+ All Categories
Home > Documents > Tone in Acatlán Mixtec nouns · TONE IN ACATLÁN MIXTEC NOUNS . by Esteban I. Méndez-Hord...

Tone in Acatlán Mixtec nouns · TONE IN ACATLÁN MIXTEC NOUNS . by Esteban I. Méndez-Hord...

Date post: 24-Jan-2020
Category:
Upload: others
View: 10 times
Download: 0 times
Share this document with a friend
121
TONE IN ACATLÁN MIXTEC NOUNS by Esteban I. Méndez-Hord Bachelor of Arts, Bishop’s University, 2004 A Thesis Submitted to the Graduate Faculty of the University of North Dakota in partial fulfillment of the requirements for the degree of Master of Arts Grand Forks, North Dakota December 2017
Transcript

TONE IN ACATLÁN MIXTEC NOUNS

by

Esteban I. Méndez-Hord Bachelor of Arts, Bishop’s University, 2004

A Thesis Submitted to the Graduate Faculty

of the University of North Dakota

in partial fulfillment of the requirements

for the degree of

Master of Arts

Grand Forks, North Dakota December

2017

ii

© 2017 Esteban I. Méndez-Hord

iii

This thesis, submitted by Esteban I. Méndez-Hord in partial fulfillment of the requirements for the Degree of Master of Arts from the University of North Dakota, has been read by the Faculty Advisory Committee under whom the work has been done and is hereby approved.

___________________________________________________________________ Keith L. Snider, Chair

___________________________________________________________________ Stephen A. Marlett

___________________________________________________________________ James S. Roberts

This thesis meets the standards for appearance, conforms to the style and format requirements of the Graduate School of the University of North Dakota, and is hereby approved.

_______________________________________________ Grant McGimpsey Dean of the Graduate School

_______________________________________________ Date

CONTENTS

Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

List of figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

List of tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv

CHAPTER

1.  INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1.  Classification, location and dialect situation . . . . . . . . . . . 2

1.2.  Language consultants . . . . . . . . . . . . . . . . . . . . . . . 9

1.3.  Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.  PERTINENT INFORMATION ABOUT GRAMMAR AND PHONOLOGY 11

2.1.  Previous work . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.2.  Tone transcriptions . . . . . . . . . . . . . . . . . . . . . . . . 12

2.3.  Syntactic structure of noun phrases . . . . . . . . . . . . . . . 13

2.4.  Noun morphology . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.4.1.  Classifier prefixes . . . . . . . . . . . . . . . . . . . . . 15

2.4.2.  Grammatical gender . . . . . . . . . . . . . . . . . . . . 22

2.5.  Consonants and vowels . . . . . . . . . . . . . . . . . . . . . . 26

v

2.5.1.  Consonants . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.5.2.  Vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2.6.  Nasalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.7.  The structure of nouns . . . . . . . . . . . . . . . . . . . . . . . 34

3.  TONE IN THE NOUN SYSTEM . . . . . . . . . . . . . . . . . . . . . 40

3.1.  Surface tone patterns . . . . . . . . . . . . . . . . . . . . . . . 40

3.2.  Hi-tone spreading across word boundaries . . . . . . . . . . . . 43

3.3.  Hi-tone spreading within the root . . . . . . . . . . . . . . . . 46

3.4.  Surface versus underlying HL . . . . . . . . . . . . . . . . . . . 48

3.5.  Surface versus underlying L . . . . . . . . . . . . . . . . . . . . 51

3.6.  Underlying tone patterns . . . . . . . . . . . . . . . . . . . . . 54

3.7.  Subject-object tonal interaction in transitive sentences . . . . . 63

3.8.  The historical source of floating tones . . . . . . . . . . . . . . 73

3.9.  Underspecification . . . . . . . . . . . . . . . . . . . . . . . . . 77

4.  CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

4.1.  Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

4.2.  Areas for further research . . . . . . . . . . . . . . . . . . . . . 86

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

vi

ABBREVIATIONS

1 first person3 third persona adjectivezoa gender of animals, round things, fruit and celestial beingscompl completivedeg degree worddiv gender of divine beingsdist distal demonstrativef femininegen genitivehon honorificHTS Hi-tone spreadinginfant infantliq gender of liquidsm masculinen nounnp noun phrasen/hon non-honorificOCP Obligatory Contour Principlealt generic genderhum gender of persons

vii

pl pluralq quantifierreal realissg singularTBU tone-bearing unitarb gender of trees and wooden things

viii

LIST OF FIGURES

Figure Page

1.  Mixtecan language groups . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.  Map of Mixtecan language groups and Amuzgo . . . . . . . . . . . . . . 3

3.  Map of Mixtec dialect areas . . . . . . . . . . . . . . . . . . . . . . . . . 7

ix

LIST OF TABLES

Table Page

1.  Where the Xayacatlán dialect of Acatlán Mixtec is spoken . . . . . . . . 5

2.  Where Northern Mixteca Baja varieties are spoken . . . . . . . . . . . . 7

3.  Grammatical gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

4.  Relationship between nouns and their gender . . . . . . . . . . . . . . . 25

5.  Consonantal surface sounds . . . . . . . . . . . . . . . . . . . . . . . . . 27

6.  Vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

7.  Roots where both vowels in the root have a shared identity . . . . . . . 29

8.  Monomorphemic nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

9.  Possible trimoraic monomorphemic nouns . . . . . . . . . . . . . . . . . 35

10.  Roots with an animal classifier prefix (§ 2.4.1) . . . . . . . . . . . . . . 36

11.  Couplets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

12.  Surface patterns of simple noun roots. . . . . . . . . . . . . . . . . . . 43

13.  Basic tone patterns in isolation . . . . . . . . . . . . . . . . . . . . . . 55

14.  Behaviour of a Hi-toned quantifier before a noun . . . . . . . . . . . . 56

15.  Behaviour of a Lo-toned quantifier before a noun . . . . . . . . . . . . 57

16.  Floating tone association from a noun onto another noun . . . . . . . 61

17.  Behaviour of /LH/ subjects . . . . . . . . . . . . . . . . . . . . . . . . . 64

x

18.  Behaviour of /H/ subjects . . . . . . . . . . . . . . . . . . . . . . . . . 66

19.  Behaviour of /HLH/ subjects . . . . . . . . . . . . . . . . . . . . . . . . 68

20.  Behaviour of /HL/ subjects . . . . . . . . . . . . . . . . . . . . . . . . . 70

21.  Behaviour of /L/ subjects . . . . . . . . . . . . . . . . . . . . . . . . . 72

22.  On the origin of floating tones in Acatlán Mixtec . . . . . . . . . . . . 75

23.  Simple nouns with an underlying /L/ tone pattern . . . . . . . . . . . 89

24.  Simple nouns with an underlying /H/ tone pattern . . . . . . . . . . . 91

25.  Simple nouns with an underlying /HL/ tone pattern . . . . . . . . . . 93

26.  Simple nouns with an underlying /LH/ tone pattern . . . . . . . . . . 95

27.  Simple nouns with an underlying /LH/ tone pattern . . . . . . . . . . . 97

28.  Simple nouns with an underlying /HLH/ tone pattern . . . . . . . . . . 100

xi

ACKNOWLEDGMENTS

Nakímani´ Tátá Sa’nu xi´ nsi´ Franco Tomás Méndez Gómez, vachi ni chiná’á-ne´ yu’u ndó’í sa’a´ Ñúu´ Davi´ xi´ sa’a´ da’an´ davi´ xi´ sa’a´ ne’ívi´ davi´. Ténakímani´ nsídáá´ ne’ívi´ davi´ vachí ni chíndéé´-ne´ yu’u kídá yúkin tútu nchichi yo’o ndo Keith L. Snider, xi´ ndo Stephen A. Marlett, xi´ ndo James S. Roberts.Nechi´ yo’o ni chindee yu’u ansa kachi palabra ko ina’i sa’a´ tutu yo’o. Te nakí-mani´ sastnu’i nuu´ nechi´ iin sa’nu nuu´ ñanchichí té mii-ne kunini-ne´ ña-ndaani nduki te ni kundu’e sa’a´ chuún yo’o. Te nakímani´ nuu´ nechi´ ni kachi xi´ansa´ ka’an´ vivii da’an yo’o.

I am especially grateful to the chair of my committee, Dr. Keith L. Snider,whose guidance in data collection, analysis and writing have been crucial and towhom I owe much of my understanding of tone. I never thought I would be blessedby having someone on my committee who had already written about the languagespoken in the very village my family is from. I am also indebted to Dr. Stephen A.Marlett, whose expertise in Oto-Manguean linguistics has proved to be invaluable,and to Dr. James S. Roberts, who guided me in my first steps in phonology andhas contributed to my understanding of tone.

While their input has been essential for this thesis, they have allowed me topursue my own analysis and writing, and any errors contained in this thesis areundoubtedly mine.

I thank the people of Xayacatlán, my uncle Efrén Méndez Peralta, who lodgedme during my stays in the village, and I especially want to thank the language con-sultants I worked with, Joel Cruz Martínez, Roberto Martínez Rojas, Élfega Méndez

xii

Lázaro, Gervacio León Peralta, Aurelia López Álvarez, Amelia Leonor Méndez Per-alta†, Leobigildo Vázquez Cruz, Gudelia Mendoza Martínez, Elia López Méndez,and four others who requested anonymity. I recall that one language consultantinsisted that her name be mentioned, since the research will be a legacy for hergrandchildren.

I thank the Estate of Dr. Kathryn J. Josserand, administered by Dr. NicholasA. Hopkins, for the permission extended to reproduce her map on Mixtec dialectareas. I thank John and Judith Williams for their information on root-final glottalstops in Ayutla Mixtec, Melquiades Gregorio Porfirio for data on Ayutla Mixtec,Laura Gittlen for her insight on nasalization in Ñumí Mixtec, Moss Doerksen forthe map of Mixtecan language groups and Amuzgo, Dr. H. Andrew Black for histechnical assistance with the XLingPaper software, and Dr. Rod Casali for histechnical assistance with the Dekereke phonology software tool. I would also liketo recognize the Mixtecanists who have infected me with their enthusiasm for theMixtec language, Ruth Mary Alexander†, Margaret H. Daly, Dr. Inga McKendry,Priscilla C. Small, Rodrigo Vázquez Peralta, Kent Wistrand and my uncle ArtemioCruz Domínguez.

I also thank the Endangered Language Fund for the financial support theyprovided through the 2011 Language Legacies Grant for the research conducted aspart of the the Xayacatlán Mixtec Multimedia and Documentation Project.

I thank my father, Refugio Méndez Domínguez, my mother, Marjory Hord deMéndez, my grandparents, Harris and Lois, and my sister, Linda, for their sup-port. I thank my wife, Laurie, and my children Yuyu, Perry and Cuquita, whoselove, patience and support have given me the strength necessary to complete thisendeavour.

xiii

Dedicated to Franco Tomás Méndez Gómez, my grandfather,

whose love for the Mixtec region (Ñúu´ Davi´), the Mixtec people (ne'ívi´ davi´)

and our language has inspired me.

ABSTRACT

Varieties of Mixtec (an Oto-Manguean language group spoken in southernMex-ico) have often been analyzed as having a three-way tonal distinction: Low, Midand High. I present evidence from original research on simple, alienable nouns,that only two lexical tones, /H/ (corresponding to the traditional Mid tone) and/L/, where some H tones are floating, are needed to describe noun roots in AcatlánMixtec. In essence, the extra-H tone (corresponding to the traditional Hi tone) onlyoccurs in derived environments, which involve the interaction of tones from twoor more morphemes. The bulk of the analysis uses H and L to represent the twotones, but in the final section I will present evidence that low tones are underly-ingly unspecified.

Acatlán Mixtec (spoken chiefly in the state of Puebla, Mexico) has attractedsome attention due to the claim made in E. Pike and Wistrand (1974) that the typo-logically uncommon tonal process known as “iterative upstep” exists in this varietyof Mixtec, and its theoretical implications potentially dispute the universal na-ture of the Obligatory Contour Principle (Snider 1988, Snider 1999 and Aronovich1994). Although upstep is beyond the scope of this thesis, this study lays thegroundwork for exploring upstep in Acatlán Mixtec in the future, especially thedescription of the raising effects of floating H tones.

I propose that only six underlying tone patterns account for all noun roots(/H/, /L/, /LH/, /HL/, /HLH/ and /LH/). I also propose that three tone processes(Hi-tone spreading, Hi-tone raising, Hi-tone lowering) and an OCP condition (the

xiv

two-slot condition, which is an instantiation of the Obligatory Contour Principledescribed in this study) account for the surface realizations of these nouns.

xv

CHAPTER 1INTRODUCTION

The variety of Acatlán Mixtec studied in this thesis is spoken by about 2900people in Xayacatlán in the state of Puebla, Mexico.1 It is one of the varieties clas-sified as mixteco de la frontera Puebla-Oaxaca (Puebla-Oaxaca border Mixtec) bythe Mexican government agency INALI (2008), and belongs to a larger group ofvarieties called Northern Mixteca Baja by Josserand (1983). Two Northern Mix-teca Baja varieties are spoken in the central part of the state of Puebla, and therest of them straddle the southern border of the state of Puebla and the northernborder of the state of Oaxaca. Unless otherwise indicated, the specific variety ofAcatlán Mixtec discussed is the Xayacatlán one, which is listed in Simons and Fen-nig (2017), along with some moribund varieties, with the ISO 639-3 code [mit].Local speakers refer to their language as da’an´ davi´ [ðaʔa ðaβi ].

Acatlán Mixtec was the first language to have been reported as having iterativeupstep tone (E. Pike and Wistrand 1974). Snider (1988), Aronovich (1994) andSnider (1999) have also analyzed the tonal system of Acatlán Mixtec based on Pikeand Wistrand’s data, but no original research has been published ever since.

Since the scope of this thesis is limited to the study of nouns in Acatlán Mixtec,the phenomenon of tonal upstep is hardly discussed, but the present study providesa basis for future research on upstep. I present evidence of the existence for theunderlying tone level Hi (H), the tone level Lo (L), and a floating Hi tone (H)that usually has a raising effect on the following word, distributed in six basic

1 This figure is based on data from INEGI (2010) for speakers 5 years and older in the municipalities ofAcatlán, Xayacatlán de Bravo, San Jerónimo Xayacatlán and Petlalcingo (Colonia El Progreso).

1

underlying tone patterns in noun roots: /H/, /L/, /LH/, /HL/, /HLH/ and /LH/.In § 3.9, I will also argue that ultimately, Lo tones are underlyingly unspecified(∅) for tone in Acatlán Mixtec, but until that point I represent the underlyinglytoneless tone-bearing units as /L/ in order to avoid confusion. This analysis hingeson the proposals, made initially in Snider (1999), that some tone-bearing unitsmay be unspecified for tone in Acatlán Mixtec, and that low tone is underlyinglyunspecified in Acatlán Mixtec.

I document the following tone processes: Hi-tone spreading, Hi-tone raisingand Hi-tone lowering.

This thesis will also be useful to further the comparative and historical studyof tone in Mixtecan languages. As discussed in Dürr (1987) and Josserand (1983),Proto-Mixtec had glottal stops in coda position, including word-finally, as evi-denced in Ayutla, Zacatepec and Santiago Nuyoó Mixtec (Pankratz and E. Pike1967:287, Dürr 1987:20 and Harris 1995:79 respectively). The disappearance ofthe glottal stop word-finally left a floating tone in several words, so I describe howthese floating tones are realized in Acatlán Mixtec. In conclusion, I demonstrateby comparing different combinations of surface realizations what the underlyingpatterns of nouns are based on original research.

1.1 Classification, location and dialect situation

Mixtec is a group of related varieties spoken by approximately 517,600 peo-ple in Mexico, according to the national census (INEGI 2015), making it the thirdmost widely-spoken indigenous language group in North America, after the Mayanlanguages (2,442,400 speakers in Mexico) and Nahuatl (1,725,600 speakers inMexico). There is also a significant Mixtec-speaking immigrant population in theUnited States. According to some estimates, reported in The San Diego Union-Tribune (2011), Mixtec speakers number around 150,000 in California, and 25,000

2

to 30,000 in New York City. Mixtec may be the most spoken indigenous NorthAmerican language group in the United States, surpassing most local indigenouslanguages in the country.2

Mixtec is a branch of the Mixtecan family, which in turn is a branch of theOto-Manguean language family. The region encompassing all Mixtecan languagesis known as La Mixteca, and it is located in three states in Mexico: Puebla, Guer-rero and Oaxaca. The Mixtecan family includes three language groups: Cuicatec,Mixtec and Triqui.

The existence of an Amuzgo-Mixtecan language family has been proposed(Campbell 1997:158), based on Kaufman (2016:8), but is not generally accepted.If such a language family exists, it would include two separate genera: Amuzgoan,and Mixtecan, as seen in Figure 1. While the exact genetic relationship betweenthe three language groups is disputed, Longacre (1961:12) states the following re-garding Mixtec, Cuicatec and Triqui: “Mixtec-Cuicatec share slightly more basicvocabulary than do any other two of the three languages that I consider to becoordinate.”

Figure 1. Mixtecan language groups and their possible position within theproposed Amuzgo-Mixtecan family

2 The most spoken indigenous language group in the United States, according to the American CommunitySurvey Report (2011), endorsed by the U.S. Census Bureau, is Navajo, with 169,000 speakers.

3

In Figure 2 we see where the Mixtecan language groups and Amuzgo are lo-cated. In the northern part of the Mixtec region we see the Northern Mixteca Bajadialect area proposed in Josserand (1983), to which Acatlán Mixtec belongs.

Figure 2. Map of Mixtecan language groups and Amuzgo (by Moss Doerksen,used with permission)

Acatlán Mixtec includes several dialects. The dialect of Acatlán Mixtec studiedhere, the Xayacatlán dialect, is spoken in three municipalities in the state of Puebla:

4

Table 1. Where the Xayacatlán dialect of Acatlán Mixtec is spokenMunicipality Coordinates PrecisionsXayacatlán de Bravo 18.235, -97.975San Jerónimo Xayacatlán 18.221, -97.914 Except in the town of

Tonahuixtla, where a sepa-rate variety is spoken.

Petlalcingo 18.133, -97.964 Only in the town of ColoniaEl Progreso. Separate fromthe Petlalcingo variety.

Due to recent migration, there are also speakers in Acatlán de Osorio (a citylocated 14 km southwest of Xayacatlán de Bravo), Mexico City, and New York City(especially in the borough of Staten Island).

Kathryn Josserand carried out a comprehensive survey of Mixtec between 1977and 1981 and published her results on Mixtec dialectology in Josserand (1983),where she classified Mixtec varieties in “dialect areas” based on phonological, mor-phological, syntactic and lexical evidence (see Figure 3). She included the follow-ing varieties in the northernmost Mixtec dialect area, Northern Mixteca Baja: Chig-mecatitlán, Santa Catarina Tlaltempan, Xayacatlán de Bravo, San Jerónimo Xay-acatlán, Totoltepec de Guerrero, Tonahuixtla, Cosoltepec, Chazumba, San MiguelIxtapan, Tepejillo, El Rosario Micaltepec and Zapotitlán Palmas.

Grimes (1974) also identified the Northern Mixteca Baja dialect area throughintelligibility testing of northern Mixtec varieties. He included Chigmecatitlán,Xayacatlán de Bravo, San Jerónimo Xayacatlán, Cosoltepec, Chazumba, Petlal-cingo and Zapotitlán Palmas.

Simons and Fennig (2017) list Chigmecatitlán [mii], Southern Puebla Mixtec[mit] and Chazumba Mixtec [xtb] as separate varieties. Josserand (1983), Grimes(1974) and Simons and Fennig (2017) recognize Chigmecatitlán as a very distinctvariety (in fact, Simons and Fennig 2017 report that it has 23% mutual intelligibil-ity with Chazumba Mixtec, reportedly the closest variety), but no published study

5

supports further subdividing the other dialects within Northern Mixteca Baja. Si-mons and Fennig (2017) report that Southern Puebla Mixtec has 87% intelligibilitywith Chazumba Mixtec, so it is clear that there is a high level of mutual intelligi-bility across the Northern Mixteca Baja, with the exclusion of Chigmecatitlán.

INALI (2008) classifies most Northern Mixteca Baja varieties under the namemixteco de la frontera Puebla-Oaxaca (Puebla-Oaxaca border Mixtec). They excludeChigmecatitlán, Tlaltempan and Atzumba from this group.

The grouping classified by INALI (2008) roughly resembles the grouping de-scribed as “Mixteco Bajo dialects” in Catecismo (1837), which says that these vari-eties are spoken in two vicaries in the Bishopric of Puebla: the Vicary of Huajuapan(including the parishes of Huajuapan, Chila, Chazumba, Huapanapan, Tequixtepeqand Huajolotitlan) and the Vicary of Acatlán (including the parishes of Acatlán,Totoltepeq, Petlaltzingo and Zacatepeq).3 Mixtec is no longer spoken by locals inHuajuapan, Chila and Huajolotitlán.

The Chigmecatitlán variety has very low intelligibility with other NorthernMixteca Baja varieties because it developed as a linguistic island of Mixtec sur-rounded by Nahuatl and Popoloca. Josserand further divides the Chigmecatitlánvariety into two dialects: Chigmecatitlán and Tlaltempan. These varieties origi-nated in the municipality of San Pedro y San Pablo Tequixtepec, Oaxaca, and de-veloped as a linguistic island of Mixtec after a migration in 1646 (Flechsig 2004:21-22).

3 Xayacatlán was in the parish of Totoltepeq at the time.

6

Table 2. Where Northern Mixteca Baja varieties are spokenMunicipality Coordinates Speakers

(INEGI2010)

Precisions

Chigmecatitlán,Puebla

18.646, -98.075 829

Santa Catarina Tlal-tempan, Puebla

18.613, -98.080 516

Acatlán, Puebla 18.202, -98.048 393 Some locals still spoke Mix-tec in the early 1900s. Cur-rent speakers are all origi-nally from other towns.

Xayacatlán de Bravo,Puebla

18.235, -97.975 695

San Jerónimo Xaya-catlán, Puebla

18.221, -97.914 1779

Petlalcingo, Puebla 18.133, -97.964 438 Only in Colonia El Progreso,Micaltepec and Tepejillo.

Santiago Chazumba,Oaxaca

18.144, -97.732 529 Extinct in San SebastiánFrontera and in the maintown, Chazumba, wherelocals no longer speakMixtec.

Cosoltepec, Oaxaca 18.142, -97.790 176San Pedro y San PabloTequixtepec, Oaxaca

18.106, -97.754 277 Except in the main town,Tequixtepec, where locals nolonger speak Mixtec.

Zapotitlán, Puebla 18.159, -97.583 954 San Pedro Atzumba.Popoloca is spoken inthe rest of the municipality.

Zapotitlán Palmas,Oaxaca

17.889, -97.818 287 Moribund.

Totoltepec de Guer-rero, Puebla

18.22, -97.855 56 Moribund.

San Miguel Ixitlán,Oaxaca

17.999, -97.773 4 Moribund.

Huajuapan de León,Oaxaca

18.003, -97.817 Unknown Towns on the northern tip ofthe municipality. Moribund.

7

Figure 3. Map of Mixtec dialect areas (by Josserand, used with permission)

8

1.2 Language consultants

The data in this thesis was collected between 2011 and 2017 as approvedby the Institutional Review Board of the University of North Dakota in projectIRB-201108-031. The data collection sessions were part of the Xayacatlán MixtecMultimedia and Documentation Project, funded by the Endangered Language Fundas part of the 2011 Language Legacies Grant.

I selected nine consultants who speak the Xayacatlán variety of Acatlán Mix-tec, one speaker of Tonahuixtla Mixtec, one speaker of Totoltepec Mixtec, onespeaker of Zapotitlán Palmas Mixtec and one speaker of Atatláhuca Mixtec. Onlythe data from Xayacatlan are discussed in this thesis, but data from other Mixtecdialects and languages contributed significantly to my understanding of AcatlánMixtec. Nine consultants expressed their desire to have their names published:Joel Cruz Martínez (Tonahuixtla), Roberto Martínez Rojas (Xayacatlán de Bravo),Élfega Méndez Lázaro (Xayacatlán de Bravo), Gervacio León Peralta (Xayacatlánde Bravo), Aurelia López Álvarez (Gabino Barreda), Amelia Leonor Méndez Peralta(Xayacatlán de Bravo), Leobigildo Vázquez Cruz (Xayacatlán de Bravo), GudeliaMendoza Martínez (Xayacatlán de Bravo) and Elia López Méndez (Xayacatlán deBravo). Four additional consultants requested that their names be kept confiden-tial.

Eight speakers of the Xayacatlán variety are from Xayacatlán de Bravo andone of them is from Gabino Barreda, a town in the municipality of San JerónimoXayacatlán. They are all bilingual in Spanish and Mixtec and speak Mixtec as theirfirst language; one of them has a limited knowledge of Spanish.

Three consultants were between the ages of 50 and 60, five between the agesof 60 and 70 and five over 70. Six were raised in monolingual households, twowere raised by parents who knew very little Spanish, two had a bilingual fatherand a monolingual mother, and three were raised in bilingual households.

9

1.3 Methodology

Recordings of linguistic data took place at my home in Puebla City and atpeople’s homes in Mixtec villages, mostly on weekends. At my home there wasnoise caused by my daughter and occasionally children’s TV shows; in the village,recordings of farm animals can be heard, as well as noise produced by visitors andfamily members of consultants. Several recordings were made outside people’shomes, among farm animals.

Most recordings were made with a Shure PG27-USB Cardioid Condenser USBMicrophone and captured directly in WAV format with Audacity. On some occa-sions, the laptop’s internal microphone was used. In the last stage of data collec-tion, I was not in Mexico, so some data were obtained over the phone.

Each language consultant provided informed consent before research began.The formwas read orally in Spanish to each consultant and the consultant’s consentwas recorded.

The data collection and analysis was based on the methodology described inSnider (In press). Data was entered into the Dekereke phonology assistant softwaredesigned by Rod Casali. The database consists of 2906 entries. Among them, 523are nouns.

10

CHAPTER 2PERTINENT INFORMATION ABOUT GRAMMAR AND

PHONOLOGY

2.1 Previous work

The first study of Acatlán Mixtec is contained in pages 3-9 of a catechismpublished in Spanish and Mixtec by an anonymous priest who did not specify whatdialect was being described (Catecismo 1899). He did mention that he was Mixtechimself, and the dialect is clearly very close to the Xayacatlán dialect. It mayhave been the Xayacatlan dialect as it was spoken in 1899, it may have been thedialect of one of the many villages surrounding Acatlan, or it may have been thenow extinct variety that was formerly spoken in the city of Acatlan.1 The authoraddresses grammatical, phonological and tonal issues.

The second study of Acatlán Mixtec was published 75 years later. Eunice Pikeand Kent Wistrand dedicated most of their article (E. Pike and Wistrand 1974) totone, focusing especially on the phenomenon of upstep. Six pages are devoted to astudy of the segmental phonology of Acatlán Mixtec from a structuralist perspec-tive, including descriptions of the phonemes of Acatlán Mixtec. This article focusessolely on the Xayacatlán dialect of Acatlán Mixtec. Three tone analyses are basedon Pike and Wistrand’s data: Snider (1988), Aronovich (1994) and Snider (1999).

Lastly, Vázquez Peralta (1997), a speaker of the Xayacatlán dialect of AcatlánMixtec, produced a pedagogical grammar where he applies the alphabet prescribed

1 Even though Acatlán Mixtec is named after the city of Acatlán de Osorio, Puebla, Mixtec is not currentlyspoken in Acatlan, except by migrants from Xayacatlan. It was named Acatlan Mixtec after the capital of thedistrict where it is spoken.

11

by Ve’e Tu’un Savi, the Academy of the Mixtec Language, to Acatlán Mixtec. Hedescribes the consonants and vowels of Mixtec, provides examples, and lays out apedagogical grammar.

Some Mixtecanists have worked on a reconstruction of Proto-Mixtec: Lon-gacre (1957), Mak and Longacre (1960), Bradley and Josserand (1982), Josserand(1983), Dürr (1987) and McKendry (2013). This thesis concentrates on the syn-chronic study of Acatlán Mixtec. Even though this discussion is dialect specific,dialectal divergence and Proto-Mixtec are occasionally addressed.

2.2 Tone transcriptions

Mixtecanists have traditionally analyzed Mixtec languages as having a three-way contrast for tones: Low, Mid and High. Acatlán Mixtec is not different inthis respect. However, my analysis does not confirm the need for a three-waycontrast, at least underlyingly. The present analysis does not include a mid tone,so the transcriptions in this thesis do not use it, not because Acatlán Mixtec has adifferent set of tones than other varieties, but rather to reflect the fact that a midtone is not warranted in the final analysis of this thesis.

The three marks used here in transcriptions are taken from IPA practice andconstitute idealized pitch targets. The grave accent ( ` ) represents low pitch, andis no different from the traditional approach and its system for marking it. Theacute accent ( ´ ) represents the higher pitch level, and is always marked as suchin this thesis; this is roughly equivalent to the mid tone of the traditional approach,which has sometimes been unmarked, and sometimes represented with a macron.Finally, the double acute accent ( ˝ ) is used to represent the “extra high” pitch,and this is roughly equivalent to the high tone of the traditional approach, whichhas been represented by the simple acute accent.

Since these are idealized pitch targets, phonetic detail is not fully described.For example, the first mora of utterances that begin with more than one low tone

12

is prominent and has a slightly higher pitch (but not as high as a high tone) thanthe low tones of the following moras. This phonetic detail is not represented inthe transcriptions.

2.3 Syntactic structure of noun phrases

The intent of this section is to provide basic syntactic information about nounphrases as background for the discussion of the phonological data in Chapter 3 onthe tone of nouns.

The basic word order of Mixtec is VSO (Macaulay (1996:101), Eberhardt (1999:1-2), Castillo García (2007:18), McKendry (2013:28)), as can be seen in (1).

(1) ni compl

ʃi i buy

[ te3m.n/hon

hwaaJuan

]sbj [ ðitatortilla

]obj

‘Juan bought tortillas.’

The following constituents were identified in the data: the quantifier phrase,the noun, the adjective phrase and the determiner. Quantifier phrases precede thehead noun, as exemplified in (2), where the head noun [ðita] ‘tortilla’ follows aquantifier phrase, whose quantifier head [kwaʔa] ‘many’ is in turn modified by adegree word.

(2) [ kʷaʔamany

sastnuʔudeg

]qp ðitatortilla

‘very many tortillas’

Within a noun phrase, the head noun may also be followed by an adjectivephrase. If the noun phrase has a determiner, it is placed at the end of the nounphrase. Both of these facts can be seen in example (3).

(3) ðitatortilla

[ ʔi ʔni hot

titi deg

]adjp ʝukadist

‘those extremely hot tortillas’

13

The noun can occur without a determiner, as seen in example (4).

(4) ni compl

ʃiʃi real.eat.1sg

ðitatortilla

‘I ate tortillas.’

Genitive constructions in Mixtec are constructed by embedding a phrase re-ferring to the possessor within the overall noun phrase. In the case of head nounswhich are alienable, the embedded element is a prepositional phrase headed bythe genitive preposition [ʃi ] that is followed by a possessive noun phrase, in thiscase the proper noun [hwaa ] ‘Juan’ (note that in Acatlán Mixtec proper namesare preceded by a classifier to indicate the age, honorific status and gender of theperson, in this case the classifier [te]), as seen in example (5).

(5) siʔi nadog

[ ʃi gen

[ te3m.n/hon

hʷaaJuan

]np ]pp

‘Juan’s dog’

In the case of head nouns which are inalienable, the embedded element is anoun phrase. Note that the possessor is expressed without a genitive preposition,as shown in (6).

(6) a. ndaʔahand

[ te3m.n/hon

hʷaaJuan

]np

‘Juan’s hand’

b. * ndaʔahand

ʃi gen

[ te3m.n/hon

hʷaaJuan

]np

‘(Juan’s hand)’

2.4 Noun morphology

Mixtec has been analyzed as having a classifier system rather than a noun classsystem (León Pasquel 1988:117, Macaulay 1987:127 and McKendry 2001:41). In

14

her dissertation on Mixtec and Tzotzil classifiers, León Pasquel (1988:20-22) ul-timately analyzes Mixtec as having a classifier system, but recognizes that Mix-tec classifiers share crucial characteristics with noun classes, to the point whereshe says that “noun classifiers are intermediate between the morphological nounclasses and the lexical numeral classifiers.”

2.4.1 Classifier prefixes

León Pasquel (1988), Gittlen and Marlett (1985:189), Macaulay (1987:127),McKendry (2001:41) and McKendry (2013:27) analyze the pre-nominal morphemepreceding some nouns as a classifier.

While McKendry (2013:38-40) concludes that nouns preceded by a classifierare in fact compounds, León Pasquel (1988:142-143) proposes the existence of“classifier phrases” instead, where the classifier is the head.

In Acatlán Mixtec, the most common pre-nominal morphemes are the classifierprefixes [si-] and [tʃi-] for animals, [nu-] for trees, [te-] for men and [te-] forliquids. There is also a rare classifier for animals, [ntsi-].

In the case of the morpheme that commonly precedes animals, de los Reyes(1593:20), Macaulay (1987:125) and McKendry (2013:36) propose it is derivedfrom the word ‘animal’. McKendry (2013:36) writes (where MXY stands for South-Eastern Nochixtlán Mixtec):

First we look at the animal classifier which in MXY is tʃi(H)-. This is mostlikely derived from the second syllable ―that is, the unstressed syllable— ofthe noun kitiMH animal. In many other varieties of Mixtec the cognate clas-sifier is ti- or tɨ-. In MXY there are many instances of where [tʃ] correspondsto [ti-] or [tɨ-] ‘animal classifier’ in other varieties.

In Acatlán Mixtec, the animal classifier prefix has historically diverged intotwo forms: [si-] and [tʃi-]. The first of these is also the last syllable of the noun

15

[kisi] ‘animal’; the second, which only occurs on a few nouns, is also historicallyderived from the Proto-Mixtec word *kɨtɨɁ ‘animal’. In (7) we see examples ofnouns with the classifier prefixes [si-] on the left column and [tʃi-] on the rightcolumn.

(7) si- tʃi-

[si-sita] ‘possum’ [tʃi-lolo] ‘tadpole’

[si-ðuʔma] ‘scorpion’ [tʃi-ɲaʔɲa] ‘lizard’

[si-tni i ] ‘mouse’ [tʃi-ni i ] ‘iguana’

[si-ðiʔu] ‘goat’ [tʃi-ðiʔi] ‘bat’

According to León Pasquel (1988:135, 146) the animal classifier has been se-mantically extended to “round shapes” and “round fruits.” In (8) we see examplesof round things with the [si-] classifier on the left column and round things withthe [tʃi-] classifier on the right column.

(8) si- tʃi-

[si-βitʃi] ‘woven palm fan’ [tʃi-mbelu ɲuʔu] ‘mushroom’

[si-kwisi] ‘thread ball’ [tʃi-kolo] ‘oblong corn-cake’

[si-ðaʔnde] ‘corn-cake’ [tʃi-muu] ‘corn smut’

[si-kii] ‘bottle’ [tʃi-kii] ‘blister’

[si-ʝoʔo] ‘root’ [tʃi-koʔo] ‘hole’

A few fruits and vegetables start with the [si-] classifier, as we can see inexample (9).

16

(9) si-

[si-kaβa] ‘plum’

[si-kumi ] ‘onion’

[si-nana] ‘tomato’

[si-kwaa] ‘guava’

[si-ʔiði] ‘avocado’

In at least one compound, the noun that the prefix is derived from is used inits full form: [kisi βiði ] ‘fruit’, where not only are the segments the same as theword ‘animal’ [kisi ], but also the underlying tone pattern, followed by the adjective‘sweet’ [βiði ].

Lastly, (León Pasquel 1988:160, 165) notes that a few celestial beings also takethe animal classifier. Acatlán Mixtec examples can be seen in (10).

(10) si-

[si-koʝaʔntʃi] ‘rainbow’

[si-tnuu] ‘star’

[si-ntsiβi] ‘star’

[si-mjaʔa] ~ [si-mi ʔa] ‘devil’

Some nouns have two forms: a prefixed form and a non-prefixed form, suchas the examples in (11).

17

(11) Prefixed form Non-prefixed form

[si-ɲuɲu] [ɲuɲu] ‘bee’

[si-ʝoko] [ʝoko] ‘wasp’

[si-koo] [koo] ‘snake’

[si-ʝoʔo] [ʝoʔo] ‘root’

[si-koʝaʔntʃi] [koʝaʔntʃi] ‘rainbow’

According to León Pasquel (1988:139-140), some classifiers have been “fos-silized” and integrated into the couplet that was once restricted to palatal-glide-initial roots through a “process of syllabic reduction”. Since this is not a cross-dialectal study, I will account for Acatlán data only, so I do not posit that any ofthese roots were glide-initial, but rather vowel-initial, as seen in (12).

(12) Full form Reduced form

*si- uku [sjuku] ‘louse’

*si- aka [sjaka] ‘fish’

*si- aʔβi [sjaʔβi] ‘nit’

*si- umi [sjumi ] ‘owl’

*si- uku [sjuku] ‘mosquito’

*si- oʔo [sjoʔo] ‘flea’

*si- atna [sjatna] ‘chicken’

*si- aβa [sjaβa] ‘bed bugs’

18

Since no evidence was found of fossilized classifiers having any influence onthe tone pattern of such nouns, these nouns are listed along with roots henceforth.

León Pasquel (1988:146) and McKendry (2013:37) also propose the existenceof a classifier prefix for trees. McKendry (2013) writes:

Another classifier prefix in MXY is nun(H)-, which appears to be a shortenedform of jutunMH ‘tree’; again the second syllable is used to form a prefix.Many words which indicate trees and some bushes have this prefix as theirinitial morpheme.

In Acatlán Mixtec, the classifier prefix for nouns referring to trees, shrubs andcacti is [nu-]. As in South-Eastern Nochixtlán Mixtec, it appears to be the lastsyllable of [ʝutnu] ‘tree’.

(13) nu-

[nu-ðitʃi] ‘Cereus cactus’

[nu-βiʔndja] ‘prickly pear cactus’

[nu-ndua] ‘guaje tree’ (Leucaena leucocephala)

[nu-kʷiʔi] ‘banana tree’

[nu-ʔisi] ‘hopseed bush’

Interestingly, in the neighbouring town of Tonahuixtla, the same prefix is [tu‑],as in the words [tu-ndua ] ‘guaje tree’ and [tu-tʃiɡŋ] ‘guamuchil tree’ (Pithecellobiumdulce, sometimes known as ‘monkeypod’ or ‘ape’s earring’ in English), in which itcan be seen that the nasalization of the prefix has been lost. Unlike CoatzoquitengoMixtec (León Pasquel 1988:138, 146), the tree classifier is not used in AcatlánMixtec before longish objects, vehicles or machines, nor is it productive.

19

In Acatlán Mixtec there are two [te-] classifier prefixes. I consider the classi-fier for liquids [te-] and the male prefix [te-] to be homonyms because they havedifferent etymologies and semantic domains.2 The classifier prefix for nouns re-ferring to liquids, appears to be the last syllable of the noun [ndute] ‘liquid’ (inProto-Mixtec *nutjaʔ), as seen in the examples in (14).

(14) te-

[te-ðaða] ‘urine’

[te-kaka] ‘lime water’

[te-tni i ] ‘sweat’

[te-kʷii] ‘water’

[te-ʝaʔa] ‘salsa’

The other [te-] classifier prefix exists in nouns referring to men, such as thosein example (15).

2 León Pasquel (1988:134-136) claims that the [te-] classifier for liquids is derived from the Proto-Mixtec*nduteʔ ‘water’ (transcribed *nduteh by León Pasquel) and that the [te-] classifier for male persons is derivedfrom the Proto-Mixtec *tee ‘man’. McKendry (2001:31-38) provides compelling arguments for consideringthe syllable [te] in contemporary Mixtec languages to be a reflex of the Proto-Mixtec syllable *tja.

20

(15) te-

[te-ðaβi] ‘Mixtec man’

[te-ndaʔβi] ‘poor man’

[te-kua] ‘blind man’

[te-ʔitʃi] ‘skinny man’

[te-ðana] ‘dangerous man’

Most nouns that refer to men start with the classifier [te‑], except for [tʲaa]or [tia] ‘man’, which is the word from which [te‑] originates. Classifier prefixesare restricted to male individuals only. Female equivalents exist, but instead of aprefix, the full-fledged noun [ɲaɁa] ‘woman’ is usually used before a second wordto complete a noun phrase, as in example (16).

(16) Nouns that refer to men and their equivalents for womenMale noun Female compound

[te-kʷika] ‘rich man’ [ɲaʔa kʷika] ‘rich woman’ woman + rich

[te-saʔnu] ‘old man’ [ɲaʔa saʔnu] ‘old woman’ woman + elderly

[te-kʷati] ‘boys’ [ɲaʔa kʷati] ‘girls’ woman + small.pl

[te-ʔi ʔi ] ‘mute’ [ɲaʔa ʔi ʔi ] ‘mute’ woman + mute

In the case of at least one female compound, the noun ‘woman’ that precedesthe noun ‘mother’ has been shortened: ‘wife’ [ɲaðiʔi ]. Although ‘woman’ is usedas a classifier, it has not become a classifier prefix in Acatlán Mixtec, but may bein transition towards becoming one.

21

Lastly, the classifier [ntsi-] is used for a small number of animals, but the originof this classifier is unknown. Only five words with the [ntsi-] classifier were foundin the Acatlán Mixtec data, all of them listed in example (17).

(17) ntsi-

[ntsi-βoɁu] ‘coyote’

[si-ntsi-ʝoo]3 ‘quail’

[ntsi-kaʔa] ‘mountain lion’

[ntsi-Ɂi i ] ‘fly’

[te-ntsjoʔo] ‘Puebla City’ Compound formed by theliquid classifier [te‑] for rivers, and thearchaic noun [ntsjoʔo] ‘humming bird’.

2.4.2 Grammatical gender

According to Corbett (1991:1-4), “the defining criterion of gender is agree-ment” and need not be based on sex. Corbett says:

The classification frequently corresponds to a real-world distinction ofsex, [...] but often too it does not (‘gender’ derives etymologically fromLatin genus, [...] and originally meant ‘kind’ or ‘sort’).

In Mixtec, the grammatical agreement system of nouns is determined by thethird person pronoun used to refer to it, rather than by the classifier on the nounitself, so the gender of pronouns is of special interest when researching Mixtecnouns.

3 Archaic word. Speakers say there have not been sightings for a long time.

22

As noted in Corbett (1991:130-131), based on Kuiper and Pickett (1974), thirdperson Mixtec personal pronouns exist not only for human referents, but also non-human referents. Kuiper and Pickett (1974) proposed the existence of four “generaltypes of referents” based on Diuxi Mixtec personal pronouns: human, deity, animaland plant.

León Pasquel (1988:138) also came to a similar conclusion, but added anothergender (or in her terminology, “kind of referent”).4 She writes:

Third person pronouns may be analyzed as stressed short morphologicalforms derived from generic nouns that designate basically five kinds of ref-erents: human, deity, animal, tree, and liquid.

In addition to León Pasquelʼs five genders, Acatlán Mixtec also has a genericgender (alt) for referents not included in any of the other genders. In fact, altis the most common gender in Acatlán Mixtec. That gives us a total of six gen-ders, listed here with the following names along with the corresponding “kindsof referents” that León Pasquel proposed: zoa (animal), arb (tree), liq (liquid),div (deity), hum (human) and alt. A summary of the relationship between thesemantic domain of nouns, the pronouns used to refer to them, and their gendercan be seen in Table 3.

4 León Pasquelʼs “tree” corresponds to Kuiper and Pickettʼs “plant”.

23

Table 3. Grammatical genderThird person pronoun Gender Semantic domain

[si] zoa Animals, some fruit andsome round objects.

[nu] arb Trees, cacti, shrubs, somewooden things and vehicles.

[te] liq Liquids.

[ʝa] div Deities, saints, priests andsacred elements of nature.

[ne, te, ɲa, βe] hum Persons.

[a] alt Others (including plants,vegetables, body parts,elements of nature, abstractconcepts, manufactured thingsand natural formations)

The gender for persons hum is in fact used here as a cover term for severalgenders based on deictic information, such as honorific status, sex and age. Thereare several third person pronouns to refer to persons: The 3.hon pronoun [ne], the3.m.n/hon pronoun [te], the 3.f.n/hon pronoun [ɲa] and the 3.infant pronoun[βe] are used to refer to persons. Examples of persons include the following: [ɲaʔa]‘woman’, [ntsii ] ‘dead person’, [tʲaa] ‘man’, [mee ] ‘baby’, [ʝua] ‘father’, [pentʃi ] ‘or-phan’, [ðito] ‘uncle’, [ðiðo] ‘in-laws’ and [latnu] ‘youngest child’. The pronounsused to refer to a person may overlap. For instance, the noun [ntsii ] ‘dead person’may be referred to with any personal pronoun. If the person was older or is re-spected, the 3.hon pronoun [ne] is used. If the speaker wants to convey contempt,or if the deceased person was unknown to the speaker, the 3.f.n/hon pronoun [ɲa]is used in the case of a woman, and the 3.m.n/hon pronoun [te] is used in the

24

case of a man. If the deceased person was a baby, the 3.infant pronoun [βe] isused.

The gender of nouns that do not have a classifier can only be ascertained byidentifying the third person pronoun used to refer to them. The gender of a noun isnot determined by the classifier used with it, since some nouns with the classifierprefixes [si-], [tʃi-] or [nu-] have the generic gender alt. Table 4 illustrates therelationship between classifiers and gender (employing the “kinds of referents”proposed by León Pasquel for mnemonic reasons).

Table 4. Relationship between nouns and their gender, based on their classifierprefix

Note that the only persons with a classifier prefix are male persons, so the3.f.n/hon pronoun [ɲa] is not used to refer to them.

25

2.5 Consonants and vowels

2.5.1 Consonants

The symbols in Table 5 are used for the consonants in impressionistic tran-scriptions in this thesis. An underlying representation of segmental data is notprovided since this thesis concentrates on the phonological analysis of tone ratherthan the segments. The surface glottal stop is included in Table 5 even though itmay not be strictly consonantal, since Macaulay and Salmons (1995) claim that theglottal stop is in fact not an underlying consonant in Mixtec, but rather a featureof the root, as discussed in § 2.7. The sounds in parentheses are rare.

26

Table

5.Co

nson

antal

surfa

cesoun

dsBilab

ialDe

ntal

Alveolar

Post-

Alveolar

Palat

alVe

larGlottal

Plosiv

ett

nkk

ʷ(ⁿɡ

(tʲ)

Nasal

mnn

ntsntʃ

Trill

(r)

Flap

(ɾ)

Frica

tive

βð

ʝ(x)

Affric

atetʃ

Approxim

ant

l

27

I discuss below two sounds used chiefly in loan words, [p] and [mb], and Iprovide some observations about the use of [l], [ɾ] and [r].

1. [p] is used primarily in obvious loan words besides a few others. For exam-ple, [pentʃi ] ‘orphan’ could be a shortened adaptation of the Nahuatl5 wordicnopiltontli ‘fatherless’.

2. [mb] is used primarily in loan words.3. In some Acatlán Mixtec words, [l], [ɾ] and [r] vary freely, as in the word

‘parasol ant’ [luɁu ~ ɾuɁu ~ ruɁu ], the word ‘hill’ [luntsi ~ ɾuntsi ~ runtsi ],and the word ‘knee’ [loɁnɡo ~ ɾoɁnɡo ~ roɁnɡo ].

2.5.2 Vowels

The surface vowels in Acatlán Mixtec found in the data collected for this thesisare given in Table 6.

Table 6. VowelsFront Central Back

Close i u

Close-Mid e o

Open a

None of the Northern Mixteca Baja dialects has the vowel [ɨ] that many otherMixtec dialects have.

Regarding Mixtec vowels, Macaulay (1996) writes the following generaliza-tion:

5 A neighbouring Uto-Aztecan language that is also known as Aztec.

28

[A]ll of the Mixtec dialects show a strong tendency toward identity of thetwo vowels in the root, as well as restrictions on possible combinations ofvowels when they are not identical.

A statistical study of Ixpantepec Mixtec in Carroll (2015:63-67) concluded thatthe preference for identical vowel quality is highly significant, even in CVCV roots.No statistical study was carried out for the purposes of this thesis, but this prefer-ence seems to be true of Acatlán Mixtec as well, as seen in the examples in Table7.

Table 7. Roots where both vowels in the root have a shared identityCV(Ɂ)CV CV(Ɂ)V ɁV(Ɂ)CV ɁV(Ɂ)V

koðo koo Ɂiði Ɂi i ‘monkey’ ‘snake’ ‘hair’ ‘hail’

luɁlu luɁu Ɂi Ɂni ʔuʔu‘marrow’ ‘parasol ant’ ‘hot’ ‘five’

2.6 Nasalization

Mixtec morphemes are underlyingly either oral or nasalized (Marlett 1992,McKendry 2001), as discussed in this section. For instance, in the examples in(18), we see that in CVV sequences, nasality is contrastive, and both vowels areeither oral or nasalized. Note that in underlying transcriptions, a superscripted N

is placed at the end of a morpheme to indicate that it is underlyingly nasalized.Since underlying tone has not been analyzed so far in this thesis, only the phoneticrealization of tone is transcribed in this section.

29

(18) Oral morphemes Nasal morphemes

a. /ðiu/ [ðiu] ‘rump’ b. /siuN/ [si u] ‘chicken’

c. /tʃii/ [tʃii] ‘young.sg’ d. /tʃiiN/ [tʃi i ] ‘nail’

If a morpheme is nasal, nasality spreads from the right edge of the morphemeto the left. To illustrate the direction of spreading, consider bimoraic roots with aconsonant in the middle. Whenever an obstruent occupies the C slot between twovowels, either both vowels will be oral (ʔVCV or CVCV), or the first vowel will beoral and the last vowel will be nasal (ʔVCV or CVCV). See the data in (19).

(19) Oral morphemes Nasal morphemes

a. /ʔiki/ [Ɂiki] ‘ridge’ b. /ʔikiN/ [Ɂiki ] ‘squash’

c. /juku/ [ʝuku] ‘grass’ d. /jukuN/ [ʝuku] ‘furrow’

e. /ðusa/ [ðusa] ‘debris’ f. /jusaN/ [ʝusa] ‘dough’

As seen in these examples, there is contrast between oral and nasal morphemes.Thus, we can conclude that Acatlán Mixtec follows the predictions made in Marlett(1992:433): “[N]asalization inMixtec is an autosegmental morpheme-level featurewhich links to the right edge of a morpheme and spreads to adjacent sonorants.”6

The only sonorants found in the middle of nasal morphemes are [m], [ɲ] or [n],and both vowels in bimoraic morphemes are nasalized if the intervening consonantis a sonorant. On the other hand, the only sonorants found in the middle of oralmorphemes are [β], [w], [ʝ] or [l], and both vowels in bimoraic morphemes areoral if the intervening consonant is a sonorant, as seen in (20).

6 Most exceptions in Acatlán Mixtec are loanwords, other exceptions may be compounds. I have found thefollowing examples: [tʃi-mandu] ‘lizard’; [si-mentu] ‘donkey’ from the Spanish word jumento ‘donkey’; [tamali ]‘tamale’ from the Nahuatl word tamalli; [meku ] ‘dirt’ from the Mexican Spanish word meco ‘reddish brown’;[si-mutu] ‘squirrel’ from the Nahuatl word motohtli; [mi lo] ‘rabbit’ (in other Mixtec languages the cognate ‘cat’is [milu] [mitu] or [mistu], according to Marlett (1992:432)) from the Nahuatl word mistontli ‘cat’.

30

(20) Oral morphemes Nasal morphemes

[ki βi] ‘day’ [tnumi ] ‘feather’

[ðaʔβa] ‘cliff’ [ɲaʔmi ] ‘sweet potato’

[nduʝu] ‘stake’ [kuɲu] ‘meat’

[nduli] ‘honey’ [nuni ] ‘corn’

[luʔlu] ‘marrow’ [ɲaʔna] ‘mask’

In order to account for some of the distributional restrictions between nasalsonorants and oral sonorants, Marlett (1992:427) proposes that labial sonorantshave a shared underlying representation and palatal sonorants also have a sharedunderlying representation. Marlett says:

I give the proposed underlying forms written informally, using w to repre-sent the labial sonorant, n to represent the coronal sonorant, y to representthe palatal sonorant, and N to represent morpheme-level nasalization.

Likewise, I consider that the sonorants [β] (occasionally [w] in Acatlán Mix-tec) and [ʝ] are the realization of /w/ and /j/ respectively in oral environments.Furthermore, [m] and [ɲ] are the realization of /w/ and /j/ respectively in nasalenvironments.7 Marlett also argues that in Mixtec languages, [nd] is the realizationof /n/ in oral environments and that [n] is the realization of /n/ in nasal environ-ments. In Acatlán Mixtec, this is not consistently true, possibly due to historicalreasons involving palatalization and alveolarization that will not be explored here.In Acatlán, [nd], [nts] and [ntʃ] can be found in oral environments, and [n] can befound in nasal environments.

7 I use /w/ to represent the underlying labial sonorant and /j/ to represent the underlying palatal sonorant.

31

To further illustrate this distribution, consider the case of bimoraic morphemesthat begin with a sonorant, where the sonorants [β], [w], [ʝ], [l], and [ɾ] are foundin oral environments only, and the sonorants [m], [ɲ] and [n] are found in nasalmorphemes only, as seen in (21).

(21) Oral morphemes Nasal morphemes

/wiu/ [βiu] ‘corn plant’ /winuN/ [mi nu] ‘goosefoot’

/juu/ [ʝuu] ‘rock’ /juuN/ [ɲuu] ‘palm leaf’

/jaʔwi/ [ʝaʔβi] ‘market’ /juʔwaN/ [ɲuʔma] ‘smoke’

/lila/ [lila] ‘film or membrane’ /nawaN/ [nama] ‘soap’

Acatlán Mixtec is one of the varieties described by Marlett (1992:429) thathave an interesting pronunciation of /t/:

[I]n a few varieties of Mixtec, the coronal obstruent t is pronounced with anasal transition before a nasalized vowel. The obstruent nevertheless stillblocks the spread of word-level nasalization.

Just as in Ñumí Mixtec (Gittlen and Marlett 1985:177) and Atatláhuca Mixtec(Marlett 1992:429), in the Xayacatlán variety of Acatlán Mixtec, /t/ has a nasalrelease before a nasal vowel, as shown in data in (22).

32

(22) Nasal morphemes

/jutuN/ [ʝutnu] ‘tree’

/tuʔuN/ [tnuɁu] ‘lie’

/tuwiN/ [tnumi ] ‘feather’

/si-tiiN/ [si-tni i ] ‘mouse’

/sastuʔuN/ [sastnuʔu] ‘very much’

In Acatlán Mixtec, nasalization spreads across glottal features within the root,which is confirmed by the fact that /t/ is realized as [tn] even when the morphemehas a glottal feature that is realized phonetically as a glottal stop in words like[tnuɁu] ‘lie’ and [sastnuɁu] ‘very much’.8

Among Northern Mixteca Baja dialects, Xayacatlán (the dialect studied here),Tonahuixtla, Totoltepec, Tlaltempan and Chigmecatitlán follow the [tn] pattern,but Cosoltepec, Tepejillo andMicaltepec (Petlalcingo), Ixtapan and Yolotepec (Tequix-tepec), Tetaltepec (Huajuapan), Ixitlán, Huaxtepec (Chazumba) and ZapotitlánPalmas retain the conservative realization of /t/ in nasal environments.

As E. Pike and Wistrand (1974) noted, word-final nasalized vowels are option-ally followed by [ŋ], as in [si-roo ] ~ [si-rooŋ] ‘junebug’.

When following /k/, a word-final nasal high vowel may be optionally (andusually) realized as [ŋ] in Acatlán Mixtec, as in example (23).

8 There is, however, an exception to this generalization: The first syllable in the noun [βaɁna] ‘sleepiness’is not nasalized, contrary to the claim made here that the glottal feature does not block nasalization. However,the word is suspect of being a compound, not only because of the nasal spreading exception, but also becauseit has a very rare tone pattern: a word with no classifier prefix with a [low-extra high] tone pattern.

33

(23) Nasal morphemes

jukuN [ʝuku] ~ [ʝuɡŋ] ‘furrow’

tʃikuN [tʃiku] ~ [tʃiɡŋ] ‘guamuchil pod’

ikiN [Ɂiki ] ~ [Ɂiɡŋ] ‘squash’

ikiN jaʔwiN [iki ɲaʔmi ] ~ [Ɂiɡŋ ɲaʔmi ] ‘tamalayota squash’(formed by ‘squash’ andand ‘sweet potato’)

Nasalization does not spread across morphemes, as can be seen in the com-pounds in (24), where one morpheme is oral, and the other is nasalized.

(24) Compounds Morphemes

[koɲuʔu] ‘censer’ [koʔo] ‘plate’ and [ɲuʔu] ‘fire’.

[βeɲuʔu] ‘church’ [βeʔe] house and [ɲuʔu] ‘fire’.

[ɲuʔiβi] ‘world’ [ɲuu] ‘town’ and an unknown element.

[ʔi ndoko] ‘placenta’ [ʔi i ] ‘skin’ and an unknown element.

[nuʔiða] ‘loom stick’ [ʝutnu] ‘stick’ and [ʔiða] ‘backstrap loom’.

2.7 The structure of nouns

Mixtec has a minimal word constraint for nouns, requiring that nouns have atleast two moras (McKendry 2001:41, Carroll 2015:56). Monomorphemic nounsare usually bimoraic. In the few cases where trimoraic nouns are not clearly poly-morphemic, they may be roots, they may have had some morphological complex-ity historically, or they may have undergone some historical phonological changeresulting in an additional segment. In Table 8 examples of noun roots can be seen.

34

Table 8. Monomorphemic nounsμμ

[ðikʷi] ‘sap’

[ʝakʷi ] ‘armadillo’

[ʔiði] ‘hair’

[ʔitu] ‘corn field’

[βiu] ‘corn plant’

[ni i ] ‘blood’

[ʔi i ] ‘salt’

[ʔi i ] ‘skin/hide’

Besides loan words,9 possible trimoraic roots include the following nouns:

Table 9. Possible trimoraic monomorphemic nounsμμμ

[sandili] ‘ancestor’

[tʃikusa] ‘a type of cactus pear’

[sakwaa] ‘deer’

[tʃoliʔi] ‘sparrow’

[ʔantsiβi] ‘heaven’

[ʔandea] ‘hell’

9 Some trimoraic loan words include the following: [ʃikama] ‘Pachyrhizus erosus’, also known as “Mexicanturnip”, from the Náhuatl word xicámatl; [tʃakwaku] ‘chimney’, from the Mexican Spanish word chacuaco;[tʃimbana] ‘bell’, from the Spanish word campana.

35

If a noun has more than two moras, the word is usually polymorphemic. Forexample, in Table 10 we can see that these trimoraic words consist of a bimoraicroot preceded by a classifier prefix.

Table 10. Roots with an animal classifier prefix (§ 2.4.1)μ-μμ μ-μμ

[si-sita] ‘possum’ [tʃi-ɲaɁɲa] ‘lizard’

[si-ndoo] ‘tarantula’ [tʃi-ni i ] ‘iguana’

[sì-dikʷi] ‘toad’ [tʃi-mi i ] ‘bumblebee’

[si-Ɂi Ɂi ] ‘skunk’ [tʃi-ðiʔi] ‘bat’

Acatlán Mixtec noun and adjective roots follow the bimoraic structure iden-tified by Mixtecanists as “the couplet,” a term applied initially to phonology byKenneth Pike in K. Pike (1948:79).10 The couplet displays the structures seen inTable 11 (transcription at a near-surface form without word-initial post-lexicallyepenthesized glottal stops):

10 He analyzed Mixtec morphemes as being disyllabic, but they are analyzed here as bimoraic roots.

36

Table 11. CoupletsStructure Roots Gloss

VV [i i ] ‘salt’

VʔV [uʔa] ‘salty’

CVV [ni i ] ‘dry corncob’

CVʔV [ɲuɁu] ‘fire’

CVCV [tata] ‘seed’

CVʔCV [ðaɁma] ‘sheet’

VCV [iðu] ‘horse’

VʔCV [i ʔna] ‘spirit’

Most noun roots in other Mixtec languages also consist of a single couplet.Macaulay (1996:27) argues that stems longer than a couplet are polymorphemic:

The couplet is subject to further affixation and/or cliticization, leading towords of more than two syllables. However, virtually all longer forms canbe analyzed as polymorphemic —if not synchronically, then diachronically.

Furthermore, Macaulay and Salmons (1995:48) argue that the root-medialglottal stop is a feature of the root, so they propose reducing the underlying coupletprofiles for the root to CVV, CVCV, CVV, VCV and VV:

37

We propose that glottalization in Chalcatongo Mixtec (as in most of theother dialects) is not a feature of vowels, nor of syllables, but rather that itis a feature of roots, i.e., of the couplet.

Likewise, Daly and Hyman (2007) say that “there is no evidence that the ʔ ofCVʔCV couplets occupies a skeletal slot.” I will not discuss the status of the root-medial glottal stop in Acatlán Mixtec in detail, but it is clear that it does not haveany discernible influence on the tone patterns of roots, so the root-medial glottalfeature will not be included in the underlying couplet profiles of roots.

In addition to the root-medial glottal feature, an “epenthetic glottalization”has been described in Carroll (2015) for Ixpantepec Mixtec. In this respect, Carrollsays:

(T)here is also epenthetic (non-contrastive) glottalization which fills an emptyonset position root-initially and at word boundaries. (...) (T)he initial con-sonant position of the root is optionally empty, but a glottal stop is regularlyinserted in this position, both utterance internally (...).

Since these phonetic glottal stops have not been included in the formal descrip-tion of several Mixtec varieties, this may be a dialect-specific case of glottalization,but it is certainly true of Acatlán Mixtec as seen in the examples in (25).

(25) a. [ʔi i /iiNone

ʔi i ]iiN/hide

‘one hide’

b. [ʔi i /iiNnine

ʔiki]iki/bone

‘nine bones’

38

Carroll (2015) also notes: “Besides targeting bare roots (...), the glottal epenthe-sis targets roots with prefixes.” The prefixed examples in (26) illustrate this inAcatlán Mixtec.

(26) a. si- ʔii coatib. te- ʔiðu horsemanc. nu- ʔiði avocado treed. te- ʔi i broth

There is consensus among Mixtecanists around the disyllabic nature of CVCVand VCV couplets. However, some Mixtecanists consider, explicitly or implic-itly, that CVV and VV couplets are disyllabic (K. Pike 1948:79, Pankratz and E.Pike 1967:287, Gittlen and Marlett 1985:179, Dürr 1987:21-22, Macaulay andSalmons 1995:39-40), and others analyze them as monosyllabic (Carroll 2015:56-57, Castillo García 2007:58-68, McKendry 2013:67-75).

Ultimately, the couplet is essentially a bimoraic structure, and whether mono-syllabic couplets exist is secondary.

39

CHAPTER 3TONE IN THE NOUN SYSTEM

3.1 Surface tone patterns

As discussed in § 2.7, simple noun roots have the following underlying syllableprofiles: VV, CVV, CVCV and VCV. In this chapter I will present evidence thatthe two basic tone elements of Mixtec, Hi (H) and Lo (L), combine to make sixunderlying tone patterns, where the superscript (H) is a floating tone that usuallyhas a raising effect on the following word: /H/, /LH/, /HL/, /HLH/, /L/, and /LH/.The tone-bearing unit (TBU) is the mora, and most noun roots are bimoraic.1 Theburden of this chapter is to analyze the tone patterns of the noun system based onbimoraic, alienable, monomorphemic nouns.

As discussed in § 2.7, whether CVV and VV roots are disyllabic or monosyllabichas been the source of some controversy. For those who analyze the bimoraiccouplet as disyllabic, the tone-bearing unit is simultaneously the mora and thesyllable, since their definition of the syllable entails that all Mixtec syllables aremonomoraic.

On the other hand, if we assume that Mixtec syllables can have more thanone mora, whereby CVV and VV roots are monosyllabic, the locus of the TBUcannot be simultaneously the syllable and the mora (which is the case for thosewho analyze all Mixtec syllables to be monomoraic). I will discuss the TBU under

1 In the few cases where trimoraic nouns are not clearly polymorphemic, they may be roots, they may havehad some morphological complexity historically, or they may have undergone some historical phonologicalchange resulting in an additional segment. Confer § 2.7.

40

the assumption that Mixtec syllables can have more than one mora in the rest ofthis section.

One of the strongest arguments for the mora as opposed to the syllable beingthe TBU unit in Acatlán Mixtec is the fact that the mora is a better predictor ofthe number of tones in a root than the syllable is. If the syllable were the basicunit of tone patterns, disyllabic roots (CVCV and VCV) would have more pitchcombinations than monosyllabic roots (CVV and VV). Tone patterns spread overthe two moras, rather than on the syllable itself.

Consider the following examples with a [LH] or [HL] pitch pattern:

(27) [HL] tone patterns [LH] tone patternsσσ σ σσ σ

CμCμ Cμμ CμCμ Cμμ

a. [31] b. [31] c. [13] d. [13][kasi] [ʝui] [ʝaka] [kaa]

‘shadow’ ‘woven mat’ dust metal

The disyllabic root (27a) has as a [HL] tone pattern, as does the monosyllabicroot (27b); disyllabic root (27c) has a [LH] pattern, as does the monosyllabic root(27d). Disyllabic words have as many pitches as monosyllabic roots, and the com-mon denominator between both syllable profiles is the bimoraic couplet, not thenumber of syllables.

For reasons discussed in § 2.7, CVʔCV couplets are included in the CVCV pat-tern, and CVʔV couplets are included with the CVV pattern in this chapter. Suchcouplets also follow the bimoraic tone patterns seen in (27), as seen in example(28).

41

(28) [HL] tone patterns [LH] tone patternsσσ σ σσ σ

CμCμ Cμμ CμCμ Cμμ

a. [31] b. [31] c. [13] d. [13][ɲaɁna] [ɲuɁu] [ʝaɁβi] [ndoɁo]‘mask’ ‘earth’ ‘market’ ‘adobe brick’

The [HL] tone pattern of disyllabic noun root (28a) has the same number oflevel tones as monosyllabic root (28b), and likewise, the disyllabic root (28c) hasthe same tone pattern [LH] as monosyllabic root (28d). As we have seen, disyllabicsimple roots do not have a greater number of tone patterns thanmonosyllabic roots.

When analyzing tone languages, it is important to control for all factors, in-cluding word class. While many nouns in Acatlán Mixtec have a classifier prefix,none of the prefixes consist exclusively of a floating tone, such as those found insome African languages. In sections § 3.2 to § 3.6 the examples presented all comefrom the alt class, and they do not take a prefix.

Simple nouns in isolation have one of the following four sets of surface tonepatterns: [HH], [LH], [HL] and [LL], as seen in Table 12.

42

Table 12. Surface patterns of simple noun roots.CVCV CVV VCV VV

[HH] [33] [33] [33] [33][ʝuʝu] [ʝoo] [Ɂiða] [Ɂii]‘dewdrop’ ‘pitcher’ ‘waistloom’ ‘hail’

[LH] [13] [13][ðusa] [ðia]‘gum’ ‘skirt’

[HL] [31] [31] [31] [31][ʝoðo] [ʝui] [Ɂiki] [Ɂii]‘plain’ ‘woven mat’ ‘bone’ ‘hide’

[LL] [11] [11] [11][ʝuku] [kwia] [Ɂii]‘grass/plant’ ‘year’ ‘salt’

Simple nouns in isolation have the surface pitch patterns [LL], [HH], [LH] and[HL]. However, the surface tone patterns cannot be taken to be the same as theunderlying pattern since there are processes that modify the underlying patterns.In particular, all nouns with the surface pattern [LL] do not behave alike in allcontexts. Similarly, all nouns with the surface pattern [HH] do not behave alikein all contexts. The [LL] surface pattern is actually the realization of two separateunderlying patterns; likewise, the [HH] surface pattern is the realization of twoseparate underlying patterns.

3.2 Hi-tone spreading across word boundaries

In Acatlán Mixtec, Hi-tone spreading (HTS) is non-iterative (it targets a singleTBU) and it applies rightward to the adjacent Lo-toned TBU, as long as the twofollowing TBUs are Lo, or if the target Lo-toned TBU is utterance-final.

43

For example, whenever the Hi-toned quantifier ‘one’ (pronounced [ʔi i ] in iso-lation) precedes a Lo-toned noun, such as ‘plant’ (pronounced [ʝuku] in isolation),the Hi tone spreads to the first TBU of the following word, as seen in example (29).

(29) /H L/ → [33  31]ʔii ʝuku [Ɂii ʝuku]‘one’ ‘plant’ ‘one plant’

HTS across word boundaries was found in q-n and n-a constructions, but somephrasal restrictions to HTS may exist, as will be discussed in § 3.7. HTS from anoun to a following word will be illustrated in example (30) with the Hi-tonednoun ‘house’ (pronounced [βeʔe] in isolation) followed by the adjective ‘spicy’(pronounced [satu] in isolation).

(30) /H L/ → [33   31]βeʔe satu [βeʔe satu]‘house’ ‘spicy’ ‘spicy house’

HTS across word boundaries can also be observed in the case of simple nounswith a [LH] tone pattern, such as [kiði ] ‘pot’.

(31) /LH L/ → [1 3   31]kiði satu [kiði satu]‘pot’ ‘spicy’ ‘spicy house’

However, HTS only applies across word boundaries provided that the two fol-lowing TBUs are Lo. I will refer to this restriction to HTS as the “two-slot condition”for reasons that will be apparent in § 3.4. In example (32), the Hi-toned quantifier‘one’ does not spread its Hi tone to the word ‘gum’ (pronounced [ðusa] in isola-tion). Hi tones target Lo-toned TBUs, but HTS is not applied because the conditionthat the two following TBUs be Lo is not fulfilled.

44

(32) /H LH/ → [33  13]ʔii ðusa [ʔii ðusa]‘one’ ‘gum’ ‘one gum’

We can also see the HTS-blocking effect of the two-slot condition in at leastone n-n compound [βeʔe kaa] ‘jail’, which is formed by the Hi-toned noun ‘house’(pronounced [βeʔe] in isolation) followed by the noun ‘metal’ (pronounced [kaa]in isolation), as seen in example (33).

(33) /H LH/ → [33   13]βeʔe kaa [βeʔe kaa]‘house’ ‘metal’ ‘jail’

The two-slot condition is a condition pervasive within Acatlán Mixtec nounphrases that requires the two following TBUs to be underlyingly toneless for tone-spreading to apply. This condition sets forth the context in which HTS can apply,thereby limiting the number of cases where it might otherwise have been triggered.

This non-application of the Hi-tone spreading rule (the two-slot condition) isattributable to the Obligatory Contour Principle (OCP) rule-blocking constraintdescribed by McCarthy (1986:220) as antigemination:2

The evidence presented up to this point has argued for only one particularinstantiation of the OCP: as a constraint on the representation of unana-lyzable morphemes in the lexicon. We now turn to some cases in whichthe OCP is enforced throughout the derivation, not to fuse sequences ofidentical elements into a single unit, as is sometimes thought, but rather toprevent the creation of such sequences.

Likewise, in Acatlán Mixtec the two-slot constraint prevents two consecutiveH tones from being contiguous.

2 The antigemination constraint was originally proposed for segmental phonology, but it applies equallyto autosegmental phonology.

45

3.3 Hi-tone spreading within the root

Not all simple nouns that surface with a [HH] tone pattern behave alike in allcontexts. Consider the simple nouns in isolation in example (34).

(34) Hi-pitched simple nouns in isolationH1 [3 3]

[ʝuʝu] ‘dewdrop’

H2 [3 3][kuɲu] ‘meat’

Even though both words are simple nouns, with the same couplet structure,from the same class alt, and with the same surface pitch in isolation, they do nothave the same underlying tone pattern.

When these nouns precede a Lo-toned adjective like ‘spicy’ (pronounced [satu]in isolation), only one of these words spreads its Hi tone onto [satu] ‘spicy’, as seenin (35).

(35) a. H1 + L [33   31][ʝuʝu satu] ‘spicy dewdrop’

b. H2 + L [33   11][kuɲu satu] ‘spicy meat’

The simple noun with the tone pattern H1 spreads its Hi tone onto a Lo-tonedadjective, as may be seen in (35a), but the simple noun with the tone pattern H2

does not. I assume that the tone pattern H1 is indeed /H/ underlyingly, so HTSapplies across the word boundary. However, since H2 nouns like ‘meat’ do notspread their Hi tone to the following word, their underlying tone must be differentfrom /H/.

46

When these underlying patterns are placed before a Hi-toned adjective like‘beautiful’ (pronounced [βiko] in isolation), we see a difference in the resultingsurface forms in example (36).

(36) H1 + H [3 3   33][ʝuʝu βiko] ‘beautiful dewdrop’

H2 + H [3 1   33][kuɲu βiko] ‘beautiful meat’

The Hi-toned noun [ʝuʝu] ‘dewdrop’ maintains its Hi tone when preceding theHi-toned adjective ‘beautiful’. However, in the case of H2 nouns like [kuɲu] ‘meat’,the first TBU of the couplet remains high-pitched, while the second TBU surfaceswith a low pitch before ‘beautiful’. I posit the underlying pattern of H2 to be /HL/:The underlyingly Lo TBU surfaces as [L] when the two-slot condition for HTS isnot met (because the Lo-toned target is immediately followed by a Hi tone), butthe underlyingly Lo TBU surfaces as [H] when the two-slot condition for HTS ismet (because the underlying Hi tone of the word is adjacent to two underlyinglyLo-toned TBUs, or the underlyingly Lo target TBU is utterance-final).

Looking back at example (35), we see that H2 nouns do not spread their Hitone across the word boundary, precisely because HTS has already taken place,non-iteratively, within the root [kuɲu] ‘meat’.

The reason why /HL/ simple nouns surface as [HH] in isolation remains un-accounted for. To explain this, the two-slot condition must be expanded:

(37) Two-slot condition (final formulation): Hi tone spreads rightward one TBU,provided that the following Lo-toned mora is followed either by anotherLo-toned mora or utterance boundary.

47

This condition was initially identified in Snider (1988:103-104) for AcatlánMixtec, but its pervasiveness beyond root-internal HTS in /HL/ roots in AcatlánMixtec is described for the first time here.

3.4 Surface versus underlying HL

If it is true that some of the simple nouns that surfaced as [HH] in isolation(referred to as H2) have an underlying /HL/ tone pattern, as is proposed in § 3.3,we cannot possibly claim at the same time that simple nouns that surface as [HL]in isolation have the same underlying tone pattern /HL/.

Consider the word [kuɲu] ‘meat’ and the word [ʝoðo] ‘plain’ in example (38).In § 3.3 I proposed that ‘meat’, which surfaces as [HH] in isolation, is underlyingly/HL/. If that is true, the word ‘plain’, even though it surfaces as [HL], cannot havethe same underlying pattern.

(38) a. HL1 [3 3][kuɲu] ‘meat’

b. HL2 [3 1][ʝoðo] ‘plain’

We could expect HTS to apply from the first to the second TBU of HL2, andyet it does not. Isolation is not a neutral environment, as we have seen, so let usnow consider example (39), where the same nouns are placed before the Lo-tonedadjective ‘spicy’.

48

(39) a. HL1 + L [3 3   11][kuɲu satu] ‘spicy meat’

b. HL2 + L [3 1   31][ʝoðo satu] ‘spicy plain’

The adjective [satu] ‘spicy’ surfaces as low in isolation and after a HL1 noun,and yet its first TBU has a high pitch after the HL2 noun [ʝoðo] ‘plain’. The differ-ence in behaviour must be due to a difference in the tone pattern of the first wordand not in the tone pattern of the following adjective.

I posit that the Hi tone in the adjective [satu] ‘spicy’ is in fact sponsored by theHL2 noun, which has an additional Hi tone after the [HL] sequence. The fact thatthe Hi tone surfaces on the following TBU in cases like (39b) is explained by thefact that this tone is floating, and is not underlyingly linked to a TBU of the root.

When roots with a HL2 tone pattern are in isolation, one would initially thinkthat the two-slot condition would not block HTS from applying to them, as seenin the case of [ʝoðo] ‘plain’ in example (38). The underlyingly Lo TBU seems to beutterance-final and is immediately preceded by a Hi tone, which normally resultsin tone spreading, but in the case of HL2 roots it does not.

The proposal that the underlying tone pattern of HL2 roots includes a floatingHi tone would explain why the Hi tone from the first TBU does not spread to thesecond TBU of the root. Roots with an underlying /HLH/ tone pattern (where (H)is a floating tone) find that HTS is immediately blocked because of the floating Hiat the end.

This is why the term “two-slot condition” is used here rather than “two-moracondition”: It does not technically require the two TBUs to the right of a Hi toneto be underlyingly Lo for HTS to apply, but rather the two slots to the right of

49

a Hi tone to be unspecified for Hi tone for HTS to apply, and floating tones areunderlyingly linked to a slot.3

We have not yet seen the effect of floating Hi tones on Hi-toned adjectives.Before looking at them, consider first the effect that a Hi tone has on an adjective,as seen in example (40).

(40) /H H/ → [3 3   33]βeʔe βiko [βeʔe βiko]‘house’ ‘beautiful’ ‘beautiful house’

/LH/ H/ → [1 3   33]kiði βiko [kiði βiko]‘pot’ ‘beautiful’ ‘beautiful pot’

/HL H/ → [3 1   33]kuɲu βiko [kuɲu βiko]‘meat’ ‘beautiful’ ‘beautiful meat’

The Hi tone from the noun and the Hi tone of the adjective both remain atthe same pitch level. In the case of /HL/ nouns, the Hi-toned adjective remainsHi. However, if we place a noun with an underlying /HLH/ tone pattern before aHi-toned adjective, the adjective surfaces with a Lo pitch, as seen in (41).

(41) /HLH H/ → [3 1   11]ʝoðo βiko [ʝoðo βiko]‘plain’ ‘beautiful’ ‘beautiful plain’

The adjective [βiko] ‘beautiful’ is realized as [HH] in isolation and after nounswith an underlying /H/ or /HL/ tone pattern. However, the same word is realizedwith a surface [LL] pattern in example (40). It follows that the difference in be-haviour must be due to a difference in the tone pattern of the first word and notin the tone pattern of the following adjective.

3 My “slot” is equivalent to the Tonal Root Node of Snider’s Register Tier Theory (Snider 1999). Under-specification will be discussed in § 3.9.

50

The low pitch of [ʝoðo] ‘plain’ in (40) does not confirm the proposition derivedfrom example (39), where I had claimed that [ʝoðo] sponsored a floating Hi tonethat was realized on the following TBU. On the contrary, we see that the otherwiseHi-toned adjective [βiko] ‘beautiful’ is realized with a low pitch instead.

Though the floating tone sponsored by [ʝoðo] ‘plain’ depresses the followingword (in fact, both TBUs of [βiko] ‘beautiful’), I still consider the floating tone to beunderlyingly Hi. The main reason is that even though the floating tones sponsoredby a noun trigger Hi-tone lowering onto a Hi-toned adjective, the floating tonehas a raising effect in other contexts, such as q-n and n-n constructions (as seenin forthcoming examples in § 3.6), as well as cases of Hi-toned nouns precedinga Lo-toned adjective, as was seen in example (39). A more theory-specific reasonresides in the claim put forward in § 3.9 that Lo tones are underlyingly unspecified,so a floating tone could only be underlyingly Hi for it to exist.

It is not clear why Hi-toned adjectives are lowered in this syntactic environ-ment. I consider this to be an area for further research.

3.5 Surface versus underlying L

As we had seen in § 3.1, one the four tone patterns of simple nouns in isolationis surface [LL]. In fact, not all nouns with a surface [LL] behave alike. Considerthe examples in (42).

(42) L1 [1 1][ʝuku] ‘grass’

L2 [1 1][ʝoðo] ‘grinding stone’

While their realizations are the same in isolation, they behave differently whenplaced before Lo-toned adjectives, as seen in example (43).

51

(43) L1 + L [1 1   11][ʝuku satu] ‘spicy grass’

L2 + L [1 1   31][ʝoðo satu] ‘spicy grinding stone’

In the case of the L1 noun [ʝuku] ‘grass’, the Lo-toned adjective [satu] ‘spicy’remains Lo. However, in the case of L2, the adjective ‘spicy’ surfaces with a [HL]surface pattern. The Hi tone can only be attributed to the previous word [ʝoðo]‘grinding stone’, so I posit that a floating Hi tone sponsored by ‘grinding stone’ isresponsible for the Hi tone on the first TBU of ‘spicy’.

Thus, the underlying tone pattern of L1 nouns is /L/ and the underlying patternof L2 nouns is /LH/, where both TBUs in the couplet are Lo, and the floating toneis realized on the following word.

Note that the effect of the floating Hi tone sponsored by a /LH/ noun onto aLo-toned adjective is identical to the effect of the floating Hi tone sponsored by a/HLH/ noun onto a Lo-toned adjective, as seen in (39).

Furthermore, the effect of the floating Hi tone sponsored by a /LH/ noun ontoa Hi-toned adjective is identical to the effect of the floating Hi tone sponsoredby a /HLH/ noun onto a Hi-toned adjective (cf. example (41)), as can be seen inexample (44).

52

(44) /L H/ → [1 1   33]ʝuku βiko [ʝuku βiko]‘grass’ ‘beautiful’ ‘beautiful grass’

/LH H/ → [1 1   11]ʝoðo βiko [ʝoðo βiko]‘grinding stone’ ‘beautiful’ ‘beautiful grinding stone’

In the case of /L/ nouns like [ʝuku] ‘grass’, the following Hi-toned adjective[βiko] ‘beautiful’ remains Hi. But since /HLH/ nouns like [ʝoðo] ‘grinding stone’have a floating tone, the otherwise Hi-toned adjective [βiko] surfaces with a lowpitch due to the same lowering process observed in (41), which was triggered bythe floating Hi tone.

Hi-tone lowering depresses the pitch of Hi-toned adjectives. Consider the ef-fects of the /L/ noun ‘chocolate’ and the /LH/ noun ‘tortilla’ on the following adjec-tive ‘hot’ (pronounced [ʔi ʔni ] in isolation) and the degree word ‘very’ (pronounced[titi ] in isolation) in example (45).

(45) /L H H/ → [1  1  3  3  3 3]ðiʔβa ʔiʔni titi [ðiʔβa ʔiʔni titi]

‘chocolate’ ‘hot’ deg ‘very hot cocoa’

/LH H H/ → [1 1  1  1   11]ðita ʔiʔni titi [ ðita ʔiʔni titi]

‘tortilla’ ‘hot’ deg ‘very hot tortilla’

The noun [ðiʔβa] ‘chocolate’ has no floating tone, so the following words main-tain their usual pitch. The word [ðita] ‘tortilla’ sponsors a floating tone, whichdepresses the pitch of the following Hi-toned TBUs down to the same level as ‘tor-tilla’.

53

3.6 Underlying tone patterns

We have now found all of the underlying patterns that must be recognized inAcatlán Mixtec noun roots: /H/, /LH/, /HL/, /HLH/, /L/ and /LH/.

These patterns account for all noun roots. Any noun roots that deviate fromthese patterns are derivations, compounds, loan words, or of unknown morpho-logical complexity.

Table 13 shows the underlying tone patterns of Acatlán Mixtec on the firstcolumn, together with examples of nouns in isolation and their surface tone in thefollowing columns. They are all simple nouns taken from the alt class.

54

Table 13. Basic tone patterns in isolationUF CVCV CVV VCV* VV**As discussed in § 2.7, VCV and VV roots have an epenthetic glottal stop root-initially.

/H/ [33] [33] [33] [33][ʝuʝu] [tʃuu] [Ɂiða] [Ɂii]‘dewdrop’ ‘work’ ‘waist loom’ ‘hail’

/LH/ [13] [13][ðusa] [ðia]‘gum’ ‘skirt’

/HL/ [33] [33] [33][nduði] [ʝoo] [Ɂitu]‘honey’ ‘pitcher’ ‘corn field’

/HLH/ [31] [31] [31] [31][ʝoðo] [ʝui] [Ɂiki] [Ɂii]‘plain’ ‘woven mat’ ‘bone’ ‘skin’

/L/ [11] [11][ʝuku] [kwia]‘grass’ ‘year’

/LH/ [11] [11] [11] [11][katʃi] [ʝoo] [Ɂiki] [Ɂii]‘cotton’ ‘month’ ‘ridge’ ‘salt’

Hi-toned quantifiers such as [ʔi i ] ‘one’, when placed before a noun, behave thesame as Hi-toned nouns in in n-a constructions (seen in § 3.2), as seen in Table14.4

4 Other /H/ quantifiers include [ðaβa] ‘some’ and [diβi ] ‘that very one’.

55

Table 14. Behaviour of a Hi-toned quantifier before a nouna. /H H/ → [33 33]

Ɂii Ɂii [Ɂii Ɂii]‘one’ ‘hail’ ‘one pellet of hail’

b. /H LH/ → [33 1 3]Ɂii kiði [Ɂii kiði]‘one’ ‘pot’ ‘one pot’

b. /H HL/ → [33  3 3]Ɂii kuɲu [Ɂii kuɲu]‘one’ ‘meat’ ‘one meat’

d. /H HLH/ → [33  31]Ɂii Ɂii [Ɂii Ɂii]‘one’ ‘skin’ ‘one skin’

e. /H L/ → [3 3  31]Ɂii ʝuku [Ɂii ʝuku]‘one’ ‘plant’ ‘one plant’

f. /H LH/ → [33 31]Ɂii Ɂii [Ɂii Ɂii]‘one’ ‘salt’ ‘one salt’

Quantifiers that surface with a [LL] tone pattern in isolation do not all behavealike. We will first look in Table 15 at how quantifiers with an underlying /L/pattern behave when placed before a noun.5

5 /L/ quantifiers include [ʔui ] ‘two’, [ʔuni ] ‘three’, [ʔuʔu] ‘five’, [ʔi ɲu] ‘six’, [ʔusa] ‘seven’, [ʔuna] ‘eight’,[ʔi i ] ‘nine’, [ʔuʃi ] ‘ten’, [saʔu] ‘fifteen’ and [ʔoko] ‘twenty’.

56

Table 15. Behaviour of a Lo-toned quantifier before a nouna. /L H/ → [11 33]

Ɂii Ɂii [Ɂii Ɂii]‘nine’ ‘hail’ ‘nine pellets of hail’

b. /L LH/ → [11 13]Ɂii kiði [Ɂii kiði]

‘nine’ ‘pot’ ‘nine pots’

c. /L HL/ → [11 33]Ɂii kuɲu [Ɂii kuɲu]

‘nine’ ‘meat’ ‘nine meats’

d. /L HLH/ → [11 31]Ɂii Ɂii [Ɂii Ɂii]

‘nine’ ‘skin’ ‘nine skins’

e. /L L/ → [11 11]Ɂii ʝuku [Ɂii ʝuku]

‘nine’ ‘plant’ ‘nine plants’

f. /L LH/ → [11 11]Ɂii Ɂii [Ɂii Ɂii]

‘nine’ ‘salt’ ‘nine salts’

Quantifiers with an underlying /L/ tone pattern have the same effect on nounsas underlyingly /L/ nouns do on adjectives: none.

However, another set of quantifiers, such as [kumi ] ‘four’ and [kwaʔa] ‘many’,have a different effect on the following noun than any pattern seen so far. Theirtone pattern will be referred to as L3 for the time being.6 These quantifiers have araising effect on the following TBU, as can be seen in the examples in (47).

6 Ultimately, I claim that it has the same underlying tone pattern as /LH/ nouns, but the explanation isforthcoming in the rest of this section.

57

(46) Floating tone association from a quantifier to a noun

a. /L3 LH/ → [1 1   6 3]kumi kiði [kumi kiði]‘four’ ‘pot’ ‘four pots’

b. /L3 L/ → [1 1   6 1]kumi ʝuku [kumi ʝuku]‘four’ ‘plant’ ‘four plants’

c. /L3 LH/ → [1 1   6 1]kumi ʝoðo [kumi ʝoðo]‘four’ ‘grinding stone’ ‘four grinding stones’

As seen in examples in (46), the first TBU of nouns with an underlying pattern/L/, /LH/ and /LH/ have an extra-high pitch, i.e. a pitch higher than a Hi pitch,and then the second TBU of the noun is realized with the same pitch that it wouldhave been pronounced had there been no raising on the preceding TBU. Note thatthe first TBU of all three nouns is underlyingly Lo.

When quantifiers with a L3 tone pattern precede a /H/ or /HL/ simple noun,it also raises its pitch to an extra-high, as seen in (47).

58

(47) Floating tone association from a quantifier to a noun

/L3 H/ → [1 1   6 6]kumi ʝuʝu [kumi ʝuʝu]‘four’ ‘drop’ ‘four dewdrops’

/L3 HL/ → [1 1   6 6]kumi kuɲu [kumi kuɲu]‘four’ ‘meat’ ‘four meats’

Note that even both Hi-pitched TBUs are raised, and not just the first TBU. Inthe case of [ʝuʝu] ‘dewdrop’ both TBUs are raised because both TBUs share a singleunderlying H tone. In the case of [kuɲu] ‘meat’, its underlying H tone was raised,and HTS still takes place to the second TBU of [kuɲu] at the same raised level.

Lastly, consider the effect of L3 on /HLH/ simple nouns, as seen in (48).

(48) Floating tone association from a quantifier to a noun

/L3 HLH/ → [1 1   6 3]kumi ʝoðo [kumi ʝoðo]‘four’ ‘plain’ ‘four plains’

The Hi tone of the noun ‘plain’ is raised to an extra-high pitch, but also thesecond TBU is raised from its usual low pitch to a Hi pitch instead. The raisingof the second TBU is in contrast with the Lo tones on second TBUs seen in (46),which remained low-pitched. Note that the resulting surface tone pattern in (48)is the same as the surface pattern of (46a), even though the target nouns each havedifferent underlying tone patterns (‘plain’ has an underlying /HLH/ tone pattern,and ‘pot’ has an underlying /LH/ tone pattern).

Since the floating tone sponsored by ‘four’ raises the pitch of the followingTBU in (46)-(48) to a level higher than the floating Hi seen in the case of nouns in

59

§ 3.4 and § 3.5, we could assume that the floating [H] tone of /LH/ nouns and thefloating extra-high tone [E] of L3 quantifiers are underlyingly different. Besides,the effects of both floating tones are patently different: while the floating tonesof /HLH/ and /LH/ nouns trigger Hi-tone lowering onto Hi-toned adjectives, L3quantifiers, on the other hand, trigger a raising effect onto /H/ and /HL/ nouns toan extra-high pitch [EE].

Observing the behaviour of the floating Hi tones of nouns in environmentsother than n-a constructions is useful in this respect. Although full sentences arenot the focus of this thesis, the case of Verb-Subject-Object constructions are ofspecial interest to the understanding of the floating tones of nouns. Consider Table16, where the /LH/ simple noun ‘tortilla’ is followed by a simple noun from eachof the underlying forms seen in this chapter.

As can be seen in Table 16, the /LH/ simple noun ‘tortilla’ triggers the sameraising process seen in the case of L3 quantifiers. Whether /LH/ nouns trigger theraising or lowering processes described in § 3.4 and § 3.5, or if they trigger theraising processes seen for L3 quantifiers seen in this section is a matter of phrasalenvironment, and does not entail that the floating tone is underlyingly different.Therefore, I also consider L3 quantifiers to have an underlying /LH/ tone pattern.

60

Table

16.F

loatin

gton

easso

ciatio

nfrom

anou

nontoa

nother

noun

a./L

LLH

H/→

[1  1 1

  1 1  6 6

]ni

ʃika

ðita

ʝuʝu

[niʃ

ikaði

taʝuʝ

u]co

mpl

‘askf

or’

‘tortilla

’‘de

w’‘Th

etortilla

askedf

ordew.

b./L

LLH

LH/

→[1 

 1 1  1 1

  6 3]

ni

ʃika

ðita

kiði

[niʃ

ikaði

takið

i]co

mpl

‘askf

or’

‘tortilla

’‘po

t’‘Th

etortilla

askedf

orpo

t(s).’

c./L

LLH

HL/

→[1 

 1 1  1 1

  66]

ni

ʃika

ðita

kuɲu

[niʃ

ikaði

taku

ɲu]

comp

l‘as

kfor’

‘tortilla

’‘m

eat’

‘Thet

ortilla

askedf

orme

at.’

d./L

LLH

HLH /

→[1 

 1 1  1 1

  6 3]

ni

ʃika

ðita

ntsiði

[niʃ

ikaði

tants

iði]

comp

l‘as

kfor’

‘tortilla

’‘co

rncob’

‘Thet

ortilla

askedf

orcorncob(s).’

e./L

LLH

L/→

[1  1 1

  1 1  6 1

]ni

ʃika

ðita

ʃitn u

[niʃ

ikaði

taʃit

n u]

comp

l‘as

kfor’

‘tortilla

’‘ov

en’

‘Thet

ortilla

askedf

orov

en(s)

.’

f./L

LLH

LH/

→[1 

 1 1  1 1

  6 1]

ni

ʃika

ðita

ʝoðo

[niʃ

ikaði

taʝoð

o]co

mpl

‘askf

or’

‘tortilla

’‘gr

inding

stone’

‘Thet

ortilla

askedf

orgrind

ingsto

ne(s)

.’

61

The raising effect to an extra-high pitch triggered by floating Hi tones wasinitially identified in E. Pike and Wistrand (1974) as “step-up terrace tone” and itconstitutes upstep according to Snider (1999:105-115). A floating Hi tone raisesthe level of all subsequent Hi tones to an extra-high level. For example, in (49),the Hi-toned words [βeʔe] ‘house’, [ʔi ʔni ] ‘hot’ and [titi ] ‘very’ can be seen at theirusual level after the underlyingly /L/ quantifier [ʔi i ] ‘nine’, but they are all raisedto a higher register than usual when they are preceded by the underlyingly /LH/quantifier [kumi ] ‘four’. Based on data in E. Pike and Wistrand (1974), there isreason to believe that this process is iterative, but the precise combinations neededto observe it will not be explored here.

(49)/L H H H/ → [11  3 3  3  3  3 3]ʔii βeʔe ʔiʔni titi [ ʔii βeʔe ʔiʔni titi]

‘nine’ ‘house’ ‘hot’ deg ‘nine very hot houses’

/LH H H H/ → [1  1   6 6   6 6  6 6]kumi βeʔe ʔiʔni titi [kumi βeʔe ʔiʔni titi]‘four’ ‘house’ ‘hot’ deg ‘four very hot houses’

If the raising process to an extra-high level triggered by a floating Hi tone,such as the one sponsored by [kumi ] ‘four’, is indeed upstep, we could expect thesecond TBU of the Lo-toned word [ʝuku] ‘plant’ seen in (50) to be realized at ahigher register as well, perhaps at the usual level of a Hi tone. However, this isnot the case, and the pitch of the last TBU of [ʝuku] is realized at the same level as[kumi ]. Given that Hi tones are clearly being upstepped in this language, the factthat this Lo tone is not upstepped might possibly be explained by positing that Lotones are underlyingly unspecified for tone, so there is no underlying Lo tone to

62

be raised to a higher register in the first place. This also is an area where furtherresearch is needed. Underspecification will not be discussed further until § 3.9.

(50)/LH L/ → [1 1   6 1]kumi ʝuku [kumi ʝuku]‘four’ ‘plant’ ‘four plants’

3.7 Subject-object tonal interaction in transitive sentences

The data collection for this thesis concentrated on eliciting mainly noun phrasedata, but some sentences consisting of a Lo-toned transitive verb, its subject andits object were also elicited in order to better understand the tonal behaviour ofnouns. Mixtec is a VSO language, so the subject np is immediately juxtaposed withthe direct object np, which allows us to observe how the tone of a subject nouninteracts with the tone of an object noun.

The Hi tone of /LH/ subject nouns follows the Hi-tone spreading behaviourseen in § 3.2, the two-slot condition seen in § 3.3, and it merges with the Hi toneof a following noun (just as the n-a constructions see in example (40) and the q-n constructions seen in Table 14). Table 17 consists of examples that display theinteraction between a subject noun having the /LH/ pattern and object nouns witheach of the six underlying patterns proposed in § 3.6.

63

Table

17.B

ehaviou

rof/

LH/s

ubjec

tsa.

/LL

LHH/

→[1  1 1  1 3  3 3]

ni

ʃika

kiði

ʝuʝu

[niʃ

ikaki

ðiʝuʝ

u]co

mpl

‘askf

or’

‘pot’

‘dew’

‘Thep

otaskedf

ordew.

b./L

LLH

LH/

→[1  1 1  1 3  1 3]

ni

ʃika

kiði

ðusa

[niʃ

ikaki

ðiðu

sa]

comp

l‘as

kfor’

‘pot’

‘gum’

‘Thep

otaskedf

orgu

m.’

c./L

LLH

HL/

→[1  1 1  1 3  3 3]

ni

ʃika

kiði

kuɲu

[niʃ

ikaki

ðiku

ɲu]

comp

l‘as

kfor’

‘pot’

‘meat’

‘Thep

otaskedf

orme

at.’

d./L

LLH

HLH /

→[1  1 1  1 3  3 1]

ni

ʃika

kiði

ntsiði

[niʃ

ikaki

ðints

iði]

comp

l‘as

kfor’

‘pot’

‘corncob’

‘Thep

otaskedf

orcorncob(s).’

e./L

LLH

L/→

[1  1 1  1 3  3 1]

ni

ʃika

kiði

ʃitn u

[niʃ

ikaki

ðiʃit

n u]

comp

l‘as

kfor’

‘pot’

‘oven’

‘Thep

otaskedf

orov

en(s)

.’

f./L

LLH

LH/

→[1  1 1  1 3  3 1]

ni

ʃika

kiði

ʝoðo

[niʃ

ikaki

ðiʝoð

o]co

mpl

‘askf

or’

‘pot’

‘grind

ingsto

ne’

‘Thep

otaskedf

orgrind

ingsto

ne(s)

.’

64

When nouns that have been analyzed as having an underlying /H/ patternin § 3.2 and § 3.3 are the subject of a transitive sentence, the expected Hi-tonespreading onto the following noun does not take place, contrary to what could beexpected. Instead, the Hi-toned noun triggers the same raising to an extra-highlevel described for /LH/ quantifiers and nouns in § 3.6, as seen in Table 18.

Even though they trigger raising to an extra-high level onto the next noun inVerb-Subject-Object constructions, there is reason to believe that the underlyingtone pattern of nouns like [ʝuʝu] ‘dewdrop’ and [βeʔe] ‘house’ is indeed /H/. Inthe first place, Hi-toned nouns behave the same way as Hi-toned quantifiers whenHi-toned nouns are placed before an adjective. Besides, cross-dialectically, theword [βeʔe] ‘house’ has been consistently identified as having the same tonal be-haviour as another Hi-toned word, the quantifier [ʔi i ] ‘one’ (for example, Alexan-der (1980:14, 239) classifies them as having a Mid-Mid pattern (which correspondsto Hi in the present analysis) and the same tonal behaviour). Both of these wordshave become the canonical Mid-toned words in Mixtecanist literature, not onlybased on their surface realization, but on their tonal behaviour (Dürr 1987:41,McKendry 2013:124, Stark et al. 2013:86-87, Carroll 2015:172).

Further investigation should shed light on whether Hi-toned nouns and Hi-toned quantifiers have the same underlying tone or not in Acatlán Mixtec. Thereason why Hi-toned subject nouns trigger extra-high raising onto object nouns isunclear at the present time.

65

Table

18.B

ehaviou

rof/

H/subje

cts

a./L

LH

H/→

[1  1 1

  3 3  6 6

]ni

ʃika

ʝuʝu

βeʔe

[niʃ

ikaʝu

ʝuβe

ʔe]

comp

l‘as

kfor’

‘dewd

rop’

‘house’

‘Thed

ewdrop

askedf

orho

use(s

).’

b./L

LH

LH/

→[1 

 1 1  3 3

  6 3]

ni

ʃika

ʝuʝu

ðusa

[niʃ

ikaʝu

ʝuðu

sa]

comp

l‘as

kfor’

‘dewd

rop’

‘gum’

‘Thed

ewdrop

askedf

orgu

m.’

c./L

LH

HL/

→[1 

 1 1  3 3

  6 6]

ni

ʃika

ʝuʝu

kuɲu

[niʃ

ikaʝu

ʝuku

ɲu]

comp

l‘as

kfor’

‘dewd

rop’

‘meat’

‘Thed

ewdrop

askedf

orme

at.’

d./L

LH

HLH /

→[1 

 1 1  3 3

  6 3]

ni

ʃika

ʝuʝu

ntsiði

[niʃ

ikaʝu

ʝunts

iði]

comp

l‘as

kfor’

‘dewd

rop’

‘corncob’

‘Thed

ewdrop

askedf

orcorncob(s).’

e./L

LH

L/→

[1  1 1

  3 3  6 1

]ni

ʃika

ʝuʝu

ʃitn u

[niʃ

ikaʝu

ʝuʃit

n u]

comp

l‘as

kfor’

‘dewd

rop’

‘oven’

‘Thed

ewdrop

askedf

orov

en(s)

.’

f./L

LH

LH/

→[1 

 1 1  3 3

  6 1]

ni

ʃika

ʝuʝu

ʝoðo

[niʃ

ikaʝu

ʝuʝoð

o]co

mpl

‘askf

or’

‘dewd

rop’

‘grind

ingsto

ne’

‘Thed

ewdrop

askedf

orgrind

ingsto

ne(s)

.’

66

As seen in § 3.4, nouns like [ʝoðo] ‘plain’ have an underlying /HLH/ pattern.When the subject noun has this tone pattern, the floating Hi tone triggers upstep,much in the same way as /LH/ nouns, as seen in Table 16. The raising effect of thefloating tone of a /HLH/ subject noun, in this case [βiko] ‘fiesta’, onto an objectnoun from each of the six underlying tone patterns can be seen in Table 19.

67

Table

19.B

ehaviou

rof/

HLH /

subje

cts

a./L

LHL

HH/

→[1 

 1 1  3 1

  6 6]

ni

ʃika

βiko

ʝuʝu

[niʃ

ikaβi

koʝuʝ

u]co

mpl

‘askf

or’

‘fiesta

’‘de

w’‘Th

efiest

aasked

fordew.

b./L

LHL

HLH

/→

[1  1 1

  3 1  6 3

]ni

ʃika

βiko

ðusa

[niʃ

ikaβi

koðu

sa]

comp

l‘as

kfor’

‘fiesta

’‘gu

m’‘Th

efiest

aasked

forgu

m.’

c./L

LHL

HHL

/→

[1  1 1

  3 1  6 6

]ni

ʃika

βiko

kuɲu

[niʃ

ikaβi

koku

ɲu]

comp

l‘as

kfor’

‘fiesta

’‘m

eat’

‘Thefi

estaa

sked

forme

at.’

d./L

LHL

HHL

H /→

[1  1 1

  3 1  6 3

]ni

ʃika

βiko

ntsiði

[niʃ

ikaβi

konts

iði]

comp

l‘as

kfor’

‘fiesta

’‘co

rncob’

‘Thefi

estaa

sked

forcorncob(s).’

e./L

LHL

HL/

→[1 

 1 1  3 1

  6 1]

ni

ʃika

βiko

ʃitn u

[niʃ

ikaβi

koʃit

n u]

comp

l‘as

kfor’

‘fiesta

’‘ov

en’

‘Thefi

estaa

sked

forov

en(s)

.’

f./L

LHL

HLH

/→

[1  1 1

  3 1  6 1

]ni

ʃika

βiko

ʝoðo

[niʃ

ikaβi

koʝoð

o]co

mpl

‘askf

or’

‘fiesta

’‘gr

inding

stone’

‘Thefi

estaa

sked

forgrind

ingsto

ne(s)

.’

68

Subject nouns with an underlying /HL/ pattern, such as [kuɲu] ‘meat’, followthe Hi-tone spreading rules within the root described in § 3.4 and follow the two-slot condition when they precede an object noun, as seen in Table 20.

69

Table

20.B

ehaviou

rof/

HL/s

ubjec

tsa.

/LL

HLH/

→[1  1 1  3 1  3 3]

ni

ʃika

kuɲu

ʝuʝu

[niʃ

ikaku

ɲuʝuʝ

u]co

mpl

‘askf

or’

‘meat’

‘dew’

‘Them

eata

sked

fordew.

b./L

LHL

LH/

→[1  1 1  3 3  1 3]

ni

ʃika

kuɲu

ðusa

[niʃ

ikaku

ɲuðu

sa]

comp

l‘as

kfor’

‘hill’

‘gum’

‘Theh

illaskedf

orgu

m.’

c./L

LHL

HL/

→[1  1 1  3 1  3 3]

ni

ʃika

kuɲu

ntʃitʃi

[niʃ

ikaku

ɲuntʃ

itʃi]

comp

l‘as

kfor’

‘meat’

‘strin

gbean’

‘Them

eata

sked

forstr

ingbean

(s).’

d./L

LHL

HLH /

→[1  1 1  3 1  3 1]

ni

ʃika

kuɲu

ntsiði

[niʃ

ikaku

ɲunts

iði]

comp

l‘as

kfor’

‘meat’

‘corncob’

‘Them

eata

sked

forcorncob(s).’

e./L

LHL

L/→

[1  1 1  3 3  1 1]

ni

ʃika

kuɲu

ʃitn u

[niʃ

ikaku

ɲuʃit

n u]

comp

l‘as

kfor’

‘meat’

‘oven’

‘Them

eata

sked

forov

en(s)

.’

f./L

LHL

LH/

→[1  1 1  3 3  1 1]

ni

ʃika

kuɲu

ʝoðo

[niʃ

ikalu

ntsiʝoð

o]co

mpl

‘askf

or’

‘meat’

‘grind

ingsto

ne’

‘Them

eata

sked

forgrind

ingsto

ne(s)

.’

70

As in all other instances, subject nouns with an underlying /L/ pattern, suchas [luntsi ] ‘hill’, have no effect on a following noun, as seen in Table 21.

The effect of /LH/ subject nouns has already been seen in Table 16. The float-ing Hi tone has the same effect as the floating Hi tone sponsored by /HLH/ subjectnouns seen in Table 19.

71

Table

21.B

ehaviou

rof/

L/subje

ctsa.

/LL

LH/

→[1  1 1  1 1  3 3]

ni

ʃika

lunts i

ʝuʝu

[niʃ

ikalu

ntsiʝuʝu

]co

mpl

‘askf

or’

‘hill’

‘drop

’‘Th

ehill

askedf

ordrop

(s).’

b./L

LL

LH/

→[1  1 1  1 1  1 3]

ni

ʃika

lunts i

ðusa

[niʃ

ikalu

ntsið

usa]

comp

l‘as

kfor’

‘hill’

‘gum’

‘Theh

illaskedf

orgu

m.’

c./L

LL

HL/

→[1  1 1  1 1  3 3]

ni

ʃika

lunts i

kuɲu

[niʃ

ikalu

ntsik

uɲu]

comp

l‘as

kfor’

‘hill’

‘meat’

‘Theh

illaskedf

orme

at.’

d./L

LL

HLH /

→[1  1 1  1 1  3 1]

ni

ʃika

lunts i

ntsiði

[niʃ

ikalu

ntsin

ts iði]

comp

l‘as

kfor’

‘hill’

‘corncob’

‘Theh

illaskedf

orcorncob(s).’

e./L

LL

L/→

[1  1 1  1 1  1 1]

ni

ʃika

lunts i

ʃitn u

[niʃ

ikalu

ntsiʃitn

u]co

mpl

‘askf

or’

‘hill’

‘oven’

‘Theh

illaskedf

orov

en(s)

.’

f./L

LL

LH/

→[1  1 1  1 1  1 1]

ni

ʃika

lunts i

ʝoðo

[niʃ

ikalu

ntsiʝoð

o]co

mpl

‘askf

or’

‘hill’

‘grind

ingsto

ne’

‘Theh

illaskedf

orgrind

ingsto

ne(s)

.’

72

3.8 The historical source of floating tones

One of the contributing factors to the difference between the number of under-lying and the number of surface tone patterns in Acatlán Mixtec is the existenceof floating tones. Floating tones are a recent development in Mixtec languages,which have been attributed to the loss of root-final glottal stops, supported by thecomparative analysis of Mixtec dialects in Dürr (1987).

The historic loss of root-final glottal stops in most Mixtec varieties was iden-tified early on by Longacre (1957:82), who claimed that Mixtec “loses final *ʔeverywhere and without trace.”7 However, the tonal trace of this historic loss wasidentified later on. In this respect, Pankratz and E. Pike (1967) state:

In fact, the role of morpheme-final and word-final glottal stop in morpho-tonemics is so important that Ayutla Mixtec may be considered to preservehere an archaic feature of considerable importance to the understanding ofthe development of Mixtec morphotonemics in its manifold dialectal varia-tions.

When E. Pike andWistrand wrote Step-up terrace tone in Acatlán Mixtec in 1974,the concept of floating tones had not yet been fully developed at that time, but theywere fully aware of the historic source of such tones, as seen in their followingstatement: E. Pike and Wistrand (1974:82)

In looking for a historical cause for the step-up tone, we note with inter-est that most Class C Acatlán morphemes which are followed by allomorphsof a higher level are cognate with morphemes of Ayutla Mixtec whose basicallomorphs end in /ʔ/, and that it is these morphemes which (in the Ayutladialect) cause changes in the tone of the following morphemes.

Dürr’s analysis of floating tones in Mixtec (at least those resulting from thehistoric loss of the root-final glottal stop) with a cover feature [±modify] is very

7 Longacre mentions an exception, but he was referring to CVʔV roots.

73

telling of the behaviour of floating tones, since some Mixtec languages have float-ing high tones, others have floating low tones, and at least two, Magdalena Peñasco(Hollenbach 2000:4) and Ixpantepec (Carroll 2015:174), have both floating highand floating low tones. This ambiguity is also true of Acatlán Mixtec, since float-ing tones in Acatlán Mixtec do not exclusively result in upstep. In fact, floatingtones in Acatlán Mixtec result in Hi-tone lowering when they are placed before aHi-toned adjective, as discussed in § 3.4 and § 3.5.

This effect of Hi-tone lowering by a floating tone stands in stark contrast withits usual raising effect before nouns and Lo-toned adjectives. I do not have anexplanation for this register effect, but it is clear that tone processes in Mixtec arenot indifferent to word category, and that Dürr’s proposal for a [±modify] feature,instead of for a floating H or L reflects the diversity of realizations of floating tonesas a reflex of the lost *Ɂ, even within a single dialect.

When comparing roots with a /LH/ pattern with roots that have been identifiedas historically having a root-final glottal stop, the results have confirmed, in mostcases, the correspondence between a floating tone and the historical root-finalglottal stop.

Just as importantly, /L/ patterns correspond in most cases with roots that havebeen identified historically as not having a root-final glottal stop.

In Table 22, words from all classes are listed, since which class a noun belongsto has no bearing on whether a word has a floating tone or not. Proto-Mixtec isrepresented in this table by Josserand’s (1983) reconstructions. Some elementsof her reconstructions have been superseded (notably by McKendry 2013), butour focus is on the root-final glottal stop that existed in Proto-Mixtec, and stillexists in Ayutla,8 Zacatepec and Santiago Nuyoó Mixtec (data with no brackets isphonetic, except in the case of Josserand’s reconstructions of Proto-Mixtec, whichare preceded by asterisks).

8 The following data is from San Felipe de Jesús, Ayutla, Guerrero. Personal communication withMelquiades Gregorio Porfirio.

74

Table

22.O

nthe

origi

noffl

oatin

gton

esin

Acatl

ánMi

xtec

Gloss

/LH /

Proto-

Ayutla

Gloss

/L/

Proto-

Ayutla

Mixt

ecMixt

ec

‘wing

’ndiʃi

*ⁿd

ixįndiʃi

‘pr

iest’

ðutu

*sutu

sutu

‘egg’

ntsiβi

*ⁿd

ɨwɨɁ

ndiβi

Ɂ‘gr

ass’

ʝuku

*yuk

u

‘nose’

ðitn i

*sitįɁ

iʃtiɁ

‘oven’

ʃitn u

*xitų

‘neck’

ðuku

*sukų

Ɂsuku

Ɂ‘sm

oke’

ɲuɁm

a*yuɁwę

ɲuɁm

a

‘eyebrow’

ðikʷa

*sɨkʷaɁ

ʃkʷaɁa

‘chocola

te’ðiɁ

βa

*sɨɁw

asiɁ

βa

‘clou

d’βik

oβik

oɁ‘ye

ar’kʷia

*kʷeɁyi

kʷiʝa

‘eagle

’tað

u*ta

sųɁ

tasuɁ

‘mud

’nᵈeɁi

*ⁿdeɁyu

nᵈjeɁ

i

‘tortilla

’ðit

a*si

taɁiʃtaɁ

‘sickness’

kʷeɁe

*kʷeɁyi

kʷeɁe

‘honey’

ɲuɲu

ɲuɲu

Ɂ‘th

read’

ʝuɁa

*yuɁwe

ʝuɁβa

‘grind

stone’

ʝoðo

ʝosoɁ

‘lard’

ðaɁa

*seɁę

ʃaɁa

‘squa

sh’

tʃiki

*yɨkɨ

Ɂiki

Ɂ‘ho

usepost’

ðiɁi

*sɨɁɨ

‘wind

’tat

ʃi*ta

tiɁtat

ʃiɁ‘fif

teen’

saɁu

*xeɁų

ʃaɁu

‘day’

kiβi

*kɨw

ɨɁkiβ

iɁ‘th

ree’

Ɂuni

*oni

uni

75

Gloss

/LH /

Proto-

Ayutla

Gloss

/L/

Proto-

Ayutla

Mixt

ecMixt

ec

‘back’

ʝata

ʃataɁ

‘six’

Ɂiɲu

*iyǫ

iɲu

‘cotto

n’katʃi

*kati

ɁkatʃiɁ

‘seven’

Ɂusa

*uxe

uʃa

‘four’

kumi

*kɨw

įɁku

miɁ

‘eigh

t’Ɂuna

*one

uɲa

‘prick

lypear’

βiɁndja

*wiɁn

deɁ

βiɁndja

‘ten’

Ɂuʃi

*uxi

uʃi

‘corn-ca

ke’

ðaɁn

d e

ʃaɁndja

‘twenty’

Ɂoko

ok

o

‘snake’

koo

*koo

Ɂko

oɁ‘tw

o’Ɂui

*uwi

uβi

‘bloo

d’nii

*n

ɨyɨɁ

niiɁ

‘nine’

Ɂii

iin

‘mon

th’

ʝoo

ʝooɁ

‘five’

ɁuɁu

*oɁǫ

uɁu

‘ston

e’ʝuu

*yuu

Ɂʝuu

Ɂ

‘ash’

ʝaa

ʝaaɁ

‘foot’

saɁa

*xeɁe

ʃaɁaɁ

‘man

y’kʷaɁa

kʷaɁaɁ

‘salt’

Ɂii

*yɨɨɁ

iiɁ

76

As seen in Table 22, there are only two cases where the floating tone does notcorrespond to the Proto-Mixtec root-final glottal stop reconstructed by Josserand(1983). In the case of [saɁa] ‘foot’, Josserand’s conclusion is probably incorrect,since the conservative Mixtec language spoken in Ayutla retains the root-final glot-tal stop. This leaves us with only one noun with a floating tone that does not cor-respond to a root with an historical root-final glottal stop, [ndiʃi ] ‘wing’. I considerthis instance to be an oddity. Ayutla seems to have lost the root-final glottal stopin the words ‘prickly pear cactus’ and ‘corn-cake’, and the root-final glottal stopin ‘eyebrow’ was probably shifted to a root-medial position due to the historicalchange in its word shape when the vowel /ɨ/ was lost.

3.9 Underspecification

In tonal languages, every TBU need not be specified for tone underlyingly.Underspecification has proved to be useful to explain tonal behaviours in vari-ous languages, and Mixtec is no exception. Regarding this development in toneanalysis, Pulleyblank (1986:104) says:

Autosegmental theory (Goldsmith (1976), Clements and Ford (1979), etc.)makes fundamental use of the assumption that tone-bearing units may beunspecified for tone underlyingly. In the typical cases, it has been sug-gested, even when a morpheme includes tones underlyingly, such tones arenot pre-linked to vowels. On the contrary, the assignment of tones to vowelstakes place as the result of conventions of association.

In the case of Mixtec languages, Snider (1999:101-102) proposed that underly-ingly toneless TBUs in Acatlán Mixtec surface with a default Lo tone, and Daly andHyman (2007) proposed that underlyingly toneless TBUs in Peñoles Mixtec surfacewith a default Mid pitch. On the other hand, McKendry (2013:327), argues thatunderspecification cannot be extended to all varieties of Mixtec and proposes that

77

South-Eastern Nochixtlán Mixtec has three tone levels that are specified underly-ingly for Lo, Mid and Hi.

As proposed by Snider (1999), I argue that in Acatlán Mixtec, Hi tones areunderlyingly specified for Hi tone, low-toned TBUs are underlyingly unspecifiedfor tone, and a low pitch is assigned to them post-lexically through a default featureassignment rule.9 This is not universal for all Mixtec languages, since some Mixteclanguages, such as Peñoles Mixtec (Daly and Hyman 2007:202), have an invertedtonal inventory. Daly and Hyman say:

There is good reason to believe that [Peñoles Mixtec] has inverted the tonesof Proto-Mixtec. A comparison with Dürr’s (1987) reconstructions revealsthe regular correspondences [...]. Dürr’s *L and *H correspond, respec-tively, to [Peñoles Mixtec] /H/ and /Ø/ in the majority tone patterns.

As seen in § 3.5 and § 3.6, the Lo tones of /L/ words do not spread to neigh-bouring TBUs, nor do they trigger any processes. In fact, it is not necessary to positunderlying Lo tones to demonstrate the tone processes present in Acatlán Mixtec.In (51), we see that if we posit that both words are unspecified for tone, a defaultfeature assignment rule correctly predicts that all unspecified TBUs will be realizedwith a low tone.

(51)

9 Some surface Lo tones are underlyingly Hi but have undergone lowering caused by a floating Hi tonesponsored by a previous word, as discussed in § 3.5 and § 3.6.

78

Furthermore, whenever underlyingly toneless TBUs undergo Hi-tone spread-ing, no trace is left of a low tone. As seen in § 3.2 and § 3.3, when Hi-tone spread-ing occurs post-lexically, the Hi tone replaces the pitch of the neighbouring TBUcompletely, leaving no Hi-falling pitch contour or any trace that the TBU was low.In (52) we see that the underlyingly /H/ quantifier [ʔi i ] ‘one’ spreads its toneacross the word boundary non-iteratively, and the default feature assignment rulecorrectly predicts that the remaining TBU will surface with a Lo tone.

(52)

Note that the derivation correctly predicts that no trace is left that the first TBUof ‘plant’ was Lo, and no Lo-tone de-linking rule is necessary in the derivation.

A derivation where a Hi tone triggers the rule-blocking effects of the two-slotcondition can be represented with a unary feature [H], as seen in (53), where HTSis not applied from the first TBU to the second TBU since the Hi tone from thesecond word blocks spreading.

79

(53)

This derivation accounts for the surface forms of phrases such as [kuɲu βiko]‘beautiful meat’, previously seen in (36), where the word [kuɲu] ‘meat’ has anunderlying /HØ/ pattern, and the word [βiko] ‘beautiful’ has an underlying /H/pattern.

However, when [kuɲu] is in isolation, the two-slot condition for HTS is met,so the Hi tone spreads within the root to the second TBU, leaving no trace of a Lotone, as can be seen in the derivation (54).

(54)

The second TBU is effectively behaving as an empty slot, where the Lo tonehas no influence on neighbouring TBUs.

To account for the surface [HL] pattern of /HLH/ words in isolation, suchas [ndutʃi ] ‘bean’, a derivation with a unary feature [H] can also account for thesurface realization, as seen in (55), where the Hi tone is underlyingly linked to thefirst TBU and the floating Hi tone is encircled to indicate that it is not linked toany TBU.

80

(55)

The two-slot condition for HTS to apply is not met due to the presence of thefloating Hi tone, so a default Lo tone is assigned to the second TBU. Note that thefloating tone is still present after the default feature assignment. However, it doesnot surface since it is not linked to any segmental structure, such as a mora, so itremains floating.

When the floating tone docks onto a Lo-toned noun, only the first TBU is raisedto an extra-high pitch. For example, in the phrase [kumi ʝuku] ‘four plants’, wherethe second word [ʝuku] ‘plant’ has an underlying /Ø/ pattern, the floating tonesponsored by ‘four’ raises the first TBU of ‘plant’ to an extra-high pitch, as seen in(56).

(56)/ØH Ø/ → [1 1   6 1]kumi ʝuku [kumi ʝuku]‘four’ ‘pot’ ‘four pots’

If the Lo tones of the second word [ʝuku] ‘plant’ were specified for Lo tone, theregister effect of the floating tone docking onto the first TBU of the target wordwould also have register effects on the following TBU. However, since the word[ʝuku] ‘plant’ is unspecified for tone, the default feature assignment rule assigns apost-lexical Lo tone to the last TBU that surfaces at the same level as the TBUs ofthe initial word [kumi ] ‘four’.

81

In the same manner, in a phrase like [kumi kiði ] ‘four pots’, the first TBU of[kiði ] ‘pot’, which has an underlying /ØH/ pattern, is raised to an extra-high level,but the last TBU retains its usual Hi pitch, as seen in example (57).

(57)/ØH ØH/ → [1 1   6 3]kumi kiði [kumi kiði]‘four’ ‘pot’ ‘four pots’

When a floating Hi tone docks forward onto the following TBU, as in (57), itis not blocked by the two-slot condition. Hi tones underlyingly linked to a TBUcannot spread in a similar way, but a floating Hi tone will dock, even if the empty(Lo) slot is followed by a Hi.

If the Lo tone of [kiði ] ‘pot’ were present underlyingly, the floating tone spon-sored by the first word would de-link or erase it, potentially having repercussionson the last TBU. However, the final TBU surfaces at a regular Hi pitch.

Lastly, consider floating tones that target words with an underlying /HØH/pattern. For example, in the phrase [kumi ndutʃi ] ‘four beans’, where the word[ndutʃi ] bean has an underlying /HØH/ pattern, the last TBU surfaces with a Hipitch, as seen in (58).

(58)/ØH HØH/ → [1  1     6  3]kumi ndutʃi [kumi ndutʃi]‘four’ ‘bean’ ‘four beans’

The source of the Hi pitch of the last TBU is unclear, but I assume that it is dueto its position between two floating tones, and that examples (57) and (58) havethe same surface tone in spite of having different underlying tone patterns.

82

The /ØH HØH/ sequence of [kumi ndutʃi ] ‘four beans’ begs the question: Whathappens to a word that follows such a sequence? A good example is the phrase[kumi ndutʃi ʔui ] ‘four split beans’ seen in (59).

(59)/ØH HØH Ø/ → [1 1      6  1  61]kumi ndutʃi ʔui [kumi ndutʃi ʔui]‘four’ ‘bean’ ‘two’ ‘four split beans’

The compound [ndutʃi ʔui ] ‘split beans’ refers to a dish that consists of beansthat have previously been split in halves and cooked. The compound consists ofthe /HØH/ noun [ndutʃi ] ‘bean’ and the /Ø/ quantifier [ʔui ] ‘two’.

The first floating tone raises the Hi tone of [ndutʃi ] ‘bean’ to an extra-high leveland the floating tone sponsored by ‘bean’ raises the first TBU of [ʔui ] ‘two’ to thesame extra-high level. In this case, the floating tone sponsored by [ndutʃi ] ‘bean’docks onto the first TBU of [ʔui ] ‘two’, and the second TBU of [ndutʃi ], insteadof surfacing with a Hi tone, as in (58), is assigned a default post-lexical Lo tone.This is because the second TBU of [ndutʃi ] is no longer located between a dockedfloating Hi tone and an undocked floating Hi tone. The surface realization of thesecond TBU of [ndutʃi ] in (59) is not unlike that seen when[ndutʃi ] is uttered inisolation, as seen previously in (55).

83

CHAPTER 4CONCLUSION

4.1 Findings

The tonal system of Acatlán Mixtec assumes a basic contrast between H and Ltone (or H and ∅, if underspecification is used); some of the H tones are specifiedas floating in the underlying structure. The tonal patterns for all noun roots arecomposed of nothing more than that. I propose that all of the extra-high tones thatoccur on the surface can be traced back, in one way or another, to the presence offloating H tones in the underlying forms.

While many varieties of Mixtec, including the Acatlán variety, have been de-scribed as having three tones (Catecismo (1899:4-5), K. Pike (1944:115), Pankratzand E. Pike (1967:291), E. Pike and Wistrand (1974:83), E. Pike and Small (1974),Macaulay (1987:32), McKendry (2013:327), Carroll (2015)), recognizing a thirdbasic tone is unnecessary to the analysis of the tone system of Acatlán Mixtec.Although the research needed in order to demonstrate the veracity of this hypoth-esis has not been pursued fully, given what we have already seen within the nounphrase, we are led to believe that this hypothesis will be confirmed.

Simple nouns only have six underlying tone patterns in noun roots: /H/, /L/,/LH/, /HL/, /HLH/ and /LH/, where (H) is a floating high tone. As proposed inDürr (1987:29), the floating H tones of noun roots are the reflex of Proto-Mixtecroot-final glottal stops.

84

In Acatlán Mixtec, Hi tones are underlyingly specified for H, low-toned TBUsare underlyingly unspecified for tone, and a low pitch is assigned to them post-lexically through a default feature assignment rule (Snider 1999:101-102). In thedata, there is no evidence that the L tone triggers any processes, or that any traceof a L tone is left whenever a Hi tone spreads onto a Lo-toned TBU.

Consequently, the six underlying tone patterns for simple nouns in AcatlánMixtec can more appropriately be expressed as /H/, /∅/, /∅H/, /H∅/, /H∅H/and /∅H/. The null symbol (∅) indicates that one or both TBUs in a bimoraic nounroot are unspecified for the unary tonal feature [H], and Hi tones are distributedover one or two TBUs, or are not underlyingly linked to any TBU, in which casethe Hi tone is floating (H).

In the course of my research, I isolated three tone processes that operate onthe underlying tone patterns in context: Hi-tone spreading, Hi-tone raising, andHi-tone lowering. Hi-tone spreading spreads H rightward non-iteratively onto asingle TBU as long as the two following TBUs are L, or if the target Lo-toned TBUis utterance-final. This process may apply across word boundaries or within thenoun, and it only applies if the “two-slot condition” is met. The “two-slot con-dition” is an OCP constraint that requires the two following TBUs to be L, or thetarget Lo-toned TBU to be utterance-final. If the condition is not met, the Hi-tonespreading rule is blocked to prevent a series of contiguous Hi tones. Hi-tone rais-ing occurs when a floating Hi-tone precedes another word. In such cases, the firstTBU of the following word, whether it begins with L or H tone, is realized withan extra-high pitch. If it begins with a H tone, any subsequent Hi-toned TBUs arerealized at the extra-high pitch of the first raised H tone. However, if the followingTBU is a Lo-toned adjective, the first TBU of the adjective is realized with a normalHi pitch. Hi-tone lowering occurs when a floating H tone precedes a Hi-toned ad-jective. In this case, both TBUs of the underlyingly Hi-toned adjective are realizedwith a Lo pitch.

85

4.2 Areas for further research

The subject of this thesis was directed to the tone of noun phrases, more specif-ically simple, alienable nouns. Further research is necessary in the following cases:

• Nouns with a classifier prefix. Most prefixes have an underlying /LH/ tonepattern (or /∅H/, if underspecification is used), except for the rare prefix[ntsi‑], which has an underlying /H/ tone pattern. Further research shouldconfirm that there need be no new patterns in noun roots that take a prefix,and the interaction between the base tone of the noun root and that of theprefix should be explored.

• Compounds. Many extra-high tones can be attributed to the tonal interactionbetween the morphemes that go into the makeup of compounds.

• Inalienable nouns. The interaction between noun roots and possessive pro-nouns requires the collection and analysis of a great array of combinations,since possessive pronouns are phonologically bound to the noun root, in somecases to the point where the possessive pronoun cannot be distinguished atthe surface from the root. Besides, the existence of floating tones in the nounroot as well as the possessive pronouns, and the further complication of thetwo-slot condition promises the researcher a fertile ground of inquiry.

Besides loanwords, nouns consisting of a bimoraic couplet hardly ever haveextra-high tones in Acatlán Mixtec.1 I believe further research will confirm thatall extra-high tones result from the tonal interaction between the morphemes thatmake up complex nouns, especially when the first morpheme has a floating Hitone. The few nouns that consist of a bimoraic couplet and have a surface extra-high tone may all be compounds.

1 The are only four exceptions in the data: [ʔi ʔi ] ‘sweat lodge’ [βaʔna] ‘sleepiness’, [toʔo] ‘stranger’ and[mee] ‘baby’.

86

The processes triggered by floating Hi tones are not the same in n-a construc-tions as in other environments. An explanation for this discrepancy was not foundin this thesis.

Further investigation should shed light on whether Hi-toned nouns and Hi-toned quantifiers have the same underlying tone or not in Acatlán Mixtec. They donot always behave alike in Acatlán Mixtec, but cross-dialectically, there is reasonto believe they may be underlyingly the same.

Within the noun phrase, the tonal behaviour of adjectives, degree words, pos-sessive pronouns and prepositions require further study.

The tone in the verbal systemwas not discussed in this thesis, except in passing.The following avenues of future research are suggested:

• Isolate the underlying tone patterns of verb roots and the morphemes thatmake up the stem.

• Interaction of words within the verb phrase, in particular between the verband the words/particles that may be used in construction with it.

• Syntactic analysis in order to isolate floating tone morphemes present in cer-tain constructions, or boundary tones which delimit certain phrases.

The possibility of studying iterative upstep specifically was largely curtaileddue to this thesis’ emphasis on nouns. The cumulative effects of upstep can bebetter observed in larger domains. Iterative upstep in Acatlán Mixtec has been ofspecial interest to linguists due to its cross-linguistic rarity, and much remains tobe analyzed regarding this tonal behaviour. Floating tones raise all subsequentHi tones to an extra-high level, and extra-high tones (which are derived and notpresent underlyingly) are raised to an even higher level when upstepped.

87

Cross-dialectal studies that take into account Acatlán Mixtec data will helpbetter understand the tonal phenomena that result from the loss of the root-finalglottal stop in most Mixtec varieties.

Finally, the proposal that Lo tones are underlyingly unspecified for tone hasgreat explanatory potential for the study of Mixtec tone patterns cross-dialectically.

88

APPENDIX AWord list with underlying tones

Table 23. Simple nouns with an underlying /L/ tone patternWord Gender English gloss Spanish gloss

ðiða alt ‘film’ ‘capa/nata’

ðusa alt ‘surface debris’ ‘residuo fluvial’

ðutu div ‘priest’ ‘sacerdote’

kʷika alt ‘comb’ ‘peine’

lila alt ‘film’ ‘capa/nata’

luntsi ~ ɾuntsi alt ‘hill’ ‘mogote’

ndutu alt ‘bud’ ‘botón’

ɲuʔma alt ‘smoke’ ‘humo’

tuku alt ‘corn-cake’ ‘memela’

ʃitnu alt ‘oven’ ‘horno’

ʝuku alt ‘grass/plant’ ‘hierba’

ðaʔβa alt ‘cliff’ ‘cantil’

ðiʔβa alt ‘chocolate’ ‘chocolate’

kʷia alt ‘year’ ‘año’

ðaʔa alt ‘lard’ ‘manteca’

89

Word Gender English gloss Spanish gloss

ðiʔi alt ‘wooden post’ ‘horcón’

kʷeʔe alt ‘illness’ ‘enfermedad’

ntʃiʔo alt ‘burrow’ ‘madriguera’

ndeʔi alt ‘mud’ ‘lodo’

nduʔa alt ‘spiderweb’ ‘telaraña’

ʃeʔe alt ‘dust’ ‘polvo’

ʃiʔi1 alt ‘mushroom’ ‘hongo’

ʝuʔa alt ‘ice/snow’ ‘hielo/nieve’

ʝuʔa alt ‘thread’ ‘hilo’

1 Attested in Gabino Barreda; archaic in Xayacatlán.

90

Table 24. Simple nouns with an underlying /H/ tone patternWord Gender English gloss Spanish gloss

ðuma alt ‘ghost’ ‘espíritu’

ðutʃi hum ‘child’ ‘bebé’

kwaʝa alt ‘dead leaves’ ‘hojarasca’

lambe alt ‘buttocks’ ‘nalgas’

landu alt ‘navel’ ‘ombligo’

lusu zoa ‘grey fox’ ‘zorro’

ndusa alt ‘vomit’ ‘vómito’

tʃitu zoa ‘cat’ ‘gato’

ʝuʝu alt ‘droplet/dew’ ‘gota/rocío’

kolo zoa ‘turkey’ ‘guajolote’

leka alt ‘seed pouch’ ‘morral’

lutʃi hum ‘child’ ‘bebé’

sjaka zoa ‘fish’ ‘pescado’

sjuku ~ sjuɡŋ zoa ‘mosquito’ ‘zancudo’

tupa alt ‘evil spirit’ ‘espanto’

luʔlu alt ‘marrow’ ‘tuétano’

ndaʔnde alt ‘mucus’ ‘moco’

xoʔlo alt ‘carnival’ ‘carnaval’

tʃio alt ‘shard’ ‘barro’

ʝoo zoa ‘moon’ ‘luna’

91

Word Gender English gloss Spanish glossβeʔe alt ‘house’ ‘casa’

tʃuu alt ‘work’ ‘trabajo’

tʃiʔa alt ‘curse’ ‘maldición’

ʃiʔi zoa ‘cicada’ ‘chiquilichi’

ʔiða alt ‘waist loom’ ‘telar de cintura’

ʔiʔna alt ‘spirit’ ‘espíritu’

ʔii alt ‘hail’ ‘granizo’

92

Table 25. Simple nouns with an underlying /HL/ tone patternWord Gender English gloss Spanish gloss

βaʝa div ‘squash blossom’ ‘flor de calabaza’

ðaʃi alt ‘dysentery’ ‘disentería’

ðiði hum ‘aunt’ ‘tía’

ðiki alt ‘jewel’ ‘alhaja’

ðiko alt ‘smell’ ‘olor’

ðito hum ‘uncle’ ‘tío’

ðoko alt ‘hunger’ ‘hambre’

kuɲu alt ‘meat’ ‘carne’

milo zoa ‘rabbit’ ‘conejo’

nduði alt ‘honey’ ‘miel’

nduʝu alt ‘stake’ ‘estaca’

nduli alt ‘honey’ ‘miel’

ntʃitja alt ‘cuajilote pod’ ‘cuajilote’

ntʃitʃi alt ‘string bean’ ‘ejote’

ɲuma alt ‘dirt’ ‘mugre’

satu alt ‘pants’ ‘calzón’

tata div ‘seed’ ‘semilla’

ʝoko zoa ‘wasp’ ‘avispa’

ʝutʃi alt ‘knife’ ‘cuchillo’

kaka alt ‘lime’ ‘cal’

93

Word Gender English gloss Spanish glosskotʃi zoa ‘pig’ ‘cochino’

sani alt ‘dream’ ‘sueño’

satu alt ‘trousers’ ‘calzón’

taka alt ‘nest’ ‘nido’

ɲaʔmi alt ‘sweet potato’ ‘camote’

βiu alt ‘corn plant’ ‘milpa’

ʝaa alt ‘song’ ‘canción’

ʝoo alt ‘pitcher’ ‘cántaro’

ʝua alt ‘pigweed’ ‘quelite’

ʝui alt ‘gully’ ‘cañada’

ndua alt ‘guaje pod’ ‘guaje’

tei alt ‘saddle’ ‘montura’

ɲuʔu alt ‘fire’ ‘lumbre’

leʔe zoa ‘rooster’ ‘gallo’

luʔu zoa ‘parasol ant’ ‘arriera’

ʔiðu zoa ‘horse’ ‘caballo’

ʔitnu alt ‘hilltop field’ ‘loma’

ʔitu alt ‘corn field’ ‘milpa’

94

Table 26. Simple nouns with an underlying /LH/ tone patternWord Gender English gloss Spanish gloss

ðiðo hum ‘in-law’ ‘suegro’

ðikʷi alt ‘sap/rubber’ ‘savia/goma’

ðusa alt ‘gum’ ‘chicle’

ʝaka alt ‘dust’ ‘polvo’

ʝute liq ‘river’ ‘río’

kiði alt ‘pot’ ‘olla’

kʷatʃi alt ‘sin’ ‘culpa’

minu alt ‘goosefoot’ ‘epazote’

sanu hum ‘daughter-in-law’ ‘nuera’

sika zoa ‘grasshopper’ ‘chapulín’

tiʃi alt ‘stomach’ ‘panza’

tasa alt ‘lightning’ ‘relámpago’

tnumi alt ‘feather’ ‘pluma’

tuntsi zoa ‘cactus flower’ ‘cacayo’

tutʃi alt ‘tendon’ ‘tendón’

ʃito arb ‘cot’ ‘catre’

ʝaʔβi alt ‘market’ ‘mercado’

ðia alt ‘skirt’ ‘falda’

kaa alt ‘metal’ ‘fierro’

tia hum ‘man’ ‘hombre’

95

Word Gender English gloss Spanish glosstjaa hum ‘man’ ‘hombre’

tʃii alt ‘nail’ ‘uña’

ðaʔa alt ‘vagina’ ‘vagina’

ðeʔe hum ‘child’ ‘hijo’

ðiʔi alt ‘upper leg’ ‘muslo’

kuʔi alt ‘banana’ ‘plátano’

kʷiʔi alt ‘banana’ ‘plátano’

loʔo alt ‘ear’ ‘oreja’

ndoʔo alt ‘adobe brick’ ‘adobe’

nuʔu alt ‘tooth’ ‘diente’

tnuʔu alt ‘mentira’ ‘lie’

96

Table 27. Simple nouns with an underlying /LH/ tone patternWord Gender English gloss Spanish gloss

βiko alt ‘cloud’ ‘nube’

ðikwa alt ‘eyebrow’ ‘ceja’

ðini alt ‘head’ ‘cabeza’

ðita alt ‘tortilla’ ‘tortilla’

ðitni alt ‘nose’ ‘nariz’

ðitʃi alt ‘Cereus cactus pear’ ‘pitaya’

ðoko alt ‘shoulder’ ‘hombro’

ðuku ~ ðuɡŋ alt ‘neck’ ‘cuello’

ʝata arb ‘plough’ ‘coa/arado’

ʝata alt ‘back’ ‘espalda’

ʝoðo alt ‘grinding stone’ ‘metate’

ʝoðo alt ‘spleen’ ‘bazo’

ʝoko alt ‘steam/temperature’ ‘vapor/calentura’

kaβa alt ‘gall bladder’ ‘vesícula’

kalu zoa ‘crow’ ‘cuervo’

katʃi alt ‘cotton’ ‘algodón’

kiβi alt ‘day’ ‘día’

nama alt ‘soap’ ‘jabón’

ndisa alt ‘sandal’ ‘huarache’

ntsiβi alt ‘egg’ ‘huevo’

97

Word Gender English gloss Spanish glossɲama alt ‘husk’ ‘hoja de maíz/cáscara’

ɲuma alt ‘wax’ ‘cera’

ɲuɲu zoa ‘bee’ ‘abeja’

siða alt ‘penis’ ‘pene’

sjaβa zoa ‘bed bugs’ ‘tlalzahuate’

sjatna zoa ‘chicken’ ‘gallina’

taðu zoa ‘roadside hawk’ ‘gavilán’

tatna alt ‘medicine’ ‘medicina’

tatʃi div ‘wind’ ‘viento’

tutnu alt ‘ear’ ‘oreja’

tʃiku ~ tʃiɡŋ zoa ‘guamuchil pod’ ‘guamúchil’

βiʔndja alt ‘prickly pear cactus leaf’ ‘penca de nopal’

ðaʔnde alt ‘corn-cake’ ‘memela’

loʔngo ~ ɾoʔngo alt ‘knee’ ‘rodilla’

taʔβi alt ‘gift’ ‘regalo’

tʃiʔndu alt ‘shell’ ‘concha’

ðiu alt ‘anus/bum’ ‘ano/nalgas’

ðoo alt ‘blanket’ ‘cobija’

ʝaa alt ‘ash’ ‘ceniza’

ʝoo alt ‘month/period’ ‘mes/menstruación’

ʝuu alt ‘stone’ ‘piedra’

98

Word Gender English gloss Spanish glosskoo zoa ‘snake’ ‘culebra’

kwii alt ‘crack’ ‘grieta’

ndea alt ‘mesquite pod’ ‘mezquite’

nii alt ‘blood’ ‘sangre’

nuu alt ‘face’ ‘cara’

ɲuu alt ‘palm’ ‘palma’

tnii liq ‘sweat’ ‘sudor’

tnuu alt ‘coal’ ‘carbón’

tʃio alt ‘upper arm’ ‘brazo’

ʃio alt ‘griddle’ ‘comal’

ʃii hum ‘ancestor’ ‘ancestro’

ðaʔa alt ‘language’ ‘idioma’

ðiʔu alt ‘money’ ‘dinero’

koʔo alt ‘plate’ ‘plato’

ndoʔo zoa ‘type of basket’ ‘talín’

saʔa alt ‘foot’ ‘pie’

sjoʔo zoa ‘flea’ ‘piojo’

ʔika alt ‘body’ ‘cuerpo’

ʔika zoa ‘cane strip basket’ ‘chiquihuite’

ʔiki alt ‘ridge’ ‘cuchilla’

ʔii alt ‘salt’ ‘sal’

99

Table 28. Simple nouns with an underlying /HLH/ tone patternWord Gender English gloss Spanish gloss

ʝaβi alt ‘century plant’ ‘maguey’

ʝakʷi zoa ‘armadillo’ ‘armadillo’

ʝuku alt ‘mountain’ ‘montaña’

ʝusa alt ‘dough’ ‘masa’

ʝutnu arb ‘tree/stick’ ‘árbol/palo’

kata arb ‘aloe vera’ ‘sávila’

kutu alt ‘copal resin’ ‘copal’

ndaku alt ‘hominy’ ‘nixtamal/pozole’

ndakwa alt ‘pus’ ‘pus’

ndiʃi alt ‘wing’ ‘ala’

ndoko alt ‘sapote fruit’ ‘zapote’

ndute liq ‘liquid’ ‘líquido’

ndutʃi alt ‘bean’ ‘frijol’

ntsiði alt ‘ear of corn’ ‘elote’

ntsiki alt ‘horn’ ‘cuerno’

ntsiki ~ ntsiɡŋ zoa ‘amaranth seed’ ‘alegría’

pentʃi hum ‘orphan’ ‘huérfano’

sjuku zoa ‘louse’ ‘piojo’

sjumi zoa ‘owl’ ‘tecolote’

titʃi zoa ‘avocado’ ‘aguacate’

100

Word Gender English gloss Spanish gloss

tutnu arb ‘firewood’ ‘leña’

tʃiki zoa ‘prickly pear’ ‘tuna’

ðiʔβi alt ‘dung’ ‘caca’

ʝui alt ‘woven mat’ ‘petate’

nii alt ‘dry corncob’ ‘mazorca’

ndei alt ‘mole sauce’ ‘mole’

ʝaʔa alt ‘chili pepper’ ‘chile’

ʝeʔe alt ‘door’ ‘puerta’

ʝoʔo alt ‘root’ ‘raíz’

ʝuʔu alt ‘mouth’ ‘boca’

luʔu ~ ɾuʔu zoa ‘parasol ant’ ‘arriera’

ndoʔo alt ‘tail’ ‘cola’

ɲaʔa hum ‘woman’ ‘mujer’

ʔiði alt ‘hair’ ‘pelo’

ʔiki alt ‘bone’ ‘hueso’

ʔiɲu alt ‘thorn’ ‘espina’

ʔii alt ‘hide/skin’ ‘cuero/piel’

ʔiu alt ‘thorn’ ‘espina’

101

REFERENCES

Alexander, Ruth Mary. 1980. Gramática mixteca de Atatláhuca. Mexico City: ElInstituto Lingüístico de Verano, A.C.

American Community Survey Report. 2011. Languages spoken at home withover 100,000 speakers older than five.http://www.census.gov/prod/2013pubs/acs-22.pdf. (November 22, 2016.)

Anonymous. 1899. Catecismo de la doctrina cristiana en la lengua mixteca por unmixteco. Puebla: Tip. del Colegio Pío de Artes y Oficios.

Aronovich, Raul. 1994. The tone system of Acatlán Mixtec and some exceptionsto the OCP. Linguistic Notes from La Jolla: Phonology Issue. San Diego:Department of Linguistics, University of San Diego. 17. 3-26.

Bradley, C. Henry and J. Kathryn Josserand. 1982. El protomixteco y susdescendientes. Anales de Antropología e Historia. 19 (2). 279-343.

Campbell, Lyle. 1997. American Indian languages: the historical linguistics of NativeAmerica. Oxford: Oxford University Press.

Carroll, Lucien. 2015. Ixpantepec Nieves Mixtec Word Prosody. San Diego:University of California dissertation.

Castillo García, Rey. 2007. Descripción fonológica segmental y tonal del mixteco deYoloxóchitl. Mexico City: Centro de Investigaciones y Estudios Superiores enAntropología Social (CIESAS) thesis.

Clements, G.N. and Kevin C. Ford. 1979. Kikuyu tone shift and its synchronicconsequences. Linguistic Inquiry. 10. 179-210.

Corbett, Greville G. 1991. Gender. Cambridge: Cambridge University Press.Daly, John P. and Larry M. Hyman. 2007. On the representation of tone in

Peñoles Mixtec. International Journal of American Linguistics. 73. 165-207.Dürr, Michael. 1987. A preliminary reconstruction of the Proto-Mixtec tonal

system. Indiana 11. 19-62.Eberhardt, Roy. 1999. Questions and inversion in Ocotepec Mixtec. Workpapers

of the Summer Institute of Linguistics, University of North Dakota. 43.

102

http://arts-sciences.und.edu/summer-institute-of-linguistics/work-papers/_files/docs/1999-eberhardt.pdf. (December 4,2017.)

Flechsig, Katrin S. 2004. Miniature crafts and their makers: Palm weaving in aMexican town. Tucson: University of Arizona Press.

Gittlen, Laura and Stephen A. Marlett. 1985. Ñumí Mixtec syllable structure andmorphology. Work Papers of the Summer Institute of Linguistics, University ofNorth Dakota. 19. 175-94.

Goldsmith, John A. 1976. Autosegmental phonology. Indiana UniversityLinguistics Club: Garland Press.

Grimes, Joseph E. 1974. Dialects as optimal communication networks. Language.50. 260-69.

Harris, Larry R. 1995. Some features of narrative discourse in Santiago NuyooMixtec. Summer Institute of Linguistics Mexico Workpapers. 11. 79-95.

Hollenbach, Barbara E. 2000. The historical source of an irregular tone-sandhipattern. Summer Institute of Linguistics, ms.

INALI (Instituto Nacional de Lenguas Indígenas). 2008. Catálogo de las lenguasindígenas nacionales: Variantes lingüísticas de México con susautodenominaciones y referencias geoestadísticas.http://www.inali.gob.mx/clin-inali/. (July 14, 2017.)

INEGI (Instituto Nacional de Estadística y Geografía). 2010. Censo de poblacióny vivienda.http://www.inegi.org.mx/est/contenidos/proyectos/ccpv/cpv2010/Default.aspx.(November 22, 2016.)

INEGI (Instituto Nacional de Estadística y Geografía). 2015. Encuesta intercensal.http://cuentame.inegi.org.mx/hipertexto/todas_lenguas.htm. (July 8, 2017.)

Josserand, J. Kathryn. 1983. Mixtec dialect history. New Orleans: TulaneUniversity dissertation.

Kaufman, Terrence. 2016. Tlapaneko-Subtiaba, OtoMangean and Hokan: whereGreenberg went wrong. University of Pittsburgh, ms.

Kuiper, Alberta and Velma Pickett. 1974. Personal pronouns in Diuxi Mixtec.Sumer Institute of Linguistics Mexico Workpapers. 1. 53-58.

León Pasquel, María de Lourdes de. 1988. Noun and numeral classifiers in Mixtecand Tzotzil. Sussex: University of Sussex dissertation.

103

Longacre, Robert E. 1957. Proto-Mixtecan. Publication 5. Indiana UniversityResearch Center in Anthropology, Folklore and Linguistics, Bloomington:Indiana University. International Journal of American Linguistics. 23. 4.

Longacre, Robert E. 1961. Swadesh’s Macro-Mixtecan hypothesis. InternationalJournal of American Linguistics. 27. 9-29.

Macaulay, Monica. 1987. Cliticization and the morphosyntax of Mixtec.International Journal of American Linguistics. 53. 119-35.

Macaulay, Monica. 1996. A grammar of Chalcatongo Mixtec. Berkeley and LosAngeles: University of California Press (University of California Publicationsin Linguistics vol. 127).

Macaulay, Monica and Joseph C. Salmons. 1995. The phonology of glottalizationin Mixtec. International Journal of American Linguistics. The University ofChicago Press. 61 (1). 38-61.

Mak, Cornelia and Robert E. Longacre. 1960. Proto-Mixtec phonology.International Journal of American Linguistics. 26. 23-40.

McCarthy, John J. 1986. OCP effects: Gemination and antigemination. LinguisticsDepartment Faculty Publication Series. 51. 207-63.

McKendry, Inga. 2001. Two studies of Mixtec languages. Grand forks: Universityof North Dakota thesis.

McKendry, Inga. 2013. Tonal association, prominence and prosodic structure inSouth-Eastern Nochixtlán Mixtec. Edinburgh: University of Edinburghdissertation.

Marlett, Stephen A. 1992. Nasalization in Mixtec languages. International Journalof American Linguistics. The University of Chicago Press. 58 (4). 425-35.

Pankratz, Leo and Eunice V. Pike. 1967. Phonology and morphotonemics ofAyutla Mixtec. International Journal of American Linguistics. 33. 287-99.

Pike, Eunice V. and Kent Wistrand. 1974. Step-up terrace-tone in Acatlán Mixtec.Advances in Tagmemics. Amsterdam: North Holland. 81-104.

Pike, Eunice V. and Priscilla C. Small. 1974. Downstepping terrace tone inCoatzospan Mixtec. Advances in Tagmemics. Amsterdam: North Holland.105-34.

Pike, Kenneth L. 1948. Tone languages. Ann Arbor: The University of MichiganPress.

104

Pike, Kenneth L. 1944. Analysis of a Mixteco text. International Journal ofAmerican Linguistics. 10. 113-38.

Pulleyblank, Douglas. 1986. Tone in lexical phonology. Dordrecht, Holland: D.Reidel Publishing Company.

Reyes, Antonio de los. 1593. Arte en lengua mixteca. Mexico City: Casa de PedroBalli.

Simons, Gary F. and Charles D. Fennig (eds.). 2017. Ethnologue: Languages of theWorld. Online version: http://www.ethnologue.com. Dallas, Texas: SILInternational.

Snider, Keith L. 1988. Towards the representation of tone: A three dimensionalapproach. In Harry van der Hulst and Norval Smith (eds.), Feature, SegmentalStructure and Harmony Processes. Dordrecht: Foris Publications. 1. 237-67.

Snider, Keith L. 1999. The geometry and features of tone. Dallas: The SummerInstitute of Linguistics and the University of Texas at Arlington, Dallas.

Snider, Keith L. In press. Tone analysis for field linguists. Dallas: SIL International.Stark, Sharon C., Audrey Johnson and Benita González de Guzmán. 2013.

Diccionario básico del mixteco de Xochapa. Mexico City: El Instituto Lingüísticode Verano, A.C.

Torrens, Claudia. 2011. Some NY immigrants cite lack of Spanish as barrier, inThe San Diego Union-Tribune. http://www.sandiegouniontribune.com/sdut-some-ny-immigrants-cite-lack-of-spanish-as-barrier-2011may28-story.html.(Nov 22, 2016.)

United commission of Misteca Baja and Montañez priests. 1837. Catecismo enidioma mixteco, según se habla en los curatos de la Misteca Baja, que pertenecen alObispado de Puebla. Puebla: Imprenta del Hospital de San Pedro.

Vázquez Peralta, Rodrigo. 1997. Gramática popular de la lengua mixteca del sur dePuebla. 11/139. (Fojas Étnicas.) Puebla: Secretaría de Cultura.

105


Recommended