waimiri Atroari examples (214-223) above, the

morpheme ia does not mark datives and directionals. It clearly marks agents and causees.

In two clauses with two different actors, ia appears after the actor who denotes more

agentivity. Although I have an example of ia in a clause that seems to be nominalized, I

would prefer to explore and get more data before claiming that in this situation ia marks

ergativity.

(229) Amyra ram aa-papa ia tymere w-o m-ini-pa.

2PRO 2PART iPOS-father AGT.PART jaguar kill-NOMLZ 2S-see-REM.P
'You saw my father kill the jaguar' or 'You saw the death of the jaguar by my father.'

Another interesting aspect of the particle ia that can be a topic for further research

is related to the notion of agency. According to Mithun (1991:516), the prototypical agent

is considered the "participant which performs, instigates, or controls the situation denoted

by the predicate," In this sense, this assumption seems to demonstrate that agentiveness

as performance/instigation and control/volitionality is not perfectly adequate for Waimiri

Atroari.

The notion of semantic agency is a complex one, as explored in Foley & Van

Valin (1984) and DeLancey (1985), among many others. Foley & Van Valin (1984:29)

characterize their general category 'actor' as "the participant which performs, effects,

instigates, or controls the situation denoted by the predicate," features shared by

prototypical agents. They characterize their general category 'undergoer' as "the

 131

participant which does not perform, initiate, or control any situation but rather is alfected

by it some way" (Foley & Van Valin 1984:29), features shared by prototypical patients.

In Waimiri Atroari, as demonstrated in examples (219) and (220) above, there is a

participant who instigates and controls the action and another who performs the action.

The participant who performs or does the action will be followed by the ia particle.

Interestingly, the participants who will perform the action do not necessarily have

volition or control over the situation.

3.2.4 Final Remarks

Ergativity is a system or pattern of casemarking which casemarks A as ERG and

S/0 as ABS (Blake 2001). Although ergativity is not a unified phenomenon, I prefer to

do more research in order to claim that Waimiri Atroari shows any feature of ergativity.

In Waimiri Atroari, S and O share no formal features, as we would expect if they were

"absolutive" elements, so I do not treat them as the same. Moreover, even when this

particle ia only occurs after A, I question if we can use it as an ergative marker.

Moreover, it is necessary to collect more data in order to verify whether Waimiri Atroari

follows the Inverse Split-S system observed by Gildea (1998). The selection of the

prefixes in intransitive verbs do not seem to be conditioned by tense or aspect where the

ergative is always found in either past tense or perfect tense (Dixon 1979:95). It also does

not seem to follow the activity versus non-activity distinction.

In the Waimiri Atroari language, first, second, and first plural inclusive and

exclusive person are ranked higher than third person. However, when second person acts

on first or first acts on second, it was observed that in some cases we have subject
 132

agreement, while other cases show object agreement. Therefore, subject and object

marking must follow this hierarchy: l=2,l+2/l+3>3. Although we have observed that full

internal pronouns as object were not obligatory, this is not true for the external

arguments. A remaining issue that needs to be studied in more detail is the behavior of

the ditransitive verbs in order to better understand the role of the indirect object. Clearly,

the dative does not function as a core argument.

 CPIAPTER 4- SYNTAX

4 Phrase Structure, Clauses, and Word Order in Waimiri Atroari

This chapter is organized as follows. Section 4.1 describes and analyzes phrase

structure in Waimiri Atroari. Section 4.2 provides a discussion about Waimiri Atroari

clausal order. Section 4.3 analyzes the topicalization construction in this language, and

finally section 4.4 provides a brief analysis of two types of subordinated clauses.

In this grammatical sketch, I argue that Waimiri Atroari is a head-right (or head-

final) language. However, there are cases in which the head is apparently allowed to

have variable position (either left or right side). As I intend to show, all these examples

seem to involve the presence of adjunctssuch as noun phrases containing adjectives,

adverbial quantifiers, and numeral words. It has been observed that adjuncts are different

fi-om complements in that they have a higher degree of positional fi"eedom so that this

apparent 'mobility' of the head could alternatively be explained as a result of the

positional fluctuation of the adjunct. Moreover, as already observed by Vieira

(1995:701) in Asurini (Tupi-Guarani family), in Waimiri Atroari quantifiers such as all,

many, and two do not belong to the functional category of determiners. Therefore, the

similarity in distribution of the head among the phrases containing adjectives, adverbial

quantifiers, and numeral words leads me to define Waimiri Atroari as a head-right

language.

As I have already mentioned, I claim that S(0)V is the basic word order, based on

statistical frequency, pragmatic factors, and descriptive simplicity. However, different

kinds of order are also attested, such as (0)VS, SVO, and (0)SV. Again, the occurrence
of word orders like SVO seems to constitute a counterexample to the claim that Waimiri

Atroari is a head-right language. Although I will not be dealing with such examples in

this paper, I suggest that a possible explanation could be related to the influence of

Portuguese in the speech of the younger speakers. The OSV and OVS word orders will

be analyzed as being the result, respectively, of the movement of the object and the whole

VP to a topic position,

4.1 Phrase Structure in Waimiri Atroari

Here, I assume that phrases are built around an element whose head is instantiated

by a major lexical class, such as N, V, or A. Second, I assume that there are at most two

projections of each class, an intermediate projection X' and a maximal projection XP,

and we can add adjuncts at any level. Since X-Bar Theory allows "Parameters''^^ (Travis,

1989:264) on the position of heads, complements, and adjuncts, I use it to explain the

phrase structure in this language. According to Greenberg (1963), there is a general word

order tendency in natural languages that tends to place modifier elements either before or

after the head. On the other hand, it is observed that the position of heads and

complements in different kinds of phrases seems not to be limited to the binary choice

where all heads must be either left or right. In other words, there could be some 'mixed

head languages,' for example Basque (Radford 1988:39).

In Waimiri Atroari, the head of the phrase occurs predominantly at the right edge

of the constituent in noun, verb, and postposition phrases. However, in the case of noun

According to Travis (1989;264), "[LJanguage variation is allowed through parameters which introduce a
limited flexibility to the system. Parameters represent the range of variation that can be found in natural
languages as well as what has to be learned by the children."
 135

phrases containing a numeral, quantifier word, or adjective, this does not seem to always

be the case. As will be demonstrated below, the fact that the head can be positioned

either to the left or to the right in this kind of phrase is probably related to the fact that

numerals, adverbial quantifiers, and adjectives are adjuncts, and, as such, can occupy

variable positions.

4.1.1 Noun phrases

In Waimiri Atroari, the simplest case of noun phrases can have a single noun (1)

or a pronoun (2). The clearest cases of NPs involving two nouns are examples of

possessive phrases, which present the order possessor-possessed (3) to (8):

(1) mabaia 'a/the papaya(s)' (2) amyry 'you'

(3) mydy i-apremy (4) Ewepe pyruwa

house REL-owner Ewepe arrow
'the owner of the house' 'Ewepe's arrow'

(5) maryba i-apremy (6) lawara mydy

song/festivity REL-owner lawara house/village
'owner of the song or of the party' 'the village of lawara'

Moreover, the examples involving noun phrases seem to have examples of both

adjunction and complements. In the trees in (7b) and (8b), the SPEC N is a possessor and

the head is the N'. Unlike English, where the possessive clitic '5 is attached to the

possessor NP, in Waimiri Atroari it is the possessed noun that receives the morphological

indication of the genitive relationship (head marking). Moreover, in cases such as

example (6) above, the noun-noun construction can be ambiguously interpreted as a

typical possessive phrase (i.e., 'the village that belongs to lawara') or a 'naming'

construction ('the village whose name is lawara'). Moreover, possessives will be always
 136

on the left because they will be either complement or specifier (Comp for inalienable and

Spec for alienable).

(7) a. Temere iee (8) a. Kaina i-yhia

jaguar tooth Kaina REL-hair
'The jaguar's tooth.' 'Kaina's hair.'

b. NP b. NP

NP N' NP N'
A 1 A 1
temere N
A
L 1
N

iee K^ina i-yhia

4.1.1.1 Noun phrases containing adjectives^"

As illustrated in the trees below, adjectives are adjunctsor, in X-Bar

terminologysisters of an N' and daughters of an N', Because adjuncts can be more

flexible in their distribution, in a noun phrase containing adjectives, the adjectives can be

Many Carib languages do not have 'adjective' as a syntactic class (part of speech). Words corresponding
semantically to adjectives are classified as nouns. As discussed in Chapter 3,1 claim that Waimiri Atroari
does have adjectives. Syntactically, adjectives can, like nouns, occur in subject or object position.
However, unlike nouns, adjectives cannot take the suffix -mi that indicates 'absence' (e.g. ety-my
'nameless'). Furthermore, only adjectives can take the emphatic suffix -pa (e.g. tamkwa-pa 'very short').
On the other hand, it is not clear whether one can use the second-position particle ram as a boundaiy
constituent in phrases of the type <Adj N>. I have to do more tests because it is not always the case that
the Waimiri Atroari consultants allow this kind of construction:

a. *kyrywy xiwia ram mixopa

snake red 2PART long
'The long red snake.'

b. wykyry sehe ram waryna wu-se txi-pia ipaikypa weri tamkwa

man tall 2PART paca kill-in.order.to go-IM.P after woman short

kymy i-eky i-akymy-se.

bacaba REL-juice REL-make-in.order.to
'The tall man went to kill paca (a kind of rodent) and the short woman went to prepare bacaba
fruit juice.'
 137

either to the left or to the right of the head noun. At this moment, I cannot determine

whether this variation is purely stylistic or whether it entails any semantic difference.

(9) a. xiwia mydy b. mydy xiwia 'beautiful house'

beautiflal house house beautiful

c. d.

N' N'

AcUP N' N' AdjP

xiwia N N xrma
'beautifiil'

mydy tydy
'house' 'house'

(10) a. taha kyrywy b. kyrywy taha

big/large snake snake big/large
'big snake' 'big snake'

(11) pana a'a n-itxi-piany [taha_kanuwa] ta

yesterday 1+3pro l+3S-go-REC.P big canoe LOG

warara bi pipe-se
turtle eggs look.for-in. order, to
'Yesterday, we went in the big canoe to look for turtle eggs.'

(12) [wykyry sehe] txi-pia waryna wu-se

man tall go-iM.P paca kill-in order to
'The tall man went to kill a paca (a kind of rodent).'

4.1.1.2 Noun phrases containing quantifier words

As with noun phrases containing adjectives, noun phrases containing quantifiers

also present a certain degree of positional variation. As shown in examples (13) to (17)

below, adverbial quantifiers can occur either to the left or to the right of the head noun.
 138

As the syntactic trees in (13c) and (13d) demonstrate, I consider such quantifiers as

adjuncts, which would explain their relative mobility.

(13) a. waha xiba xiba waha

many fish fish many
'many fish' 'many fish'

c. d.

AdvP AdvP

waha X X waha
'many' i | 'many'
xiba xiba 'fish'

(14) a. wapy kinja wyty ipo-piany

many people meat look.for-REC.P
'Many people hunted.'

b. kinja wapy wyty ipo-piany

people many meat look.for-REC.P
'Many people hunted.'

(15) tahkome wapy n-oosa-pa kamakaxi taka

elders many 3-climb-REM.P tree (sp.) AL

xirikiki baka-paiky
parakeet kill-T/A
'Many elders climbed trees to kill parakeets.'

(16) kinja wyty ipo-piany wapy.

people meat look.for-REC.P many
'People hunted a lot.'

(17) njawa nyn-pa waha kipety tarara many

rain come-REM many wind thunderstorm too
"It rained a lot, with wind and thunderstorm as well."
 Although examples (18), (19), and (20) are not examples of adverbial quantifiers,

I provide them to show that other kinds of adverbs behave in the same way, presenting

the same mobility.

(18) mamyhkypa a'a n-y-sapa kwata wu-se

tomorrow l+3prol+3S-go-T/A spider.monkey kill-in.order.to
"Tomorrow we will go to kill a spider monkey."

(19) a'a n-y-sapa kwata wu-se mamyhkypa

l+3pro l+3S-go-T/A spider monkey kill-in.order.to tomorrow
"We will go to kill a spider monkey tomorrow."

(20) a'a n-y-sapa mamyhkypa kwata wu-se

l+3prol+3S-go-T/A tomorrow spider.monkey kill-in.order.to
"We will go tomorrow to kill a spider monkey."

Interestingly, the position of an adverbial quantifier such as waha 'many, a lot'

and wapi 'many, a lot' seems to be free when they modify a noun phrase (examples 13-

15). However, these adverbial quantifiers seem to occur preferentially in post-verbal

position when modifying a verb phrase (examples 16 and 17). Waimiri Atroari lacks

determiners that correspond to each, every, most, and some, a fact that suggests the

absence of a class of D-quantifiers^^ in this language. As mentioned in the introduction,

in Waimiri Atroari quantifiers such as all, many, and two do not belong to the functional

category of determiner; therefore, I prefer to think of them as adverbs.

According to Bach et al (1995), D-quantifier is associated with determiner-like elements where the
scope is restricted to NPs in specific positions.
 140

4.1.1.3 Noun phrases containing numerals

The native lexicon of Waimiri Atroari has only three numeral words,18 whose

meaning is not generally restricted to mathematical quantities. The expression cmini ~

awinini -awinihe -awynihe' means 'alone' and also 'one;' the term typytyna means 'a

couple,' 'a pair,' or 'two;' the word for 'three' is takynynapa. Thus, traditionally the

kinja counted only up to three; amounts higher than three were referred to simply as

'several, many'. Today, with the modern necessity for handling money and the

introduction of western mathematical concepts through the village schools, the kinja

started using Portuguese loanwords to refer to numbers higher than three. These

borrowed numerals occur in the same position as the native words meaning 'one,' 'two,'

or 'three.' Less commonly, Portuguese numerals for 'one,' 'two,' or 'three' may also be

used instead of the native words, especially by the younger speakers (25). As shown in

the examples below, numeral words can occur before a noun (21-23, 26), after a noun

(24, 26), or by itself, after a verb (25).

(21) typytyna karyka

two chicken
'two chickens'

(22) awynihe petxi Kwawura i-aryka-pa ty-kyda tohnaka

one pig Kwawura REL-put-REM 3REFL-back over
'One wild pig put Kwawura on his own back.'

(23) takynyny pahky kaminja n-apynaka.

three only non-native 3-escape
'Only the three white men escaped.

'^.I am using the term 'niuneral words' instead of 'numerals' because I have no evidence for the existence
of numerals as an independent part-of-speech in this language.
This sentence was t^en from a text narrating a fight between the kinja and the non-natives. The three
non-Indian characters mentioned in tliis sentence have already been introduced in an earlier passage of the
 141

(24) weri samka ka-pia takynynapa

woman hammock make-IM three
'The woman made three hammocks.'

(25) amy kinja dezessete apytyhy

other people seventeen behind

amy kinja dezessete nate'me

other people seventeen behind
'Seventeen people were in front, and seventeen were behind.'

(26) dois kinja xiba myry-myry-pia quatropahky.

two people fish redup^ -fish-lM four only
'Two people caught only four fish.'

It is not totally clear what motivates this variation in the position of the numeral

words. However, from the examples shown above, it seems that the variation may be

related to issues of specificity and definiteness. That is, if the speaker thinks the listener

already knows and can identify the particular referent which will be talked about, the

speaker will codify such referents as definite and specific. According to Diesing (1992)

and Diesing & Jelinek (1995), there is a mapping between argument structure and

information structure. By information structure, Diesing (1992:58) refers to "the

organization of the clause with respect to presuppositional (familiar) vs. information new

to the discourse." The Waimiri Atroari language clearly follows this mapping when

topicalizing some arguments of the clause and also seems to make an association between

definiteness and old information versus indefiniteness and new information through the

relative position of the numeral word in a noun phrase. When in specific and definite

text. Therefore, the noun phrase taMnini pahki kamijia 'only [the] three White men' is clearly definite in
this context, as shown by the English translation provided above.
This is an example of reduplication in Waimiri Atroari. Reduplication in tliis language is bimoraic,
occurring with verb stems to indicate repetition or continuation; See section 2.5.4 and 2.6.2.
 142

contexts, the numeral word seems to prefer the left side of the noun (21-23, 26).

However, in unspecific contexts, the numeral word is positioned either post-verbally (25)

or at the right side of the noun (24, 26). In this sense, the numeral words in (24) and (26)

behave like the adverbial quantifier in example (15), inasmuch as the speaker is not

talking about a specific group of people or hammocks. It is true that examples (21) and

(25) are potentially problematic for my assumptions, since it is not clear if the consultant

is talking about two specific chickens or a specific group of seventeen kinja. Therefore,

this is a hypothesis to be fiirther investigated.

4.1.2 Verb phrases

In Waimiri Atroari, a VP can be constituted by a verb alone (28) or a verb

preceded by a NP (27). The VP can move to before the subject NP through topicalization

(27) and cannot have its components separated, except in the OSV context when the

object moves alone to a topic position (as will be discussed below in example 30). In

Waimiri Atroari there is a second-position particle,^^ ram, which can be used as a

criterion to test the constituency of a given phrase (See section 3.1.6.2). The particle ram

can never intervene between two elements of the same phrase (27c). Furthermore, since

ram is a second-position particle, it can be usefiil in determining which elements in a

given sentence were moved, as in example (27a) below.

(27) a, tahkome i-iny-pia ram Irikwa

elders REL-eat-lM 2PART Irikwa
'Irikwa (a mythological entity) ate the elders.'

According to the syntactic framework I am adopting here, based on HaJpem & Zwicky (1996), the first
element is the first immediate constituent of the clause, such as a complement or argument of the verb, an
adverbial modifier, or other clausal constituent.
 143

.ikome i-iny-pia

c, *[tahkome ram i-iny-pyia] Irikwa.

elders 2PART REL-eat-IM Irikwa
'Irikwa (a mythological entity) ate the elders.'

(28) ka-ky/
speak-IMP
'Speak!'

(29) bahinja maia kynke

children knife break-T/A
'The children are breaking the knife.'

(30) woky i-eky kra h-ee-ia

banana REL-juice IPRO 1-drink-T/A
'I drink the banana juice.'

Example (30) above illustrates the only context where the VP is separated by the

subject, when the object undergoes topicalization. (For more details, see section 2.1 on

topicalization.) Generally, in more 'unmarked' situations, nothing can intervene between

the object and the verb.

4.1.3 Postpositional phrases

In Waimiri Atroari, some postpositions can inflect for person, taking the same

series of markers used to indicate the possessor on nouns and the object on transitive

verbs (See table 3.10). The syntactic link between a postposition and its noun phrase
 144

object is as strong as that between the elements of the noun and verb phrases: nothing can

intervene between them.

(31) PostP

Post'

NP Post
A
syna ke
'with water'

The tree above demonstrates that the head is always to the right in postpositional

phrases, just as with noun and verb phrases. The examples below reinforce my claim that

nothing can separate the postpositions from their complements.

(32) iakypa a'a ny-djkie-pa tapinja ta

then 1+3PRO l+3S-squeeze-REM.P sieve LOC
'Then, we squeezed (the manioc) in the sieve.'

(33) samka tyhnaka

hammock over
'over the hammock'

(34) impa a'a n-ike'ia-pa meie impary axinjaty tyhnaka

then 1+3PR0 1+3-bake-REM beiju then oven on
'Then we baked the beiju (kind of manioc tortilla) on the oven.'

(35) iakypa a'a minja pitxi-pia maia ke

After 1+3PRO manioc peel-lM.P knife nsfSTR
'Then we peeled the manioc with the knife.'

(36) aa ram xiba h-iri-pia ka inaka

IPRO 2PART fish 1-give-IM.P 3PRO DAT
'I gave fish to him.'

(37) amyra ram aa=inaka xiba m-yry-pia

2PRO 2PART 1-DAT fish 2A-give-lM.P
'You gave fish to me.'
 145

(38) pamwe aa=i-iry-py-pia, woky yry-ky mahta inaka

Pamwe l=REL-tell-CAUS-lM,P banana give-MP Marta DAT
'Paruwe told me; "Give the banana to Marta.'"

Based on the different types of phrases shown above, I argue that Wairoiri Atroari

is a head-right language. In verb phrases, noun phrases with possessives, and

postposition phrases, Waimiri Atroari presents a typical case of head right. However, in

phrases with adjuncts, such as noun phrases containing adjectives, adverbial quantifiers,

and numeral words, the relative position of the head seems to vary; a) depending on the

kind of information that the speaker intends to convey, such as specificity and

definiteness (adverbial quantifiers and numeral words);^^ or b) whether it is modifying a

noun or a verb (adverbial quantification phrases).

4.2 Waimiri Atroari Clausal Order

I have shown in the first section of this chapter that Waimiri Atroari is a head-

right language. In the next section, I intend to extend the analysis to show that this

language is also 'specifier-first.' According to Radford (1997;90), "Since although the

notion specifier is central to much contemporary work in syntax, it is hard to identify any

common set of properties which all specifiers share because of the disagreement among

linguists of what is its fLinction." Here, I assume the view that "specifier position is used

to mark a particular grammatical function, the subject fijnction" (Chomsky 1991). As

illustrated in the trees below, Waimiri Atroari allows different kinds of clausal order,

such as SOV, SVO, SV, VS, OSV, and OVS. However, as demonstrated in the tree in

In relation to this issue, I have to do more tests and collect more data to check this hypothesis.
 146

(39), in the majority of cases the head of the clause is on the right and the specifier is in

the first position.

(39) a. VP b. VP c. VP

Spec V Spec V

V Comp Comp V V

d.VP e. VP

V Spec V Spec

V Comp V

Like English, Waimiri Atroari is a 'specifier-first' language, since the SPEC is

mostly positioned before the head. The only exceptions are in the OSV and VS orders.

In the OSV context, topicalization destroys the order of the SPEC in relation to the head,

leaving behind a trace. However, at this point in the analysis, it is not clear to me what

motivates the VS order.^^ It is possible that verbal topics offer an area of exploration for

this word order.

In this grammatical sketch, I claim that SOV is the basic order based on three

factors: statistical frequency, descriptive simplicity;^^ and pragmatic aspects (such as

distinctions between old and new information, etc.). Waimiri Atroari tends to put old

information on the left side of the sentence. Waimiri Atroari shows what Mithun

In transitive sentences, Waimiri Atroari does not allow verb-initial orders.

Chomsky (1965:127) shows that the preferred order for potentially ambiguous clauses can generally be
considered the most neutral one. As he states it, "Simple, declarative, active clauses with no complex verbs
or noun phrases" seem to exhibit a neutral order.
 147

(1992:31) call 'newsworthiness:' even when both arguments (subjects and objects) are

equally provided at the same time or both are new (none of them was presented in

previous discourse or context), the speakers of Waimiri Atroari tend to prefer the subject

first (see texts in Appendix).

Looking at a corpus of eleven texts of different kinds (mythological, technical,

reported stories, and others), I analyzed all sentences that presented both one- and two-

place predicates with overt non-pronominal arguments and pronominal arguments. From

a limited sample of ninety-five sentences, the proportion of word order variation found

was the one presented in Table 4.1 below:

Table 4.1Frequency of occurrence of each clausal order

SOV 42
sv 25
OVS 10
svo 8
osv 5
VS 5

Taking the position of the specifier in relation to the head, even with the

possibilities of OVS and VS (fifteen occurrences), in the majority of cases (eighty

occurrences) reinforce the 'specifier-first' analysis. Although statistical predominance

may not be sufficient to establish the basic word order of a language, as pointed out by

Derbyshire (1977), this criterion is also corroborated by other facts of the Waimiri

Atroari language. Here, I demonstrate that at surface structure the basic word order

appears in different types of constructions, reinforcing the assumption that SOV shows

the 'descriptive simplicity' pointed out by Chomsky (1965). The examples below also

show the preference for 'SPEC-first' and head-right constructions.

 148

Transitive context
(40) aa ram ka h-ini-pia
1 2PART 3 1-see-lM
'I saw him.'

(41) ka ram aa=i-ini-pia

3 2PART l=REL-see-IM
'He saw me.'

(42) kipeti wiwe pyrykia-pa waha

wind tree bring. down-REM.P many
'The wind brought down many trees.'

(43) warakaxi myryky ram kixinja i-mah-pa

Warakaxi son 2PAJR.T sand REL-throw-REM.P

Kamiahara myryk-eme i-eba taka

Kamiahara son-DEV REL-eyes AL
'Warakaxi's son threw sand on the eyes of Kamiahara's son.'

As you can observe, examples (40) to (43) were taken from elicitation data and

from stories; they show the preference for SOV,

Stative predicates (with copula)

(44) Ananiwa-beme ram ietypy-pa na
bird (sp.)-DEV 2PART sick-emph cop
'Anaruwa is sick.'

Quotative context
(45) mawa njy i-tee-pa, n-ootxi-pa ke-pa, apia myre?
Mawa noise REL-hear-REM.P 3-go.down-REM.P say-REM.P what that
'Mawa heard a noise, waited, and said: "What's that?"'

Since X-Bar Theory does not allow or cannot generate sentences of the type

'OSV inasmuch as the SPEC of IP cannot intervene between the verb (head) and the

complement object, in this grammatical sketch I show that the OSV and OVS orders

result respectively from the object and the VP movement to a topic position.
 149

4.3 Topicalization

According to Payne (1997:270), "Topic is what the sentence is about, it is the old,

given, or known information." In Waimiri Atroari, topical elements tend to appear in

initial position, a fact that is very common cross-linguistically. The topic construction in

Waimiri Atroari is used to turn the attention to a definite object in order to avoid being

focused and interpreted as new information, as predicted by Diesing (1992).25

Aissen (1992:43), analyzing the position of topic and focus in Mayan, claims that

sentences with two NPs before the verb, (SOV) and (OSV), involve the focus of one NP

and the topicalization of the other. She shows that while SOV represents subject

topicalization and object focus (topic is S-initial), OSV order must represent object

topicalization and subject focus (focus position is preverbal). Taking part of this

approach into consideration, I claim that the OVS order in Waimiri Atroari results from a

left movement of the VP in order to reach topic position, and unlike Mayan, the OSV

order results from the leftward movement of the object alone.

4.3.1 OVS order (movement of the whole VP)

Unlike Hixkaryana (Derbyshire 1977:595), an OVS language of the same family

which moves the subject to the initial position in order to express topic, in Waimiri

Atroari OVS is not neutral. Instead, it is a marked order where what is moved to the

topic position is the whole VP. In this paper, I argue that the element to be topicalized

moves into the SPEC position within a TopP (Topic Phrase) constituent headed by a topic

head. The text fragments below illustrate cases of topicalization:

See texts in tlie Appendix.

 150

(46) a. tapesa kixinja weiaky tyiyry n-o'm-pa

shallow beach when? Tyiyry 3S-dive-REM.P

tapiwutape ta n-o'm-pa
Tapiwutape LOC 3S-dive-REM.P
'In the shallow beach of the Tapiwutape lake Tyiyry dived.'

b. tyiyiy i-yhia i-erekyty-pa xiriminja

Tyiyry REL-hair REL-cut-REM.P Xiriminja
'Xiriminja cut Tyiyry's hair.'

In example (46b), the VP tyiyry i-yhia i-erekyty-pa 'tyiyry's hair cut' is

topicalized probably because it contains old informationthat is, the noun tyiyry

'Tyiyry' is introduced earlier in the text (46a). As a result of the topicalization of the VP,

the subject NP xiriminja 'm3 'thological entity,' which is new information, is introduced at

the end of the sentence.

A similar case occurs in the examples in (47) below. In (47a), the subject NP

ianana 'lanana' and the object NP tahkome 'elder(s)'are both introduced for the first time

in the unmarked, SOV order. In (47b), the NP tahkome occurs again as the subject of an

intransitive verb. In the following two sentences, (47c) and (47d), the NP tahkome is

already considered old information; therefore, the VP in which it occurs is topicalized.

This situation is represented in (48) below.

(47) a. ianana tahkome bake-paiskixki xirikiki pyky

lanana elders shoot/kill-REM parakeet parakeet because
'Ianana killed the elders because of the parakeets.'

b. tahkome wapy n-oo-sa-pa kamakaxi taka

elders many 3-climb-rem kamakaxi.tree AL
 151

xirikiki baka-paiky
parakeet shoot/kill-T/A
'Many elders climbed the kamakaxi tree in order to kill the parakeet.'

impa tahkome ipia-pa ianana xirikiki baka taka

then elders find-REM lanana parakeet shoot/kill AL
'Then lanana found the elders killing parakeets.'

d. impa tahkome bake-pa ianana ebapy tapary

then elders shoot/kill-REM lanana eye LOG
'Then lanana shot the elders in their eyes.'

(48) OVS (through VP topicalization)

TopP (46.b) opP (47.d)

[ V P ] Top' [VP]

p IP Tbp JP

0 xiriminja 0 lanana

0 V
DP V
[tyiyry i-yhia i-erekytypa] [tahkome bakepa]

4.3.2 OSV order (movement of the object)

As X-bar theory cannot deal with OSV order, I claim that this order results from

the dislocation of the object to topic position. Diesing (1995:126) claims that in many

languages, object shift depends on information structure, in particular something like the

contrast between specific (definite) and non-specific (indefinite) information. As a result,

objects move for configurational or interpretational reasons rather than for case-checking.

According to my Waimiri Atroari consultants, this order happens when they want to
 152

emphasize who did the action. Therefore, as proposed by Diesing (1995), I assume that

in the OSV order, the object undergoes topicalization. The subject position in this order

is restricted to (taking only the kra ~ kara form) and 2"'^ person pronouns. The only

exception for this restriction happens with quotative sentences, where the subject can be a

proper noun.

(49) [ampa kinja teneriki na-ky], kybina ke-pa

other people afraid COP-PAST Kybina say-REM
'"The other persons were afraid," said Kybina.'

I leave aside these quotative sentences since they are well known for exhibiting a

marked order (Branigan and Collins 1993). I assume that OSV order in the standard case

can best be explained by Wackernagel's Law, which claims that clitics, particles, and

pronouns must appear in second position in a clause (Halpem & Zwicky 1996).

According to Kaisse (1981), recent research on clitics shows that languages may vary

with respect to their definition of second-position particle. Waimiri Atroari is one of

those languages, such as Bulgarian, Luiseno, and Serbo-Croatian, that allow either the

first word or the first constituent of the clause to define 'second position' (see footnote

9)26

(50) [merepy i-eky] kra h-ee-ia

merepy fruit REL-porridge iPRO lA-drink-T/A
'I drink merepy porridge.'

(51) maryma kra h-yn-iany

piranha IPRO lA-eat-T/A

As illustrated above in examples (50) and (51), the pronoun indicating first person singular kra can occur
after a word or a constituent. On the other hand, second position particle ram can appear not only after a
word and a constituent, but also after a whole clause.
imka typotxe a=wutoty m-itxirikw-epa ipaikyparam tymeri a=w-iapa
if fast 2-run 2-stop-T/A after 2PART jaguar 2=kill-T/A
'If you stop running fast, the jaguar will catch you.'
 153

'I am eating piranha.'

(52) *Maryma ram kxa h-yn-iany

piranha 2PART iPRO lA-eat-T/A
'I am eating piranha.'

In the tree below, I demonstrate that the heavy element moves to the left and the

light element stays in the right, in the second position. The pronoun kra is in fact in the

second position, inasmuch as the second-position particle ram cannot co-occur in this

example (see example 52). They compete for the same slot in the tree, the Top slot.

They are in complementary distribution.

(53) OSV (through topicalization of the O)

TopP

[ O ] Top'

Top

kra

aryma h-yni-any

4.4 Adverbial subordinated clauses

According to Payne (1997:316) and Thompson & Longacre (1985:171), adverbial

clauses are those which serve an 'adverbial function:' they modify a verb phrase or a

whole clause. Adverbial subordinated clauses behave as adjuncts inasmuch as they

simply add some extra information to the clauses. Thompson & Longacre (1985:172)
 154

explain that there are three types of subordinated clauses: those which function as noun

phrases (complements); those which function as modifiers of nouns (relative clauses);

and those which function as modifiers of verb phrases or entire propositions (adverbial

clauses).

In this section, I briefly direct the analysis to two types of subordinated clauses:

clauses that indicate time and location, and what Meira (1999) and Gildea (1998) call

'nominalized clauses.' Here, I want to demonstrate that these clauses work exactly like

the simple phrases illustrated above.

4.4.1 Temporal subordinated clauses^^

This type of clause can occur either at the beginning or at the end of the sentence.

Thompson & Longacre {op. cit.) claim that in the world's languages, there are typically

three devices for marking subordinated clauses: subordinating morphemes, special verb

forms, and word order. Waimiri Atroari uses only the first two devices. As shown in the

example below, temporal adverbial clauses in Waimiri Atroari are introduced by the

word imka 'when, if As Thompson & Longacre {op. cit, 193) mention, there are

several languages, such as Indonesian and certain languages of Papua New Guinea,

which do not make a distinction between if clauses and when clauses. Waimiri Atroari

seems to be one of these languages, since, as we have mentioned, imka can be either

translated as 'when' or 'if (see another example provided in footnote 16.)

(54) a. [keme iki-pia namijia] impa ni-tam-pia

3 bite-lM.P dog then 3s-cry-lM.P
'When the dog bit him, he cried.'

In this paper, as you note in the trees below, I am not dealing with INFL (I am not representing it in the
trees).
 155

As you can observe in the tree below, in the temporal adverbial clause, the VP

undergoes topicalization; it does not show the basic order SOV, but the marked OVS.

This example was given after I had asked about other examples: kerne 'he' is old

information. Moreover, you can note that the C slot can be empty or filled by the word

imka 'if, when.'

b CP
CP(AdvP; CP

C (Adv) impa nytampia

imka opP

[VP] Top'
Top

ndminja
[kerne ikipia]

4.4.2 Locative subordinated clauses

Unlike temporal clauses, which can occur either at the beginning or at the end of a

sentence, locative subordinate clauses, introduced by the construction apiyapi yaki ~

epiapi yaki, can only occur at the end of the sentence, as shown by examples (55) and

(56) below:
 156

(55) ka mepr-eme wu-pa epiapy yaky mydy pahsapy na-ky

3 tapir-DEV kill-REM where house old COP-PAST
'He killed the tapir where the old village was.'

(56)
a. ka ram ase mydy iam-ia apiapy iaky temer-eme wu-pa
3 2PART new house build-T/A where jaguar-DEV kill-REM
'He builds the new village where he killed the jaguar.'
b. IP

ase mydy ia

apiapyyaky

temereme

wupa
 157

4.5 Nominalized clauses

Nominalized clauses present the same parameter position of the head shown by

the more typical NPs described above. As shown by examples (57) and (58) below, the

head in these nominalized clauses is marked by the nominalizer sufiBx -o, while the NP

corresponding to the subject of a transitive clause is marked by the particle ia. Unlike

independent clauses, which may present word-order variations due to phenomena such as

topicalization, nominalized clauses present a strict SOV order, a fact that gives flirther

support to the postulation of SOV as the basic word order in Waimiri Atroari.

(57) ka ram aa=y-akyny ia woky i-akym-o i-ini-piany

3 2PART l=REL-sister AG banana REL-make-NOM REL-see-REC
'He saw that my sister made banana porridge' or 'He saw the making of banana
porridge by my sister.'
(58)
a. amyra ram aa=papa ia tymeri w-o28 m-ini-pa
2 2PART 1-father AG jaguar kill-NOM 2-see-REM
'You saw that my father killed the jaguar' or 'You saw the killing of the jaguar by
my father.'
b.

amyra

C minipa

aapapa w-o

The root of the verb 'to kill' is -wm and the root of the verb 'to make' is -aJdmi'. The last vowel of the
root is dropped with the suffixation of the nominalizer -o. The nominalizer -o is also observed in other
Carib languages (Gildea 1998).
 I have provided an analysis of the Waimiri Atroari phrase structure and word

order, as well as a brief overview of the distribution of the heads in some subordinated

and nominalized clauses, accounting for the head and specifier parameter. I have argued

that Waimiri Atroari belongs to a parametric class of languages that exhibits the

following characteristics at spell-out: (a) head-right when not involving adjuncts, (b)

specifier-first, and (c) topicalization-left based on pragmatic factors (distinction between

old versus new information) that have grammatical consequences.

Less usual word orders, such as OVS and SOV, are accounted for by postulating a

movement resulting in the topicalization of the whole VP or the object alone,

respectively. Although I do not analyze cases of SV and SVO orders in this paper, I

speculate that SVO order could be a result of Portuguese influence, inasmuch as its

occurrence is very reduced, being more common among the younger male speakers, who

present a higher degree of bilingualism than elders and women. Therefore, this, as well

as the factors underlying the occurrence of SV word order, is one of the hypotheses to be

fiarther investigated.
 ABBREVIATIONS

A 'subject of transitive verb'

ABL 'ablative'
AGT.PART 'agentivity particle'
ADJ 'adjective'
ADV 'adverb
AL 'allative'
CAUS 'causative'
COP 'copula'
DAT 'dative'
DES 'desiderative
DEV 'devaluative'
DISJ 'disjunctive particle'
EMPH 'emphatic'
IMM.F 'imminent future'
IM.P 'immediate past'
IMP 'imperative'
INT 'interrogative'
INSTR 'instrumental'
INSTR.N 'instrumental nominalizer'
LOG 'locative'
NEG 'negation'
NOMLZ 'nominalizer'
2PART 'second-position particle'
O 'object'
POS 'possession'
PRO 'pronoun'
REC.P 'recent past'
REM.P 'remote past'
REDUP 'reduplication'
REFLX 'reflexive'
REL 'relational prefix'
REM.F 'future remote'
REM.P 'remote past'
S 'subject of intransitive verb'
T/A 'tense/aspect
VAL 'valuative'
VERBL 'verbalizer'
1 'first person singular
2 'second person sing, or pi'
3 'third person sing.or pi.'
1+2 'first person plural inclusive'
1+3 'first person plural exclusive'
 160

APPENDICES

Al. Tyiyry ikaa^^

'Tyiyry's story'

tapesa kixinja weiaky tyiry n-o'm-pa

shallow beach when? Tyiyry 3s-dive-REM.P

tapiwutape ta n-o'm-pa.
Tapiwutape LOG 3s-dive-REM.P
'In the shallow beach of the Tapiwutape lake Tyiyry dived.'

syna i-apo'o tyiyry i-yhia i-erekyty-pa xiriminja

water REL-inside Tyiyry REL-hair REL-cut-REM.P Xiriminja
'Inside the river, Xiriminja cut Tyiyry's hair.'

3. aminjaky iakypa xiriminja ny-nypykwa-pa amehepie' taka

later after Xiriminja 3s-appear-REM.P Amehepie' AL
'Later on, Xiriminja appeared at the Amehepie' village.'

njawa n-ym-pa waha kipety tarara many

rain 3S-come-REM many/much wind thunder/lightning also
'It rained a lot with thunder and lightning also.'

kipety wiwe pyrykia-pa waha

wind tree bring. down-REM.P many
'The wind brought down many trees.'

kinja pyruwa ke n-itxikwa-pa

people arrow INSTR 3A-shoot-REM.P
'The persons shot with arrows.'

7. paryna xiriminja pykwa-pa iakypa maiahka n-itxikwa-pa,

Paiyna Xiriminja shoot.arrows-REM.P after Maiahka 3A-shoot-REM.

am pa kinja many
other people also
'After Paryna shot Xiriminja, Maiahka also shot him, and other people did too.'

These stories, traditional tales of the Kinja people, were narrated by Dauna, a male shaman and
storyteller today in his late fifties, in the village of Kaminjanyty in 1990/1993. I collected lanana story,
kaapy tahkome karykapa, wyty ikaa (written by Waraie), and meie ikaa (written by the women of Alalaii's
village). The other stories were collected by Carlos Augusto Queiroz (ex-professor of the Alalaii's village)
 161

8. maiahka xiriminja i-yhy i-aa-pa itxi taka

Maiahka Xiriminja REL-head REL-take-REM.P jungle AL
'Maiahka took Xiriminja's head to the jungle.'

9. kyta-hkypa xiriminja iee i-aa-pa iakypaiee wepy

putrefy-after xiriminja teeth REL-take-REM.P after teeth ?

iry-pa wymy ida-typyme.

make-rem.p lice comb-used to
'After that Xiriminja's tooth putrefied, he took his tooth and made a comb to take
lice.'

10. tyiyry araky nyryna kyky ta na

Tyiyry today alive under LOG COP
'Today, Tyiyry is alive under (the river).'

A2. Kyrywaky Weriri Ikaa

Parrot woman's story
'The Parrot woman's story'

1. pana kinja wapy na-ky maryba ka-huwa

formely people many COP-T/A party make-NEG
'Formely, the people did not make many parties.'

2. iakypaka kyrywaky i-meky ny-pytxima-pa

then ? parrot REL-daughter 3 S-transform-REM.P

ny-synypykwa-pa kinj-e'me pa piyhy mydy taka

3S-appear-REM.P people-VAL as Piyhy village AL
'Then, the parrot's daughter became a person and appeared at Piyhy's village'

3. paxe nu-waty-pa iry mana ka

Paxe 3S-marry-REM.P 3PRO with/and 3PRO
'Paxe married her.'

4. kyrywaky pyty paxe i-nepa-pa maryba kapry pyky

parrot wife Paxe REL-teach-REM.P party/song make how
'The parrot's woman (Paxe's wife) taught Paxe how to sing and make parties.'

5. iakypa ny-tyn-pa kinja many kapy te'xy n-eeni-pa

then 3S-go-REM.P people with NEG DESID 3S-stay-REM.P
'Then she went off, she did not want to stay with the people.'
 162

6. maryba kapy-se n-y-sapa kaapa taka iry

song/party make-in order to 3S-go-T/A field al 3PRO

ty-iymy ipo-se na.

3REFLX-father find-in order to COP
'She went to sing in the garden in order to meet her father.'
7. paxe sakyna wapy i-amapy-se n-eepy-sapa mydy taka
Paxe angry many REL-call-in order to 3S-retum-T/A house AL
'Paxe very angry called her to return to the house.'

8. iry n-aryma-pa te'xy n-eeni-pa iakypapaxe

3PRO 3 S-come back-REM.P DES 3S-stay-REM.P then Paxe

n-ipykwapy-sapa pyruwa ke
3S-kill-T/A arrow INSTR
' She did not want to come back then Paxe killed her with arrows.'

9. paxe ka-pa t-aska ia piyhy mydy tany

Paxe say-REM.P 3REFLX-relatives to Piyhy village from

ke-pa: kaapy taka kamapal

say-REM.P heaven AL let's go
'Paxe said to his relatives fi'om Piyhy's village: "Let's go to the sky!"'

10. waiamy kinja-py kaapy pykwa-pa

tortoise people-as sky shoot arrow-REM.P
'The tortoise person shot an arrow to the sky.'

11. iakypa pyruwa i-asyty pykwe-pa kixi tyhnaka n-otxia-pa.

Then arrow REL-butt shoot-REM.P floor on 3S-climb down-REM.P
'Then he was shooting an arrow into the other arrow's butt until they touched the
floor.'

12. iare kinja-py kyrywy

Tare people-as snake
'Tare, the snake that was a person.'

13. iry n-oony-pa kykynora i-aa-pa kaapy

3PRO 3 S-climb-REM.P Sp. of leave Rel-take-REM.P sky
'He climbed to the sky using the leaves.'

14. waiamy tyty-sy n-eepy-sapa

tortoise ?-DES 3S-retum-T/A
'The tortoise wanted to return (climbed to the heaven).'
 163

15. wook-uwa-pa n-eeni-pa

climb-NEG-REM.P 3S-stay-REM.P
'She could not climb, then she stayed.'

16. mawa kinja-py iry ia ke-pa: yrypy we'ky ka-pa

Mawa people-as 3PRO to say-REM.P here IMP say-REM.P
iakypa n-oo-sapa.
then 3S-climb-t/a
'Mawa who was a person said to him to be here, then he climbed.'

17. kwata ny-kypia-pa ny-kyhy-sapa tykyry pyky

spider monkey 3S-experiment-REM.P 3S-try-T/A ? how
'The spider monkey tried to climb.'

18. mawa kykynora kyty-pa iakypa iry n-yma-pa itxi iary

Mawa Sp.leave cut-REM.P then 3PRO 3S-fall-REM.P jungle tree

tyhnaka araky ipy na.

over today there COP
'Mawa cut the leave, then he (the monkey) fell over a tree in the jungle and stayed
there until now.'

A3. lanana ikaa30

'Xanana's story'

1. iananatahkome bake-pa iskixiki xirikiki pyky

lanana elders shoot/kill-REM.P parakeet (sp.) parakeet(sp.) because
'lanana killed the elders because of the parakeets.'

2. tahkome wapy n-oo-sa-pa kamakaxi taka

elders many 3-climb-REM.P kamakaxi.tree AL

xirikiki baka-paiky
parakeet shoot/kill-T/A
'Many elders climbed the kamakaxi tree in order to kill the parakeets.'

3. impa tahkome ipia-pa iananaxirikiki baka taka

then elders find-REM.P lanana parakeet shoot/kill AL
'Then lanana found the elders killing parakeets.'

lanana is a mythological entity, the owner of tlie forest. He has as a pet parakeet {firikiki in Waimiri
Atroari).
 164

impa tahkome bake-pa iananaebapy tapary

then elders shoot/kill-REM.P lanana eye LOG
'Then lanana shot the elders, at their eyes.'

iakypaiananawehe warykypa i-apremy i-amyrysa-pa

later lanana arrow luck REL-owner REL-miss-REM.P
'Later on, lanana's arrow missed the lucky person.'

iakypa ianana ke-pa: wa-pesa wura aa-wehe wy-se

after lanana say-REM.P NEG-in the time ? iPOS-arrow go-in order to
'After lanana said: "-once there aren't animals anymore my arrows will
disappear.'"

impa tahkome ikehepy i-emxa-pa iakypa

then elders dead REL-put.together-REM.P after

n-ymia-pa arawuta ike piempary

3A-lash-REM.P monkey ? like
'Then [lanana] put together the dead elders and later lashed them as one lashes
monkeys.'

8. impa tahkome ikehepy i-aa-iaa-pa ty-mydy taka

then elders dead REL-take-REDUP-REM.P SREFLX-house AL
'Then lanana took the dead elders to his house.'

impa warakypa i-apremy n-ytxia-pa n-ahtapa myry-kyhpa

then lucky RET .-owner 3S-go-REM.P 3S-at the back that-after

n-y-sapa takynyt-aska ikehe i-ary ini-apa.

3S-go-T/A ? 3REFLX-relatives dead 3REL-take see-t/a
'Then, that lucky person went at lanana's back observing to see where lanana
would take his dead relatives.'

10. impa iananany-bia-pa ty-mydy taka

then lanana 3S-come-rem.p 3REFLX-house AL
'Then lanana came to his own house.'

11. warakypa i-apremy tahkome ikehe iaba i-te-pa.'

lucky REL-owner elder dead ? REL-hear-REM.P

Hepe! Hepe! Hepe! Hebe! ianana ke-pa tahkome ikehe iabepa

Hepe! Hepe! Hepe! Hebe! lanana say-rem.p elder dead ?
 165

ianana
lanana
'Then the lucky person heard lanana happily saying to the dead elders:-"hepe!
hepe! hepe! hebe.'"

12. impa warakypa i-apremy peri i-kysa-pa wiwe ke

then lucky REL-owner door REL-measure-REM.P wood INSTR
'Then that lucky person measured the size of the door of lanana's house with a
piece of wood.'

13. impa peri i-kyhypy n-ame-pa mydy taka warakypa

then door REL-measurement 3S-take-REM.Phouse AL lucky

i-apremy
REL-owner
'Then the lucky person took the door's measurement from lanana's house to the
village.'

14. impa tximtxa byia-pa talikome

then wood plank make-REM.P elder
'Then the elder made a wood plank to close the entrance.'

15. impa tximtxa byhy i-aaia-pa impa tahkome ianana peri

then wood plank make REL-take-REM.P then elder ianana door

i-tapia-pa
REL-close-REM.P
'Then the elder took the wood plank and make a door to close ianana's entrance.'

16. impa ianana beme txitmxa pykwapy-pysa

then ianana poor wood plank shoot arrow-T/A
'Then the poor ianana was locked in and he was shot arrow.'

Note: lanana lived in a hollow tree.'

17. impa tahkome waty tamyke-pa

then elder fire put-REM.P
'Then the elders put fire in lanana's house.'

18. impa tahkome waty i-atxia-pa wahpary

then elder fire rel-make-rem.p many
'Then the elders make a big fire.'
 166

19. txamka-ky! txamka-ky! ke-pa tahkome waha i-aty-ky!

make fire-IMP make fire-lMP say-REM.P elder many REL-put-iMP

i-aty-ky! ke-pa tahkome

rel-put-imp say-REM.P elder
"Make the fire!" "make the fire!" said the elder "Put more fire wood!" "Put more
fire wood!" said the elders to the others.

20. impa iananany-kyrysa-pa tykatyka aa=pap-eme

then ianana 3 S-bum-REM.Psay iPOS-father-DEV
'Then they burned Ianana, said my father.'

21. impa tahkome kirikwa-hkypa n-aryme-pa ty-mydy taka

then elder bum-after that 3 S-come back-REM.P 3REFLX-house AL
'Then after they burned lanana's house, they returned to their own house.'

22. impa amiakypa tahkome n-ysa-pa ikry tyhy ini-se

Then later elder 3 S-come back-REM.P ashes LOC see-in order to
'Then they come back to see lanana's ashes.'

23 impa tahkome ianana myryky maryba kapry i-te-pa

then elder ianana son song sing REL-hear-REM.P
'Then the elders heard lanana's son sing.'

24. -hensin hensin mensi taba ieni piky many taba iene. ..
hensin hensin mensi taba ieni piky many taba iene
'Hensin, hensin mensi taba iene piky many taba iene said lanana's son.'

Note: I tried to get the translation fijr this song, but even my consultants could not
translate.

25. impa puwan-pa ianana myryky i-ape'ia-pa kiwyry

then rear-REM.P ianana son REL-take-REM.P tree hole

ny-tytapah-ky
3S-held-PAST
'Then the elders took lanana's son who was hold in the tree hole to rear.'

26. impa tahkome n-aaia-pa many

then elders 3S-take-REM.Pwith
'Then they took him with them.'
 167

27. impa tahkome ny-puwasa-pa

then elders 3S-rear-REM.P
'Then the elders took care of lanana's son until he grow up.'

28. ipaikypa ty-puwa-hkypa ianana myryky mepry syh-kwa-pa

then 3REFLX-grew-after that ianana son tapir hair-take-REM.

impa timatrepitxi mepri syhy i-aaia

then adopted father tapir hair REL-take
'Then after he grow up, he plocked tapir's hair to show his adopted father (to
show that closed to there they could find tapirs).'

29. impa typuwa-hkypa ianana myryky mepry pykwe-pa axiwi

then grew after that ianana son tapir shoot-REM.P agouti

ikehe warypahky tete ianana myryky mytxia-pa patuwa ia

dead look like ? inana son wrap-REM.P pataua leaf

taka
AL
'Then after he was grown up, he hunted a tapir and wrapped it with pataua leaves
to look like agouti.'

impa n-insa-pa typuwanypytxi mepri ikehe ini-sapa ianana myryliy

then 3S-bring-REM,P adopted mother tapir dead see-T/A ianana son
'Then he brought the tapir to show his adopted mother.'

impa typuwanpa i-kyh-sapa heme! heme! ka-kwapy ma

then adopted mother REL-order-t/a heme! heme! say-MP mom
'Then (before she could open the wrapped thing) he told his adopted mother to
say "heme!" "heme!" say mom.'

32. Impa xana-terepy ke-pa ianana myryky ia bahinja syba

then mother-? Say-rem.p ianana son to small bad

amyry m-inehetxa axiwi sybary myra kapy inime mepri inehty taha
2PRO 2A-bring agouti bad ? NEG husband tapir bring big

myry ieny inehty

? basket bring
'Then her mother said: This is not tapir, this looks more like agouti.' "When
my husband would bring tapir, it was in a very big basket.'
 168

33. impa xanypy tamtypy sykwe-pa impa mepri ikehe n-adaryke-pa

then mother string cut-REM.P then tapir dead 3S-spreadREM

taham-pa-ry
big-EMPH-?
'Then when she cut the string that tied the bundle, the bundle spread open
showing the real size of the tapir.'

34. impa nysakome iaxinja-tyrypy i-kynke-pa mepry se-py tyruwa many

then elder woman griddle-/ REL-break-REM.P tapir foot-? pan and
'Then when the eldery woman opened the bundle, the tapir's foot kicked the
griddle and the ceramic pan too.'

35. impa iananamyryky texiba-pa na-pa

then ianana son sad-emph cop-rem.p
'Then Ianana myryky became very sad.'

36. impa iananamyryky ke-pa: -hepe! hepe! hebe! tykatry-ky

then ianana son say-REM.P -hepe! hepe! hebe! say-PAST

aa=mam-eme aa=mama kapy sybary amyry ke-pa

IPOS-mother-DEV IPOS-mother NEG bad 2PRO say-REM.P
'Then lanana's son said: You are not my mother!" "My true mother said:-
hepe! hepe! hebe!'"

37. impa tykry tykry tykry ka i-te-pa ianana myryky

then tykry tykry tykry EVID REL-hear-REM.P ianana son
'Then lanana's son heard: -tykry, tykry,tykry.'

38. impa ianana myryky ke-pa: "-aa=iaka ky n-yt-e"

then ianana son say-REM.P IPOS-uncle this 3S-sing-T/A
'Then lanana's son said: "-This is my uncle singing.'

39. ipaikypa kate' iananamyryky n-y-sapa

after ? ianana son 3S-go-T/A
'Then he left and never come back.'

A4. Kaapy tahkome karykapa

'The sky fell over the elders'

1. Pinapa tete tahkome pyna mepri n-ytyta-pa pakia many

close ? elders place tapir 3S-come-REM.P pig and/with
'The tapirs and wild pig came close to the place (village) of the elders.'
 169

2. wahpaky tete tahkome pyna nykwa n-yhtxa-pa

many ? elders place animals 3S-come-REM.P

kaapy ia tybekia-paiky.
heaven to tell-after
'Many animals came close to the elder's place,(somebody telling to other that this
happened before.'

3. Impa tahkome kabaha iee ini-apa pakia iee waryparyky

Then elders armadillo tooth see-T/Apig tooth as
'Then the elders saw the armadillo's teeth as the wild pig's teeth,'

Note; The elders perceived it because the armadillo's teeth are small, and they
were big as the wild pig's teeth.

4. impa ty-periraipy mydy ini-pysapa

Then 3REFLX-brother-in-law house see-T/A
'Then He saw the house of his own brother-in-law.'

5. impa tahkome ke-pa

then elder say-REM.P
'Then the elders say:'

6. - moo kytapymykyme mydypy ka-pysapa tahkome

-there ? that house say-T/A elder

ty-perimipy mydy pia.

REFLX-brother-in-law house ?
"-There, is the house of that people" said the elders to his brother-in-law.'

Note: The villages were very distant from each other, but a big fire had made a
big clearing in the jungle that made it possible to see from a village to another.'

7. ipaikypa kate' tahkome bekie-pa kaapy

after that ? elders fall-REM.P heaven
'After that, the sky fell on the elders.'

8. he'pia! he'pia! ka-pysapa tahkome irany xia

poor us poor us say-T/A elder what ?

ka-pysapa tahkome kaapy ia tywu-paiky.

say-T/A elder heaven to kill-after that
"Poor us, poor us" said the elders. "What is happening!" said the elders. After that
the sky fell on the village and killed them.'
 170

9. piria wukapry tete n-opyna-pa.

type of wood support ? 3S-save-REM.P
'Only a man and a woman (siblings) that protected themselves in the central pool
of the house were saved.'

10. ie'ximpa ampa mydy tanypywu-iapa

all other house ABL die-T/A
'All the other people from the other houses died.'

11. impa yry itxiri tyhnaka n-apyke-pa tykaty aa-pap-eine

then this land above 3S-pass-REM.Psay iPOS-fkther-DEV
'Then (those people that were saved) passed by the house's support to the new
land said my father.'

Note: The house's central pole made a hole in the sky. Through this hole the
people who were saved exited, went above the old sky and built a new land.
Today the Waimiri Atraori people live on the top of sky.

12. takrehen myky kaapy n-ybixkwa-pa n-esa-pa tykaty tete

take time 3PRO heaven 3S-hurt-REM.P 3S-recovery-REM.P say ?

tahkome etypa myry kaapy tykaty tahkome

elder hot this heaven say elder
'They took time to recover from all the injuries caused by the sky's falling said
the elders.. .the elders said that the sky was very hot.'

13. piria wukrapy tete n-iemkwa-pa araky kwetypyme

type of wood support ? 3S-reproduce-REM.P today ?
'That brother and sister who survived, they reproduced. It is because of their
children that we are here today.'

14. impa takrehen myry ty-samka i-pyia-pa.

then take time this 3REFLX-hammock REL-find-REM.P
'Then they took time to find their own hammock.'

15. impa takrehen ny-patyme-pa ty-kaapa i-akytxia-pa

then take time l+3S-find-REM.P SREFLX-garden REL-make/cut-REM.P
'Then they took time to find a place to make their own garden plantation.'

16. impa ty-mydy i-amyia-pa tahkome takrehen-pary ty-pyty-pe'me

then 3REFLX-house REL-build-REM.P elder take time-? 3REFLX-wife-DEV

many pahky ty-kaapa i-akytyia-pa

and/with ? 3REFLX-garden REL-make/cut-REM.P
 171

'Then the elders (the brother and sister that survived) took a long time to build
their own house and make their garden plantation too, his wife also helped him to
make the garden plantation.'

17. wury kapy kypahky ty-kaapa i-akytyia-pa topy kypahkyn

axe NEG made of 3REFLX-garden REL-make-REM.P stone made of

pakia iee kypahkyn waryna iee kypahky ty-kaapa

wild pig tooth made of Sp. of rodent tooth made of 3REFLX-garden

i-akytyia-pa tahkome piemekyry-pahky

REL-make/cut-REM.P elder slowly-/
'The elders cut the garden plantation with the axe that was not made of steel, but
made of stone, wild pig teeth, and paca teeth.. .they cut the garden very slowly.'

18. impa iry iapykahp-e'me ny-kwahsa-pa ny-kwahsa-pa

then 3PRO survivor-VAL 3S-have sex-REM.P 3S-have sex-REM.P

ny-kwahsa-pa iry iapykahp-e'me n-iemkwa-pa

3S-have sex-REM.P 3PR0 survivor-VAL 3S-reproduce-REM.P
'Then they, the survivors, had sex, had sex, had sex and produced many children'

19. impa niriky n-yme-pa e'nypeme ampa n-yme-pa ampa

then son 3s-born-REM.Phelp other 3s-bom-REM.Pother

n-yme-pa.
3s-born-REM.P
'Then many sons were born to help them.'

20. impa tahkome n-iemkwe-pa piemkyry-pahky kaapy

then elder 3S-reproduce-rem.p slowly-? Heaven

wumaha ikaa pykyry many

falUng story tell and/with/too
'Then the elders reproduced very slowly, and for each child who was bom, they
told the story of the sky's falling.

21. ipaikypa iry i-aaia-pa Mawa tykatyka aa-pap-eme myry

After that 3PRO 3REL-take-REM.P Mawa say IPOS-father-DEV 3PRO

pykyry-maty
tell-?
'After that Mawa took that couple who survived said my father.'
 172

22. tykaty aa-pap-eme a'a i-nepiaka-ty tymatry-ky tete wapy

say iPOS-father-DEv 1+3PRO REL-? fall-T/A ? many

kaapy tymatry-ky tete wapy typohinji

heaven fall-T/A ? many hke
'The sky fell a lot, it liked to fall... said my father to make us afraid.'

23. iawura iawura iawura ky tete tahkome baka-ty wapy kaapy

always always always ? ? elder kill-? many heaven
'The sky always, always, always falls.. .killing many elders.'

Note: The elders said that the sky had fallen three times already. The last time was
when this pair of siblings survived.

24. iry patahkan-pa kanji wumaha n-oky-sa tykaty kara

3PRO substitute-REM.P this falling 3S-stay-T/A say ?

aa-pap-eme
iPOS-fkther-DEV
'That fallen sky was substituted by this sky that we have today... said my father.'

Note: This heaven that we have today will fall as well said the kinja

A5. Meie Ikaa

'The story of manioc bread'

1. kokyne a'a n-itxi-pia minja i-kwa-se

early 1+3PRO l+3S-go-LM.P manioc REL-take-in order to

kaapa taka
garden AL
'Early in the morning, we went to the garden to collect manioc tubers'

2. impa n-arem-pia mydy taka minja i-ne-pia

then 1+3S-come back-IM.P house AL manioc REL-?-IM.P

wyiepe ta
jamaxi(kind of basketery) Loc
'Then we came back to the house with the jamaxi full of manioc.'

3. iakypa a'a minja pitxi-pia maia ke

after 1+3PRO manioc peel-M.P knife INSTR
'After that we peeled the manioc with a knife,'
 173

4. iakypa a'a n-ikin-pia syna ke paxa taka

after l+3PROl+3-wash-LM.P water INSTR bowl AL
'After that we washed the manioc with water in the bowl.'

5. iakypa a'a ny-dexkie-pa tapinja ta

after 1+3PRO l+3S-squeez-REM.P sieve . Loc
'After that we squeezed the manioc in the sieve.'

6. iakypa a'a ny-ria-pa wiwi taka

after 1+3PRO l+3s-put-REM.P wood AL
'After that we put the manioc in the press.'

7. impa wiwi tapary n-emine-pa minja behe

then wood Loc 3S-dry-REM.P manioc dough
'Then we dried the dough.'

8. impa a'a n-ike'ia-pa meie impa-ry axinjaty tytinaka

then l+3PROl+3S-toast-rem.p bread then stone griddle on
'Then we toasted it into flat bread on the gridle.'

9. impa ny-tyky-pia.
then l+3S-finish-LM.P
'Then we finished.'

A6. Wyty Ikaa

'The story of the meat.'

1. kinja ram wyta-ha bykw-e waty taka

person 2PART meat-? cook-T/A fire AL
'Kinja cooks the meat in the fire.'

2. impa-ry n-iki-e tamxi-pesa waty tanyme

then-? 3S-take out-T/A soft-in the time fire ABL
'Then, when the meat become soft, they take it fi'om the fire.'

3. impa-ry kinja n-ynia-pa atyke-hkypa kinja n-iki-epa

then-? person 3S-put-REM.P cold-after that person 3S-take out-T/A

bixuwa taka
bowl AL
'Then, they took the meat from the pan and put it in a bowl.
 174

4. impa-ry bahinja n-ape'-ia tynaka

then-? child 3S-take-T/A ?
'Then the child takes the meat with the hands.'

5. naminja ny-tyt-epa bahinja pyna

dog 3S-come-T/A child close
'The dog came close to the child.'

6. iakypa bahinja ke-pa:

after child say-REM.P
'After that the child said:'

7. -piare! naminja wyty wanpa naminja

-piare dog meat NEG dog
Don't you have food, dog!'

8 ny-tyky-pia
3S-finish-LM.P
'Finished.'
 175

B. Part of Material elaborated by the Waimiri Atroari teachers after I explained to

them points of articulation

Kara mo pyky wenpa-typy

Language sound how study-used to
'What is used to study the sounds.'
(Phonology)

KYNYRY (tongue)
Kynyry nata (tip of the tongue)
Kynyry sopry (center of the tongue)
Kynyry iee kysa (laterals of the tongue)
Kynyry kadyhyry (back of the tongue)

KIEE PYNY WOTYKA (alveolar ridge)

Teeth cloth cover

KIEE IEPRY DYWE (palate)

Teeth
Places of articulation:

BILABIAL- kypyta bixi nytybaske sypyky naka. (The two lips get together);
P , b, kapaiky kypj^a bixi nybaske imany kynyraty tuwaha na kiee iepry pyky naka
wotyrykwaharyky na.
" When we say /p/ and Ihl our lips will get together and our tongue will be in the
middle, it will not touch the palate"

fMJ iatykapaiky ram kynata nyta ta napyke neky wapy.

" When we say /m/ the sound leaves through the nose."

AV/ kapaiky ram kareme wotyrykwaharyky kypyta bixi na.

"When we say /w/, the lips do not keep together the whole time"

ALVEOLAR- Kynyry nata notyrykwe kiee pyny wotyka pyky naka - The tip of the
tongue touch behind of the teeth, on the alveolar ridge)
 176

C. More Verbal Paradigms

Sa So
c. 1 -ikiam- 'spill' c.2 -tyryma- 'run away, escape'

Awy h-ikiam-pia 'I spilled' aa wy-tyryma-ky 'I ran away'

Amyry m-ikiam-pia 'you spilled' amyry my-tyryma-ky 'you ran away
Iry n-ikiam-pia 'she/he spilled' ka ny-tyryma-ky 'he ran away'
A'a n-ikiam-pia 'we spilled' (1+3) a'a ny-tyryma-ky 'we ran away'
kyky hy-tyryma-ky 'we ran away'

C.3 -pyty- ' think' C.4 -synehky- 'disappear'

Awy hy-pyty-pia 'I thought' wy-synehky-piany 'I disappeared'

Amyry my-pyty-pia 'you thought' my-synehky-piany 'you disappeared'
Ka ny-pyty-pia 'she/he thought' ny-synehky-piany' she disappeared'
A'a ny-pyty-pia 'we thought' (1+3) ny-synehky-piany 'we disappeared'
hy-synehky-piany 'we disappeared'
C.5 -ek- 'defecate'
aa w-eky-pia 'I defecated'
amyry m-eky-pia 'you defecated'
ka n-eky-pia 'she/he defecated'
a'a n-eky-pia 'we defecated'
kyky h-eky-pia 'we defecated'

C.6 Verb 'to push'

1A30 Aa ram k-eme h-aape'-pia
iPRO 2PART 3PRO-DEV lA-push-IM.P
'I pushed him'
2A30 Amyra ram k-eme m-aape'-pia
2PR0 2PART 3PRO-DEV 2A-push-IM.P
'You pushed him'
3A30 Ka ram k-eme i-aape'-pia
3PR0 2PART 3PRO-DEVREL-push-IM.P
'She/he pushed him'
1+2A30 Kyka ram k-eme h-aape'-pia
1+2PRO 2PART 3PRO-DEV l+2A-push-iM,p
'We pushed him'
1+3A30 A'a ram k-eme a'=i-aape'-pia
1+3PRO 2PART 3PRO-DEV l+3A-push-iM.P
'We pushed him'
3A1+20 Ka ram k-aape'-pia
3PRO 2PART 20-push-IM.P
'He pushed you'
1+3 A20 A'a ram a=i-aape'-pia
1+3PR0 2PART 20-push-IM.P
'We pushed you.'
 Ill

C.7
1A30 Aa ram ka h-irima-py-piany.
IPRO 2PART 3PR0 lA-rest-CAUS-REC.P
'I made him/her rest.'
2A30 Am5Ta ram ka m-irima-py-piany.
2PR0 2PART 3PRO 2A-rest-CAUS-REC.P
'You made him/her rest.'
3A30 Ka ram ka 0-irima-py-piany.
3PR0 2PART 3PRO 0-rest-caus-REC.P
'She/he made him rest.'
3A20 Ka ram a-irima-py-piany.
3PRO 2PART 20-rest-CAUS-REC.P
'She/he made you rest.'
1A20 Aa ram k-irima-py-piany.'
IPRO 2PART 20-rest-CAUS-REC.P
'I made you rest.

C.8 Verb to voitiit/ threw up

Isg Aa hu-wen-ta-pa
IPRO IS-vomit-VERBL-REM.P
'I threw up.'
2sg amyry mu-wen-ta-pa
2PR0 2S-vomit-VERBL-REM.P
'You threw up.'
3sg/pl mykyky nu-wen-ta-pa
3PR0 3S-vomit-VERBL-REM.P
'He/she/they threw up.'
1+2 incl kyky hu-wen-ta
1+2PR0 l+2S-vomit-VERBL-REM.P
'We threw up.'
1+3 excl a'a nu-wen-ta-pa.
1+3PR0 l+3S-vomit-VERBL-REM.P
'We threw up.'
D. List of animals and plants found in this sketch with their scientific names

Animals:

Akenbehe 'Sp. of armadillo' (Priodonts Maxims)

Akra' 'Sp. offish' (Aequidens)
Amakra 'tucunare fish' (Cichla Ocellaris)
Amana 'Amazonian dolphin'
Arawata 'Guariba monkey' (Ateles Paniscus)
Danja 'cicada'
lake 'alligator' (Caiman-crocodilus Crocodilus)
Fky 'Sp. of ant'
Inaxixi 'bat'
Kabaha 'Sp. of armadillo' (Dasypiis sp.)
Kadjiwi 'worm'
Kyiywaky 'parrot' (Amazona sp. Psittacidae)
Kyrywy 'any kind of snake'
Kwata 'spider monkey' {Ateles Paniscus)
Maba 'macaw' (Am ararauna-Psittacidae)
Makrykry 'pernilong'
Mare'e 'guam-bird' (Penelope Sp. Cracidae)
Mepry 'tapir' (Tapirus Terrestris)
Pakia 'Sp. of wild pig' (Tayassu Tajuca)
Petxi 'Sp. of wild pig' (Tayassu Pecari)
Peepe 'butterfly'
Poopo 'moth'
Pyryry 'frog'
Sadada 'shrimp'
Sady 'crab'
Sana 'termite'
Suweri 'deer' (Mazana Americana)
Tabe'e 'capybara' (Hydrochoerus capyhara)
Temere 'jaguar' (Panthera onca)
Waiamy 'tortoise' (Geochelone carbonaria)
Warara 'turtle' (Podocnemis spp.)
Waryna 'paca' (Agouti paca)
Wuky 'mutum bird' (Crax alector-cracidae)
Xeri 'skate fish'
Xirikiki 'parakeet' (Pyrrhura sp. Psittacidae)
 179

Fruits and Plants

Anahkwa-anaskwa 'Abiorana fruit' (Sapotacea)

Bixuwa 'cuieira' {Crescentia cujete, Bignoniaceae)
Hiri 'caju fruit' {Anarcadium Occidentale, Anacardiaceae)
Kymy 'bacaba berry' (Oenocarpus bacaba, palmae)
Mabaia 'papaya' (Carica papaya, caricaceae)
Merepy 'palm fruit' {Bactris gaspaes, palmae)
Mixi 'buriti fruit' (Mauritia flexuosa, palmae)
Nana 'pineaple' {Ananas Comosus, bromeliaceae)
Patuwa 'pataua berry' {Jessenia bataua, palmae)
Tetyky 'brazilian nut' {Bertholletia excelsa, Lecythidaceae)
Woky 'banana' {Musa spp, Musaceae)
 180

REFERENCES

AISSEN, Judith L. 1992. Topic and focus in Mayan. In Language 68.1.43-79.

ANDERSON, Stephen R. 1992. A-morphous-morphology. Cambridge: Cambridge Press.

ARCHANGELI, Diana. 2000. What Does OT tell us about language? Talk given at 12"'
SCIL at University of Arizona, Tucson. (Handout).

BAINES, S.G. 1991. a FUNAI quern sabe Frente de atragdo Waimiri AtroarL Belem:
Museu Paraense Emilio Goeldi/CNPq/SCT/PR.

. 1992. La raison politique de I'ignorance ou l'ethnologic interdite chez lezWaimiri

Atroari. Recherches Amerindienees au Quebec, V.xxii, n.l, pp 65-78.

. 1994. Epidemics, the Waimiri Atroari Indians and the politics of demography.
SerieAntropologia, 162, Brasilia: Departamento de Antropologia, Universidade de
Brasilia.

. 1997. Uma tradigao indigena no contexto de grandes projetos: os Waimiri

Atmariln Anudrio Antropologico 96. Rio de Janeiro, Tempo Brasileiro.

. 1999 "Imagens de lideran9as indigena e o Programa Waimiri Atroari: indios e usi-

nas hidroeletricas na Amazonia. In Serie Antropoldgica, n.246. Brasilia: Universi
dade de Brasilia.

BANDEIRA, A. 1926. Jauapery. Manaus.

BARBOSA RODRIGUES, J. 1885. Pacificagao dos Crichanas. Rio de Janeiro: Imprensa

Nacional.

BASSO, Ellen. 1977. Carib speaking Indians: culture, society, and language. In
Anthropological Papers of The University of Arizona, 28. Tucson: The University
of Arizona Press.

BASSO, Ellen. 1985. A musical view of the universe. Philadelphia: University of Pennsyl
vania. Press.

. 1987. In favor of deceit. Tucson: University of Arizona Press.

. 1995. The last cannibals: a South America oral history. Austin: University Press.

BLAKE, Barry. 2001. Case. Cambridge University Press.

 181

BRANIGAN, Philips and COLLINS, Chris. 1993. Verb movement and the quotative
construction in English. In Bobaljik, Jonathan and Philips Branigan (editors), MIT
Working Papers in Linguistics 18. Cambridge: MIT.

BRUNO, Ana Carla. 1995a. Proposta ortografica para lingua Waimiri Atroari. Manaus:
PWA.

. 1995b. Person subject and object marking in Waimiri Atroari. Unpublished

manuscript.

. 1996/1998. Fieldwork notes. Waimiri Atroari Reservation.

. 1999. The causative construction in Waimiri Atroari. Final paper for the Lexical
Semantics class. Tucson: University of Arizona.

. 2000. Reduplication in Waimiri Atroari. Department of Linguistics master's

paper. Tucson: University of Arizona.

CARNIE, Andrew and E GUILFOYLE, Eithne (editors). 2000. The Syntax of verb initial
languages. New York: Oxford University Press.

CARNIE, Andrew. 2001. Syntax. Oxford: Blackwell.

CARVALHO, Jose P. F. 1982. Waimiri Atroari: a hitdria que ainda nao foi contada.
Brasilia.Edifao do autor.

CHOMSKY, Noam. 1965. Aspects of the Theory of Syntax. Cambridge: MIT Press.

CHOMSKY, Noam. 1970 Remarks on nominalizations. In R. Jacobs and P. Rosenbaum

(eds). Readings in English Transformational Grammar. Waltham: Ginn, pp. 184-
221.

CHOMSKY, Noam 1991. Some notes on economy of derivation and representation. In R.

Friedin (ed.) Principles and parameters in comparative grammar. Cambridge:
MIT Press, pp.417-54.

DERBYSHIRE, Desmond. 1961. Hixkaryana (Carib) Syntax Structure. International

Journal of American Linguistics 27.125-42, 226-36.

. 1977. Word order universals and the existence of OVS languages. Linguistic
Inquiry 8.590-99.

. 1985. Hixkaryana and Linguistic Typology. Texas: Summer Institute of

Linguistics and University of Texas.
 182

DELANCEY, Scott. 1985. On active typology and nature of agentivity. Relational typo
logy, ed. by Frans Plank, 47-60. Berlin:Mouton.

DFFISING, Molly. 1992. Indefinites. Cambridge; MIT Press.

DffiSiNG, Molly and JELINEK, Eloise. 1995. Distributing arguments. Natural Languages
Semantics 3.123-176.

DIXON, R.M.W. 1979. Ergativity. Language 55.59-138.

Do VALE, Carmen Rezende. 2002. Waimiri Atroari em festa e.Maryba na floresta.

Manaus. Universidade Do Amazonas. Dissertagao de Mestrado em Antropologia.

ESPINDOLA, Claudia Voigt. 1995. O sistema Medico Waimiri Atroari: Concepfoes e

praticas. Florianopolis, Universidade Federal de Santa Catarina, Disserta9ao de
Mestrado em Antropolocia Social.

FOLEY, William A., and VAN VALIN, Robert D., Jr. 1984. Functional and Universal
Grammar. Cambridge: Cambridge University Press.

FRANCHETO, Bruna. 1986. Falar Kuikuro: Estudo etnolinguistico de um grupo Carib do

Alto Xingu. Rio de Janeiro, Museu Nacional. Tese de Doutorado.

GILDEA, Spike. 1995. Comparative Cariban syllable reduction. International Journal

ofAmerican Linguistics 61:62-102

. 1998. On reconstructing grammar: comparative Cariban morphosyntax. Oxford

Studies in Anthropological Linguistics 18. Oxford University Press.

GREENBERG, J.H. 1963. Universals of languages. MIT Press: Cambridge,MA.

HALPERN, Aaron. 1992. Topics in the placement and morphology of clitics. PhD
dissertation. Stanford, CA: Stanford University.

HALPERN, Aaron and ZWICKY, Arnold. 1996. Approaching second: second position
clitics and related phenomena. Stanford: CSLI Publications.

HAYES, Bruce. 1989. Compensatory lengthening in moraic phonology. Linguistic Inqui

ry 2Q\ 253-305p.

. 1995. Metrical stress theory: principle and case of studies. Chicago: University of
Chicago Press.

HARLEY, Heidi. 1995.SASE Bizarre; The structure of Japanese causative. MIT. 13p. ms.
 183

HILL, Joseph and HILL, Tamara, 1985.Phonological sketch of Waimiri Atroari (Cariban).
Unpublished manuscript. Missao Evangdica da Amazonia.

HOFF, Berend J. 1978. The relative order of the Carib finite verb and its nominal depen
dents. Studies on Fronting. Frank Jansen,ed., pp 11-27. Dordrecht, Netherlands;
Foris.

. 1986. Evidentiality in Carib69:49-103

JACKENDOFF, Ray. 1977. X' Syntax: A Study of Phrase Structure. Linguistic Inquiry
Monograph. Cambridge; MIT Press.

JELINEK, Eloise. 1984, Empty categories, case, and non-configurationality. Natural

Language and Linguistic Theory 2; 39-76.

KAISSE, Ellen M. 1981. Luiseno particles and the universal behavior of clitics. Linguistic
Lnquiry 12;424-34.

KAGER, Rene. 1999. Optimality Theory. Cambridge; Cambridge University Press.

KLOOS, Peter 1971. The Maroni River Caribs. Assen. The Netherlands; Gorcum.

LACERDA, Edith Maria Neves. 1991. Waimiri Atroari; observagSes linguisticas.

Unpublished manuscript. Manaus, Brazil; Programa Waimiri Atroari.

LACERDA, Edith and QUEIROZ, Carlos Augusto. 1991. Waimiri Atroari preliminary
dictionary. Manaus; PWA.

. 1996. Observafoes sobre morfologia e sintaxe da lingua Waimiri Atroari. Unpu

blished manuscript. Rio de Janeiro. Museu Nacional.

LEVIN, Beth. 2000. Aspects, lexical semantic representation, and argument expression.
Talk given at the Linguistic Colloquium at the University of Arizona,
Tucson.(Handout).

MARANTZ, Alec. 1982. Linguistic Inquiry J3. 453-83.

MATTEI MULLER, Marie-Claude. 1992. Yoroko: a Panare shaman's confidence. Caracas;

Armitano Editores.

MCCARTHY, John 1986. Prosodic Morphology, ms., University of Massachusetts,

Amherst, and Brandeis University, Waltham, Massachusetts.
 184

MCCARTHY, John and PRINCE, Alan. 1994. The emergence of the unmarked: Optimality
Theory. University of Massachusetts, Amherst, and Rutgers. 43p.ms.

. 1997. Faithfulness and identity in prosodic morphology. [ROA,

http://ruccs.rutgers.edu/roa.html].

MEIRA, Sergio. 1995a. The Accidental intransitive split in the Cariban family, ms

. 1995b. The Split-S system in Tiriyo: semantic versus morphological

explanations.32pp.ms

. 1999. A grammar of Tiriyo. Doctoral thesis. Houston: Rice University.

. 2002. Word class systems in Cariban Languages.ms.

MESTCR, R.Armin. 1988. Studies in Tier Stj'ucture. New York: Garland Publishing.

MILLEKEN, W.; MILLER, R. P. et al. 1992. Ethnobotany of the Waimiri Atroari Indians of
Brazil. Great Britain: The Royal Botanic Garden, Kew, Whistable Litho,

MITHUN, Marianne. 1991. Active/agentive case marking and its motivations. Language
61: 510-546,

. 1992. Is basic word order universal? In Pragmatics of word order flexibility,

Doris Payne (ed). Philadelphia: John Benjamins, pp. 15-61.

MONTE, Paulo Pinto. 1992. Etno-historia Waimiri Atroari (1883-1962). Dissertacao de

Mestrado em Ciencias Sociais, Sao Paulo, Pontificia Universidade Catolica de
Sao Paulo.

PAYNE, Doris and Marianne Mithun (editors). 1992. Pragmatics of Word Order
Flexibility. Typological Studies in Language 22. Philadelphia: John Benjamins.

PAYNE, Thomas. 1997. Describing Morphosyntax: A Guide for Field Linguists.

Cambridge: Cambridge University Press.

PRINCE, Alan and SMOLENSKY, Paul. 1993. Optimality Theory. Constraint interaction in
generative grammar. Ms., Rutgers University, New Brunswick and University of
Colorado, Boulder.

RADFORD, Andrew. 1988. Transformational Grammar. Cambridge: Cambridge

University Press.
 185

RADFORD, Andrew. 1997. Syntactic Theoiy and the structure of English: A minimalist
approach. Cambridge: Cambridge University Press.

RODRIGUES, Aryon. 1994. Grammatical affinities among Tupi, Carib, and Macro-Je.
Unpublished manuscript. Brasilia; Universidade de Brasilia.

SUVA, Marcio F. da. 1993. "O parentesco Waimiri Atroari: algumas observa9oes prelimi-
nares" In Viveiros de Castro, Eduardo de & Cunha, Manuela Carneiro de
{Or^s.)Amazonia etnologia e historia indigena. Sao Paulo, Nucleo de Historia
Indigenae de Indigenismo USP/FAPESP. p.211-228.

, 1995. "Sistemas dravidianos na Amazonia; o caso Waimiri Atroari"In Viveiros de

Castros, Eduardo (Org.), Antropologia do Parentesco estudos amerindios. Rio de
Janeiro, UFRJ. p.25-60

THOMPSON, Sandra and LONGACRE, Robert. 1985. Adverbial clauses. In Shopen,

Timothy (editor), Language Typology and Syntactic Description, v.2. Cambridge;
Cambridge University Press.

TRAVIS, Lisa. 1989. Parameters of phrase structure. In Baltin, Mark and Kroch, Anthony
(editors). Alternative Conceptions of Phrase Structure. Chicago; Chicago
University Press.

VAN VALIN, Robert D., Jr. 1990. Semantic parameters of Split Intransitivity. Language
66, 221-260.

VIEIRA, Marcia Damaso. 1995. The expression of quantification in Asurini do Trocara;

evidence against the universality of determiner quantification. In Bach, Emmon;
Jelinek, Eloise; Kratzer, Angelika; and Partee, Barbara (editors). Quantification in
Natural Languages. Boston; Kluwer Academic Publishers.

WUNDERLICH, Dieter and LAKAMPER, Renate. 2001. On the interaction of structural and
semantic case. Lingua. ILL; 277-418.