Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Tone gestures and constraint interaction in Sierra Juarez Zapotec
(USC Thesis Other)
Tone gestures and constraint interaction in Sierra Juarez Zapotec
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
TONE GESTURES AND CONSTRAINT INTERACTION
IN SIERRA JUAREZ ZAPOTEC
by
Laura Tejada
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree of
DOCTOR OF PHILOSOPHY
(LINGUISTICS)
May 2012
Copyright 2012 Laura Tejada
ii
Dedication
To my children, Gabriel Armando and Sarayu Amalia,
for brightening my world each and every day.
iii
Acknowledgments
This dissertation would have been impossible without the help and inspiration of
many people. First and foremost I would like to thank my dissertation chair Rachel
Walker, for all her hard work and helpful encouragement along the way. Rachel is truly
an amazing scholar and teacher. I would also like to thank Louis Goldstein for inspiring
me with his knowledge and intelligence, and for always being patient with me as I
discovered the world of gestures. Thanks to Dani Byrd for teaching my first ever
phonetics class, and for being a role model as a parent and a scholar. A special thanks
goes to Pam Munro for inspiring me to carry out fieldwork on Zapotec and for helping
me along the way, and to Larry Hyman, Melissa Frazier, and Mario Saltarelli for their
help and guidance. I am grateful for the support and encouragement of the community of
scholars who work on Zapotec and related languages, and especially Christian DiCanio,
Aaron Broadwell, Michael Galant, Aaron Sonnenchein, Christina Esposito, and
Rosemary Beam de Azcona. Thanks to Angela Della Volpe, who first encouraged me to
attend graduate school and pursue my dream. Graciela Bautista and Donaldo Pérez merit
special praise for their help in providing the data and for helping to watch my daughter
Sarayu as I worked to record and transcribe our fieldwork sessions.
Thanks also to the wonderful group of scholars and students who shared time with
me at USC: Hector Velásquez, Michelle Har-Kim, Erika Varis Doggett, Michael
Shepherd, Magdalena Pire-Schmidt, Ben Parrell, Daylen Riggs, Katy McKinney-Bock,
David Li, Michael Proctor, Alvaro Cerrón-Palomino, Michal Temkin-Martínez, Janet
Anderson, Kate Hwang, Canan Ipek, Xiao He, Erin Tavano, Christina Hagedorn, and
iv
Ann Sawyer. Special thanks to Stephanie Huang for sharing parenting tips along the way
and for helping me through syntax class our first year.
None of this would have been possible without the help of Joyce Perez, who
always knew what I needed to do and when, and who encouraged me along the way.
Thanks to Yi-Hsien Walker and Aaron Walker for being my friends and for teaching my
son Gabriel how to be a true USC Trojans fan.
Thanks to my parents Daryl and Sally for their support and love, and to other
members of my family: Susan, Jane, John, Kristen, David, Bill, Marge, Sam, and Renata.
My husband Armando deserves special praise for his support and love, not only for
seeing me through the birth of our two children but also for supporting me through all the
ups and downs of graduate school, new jobs, and our life together. I could not have done
much at all without his love and support.
v
Table of Contents
Dedication.......................................................................................................... ii
Acknowledgments............................................................................................. iii
List of Tables ................................................................................................. viii
List of Figures .................................................................................................. xi
List of Abbreviations ...................................................................................... xv
Abstract ........................................................................................................ xvii
Chapter 1 Introduction................................................................................ 1
1.1 Main Goals of the Dissertation ..................................................... 1
1.2 Introduction to Sierra Juárez Zapotec ........................................... 5
1.2.1 Zapotec Languages .......................................................... 5
1.2.2 Previous Work on SJZ and other Zapotec Languages........ 5
1.2.3 Speakers of SJZ ................................................................ 9
1.2.4 The Data and the Speakers................................................ 9
1.2.5 SJZ Inventory ................................................................. 12
1.2.6 Vowels ........................................................................... 12
1.2.7 Stress and Syllable Type................................................. 14
1.2.8 Lexical Tone................................................................... 15
1.2.9 Pronouns and the Structure of Verbs............................... 19
1.3 Introduction to Articulatory Phonology and Optimality Theory.. 28
1.3.1 Optimality Theory .......................................................... 28
1.3.2 Intro to AP and Coupling................................................ 32
1.3.3 Tone Gestures................................................................. 41
1.3.4 Deactivation of Tones..................................................... 46
1.4 Chapter Overviews..................................................................... 48
1.4.1 Chapter 2 Overview........................................................ 48
1.4.2 Chapter 3 Overview........................................................ 49
1.4.3 Chapter 4 Overview........................................................ 50
1.4.4 Appendix........................................................................ 51
1.5 Conclusion ................................................................................. 52
Chapter 2 Tonal Data from Sierra Juárez Zapotec................................. 53
2.1 Introduction................................................................................ 53
2.2 First-person Singular Clitic......................................................... 56
2.3 Third-person Singular Non-formal ............................................. 66
vi
2.4 The Floating H of the First-person Singular................................ 69
2.4.1 Toneless Aspect Markers in Other Verb Classes ............. 77
2.4.2 Contour Tones as Result of Floating H ........................... 87
2.4.3 No Contour on Short Vowel ..........................................104
2.4.4 No Change When Root is Already H .............................117
2.4.5 Toneless Causative Marker /di/......................................120
2.4.6 Verbs with Incorporated Nouns .....................................126
2.4.7 Summary ......................................................................136
2.5 Spreading tones in SJZ ..............................................................137
2.5.1 Rightward Spreading of Verb-final H onto Clitics..........138
2.5.2 Clitics Whose Underlying Tone is H..............................144
2.5.3 The Clitic /é/..................................................................148
2.5.4 L Tones do not Spread onto Clitics ................................152
2.5.5 Potential H Tones Spread onto Toneless Syllables .........156
2.5.6 Glottal Stops Block Tone Spreading ..............................162
2.5.7 Summary .......................................................................165
2.6 Conclusion ................................................................................166
Chapter 3 Analysis of Floating Tone and Tone Spreading .....................169
3.1 Introduction...............................................................................169
3.2 Constraints for the Floating Tone ..............................................170
3.3 Falling Contour Tone as the Result of the Floating H in SJZ .....199
3.4 Floating Tone Appears on Toneless Aspect Markers in SJZ ......212
3.4.1 Main Constraint Rankings Summary .............................218
3.4.2 Tone Deactivation: Preview...........................................220
3.5 Tone Spreading in SJZ ..............................................................234
3.5.1 Glottal Stops Block Rightward Tone Spreading .............246
3.6 Macuiltianguis Zapotec Tones...................................................251
3.6.1 Tone Spreading in MacZ................................................251
3.6.2 Floating Tone in MacZ ..................................................264
3.7 Conclusion ................................................................................271
3.8 Summary of Constraint Definitions ...........................................275
Chapter 4 Case Studies and Future Directions........................................277
4.1 Introduction...............................................................................277
4.2 Typology of Deactivation Constraints .......................................278
4.3 Case Studies..............................................................................294
4.3.1 Margi.............................................................................295
4.3.2 Chilungu........................................................................303
4.3.3 Chinese Neutral Tone ....................................................306
4.3.4 Lithuanian Pitch Accent.................................................311
4.4 Autosegmental Phonology.........................................................314
4.5 Peak Delay Theory and Optimal Domains Theory.....................328
4.5.1 Optimal Domains Theory...............................................328
vii
4.6 Future Work..............................................................................333
4.6.1 Peak Delay Theory .......................................................338
4.7 Summary...................................................................................344
References ......................................................................................................347
Appendix: SJZ Verb Reference ...................................................................356
viii
List of Tables
Table 1.1 SJZ Phonemes and Orthography................................................. 13
Table 1.2 Verb Classes and Aspect Marker Allomorphs............................. 21
Table 1.3 Dependent pronouns which cliticize to verbs in SJZ ................... 23
Table 1.4 Dependent pronouns used by DP................................................ 24
Table 1.5 Independent pronouns in SJZ ..................................................... 27
Table 1.6 Constraint X >> Constraint Y >> Constraint Z ........................... 31
Table 2.1 Placement of the Floating H Tone .............................................137
Table 3.1 PRES-COUP-IO ..........................................................................174
Table 3.2 PRES-COUP-IO >> TONE-TO-STRESS .........................................180
Table 3.3 MAX-T >> TONE-TO-STRESS .....................................................182
Table 3.4 PRES-COUP-IO >> ACTIVE-INITIAL-SYLL...................................184
Table 3.5 ACTIVE-INITIAL-SYLL and toneless initial syllable .....................185
Table 3.6 *MULTI-TONE-SHORT-VOWEL >> MAX-T ................................189
Table 3.7 COUPLE-TO-PROM-SYLL >> MAX-T ..........................................190
Table 3.8 MAX-T >> * MULTI-TONE ........................................................191
Table 3.9 Floating H replaces underlying L tone on verb root in SJZ ........192
Table 3.10 Floating H replaces underlying H tone on stressed syllable
of verb root ...............................................................................194
Table 3.11 No Contour on open syllable with short vowel .........................196
Table 3.12 Contour forms when stressed syllable of verb root has long
vowel .......................................................................................198
Table 3.13 PRES-COUP-IO >> *C-T, *T-T >> *V-T....................................205
Table 3.14 Lexical contour tone is unchanged due to PRES-COUP-IO ..........206
ix
Table 3.15 Max-T >> *C-T, *T-T >> *V-T ................................................207
Table 3.16 *C-T, *T-T >> *V-T ................................................................208
Table 3.17 MAX-T >> ACTIVE-INITIAL-SYLL .............................................212
Table 3.18 Floating H appears on toneless aspect marker............................214
Table 3.19 Mid-toned aspect marker and long vowel on verb root ..............217
Table 3.20 Tone Deactivation Constraints and evaluation ...........................224
Table 3.21 Floating H tone goes to stressed syllable and aspect marker
spreads rightward .....................................................................227
Table 3.22 Tone of aspect marker spreads rightward in SJZ .......................230
Table 3.23 High-toned aspect marker spreads onto toneless syllable ...........232
Table 3.24 Tone of aspect marker spreads rightward due to effects of
*SELF-DEACT ........................................................................... 235
Table 3.25 Verb-final H tone spreads onto toneless clitic ...........................236
Table 3.26 Verb-final L tone spreads onto toneless clitic ............................238
Table 3.27 M tone spreads onto a toneless syllable .....................................239
Table 3.28 Verb-final H tone spreads onto clitic .........................................241
Table 3.29 Verb-final tone spread and deactivation.....................................243
Table 3.30 Tone gesture of aspect marker does not spread onto root ...........245
Table 3.31 H tone spread blocked by glottal stop ........................................247
Table 3.32 Clitic does not undergo spreading from verb root ending
with glottal stop ........................................................................248
Table 3.33 Potential H tone spreads rightward onto root in MacZ ...............255
Table 3.34 Potential H tone spreads rightward to delete tone of root
in MacZ ....................................................................................257
Table 3.35 Rightward L tone spread onto toneless syllable in MacZ ...........262
Table 3.36 Rightward H spread onto toneless syllable in MacZ ..................263
x
Table 3.37 Potential and habitual forms for 1s and 3s in MacZ ...................265
Table 3.38 Floating H appears on stressed syllable in MacZ .......................266
Table 3.39 Floating tone forms a falling contour on stressed syllable
in MacZ ....................................................................................269
Table 4.1 Input/Output candidates for OTSoft ..........................................281
Table 4.2 Input/Output candidates for OTSoft grammar 2 ........................283
Table 4.3 Input/Output candidates for OTSoft grammar 3 ........................284
Table 4.4 Input/Output candidates for OTSoft grammar 4 ........................286
Table 4.5 Input/Output candidates for OTSoft grammar 1 with DEP-T ....288
Table 4.6 Input/Output candidates for OTSoft grammar 2 with DEP-T ......289
Table 4.7 Input/Output candidates for OTSoft grammar 3 with DEP-T .....291
Table 4.8 Grammars discovered by OTSoft .............................................293
Table 4.9 H verb with L suffix in Margi ...................................................298
Table 4.10 Toneless suffix attaches to toned root in Margi..........................299
Table 4.11 Verb root takes tone of suffix in Margi......................................301
Table 4.12 Verb root does not take suffix tone in Margi..............................302
Table 4.13 H tone spread in Chilungu.........................................................305
Table 4.14 Mandarin Neutral Tone Syllables ..............................................309
Table 4.15 L tone realization of toneless syllable in Weinan Chinese..........310
Table 4.16 Lithuanian pitch accent..............................................................313
Table 4.17 WSA and BA constraints in ODT..............................................330
xi
List of Figures
Figure 1.1 Otomanguean Stock .................................................................. 6
Figure 1.2 Basic Overview of OT................................................................ 30
Figure 1.3 Gestural score for syllable /pan/ ................................................. 34
Figure 1.4 In-phase and anti-phase clocks for “mad” .................................. 35
Figure 1.5 CVC coupling relationships........................................................ 38
Figure 1.6 C-center effect............................................................................ 39
Figure 1.7 From coupling graph to output speech........................................ 40
Figure 1.8 Mandarin Coupling Graph of Tone 1, 2, and 3........................... 42
Figure 1.9 Mandaring Coupling Graph of Tone 4 ....................................... 42
Figure 1.10 Mandarin Tone Gestures (Gao, 2008) ........................................ 43
Figure 2.1 [ka!"bi#=a#$].................................................................................. 58
Figure 2.2 Spectrogram of [ka!"bi#=a#$] ....................................................... 59
Figure 2.3 [wá$ní=á$]................................................................................ 61
Figure 2.4 [í-látsu#$=a#$].............................................................................. 65
Figure 2.5 Spectrogram of [í-látsu#$=a#$].................................................... 65
Figure 2.6 [be%-&átta%=$]............................................................................... 71
Figure 2.7 [be%-&a%tta%=lu%$]............................................................................ 72
Figure 2.8 [rí-látsu#$=a#$] ............................................................................ 79
Figure 2.9 Spectrogram of [rí-látsu#$=a#$] .................................................. 80
Figure 2.10 [ri-la%tsu#$=lu%$] ........................................................................... 81
xii
Figure 2.11 [u%-ta!"'i#=a#$] ............................................................................... 88
Figure 2.12 [u%-ta%"'i#=lu%$] .............................................................................. 89
Figure 2.13 [ga!"'i#=lu%$]................................................................................. 91
Figure 2.14 [ga!"'i#=a#$].................................................................................. 92
Figure 2.15 [gú-dí-ba!"gá$]............................................................................. 94
Figure 2.16 [gú-dí-ba#"gá$=lu%$]..................................................................... 94
Figure 2.17 [be%-di%-ba!"gá$] ............................................................................. 96
Figure 2.18 [be%-di%-ba#"gá$=lu%$]..................................................................... 96
Figure 2.19 [rú-dí-ba!"gá$].............................................................................. 98
Figure 2.20 [ru-di-ba#"gá$=lu%$] ..................................................................... 98
Figure 2.21 Spectrogram of [ru-di-ba#"gá$=lu%$] ........................................... 99
Figure 2.22 [be%-di%-bí #"tsi%=a#$] ......................................................................102
Figure 2.23 Spectrogram of [be%-di%-bí #"tsi%=a#$] .......................................... 102
Figure 2.24 [be%-di%-bi"%tsi%=lu%$] ......................................................................103
Figure 2.25 [gú-dét(u%=lu%$]..........................................................................115
Figure 2.26 [é-júnna%=$] ..............................................................................119
Figure 2.27 [é-júnna%=lu%$] ...........................................................................120
Figure 2.28 [be%-día#-loá=$] ..........................................................................132
Figure 2.29 [be%-di#a#-lo!"=lu%$] ........................................................................133
Figure 2.30 Spectrogram of [be%-di#a#-lo!"=lu%$]............................................. 133
xiii
Figure 2.31 [wá$ní=bí] ...............................................................................145
Figure 2.32 [í-bí&i%=lé].................................................................................147
Figure 2.33 [tá-ne#llé=lé].............................................................................147
Figure 2.34 [u%-ta##"bi#=é]................................................................................149
Figure 2.35 [u%-ta%"ní=e]................................................................................151
Figure 2.36 [gú-di#a#=tu%$].............................................................................153
Figure 2.37 [í-bí&i%=lu%$]...............................................................................159
Figure 2.38 [ri-bi&i%=lu%$] .............................................................................161
Figure 2.39 [be%-ni%bíá$=tu%$] ........................................................................163
Figure 2.40 [gá$=a)] ....................................................................................164
Figure 3.1 Coupling of [babá] and [bába] ...................................................171
Figure 3.2 Coupling between C, V, and T in Mandarin Tones 1, 2
and 3 .........................................................................................200
Figure 3.3 Mandarin rising tone .................................................................201
Figure 3.4 Mandarin Tone 4.......................................................................201
Figure 3.5 Proposed coupling graph for SJZ lexical rising tone ..................202
Figure 3.6 Proposed coupling graph for SJZ lexical falling tone.................202
Figure 3.7 SJZ falling contour as result of floating H .................................203
Figure 3.8 Lexical H tone gesture in SJZ....................................................210
Figure 3.9 Floating H tone gesture coupled in syllable with no lexical
tone...........................................................................................210
Figure 3.10 Constraint lattice for SJZ floating tone ......................................220
Figure 3.11 Constraint lattice for SJZ tone deactivation................................250
xiv
Figure 3.12 Constraint lattice for MacZ........................................................259
Figure 3.13 Constraint lattice for MacZ tone spreading ................................250
Figure 3.14 Constraint lattice for MacZ floating tone ...................................270
Figure 4.1 Sample Grammar 1....................................................................282
Figure 4.2 Sample Grammar 2....................................................................299
Figure 4.3 Sample Grammar 3 ...................................................................285
Figure 4.4 Sample Grammar 4 ...................................................................286
Figure 4.5 Grammar 1 with DEP-T .............................................................288
Figure 4.6 Grammar 2 with DEP-T .............................................................290
Figure 4.7 Grammar 3 with DEP-T .............................................................291
Figure 4.8 Ranking Lattice for Margi ........................................................303
Figure 4.9 Toneless syllables in Mandarin..................................................307
Figure 4.10 Autosegmental representation....................................................316
Figure 4.11 Change in association................................................................317
Figure 4.12 Vowel deletion and tone reassociation.......................................318
Figure 4.13 Multiple association ..................................................................318
Figure 4.14 Many-to-one..............................................................................319
Figure 4.15 Well-formedness and violations ................................................320
Figure 4.16 More syllables than tones ..........................................................322
xv
List of Abbreviations
1,2,3 first, second, third person
AP Articulatory Phonology
AZ Atepec Zapotec
C Consonant (gesture)
COM completive
EXCL exclusive
FAM familiar
FORM formal
GLO Glottis
H high tone
HAB habitual
INCL inclusive
IPA International Phonetic Alphabet
L low tone
M mid tone
MacZ Macuiltianguis Zapotec
NFML non-formal
OT Optimality Theory
P plural
POT potential
S singular
xvi
SJZ Sierra Juárez Zapotec
sthg something
T Tone (gesture)
TB Tongue Body
TT Tongue Tip
V Vowel (gesture)
= clitic boundary
- affix boundary, bound root
. separates complex gloss
/ fused morphemes
xvii
Abstract
This dissertation examines floating tones and tone gesture deactivation in Sierra
Juárez Zapotec (SJZ), and provides an Optimality Theoretic account of tonal spreading
and placement using insights from Articulatory Phonology. While the data portion of the
dissertation is drawn from SJZ, the approach has broader implications for theories of
tonal representation and analysis.
Constraints that prohibit tone deactivation except in specific locations such as at
an utterance boundary, at a following anti-phase coupled tone, or with the vowel of a
stressed syllable are proposed. Constraints that reference the coupling point (onset) of a
tone are proposed which can draw a tone gesture to couple to a prominent position such
as the initial syllable. Constraints are also proposed which prohibit the overlap of
activation of tone gestures with other vowel or consonant gestures. In this way, the
dissertation proposes that both the onset and endpoint of a tone may be controlled via
constraints in the grammar.
Fieldwork by the author is presented which illustrates that in SJZ the first-person
singular form of transitive verbs has a high tone that does not appear on the first-person
singular suffix. Instead, this high tone appears on the stressed syllable of the verb root,
and, in cases where the aspect marker is toneless, on the aspect marker as well as the
stressed syllable of the verb root. Because the tone appears in locations separate from the
first-person singular suffix, it is called a floating tone. The constraints proposed in the
dissertation account for the placement of this floating tone while making testable
predictions for other languages.
xviii
The proposed constraints suggest that tone spreading should normally be
rightward due to the unique deactivation properties that tone gestures possess (as distinct
from consonant or vowel gestures). Leftward spreading is predicted to occur only when a
tone is drawn to couple to a prominent position such as with the vowel of the initial
syllable. In cases where a tone changes its coupling (onset) and then spreads rightward,
the output pattern will appear to be bidirectional. In cases where the onset of a tone does
not move, spreading is predicted to be rightward until the tone encounters a deactivator.
The predictions made by the constraints are tested via OT analyses of other languages.
Several areas where these constraints could be applied in future research are discussed.
1
Chapter 1 Introduction
In this dissertation I present new data on tone patterns in Sierra Juárez Zapotec
and propose novel constraints that govern deactivation in tone gestures. I propose a new
definition of floating tones based on their gestural representation. I examine a process in
Sierra Juárez Zapotec where a floating tone is assigned to the first-person singular form
of transitive verbs. I also examine processes of rightward tone spreading in the language
and argue that the pattern of rightward spreading onto toneless syllables emerges due to
the way that tone gestures are deactivated. The deactivation constraints I propose in this
dissertation account for the tone patterns I present and predict that tone spreading should
be iterative in nature where it occurs. This work raises several new areas of research due
to the rich representations that are predicted by a coupling model of tone gestures.
1.1 Main goals of dissertation
This dissertation has three main goals. One is to make a contribution to what is
known about Zapotec languages. Although the language I discuss here has been
documented via a dictionary and a grammar, less work has been done on its tone sandhi
processes. My work provides data that are important for the theoretical issues
surrounding tones that I study in this work, and also for scholars who work with tone
languages or specifically with Zapotec. The other two goals of this dissertation are to
make a theoretical contribution that expands the notion of what it means to be a floating
tone and to propose phonological constraints that operate to control patterns of gestural
deactivation in tones. My work is an approach that expands upon the traditional
2
autosegmental representation of tone. I argue that while autosegmental representations
capture the mobility of tones, they do not provide an explanation for the perseveratory
nature of tone spreading. My approach accounts for the perseveratory nature of tone
spreading by proposing deactivation constraints that operate on tone gestures. This
approach obtains the asymmetry of tone spreading, which, all else being equal, tends to
be rightward rather than leftward. The activation typology of gestures and their
deactivation are most important to my proposal. These elements of Articulatory
Phonology (Browman and Goldstein 1992) allow for the novel analysis that I present.
In chapter 2 I present several processes of tonal behavior that were discovered as
a result of fieldwork with a native speaker of Sierra Juárez Zapotec (SJZ) from the town
of San Juan Atepec. More specifically, I examine a high (H) tone which is associated
with the first-person singular inflection. This floating H tone interacts in interesting ways
with the underlying verb root and prefixal aspect markers. The floating H tone appears
only in transitive verbs (those that may take an object) in the database of 43 verb
paradigms which were elicited and transcribed as part of this study.
1
To illustrate,
consider the examples in data set (1):
(1a) be% - &a%tta% (1b) be%-&átta%=$
COM-iron COM-iron=1s
‘ironed’ ‘I ironed’
(1c) be%-&a%tta%=lu%$ (1d) be%-&a%tta%=bí
COM-iron=2s.INF COM-iron=3s.FAM
‘You ironed’ ‘He ironed’
1
The appendix contains the full database with transcriptions for all verbs, aspect markers,
and enclitic pronouns.
3
Examples (1a) through (1d) illustrate that the first-person singular form is marked not
only with a segmental clitic, but also with a H tone that appears in example (1b) on the
first syllable of the verb root. All of the enclitics here mark subject agreement, and the
other subjects lack this H tone. In chapter 2 I present numerous examples of the floating
H tone and discuss the factors that affect its placement, such as the underlying tone(s) of
the aspect marker prefix and verb root. I show that when the aspect marker is toneless,
the floating H tone of the first-person singular appears in that location (word-initial) as
well as on the stressed syllable of the verb root. In cases where the aspect marker already
has a tone, the floating H just appears on the stressed syllable of the verb root.
There are also processes of rightward spreading of tones in SJZ. Verb-final H
tones spread rightward onto clitics regardless of whether these clitics have an underlying
tone. Tones also spread rightward onto syllables that have no underlying tone. An
example is shown in (2):
(2) í-bi&i%=lu%$ í- bí&i%=lu%$
POT-fall.down=2sg.INF POT-fall.down=2sg.INF
‘you will fall down’ ‘you will fall down’
In example (2), the toneless initial syllable of the verb root becomes H when a H-toned
aspect marker attaches to the root. In chapter 3, I present an analysis of the SJZ data in
which I propose that tones are articulatory gestures whose deactivation is controlled by
the phonological grammar. I argue that tones spread rightward in SJZ because they are
deactivated only at the onset of a following anti-phase coupled tone or by the end of the
utterance. My approach explains why leftward spreading would not occur unless driven
4
by a positional target, because under my approach tones are deactivated at the onset of a
following anti-phase coupled tone, by the end of an utterance, or, in some languages via
their own articulatory clock. Tones could couple to a positional target (such as the initial
syllable) and then remain active, however. Coupling controls when a tone gesture
becomes activated, so early activation could only happen in cases where the coupling
relationship was moved earlier, such as if tones coupled to a positional target. In this way
my approach explains leftward spreading only under certain conditions but otherwise
predicts rightward spreading to be the norm.
In chapter 4, I give a factorial typology of the new deactivation constraints I
propose and highlight four case studies which illustrate how different rankings of these
constraints can give rise to different grammars. I also compare and contrast my approach
with previous analyses of tone. I conclude chapter 4 with future research topics that this
dissertation has brought to light.
This dissertation has a dual function in that it brings to light new data and then
uses these data to propose a novel analysis. The analysis is based upon the theoretical
paradigms of Optimality Theory (Prince and Smolensky 1993) and Articulatory
Phonology (Browman and Goldstein 1992). In this approach, I assume the phonological
grammar consists of a set of strictly ranked but violable constraints, and that these
constraints can make reference to the gestural composition of input and output
candidates.
Before moving to the data itself or to the analysis, it is necessary to gain insight
into the language that has formed the basis for my ideas about tone spreading and tone
5
deactivation. In the next section I present background information about SJZ, its
speakers, and the manner in which the data of chapter 2 were elicited.
1.2 Introduction to Sierra Juárez Zapotec
This section presents an introduction to SJZ. First, I discuss the Zapotec language
family and show how the language I discuss in this dissertation is related to other Zapotec
languages and the Otomangean language family as a whole. I give an overview of the
language as it has been described and analyzed previously. I discuss the phonemic and
tonal inventory of the language as well as the pronoun system. Finally, I preview a
discussion of the floating tone which has been reported previously for SJZ (Bickmore and
Broadwell 1998) and for Macuiltianguis Zapotec (Broadwell 2000). The floating tone of
the first-person singular is discussed in detail in chapter 2 and is one of the central foci of
the analysis in chapter 3.
1.2.1 Zapotec Languages
Although referred to as SJZ in this dissertation, the work presented here examines
data gathered from a speaker of Atepec Zapotec, which is the Zapotec spoken in the town
of San Juan Atepec. This section will describe SJZ’s relation to other Zapotec and
Otomanguean languages and clarify the terms that I use throughout the rest of the
dissertation.
Both Atepec Zapotec (AZ) and Macuiltianguis Zapotec (MacZ) are often referred
to as Sierra Juárez Zapotec because they are spoken in the Sierra Juárez region of
6
Oaxaca, Mexico. Both belong to the Zapotecan family of languages, which in turn
belongs to the Otomanguean stock that consists of over 174 languages and more than
seven language families (Gordon 2005). Figure 1.1 illustrates the different language
families of Otomanguean stock:
Figure 1.1 Otomanguean Languages
Amzugo(3) Chiapaneco Chorotega Chinantec Malatzinca Mazahua Otomí (9) Chicimeca- Pame
(14) (2) (2) Jonaz (3)
Matlatzincan Otomian Chichimec Pamean
Amuzgoan Chiapanec-Mangue Chinantecan Otopamean
Oto-Manguean (174)
Popolocan Zapotecan Mixtecan
Chocho-Popolocan Ixcatecan Mazatecan Mixtec-Cuicatec Trique
Chochotec Popoloca Ixcatec Mazatec (8) Chatino (6) Zapotec (58) Cuicatec Mixtec (53) Trique
(7) (3)
As illustrated in Figure 1.1, Zapotec itself is divided into 58 languages (Gordon 2005).
These languages are often divided into northern, eastern, western, southern, and central
branches (Foreman 2006: 8). Because they are spoken in the Sierra Juárez mountain
range, both Atepec Zapotec and Macuiltianguis Zapotec belong to the northern branch of
Zapotec. The division between language and dialect is less clear for these two varieties,
however. Foreman (2006) describes MacZ as belonging to a dialect chain, where the
7
Zapotec of the region varies from town to town and where mutual intelligibility decreases
across greater distances. He also points out that MacZ speakers he has worked with report
being able to understand the Zapotec of people from Atepec, as well as the nearby towns
of Abejones, Analco, and Jaltianguis. While the dictionary written by Nellis and Nellis
(1983) refers to the language I present in this dissertation as Sierra Juárez Zapotec and
while I refer to it throughout as SJZ, it is important to point out that MacZ and Atepec
can both be considered varieties of SJZ. Further work could help to determine all of the
differences between the two, although Foreman includes a helpful discussion of some of
the morphosyntactic differences. All future references to SJZ should be understood to
refer specifically to Atepec Zapotec for this dissertation. In agreement with Foreman’s
conclusions, I consider MacZ and Atepec to be dialects of the same language, and I
consider Zapotec varieties spoken farther away as distinct languages. Further work on the
primary differences between these variants would be useful.
1.2.2 Previous Work on SJZ and Other Zapotec Languages
SJZ has been well documented, especially compared to numerous other dialects
and languages of Zapotec which have received little or no attention in published
literature. The materials available on SJZ include an extensive dictionary produced by
Nellis and Nellis (1983), as well as an excellent grammar by Bartholomew (1983). Two
master’s theses have also been written. Marks (1976) examines the interaction between
verb morphology and tone, while Gibbs (1977) investigates discourse elements of SJZ.
Bickmore and Broadwell (1998) provide a theoretical analysis of the interaction between
8
tone and verb morphology. Foreman (2006) explains the morphsyntax of subjects in
Macuiltianguis Zapotec (MacZ). I will briefly comment on the analyses presented by
these authors in a following section.
There is also a variety of descriptive materials available on related Zapotec
languages. Some works on Northern Zapotec languages such as Yatzachi el Bajo Zapotec
include a grammar (Butler 1980) and dictionary (Butler 1997). For Zoogocho Zapotec
there is both a dictionary by Long and Cruz (1999) and grammatical description of the
language by Sonnenschein (2004). For Yalálag Zapotec there is a book on verbs (López
and Newberg 1990) as well as an excellent dissertation on the phonetics and phonology
of the language by Avelino (2004).
For even more distantly related Zapotec languages, the following materials are
available. Córdova’s dictionary (1987 [1578b]) and grammar (1886 [1587a]), which
document Valley Zapotec, represent the earliest documentation of a Zapotec language.
More recent publications include a dictionary by Stubblefield and Stubblefield (1990)
and grammars (Briggs 1961, Stubblefield and Hollenbach 1991) of Mitla Zapotec.
Finally, San Lucas Quiaviní has an excellent dictionary (Munro and Lopez et al. 1999) as
well as two dissertations, one by Lee (1999), and the other by Galant (1998).
For Southern Zapotec languages, there are two main sources. One is a generative
analysis of syntax in Quiegolani Zapotec by Black (2000), and the other is a grammar of
Coatlán Loxicha Zapotec by Beam de Azcona (2004). For the eastern group, Isthmus
Zapotec is documented via articles, a dictionary (Pickett et al. 1959) and a popular
grammar (Pickett et al. 1998).
9
While there may be more than 50 Zapotec languages, not all of these have been
documented. As should be obvious from the list above, SJZ is well documented in the
sense that there is both a dictionary and a grammar available for the language.
1.2.3 Speakers of SJZ
As introduced in section 1.2.1, SJZ is an Otomanguean language spoken in the
town of San Juan Atepec in the Ixtlán district of Oaxaca, Mexico. The town is located at
a latitude of 17° 26” and a longitude of 96° 33”. Nellis and Nellis (1983), the authors of
the only dictionary written for this language, give a population estimate of 5,000. On the
other hand, a more recent publication by Bickmore and Broadwell (1998) states that there
are about 10,000 speakers. The Ethnologue (Gordon 2005), gives a population estimate
of 4,000 with 150 monolingual speakers for the Sierra Juárez variety. By contrast, the
speaker who provided all of the data for chapter 2 estimates that there are only about
1,600 fluent speakers remaining in the town of Atepec, although he noted that there is a
large expatriate community living in the United States, many of whose members still
speak or at least passively understand the language. (Donaldo Pérez, personal
communication). This suggests there are somewhere between 2,000 and 5,000 remaining
speakers of SJZ.
1.2.4 The Data and Speakers
All of my fieldwork has taken place with speakers living in Los Angeles. I have
primarily worked with one native speaker: Donaldo Pérez (DP). DP acquired Spanish at
10
the age of 10 before which time he was a monolingual Zapotec speaker, and he regularly
speaks Zapotec with friends living in Los Angeles. The data in this dissertation reflect his
idiolect in the sense that the recordings were made only of his speech. Occasionally, I
verified data with other Zapotec speakers living in Los Angeles including Graciela
Bautista and Azarael Pérez, both of whom acquired Zapotec as their first language before
learning Spanish. Their speech matched DP’s, so the transcriptions I give in chapter 2
may be considered composites which represent the speech of three individuals, although
the pitch contours I show are only from the recordings made with DP.
My data were collected using standard linguistic elicitation. The language of
elicitation was always Spanish since the speakers with which I worked are not
particularly comfortable with English. Two of the three have at least passive knowledge
of English, however, meaning that they can understand and use it to some extent,
although their preferred languages are Spanish and Zapotec. In fieldwork sessions, I
asked the speakers to verify whether certain Zapotec verb forms were correct or not. I
first obtained these examples by selecting verb roots from the Nellis and Nellis dictionary
and combining them with the appropriate aspect markers for their verb class. I associated
each verb root with its aspect marker with the help of the Bartholomew grammar (1983).
If the speakers judged the forms as correct, I proceeded to make recordings with DP. If
corrections were needed, we noted these in written form and then continued to make
recordings of the modified forms. At no time did we develop an orthography for marking
tone, as I felt this would potentially bias their pronunciation. To write the verbs in
question, we used an adapted version of the orthography used in Nellis and Nellis (1983:
11
xvi). Most of the time, the speakers agreed with the forms given in Nellis and Nellis
(1983) and Bartholomew (1983). Only minor changes were ever needed.
In this chapter and in all following examples, I generally cite data as shown in
examples (3) and (4), where the underlined syllable is stressed:
(3) ú-ká$ná=$ (4) ú-ka#$ná=lú$
POT-leave.sthg=1s POT-leave.sthg=2s
‘I will leave it’ ‘You will leave it’
The first line is the phonetic transcription in International Phonetic Alphabet (IPA).
Throughout this dissertation, geminate consonants are written with two symbols, while
long vowels are written with “ ː ” after the vowel. In the next two lines the example is
broken down into its component morphemes with an English gloss underneath. Bound
morphemes are separated by an “ = ” if they are clitics, and by a “ –" if they are affixes or
bound roots. Fused morphemes,
2
which are typically verbs and their aspectual prefixes,
are shown with a “ / ” in the gloss. When the translation of a morpheme requires more
than one word in English, I have placed a period between the two words of the complex
gloss. In the examples shown above, the verb was translated by the native speaker as
“dejar” (Spanish) which in this case translates to English as ‘leave (something)’. Thus I
have used two English words in the gloss for this example.
2
By fused morphemes I refer to aspectual prefixes that consist only of a consonant or a
consonant plus a tone that appear with a vowel-initial verb root. These types of fused
morphemes have been identified as fused by scholars like Foreman (2006) for
Macuiltianguis Zapotec.
12
1.2.5 SJZ Inventory
In this section I briefly introduce the phoneme inventory of SJZ. All the data I cite
in this section are from Bartholomew (1983). The fieldwork I conducted did not uncover
significant differences between the speakers who worked with me and the descriptions
given in Nellis and Nellis (1983) or Bartholomew (1983). Some lexical items are slightly
different in their composition, and on occasion lexical items differed in their tones, but
the phonemic and tonal inventory reported by these two sources match the consultants’
pronunciation. Furthermore, the focus of my recordings was not to analyze or transcribe
the segmental inventory so much as it was to observe the presence and relationship of
tones. Interested readers are referred to Bartholomew (1983) for an exhaustive discussion
of the segmental inventory of the language.
1.2.6 Vowels
The vowel inventory of SJZ includes /a/, /e/, /i/ /o/, /u/, as well as nasalized
versions of each: /a)/, /e)/, /i)/, /o)/, /u)/. The following diphthongs also exist: /ia/, /iu/, /ie/,
/ua/, /ue/, /ui/. The chart in Table 1.1 shows the consonant inventory in IPA as given in
Bartholomew. The orthographic symbols used in the Nellis and Nellis dictionary (1983)
to represent these contrasts is given in parentheses next to the IPA symbol. When I cite
my own data in examples throughout the dissertation, I use IPA and not the Zapotec
orthography developed in Nellis and Nellis, with the exception that I write geminates
using two symbols: /kk/, /tt/. Aside from this, all markings are standard IPA.
13
Table 1.1 SJZ phonemes and orthography
Bilabial Labio-
dental
Inter-
dental
Dental Alveolar Retro-
flex
Palatal Velar Labio-
velar
Glo-
ttal
Long vcls.stop p"(pp)
t*" (tt)
k"
(cc,qqu)
k
w
" (ccu)
Short vcls.stop p (p)
t* (t)
k
(c,qu)
k
w
(cu)
$ (')
Short voiced
stop
b (b)
d (d)
g
(g/gu)
Long vcls.
affricate
t*s*" (tts) t(" (ch.)
Short vcls.
affricate
t*s* (ts) t( (ch)
Long vcls.
fricative
+"
(th.)
&" (xx)
Short vcls.
fricative
f +
(th)
s* (s)
& (x)
x (j)
Voiced fricative
v
' (y))
Long vcd. Nasal m"(mm)
n*" (nn)
Short vd. nasal M
n*
, (ñ)
Voiced tap
- (r)
Voiced trill
r (rr)
Long vd.
Lateral approx
l*" (l.l)
Short vd.
Lateral approx
l* (l)
Long vd. approx
j" (yy)
Short vd.
approx
w (hu)
j (y)
Table 1.1 shows that there are both single and geminate varieties of stops, affricates, and
glides. I refer to these as “short” and “long” in Table 1.1. These consonants are normally
referred to as lenis vs. fortis in the Zapotec literature. Future work on SJZ could
investigate the phonetic properties of this distinction.
14
1.2.7 Stress and Syllable Type
There are three syllable types in SJZ: CV, CV$, and CV$V.
3
Each word has a
stressed syllable, and a compound word may have two stressed syllables. The main
prosodic stress usually falls on the first syllable of the root (Bartholomew 1983: 339).
Verb prefixes are not stressed, however. If the verb root begins with a vowel, the aspect
marker (prefix) combines with the verb root and the primary stress is then on the first
syllable which combines the prefix and root material. As will be evident in the data in
chapter 2, and in the appendix to the dissertation, this process happens frequently in
certain verb classes, where the aspect marker consists of a consonant which then
combines with a vowel-initial root.
Lexical items that are compound words are stressed on the first syllable of each
root; for this reason they have as many main stresses as there are roots in each compound
word. Some examples are below in (5), where the stressed syllables have been underlined
(Bartholomew 1983: 339):
(5)
[etta &tílá] ‘bread’ (Castilian tortilla, &tílá =Castilla)
[ja$a &tílá] ‘soap’ (Castilian soap Lily)
[tsu$ú latsi$] ‘to like’ (be in the heart)
[kká latsi$] ‘to wish for’ (where latsi$= heart or being)
3
In later examples I have transcribed geminate consonants (due to their length), but leave
aside the question of whether the language described here has geminates or a lenis/fortis
distinction. Previous studies have described the contrast as lenis vs. fortis, with fortis
consonants being longer than lenis (Bartholomew 1983). If the consonants in question
were indeed moraic geminates, there could be consequences for the range of syllable
types in the language.
15
In the data section of chapter 2 and in all other examples of data I have collected, the
stressed syllable is underlined. As I show in chapter 2, all of the verb roots I recorded had
their primary stress on the first syllable. This was evident due to slight lengthening of the
stressed syllable compared to the unstressed syllable.
4
As illustrated in the SJZ dictionary
(Nellis and Nellis (1983), there are verb roots with non-initial stress. I underline the
stressed syllable in all examples in this dissertation to illustrate its placement but do not
make the claim that stress placement is predictable. As shown by Nellis and Nellis (1983)
some roots have stress elsewhere. It would be valuable in the future to investigate the
realization of the floating tone in forms with non-initial stress.
1.2.8 Lexical Tone
SJZ is described in the Bartholomew grammar and the Nellis and Nellis
dictionary as having L (low), H (high), M (mid), rising, and falling lexical tones.
Although Bartholomew does not discuss them, I also propose that SJZ has syllables that
are unspecified for tone, which emerge with variable tone, depending on the surrounding
tonal environment. If toneless syllables do not otherwise acquire a tone, then they are
realized as mid. One way that the difference between unspecified and specified M tones
becomes apparent is when the floating H tone of the first-person singular either forms a
contour in the stressed syllable (when there is an underlying specified tone) or becomes a
simple H (when there is not an underlying tone affiliated with that syllable). Another way
4
Lengthening of stressed syllables is also noted by Marks (1976: 117). Future
experimental work is needed to verify this more concretely.
16
that toneless syllables become evident is that they undergo tone spreading and acquire the
tone of whatever syllable precedes them. Thus when the initial syllable of a verb root is
toneless, it presents the same tone as the aspect marker which is attached to it. These
patterns are discussed in chapter 2. Here I present examples given by Bartholomew
(1983: 340-341) showing that SJZ has H, M, L, rising and falling tones and that these are
lexically contrastive.
Bisyllabic words may have both syllables with the same tone or may have
syllables with different tones. This is illustrated in examples from data set (6):
(6) Bartholomew (1983: 340-341)
HH /íjjá/ ‘rock’
MM /i%jja%/ ‘rain’
LL / i#jja#/ ‘flower’
HL /tsíttsi#/ ‘white’
HM /jétti%a%/ ‘cane’
LM /ja#$la%/ ‘first’
LH /be#ttsí$/ ‘louse’
MH /tsu%mmí/ ‘basket’
ML /da%na#/ ‘sister (of a man)’
In addition to the three level tones, there are two contour tones: rising and falling. The
Nellis and Nellis dictionary writes these contours using two vowel symbols (although I
cite them here using the IPA symbolism for long vowels). This reflects the fact that
contour tones are twice as long as single tones – a point also noted by Marks (1976)
which is consistent with the data I recorded. As I discuss in greater detail in chapter 2,
17
there does not seem to be a difference on the in the output between LH and MH, or
between HM and HL. That is, all lexical falling tones are realized approximately HM and
all lexical rising tones are realized approximately as MH. When the floating H tone of the
first-person singular is the cause of the tonal contour, as I show in chapter 2, there does
not appear to be a difference between contours that are plausibly composed of HL versus
HM. More phonetic studies would be useful to investigate the difference between these
two types of contours, and between short and long vowels with tones. While Nellis and
Nellis (1983) identify contours with long vowels, they do not mark vowel length for
words with level tones. In some instances, I recorded and transcribed long vowels (such
as /na%"ga$/ ‘ear’) on words without contour tones. I also transcribed a long H tone in the
verb ‘to sow crops’: /rá"dá/. Because the focus of my fieldwork was on verbs, and
because the dictionary does not list vowel length, it is difficult to know how common (or
rare) long vowels may be with level tones. This question should be pursued in future
fieldwork and via experimentation because the question of long H tones plays into the
theoretical arguments I make in chapters 3 and 4 (see section 3.3). In this dissertation, I
will propose that long level tones do occur in SJZ, with the caveat that this needs to be
verified systematically in future studies.
The examples given by Bartholomew (1983: 341) to illustrate contour tones are
found in data set (7):
18
(7)
Rising tone: /tu#."/ ‘mecate’
Falling tone: /tte!"/ ‘to pass’
Falling-M /ja!"ni%/ ‘neck’
M-Rising /be%ja#."/ ‘cactus’
Some additional words which are near minimal sets are shown in data set (8):
(8)
HH /bé$já/ ‘a kind of very large fly’
M-rising /be%ja#."/ ‘cactus’
MH /be%$já/ ‘mushroom’
HL /dá$a#/ ‘corn (elote)’
rising /da#."/ ‘bean’
falling /da!"/ ‘lard’
LL /be#l"a#/ ‘snake, serpent’
HH /bél"á/ ‘fish’
MMH /be%lla%$á/ ‘ravine’
MLH /be%lla#$á/ ‘coarsely ground corn (maíz desquebrajado)’
We can see from the examples presented here that the language has lexical tones which
may be L, M, H, rising or falling. None of the data I present in chapter 2 conflict with the
information given by Bartholomew or Nellis and Nellis and for this reason I will not
delve into further description of the language here. I refer the reader to those sources for
more discussion of the segmental inventory.
19
1.2.9 Pronouns and structure of verbs
Here I will discuss the verb construction which is the focus of the rest of the
dissertation. First I discuss the verb classes, and then the independent and dependent
pronouns which either precede or attach to the verb root.
The forms which are of greatest interest and which I examine in detail in chapter
2 include an aspect marker, verb root, and a pronoun which cliticizes to the verb. First,
consider the example in (9):
(9) (la%bí) ú-ka#$ná=bí
5
3s.FAM POT-leave.sthg=3s.FAM
‘He will leave (it)’
The output form has the potential aspect marker /ú/ attached to the verb root /ka#$ná/, to
which the clitic /bí/ is then attached. All of the forms that I recorded and list in the
appendix, and all of the data I present in chapter 2 are in this format of aspect marker +
verb root + clitic. The data examples I discuss most extensively involve tone spreading
and interaction in verb phrases of this type.
6
It is possible in SJZ to have a noun subject
following the aspect marker + verb root, but since my interest was in observing changes
in the subject-marker clitics, I recorded only those forms with attached subject markers.
There are 12 verb classes identified in the Nellis and Nellis dictionary, each with
different segmental and tonal allomorphs for the potential, completive, and habitual
5
This is a transitive verb – it assumes an object is being left and was elicited in the
context ‘to leave a book behind’.
6
I leave investigation of longer phrases and their tonal patterns for future work.
20
aspect markers.
7
Bartholomew refers to these as future, preterit, and present tense
markers (futuro, preterito, y presente) (1983: 387), but I assume that these are better
described as potential, completive, and habitual aspect markers (Foreman, 2006). In
Table 1.2 I show these allomorphs along with the class names that Bartholomew has
assigned to them. In the appendix to this dissertation, I have listed these classes using a
numerical system instead, and thus I place numbers beside each classification that
Bartholomew has given here. I have also modified this chart so that the examples are
given in IPA:
7
Marks (1976: 60) identifies 5 verb classes only, but her work was published prior to the
more extensive dictionary and grammar of Nellis and Nellis (1983) and Bartholomew
(1983).
21
Table 1.2 Verb Classes and Aspect Marker Allomorphs
Class Potential Habitual Completive Example Gloss
1. IA
/gú/ /ru/ /be%/ gú-k
w
e%dá=
‘to flee’
2. IB
/w/ /ru/ /be/ wé$=
‘to hit’
3. IIA
/í/ /ri%/ /bi/ í-lá$ni=
‘to see’
4. IIB
/í/ /ri/ /gu/ í-bi&i%=
‘to fall
down’
5. IIC
/é/ /re/ /be%/ é-júnna%=
‘to return’
6. III
/ (zero) + H /ri/ /gu%/ ka!!"ni%=
‘to dig’
7. IVA
/gá/, /t(á/ /rá/ /gu/ gáda%=
‘to plant’
8. IVB
/g/ + H /r/ /gut/ gá$a%=
‘to enter’
9. VA
/ts/, /t(/ +H /r/ /w/ tsía%=
‘to go’
10. VB
/g/ + H /r/ /wi/ gá$=
‘to strike’
11. VI
/kká/ /kka/ /ukk
w
a/ kká=
‘to be’
12. VII
/tá/ /rita/ /ta%/ tá-ne%llé$=
‘to visit’
Notice that in Table 1.2, I have marked M tones with a macron, while toneless syllables
are unmarked. In the potential column, some aspect markers consist of a consonant plus a
H tone, as in verb class 8. Also, the example column only shows the verb root with the
potential aspect marker. So for row 1, which shows the aspect markers for class 1 (1A in
Bartholomew’s system), the aspect markers are /gú/ for the potential, /be%/ for the
completive, and /ru/ for the habitual. This means that the example verb root will be /gú-
k
w
e%dá=/ in the potential, to which clitics will then attach, but /be%-k
w
e%dá=/ in the
completive. The habitual will be /ru-k
w
e%dá=/, to which the clitics will be added. As will
become clearer in chapter 2, I have determined which aspect markers are underlyingly
22
toneless based on where the floating H tone of the first-person singular appears. The
Bartholomew grammar lists both M tones and what I call underlying toneless syllables as
M, and does not mark them orthographically. It is important to point out that Table 1.2
gives a general overview of the aspect markers – that is, I assume most verbs from each
class will take the aspect markers shown in Table 1.2. Slight variations that I uncovered
are found in the appendix, such as an optional /g/ before the potential aspect marker /ú/ in
some cases. The most important detail to recognize from Table 1.2 is that there are 12
verb classes, and that each class is associated with its own aspect markers that vary both
segmentally and tonally. All of the verbs in the Nellis and Nellis dictionary are listed with
their appropriate verb class.
Each verb appears with one of these aspect markers according to its class. In cases
where the verb root is vowel-initial and the aspect marker consists of just a consonant,
this consonant will attach directly to the verb stem and will become part of the initial
syllable. Bartholomew gives a very detailed description of each verb class, and includes
many illustrative examples. I refer the reader to her work (1983: 387-399) for a more
thorough overview of the segmental properties of each class. The appendix of this
dissertation includes at least two examples from each of these verb classes, and in some
instances many more.
8
As I show, the tonal properties of the aspect markers affect the
placement of the floating H of the first-person singular in important ways.
8
In my database, verb class 1 (1A) has the most members. It appears that this reflects an
overall pattern in the language where verb class 1 is in fact most common. Marks (1976:
60) also noted this fact.
23
Next, I will explain the dependent pronouns which attach to the right edge of the
verb stem. As I discuss in detail in chapter 2, these undergo the results of tone spreading
when the verb root ends in a H tone. Here I will give a brief description of the pronouns
which Bartholomew calls “attached” pronouns. Again, recall the example:
(10) (la%bí) ú-ka#$ná=bí
3s.FAM POT-leave.sthg=3s.FAM
‘He will leave something behind’
The pronoun that Bartholomew identifies as independent is /la%bí/, while the attached
(dependent) pronoun is simply /bí/. Table 1.3 lists the dependent pronouns given by
Bartholomew (1983: 346). As I will discuss below, this differs slightly from the forms
reported by the speaker DP.
Table 1.3 Dependent pronouns which cliticize to verbs in SJZ
Singular Plural
1 exclusive
=a$, =te$ =tu%$
1 inclusive
=rí$u%
2 informal
=lu%$ =lé
2 formal
=k
w
ía%$lu%$, =é = k
w
ía$lé, =lé, =ké
3 familiar
=bí =ka%bí,
3 formal
=é, =bí =ké
3 non-formal
=a)
=ka)
24
In the original source this table is named “pronombres dependientes que indican sujeto”
or dependent pronouns that indicate subject. These are the pronouns that cliticize to
verbs. In the table above, some pronouns have more than one allomorph, which I have
shown by listing all possibilities separated by a comma between each. For example, the
second-person singular formal pronoun can be either /k
w
ía%$lu%$/ or /é/. The speaker DP
prefers /é/ but noted that both are acceptable with no difference in meaning. DP also
prefers /gé/ for the second-person formal plural. In his speech, it seems that the second
and third-person formal are the same for the singular and plural. That is, he prefers to use
/é/ for the second and third-person singular formal, and to use /gé/ for the second and
third-person person plural formal. To make this clearer, I include a table that shows only
the forms that DP uses for the dependent pronouns. These are shown in Table 1.4:
Table 1.4 Dependent pronouns used by DP
Singular Plural
1 exclusive
=a$, =te$
9
=tu%$
1 inclusive
=rí$u%
2 informal
=lu%$ =lé
2 formal
=é =gé
3 familiar
=bí = ga%bí
3 formal
=é =gé
3 non-formal
=a)
=ga) #
9
The allomorph /te$/ occurs only in subject oriented stems, and only appeared twice in
my database of 43 verbs (located in the appendix).
25
The examples in Table 1.4 show that DP prefers the same pronoun for the second-and
third person formal, and also that he has voicing of the initial consonant of the pronouns
for the third-person plural. For example, where Bartholomew gives /ka%bí/ DP has /ga%bí/
instead.
It is also important to comment here on the names I assign to the different
pronouns. Bartholomew identifies these pronouns differently than I have done. For
example, she separates the second-person into familiar and respectful, and the third-
person into familiar, respectful, and impersonal. I have changed these terms slightly, in
that I refer to the second-person informal and formal, and to the third-person familiar,
formal, and non-formal.
According to the speakers, the non-formal pronoun is used as a default when
neither the third-person familiar nor the third-person formal is appropriate. Thus the
familiar pronouns are used for those of similar age or who are younger than the speaker,
while formal pronouns are used for those older than the speaker or who otherwise
deserve respect (such as people in positions of authority). The non-formal pronouns for
the third-person might be used in cases where the person being referred to is neither
much younger nor much older than the speaker, but is not a friend or acquaintance. This
may be why Bartholomew calls these pronouns “impersonal” because they are non-
specific and a kind of default pronoun for those who do not fit into any other category.
Marks (1976: 67-68) provides a helpful discussion on this issue which I will
summarize here, since it makes the division of these pronouns clearer. Marks notes that
Proto-Zapotec is posited to have had 13 pronominal categories. In Proto-Zapotec, second-
26
person singular and plural were divided into the categories familiar and reverential, while
third-person singular and plural were divided into human, animal, and inanimate. As I
have already discussed, SJZ also has 13 pronominal categories. The first and second-
person in SJZ correspond closely to the categories posited for Proto-Zapotec. In the third-
person it appears that a semantic shift has taken place. The categories “inanimate” and
“human-common” have fused with “animal”, resulting in a loss of the animal/inanimate
distinction, and with “human only” expressing a meaning above “human common” – i.e.
respect. Within the “human” category there has been a split in the expression of respect,
similar to that which has occurred in the dialect of Choapan Zapotec, creating a
distinction between respect and high respect. Marks argues that the morpheme which
expresses respect is the most common form that is used when referring to people and can
be used when referring to people with whom the speaker is on familiar terms. The
common form, which is marked by the animal indicator, is used to refer to animals,
inanimate objects, and common people. Thus Marks’ analysis is that there are three
distinctions in the third-person: high respect, respect, and common. This background
helps to make the speakers’ statements about the use of these pronouns more
understandable. Thus what I refer to as “familiar” is what Marks calls “respectful” – the
pronoun /bí/. What she calls “common” I refer to as the non-formal /a)/. Finally, what she
terms “high-respect” I refer to as “formal” - /é/.
Now, I will turn to a discussion of the independent pronouns. Table 1.5 gives the
independent pronouns as cited in Bartholomew (1983: 347). I have noted differences
between DP and the forms cited in Bartholomew via parentheses where applicable:
27
Table 1.5 Independent pronouns in SJZ
Singular Plural
1 exclusive inte#$ (nte#$) intu#$ (ntu#$)
1 inclusive rí$u%
2 informal lu%$ lébi%$i% (léi#)
2 formal k
w
ía%lu%$ k
w
ía%$lé
3 familiar la%bí la%ka%bí (la%ga%bí)
3 formal le#." la%ké (la%gé)
3 non-formal la)
la%ka) (la#ga) # )
When questioned about the second-person informal plural pronoun /lébi%$i%/, DP stated that
the older generations pronounced it that way, but that most people now use /léi#/ instead.
DP also differs from Bartholomew in that he uses a voiced velar stop for all of the plural
third-person pronouns. We also saw that this was the case for the attached, dependent
pronouns discussed earlier. In addition, DP does not have an initial vowel in the first-
person singular or plural. These differences have been indicated in Table 1.5 via
parentheses.
I will not make much reference to the independent pronouns after this
introduction, as they do not play a role in the tonal interactions which I examine in
chapter 3. It is important to point out that they are different from the attached pronouns I
showed in Table 1.4, however. Now that I presented an overview of SJZ, I move to the
theoretical paradigms upon which my analysis in chapters 3 and 4 is based.
28
1.3 Introduction to Articulatory Phonology and Optimality Theory
In this section I will explain the basic principles of Articulatory Phonology (AP)
and Optimality Theory (OT). An important aspect of this section is to explain the need
for constraints that turn off (deactivate) tone gestures. At the moment there is no theory
about how this process occurs. The theory of tone deactivation that I present in chapter 3
is based on a framework that combines the formalisms of both AP and OT. First, I will
present the basics of the OT model and then I will move to a discussion of AP. Finally, I
include a section which deals specifically with tone deactivation and the deactivation of
other types of gestures.
1.3.1 Optimality Theory
This section will give an overview of OT (Prince and Smolensky, 1993). This
dissertation is couched within OT, and the constraints make reference to parts of the AP
model as well. The following section will present the basics of AP.
The OT framework is a constraint-based model where violable constraints are
ranked to determine the optimal output forms. For the data under study in this work, the
relevant constraints will operate over a series of segments with tones, such as aspect
markers and verb roots, and their ranking will predict the optimal outcome or the final
product with respect to tone sandhi or tone perturbations. In OT, constraints are universal,
while their ranking is language-specific. This leads to the many grammars that we see in
the world’s languages. Thus the constraints I propose in chapter 3 are ranked a certain
29
way for SJZ, but they are ranked differently for other languages with different tone
patterns.
The three main components of the OT system are GEN, CON, and EVAL. Not
surprisingly, GEN is a function that generates a set of candidate outputs based on a given
input or an underlying representation. GEN produces an exhaustive set of candidates for
evaluation. These candidates can be identical to the input, slightly different, or totally
different. A restriction on GEN is that the candidates have to be composed of universal
elements like prosodic and segmental structures. An essential assumption that I make in
this work is that GEN produces candidates comprised of articulatory gestures and
coupling relationships.
CON is a set of violable and universal constraints. These constraints are
considered to be present in the grammar of all languages, but their rankings differ across
languages. Constraints belong to one of two categories: markedness and faithfulness. The
well-formedness of a candidate is decided by markedness constraints, while the
faithfulness constraints favor candidates that most closely resemble aspects of the input
(McCarthy and Prince, 1995). The interactions between markedness and faithfulness
(well-formedness vs. similarity to the input) or between markedness constraints is what
determines the output of the grammar.
The output representation and optimal candidate for each input is determined by
EVAL, which is the mechanism by which candidate outputs are evaluated against the
constraint ranking in the grammar. EVAL is the function defined by composing all of the
constraints in the order in which they are ranked (Prince 2004). That is, EVAL starts with
30
the constraint that is ranked highest (say, constraint 1) and extracts the subset of
candidates that is most favored by constraint 1. This subset is then passed along to the
next constraint in the ranking, constraint 2, whereupon the same thing happens: the subset
of candidates which is most favored by the ranking is located and the rest is discarded.
An optimal candidate is selected when all other candidates have been discarded. What
determines the best output of a grammar is the least costly violation of the constraints.
Figure 1.2 shows a schematic representation of this process (from McCarthy 2002):
Figure 1.2 Basic Overview of OT
Input GEN candidate outputs EVAL optimal output
To make this idea clearer, consider a case where CON has only three constraints, which I
will call CONSTRAINT X, CONSTRAINT Y, and CONSTRAINT Z. GEN produces an infinite
set of candidates, but I will only consider 4 of them here. I will call these Candidate a,
Candidate b, Candidate c, and Candidate d. The tableau in Table 1.6 illustrates how these
four candidates would be evaluated given the ranking of these constraints. In the tableau,
constraints are listed in their ranking order from left to right, with the highest-ranked
constraint in the left most column. When a candidate incurs a violation of a constraint it
is assigned an asterisk (*). A fatal violation, where the candidate loses out altogether, is
shown with an exclamation point following the asterisk (*!). The rightward pointing hand
indicates the winning candidate. This is illustrated in Table 1.6 below:
31
Table 1.6 Constraint X >> Constraint Y >> Constraint Z
10
/input/ Constraint X Constraint Y Constraint Z
☞ a. Candidate a *
b. Candidate b **!
c. Candidate c * *!
d. Candidate d *!
In Table 1.6, candidate d) is eliminated right away because it violates the highest-ranked
constraint. This is shown with the symbol (*!) in the table above. Candidates a) and c) tie
on CONSTRAINT Y, as each candidate has one violation of the constraint. However,
candidate c) has an additional violation of CONSTRAINT Z and is eliminated. Candidate b)
is ruled out because it has one more violation of CONSTRAINT Y than the winning
candidate. Winning candidate a) shows that it is acceptable for a winning candidate to
violate some constraints, as long as it does not incur more violations than any of the other
candidates prior to their elimination from the evaluation.
Finally, it is important to mention the principle of Richness of the Base (Prince
and Smolensky, 1993). Richness of the Base is the idea that languages cannot differ
systematically in their lexicons, which form the input to the grammar. That is, the base or
lexicon of the grammar contains many diverse forms because it is not subject to any
language-particular restrictions. All aspects of well-formedness come under the control of
EVAL and the constraint hierarchy so that systematic differences between languages are
obtained via differences in constraint ranking. It is important to point out that the theory
of Richness of the Base does not override the theory of representations. One is free to
10
The symbol “>>” means “dominates”. So Constraint X dominates Constraint Y is
written as Constraint X >> Constraint Y.
32
assume universal restrictions on inputs without violating the principle of Richness of the
Base (McCarthy 2008). As I present in later sections, the universal restrictions I propose
(candidates being comprised of coupling relationships and gestures) do not violate
Richness of the Base.
1.3.2 Introduction to AP and coupling
As I discussed in the previous section, constraints in OT have been used to assign
violations to output forms. In approaches that assume feature-based representations, these
constraints make reference to segments which have features that distinguish them
(Jacobson, Fant, and Halle 1963; Chomsky and Halle 1968; Clements 1990; Keating
1988; Gordon 2001, etc.). Some of these features can be [voice], [nasal], [palatal], etc.
Features are considered to be abstract cognitive units which can be combined in different
ways within segments to produce lexical contrast(s) in each phonological system. In this
dissertation I adopt the AP model because of the richer set of representations it allows.
For example, the gestural approach makes reference to coupling which controls the
location and onset of tone and other gestures, as well as the deactivation or endpoint of a
tone. This allows more specific predictions that the tendency of tones will be to spread
rightward due to their unique deactivation typology. As discussed in the following
paragraph, the AP model uses gestures rather than features as the primary units.
The AP model (Browman and Goldstein, 1992, 1995; Byrd 1996; Byrd and
Saltzman, 2003) is different from featural analyses. AP views the units of contrast in the
phonological system as articulatory gestures, rather than as features that are organized
33
into or associated with segments. In this system, phonological “segments” are modeled as
highly cohesive coordinations between articulatory gestures (Byrd, 1996). This means
that each ‘segment’ in the traditional sense can be comprised of several articulatory
gestures.
Each articulatory gesture in this model is a unit of information as well as a
discrete unit of constriction action that uses an organ of the vocal tract. These are the
glottis, velum, tongue tip and body, and the lips. Thus, when considered as an
information unit, the presence, location, or constriction degree of a specific gesture can
be contrastive for a particular lexical item or utterance.
As a unit of action, each gesture entails the formation and release of a constriction
in the vocal tract. Gestures are understood as a particular type of dynamical system with
specific temporal and spatial properties. Because gestures are part of a dynamical system,
they can be defined using the task dynamic model (Saltzman, 1986; Saltzman and Kelso
1987; Saltzman and Munhall, 1989). According to this model, a gesture refers to the
articulators like tongue tip or jaw. Gestures control coordination of a set of articulators to
achieve the constriction goal(s) for an organ that is represented in tract variable
coordinates. There are a total of eight tract variables in this model, and the tract variables
corresponding to oral gestures are arranged in pairs to represent the two dimensions of an
oral constriction: LP (lip protrusion) and LA (lip aperture) for LIP gesture, and CL
(constriction location) and CD (constriction degree) for TT (tongue tip) as well as TB
(tongue body) gestures. The other two tract variables are VEL (velic aperture) and GLO
(glottal aperture). These two tract variables correspond to VELUM and GLOTTIS
34
gestures respectively. These tract variables specify constriction goals in terms of the
location and degree for oral gestures and openness for VELUM and GLOTTIS gestures.
In the AP framework, gestures are the smallest unit that can compose a vowel or
consonant. For example, vowels and consonants can be combinations of several gestures
that are glued together via their coupling relations. An example might be the consonant
/n/, which is composed of both a TT(tongue tip) and a VELUM gesture. The relationship
between these two gestures (how they are coordinated) is defined by their coupling
relation. Coupling can be either in-phase, anti-phase, or another less-stable coordination
pattern. A gestural score shows the coordinated structure of gestures as a function of
time. For example, the syllable /pan/ is shown in the gestural score in Figure 1.3:
Figure 1.3 Gestural score for syllable /pan/
VELUM
TT
ORGANS TB
LIPS
GLO
This figure shows each gesture with a box; the length of the box corresponds to the
duration of the corresponding gesture. The property of duration is what sets gestures apart
from features. Note also that the gestural score is a consequence of coupling among the
gestures and as such, their gestural frequencies (relative speed and rate of oscillation of
STOP
NARROW PHARYNGEAL
STOP
WIDE
WIDE
35
the gestures) in some particular context. The coordination relations between various pairs
of gestures are what determine the timing of each gesture relative to the others. In-phase
coupling relationships are shown with solid lines, while anti-phase is shown via a dashed
line. Gestures which are coupled in-phase begin simultaneously, while gestures that are
anti-phased coupled are sequentially timed. In Figure 1.3, the tongue body (TB) and lip
(LIPS) gestures are in-phase coupled, meaning they begin simultaneously. The tongue
body (TB) and tongue tip (TT) gestures are anti-phase coupled, meaning that the TT
gesture occurs after (later in time than) the TB gesture. To better illustrate the in-phase,
anti-phase relationship, consider the diagram in Figure 1.4, which illustrates in-phase and
anti-phase clocks for the word ‘mad’:
Figure 1.4 In-phase and anti-phase clocks for “mad” (after Goldstein 2011).
36
Figure 1.4 shows a green (vertical) line and synchronized clock picture for those gestures
which are in-phase. Instead of illustrating a gestural score, this particular picture
illustrates the idea of synchronized clocks which are either in-phase (0 degrees) or anti-
phase (180 degrees opposite to each other). To form the /ma/, the velum is wide, pharynx
is wide, and the lips are closed. These gestures are all synchronous and in-phase as shown
by the green (vertical) lines and clocks with the same position. The tongue tip articulator
that achieves closed alveolar, shown by the red (45 degree) line and anti-phase clock, is
anti-phase coupled to the pharyngeal gesture. Figure 1.4 shows in a simplified way that
the gestures which comprise the traditional segments /ma/, are in-phase coupled to each
other, while the gesture which represents the coda /d/ is anti-phase coupled to the vowel.
Goldstein et al. (2006) note that consonant and vowel gestures are hypothesized to
be coordinated in either of the intinsically stable modes: in-phase (the most stable) and
anti-phase. Goldstein et al. (2006: 229) point out that syllable initial consonant gestures
and their following vowels are coordinated in-phase with one another. In Figure 1.3, then,
the TB gesture is the vowel gesture which is in-phase coupled to the syllable initial
consonant gesture (the labial gesture (LIPS)). The important thing to notice in Figure 1.3
is that what more traditional phonology identifies as /p/ + /a/ + /n/, the AP model views
as a combination of tongue-tip, labial, tongue body, glottal, and velum gestures which are
coordinated in a specific way.
11
In AP, the idea is that gestures are coordinated in certain
ways within a given gestural constellation or unit, which is a lexical property that
contrasts one lexical item with another.
11
There are no segments in the traditional sense in AP. Instead, there are only gestures
which cohere into larger ‘molecules’ via coupling relations.
37
A planning model of speech production is also part of the larger theoretical
framework of AP. This planning model is described in Goldstein et al., 2006. As I
mentioned above, in the planning model, dynamic coupling is used like the glue that
holds individual gestures together. Each individual gesture has an oscillator with its own
cycle (we can think of this as each gesture having its own clock) and each gesture is
activated by a defined phase (0 degrees) of its own clock. This means that two gestures
can be coordinated with each other by coupling their corresponding clocks with a relative
phase during the planning stage. Phasing relationships can also be understood as the
degree of cohesion between two gestures, and the coupling as the bonding strength that
exists between them (Browman and Goldstein, 2000; Goldstein et al. 2006). Gestures
prefer to couple in the optimally stable mode. This mode is in-phase, as I described
previously for Figure 1.3 and illustrated via clocks in Figure 1.4. Sometimes, gestures can
also couple in anti-phase mode or in neither of these two intrinsically stable modes. As I
showed Figure 1.3, for the word /pan/, the gestural pairs of LIP and GLO, TT and
VELUM, and LIP and TB are in-phase coupled, while the gestures of TT and TB are
anti-phase coupled. The same idea was illustrated in Figure 1.4 via the green and red
lines for the word ‘mad’.
The AP model also makes specific predictions about syllable structure based on
the coupling relationships of C and V gestures. In general, consonant and vowel
constrictions involve distinct organs, so that C and V gestures can be produced
concurrently. The tongue body (TB) is the major organ for the production of vowel
gestures, with the lips playing a role as a more minor organ. Consonant gestures include
38
constrictions of all the organs, with the tongue tip (TT) organ being the one most
frequently involved in consonants. Goldstein el al. (2006) propose the C-V coupling
hypothesis. This theory holds that syllable structure emerges from the intrinsically stable
modes of coupling. The C and V gesture(s) in onset exhibit in-phase mode, while the V
gesture and coda C gesture (s) exhibit anti-phase mode. This means that in a CVC
syllable, we can represent the coupling relations as shown in Table 1.3. The solid line
represents in-phase coupling, while the dashed line shows anti-phase coupling
Figure 1.5 CVC coupling relationships
C V --------------C
The vowel is in-phase coupled with the onset consonant, and anti-phase coupled to the
coda consonant. There is no relationship assumed between the two consonant gestures.
This brings up another point which might lead us to ask how gestures are coupled
when there is a complex onset, as in the English word /spa/. For these cases, such as
when there is a CCVC sequence, the coupling relations between gestures can be in
competition with one another. This is known as the competitive coupling hypothesis
(Browman and Goldstein, 2000). In a case like C
1
C
2
VC
3
, then, both C
1
and C
2
are in-
phase coupled to the V gesture. However, C
1
and C
2
are anti-phase coupled to each other
as well. This means that the C-C anti-phase coupling relation is in competition with the
C-V in-phase coupling relation. When this occurs, the leftmost C shifts earlier relative to
the vowel, and the rightmost C shifts later relative to the vowel. This is illustrated in
Figure 1.6:
39
Figure 1.6 C-center effect
C
1
V
C
2
Figure 1.6 shows that the onset of the V gesture coincides with the mid point of the two C
gestures. When this shifting occurs, it is known as the “C-center effect” (Browman and
Goldstein, 1988, 2000). This C-center effect will apply even when there are more than
two consonants in onset position, although this is not relevant for the Zapotec data I
present in chapter 2. When there are consonant clusters in coda position, such as a
sequence like CVCC, the V gesture is hypothesized to be anti-phase coupled only with
the first C gesture. No competition is assumed to exist between the consonants that form
consonant clusters. The second C in a VCC sequence is only anti-phase coupled to the
preceding consonant.
The gestural theory that I have discussed so far (Browman and Goldstein 1986,
1990a, 1990b, 1992, 1995) and the task dynamic model of speech production (Saltzman
1986; Saltzman and Kelso, 1987; Saltzman and Munhall 1989) are used together to
model speech production. There are two levels to this model: the level of inter-gestural
coordination and the level of inter-articulator coordination. These two levels are both
40
needed; the abstract coupling graph of an utterance is taken at the inter-gestural level, and
then the output is generated from the inter-articulator level. I include a schematic
representation of this process in Figure 1.7 , which is adapted from Goldstein et al. 2006.
Figure 1.7 From coupling graph to output speech
Lexicon
Intergestural
Coupling
Graph
Gestural Tract/
prosody Planning Constriction
Oscillator Variables
Variables
Output Speech
rate
Activation Model
Variables Articulator
Gestural Variables
Score
Inter-gestural Inter-articulator
Coordination Coordination
Figure 1.7 is divided into two sides: the inter-gestural coordination and the inter-
articulator coordination. At the inter-gestural coordination level on the left side, this
41
figure shows that the gestural score is generated by the gestural oscillator variables and
activation variables, and is then sent as input to the inter-articulator coordination level.
Note that rate of speech as well as prosodic context can affect the gestural score, as
shown in Figure 1.7. So the left side of this diagram, the inter-gestural level, is the
planning level in this model. On the right hand side of the diagram is the inter-articulator
level, where the tract variables define the model articulator variables and execute the
articulatory gestures according to the gestural score input. For present purposes, the
important thing to take from this diagram is that the coupling graph is what generates the
gestural score, since this is relevant to the constraints I propose in chapter 3. Constraints
in chapter 3 will reference both the coupling graph as well as the overlap of gestural
activation by at least 25% of the activation duration of a gesture. Now that I have
reviewed the basics of the AP model, including the task dynamic model, I move to a
discussion of tones in AP.
1.3.3 Tone Gestures
The notion of an articulatory gesture is extended to encompass tones in Gao
(2008) and Mu0cke et al. (to appear). Gao’s work extends the coupled oscillator model for
intergestural timing to analyze the lexical tones of Mandarin Chinese and is the first work
to consider tones as articulatory gestures. I highlight her work here because the inventory
of Mandarin tones is similar to Zapotec with the exception that the Zapotec data I discuss
in chapter 2 also have a lexical M tone. Two invariant tone gestures with High (H) and
Low (L) pitch targets, respectively, are used to represent the four tones in Mandarin. In
42
Gao’s work, tone gestures are viewed as dynamic events that couple with other gestures
(such as consonant and vowel gestures) in various coupling modes. A simple H tone is
presumed to have a H target, while a L tone has a low target.
Gao proposes that Tone 2 (rising) in Mandarin is composed of L and H tone
gestures that are coupled in-phase, so that the onsets of both tones are synchronous. The
L tone is presumed to have a shorter activation duration, which then leads to the output
rising from L to H. For Tone 4 in Mandarin (falling), Gao proposes that the H and L tone
gestures are coupled in an anti-phase (sequential) relationship to each other, although
both are in-phase with the vowel. Figure 1.8 shows the coupling graph for Tones 1, 2, and
3, while Figure 1.9 shows the coupling graph for Tone 4. I show the gesture scores for
these tones in Figure 1.10.
Figure 1.8 Mandarin Coupling Graph of Tone 1,2, and 3 (Gao, 2008)
C-----------------T
V
Figure 1.9 Mandarin Coupling Graph of Tone 4 (Gao, 2008)
C----------T1----------T2
V
43
Figure 1.10 Mandarin Tone Gestures (Gao, 2008)
Tone 1
Tone 2
Tone 3
Tone 4
Tone 1, the level H tone, is comprised of only one gesture with an articulatory target of
H. Tone 2, the rising tone, is comprised of both L and H gestures which are in-phase
coupled to each other. Tone 3, the L tone, has only one tone gesture. Finally, the falling
tone, tone 4, has a H and L gestures which are coupled anti-phase. This information is
discussed again in chapter 3, but here I highlight Gao’s findings because her work is so
far the only study that has situated lexical tone gestures within the larger AP model, and
because her work finds evidence for the different ways that tone gestures can couple with
the underlying C and V gestures in a word.
For syllables with only one tone (either tone 1 (H) or tone 3 (L)), Gao’s work
showed that gestures occur in the order C-V-T, where T=Tone. The onset of the
consonant gesture occurred about 50 ms before the onset of the vowel gesture, while the
L
H
L
H
H L
44
onset of the tone gesture occurred after the vowel gesture by approximately 50 ms. This
showed that tone gestures act like consonant gestures in Mandarin and that they trigger
the C-center effect. These hypotheses were borne out via collection of both kinematic and
acoustic data, as well as via simulations generated with the Task-Dynamic Application
(TADA) (Nam et al. 2005). For tone 4 (falling from H to L), Gao showed that the H tone
was synchronized with the V, while the C preceded and the L followed. This is evidence
that the C gesture, H tone gesture, and L tone gesture are all coupled anti-phase to one
another, but in-phase to the vowel. As I discuss in chapter 3, these details about contour
tones in Mandarin are important for the constraint-based analysis I present for the
Zapotec data. Gao’s work is important because she demonstrates that tones act like other
gestures in terms of their coupling.
Mu0cke et al (to appear) examined the coupling of tone and constriction gestures in
pitch accent. Gao’s (2008) work on tone gestures was extended to see how pitch accents
are temporally coordinated with oral constriction gestures and with other (tone) gestures
in German and Catalan. Note that these two languages lack tone gestures which are part
of the lexical specification of words, since they have intonational pitch accents only but
not lexical tones like Mandarin. The authors note that an important difference between
languages with pitch accents and those with lexically specified tones is that the coupling
of pitch accent tones to syllables must happen after the coupling has been established
between the C and V gestures. This applies to languages like Catalan or German.
12
12
This leaves aside the question of languages like Japanese, where the pitch accent
functions like a lexical tone. More study of the tonal alignment and gestural
45
Therefore, the coordination of a tone gesture with a given syllable might not change the
intrasyllabic coupling relationships that define that syllable (Mueke et al, to appear, 5).
The authors found differences between the two languages. For Catalan pitch
accents, they found that the H tone gesture of the pitch accent was coupled in-phase with
the accented V gesture so that both gestures began simultaneously. For German, they
found that both L and H tones are in-phase with the accented V gesture and furthermore
that the H and L gestures engage in competitive coupling whereby the H tone gesture
shifts rightward to make room for the preceding L gesture. From this the authors make
important points relevant to the floating Zapotec tones discussed in this paper. First, they
note that when a tone gesture is added to a syllable in Catalan it produces no effect on the
C-V coordination, which differs from Mandarin, where adding a tone gesture causes the
C to precede the V because of the competitive coupling relationship. The authors draw a
distinction between the lexical tone gestures of Mandarin (which function and cause
coupling effects like consonants do) and the prosodic tone gestures in Catalan, which
appear to couple to V gestures without changing the within-syllable coupling relations.
For German, competitive coupling between the tone sequence L-H and the V gesture
causes a delay in the onset of the H tone gesture with respect to the vowel. But, the C-V
timing is not affected. The authors note that both Catalan and German demonstrate that
post-lexical tone gestures do not affect within-syllable coupling relations. They do not,
however, speculate about pitch accent languages like Japanese which are lexical.
representation of Japanese-type pitch accent systems would be useful. At the moment it is
not clear how the two types would differ, if at all.
46
The study of tonal perturbation in SJZ described in chapter 2 and analyzed in
chapter 3 proposes that the tones of this language can be modeled using the AP
framework. I argue that the floating tone of the first-person singular attaches after the
relationships between the underlying tone gesture, C, and V gestures have already been
established, much like the intonational pitch accents described for Catalan and German.
However, since SJZ also has lexical tones (which Catalan and German lack), I propose
that tone gestures are specified as part of the coupling graph for lexical items in SJZ. I
assume that tones that are not floating are represented as part of the coupling graphs for
lexical items so that they are coordinated with the vowel gestures with which they show
affiliation. I propose that the floating tone will always couple in-phase to the V gesture in
a given syllable where it attaches, just as was shown for Catalan and German pitch
accents. The OT constraints proposed in the analysis section of this paper incorporate
these ideas by making reference to tone gestures.
1.3.4 Deactivation of Tones
The current AP model holds that each gesture is associated with a planning
oscillator, or clock, which triggers that gesture’s activation. The relative phase of the
oscillators (and the timing of triggering) is then controlled by coupling the clocks to one
another. However, there is no specific theory of deactivation. It is assumed that gestures
of a given type will be deactivated at a particular clock phase. This clock phase is
different for C and V gestures, so V gestures are active for a time after they effectively
get to their targets. To give an analogy, assume we have a traditional wall clock. We can
47
then imagine that a gesture is activated at 12:00, and then at 3:00, it is deactivated. We
assume that each gesture has this ‘deactivation time’ already set into its clock and thus
that if no outside force acts upon it, it will deactivate at 3:00. The theory that this
dissertation presents is that tone gestures have a fundamentally different deactivation
typology. Instead of just having a “set” time such as 3:00 in their clocks which tells them
when to deactivate, a tone gesture has the potential to be deactivated by an outside force.
In SJZ, for example, constraints interact so that tones are only deactivated in this way.
13
I
propose that one possible cause for deactivation is the presence of a following anti-phase
coupled tone gesture. I propose that in the output all tone gestures are anti-phase coupled
to one another and thus that a following tone gesture can cause the preceding tone gesture
to be deactivated. I propose that anti-phase coupling of tones in the output is dictated by a
constraint on representations on GEN. That is, if an output contains 3 tones, they will be
anti-phase coupled in sequence, so that the second is anti-phase to the first, and the third
to the second. I also propose that the end of an utterance must deactivate a tone gesture.
Thus, in some languages, I propose that a tone will deactivate when it reaches a following
anti-phase coupled tone.
14
In some cases, a tone could remain active for longer than 360
degrees of its own clock but be deactivated instead by an utterance boundary (as would
happen if there is no following anti-phase coupled tone). Yet another possibility is for a
13
The rankable nature of constraints in OT means that different languages can have
different deactivation typologies, depending on how the constraints are ranked for each
language.
14
In a language where all tones are anti-phase coupled, where all tones are deactivated by
a following tone, and where all syllables must have a tone, it would be difficult to see the
difference between deactivation via a tone’s own clock and deactivation via a following
tone. That is, all tones would “turn off” at the onset of a following tone.
48
language to deactivate all tones at 360 degrees of their own clock regardless of whether a
following tone is present in the output or whether there is an utterance boundary. As I
show in chapter 3, these possibilities can be modeled using an OT constraint that
penalizes deactivation that occurs other than via a following anti-phase coupled tone or
an utterance boundary. Constraints also penalize overlapping activation of tone gestures
when deactivation does not occur by 360 degrees of a gesture’s own clock.
15
By
proposing OT constraints that control deactivation and penalize overlapping activation, I
can capture the patterns of the SJZ data, as I show in chapter 3, while allowing for the
possibility of other typologies for tone patterns in other languages, as I discuss in chapter
4.
1.4 Chapter Overviews
In what follows I will briefly present an overview of each chapter of the
dissertation.
1.4.1 Chapter 2 Overview
Chapter 2 includes the result of my fieldwork on SJZ with a native speaker of the
language. Chapter 2 presents two main processes of tonal perturbations: the floating H
tone of the first-person singular, and two processes of rightward tone spreading. Chapter
2 gives specific examples which show that when an aspect marker is toneless, the
15
Deactivation of a tone at 360 degrees of its own clock will be at the end of the syllable
with the vowel to which the tone gesture is coupled.
49
floating H tone of the first-person singular will appear on the aspect marker as well as on
the stressed syllable of the verb root, and on any syllables that intervene between the two.
In some cases, a contour tone forms on the stressed syllable of the root, and I give
examples where this occurs and where it does not, with explanations of their relevant
differences. For rightward tone spreading, I show that there are two processes at work.
One of these is rightward spreading from a verb-final H tone onto the attached clitics.
This spreading is very consistent and happens for 12 of the 13 pronouns in SJZ.
16
The
second process of rightward spreading is from a toneful syllable onto a toneless syllable.
As I show in chapter 2, a toneless syllable will take whatever tone is to its immediate left.
Thus, chapter 2 is a summary of the most noteworthy tonal processes I have found in
SJZ. These tonal processes motivate the theoretical approach I adopt in chapter 3. In each
example, I give the underlying form(s) of the verb root, aspect markers, and clitics, and
then show how the tonal pattern changes when these are combined.
1.4.2 Chapter 3 Overview
Chapter 3 is the section of the dissertation where I develop an OT analysis of the
patterns which I present in chapter 2. This chapter begins with a detailed analysis of the
placement of the floating H tone of the first-person singular. I present constraints that call
for the floating H to be realized in word-initial position and to appear on the stressed
syllable. In my analysis I assume that the floating H tone is a H tone gesture which does
not have a coupling relationship established in the input. Where this H tone is ultimately
16
The exception is one high-toned enclitic pronoun that undergoes a lowering process.
50
coupled is then determined by the constraints I propose and their ranking. In chapter 3 I
also propose constraints that govern the deactivation of tone gestures. These constraints
penalize tone deactivation when it occurs anywhere except for at the onset of an anti-
phase coupled tone or a phrase boundary. I show how these constraints can account for a
grammar where tones asymmetrically spread rightward onto toneless syllables. Finally, I
include a description of the related dialect MacZ in chapter 3 and show how the
constraints I propose for SJZ can also account for this dialect.
1.4.3 Chapter 4 Overview
Chapter 4 presents a factorial typology for the deactivation constraints I present in
chapter 3. I also include typologies that include constraints governing the coupling of
tones between input and output. This chapter presents a comparison of my analysis to the
autosegmental rules which have been used to account for tones and their spreading. I
show that while autosegmental rules are effective at capturing the tonal patterns exhibited
by many languages, they fail to explain the rightward or perseveratory bias of tone
spreading. Hyman (2007) argues that tones spread rightward far more often than they
spread leftward, and he argues that cases of leftward spreading usually occur due to
suffix control, where a suffix spreads its tone onto a root with which it is affiliated. This
type of leftward spreading might also be due to an edge prominence effect. Chapter 4
argues that constraints governing the deactivation of tone gestures provide a more
effective approach than a rule that says “spread-right” for which we could expect an
equivalent counterpart “spread-left”. I also include three brief case studies in chapter 4
51
that analyze tone patterns in Chilungu, Margi, two Chinese dialects, and pitch accent
languages such as Lithuanian. These case studies show how different rankings of the
constraints I present in chapter 3 can account for different phonological grammars for
tone.
1.4.4 Appendix
In the appendix, for each verb that I recorded with the speaker DP, I give the
underlying form of the verb root along with the output tonal patterns for each pronoun in
each of the three aspects (potential, completive, and habitual). I also include information
from Bartholomew (1983) giving the verb class which the particular root belongs to, as
well as the page number where the verb is located in the Nellis and Nellis dictionary. In
most cases my transcriptions are very similar to those given by Nellis and Nellis, but my
appendix is more thorough in the sense that I include the first-person singular forms for
all three aspects, while Nellis and Nellis only give the potential form of each verb in the
first-person singular. I also show the tonal perturbations that occur when various verb
roots attach to the 13 different pronouns. Although Bartholomew’s grammar gives an
excellent description of each of the dependent pronouns, and of the segmental variation in
verb roots, she does not show each verb with the changes that occur after the pronouns
attach.
52
1.5 Conclusion
This chapter has presented a basic introduction to the dissertation as a whole, as
well as to the language SJZ, its speakers, and its inventory. I have briefly explained both
Optimality Theory and Articulatory Phonology, and how my approach addresses the
deactivation of tone gestures in SJZ. As I discussed in the brief summary of each chapter,
I will now move to a detailed description of the tonal patterns in SJZ. The following
chapter 2 will present the data which are the focus of much of chapter 3, and which form
an empirical foundation for my gestural approach to tones.
53
Chapter 2 Tonal Data from Sierra Juárez Zapotec
2.1 Introduction
The goals of this chapter are to present the data that have resulted from my
fieldwork with a native speaker of SJZ and to expand our understanding of the first-
person singular marking in the language. While I presented a very basic overview of the
phonological structure of SJZ in chapter 1, here I will focus on the first-person singular
and its tonal marking. I also discuss patterns of rightward spreading which affect all verbs
and their clitics, regardless of the particular pronoun in question. These data add to the
description of the tonal patterns in SJZ and expand upon the data available in the Nellis
and Nellis dictionary (1983) and the Bartholomew (1983) grammar. They are also
important because they form the basis of the analysis I present in chapter 3.
In this chapter I will first discuss segmental and tonal variation of the first-person
singular enclitic. Next, I describe and explain the location of the floating H tone of the
first-person singular. As I will argue in this chapter, the first-person singular form of
transitive verbs is marked with both a segmental affix (an enclitic) and a floating H tone.
I speculate that only transitive verbs have this floating H tone but leave the question open
for further investigation in the future.
17
My findings support the claim originally made by
Bickmore and Broadwell (1998) that only transitive verbs have the floating tone in SJZ,
since verbs such as those that are glossed as ‘fall down’, and ‘awaken’ did not show this
17
In my database of 43 verbs, all transitive verbs have the floating tone. A larger
database of 100-200 verbs would be desirable.
54
floating tone in my elicitation sessions. Broadwell (2000) makes a similar claim for
transitives in Macuiltianguis Zapotec. I do not claim to have definitely answered the
question about the relationship between transitivity and the floating tone, however. It
would be useful to determine whether this special marking occurs for imperatives, for
example. My work supports the claim that 1sg transitive verbs have this floating tone and
that some intransitives lack it, but a larger database from which to draw data would be
useful, especially so that additional speakers could be included.
It seems clear based on the data I have collected that it is only the first-person
singular form of verbs that have a floating H tone. The first-person plural forms do not
show the floating H tone, nor do any other person markers in the language. Broadwell
(2000) and Beam de Azcona (2004) have discussed similar findings for other Zapotec
languages, but it would also be interesting to investigate why the first-person in particular
is special. Work by Wechsler (2010) proposes that first- and second-person pronouns are
grammatically specialized for self-ascription by the speaker and addressee, respectively,
and proposes a theory that while requiring self-ascription for these pronouns, does not
require grammatical reference to the speaker alone (p. 362). Although Wechsler’s work
does not answer the question of why only the first-person singular would be specially
marked in Zapotec, it does show that the first- and second-person categories may be
special or different from other pronouns. Other work showing that first- and second-
person plural pronouns have associative semantics rather than regular plural semantics
include Moravcsik 1978:356, Greenberg 1988: 14; Noyer 1992:31; Cysouw 2003, and
Siewierska 2004:82-83.
55
The floating H tone is attracted to the stressed syllable of the verb root, but may
also appear on underlyingly toneless aspect markers which precede the root. Contour
tones form on the stressed syllable of the verb root in cases where there is an underlying
L or M tone on the stressed syllable and when it contains a long vowel. In these cases, a
falling contour is formed when the floating H is realized on the stressed syllable. In a few
cases I show pitch tracking examples that illustrate the transcriptions I give in this
chapter. For the speaker DP, whose recordings were used to generate all of the pitch
tracks shown, a H tone is generally between 140 and 160 Hz, a M tone is around 100-120
Hz, and a L tone is around 80-90 Hz. These pitch ranges are consistent for the examples
from DP that I show, but the ranges do not represent all speakers of SJZ. In cases of
downdrift over an utterance, H tones may decrease by as much as 15 Hz across the
utterance. In some cases, a H tone that occurs utterance-finally will only measure at 125
Hz. While the pitch measurement itself is lower in these cases, a H tone is still higher
than a M tone (which may also shift downwards and measure at 105 Hz).
In addition to the processes that are evidenced in first-person singular forms, I
also discuss processes of rightward tone spreading in the language. Verb-final H tones
spread rightward onto attached clitics, and do so consistently regardless of the verb class
or the underlying tone of the clitic. There is also a process where toneless syllables
receive their tones due to perseveratory spreading from a previous syllable. In my data
this has become apparent for the causative marker /di/ and occasionally for verbs which
have a toneless initial syllable. Each of these particular processes will be examined in
detail below, along with relevant examples. Although I have given at least two verbs for
56
each class where possible, this does not mean the examples are limited to these particular
verbs. More examples which follow the same patterns are located in the appendix.
2.2 First-person Singular Clitic
Although the focus of this chapter is the tonal patterns of the language, I
uncovered some segmental (and tonal) variation in the first-person singular clitic. I will
present the patterns here so that in later examples and in the analysis chapter, the
variation in first-person singular forms is clearer. The recordings I have done suggest that
the underlying form not including the floating tone of the first-person singular marker is
/=a$/.
18
As I will show, the verb root is also marked with a H tone. The H tone and its
eventual placement in the output is complex, so I discuss it exclusively in an upcoming
section.
While the H tone always appears as long as the verb is transitive, the segmental
and tonal content of the first-person singular clitic varies somewhat. I argue that the 1 sg
clitic lacks its own underlyingly coupled tone, as I show in this section. Bartholomew
(1983) found that the phonetic realization of this person marker varies according to the
environment in which it appears, and my data support this conclusion. Bartholomew
gives the underlying form of this clitic (not including the floating tone) as /=$a#$/, while I
propose that it is /=a$/ plus a floating H tone. The data I have collected suggest that the
18
This is the most frequent form. Some verbs take the dative suffix /te$/ which marks
1sg. See Bartholomew (1983: 349) for a discussion of this suffix.
57
realization is [=a#$], [=$], [=a$], or [=á$] depending on context.
19
The following
paragraphs will explain my arguments in favor of /=a$/ as the underlying segmental form
for this part of the first-person singular morpheme.
If the verb-final vowel is /a/ with any tone, then the clitic is reduced to a glottal
stop. There is no noticeable lengthening to suggest an additional vowel /a/ is added to the
clitic in these cases. For verbs that end with a vowel other than /a/, then the clitic is
realized as /=a$/ with various tones. The tone which appears on the clitic is determined
by the tone of the final vowel of the verb root. If the verb-final tone is L, then the 1sg
clitic will have a L tone. Likewise it will have a M or H tone if the verb-final tone is M or
H. For this reason, I posit that while there is a floating H tone associated with the first-
person singular, the first-person singular enclitic /=a$/ does not have an underlying tone
coupled to it. As I will discuss in the following section on tone spreading, I posit that
rightward spreading from a toneful to a toneless syllable is common in the language. I
will give other examples in this chapter where the same property is exhibited in forms
other than the first-person singular.
Here I present examples from my database of 43 verbs that show the various
forms of the first-person singular enclitic. Consider first that the L tone of a verb-final
vowel spreads rightward only onto the first-person singular clitic. Examples (1) and (2)
19
I have not recorded any examples where this clitic appears as /=ja$/, so I do not
comment on that particular form. Bartholomew gives this as another possibility for the
first-person singular clitic.
58
illustrate that the second person informal /lu%$ / has an underlying M tone, whereas the
first-person singular does not. As shown in these examples, the second-person informal
does not undergo spreading.
(1a) ka!"bi#=a#$ (1b) ka!"bi#=lu%$
POT/paint=1s POT/paint=2s.INF
‘I will paint’ ‘You will paint’
(2a) g-a!"'i#=a#$ (2b) g-a!"'i#=lu%$
POT-call=1s POT-call=2s.INF
‘I will call’ ‘You will call’
Examples (1a) and (2a) show that when the verb ends in a low-toned vowel, this low tone
spreads onto the first-person singular clitic. Figure 2.1 illustrates the L tone on the verb-
final vowel and the clitic and Figure 2.2 shows the spectogram for this token:
Figure 2.1 [ka!"bi#=a#$]
59
Figure 2.2 Spectrogram of [ka!"bi#=a#$]
In Figure 2.1, the initial portion of the clitic vowel /a/ is L, but then the speaker’s voice is
becomes creaky and the second half of the vowel the vowel /a/ did not pitch-track
correctly in this example. Figure 2.2 shows the spectrogram to illustrate the creaky voice
in this example. The arrow in Figure 2.2 points to the period of irregular glottal pulses.
Now, consider examples of verbs which end with a vowel other than /a/ and
which have a mid tone. In this case, I posit that the M tone of the final vowel of the verb
root spreads onto the first-person clitic.
(3a) í-bí&i%=a%$ (3b) í-bí&i%=lu%$
POT-fall.down=1s POT-fall.down=2s.INF
‘I will fall down’ ‘You will fall down’
60
(4a) í-láni%=a%$ (4b) í-láni%=lu%$
POT-hug=1s POT-hug=2s.INF
‘I will hug (someone)’ ‘You will hug (someone)’
(5a) é-ba%ni%=a%$ (5b) é-ba%ni%= lu%$
POT-wake.up=1s POT-wake.up=2s.INF
‘I will wake up’ ‘You will wake up’
All three of these examples have a final /i/ on the verb root, and the first-person singular
clitic emerges with a M tone also in all three cases.
When the verb-final vowel has a high tone and is a vowel other than /a/, the first-
person singular enclitic emerges with a H tone. This is illustrated in examples (6) and (7).
As I discuss later in this chapter, spreading of verb-final H tones onto clitics happens
regardless of whether the clitic has an underlying tone.
(6a) wá$ní=á$ (6b) wá$ní=lú$
POT/carry=1s POT/carry=2s.INF
‘I will carry’ ‘You will carry’
(7a) tá-néllí=á$ (7b) tá-ne#llé=lú$
POT-visit=1s POT-visit=2s.INF
‘I will visit’ ‘You will visit’
The examples in (6) and (7) show that both first person singular clitic and the second-
person singular informal clitic are high-toned due to spreading from the final vowel of the
verb root. Example (7a) also illustrates that a verb-final vowel /é/ will become /í/ before
61
the vowel /a/. A pitch tracking example demonstrating this phenomena is found in Figure
2.3:
Figure 2.3 [wá$ní=á$]
Figure 2.3 illustrates that when the verb-final vowel has a H tone, then the first-person
singular clitic will also emerge with a H tone. Both the final /i/ of the verb root and the /a/
of the clitic are realized with a pitch around 150 Hz. As I will show in the following
sections, rightward spreading of tones onto toneless syllables is very common, and is
predicted by the analysis that I present in chapter 3.
When the final vowel of the verb root is /a/, the clitic is reduced to a glottal stop.
This is shown in examples (8) and (9):
20
20
For examples (8a) and (9a), it is not clear where best to mark the clitic boundary.
Perhaps this is due to an OCP-type effect with the vowel /a/, such that the verb-final
vowel /a/ and the suffix /a/ are fused. It is also possible that /a + a/ is repaired via fusion
without an OCP basis. It is also possible that one of the /a/ vowels has been deleted.
62
(8a) kába#=$ (8b) kába#=lu%$
POT/shave.wood=1s POT/shave.wood=2s.INF
‘I will shave wood’ ‘You will shave wood’
(9a) gú-día#=$ (9b) gú-di#a#=lu%$
POT-write=1s POT-write=2s.INF
‘I will write (it)’ ‘You will write (it)’
In these two cases, the final vowel of the verb root is low toned. The same pattern
emerges when the final vowel /a/ has a M tone. This is illustrated in examples (10)
through (11):
(10a) é-júnna%=$ (10b) é-júnna%=lu%$
POT-return=1s POT-return=2s.INF
‘I will return (it)’ ‘You will return (it)’
(11a) ú-&átta%=$ (11b) ú-&átta%=lu%$
POT-iron=1s POT-iron=2s.INF
‘I will iron’ ‘You will iron’
The same pattern is true for high-toned final vowels which are illustrated in (12) and
(13):
(12a) k
w
édá=$ (12b) k
w
édá=lú$
POT/wait=1s POT/wait=2s.INF
‘I will wait’ ‘You will wait’
(13a) ú-ká$ná=$ (13b) ú-ká$ná=lú$
POT-leave.something=1s POT-leave.something=2s.INF
‘I will leave (it)’ ‘You will leave (it)’
63
Examples (12b) and (13b) show a process of rightward spread of H tones, where a verb-
final H tone spreads onto the second-person singular clitic. Examples (12a) and (13a)
illustrate that when the verb-final vowel is /a/, regardless of its tone, the first-person
singular clitic is realized as a glottal stop.
There is one more form in which the first-person singular clitic appeared in my
database. This was [=a#$] for verbs which end in a glottal stop. On one hand, /=a#$/ could
be taken to be the underlying form of this clitic, since I assume that no tone spreading can
occur through a glottal stop (as evidenced by the fact that in other places in the language,
tones do not spread rightward through a glottal stop –see section 2.5.6. However, it is
also the case that having a glottal stop on either side of a vowel could cause creaky voice,
which would lower pitch considerably, causing a L tone.
21
I propose that the presence of
two glottal stops is what causes the vowel of the clitic to be low in these cases, and that
the underlying form of this clitic is toneless /=a$/ which becomes [=a#$] when it is
attached to a verb with a final glottal stop. Examples (1) through (7) which I presented
earlier also support my argument that /=a$/ is toneless because L, M, and H tones spread
onto it, while only H spreads onto /lu%$/ because /lu%$/ has an underlying tone of its own.
Examples of verbs that end with a glottal stop, and the form of the first-person
clitic that results are shown in examples (14) and (15) below:
21
For speaker DP, creaky voice results in very low pitch measurements or in failure of
the pitch tracking algorithm in some cases. The relationship between pitch and non-
phonemic creak is a topic for further analysis.
64
(14a) í-látsu#$=a#$ (14b) í-la%tsu#$=lu%$
POT/pull.out=1s POT/pull.out=2s.INF
‘I will pull (it) out’ ‘You will pull (it) out’
(15a) be%-di%-tsá$=a#$ (15b) be%-di%-tsa#$=lu%$
COM-CAUS-full=1s COM-CAUS-full=2s.INF
“I filled (it) up” ‘You filled (it) up’
Examples (14a) and (14b) show that the first-person singular clitic has a low tone when it
follows a glottal stop, while the second-person singular clitic has its usual M tone when
following a verb root that ends with a glottal stop. In the pitch tracking examples for
these forms, the first-person singular clitic is normally realized with creaky voice – this
causes the pitch-tracking algorithm to fail, due to the irregular glottal pulses as illustrated
in Figure 2.4. A spectrogram for this example is shown in Figure 2.5 so that the creaky
voice is visible:
65
Figure 2.4 [í-látsu#$=a#$]
Figure 2.5 Spectrogram of [í-látsu#$=a#$]
66
In Figure 2.4, the speaker’s voice during the final vowel of the verb root and during the
first-person singular was creaky. I attribute this to the verb-final glottal stop and the
glottal stop at the end of the first-person singular clitic. In figure 2.5, an arrow points to
the creaky voice and irregular glottal pulses that are illustrated in the spectrogram.
In this section on the enclitic form of the first-person singular, I have argued that
the clitic is underlyingly toneless and that it obtains its tone due to rightward spreading
from the verb-final tone. The exceptions to this occur when the verb-root ends with the
vowel /a/ of any tone, in which case the clitic is reduced to a glottal stop. As I illustrated
in Figure 2.4 and Figure 2.5, the glottal stop of the first-person singular can at times
cause creakiness and thus reduce pitch. I assume this is a phonetically driven process that
is separate from phonological tone spreading. In summary, I assume that the verb-final
vowel /a/ has a phonological tone even when its syllable is closed by a glottal stop, but
that the clitic has a phonologically toneless /a/, as argued throughout this section.
2.3 Third-person Singular Non-formal
Similar evidence of tonelessness is evident for the third-person non-formal clitic
/a)/.
22
I mention this clitic here because it will be discussed in my theoretical arguments in
chapter 3. My data on the third-person singular clitic differ from the report of different
realizations that Bartholomew gives in her grammar. I present her findings on the third-
person singular clitic in more detail in chapter 1. Here, I will discuss the patterns found in
22
In the related dialect of Macuiltianguis Zapotec, this clitic is /na#/, which easily explains
its nasalization in SJZ. However, this would also mean that the two dialects differ in the
tonal specification of this clitic.
67
the data from the speaker DP. For him, the clitic /a)/ will take the tone of the verb-final
vowel. In cases where the verb-final vowel is already /a/, nasalization will be added to the
verb-final vowel, but no lengthening occurs. Thus the verb-final vowel /a/ and the clitic
/ã/ appear to merge. An additional complication is that this morpheme can cause a
segmental change in the vowel of the root, although it will still take the same tone that the
verb-final vowel has.
My data show that verb-final /e/ will become /i/ when followed by the third-
person singular non-formal clitic. This is the same pattern that Bartholomew noted (1983:
352). Examples illustrating these effects from my database are found in (16):
(16a) tá-ne#llí=a) . (16b) tá-ne#llé=lú$
23
POT-visit=3s.NFML POT-visit=2s.INFML
‘He will visit’ ‘You will visit’
This is a case where the clitic induces a change on the verb-root itself. This vowel quality
change does not affect the tonal properties however, since the verb-final vowel is H in
both cases.
We can see more clearly that the third-person singular lacks an underlying tone of
its own by observing examples (17a) and (17b), where the verb-final tone is L.
23
As I will discuss in a later section of this chapter, verb-final H tones spread rightward
onto clitics even if these clitics already have an underlying tone. This is what causes /lu%$/
to be H-toned in this example.
68
(17a) ka!"bi#=a) # (17b) ka!"bi#=lu%$
POT/paint=3s.NFML POT/paint=2s.INFML
‘He will paint’ ‘You will paint’
In (17a) the third-person singular marker has the same tone as the verb-final vowel. This
suggests that the verb-final tone has spread rightward onto the clitic. As shown in (17b),
this does not happen in the second-person singular, where the clitic /lu%$/ has an
underlying M tone. As I will show in following sections in this chapter and in the
analysis chapter, rightward spreading onto toneless syllables, whether in the verb root or
in clitics, is common in this language. In (17b), I assume the tone of the second-person
singular has prevented rightward spreading of the verb-final L tone.
Finally, I will show an example that illustrates what happens when the verb-final
vowel is already /a/. In this case, nasalization is added to mark the third-person non-
formal. Observe the examples in (18a) and (18b):
(18a) k
w
éda
. ) (18b) k
w
édá=lú$
24
POT/wait/3s.NFML POT/wait=2s.INFML
‘He will wait’ ‘You will wait’
Example (18a) shows that the final vowel of the verb root has become nasalized.
Presumably, the segmental portion of this clitic is not separately produced because there
would then be two /a/ vowels next to one another. In this case, the nasalization appears to
24
This example shows another case where the verb-final H tone spreads rightward onto
the clitic, making it too become H.
69
be the only marking of the third-person singular non-formal morpheme, or, as mentioned
before, the two /a/s have merged.
Both the first and third-person singular clitics obtain their tones from the final
vowel of the verb root. In some cases these clitics are reduced to simply adding a glottal
stop for the first-person singular, and just to adding nasalization for the third-person
singular. In cases where this reduction does not occur, the tone spreading is easier to
observe. As I will show in chapter 3, the cases where no reduction occurs but where tone
spreading is evident are good examples where the constraints I develop in this
dissertation can be applied. Finally, I do not posit that any of the other 11 pronouns in the
language lack underlying tones. I argue that only the first-person singular and the third-
person singular are underlyingly toneless.
Now that I have described the segmental content of the first-person singular
morpheme and its allomorphs along with the realizations of the third-person singular
morpheme, I will move to a discussion of the floating H tone of the first-person singular.
As I explained in the introduction, the first-person singular form of transitive verbs is
marked at the right edge with an attached clitic, and with a H tone that appears in various
locations on the verb root or aspect marker. The exact location where this floating H tone
appears is the subject of the next section.
2.4 The Floating H Tone of the First-person Singular
The first-person singular has a floating H tone that appears in transitive verbs. It is
absent in verbs that are intransitive, and for verbs which can be either transitive or
70
intransitive, it only appears in the transitive version. I call this tone a floating tone
because it is not realized on the suffix vowel and it is absent on verbs with other
pronouns of the language. One might also refer to this as a tonal morpheme, since it acts
as part of the first-person singular morpheme. First-person singular verbs have this tone
in addition to the enclitic morpheme I already discussed in this chapter. I start with an
example from verb class 1 where the floating H tone appears on the stressed syllable of
the verb root. I present the underlying form of the verb root by itself, along with its
output form shown in brackets, as in (19a), and then I show the first, second, and third-
person singular forms for comparison, as in (19b-d).
(19a) /ú-&a%tta%/ [ú&a%tta%] (19b) ú-&átta%=$
POT-iron POT-iron=1s
‘will iron’ ‘I will iron’
(19c) ú-&a%tta%=lu%$ (19d) ú-&a%tta%=bí
POT-iron=2s.INF POT-iron=3s.FAM
‘You will iron’ ‘He will iron’
All three examples have a H toned potential aspect marker /ú/, followed by the verb root
and then the appropriate clitic. The difference in example (19b) is that the first-person
singular has a H tone on the stressed syllable of the verb root. Notice that the second and
third-person singular shown in examples (19c) and (19d) lack this H tone. Although I do
not show them here, the H tone is also absent in the other 10 pronouns of the language
(see the appendix for the full paradigm). Since the H tone only appears for the first-
person singular, I argue that it is associated only with the first-person singular and not
71
with the other 10 pronouns of the language. The same pattern emerges in the completive
aspect of the same verb. This is illustrated in examples (20a) through (20d).
(20a) /be%-&a%tta%/ [be%&a%tta%] (20b) be%-&átta%=$
COM-iron COM-iron=1s
‘ironed’ ‘I ironed’
(20c) be%-&a%tta%=lu%$ (20d) be%-&a%tta%=bí
COM-iron=2s.INF COM-iron=3s.FAM
‘You ironed’ ‘He ironed’
Again, only the first-person singular has a H tone on the stressed syllable of the verb root.
This can be seen by comparing Figure 2.6 and Figure 2.7, which show examples of the
first-person singular and second-person singular, respectively:
Figure 2.6 [be%-&átta%=$]
In Figure 2.6 the floating tone appears on the stressed syllable. Compare with Figure 2.7:
72
Figure 2.7 [be%-&a%tta%=lu%$]
Figure 2.7 shows that for the second-person singular there is no floating H tone on the
stressed syllable of the verb root.
I posit that because the aspect marker /be%/ has a M tone, the floating tone cannot
appear in this location and instead is only attracted to the stressed syllable of the verb root
in the first-person singular. As I show in an upcoming section, when the aspect marker is
toneless we can expect the floating tone to be realized in that position as well as on the
stressed syllable. For the verb /&a%tta%/, I have shown here that the first syllable of the root
has an underlying M tone and that the floating H replaces this in the first-person singular.
73
Another verb which shows the floating H tone of the first-person singular is /+i#a#/
‘to close’.
25
This verb is illustrated in examples (21b) through (21d). I show the
underlying representation in (21a).
(21a) /ú -+i#a# / [ú+i#a#] (21b) ú-+ía#=$
POT-close POT-close=1s
‘will close’ ‘I will close’
(21c) ú-+i#a#=lu%$ (21d) ú-+i#a#=bí
POT-close=2s.INF POT-close=3s.FAM
‘You will close’ ‘He will close’
Example (21b) shows that the floating H tone has appeared only on the stressed syllable
of the verb root. Since this particular verb has L tones specified on two syllables of the
verb root, the floating H appears in place of the L tone of the first syllable. I posit that a
contour is not formed because the stressed syllable is short (the same is true for the root
/&a%tta%/). Examples (21b) through (21d) also illustrate that this H tone is unique to the first-
person singular, since none of the other pronouns have it.
The completive aspect of this verb shows the same pattern: the stressed syllable
has a H tone in the first-person singular which is not found in the other person markers.
This becomes apparent by examining the following forms in (22):
25
This verb was given by DP as “cerrar” and is transitive. It was elicited in the context
of ‘close a door’ or ‘close a window’.
74
(22a) /be%-+i#a#/ [be%+i#a#] (22b) be%-+ía#=$
COM-close COM-close=1s
‘closed’ ‘I closed’
(22c) be%-+i#a#=lu%$ (22d) be%-+i#a#=bí
COM-close=2s.INF COM-close=3s.FAM
‘You closed’ ‘He closed’
Here I posit that the specified M tone of the completive aspect marker /be%/ blocks the
floating tone from appearing on the same syllable, and that the floating tone may only
appear on the stressed syllable of the verb root. I give an analysis of this phenomenon in
chapter 3.
A different pattern emerges for the habitual aspect of this verb class. I posit that
the habitual aspect marker is toneless and therefore that the floating H tone appears on
the habitual prefix /ru/ in addition to the stressed syllable of the verb root. This is shown
in the data in (23) and (24). In these examples, the toneless aspect marker
26
becomes H in
the first-person singular forms only.
(23a) /ru-&a%tta%/ [ru&a%tta%] (23b) rú-&átta%=$
HAB-iron HAB-iron=1s
‘irons’ ‘I iron’
(23c) ru-&a%tta%=lu%$ (23d) ru-&a%tta%=bí
HAB-iron=2s.INF HAB-iron=3s.FAM
‘You iron’ ‘He irons’
26
Recall that toneless vowels are realized with a M pitch (about 100 Hz for speaker DP).
75
(24a) /ru - +i#a#/ [ru+i#a#] (24b) rú-+ía#=$
HAB-close HAB-close=1s
‘closes ‘I close’
(24c) ru-+i#a#=lu%$ (24d) ru-+i#a#=bí
HAB-close=2s.INF HAB-close=3s.FAM
‘You close’ ‘He closes’
Data sets (23) and (24) show that the floating H tone appears not only on the stressed
syllable of the verb root but also on the habitual aspect marker in the first-person singular
forms. This pattern occurred for all twelve verbs in my database from verb class 1. I take
this as evidence that while the potential aspect marker /gú/
27
has an underlying H tone
and the completive aspect marker /be%/ has an underlying M tone, the habitual /ru/ is
toneless.
The same pattern may be found in all of the verbs from class 1. To further
demonstrate, I include an additional verb in all three aspects. (25a) – (25l) show the
potential, completive, and habitual aspects for the verb root / ka#$ná/ ‘to leave (an
object)’:
(25a) /ú-ka#$ná/ [úka#$ná] (25b) ú-ká$ná=$
POT-leave POT-leave=1s
‘will leave (it)’ ‘I will leave (it)’
27
For this verb class, the speaker varied between /gú/ and /ú/ for the potential aspect
marker.
76
(25c) ú-ka#$ná=lú$
28
(25d) ú-ka#$ná=bí
POT-leave=2s.INF POT-leave.sthg=3s.FAM
‘You will leave (it)’ ‘He will leave (it)’
(25e) /be% - ka#$ná/ [be%ka#$ná] (25f) be%-ká$ná=$
COM-leave COM-leave.sthg=1s
‘left (it)’ ‘I left (it)’
(25g) be%-ka#$ná=lú$ (25h) be%-ka#$ná=bí
COM-leave.sthg=2s. INF COM-leave.sthg=3s.FAM
‘You left (it)’ ‘He left (it)’
(25i) /ru - ka#$ná/ [ruka#$ná] (25j) rú-ká$ná=$
HAB-leave HAB-leave.sthg=1s
‘leaves (it) ‘I leave (it)’
(25k) ru-ka#$ná=lú$ (25l) ru-ka#$ná=bí
HAB-leave.sthg=2s. INF HAB-leave.sthg=3s.FAM
You leave (it)’ ‘He leaves (it)’
Example (25b) shows that in the potential the floating H only appears on the stressed
syllable of the verb root. Example (25f) shows that in the completive, the floating tone
only appears on the stressed syllable of the verb root. Finally, example (25j) shows that in
the habitual aspect the floating tone appears on the aspect marker /ru/ as well as on the
stressed syllable of the verb root. I take this evidence from verb class 1 as an indication
that the potential aspect markers /gú/ and /ú/ have an underlying H tone and that the
completive aspect marker /be%/ has an underlying M tone. Because the floating H tone
28
H tone spreading onto clitics is discussed in following sections.
77
appears on the habitual aspect marker /ru/, I assume that the habitual aspect marker in
verb class 1 is underlyingly toneless.
2.4.1 Toneless Aspect Markers in other Verb Classes
To illustrate that toneless aspect markers always take the floating H tone in first
singular subject forms, I will now present examples from other verb classes in the
database. This is to illustrate that the patterns I have presented thus far are not specific or
exclusive to verb class 1. Whenever there is a toneless aspect marker, it will be realized
with a mid pitch with other person markers, but as H when the 1 sg attaches.
29
Verb class
4 has the underlying forms /í/ for the potential aspect, /gu/ for the completive, and /ri/ for
the habitual. Several verbs from this class show that the floating H tone of the first-person
singular appears on both the aspect marker and the verb root in the habitual and the
completive aspects. This is evidence that the habitual and completive aspect markers are
underlyingly toneless in this class.
First, consider the verb /la%tsu#$/ ‘to pull out’.
30
Examples (25a) through (25d)
show that for the first-person singular, only the stressed syllable of the root becomes H.
The potential aspect marker itself is already high-toned. This is shown in data set (26):
29
If there were L-toned aspect markers, I predict that the floating H would not appear on
these but rather would go to the stressed syllable of the verb root. The floating H is only
attracted to toneless aspect markers.
30
This verb is transitive – sacar/ extraer – ‘to pull out or remove an object’.
78
(26a) /í-la%tsu#$/ [íla%tsu#$] (26b) í-látsu#$=a#$
POT-pull.out POT-pull.out=1s
‘will pull out’ ‘I will pull out’
(26c) í-la%tsu#$=lu%$ (26d) í-la%tsu#$=bí
POT-pull.out=2s. INF POT-pull.out=3s.FAM
‘You will pull out’ ‘He will pull out’
The H tone of the potential aspect marker does not spread rightward onto the verb root.
This shows that the first syllable of the verb root has an underlying M tone.
The pattern is different for the other two aspects however, because the aspect
markers lack a H tone in forms that are not first-person singular. I assume this is because
the completive and habitual aspect markers do not have underlying tones for this verb
class. The examples in (27a) through (27d) illustrate the completive aspect for the verb
/la%tsu#$/. Notice that the floating H tone of the first-person singular appears on both the
aspect marker and the stressed syllable of the root in the completive"
(27a) /gu-la%tsu#$/ [gula%tsu#$] (27b) gú-látsu#$=a#$
COM-pull.out COM-pull.out=1s
‘pulls out’ ‘I pulled out’
(27c) gu-la%tsu#$=lu%$ (27d) gu-la%tsu#$-bí
COM-pull.out=2s. INF COM-pull.out=3s.FAM
‘You pulled out’ ‘He pulled out’
This data set shows that the completive aspect marker for this class is underlyingly
toneless. I have shown this aspect marker without a tone marking in (27a), (27c), and
79
(27d), and it is realized with something similar to a M tone. The evidence that this aspect
marker is toneless is found in example (27b), where the floating H tone appears both on
this syllable and on the stressed syllable of the verb root. The same pattern emerges in the
habitual aspect, as shown in data set (28):
(28a) /ri-la%tsu#$/ [rila%tsu#$] (28b) rí-látsu#$=a#$
HAB-pull.out HAB-pull.out=1s
‘pulls out’ ‘I pull out’
(28c) ri-la%tsu#$=lu%$ (28d) ri-la%tsu#$=bí
HAB-pull.out=2s. INF HAB-pull.out=3s.FAM
You pull out’ ‘He pulls out’
To compare the forms in (28b) and (28c), a visual pitch tracking example is useful. First,
consider the first-person singular form shown in Figure 2.8 and its spectrogram in Figure
2.9:
Figure 2.8 [rí-látsu#$=a#$]
80
Figure 2.9 Spectrogram of [rí-látsu#$=a#$]
In this case the final syllable of the verb root and the first-person singular marker were
realized with creaky voice. As shown earlier in Figure 2.5, irregular glottal pulses cause
the pitch tracking algorithm to fail. In Figure 2.8 the pitch tracking line is not shown due
to the creaky voice. A spectrogram with an arrow to illustrate the irregular glottal pulses
is shown in Figure 2.9. The important detail to notice from this example is the elevated
pitch on the aspect marker /ri/ and the initial syllable of the verb root /la/. Compare
Figure 2.8 and Figure 2.9 with the second-person singular form shown in Figure 2.10:
81
Figure 2.10 [ri-la%tsu#$=lu%$]
In Figure 2.10, we can observe that there is no elevated pitch on the aspect marker or the
initial syllable of the verb root. Figure 2.8 and Figure 2.10 demonstrate visually that there
is a H tone in the first-person singular and that this H tone appears on the aspect marker
/ri/ in addition to the stressed syllable of the verb root.
Because the floating H appears on the completive and habitual aspect markers
only for the first-person singular, it must be the case that the completive and habitual
aspect markers do not have their own underlying tones, as I argue here. I propose that
since nothing is present to block the floating H of the first person singular, it can then
appear in this position as well as on the stressed syllable of the verb root. This
phenomenon is analyzed using the constraints I develop in chapter 3.
To further support my claim about toneless aspect markers and the floating H tone
of the first-person singular, consider verb class 6. For this class, only the habitual aspect
marker lacks an underlying tone. This verb class demonstrates exactly the same pattern
82
we have seen thus far: the floating H tone of the first-person singular shows up on the
toneless aspect marker as well as on the stressed syllable of the verb root. Here I will
present two different verbs from this class to illustrate the pattern.
First, consider the completive forms of the verb ‘to wait’
31
shown in examples
(29a) through (29d):
(29a) /u%-ledá/ [u%ledá] (29b) u%-lédá=$
COM-wait COM-wait=1s
‘waited’ ‘I waited’
(29c) u%-le%dá=lú$ (29d) u%-le%dá=bí
COM-wait=2s.INF COM-wait=3s.FAM
‘You waited’ ‘He waited’
In example (29b) we observe that the floating tone of the first-person singular appears
only on the stressed syllable of the verb root. For this verb class I posit that the
completive aspect marker /u%/ has an underlying M tone and that this explains why the
floating H tone fails to appear in this position.
Now compare the examples given in data set (30) These forms illustrate the
habitual form for this verb, along with the placement of the floating H tone in this aspect:
(30a) /ri–bedá/ [ribedá] (30b) rí-bédá=$
HAB-wait HAB-wait=1s
‘waits’ ‘I wait’
31
This verb is transitive. DP gave the Spanish translation as ‘esperar’ as in ‘wait for a
person’.
83
(30c) ri-bedá=lú$ (30d) ri-bedá=bí
HAB-wait=2s.INF HAB-wait=3s.FAM
‘You wait’ ‘He waits’
Example (30b) shows that both the habitual aspect marker and the stressed syllable of the
verb root have a H tone for the first-person singular. This must arise because the habitual
aspect marker lacks an underlying tone and is available for the floating tone. Note that in
the second and third-person singular forms for this verb both the aspect marker and the
stressed syllable of the verb root are realized with a mid-like pitch.
Finally, consider the potential aspect of this verb, shown in (31a) – (31d):
(31a) /k
w
+H + edá/ [k
w
édá] (31b) k
w
édá=$
POT/wait POT/wait=1s
‘will wait’ ‘I will wait’
(31c) k
w
édá=lú$ (31d) k
w
édá=bí
POT/wait=2s.INF POT/wait=3s.FAM
‘You will wait’ ‘He will wait’
I posit that the verb root is /edá/ and that the potential aspect is marked by /k
w
/ plus a H
tone. The consonant /k
w
/ combines with the vowel-initial verb root and the H tone
appears on the first syllable of the verb root. Another possibility is that the initial
consonant of the root is /w/ and the potential is marked by /k/. Further work is needed to
determine which analysis is correct. However, we can see that in (31b), (31c), and (31d),
since the potential marker has a H tone, we do not see any difference in the output
84
between the first-person singular and the second- or third-person singular forms. All three
have a H tone on the first syllable of the verb root.
A second example from this verb class shows exactly the same tonal pattern: the
potential aspect marker has an underlying H tone, the completive aspect marker has an
underlying M tone, and the habitual aspect marker is toneless. First, consider the verb ‘to
dig’. This verb also has some segmental variations in the root which will become clear
when we examine the underlying form for each aspect. In the completive aspect, the
floating H of the first-person singular appears only on the stressed syllable of the verb
root and forms a contour on this syllable. I discuss contour formation in detail in the
following section. Here my focus is on the fact that the floating H tone does not appear
on the aspect marker, but rather on the stressed syllable of the verb root. This is
illustrated in examples (32a) – (32d):
32
(32a) /u%-ta%"ní/ [u%ta%"ní] (32b) u%-ta!"ní=á$
COM-dig COM-dig=1s
‘dug’ ‘I dug’
(32c) u%-ta%"ní=lú$ (32d) u%-ta%"ní=bí
COM-dig=2s.INF COM-dig=3s.FAM
‘You dug’ ‘He dug’
In (32b) I posit that the floating H of the first-person singular has combined with the
underlying M tone of the verb root to form a falling contour tone. This falling tone is not
present in examples (32c) and (32d), which is evidence that (32b) is special because of
32
Note that in the habitual, the verb root begins with /g/.
85
the first-person singular clitic. Again, I assume that the floating tone does not appear on
the completive aspect marker because this syllable already has a specified M tone. Note
also that for the completive aspect, the verb root begins with the consonant /t/. Example
(33) shows that in the habitual aspect the verb root begins with /g/.
In the habitual aspect of this verb, we can see the same pattern as was apparent in
earlier examples: when an aspect marker is toneless, the floating tone appears there as
well as on the stressed syllable of the verb root. This is illustrated in (33a) – (33d):
(33a) /ri –ga%"ní/ [riga%"ní] (33b) rí-ga!"ní=á$
HAB-dig HAB-dig=1s
‘digs’ ‘I dig’
(33c) ri-ga%"ní=lú$ (33d) ri-ga%"ní=bí
HAB-dig=2s.INF HAB-dig=3s.FAM
‘You dig’ ‘He digs’
The example in (33b) shows that there is a falling tone on the stressed syllable of the verb
root in addition to a H tone on the habitual aspect marker /ri/. This illustrates that the
floating H is attracted to both positions. This happens regardless of whether there is an
underlying tone in that position or not (as long as the stressed syllable is long). The
floating H tone will also appear on a toneless aspect marker and become word initial
where possible, as illustrated in this data set.
Finally, consider the potential aspect of the verb ‘to dig’. For this verb class, the
potential aspect is marked by the consonant /k/ and a H tone. The /k/ will attach to the
verb root and appear in the initial syllable along with the H tone of the potential aspect.
86
As illustrated in examples (34a) –(34d), this creates a contour tone for the first, second,
and third-person singular:
(34a) /k - H – a"%ní/ [ka!"ní] (34b) ka!"ní=á$
POT/dig POT/dig=1s
‘will dig’ ‘I will dig’
(34c) ka!"ní=lú$ (34d) ka!"ní=bí
POT/dig=2s.INF POT/dig=3s.FAM
‘You will dig’ ‘He will dig’
All pronominally inflected forms have a falling contour in the potential aspect for this
verb. I argue that this is due to the combination of the H of the potential with the
underlying M tone of the verb root. As illustrated in (34b), the stressed syllable of this
particular verb shows no difference in the output between the first-person singular form
and the forms for the other pronouns.
From these examples I conclude that for verb class 6, only the habitual aspect
marker is toneless. We have seen that the floating tone of the first-person singular will
appear on this toneless aspect marker in addition to the stressed syllable of the verb root.
We saw previously that for verb class 1, only the habitual aspect marker is toneless, while
for verb class 4, both the habitual and completive aspect markers are toneless. From these
examples, I conclude that whenever there is a toneless aspect marker preceding the verb
root, the floating tone of the first-person singular will appear in this location in addition
to on the stressed syllable of the verb root. For an exhaustive exemplification for other
87
verb classes, see the appendix, and for underlying representations of the aspect markers,
see chapter 1, Table 1.2.
2.4.2 Contour Tones as Result of Floating H
In this section I present examples that show how and when contour tones are
formed on the stressed syllable of the verb root. This only occurs in 8 out of the 43 verbs
in the database but merits discussion because it is only the first-person singular forms
which demonstrate these falling tones.
33
Presumably the falling tones in this location are
due to the effects of the floating H tone associated with the first-person singular
morpheme. These contours only form when the underlying tone of the stressed syllable of
the verb root is M or L, and only when the syllable is long. The following paragraphs will
give examples of verbs that exhibit contour tones when the floating H is added, as well as
examples where a contour tone does not form. In each case I will explain why I posit the
contour is permitted or blocked.
The floating H of the first-person singular combines with an underlyingly
specified M tone. In such cases the stressed syllable of the verb root has a falling contour
for the first-person singular, as evidenced by the verb ‘to call’, which belongs to class 7.
The completive aspect of the verb is illustrated in examples (35a) – (35d):
33
Recall that falling tones can occur lexically – the falling tones mentioned here in
relation to the first person singular have to do with alternations in the stressed syllable.
88
(35a) /u%-ta%"'i#/ [u%ta%"'i#] (35b) u%-ta!"'i#=a#$
COM-call COM-call=1s
‘called’ ‘I called’
(35c) u%-ta%"'i#=lu%$ (35d) u%-ta%"'i#=bí
COM-call=2s.INF COM-call=3s.FAM
‘You called’ ‘He called’
Example (35b) shows that the floating H of the first-person singular has combined with
the underlying M tone of the verb root to form a falling tone. Examples (35c) and (35d)
both lack a contour tone in this position, and this is what tells us that the floating H is
present in example (35b). As in previous examples, the floating tone is absent on the
completive aspect marker because this syllable already has an underlying tone. A picture
of the pitch track for example (35b) is found in Figure 2.11:
Figure 2.11 [u%-ta!"'i#=a#$]
89
Figure 2.11 illustrates the presence of the falling contour. Figure 2.12 shows the pitch
contour for the second-person singular, where there is no contour on the stressed syllable
of the root:
Figure 2.12 [u%-ta%"'i#=lu%$]
Figure 2.11 and Figure 2.12 support the conclusion that only the first-person singular
form emerges with a falling contour on the stressed syllable of the verb root.
The presence of the floating tone is also apparent when we consider the habitual
aspect for this same verb. This is illustrated in examples (36a) through (36d):
34
(36a) /r-a%"'i#/ [ra%"'i#] (36b) ra!"'i#=a#$
HAB/call HAB/call=1s
‘calls’ ‘I call’
34
In the present aspect the verb root is vowel-initial, as shown here.
90
(36c) ra%"'i#=lu%$ (36d) ra%"'i#=bí
HAB/call=2s.INF HAB/call=3s.FAM
‘You call’ ‘He calls’
For this verb class the habitual aspect marker is just /r/, which then combines with the
vowel-initial verb root. Example (36b) shows that the floating tone of the first-person
singular has combined with the specified M tone of the verb root to form a falling
contour. This contour tone is absent in the second and third-person forms.
Finally, consider the potential aspect, where all the pronouns have a falling
contour in the first syllable. I argue that this occurs because the H tone of the potential
combines with the underlying M tone of the initial syllable of the verb root. The potential
forms are illustrated in (37a) – (37d):
(37a) /g -H-a%"'i#/ [ga!"'i#] (37b) ga!"'i#=a#$
POT/call POT/call=1s
‘will call’ ‘I will call’
(37c) ga!"'i#=lu%$ (37d) ga!"'i#=bí
POT/call=2s.INF POT/call=3s.FAM
‘You will call’ ‘He will call’
A comparison of (37b) with (37c) and (37d) shows that the first-person singular is no
different on its initial syllable than the other persons. All of the pronouns for this verb
have a falling tone on the initial syllable. To illustrate this visually, I include figures that
compare the first-person singular with the second person singular.
91
Figure 2.13 [ga!"'i#=lu%$]
Figure 2.13 shows that the second-person singular form has a falling contour on the
initial syllable of the verb root. As I argued above, this is due to the H tone of the
potential combining with the underlying M tone of the verb root. The same pattern
emerges in the first-person singular, shown in Figure 2.14:
92
Figure 2.14 [ga!"'i#=a#$]
Verb class 1 also contains examples of contours that can form when the floating H
of the first-person singular attaches. Here I present two of these examples. The first is
from the verb ‘to cover’, and in this case a floating H combines with an underlying L tone
to form a falling contour tone. When the floating H combines with an underlying L, a
falling contour results which is quite similar to the falling contour that resulted from the
combination of a floating H tone with an underlying M tone. The potential aspect is
shown in data set (38):
(38a) /gú -di-ba#"gá$/
35
[gúdíba#"gá$] (38b) gú-dí-ba!"gá$
POT-CAUS-cover POT-CAUS-cover=1s
‘will cover’ ‘I will cover’
35
The intransitive of this same verb is /í-ba#"gá$/, meaning ‘cover oneself’ – for example
with a blanket, jacket, etc. Used in the example above, it means to cause something to
be covered.
93
(38c) gú-dí-ba#"gá$=lu%$ (38d) gú-dí-ba##"gá$=bí
POT-CAUS-cover=2s.INF POT-CAUS-cover=3s.FAM
‘You will cover’ ‘He will cover’
The tone of the causative marker /di/ is discussed in more detail in a following section.
Here I assume that it is underlyingly toneless as shown in (38a), and that it becomes H in
examples (38b), (38c) and (38d) due to spreading from the potential aspect marker /gú/,
which is also H. Example (38b) also shows a contour tone on the stressed syllable of the
verb root. I assume this contour tone results when the floating H tone becomes affiliated
with a syllable that also presents a L tone. The fact that the stressed syllable of the verb
root is actually L in its underlying form can be seen by comparing the examples of the
second and third-person singular with the first-person singular.
To illustrate the difference between the first-person singular and the second-
person singular, compare the pitch tracks of these two forms:
94
Figure 2.15 [gú-dí-ba!"gá$]
Figure 2.15 illustrates that the causative marker /di/ is H and that the stressed syllable of
the verb root has a falling contour tone. Compare this figure with Figure 2.16:
Figure 2.16 [gú-dí-ba#"gá$=lu%$]
95
Figure 2.16 shows that while the syllable /di/ is high in the second-person singular, the
stressed syllable of the verb root has a L tone with no contour.
A similar pattern emerges in the completive aspect, where the floating tone
appears on the stressed syllable of the verb root to form a falling tone in the first-person
singular. This is illustrated in examples (39a) through (39d):
(39a) /be%-di-ba#"gá$/ [be%di%ba#"gá$] (39b) be%-di%-ba!"gá$
COM-CAUS-cover COM-CAUS-cover-1s
‘covered’ ‘I covered’
(39c) be%-di%-ba#"gá$=lu%$ (39d) be%-di%-ba#"gá$=bí
COM-CAUS-cover=2s.INF COM-CAUS-cover=3s.FAM
‘You covered’ ‘He covered’
Example (39b) shows that the floating H tone of the first-person singular has combined
with the underlying L tone to form a falling contour in the first-person singular but not in
the other forms. These examples also show that the causative marker /di/, which is
toneless itself as shown in (39a) will take whatever tone is to its immediate left. In the
examples in data set (39), /di/ has a M tone, presumably because the completive aspect
marker which precedes it is also mid-toned and this M tone has spread rightward. Recall
that earlier examples from this chapter showed that /be%/ has a M tone in this verb class.
Examples (39b) and (39c) are compared visually in the following figures. First, observe
the contour tone in the first-person singular form:
96
Figure 2.17 [ be%-di%-ba!"gá$]
Figure 2.17 illustrates that the completive aspect marker and the causative /di/ are
realized with a M tone, but that the stressed syllable of the verb root has a falling contour
tone in the first-person singular form. Compare this with Figure 2.18, which illustrates
the second-person singular form:
Figure 2.18 [be%-di%-ba#"gá$=lu%$]
97
Figure 2.18 shows there is no H tone on the stressed syllable of the verb root in the
second-person singular form.
In the habitual aspect, of this same verb, ‘to cover’, the floating H tone is realized
on the habitual aspect marker, the causative marker, and on the stressed syllable of the
root. Observe these examples in data set (40):
(40a) /ru-di-ba#"gá$/ [rudiba#"gá$] (40b) rú-dí-ba!"gá$
HAB-CAUS-cover HAB-CAUS-cover=1s
‘covers’ ‘I cover’
(40c) ru-di-ba#"gá$=lu%$ (40d) ru-di-ba#"gá$=bí
HAB-CAUS-cover=2s.INF HAB-CAUS-cover=3s.FAM
‘You cover’ ‘He covers’
Example (40b) suggests that the floating H tone of the 1 sg. is initiated with the
beginning of the word, and that it persists until it reaches the underlying L tone of the
verb root. Example (40b) also exhibits a contour tone on the stressed syllable of the verb
root. Since this verb belongs to class 1, it should show the floating tone on the habitual
aspect marker because this aspect marker is toneless. These patterns are illustrated in
Figure 2.19 below.
98
Figure 2.19 [rú-dí-ba!"gá$]
Figure 2.19 shows that in the habitual aspect, the floating H tone is realized on the initial
syllable /ru/. It also appears that /di/ has a H tone, and that there is a falling contour on
the stressed syllable of the root. Each successive high syllable is lower than the preceding
one. Compare this with the example in Figure 2.20:
Figure 2.20 [ru-di-ba#"gá$=lu%$]
99
Figure 2.21 Spectrogram of [ru-di-ba#"gá$=lu%$]
Figure 2.20, and Figure 2.21, which illustrate the pitch contour and spectrogram for the
second-person singular, indicate that there is not a contour on the stressed syllable of the
verb root. The spectrogram is included in Figure 2.21 because of slight creakiness on the
final syllable of the verb root and on the second-person singular clitic. These periods
have been marked with arrows in Figure 2.21. Both the initial syllable of the word and
the causative marker /di/ are realized with a pitch that is neither H nor L.
Another verb with a similar tonal pattern is ‘to dry (something)’. This verb also
has the causative marker, /di/. Here we see the same pattern on the first-person singular
that we observed for data set (40) for the verb ‘to cover’. The potential aspect of the verb
‘dry (something)’ is shown in data set (41):
(41a) /gú-di-bi%"tsi% / [gúdíbi%"tsi] (41b) gú-dí-bí #"tsi%=a#$
POT-CAUS-dry POT-CAUS-dry=1s
‘will dry (it)’ ‘I will dry (it)’
100
(41c) gú-dí-bi%"tsi%=lu%$ (41d) gú-dí-bi%"tsi%=bí
POT-CAUS-dry=2s.INF POT-CAUS-dry=2s.FAM
‘You will dry (it)’ ‘He will dry (it)’
The contour that appears in example (41b) suggests that the stressed syllable of the verb
root has an underlying M tone. When the floating H tone combines with this underlying
M tone, a falling contour is formed. Here I assume that the floating H tone of the first-
person singular cannot appear on the potential aspect marker because the potential aspect
marker already has a specified H tone. If I assumed that the floating H tone of the first-
person singular replaced the underlying H tone of the potential aspect, then I would have
no explanation as to why this does not occur in the completive aspects that we have seen
above. As with previous examples that I have presented, I assume that the causative
marker /di/ is high-toned due to spreading from the potential aspect marker /gú/. This is
because /di/ is high in the second and third-person forms and not just in the first-person
singular.
A similar pattern emerges for the completive aspect, where it is only the first-
person singular form that has a contour on the stressed syllable of the verb root. In this
data set, the causative marker /di/ is mid-toned due to spreading from the aspect marker.
This is illustrated in data set (42):
(42a) /be%-di-bi%"tsi%/ [be%di%bi"%tsi%] (42b) be%-di%-bí #"tsi%=a#$
COM-CAUS-dry COM-CAUS-dry=1s
‘dried (it)’ ‘I dried (it)’
101
(42c) be%-di%-bi"%tsi%=lu%$ (42d) be%-di%-bi"%tsi%=bí
COM-CAUS-dry=2s.INF COM-CAUS-dry=2s.FAM
‘You dried (it)’ ‘He dried (it)’
Example (42b) shows that the floating H is realized only in to the stressed syllable of the
root, and, moreover, that when it attaches in this location it forms a contour by combining
with the underlying lexical tone. The second and third-person singular both lack a
contour tone on the stressed syllable. The data show that the underlying M tone of the
completive aspect marker /be%-/ blocks the floating H tone from appearing word initially,
and therefore it must appear only on the stressed syllable of the verb root. This verb also
shows a case where a floating H combined with the underlying M gives a falling contour.
A comparison of the first and second-person singular forms are given in the following
figures. Figure 2.22 shows the first-person singular form, and Figure 2.23 gives the
spectrogram to confirm the final two syllables (which pitch-tracked poorly due to
voicelessness and creakiness). As in previous cases, arrows mark irregular or absent
vocal fold pulses in the spectrogram. Figure 2.24 shows the second-person singular form.
102
Figure 2.22 [be%-di%-bí #"tsi%=a#$]
Figure 2.23 Spectrogram of [be%-di%-bí #"tsi%=a#$]
103
Figure 2.24 [be%-di%-bi"%tsi%=lu%$]
Finally, by examining the habitual aspect of this same verb we can notice that the
floating H tone will appear on the toneless aspect marker /ru/, as well as on the stressed
syllable of the verb root. I posit that the causative marker /di/ is also H due to spreading
of the H tone from the initial syllable. This is demonstrated in data set (43):
(43a) /ru -di -bi%"tsi%/ [rudibi"%tsi%] (43b) rú-dí-bí #"tsi%=a#$
HAB-CAUS-dry HAB-CAUS-dry=1s
‘dries (it)’ ‘I dry (it)’
(43c) ru-di-bi"%tsi%=lu%$ (43d) ru-di-bi"%tsi%=bí
HAB-CAUS-dry=2s.INF HAB-CAUS-dry=2s.FAM
'You dry it’ ‘He dries (it)’
The floating tone of the first-person singular is found in example (43b) but is absent in
examples c) and d) in this set. Example (43b) illustrates that the floating H will appear on
104
the initial syllable, on the causative marker, and on the stressed syllable of the verb root. I
posit that in this case the floating H is present on the initial syllable since there is no tone
marker in this position to block it, and that the tone gesture then stays active through the
stressed syllable to form a contour. This type of rightward spreading behavior is
discussed in more detail in the analysis chapter.
2.4.3 No Contour on Short Vowel
In some cases, such as when the stressed syllable of the verb root contains a short
vowel, a contour tone does not form when the floating tone of the first-person singular is
added to a verb stem. This is the case even when there is a specified tone underlying on
the stressed syllable of the root, and in these cases the floating H tone is realized in place
of the verb’s underlying tone. I illustrate this in data set (44), which shows the potential
form of the verb ‘leave (something)’.
(44a) /ú-ka#$ná / [úka#$ná] (44b) ú-ká$ná=$
POT-leave POT-leave.sthg=1s
‘will leave (it)’ ‘I will leave (it)’
(44c) ú-ka#$ná=lú$ (44d) ú-ka#$ná=bí
POT-leave.sthg=2s.INF POT-leave.sthg=3s.FAM
‘You will leave (it)’ ‘He will leave (it)’
The output next to example (44a) shows that the stressed syllable of the verb root has an
underlying L tone. Examples (44c) and (44d) verify the same fact. However, when the
floating H of the first-person singular attaches, as shown in example (44b), the stressed
105
syllable of the verb root becomes H instead of forming a contour. I claim that this
happens because contour tones are blocked from occurring on short vowels. The same
pattern appears for this verb in both its completive and habitual aspects, as I will illustrate
in the following two data sets. The potential aspect is illustrated in (45):
(45a) /be% -ka#$ná / [be%ka#$ná] (45b) be%-ká$ná=$
COM-leave COM-leave.sthg=1s
‘left (it)’ ‘I left (it)’
(45c) be%-ka#$ná=lú$ (45d) be%-ka#$ná=bí
COM-leave.sthg=2s. INF COM-leave.sthg=3s.FAM
‘You left (it)’ ‘He left (it)’
For example (45b) we would have predicted a falling contour if the stressed syllable were
open and long, but instead see a simple H tone is the result of adding the floating H tone.
In example (45b), I posit that the floating H tone does not appear on the completive
aspect marker because this syllable already has an underlying tone of its own. Finally, we
can observe both the blocking of the contour tone on the stressed syllable as well as the
appearance of the floating H tone on the toneless aspect marker in example data set (46):
(46a) /ru -ka#$ná / [ruka#$ná] (46b) rú-ká$ná=$
HAB-leave HAB-leave.sthg=1s
‘leaves (it)’ ‘I leave (it)’
(46c) ru-ka#$ná=lú$ (46d) ru-ka#$ná=bí
HAB-leave.sthg=2s. INF HAB-leave.sthg=3s.FAM
‘You leave (something)’ ‘He leaves (something)’
106
Example (46b) shows the pattern which is by now familiar: the floating H tone appears
on /ru/ in the first-person singular because this aspect marker is toneless. It also replaces
the underlying L tone of the stressed syllable of the verb root. No contour is formed since
the vowel is short.
Another verb which shows the same pattern is “to enter”.
36
This verb has a
floating H tone for the first-person singular, and the underlying tone of the first syllable
in the verb root is L. The completive aspect is illustrated in data set (47):
(47a) /gu% - ta#$a%/ [gu%ta#$a] (47b) gu% - tá$=a#$
37
COM/enter COM/enter=1s
‘entered’ ‘I entered’
(47c) gu% - ta#$a%=lu%$ (47d) gu% -ta#$a%=bí
COM/enter=2s.INF COM/enter=3s.FAM
‘You entered’ 'He entered’
Here I posit that the underlying form of the verb root in the completive aspect is /ta#$a%/, as
shown in (47a), and that the first syllable receives the primary prosodic stress. Since the
stressed syllable contains a short vowel, no contour is formed when the floating H of the
first-person singular is realized. This is shown in example (47b), where the stressed
syllable has a H tone.
36
According to DP, this verb can be used in a transitive sense with the meaning ‘to enter
a doorway’ or ‘enter a car’. As a transitive, it has a floating H tone.
37
As noted also by Bartholomew (1983), when the 1 sg enclitic attaches to a verb root
ending in the sequence V$V, the final vowel of the verb root is lost.
107
For the habitual aspect, this verb class has only /r/ as its aspect marker. Consider
the examples below in data set (48):
(48a) /r - a#$a% / [ra#$a] (48b) r-á$=a#$
HAB/enter HAB/enter=1s
‘enters’ ‘I enter’
(48c) r-a#$a%=lu%$ (48d) r- a#$a%=bí
HAB/enter=2s.INF HAB/enter=3s.FAM
‘You enter’ ‘He enters’
For the habitual aspect, we see exactly the same pattern as we did for the completive,
with the exception that the habitual aspect marker is not associated with a H tone. Thus
the /r/ appears in the onset of the initial syllable of the verb root as shown in (48c) and
(48d). For the second-and third-person singular, the initial syllable of the root remains L.
The first-person singular form shown in (48b) has replaced the L tone of the root with the
floating H of the first-person singular. I posit that a contour is not formed in this case
because the initial vowel of the verb root is short.
Finally, consider the potential aspect of this verb which is illustrated in (49a) –
(49d). For this class, the potential aspect marker is /g/ and a H tone, as I have illustrated
in (49a). For all the forms in data set (49), the H tone of the potential replaces the
underlying lexical tone. Both the H tone of the potential and the consonant /g/ appear in
the first syllable of the verb root.
108
(49a) /g - H - a#$a% / [gá$a%] (49b) g-á$=a#$
POT/enter POT/enter=1s
‘will enter’ ‘I will enter’
(49c) g-á$a%=lu%$ (49d) g-á$a%=bí
POT/enter=2s.INF POT/enter=3s.FAM
‘You will enter’ ‘He will enter’
In the potential aspect, we do not see a difference in the output form of the stressed
syllable between the first-person singular and the other two forms shown in (49c) and
(49d). I posit that the H tone of the potential fails to form a contour tone in examples
(49c) and (49c) because the initial syllable of the verb root contains a short vowel. The H
tone associated with the potential aspect then replaces the underlying (L) lexical tone.
The examples I have presented for this verb illustrate that the floating H of the
first-person singular will not combine with an underlying tone to form a contour when the
vowel on which it appears is short. When the vowel of the stressed syllable is short, the
floating H of the first-person singular appears and no signs of the underlying lexical tone
are left. As I showed in data set (49), a contour also fails to form due to the effect of the
H tone of the potential if the syllable where it is realized is short. Thus we can expect that
in verb classes with only a consonant and H tone for the potential marking (but no
vowel), the H tone of the potential will form a contour on the vowel where it appears if
this vowel is long, but will delete the lexical tone in cases where the vowel is short.
Recall that in the examples shown in (37), an aspect marker with a H tone attached to a
vowel-initial verb root with a long vowel to form a contour.
109
The previous examples have shown cases where I posit that the stressed syllable
of the verb root contains a short vowel, and that this blocks the formation of contours
with the floating H of the first-person singular or the H of the potential. To illustrate that
this also happens when the stressed syllable is not followed by a glottal stop, I present the
data in (50).
38
Notice here that the consonant /ll/, a fortis consonant according to
Bartholomew, is one that may begin a syllable, as evidenced by the many lexical items
that have this as their initial consonant (1983" 162-164).
(50a) /tá - ne#llé / [táne#llé] (50b) tá-néllí=á$
POT-visit POT-visit=1s
‘will visit ‘I will visit’
(50c) tá-ne#llé=lú$ (50d) tá-ne#llé=bí
POT-visit=2s.INF POT-visit=3s.FAM
'You will visit' 'He will visit'
Examples (50a), (50c), and (50d) show that the underlying tone of the stressed syllable of
the verb root is L. These examples also show that the H tone of the potential aspect
marker does not spread rightward onto the first syllable of the verb root. (This differs
from the previous example in (49a) – (49d), where I showed that when the potential
aspect marker has no vowel of its own, the H tone of the potential aspect marker is
realized on the verb root). Example (50b) demonstrates that for the first-person singular,
the floating H tone replaces the underlying L tone of the verb root. I posit that this occurs
38
Here I adopt Bartholomew’s segmentation of the aspect marker and root (1983: 387).
An alternate analysis would be that /ta/ is part of the stem and the aspects are indicated
solely by tone. Future work should examine additional verbs from this class.
110
since the vowel is short. If the vowel were long, I would predict the formation of a
contour instead.
The same pattern emerges for the completive aspect shown in data set (51)"
(51a) /ta% - ne#llé/ [ta%ne#llé] (51b) ta%-néllí=á$
COM-visit COM-visit=1s
‘visited’ ‘I visited’
(51c) ta%-ne#llé=lú$ (51d) ta%-ne#llé=bí
COM-visit=2s.INF COM-visit=3s.FAM
'You visited' 'He visited'
Notice that for the completive aspect shown here, the segmental part of the aspect marker
is the same as the potential, it just differs in tone. The completive marker is mid-toned,
while the potential marker is high-toned. Example (51b) shows that the floating H of the
first-person singular replaces the underlying L tone and that no contour is formed. In the
habitual aspect, the same pattern emerges, with the exception that the floating tone of the
first-person singular also appears on the toneless aspect marker. This is illustrated in data
set (52)"
(52a) /rita - ne#llé / [ritane#llé] (52b) rítá-néllí=á$
HAB-visit HAB-visit=1s
‘visits’ ‘I visit’
(52c) rita-ne#llé=lú$ (52d) rita-ne#llé=bí
HAB-visit=2s.INF HAB-visit=3S.FAM
‘You visit’ ‘He visits’
111
Example (52b) shows that the floating H of the first-person singular is initiated with the
beginning of the word. It appears on both syllables of the aspect marker (for this class it
is /rita/). The floating H also appears on the stressed syllable of the root and replaces the
underlying L tone. Examples (52a), (52c) and (52d) illustrate that the stressed syllable of
the verb root is L, and that the habitual aspect marker /rita/ is toneless in the second and
third-person singular.
The verb /kka%bí/ ‘to answer (it)’ is an example where a contour fails to form on an
open syllable that has a short vowel. I include the following examples to illustrate that
contours fail to form in cases other than when a fortis consonant (like /ll/) or a glottal stop
follow the vowel in question. First, consider the potential aspect illustrated in (53).
(53a) /é- kka%bí/ [ékka%bí] (53b) é-kkábí=á$
39
POT-answer (it) POT-answer=1s
‘will answer (it)’ ‘I will answer (it)’
(53c) é-kka%bí=lú$ (53d) é-kka%bí=/
POT-answer=2s.INF POT-answer=3S.FAM
‘You will answer (it)’ ‘He will answer (it)’
In the potential aspect examples shown in (53), we can see that the underlying
tone of the verb root's initial syllable is M. Therefore no spreading occurs from the H-
toned potential aspect marker. Of the three examples shown, only (53b), the first-person
singular, shows a H tone on the stressed syllable of the verb root. Importantly, no contour
is formed in this case. I argue that this happens because the vowel is short and cannot
39
Spreading of H tones from the final vowel of the verb root onto clitics is discussed in
section 2.5.1.
112
support a contour tone. This verb also illustrates an OCP type effect, in that the third-
person singular familiar morpheme is not realized, as shown in (53d). I posit that this
occurs so that speakers can avoid two /bí/ sequences in a row.
The completive aspect is shown in (54)"
(54a) be%-kka%bí [be%kka%bí] (54b) be%-kkábí=á$
COM-answer (it) COM-answer=1s
‘answered (it)’ ‘I answered (it)’
(54c) be%-kka%bí=lú$ (54d) be%-kka%bí=/
COM-answer=2s.INF COM-answer=3S.FAM
‘You answered (it)’ ‘He answered (it)’
Data set (54) illustrates the same pattern as (53), where only the first-person singular
example has a H tone on the stressed vowel of the verb root. As (54b) illustrates, no
contour is formed. I argue that this results when the H of the first-person singular replaces
the underlying lexical tone. Finally, consider the habitual aspect shown in (55)"
(55a) re-kka%bí [rekka%bí] (55b) ré-kkábí=á$
HAB-answer (it) HAB-answer=1s
‘answers (it)’ ‘I answer (it)’
(55c) re-kka%bí=lú$ (55d) re-kka%bí=/
HAB-answer=2s.INF HAB-answer=3S.FAM
‘You answer (it)’ ‘He answers (it)’
Data set (55) shows the same pattern with the exception that the habitual aspect marker is
toneless. Therefore the floating H of the first-person singular appears on it as well as on
113
the first syllable of the verb root , as shown in (55b). For the other persons illustrated
here, the aspect marker is realized with a mid-like pitch although it has no tone.
These examples have shown that when the verb root has a stressed syllable that
contains a short vowel, the floating H tone of the first-person singular will not form a
contour tone but instead will replace the underlying lexical tone. As I argue in the
analysis chapter, I posit that this results from a grammatical requirement to preserve the
underlying lexical tone if possible. When this is not possible, then the lexical tone is
deleted in favor of the floating H tone of the first-person singular.
As expected, contour tones also do not form in cases where there is no underlying
tone on the stressed syllable of the verb root. Some examples of this pattern were
presented at the beginning of the chapter and for this reason I will illustrate this point
only briefly here. I also demonstrate that toneless syllables undergo spreading effects
from tones which precede them.
In the verb 'to fold', I posit that the first syllable of the root is toneless. To
illustrate this I will first present examples from the habitual aspect. These are shown in
(56)"
(56a) /ru - det(u% / [rudet(u%] (56b) rú-dét(
w
=a#$
HAB-fold HAB-fold=1s
‘folds’ 'I fold’
(56c) ru-det(u%=lu%$ (56d) ru-det(u%=bí
HAB-fold=2s.INF HAB-fold=3s.FAM
'You fold' ‘He folds’
114
Again, only the first-person singular has a H tone. In example (56b) we can see that this
H tone appears on both the stressed syllable of the verb root and on the toneless aspect
marker. I take the lack of a H tone on the /de/ syllable in (56c) and (56d) as evidence that
this syllable is toneless and that it only becomes high-toned in the presence of the first-
person singular floating H tone, or due to spreading from a preceding H tone such as the
potential aspect marker.
Evidence that the initial syllable of the verb root is toneless comes when we
examine the potential and completive aspects. Data set (57) shows that the toneless initial
syllable of the verb root becomes high-toned due to spreading from the H tone of the
potential aspect marker. This means that the first, second and third-person singular forms
have a H tone on the stressed syllable of the verb root.
(57a) /gú - det(u%/ [gúdét(u%] (57b) gú-dét(
w
=a#$
40
POT-fold POT-fold=1s
‘will fold’ ‘I will fold’
(57c) gú-dét(u%=lu%$ (57d) gú-dét(u%=bí
POT-fold=2s.INF POT-fold=3s.FAM
'You will fold' 'He will fold'
Example (57a) shows that the underlying form of the potential aspect marker is /gú/, and
that the verb root has an initial syllable which is toneless. As shown in (57b), (57c) and
40
This example also shows that the first-person singular clitic causes a change to the verb
root – the vowel /u/ is reduced to just rounding on the consonant /t(/. It would be
interesting in the future to analyze all of the segmental changes which occur when the
clitics attach to the verb root.
115
(57d), all three pronouns have a H tone on the initial syllable of the verb root. I posit that
this H tone results from rightward spreading from the potential aspect marker onto the
toneless syllable of the verb root for the second and third-person singular. We would
have expected the first-person form shown in (57b) to have a H tone on the stressed
syllable of the verb root due to the floating H tone. To illustrate the spreading from the
potential aspect marker onto the verb root, I present the pitch track of the second-person
singular form. This is given in Figure 2.25"
Figure 2.25 [gú-dét(u%=lu%$]
While the pitch of the syllable /de/ declines, it is still much higher than the final two
syllables, which are mid-toned. This is evidence that the initial syllable of the verb root is
toneless and that the H tone of the potential marker has spread rightward.
116
Although the first, second, and third-person singular forms in data set (57) all
showed a H tone on the stressed syllable of the verb root, the situation differs in the
completive aspect. For data set (58a) –(58d), I posit that the other person markers have a
M tone on the initial syllable of the verb root due to spreading from the mid-toned aspect
marker, while the first-person singular has a H tone on the stressed syllable due to the
floating H.
(58a) /be% - det(u% / [be%de%t(u%] (58b) be%-dét(
w
=a#$
COM-fold COM-fold=1s
‘folded ‘ ‘I folded’
(58c) be%-de%t(u%=lu%$ (58d) be%-de%t(u%=bí
COM-fold=2s.INF COM-fold=3s.FAM
'You will fold' 'He will fold'
(58a) – (58d) show that the floating H of the first-person singular will appear on the
toneless syllable of the verb root. The second and third-person forms have a M tone on
this syllable, presumably due to spreading from the M-toned aspect marker /be%/. This
suggests that the realization of the floating H tone takes priority over tone spreading.
I have shown in this section that the floating H can only form a contour tone when
the underlying syllable of the verb root is long, and when it possesses a specified tone. If
these conditions are not met, then the floating H replaces the underlying tone if there is
one. If there is no underlying tone present in the input, then the floating H simply appears
on the stressed syllable of the root.
117
2.4.4 No Change When Verb Root is Already H
Up to this point it has been shown that at times the floating H tone of the first-
person singular is realized but does not induce a change in the output, such as when the
potential aspect marker appears on a vowel-initial verb root for all pronouns. Here I will
present and comment on additional examples where the floating tone of the first-person
singular is not immediately obvious. This occurs when the stressed syllable of the verb
root has an underlying H tone. In these cases there is no change for the first-person
singular form. The stressed syllable of the root does not have an exaggerated high pitch.
This suggests that for the first-person singular, speakers do not make an effort to raise a
high pitch even higher than a H tone that is underlying affiliated with the stressed
vowel.
41
We know that the floating tone is present, because it will appear on the toneless
aspect marker just as it does for other verbs whose stressed syllable has a tone other than
H. This is illustrated in the verb 'return something'. The habitual aspect is given in (59)"
(59a) /re - júnna%/ [rejúnna%] (59b) ré-júnna%=$
HAB-return HAB-return.sthg=1s
‘returns (it)’ ‘I return (it)’
(59c) re-júnna%=lu%$ (59d) re-júnna%=bí
HAB-return.sthg=2s.INF HAB-return.sthg=3s.FAM
‘You return (it)’ ‘He returns (it)’
41
Note that Beam de Azcona (2004) reports extra-H pitch in some instances due to the
effects of the first-person singular morpheme and potential aspect in Coatec Zapotec. She
refers to this as ‘upstep’.
118
Example (59b) shows that the first-person singular has a H tone on the toneless aspect
marker and on the stressed syllable of the verb root. I assume that the floating tone is
realized on the verb root, even though it is not distinct from the underlying lexical H tone.
These forms give another example demonstrating that when toneless aspect markers are
available, they too will realize the floating H tone of the first-person singular.
In the completive and potential aspects, there is no difference between the first-
person singular and the other pronouns, except for the enclitics. The tones of the aspect
marker and initial syllable of the verb root are the same for all three. This is illustrated in
data set (60):
(60a) /é - júnna% / [éjúnna%] (60b) é-júnna%=$
POT-return POT-return.sth=1s
‘will return (it)’ ‘I will return (it)’
(60c) é-júnna%=lu%$ (60d) é-júnna%=bí
POT-return.sthg=2s.INF POT-return.sthg=3s.FAM
‘You will return (it)’ ‘He will return (it)’
(60e) /be% - júnna% / [be%júnna%] (60f) be%-júnna%=$
COM-return COM-return.sthg=1s
‘returned (it)’ ‘I returned (it)’
(60g) be%-júnna%=lu%$ (60h) be%-júnna%=bí
COM-return.sth=2s.INF COM-return.sthg=3s.FAM
‘You returned (it)’ ‘He returned (it)’
In (60a) – (60d), all three pronouns have a H tone on the potential aspect marker and a H
tone on the first syllable of the verb root. Likewise, in (60e) through (60h), all three
119
pronouns have a M tone on the completive aspect marker and a H tone on the first
syllable of the verb root. In Figure 2.26 and Figure 2.27, below, I give a comparison of
the first-person singular and the second-person singular in the potential aspect. This
shows that the first-person singular does not have an extra-high tone on its stressed
syllable.
Figure 2.26 [é-júnna%=$]
120
Figure 2.27 [é-júnna%=lu%$]
From this example we may conclude that in verbs with a specified H tone on the
stressed syllable of the root, no change in the output realization will occur when the
floating H of the first-person singular attaches. For some verbs the floating tone does not
induce an alternation for tone in the stressed syllable in the potential aspect, whereas for
verbs with an underlying H tone on the first syllable of the root, the floating tone does not
produce tonal alternation in the potential, habitual, or completive aspect. For these verbs,
it is only when they happen to occur with a toneless aspect marker that any evidence is
available in the output for the floating tone of the first-person singular.
2.4.5 Toneless Causative Marker /di/
I posit that the causative marker /di/ is toneless in SJZ. In the examples I comment
on here as well as in others located in the appendix, /di/ has the same tone as the syllable
which precedes it. I take this as evidence that /di/ is toneless and that its tone is acquired
121
via spreading from the preceding syllable. First, consider the examples which are
presented in (61). This shows rightward spreading from the potential aspect marker /gú/
onto toneless /di/.
(61a) /gú-di-bi%"tsi%%/ [gúdíbi%"tsi%%] (61b) gú-dí-bi%"tsi%%=lu%$
POT-CAUS-dry POT-CAUS-dry=2s.INF
‘will dry (it)’ ‘You will dry (it)’
(61c) gú-dí-bi%"tsi%=bí (61d) gú-dí-bi%"tsi%=a)
POT-CAUS-dry=3s.FAM POT-CAUS-dry=3s.NFML
‘He will dry’ ‘He will dry’
Notice that none of the examples above contain a first-person singular, so we do not need
to consider the floating tone of the first-person singular clitic. In examples (61b) through
(61d), the H tone of the potential aspect marker /gú/ spreads rightward onto /di/ and
causes it to become H as well.
The same is true when we consider the completive aspect. In data set (62), the
mid-toned completive aspect marker precedes the causative /di/, and causes it too to
become mid-toned. This is illustrated in examples (62a) – (62d).
(62a) /be%-di-bi%"tsi%%/ [be%di%bi%"tsi%%] (62b) be%-di%-bi%"tsi%=lu%$
COM-CAUS-dry COM-CAUS-dry=2s.INF
‘dried’ ‘You dried’
122
(62c) be%-di%-bi%"tsi%=bí (62d) be%-di%-bi%"tsi%=a)
COM-CAUS-dry=3s.FAM COM-CAUS-dry=3s.NFML
‘He dried’ ‘He dried’
Here I posit that the completive aspect marker /be%/ has an underlying M tone. By
comparing the examples in (62) with the previous data set that illustrated the potential
aspect, it becomes clear that the tone of /di/ varies depending on its context. When the
aspect marker with a specified M tone precedes /di/, then /di/ has a M tone also.
42
When a toneless aspect marker precedes /di/ then both the aspect marker and /di/
will be realized with a mid-like pitch in the output. I posit that a phonologically toneless
syllable will always emerge with a M tone, that is, a tone that is neither H nor L. A
toneless aspect marker which precedes the causative /di/ is shown in data set (63). Here, I
have not transcribed a tone on either the aspect marker or /di/:
(63a) /ru-di-bi%"tsi%%/ [rudibi%"tsi]%% (63b) ru-di-bi"%tsi%=lu%$
HAB-CAUS-dry HAB-CAUS-dry=2s.INF
‘dries’ ‘You dry’
(63c) ru-di-bi%"tsi%=bí (63d) ru-di-bi%"tsi%=a)
HAB-CAUS-dry=3s.FAM HAB-CAUS-dry=3s.NFML
‘He dries’ ‘He dries’
The examples from the habitual aspect show that /di/ is realized with neither a H nor a L
tone. Presumably this is because both the habitual aspect marker /ru/ as well as the
42
While none of the aspect markers in SJZ are low-toned and I therefore cannot show L-
tone spreading onto /di/, I predict that this would happen if a L-toned aspect marker
preceded /di/.
123
causative marker /di/ lack underlyingly specified tones and consequently are realized as
M. These examples show that the verb ‘to dry’ is one case where /di/ emerges with the
same tone of the syllable to its immediate left.
The variable nature of /di/ is not limited to just one verb however. All verbs in the
database that include this prefix exhibit the same pattern. Another example is the verb ‘to
cover’, shown data set (64). Here I illustrate the potential and completive aspects
together:
(64a) /gú-di-ba##"gá$/ [gúdíba##"gá$] (64b) gú-dí-ba##"gá$=lu%$
POT-CAUS-cover POT-CAUS-cover=2s.INF
‘will cover’ ‘You will cover’
(64c) gú-dí-ba#"gá$=bí (64d) gú-dí-ba#"gá$=a)
POT-CAUS-cover=3S.FAM POT-CAUS-cover=3S.NFML
‘He will cover’ ‘He will cover’
(64e) /be%-di-ba##"gá$/ [be%di%ba##"gá$] (64f) be%-di%-ba##"gá$=lu%$
COM-CAUS-cover COM-CAUS-cover=2s.INF
‘covered’ ‘You covered’
(64g) be%-di%-ba#"gá$=bí (64h) be%-di%-ba#"gá$=a)
COM-CAUS-cover=3S.FAM COM-CAUS-cover=3S.NFML
‘He covered’ ‘He covered’
(64i) /ru-di-ba##"gá$/ [rudiba##"gá$] (64j) ru-di-ba##"gá$=lu%$
HAB-CAUS-cover HAB-CAUS-cover=2s.INF
‘covers’ ‘You cover’
(64k) ru-di-ba#"gá$=bí (64l) ru-di-ba#"gá$=a)
HAB-CAUS-cover=3S.FAM HAB-CAUS-cover=3S.NFML
‘He covers’ ‘He covers’
124
Examples (64a) – (64d) illustrate that in the potential aspect, /di/ has a H tone. I argue
that this is due to spreading from the high-toned potential aspect marker. In the
completive aspect illustrated with examples (64e) – (64h), /di/ has a M tone. Finally, the
habitual aspect illustrated in examples (64) – (64l) shows that /di/ is realized with no
phonological tone. Notice that the L tone of the first syllable of the verb root does not
spread leftward onto /di/ nor onto /ru/. If spreading were occurring just from a toned to a
toneless syllable without preference for directionality, then we would have expected the
L tone of the root to spread leftward to the toneless causative and aspect marker. Since
this does not occur, I argue that tonal spreading is only rightward in SJZ. Rightward
spreading is discussed in greater detail in section 2.5.
Finally, the verb ‘to fill’ – or ‘to make something full’ – shows the same pattern
whereby /di/ obtains its tone due to the syllable to its immediate left. Consider the
examples in (65). Here, all three aspects are shown together, with each individual marked
via a different letter.
(65a) /gú-di-tsa#$/ [gúdítsa#$] (65b) gú-dí-tsa#$=lu%$
POT-CAUS-fill POT-CAUS-full=2s.INF
‘will fill (it)’ ‘You will fill (it)’
(65c) gú-dí-tsa#$=bí (65d) gú-dí-tsa#$=a)
POT-CAUS-full=3S.FAM POT-CAUS-full=3S.NFML
‘He will fill (it)’ ‘He will fill it’
125
(65e) /be%-di-tsa#$/ [be%di%tsa#$] (65f) be%-di%-tsa#$=lu%$
COM-CAUS-fill COM-CAUS-full=2s.INF
‘filled (it)’ ‘You filled (it)’
(65g) be%-di%-tsa#$=bí (65h) be%-di%-tsa#$=a)
COM-CAUS-full=3S.FAM COM-CAUS-full=3S.NFML
‘He filled (it)’ ‘He filled it’
(65i) /ru-di-tsa#$/ [ruditsa#$] (65j) ru-di-tsa#$=lu%$
HAB-CAUS-fill HAB-CAUS-full=2s.INF
‘fills (it)’ ‘You fill (it)’
(65k) ru-di-tsa#$=bí (65l) ru-di-tsa#$=a)
HAB-CAUS-full=3S.FAM HAB-CAUS-full=3S.NFML
‘He fills (it)’ ‘He fills it’
Examples (65a) through (65d) show that in the potential, /di/ has a H tone. In the
completive, as shown in (65e) – (65h), /di/ has a M tone. The habitual examples show
that /di/ emerges as toneless and that no spreading occurs from the syllable to the
immediate right of /di/.
These examples show that in the habitual aspect /di/ is realized with a mid-like
pitch. It is important to point out that the specified L tone of the verb root does not spread
leftward onto the causative marker /di/ nor onto the toneless aspect marker /ru/. As I
discuss in the analysis chapter 3, this pattern is predicted by my analysis of tone gesture
deactivation.
These examples in the data set (65) shows the same pattern that was demonstrated
for the other two verbs in the database: the causative marker /di/ will exhibit whatever
tone is to its immediate left. I suggest that this is because the causative marker /di/ does
126
not have an underlying, lexically specified tone. Thus I would expect that in cases where
a low-toned aspect marker precedes /di/ that /di/ would then be low-toned as well. Since
there are no L-toned aspect markers in SJZ, there are no cases I have found where this
occurs. I posit that spreading is only rightward in this language and that toneless syllables
will emerge as M if not acted upon by another tone. Thus if a toneless syllable is not
preceded by any tone, then it will have a pitch that is neither H nor L. Otherwise, a
toneless syllable will suffer the effects of spreading from a preceding tone. As we have
seen based on the examples from this section, a preceding H tone causes a toneless
syllable to become H and a preceding M tone causes a toneless syllable to be realized as
M.
2.4.6 Verbs with Incorporated Nouns
There are only three verbs in the database with incorporated nouns, but all three
show a similar pattern: the floating H of the first-person singular appears on the verb root
in a way similar to what has been shown for other verbs, and a H tone appears on the
incorporated noun itself. An example of a verb with an incorporated noun is ‘to listen’,
which is literally ‘to put ear’. In this case, a H tone appears on the verb ‘to put’ as well as
on the noun ‘ear’.
First I will present the verb ‘to listen’. The potential forms are shown in the
examples in data set (66):
127
(66a) /gú-da%"-na%"ga%$/ [gúda%"na%"ga%$] (66b) gú-da!"-na%"gá$
POT-put.ear POT-put.ear=1s
‘will listen’ ‘I will listen’
(66c) gú-da%"-na%"ga%$=lu%$ (66d) gú-da%"-na%"ga%$=bí
POT-put.ear=2s.INF POT-put.ear-3s.FAM
‘You will listen’ ‘He will listen’
Example (66b) has a falling contour tone on the verb root /da%"/. I propose that this occurs
because the floating H combines with an underlying M tone on the verb root. Second,
notice the H tone on the last syllable of 'ear'. The first-person singular clitic which usually
appears as either /$/ or /a$/ has been dropped, presumably because the verb ends with this
same sequence.
43
The noun /na%"ga%$/ appears instead as [na%"gá$] in the first-person
singular but not in the second or third-person singular. The underlying lexical M tone is
deleted and replaced with a H tone in the noun; this is because a contour cannot be
formed in a syllable with a short vowel.
Example (66b), when compared with (66c) and (66d) might seem to indicate that
these types of verbs are double-marked in the sense that both the verb root and the noun
are marked with a floating H tone. Since a specified M tone intervenes, it cannot be the
case that one H tone has attached to the verb root but stayed active to cause the pattern in
(66b). I speculate, however, that the H tone on the noun is due to a separate process that
marks inalienably possessed nouns. I propose that the H tone on the noun may be part of
43
Future work should examine segmental alternations of this type.
128
a dependent possessive pronoun.
44
There is evidence that the incorporated nouns are
marked with the possessive affixes and not the regular subject pronouns. This is apparent
in the third-person singular and plural non-formal. The verb ‘to listen’ for this pronoun is
given in examples (67a) and (67b) below:
(67a) gú-da%"-na%%"g=i# (67b) gú-da%"-na%%"g=gi) #
POT-put.ear=3s.NFML POT-put.ear=3s.NFML
‘He will listen’ ‘They will listen’
Recall from previous examples that I have presented in this chapter that the expected
subject clitic for the third-person singular non-formal is /a)/. As presented in the
introduction and seen throughout the appendix, the usual form of the third-person plural
non-formal is /ga) # /. Examples (67a) and (67b) are instead marked with possessive
pronouns (Bartholomew, 1983: 354). In SJZ, the subject markers and the possessive
markers are the same for 11 of the 13 pronouns – only the third-person singular and
plural have a different marker.
This example suggests that incorporated inalienable nouns are obligatorily
possessed in SJZ.
45
Evidence from the Bartholomew grammar (1983) suggests that nouns
which are marked for possession also have a floating H tone in the first-person singular.
This is likely to be a separate process from the floating H that appears on verbs. For
44
This is translated from pronombre posesivo dependiente (Bartholomew 1983)
45
Beam de Azcona (2004) reports a similar marking of inalienably possessed nouns in
Coatlán Loxicha Zapotec.
129
example, Bartholomew gives examples such as /k
w
e#$e#/ ‘back’ versus /k
w
í$a#$/ ‘my back’.
The first-person singular form appears to show an additional H tone which is absent in
the unmarked form (1983: 354-356). Unfortunately I did not collect data on these types
of nouns which are obligatorily possessed, but such data would disambiguate the
situation. If nouns are marked with a H tone in the first-person singular – as in, ‘my ear’
but not marked specially for ‘your ear’ or ‘his ear’, then we could conclude that there are
two similar floating H-tone marking processes in the language: one for verbs and another
for nouns.
Next, consider the completive aspect of this same verb. This is shown in data set
(68):
(68a) /be%-da%"-na%"ga%$/ [be%da%"na%"ga%$] (68b) be%-da!"-na%"gá$
COM-put.ear COM-put.ear=1s
‘listened’ ‘I listened’
(68c) be%-da%"-na%%"ga%$=lu%$ (68d) be%-da%"-na%"ga%$=bí
COM-put.ear=2s.INF COM-put.ear-3s.FAM
‘You listened’ ‘He listened’
We see the same pattern for the completive aspect as demonstrated for the potential: the
first-person singular has a H tone that appears on the verb root /da%%"/. The floating H
combines with the underlying M tone to form a falling tone. Then, there is an additional
H tone that appears on the incorporated noun /na%"ga%$/ to give /na%"gá$/ only for the first-
person singular form.
130
Finally, consider the habitual aspect of this verb. The examples in data set (69)
show that the floating H of the first-person singular appears on the toneless aspect marker
/ru/ in addition to the verb root /da%"/:
(69a) /ru-da%"-na%"ga%$/ [ruda%"na%"ga%$] (69b) rú-da!"-na%"gá$
HAB-put.ear HAB-put-ear=1s
‘listens’ ‘I listen’
(69c) ru-da%"-na%"ga%$=lu%$ (69d) ru-da%"-na%%"ga%$=bí
HAB-put-ear=2s.INF HAB-put-ear-3s
‘You listen’ ‘He listens’
Thus the verb in example (69b) follows all of the patterns that would be expected based
on the data that have been presented thus far. The difference is that example (69b) also
shows a H tone on the incorporated noun /na%"gá$/.
Since I have already shown the possessive marker on this verb for the third-
person non-formal, I will not show examples of that pronoun in the following examples
so that these data are consistent with the data presented in the rest of this chapter. I posit
that in this example and future examples, the first-person singular has a H tone on the
noun because of a separate process of H tone marking used for inalienably possessed
nouns.
Next I will give examples from the verb ‘to cross oneself’. This verb is a
combination of the verb ‘write’ plus the noun ‘face’. It too has a H tone that appears on
the verb and a H tone that appears on the incorporated noun.
131
(70a) /gú-di#a#-lo!"/ [gúdi#a#lo!"] (70b) gú-día#-lo%=á$
46
POT-write.face POT-write-face=1s
‘will cross self’ ‘I will cross myself’
(70c) gú-di#a#-lo!"=lu%$ (70d) gú-di#a#-lo!"=bí
POT-write-face=2s.INF POT-write-face=3s.FAM
‘You will cross yourself’ ‘He will cross himself’
Here I posit that the verb root ‘to write’ is /di#a#/. In the first-person singular, the floating H
tone appears on the vowel /i/, as would be expected. Notice that the noun ‘face’ has a H
tone in example (70b) that is not present in the other two examples. This noun is /lo!"/
with a falling tone in (70c) and (70d) but changes to /loá/ in the first-person singular. It
appears either that the first-person singular marker for nouns is /á$/, or that the first-
person singular marker for nouns has caused a segmental change in the noun /lo!"/
Nonetheless this example appears to have a H tone that is specific to the first-person
singular. The other two forms do not have this H tone.
The completive forms of this same verb are shown in the examples in data set
(71):
(71a) /be%-di#a#-lo!"/ [be%di#a#lo!"] (71b) be%-día#-lo=á$
COM-write.face COM-write.face=1s
‘crossed self’ ‘I crossed myself’
46
It is unclear how to segment this form. The vowel of /lo!"/ ‘face’ is now short, with a M
tone, and a high-toned vowel /á/ appears.
132
(71c) be%-di#a#-lo!"=lu%$ (71d) be%-di#a#-lo!"=bí
COM-write.face=2s.INF COM-write.face=3S.FAM
‘You crossed yourself’ ‘He crossed himself’
Example (71b) shows the floating H tone of the first-person singular on the first syllable
of the verb root. Here, the vowel /i/ has a H tone in example (71b) but not in examples
(71c) or (71d). Example (71b) also illustrates that the vowel of the word for ‘face’ has
acquired a H tone that is not present in the other pronouns. A comparison of the first and
second-person singular forms is found in Figure 2.28 and Figure 2.29. For the second-
person singular, a spectrogram is also included in Figure 2.30 with an arrow to indicate a
slight period of glottal irregularity.
Figure 2.28 [be%-día#-lo%á=$]
133
Figure 2.29 [be%-di#a#-lo!"=lu%$]
Figure 2.30 Spectrogram of [be%-di#a#-lo!"=lu%$]
Finally, consider the habitual aspect shown in examples (72a) through (72d):
134
(72a) /ru-di#a#-lo!"/ [rudi#a#lo!"] (72b) rú-día#-loá=$
HAB-write.face HAB-write-face=1s
‘crosses self’ ‘I cross myself’
(72c) ru-di#a#-lo!"=lu%$ (72d) ru-di#a#-lo!"=bí
HAB-write-face=2s.INF HAB-write-face=3S.FAM
‘You cross yourself’ ‘He crosses himself’
Example (72b) shows that the floating H of the first-person singular appears on the
toneless aspect marker /ru/ as well as on the first syllable of the verb root. There is also
an additional H tone which appears on the incorporated noun /lo!"/. This noun then
becomes /loá$/ after being marked for the first-person singular.
While the first two verbs with incorporated nouns presented here belong to verb
class 1, there is also an example of a verb with an incorporated noun from verb class 7.
This verb class has the aspect markers /gá/ for the potential, /rá/ for the habitual, and /gu/
for the completive. I assume that /gu/ is underlyingly toneless because the floating tone
of the first-person singular appears on this aspect marker. Data set (73) shows the
potential forms of the verb ‘to get dressed’:
(73a) /gá-kku#$-'o!"/ [gákku#$'o!"] (73b) gá-kkú$-'ó=á$
POT-put.clothes POT-put-clothes=1s
‘will get dressed’ ‘I will get dressed’
(73c) gá-kku#$-'o!"=lu%$ (73d) gá-kku#$-'o!"=bí
POT-put-clothes=2s.INF POT-put-clothes=3s.FAM
‘You will get dressed’ ‘He will get dressed’
135
Example (73b) shows that the verb root /kku#$/, ‘to put’, has become /kkú$/ in the first-
person singular. This shows that the underlying L tone has been replaced by the floating
H of the first-person singular. The underlying L is replaced because the vowel is short
and no contour may form. Meanwhile, the noun for ‘clothing’ has been marked for the
first-person singular. This gives /'óá$/ ‘my clothing’. Regardless of whether we consider
the vowel /a/ to be part of the noun or part of the 1 sg marker, this example shows that
there are two H tones in example (73b) that are absent in examples (73c) and (73d).
The completive forms for this verb show the same pattern: a H tone appears on
the verb root and on any available aspect marker for the first-person singular. The
incorporated noun is also marked with an additional H tone in the first-person singular.
Observe the members of data set (74):
(74a) /gu-kku#$-'o!"/ [gukku#$'o!"] (74b) gú-kkú$-'óá=$
COM-put.clothes COM-put.clothes=1s
‘got dressed’ ‘I got dressed’
(74c) gu-kku#$-'o!"=lu%$ (74d) gu-kku#$-'o!"=bí
COM-put.clothes=2s.INF COM-put.clothes=3s.FAM
‘You got dressed’ ‘He got dressed’
The first-person singular form shown in (74b) has a H tone on the toneless aspect marker
/gu/, as well as on the verb root. We can also observe from this example that the L tone of
the verb root does not spread leftward onto the toneless aspect marker. Of course, the
noun ‘clothes’ is also marked with a H tone in the first-person singular.
136
Finally, consider the habitual aspect shown in data set (75). This demonstrates
that for this particular verb class (7), the habitual aspect marker has an underlying H tone.
(75a) /rá-kku#$-'o!"/ [rá-kku#$-'o!"] (75b) rá-kkú$-'óá=$
HAB-put.clothes HAB-put-clothes=1s
‘gets dressed’ ‘I get dressed’
(75c) rá-kku#$-'o!"=lu%$ (75d) rá-kku#$-'o!"=bí
HAB-put-clothes=2s.INF HAB-put-clothes=3s.FAM
‘You get dressed’ ‘He gets dressed’
Example (75b) shows a H tone on the stressed syllable of the verb root. The floating tone
cannot appear on the aspect marker since this marker already has an underlying lexical
tone. The noun is also marked, as we have seen for other verbs with incorporated nouns.
All three of the incorporated noun verbs shown here suggest that in SJZ, both
verbs and nouns are specially marked for the first-person singular. The forms shown here
conform to the other processes that have been described for the interaction between the
floating tone and verb roots, with the additional trait that they also show a H tone on the
incorporated noun. Further fieldwork is needed to confirm if this is caused by a form of
H-tone marking on possessed nouns, but I tentatively expect that this is the case.
2.4.7 Summary
This concludes the section on the placement of the floating H of the first-person
singular. A summary of the tonal processes we have seen is given in Table 2.1:
137
Table 2.1 Placement of the Floating H Tone
Circumstance Floating tone
Aspect marker has a tone and stressed
syllable of verb root has no tone
Appears on stressed syllable of
verb root
Aspect marker has a tone and stressed
syllable contains a short vowel with any
tone
Replaces underlying tone of
stressed syllable
Aspect marker has a tone and stressed
syllable is a long vowel with a M or L
tone
Appears on stressed syllable of
verb root to form a falling
contour
Aspect marker has a tone and stressed
syllable of verb root is long or short vowel
with a H tone
No change is evident in the
output: stressed syllable of
verb root still has H tone
Aspect marker is toneless and stressed
syllable has any of the above conditions
Appears on aspect marker
AND on stressed syllable of
verb root according to the
conditions above
Aspect marker is toneless, followed by
toneless causative marker followed by
verb root with any of the above conditions
Appears on aspect marker,
causative marker, and stressed
syllable of verb root according
to the conditions above.
2.5 Spreading Tones in SJZ
This section will examine the spreading behavior of tones in SJZ. These processes
of tone spreading are independent of the behavior of the floating H tone of the first-
person singular. In many cases, a particular verb may show the floating H of the first
person singular as well as rightward spreading of verb-final H tones onto the clitics. This
section on tone spreading is separated into individual sections that highlight the following
processes: rightward spreading of verb-final H tones onto clitics, failure of verb-final L
tones to spread onto clitics, downstepping only of the clitic /é/ when following a verb-
final H tone, and rightward spreading from the aspect markers onto toneless syllables in
138
the verb root. Importantly, these rightward spreading processes are distinct from root
control, because it is not simply tones that spread from the root to attached clitics. Rather,
prefix tones spread rightward onto the root and root tones spread rightward onto the
suffix. Each case is illustrated with at least two examples from the database of SJZ verbs.
The reader is referred to the appendix for additional examples of these processes.
2.5.1 Rightward Spreading of Verb-final H onto Clitics
Throughout the database in every instance in which a verb root ends with a high-
toned vowel, this H tone spreads rightward onto the attached clitics. When a verb ends
with a H tone, the first-person singular /a$/, the second-person singular informal /lu%$/,
the third-person singular non-formal, /a)/, the first-person plural exclusive /tu%$/, and the
third-person plural non-formal /ga) # / all become high. To illustrate, I have placed the
underlying forms on the left side, and the outcome of their combination on the right. So,
for example, in data set (76) below, (a) has the underlying forms and the arrow indicates
the output forms. The glosses and translation are only given for the output form.
(76a) /k
w
édá =a$/ k
w
édá=$
47
POT/wait=1s
‘I will wait’
47
It is unclear how to segment this form, since we do not know whether the verb-final /a/
has dropped, or whether it is the clitic’s /a/ which has been dropped. Or, the two could
have merged.
139
(76b) /k
w
édá = lu%$/ k
w
édá=lú$
POT/wait=2s.INF
‘You will wait’
(76c) /k
w
édá =a)/ k
w
éda) .
POT/wait=3s.NFML
‘He will wait’
(76d) /k
w
édá = tu%$/ k
w
édá=tú$
POT/wait=1p.EXCL
‘We will wait’
(76e) /k
w
édá = ga) # /
k
w
édá=ga) .
POT/wait=3p.NFML
‘They will wait’
Thus, (76a) – (76e) all show the underlying forms. The output forms shown next to each
via the arrow show that all of these clitics have become H under the influence of the
verb-final H tone. I posit that this tone has spread rightward and has replaced the
underlying lexical tone of each of these clitics. In the first-person singular and third-
person non-formal, perhaps due to an OCP-type effect, the two /a/ vowels have either
merged or undergone deletion so that only one /a/ is realized. In both cases however the
tone is H. In the additional examples with this verb I leave these two forms out, since the
spreading effect is easier to see in persons other than these two.
To illustrate that it is not the H tone of the potential aspect marker causing the
clitics to become H, I will show the completive aspect of the same verb. This makes it
clear that the change in tone in the clitics is due to the verb-final tone. Observe the
examples in data set (77):
140
(77a) /u%-le%dá = tu%$/ u%-le%dá=tú$
COM-wait=1p.EXCL
‘We waited’
(77b) /u%-le%dá = lu%$/ u%-le%dá=lú$
COM-wait=2s.INF
‘You waited’
(77c) /u%-le%dá = ga) #/
u%-le%dá =ga) .
COM-wait=3p.NFML
‘They waited’
Examples (77a) through (77c) show that because this particular verb ends in a H tone, it
will spread this H tone rightward onto the attached clitics regardless of the aspect. The
fact that the completive aspect behaves the same way as the potential shows us that it is
not just the H tone of the potential spreading all the way through the verb root and onto
the clitic. This is confirmed by considering the habitual aspect shown in examples (78a)
through (78c):
(78a) /ri-be%dá = tu%$/ ri-be%dá=tú$
HAB-wait=1p.EXCL
‘We wait’
(78b) /ri-be%dá = lu%$/ ri-be%dá=lú$
HAB-wait=2s.INF
‘You wait’
(78c) /ri-be%dá = ga) #/
ri-be%dá=ga) .
HAB-wait=3p.NFML
‘They wait’
141
Again, examples (78a) through (78c) illustrate that the final vowel of the verb has a
specified H tone and that this H tone spreads rightward onto the clitic in three aspects.
Another verb which illustrates the same pattern of rightward H spread onto clitics
is ‘to carry’. Since this particular verb root does not end with the vowel /a/, tone
spreading is easier to see onto the first-person singular and third-person non-formal
clitics, which are included for this verb. I will first present the completive aspect in (79):
(79a) /bi-a$ní =a$/ bí-á$ní=á$
COM-carry=1s
‘I carried’
(79b) /bi-a$ní = lu%$/ bi-a$ní =lú$
COM-carry=2s.INF
‘You carried’
(79c) /bi-a$ní = a)/ bi-a$ní=a) .
COM-carry=1s.NFML
‘He carried’
(79d) /bi-a$ní = tu%$/ bi-a$ní =tú$
COM-carry=1p.EXCL
‘We carried’
(79e) /bi-a$ní = ga) # /
bi-a$ní =ga) .
COM-carry=3p.NFML
‘They carried’
The completive aspect shown in examples (79a) through (79e) shows that it is only the
second syllable of the verb that is underlyingly H-toned and that this H tone spreads
rightward onto the attached clitics. Importantly, there is no leftward spread from the verb-
142
final H tone onto the toneless initial syllable or aspect marker. The same pattern of
rightward spread from the final vowel of the verb root onto the aspect marker holds true
for the habitual and potential aspects. Due to the large number of examples, I will show
the aspects separately in (80) and (81).
(80a) /ru-a$ní =a$/ rú-á$ní=á$
HAB-carry=1s
‘I carry’
(80b) /ru-a$ní = lu%$/ ru-a$ní=lú$
HAB-carry=2s.INF
‘You carry’
(80c) /ru-a$ní = a)/ ru-a$ní=a) .
HAB-carry=3s.NFML
‘He carries’
(80d) /ru-a$ní = tu%$/ ru-a$ní =tú$
HAB-carry=1p.EXCL
‘We carry’
(80e) /ru-a$ní = ga) #/
ru-a$ní =ga) .
HAB-carry=3p.NFML
‘They carry’
Data set (80) is straightforward with the exception of example (80a). In the case of the
first-person singular, the aspect marker /ru/ and the initial syllable of the verb root
become H due to the floating H tone. The final H tone then spreads onto the clitic. In
examples (80b) – (80e) we can see that the verb-final H tone does not spread leftward and
143
instead spreads rightward onto the attached clitics. Examples in (81) illustrate the
potential aspect:
(81a) /w-á$ní = a$/ wá$ní=á$
POT-carry=1s
‘I will carry’
(81b) /w-á$ní = lu%$/ wá$ní =lú$
POT-carry=2s.INF
‘You will carry’
(81c) /w-á$ní = a)/ wá$ní=a) .
POT-carry=1s.NFML
‘He will carry’
(81d) /w-á$ní = tu%$/ wá$ní =tú$
POT-carry=1p.EXCL
‘We will carry’
(81e) w-á$ní = ga) # /
wá$ní =ga) .
POT-carry=3p.NFML
‘They will carry’
The examples in (81) show that the potential aspect spreads its H tone onto the initial
syllable of the verb root. Examples (81a) through (81e) also show that the verb-final H
tone spreads onto the attached clitic.
The examples from this section have shown that the second syllable of the verb
root is high and that this H tone spreads rightward onto the attached clitic. Although there
are no cases of a monosyllabic H-toned verb root in the database I recorded, these types
of verb roots do exist, and I would expect such roots to show a similar spreading process.
144
There are several examples of disyllabic verbs ending in a H-toned vowel in the
appendix, and the reader will notice throughout this chapter that in cases where the verb
ends in a H tone, the clitics also have H tones. This is also true for clitics whose
underlying tone is H. These are the focus of the next section I discuss.
2.5.2 Clitics Whose Underlying Tone is Hː
Clitics such as the second-person plural /lé/, the third-person singular familiar /bí/,
and the third-person plural familiar /gé/ do not change their tones when they attach to a
verb ending in a H tone. Earlier in this chapter when I discussed Figure 2.19, I mentioned
that across an entire word (prefix, root, and suffix), each consecutive H tone was slightly
lower than the preceding one, a pattern I attributed to downdrift. This downdrift is not
apparent when a H-toned clitic attaches to a verb root with a final H tone. I do not rule
out that it could occur, but in the database I collected, there is no evidence of a change to
the two consecutive H tones for clitics with a CV structure. (The third-person singular
formal clitic which consists of just a vowel /é/ is discussed in a separate section.) To
illustrate that verbs ending in a H tone do not cause tone deletion or other noticeable
perturbations of the H-toned clitic, consider the examples in (82).
(82a) /wá$ní + bí/ wá$ní=bí
POT/carry=3s.FAM
‘He will carry’
145
(82b) /wá$ní + lé/ wá$ní=lé
POT/carry=2P.INF
You all will carry’
(82c) /wá$ní + gé/ wá$ní=gé
POT/carry=3P.FML
‘They will carry’
The examples in (82) illustrate that clitics which are underlyingly H do not change this
tone when the verb ends in a H tone. For example, the third-person singular familiar /bí/
is always H, regardless of the tones of the verb. A pitch track is given in Figure 2.31:
Figure 2.31 [wá$ní=bí]
The data sets I present in (83) and (84) contain a verb with a final M tone and a
verb with a final H tone. The clitics shown here are lexically specified as H, and they are
not altered when they attach to a verb with a final H tone.
146
(83a) /í-bí&i% = bí/ í-bí&i%=bí
POT-fall.down=3s.FAM
‘He will fall down’
(83b) /í-bí&i%=lé/ í-bí&i%=lé
POT-fall.down=2P.INF
You all will fall down’
(83c) /í-bí&i%=gé/ í-bí&i%=gé
POT-fall.down=3P.FORM
‘They will fall down’
(84a) /tá-ne#llé + bí/ tá-ne#llé=bí
POT-visit=3s.FAM
‘He will visit’
(84b) /tá-ne#llé + lé/ tá-ne#llé=lé
POT-VISIT=2P.INF
You all will visit’
(84c) /tá-ne#llé=gé/ tá-ne#llé=gé
POT-VISIT=3P.FORM
‘They will visit’
Here we see that the verb ending in a H tone does not affect the H tone of the clitic. The
clitics illustrated here are H regardless of whether they follow a verb that ends in a M or
H tone. This is illustrated in the figures below. First, observe the clitic /=lé/ when it
cliticizes to a verb that ends with a M tone. The pitch track in Figure 2.32 is from
example (83b) of the data set:
147
Figure 2.32 [í-bí&i%=lé]
Figure 2.32 illustrates that /lé/ has H tone when it attaches to a verb ending in a M tone.
Now, compare Figure 2.32 with a visual example of this same clitic following a verb that
ends with a H tone. The pitch track in Figure 2.33 is from example (84b) in the data set:
Figure 2.33 [tá-ne#llé=lé]
148
From these examples I posit that the clitics illustrated here are underlyingly high-toned
and that they remain H when following a verb that ends in a H tone.
There is one exception to this pattern, and that is the clitic /é/, which corresponds
to the second and third-person singular respectful. I discuss a process of lowering that
affects this clitic in the next section.
2.5.3 The Clitic /é/
The clitic /é/ does not interact with verbs that have a final H tone the same way as
the other H-toned clitics. Whereas the other high-toned clitics remain H when they attach
to a verb that ends in a H tone, /é/ is different because its pitch lowers to the same level as
a M tone would normally have. This creates a falling tonal contour on the last two vowels
of the word. Here I will point out two cases where this occurs. I also include some
examples of pitch tracking to show the process visually.
First I will give examples where /é/ appears in its usual form. This occurs when
/é/ attaches to verbs that end in a low-toned vowel, a mid-toned vowel, or a glottal stop.
Recall that this clitic corresponds to both the second-person singular formal, and to the
third-person singular respectful.
(85a) /ka!"bi# = é/ ka!"bi#=é
POT/paint=2s.FORM
‘You will paint’
(85b) /u%-ta##"bi# =é/ u%-ta##"bi#=é
COM/paint=2S.FORM
‘You will paint’
149
(85c) /r-a#"bi# = é/ r-a#"bi#=é
HAB/paint=2S.FORM
‘You will paint’
Examples (85a) through (85c) show the potential, completive, and habitual aspects for the
verb ‘to paint’. Here the clitic /é/ appears with its normal H tone in all three cases. A
pitch track example of (85b) is shown in Figure 2.34:
Figure 2.34 [u%-ta##"bi#=é]
Figure 2.34 shows that /é/ retains its H tone when it follows a verb ending in a L tone.
Contrast these examples with those from the verb ‘to wait’, illustrated in the
examples of (86). Here I have included a third-person singular familiar just for
comparison. These examples illustrate that the clitic /é/ loses its H tone in this context.
150
(86a) /k
w
édé = é/ k
w
édé=e
POT/wait=2s.FORM
‘You will wait’
(86b) /k
w
édá= bí/ k
w
édá=bí
POT/wait=3S.FAM
‘He will wait’
(86c) /k
w
édé = é/ k
w
édé=e
POT/wait=3s.FORM
‘He will wait’
Example (86a) shows that when the second-person singular clitic attaches to the verb
root, the clitic is lowered to a mid-toned vowel. Example (86b) gives the third-person
singular familiar form and illustrates that the clitic /bí/, despite having a H tone
underlyingly, does not lower this H tone after attaching to the verb root. The example in
(86c) shows the third-person singular formal form of the verb, again with lowering of the
clitic /é/. It is important to point out that the speaker DP uses the same clitic to refer to
both the second-person singular formal and the third-person singular formal. Each
pronoun was recorded separately during the elicitation session and both demonstrated the
same lowering pattern. Since no tone markings were used in the written format that the
speakers and I developed, I assume this tone-lowering process is automatic for the
speaker and thus part of the grammar.
A second example verb showing downstepping of the clitic /é/ is shown in
examples (87a) – (87c):
151
(87a) /u%-ta%"ní = é/ u%-ta%"ní =e
COM-dig=2s.FORM
‘You dug’
(87b) /u%-ta%"ní = bí/ u%-ta%"ní =bí
COM-dig =3S.FAM
‘He dug’
(87c) /u%-ta%"ní =é/ u%-ta%"ní =e
COM-dig=3S.FORM
‘He dug’
Examples (87a) and (87c) illustrate again that the clitic /é/ is lowered when it attaches to
a verb which ends with a H tone. A pitch tracking example is shown in Figure 2.35
Figure 2.35 [u%-ta%"ní=e]
In Figure 2.35 the aspect marker and initial vowel of the root are both mid-toned. The
final vowel of the root has a H tone, but then we can observe that /é/ is lowered, although
152
it does not become as low as the preceding M tones. There are many more examples like
this in the appendix, but the examples shown here illustrate a general pattern: when high-
toned clitics consist of a consonant and a vowel they will not be lowered when attaching
to a verb that ends in a H tone. When high-toned clitics consist of just a vowel, the tone
on this vowel is then lowered when it attaches to a verb ending in a H tone. It is unclear
what would drive an OCP type process that applies only to consecutive H-toned vowels.
It is also unclear whether the process involves deletion or simply lowering without
deletion. This topic should be investigated in future work with additional data from other
speakers. Perhaps there is a restriction on consecutive H-tones without an intervening
consonant in the language.
2.5.4 L Tones do not Spread onto Clitics
Thus far we have seen that verb-final H tones spread rightward onto attached
clitics. One might wonder if all verb-final tones spread rightward onto the attached clitics
in some type of root control scenario. Here I present examples illustrating that verb-final
L tones do not spread rightward. Therefore it must be the case that only H tones spread
rightward onto the clitics. The exception of course is when clitics lack an underlying
tone, as I discussed in an earlier section of this chapter.
First, consider the examples in data set (88):
(88a) /gú - di#a# = lu%$/ gú-di#a#=lu%$
POT-write=2s.INF
‘You will write’
153
(88b) /gú - di#a# = bí/ gú-di#a#=bí
POT-write=3s.FAM
‘He will write’
(88c) /gú - di#a# = tu%$/ gú-di#a#=tu%$
POT-write=1p.EXCL
‘We will write’
Example (88a) demonstrates that the second-person singular infomal clitic /lu%$/ will
remain mid-toned even following a verb that ends in a L tone. Likewise, the third-person
singular familiar clitic /bí/ shown in example (88b) remains H and does not alter its tone
when it attaches to a verb root ending in a L tone. The same is true where the first-person
plural exclusive clitic maintains its M tone. A visual example of (88c) is found in Figure
2.36:
Figure 2.36 [gú-di#a#=tu%$]
154
Figure 2.36 illustrates that the M tone of the clitic /tu%$/ is maintained in this example.
Although they are not included here, both the completive and habitual forms of this verb
demonstrate that the verb-final L tone fails to spread rightward onto the attached clitics.
If there were a general process in the language whereby the verb-final L tones spread
rightward onto these clitics, then we would expect all of the examples shown here to have
L tones on the clitics as well. The fact that this does not happen suggests that there is no
process of rightward L spread in the language.
An additional verb which illustrates that verb-final L tones do not spread
rightward onto the attached clitics is given in data set (89):
(89a) /ka!"bi# = lu%$/ ka!"bi#=lu%$
POT/paint=2s.INF
‘You will paint’
(89b) /ka!"bi# = bí/ ka!"bi#=bí
POT/paint=3s.FAM
‘He will paint’
(89c) /ka!"bi# = tu%$/ ka!"bi#=tu%$
POT/paint=1p.EXCL
‘We will paint’
This data set shows the potential aspect of a verb with a final L tone. Example (89a)
shows that the second-person informal will maintain its underlying M tone and that it will
not become L under the influence of the verb-final tone. Example (89b) illustrates that
the underlying H tone of /bí/ will be maintained and that it is not lowered to a M tone or
155
changed to a L tone. Finally, example (89c) shows that the underlying M tone of /tu%$/ is
maintained and that the verb-final L tone does not spread. Additional examples showing
that clitics are not altered by a verb-final L tone may be found in the appendix.
The same is true for verb roots that end in M tone. The examples in (90) illustrate:
(90a) / t(á-++a% = lu%$/ t(á-++a%=lu%$
POT/get.up=2s.INF
‘You will get up’
(90b) / t(á-++a%=bí/ t(á-++a%=bí
POT/get.up=3s.FAM
‘He will get.up’
(90c) / t(á-++a%=tu%$/ t(á-++a%=tu%$
POT/get.up=1p.EXCL
‘We will get up’
The example in (90) ends with a M tone and this is shown not to affect the tone of the
following clitics. There are additional examples in the appendix such as /&a%tta%/ ‘iron’,
/det(u%/ ‘fold or twist’, /júnna%/ ‘return something’, and many others. Since M and L tones
do not spread rightward onto the attached clitics if these have an underlying tone, I posit
that there is only a process of rightward spreading of H. This point will be discussed at
length in the analysis chapter of the dissertation.
156
2.5.5 Potential H Tones Spread onto Toneless Syllables
Thus far it has been shown that verb-final H tones spread rightward onto attached
clitics. Given the theoretical approach that is discussed in chapter 3, we might also expect
the H tone of the potential to spread rightward onto toneless syllables in the root. This
does indeed occur, as I discuss in this section. I would also expect spreading in cases
where a prefix is low-toned and it attaches to a verb root whose first syllable is
underlyingly toneless. Unfortunately there are not enough examples in the database to
verify whether this is in fact the case.
48
It does seem clear that for many verb classes with
a H tone on the potential aspect marker, this aspect marker will spread its H tone
rightward onto the a toneless syllable of the root.
To observe this fact, consider examples from the verb ‘to fold’. The completive
aspect is shown in (91a) – (91d). Here, it is only the first-person singular shown in
example (91b) that has a H tone on the first syllable of the verb root. I assume that this H
tone is the floating H of the first-person singular.
(91a) /be% - det(u%/ [be%de%t(u] (91b) be%-dét(
w
=a#$
‘folded’ COM-fold=1s
‘I folded’
(91c) be%-de%t(u%=lu%$ (91d) be%-de%t(u%=bí
COM-fold=2s.INF COM-fold-3s.FAM
‘You folded’ ‘He folded’
48
This does occur in Macuiltianguis Zapotec, a related dialect that I discuss in chapter 3.
157
Note that examples (91c) and (91d) lack a H tone on the syllable /de/. Instead, these are
mid-toned due to spreading from the mid-toned aspect marker. This syllable could not be
underlyingly mid-toned, or a falling tone would be expected in the 1sg form.
The same pattern emerges for the habitual aspect of this verb, which is presented
in examples (92a) through (92d):
(92a) /ru - det(u%/ [rudet(u%] (92b) rú-dét(
w
=a#$
‘folds’ HAB-fold=1s
‘I fold’
(92c) ru-det(u%=lu%$ (92d) ru-det(u%=bí
HAB-fold=2s.INF HAB-fold-3s.FAM
‘You fold’ ‘He folds’
Again, we can see that it is only the first-person singular shown in example (92b) that has
a H tone on the stressed syllable. Evidence supporting that the initial syllable of the verb
root is toneless is found in the potential aspect, shown below in (93a) – (93d):
(93a) /gú - det(u%/ [gúdét(u%] (93b) gú-dét(
w
=a#$
‘will fold’ POT-fold=1s
‘I will fold’
(93c) gú-dét(u%=lu%$ (93d) gú-dét(u%=bí
POT-fold=2s.INF POT-fold-3s.FAM
‘You will fold’ ‘He will fold’
In example (93b), I argue /de/ has a H tone due to the floating tone of the first-person
singular. However, examples (93c) and (87d) show a H tone on this syllable due to
158
spreading from the high-toned potential aspect marker /gú/. The reason I propose for why
this syllable obtains its H tone due to spreading is that when a mid-toned or a toneless
aspect marker precedes /de/, it does not have a H tone as it does in examples (93c) and
(93d).
Additional examples from the database show the same pattern in other verbs. In
these cases the H tone of the potential aspect marker spreads rightward onto toneless
syllables in the root. To date I have not encountered an example where both syllables in
the root are toneless, but I do not rule out that possibility. In such cases I would expect
the H tone of the potential aspect marker to spread all the way through the verb root until
it reached a clitic with a specified tone. Therefore, whether or not the potential aspect
marker spreads its tone rightward onto the verb root can be used as a diagnostic to
determine whether or not the verb root has specified tones.
Data set (94) shows an additional case of rightward spreading from the high-toned
potential aspect marker onto the verb root. This verb is intransitive, so there is no H tone
of the first-person singular. Thus I assume that the H tone on the first syllable of the root
is due to spreading from the potential marker in all three cases.
(94a) /í - bi&i% / [íbí&i%] (94b) í-bí&i%=a%$
POT-fall.down POT-fall.down.=1s
‘will fall down’ ‘I will fall down’
(94c) í-bí&i%=lu%$ (94d) í-bí&i%=bí
POT-fall.down=2s. INF POT-fall.down=3s.FAM
‘You will fall down’ ‘He will fall down’
159
Here, in all three examples, the potential H tone of /í/ spreads rightward onto the syllable
/bi/ of the verb. I assume that this verb has a toneless initial syllable because it undergoes
spreading and becomes H when the potential aspect marker attaches to it. This is
illustrated in Figure 2.37, where the syllable /i/ is clearly H.
Figure 2.37 [í-bí&i%=lu%$]
The example I show in Figure 2.37 is from the second-person singular so there is no
floating H tone to contend with.
The completive and habitual aspects of this verb are shown in (95):
(95a) /u% - bi&i%/ [u%bi%&i%] (95b) u%-bi%&i%=a%$
COM-fall.down COM-fall.down.=1s
‘fell down’ ‘I fell down’
160
(95c) u%-bi%&i%=lu%$ (95d) u%-bi%&i%=bí
COM-fall.down=2s.INF COM-fall.down=3s.FAM
‘You fell down’ ‘He fell down’
(95e) /ri - bi&i%/ [ribi&i%] (95f) ri-bi&i%=a%$
HAB-fall.down HAB-fall.down.=1s
‘falls down’ ‘I fall down’
(95g) ri-bi&i%=lu%$ (95h) ri-bi&i%=bí
HAB-fall.down=2s.INF HAB-fall.down=3s.FAM
You fall down’ ‘He falls down’
Examples (95b) and (95f) show that there is no floating H tone for this verb since it is
intransitive. The rest of the examples in this data set illustrate that the initial syllable of
the verb root is not H. This is also visible if we examine the pitch track of example (95g)
found in Figure 2.38:
161
Figure 2.38 [ri-bi&i%=lu%$]
49
Because the examples in data set (95) and in Figure 2.38 do not have a H tone on the first
syllable of the root, I conclude that the H tone on this syllable in the potential aspect is
due to rightward spreading from the high-toned potential aspect marker. I posit that
examples (95b) through (95d) have a M tone on the initial syllable of the verb root and
that examples (95f) through (95h) remain toneless in the output.
As discussed further in the analysis chapter 3, I would expect any tone (L, M, H)
to spread rightward onto a toneless syllable. As we have seen here, when a M tone
spreads rightward there is no evidence in the output, because we expect both a M tone
and a toneless syllable to have approximately the same pitch. The rightward spreading
only becomes obvious when a H tone spreads rightward. Likewise, we would expect that
when a low-toned aspect marker attaches to a verb with an initial toneless syllable, this L
49
It is likely that the syllable /bi/ is very slightly higher in pitch than /ri/ because /bi/ is
the stressed syllable. However neither tone is phonologically H.
162
tone would also spread rightward. See the appendix for additional examples where the
tone of the aspect marker spreads rightward onto a toneless syllable in the verb root.
2.5.6 Glottal Stops Block Tone Spreading
The data I have collected suggest that when a verb root ends in a glottal stop, the
H tone of the final syllable of the verb root fails to spread rightward. This is the case at
normal speech rate, although in some cases at a very fast rate of speech the H tone does
spread through the glottal stop.
50
I will leave the question of rate-dependent spreading for
further work and argue here that glottal stops block rightward tone spreading. I posit that
glottal stops can have this effect because they cease the vibration of the vocal folds.
Consider the examples in data sets in (96) and (97) from two different verbs:
(96a) /gú-di-ba##gá$/ [gúdíba##gá$] (96b) gú-dí-ba##gá$=lu%$
POT-CAUS-cover POT-CAUS-cover=2s.INF
‘will cover’ ‘You will cover’
(96c) gú-dí-ba#gá$=a) (96d) gú-dí-ba#gá$=tu%$
POT-CAUS-cover=3s.NFML POT-CAUS-cover=1p.EXCL
‘He will cover’ ‘We will cover’
(97a) /be%-níá$/ [be%níá$] (97b) be%-níá$=lu%$
COM-know.fact COM-know.fact=2s.INF
‘knew a fact’ ‘You knew a fact’
(97c) be%-ni%bíá$=a) (97d) be%-ni%bíá$=tu%$
COM-know.fact=3s.NFML COM-know.fact=1s.EXCL
‘He knew a fact’ ‘You knew a fact’
50
Perhaps this depends on how fully the glottal stop is realized. A larger study with more
speakers would be useful.
163
These examples illustrate that the clitics do not become H when they attach to verbs that
end with the sequence /V
. .$/. Recall from previous examples that when the final vowel of
the root has a H tone, this H tone spreads rightward onto attached clitics and replaces
their underlying tones. The examples I present here suggest that when it is present the
glottal stop blocks this process of rightward tone spreading. A visual example of this is
given in Figure 2.39.
Figure 2.39 [be%-ni%bíá$=tu%$]
This figure shows that while the final syllable of the verb root is H, the attached clitic still
appears with its usual M tone.
Another verb from a different class illustrates the same pattern where a glottal
stop blocks rightward H spread. This is illustrated in data set (98):
164
(98a) /gá$/ [gá$] (98b) gá$=lu%$
POT/make.enter POT/make.enter=2s.INF
‘will make enter’ ‘You will make enter’
(98c) gá$=a)) (98d) gá$=tu%$
POT/make.enter=3s.NFML POT/make.enter=1p.EXCL
‘He will make enter’ ‘We will make enter’
These examples show the same pattern: the H tone of the verb stem fails to spread
rightward onto the attached clitics. I posit this is due to the presence of the glottal stop.
An additional pitch track of example (98c) is given in Figure 2.40.
Figure 2.40 [gá$=a)]
Figure 2.4 shows that the clitic / a))/ emerges without a H tone. Although I have only
included a few clitics in the examples in this section, the same pattern holds for all 13
pronouns when they attach to this verb - no rightward tone spreading occurred from the
165
verb-final H tone onto any clitic. I take this as evidence that glottal stops block rightward
tone spreading in the language.
2.5.7 Summary
This section will briefly review the tonal spreading processes discussed in section
2.5. Table 2.2 summarizes the conditions and outcomes for tone spreading.
Table 2.2 Summary of tone spreading processes
Circumstance Outcome
Verb root ends with a H tone This H tone spreads onto
attached clitics causing
toneless, M, and L clitics to be
realized as H.
Verb root ends with a H tone No change is observed on CV
clitics with an underlying H
tone.
Verb root ends with a H tone The clitic /é/ undergoes a type
of lowering process.
Verb root ends with a M or L tone Thesse M and L tones do not
spread onto attached clitics
unless the clitics are
underlying toneless. In most
cases no change is evident in
the output.
H tone associated with the potential aspect
precedes a toneless syllable in the verb
root.
The toneless syllable becomes
H. Spreading is rightward from
aspect marker to root.
M tone associated with the completive or
habitual aspect marker precedes a toneless
syllable in the verb root.
The toneless syllable becomes
M. Spreading is rightward
from aspect marker to root.
Verb root ends in a glottal stop No tone spreading occurs
through the glottal stop.
166
2.6 Conclusions
This chapter has presented several tonal processes in SJZ. First, the unique
patterns of the first-person singular were discussed. I described the segmental variation in
the first-person singular clitic, and I discussed evidence that this clitic is underlyingly
toneless. Then, I discussed the floating H tone associated with the first-person singular. In
this section it was noted that transitive verbs have a H tone when they are used in the
first-person singular. In my database, this H tone was absent in intransitive verbs and in
persons other than the first-person singular.
I have termed the H tone that appears in first-person singular transitive
constructions a floating tone. This floating tone normally appears on the stressed syllable
of the verb root. If the stressed syllable of the verb root has a short vowel, then the
floating H of the first-person singular will replace the underlying lexical tone. If the
stressed syllable of the verb root has a long vowel, and if it has a specified M or L tone,
then the floating H of the first-person singular will form a falling tone when it combines
with the underlying lexical tone. There does not seem to be a distinction between
contours that result from H +L and those that result from H + M. Both are realized
approximately as HM. A phonetic study or perception experiment would be useful to
determine more precisely whether these two contours are distinct phonetically in different
environments.
I presented examples that illustrate that when a toneless aspect marker precedes
the verb root, the floating high tone of the first-person singular will appear in this location
as well as on the stressed syllable of the verb root. In no cases did the floating tone only
167
appear on the toneless aspect marker. In every case where it appeared on a toneless aspect
marker it also appeared on the stressed syllable of the verb root.
Two separate cases of rightward tone spreading were also presented. First, it was
shown that verb-final H tones spread rightward onto the clitics. This is obvious only
when the clitic does not already have a specified H tone. Thus mid and low-toned clitics
become H when they follow a verb that ends with a H tone. While this rightward
spreading occurs with verbs ending in H tones, the evidence points to verb-final M and L
tones not spreading to override exisiting tones in the same way. The H tone of the
potential also spreads rightward onto toneless syllables in the verb root. This process is
easy to detect in the language because the potential aspect always has a H tone, regardless
of the verb class. I argue that toneless syllables will obtain their tone due to rightward
spreading from whatever tone precedes them. Finally, it was shown that glottal stops
block rightward tone spreading. I argued that this occurs because cessation of glottal
vibration may impede tone production and thus tone spreading.
Although this chapter has presented many tonal perturbations and patterns of tone
spreading, I do not claim that these are the only tone processes which occur in the
language. While my database of 43 verbs was useful to determine which aspect markers
are underlyingly toneless and to observe the various ways that the floating H tone can
combine with or delete an underlying lexical tone, there may well be additional facts that
emerge as more fieldwork is completed on this language.
As I will show in the analysis chapter, the two main points from this chapter
concerning floating tones that are the focus of my analysis of SJZ are that the floating H
168
tone is drawn to the word-initial syllable and to appear on the stressed syllable of the verb
root. A second focus of my formal analysis of SJZ is the rightward spreading of tones
that occurs from toneful to toneless syllables as a result of their unique deactivation
typology. As I discuss in the next chapter, I argue that rightward spreading to toneless
syllables is driven by specific constraints that govern the deactivation of tone gestures.
The deactivation typology of tone gestures results not only in spreading from toneful to
toneless syllables, but in rightward spreading in general. Chapter 4 lays out the
typological consequences of this rightward spreading tendency.
169
Chapter 3 Analysis of Floating Tone and Tone Spreading
3.1 Introduction
This chapter will present an Optimality Theoretic analysis of the Sierra Juárez
Zapotec data presented in chapter 2. I begin by analyzing the floating tone of the first-
person singular with constraints that make reference to tone gestures. Then I move to an
explanation of rightward tone spreading. The gestural model of speech upon which these
constraints are based was introduced in chapter 1. Here I combine the representations of
articulatory gestures with the rankable and violable nature of constraints in Optimality
Theory.
The goals of this chapter are to propose a constraint system that uses an
Articulatory Phonology basis to account for tone spreading and the location of the
floating H tone of the first-person singular in SJZ. I propose a new mechanism to account
for tone spreading and show that constraints governing deactivation of gestures can
account for the rightward spreading from toned onto toneless syllables in SJZ.
Among the patterns I account for are the location of the floating tone of the first-
person singular. As I showed in chapter 2, this floating tone will appear on toneless
aspect markers if they are available, and always on the stressed syllable of the verb root. I
argue in this chapter that the floating tone acquires its relationship with the underlying
segments via the constraints I propose. I also discuss the effects of tone gestures that stay
active for more than one syllable because these effects emerge in some of the same
examples I will use to explain the placement of the floating H tone gesture. In section 3.5
170
of this chapter I then discuss tone spreading behaviors exclusively to further illustrate and
account for those patterns.
3.2 Constraints for the floating tone
This section introduces several constraints and begins an analysis of the
placement of the floating H tone of the first-person singular. First I present new
constraints that use Articulatory Phonology representations. These new constraints make
use of AP representations in a way that can explain the rightward bias in tone spreading
and account for the placement of the floating H tone in SJZ. This is accomplished by
referencing both the onset (coupling location) as well as the end point (point of
deactivation) of a tone. Constraints also work to prohibit the overlapping of gestural
activation. This approach is unique in that it captures the data from SJZ and makes cross-
linguistic predictions about tone spreading and its relation to prominent syllables. After
presenting the new constraints, I show how the ranking between these constraints
accounts for the patterns observed in the SJZ data. Finally, I give examples from chapter
2 and illustrate how the constraints I propose interact to obtain appropriate output forms.
The first constraint I propose acts to hold underlying tone gestures to the coupling
relationships they have in the input. Its definition is given in (1):
(1) PRESERVE-COUPLING–IO: Let X and Y be gestures in the input, and X’
and Y’ be correspondents of X and Y in the output, respectively.
If X is coupled with Y in the input, then X’ is coupled to Y’ in the
output.
171
This constraint, which I abbreviate as PRES-COUP-IO, works to ensure that coupling
relationships in input forms are maintained in output forms. This constraint formulation is
similar to the IDENT constraints proposed first by McCarthy and Prince (1995), with the
exception that it references gestures. Since coupling is the glue that holds gestures
together in the Articulatory Phonology model, I propose a constraint that ensures that
their preservation is part of the phonological grammar. With reference to tone, this
constraint will also prevent leftward movement or ‘spread’. For example, in a word like
/babá/, where the last vowel has a H tone, I assume that this tone gesture is coupled to the
second consonant /b/ and to the vowel /a/. Therefore if an input such as /babá/ became
[bábá] in the output, with leftward tone spread, I assume the tone gesture has coupled to a
new location and consequently violated PRES-COUP-IO. This is illustrated in Figure 3.1
below:
Figure 3.1 Coupling of [babá] and [bábá]
a. [babá] b. [bábá]
C C H C H C
V V V V
In (a), there are two CV syllables, and the second syllable has a tone gesture coupled in-
phase with the vowel and anti-phase with the consonant. In-phase coupling is shown with
172
a solid line and anti-phase coupling is shown with a dashed line. These figures do not
show activation of gestures, just coupling, so in (b), I assume that while the tone gesture
is coupled to the vowel and consonant in the first syllable, it remains active through the
second syllable. Thus the output like [bábá] would have one tone gesture coupled to the
/b/ and the /a/ of the initial syllable, but which remained active through the second
syllable.
The constraint PRES-COUP-IO does not directly replace autosegmental faith
constraints. Rather, I make different assumptions about the representations which
necessitate different constraints. Yip proposes constraints like those defined below in (2)-
(3) (2001: 83):
(2) *ASSOCIATE: No new association lines.
(3) *DISSOCIATE: No removal of association lines.
The constraints *ASSOCIATE and *DISSOCIATE make reference to the association lines
used in autosegmental representations. Since I do not assume this type of framework, I do
not penalize the removal or insertion of association lines. I compare the approach adopted
in this chapter with autosegmental analyses of tone in chapter 4. Here, it is sufficient to
state that PRES-COUP-IO penalizes changing coupling relationships in the respect that it
penalizes a loss in coupling relationships between input and output. In an autosegmental
representation, shifting the position of a tone or spreading it would have incurred
violations of *DISSOCIATE (for having removed a tone’s association to a syllable) and
173
*ASSOCIATE (for the new tone bearing unit). PRES-COUP-IO is not violated when a new
tone is added, unless this would cause changes to the underlying coupling relationships.
Again consider the example /babá/ becoming [bábá]. Under Yip’s constraints, this
movement would violate *ASSOCIATE, whereas it violates PRES-COUP-IO under the
analysis that I adopt here. In a case of /bába/ becoming [bábá], under Yip’s constraints a
violation of *ASSOCIATE would occur, whereas there would be no violation of PRES-
COUP-IO under my analysis. If there is no coupling present in the input, and then
coupling is established between a tone and another gesture in the output, I do not assume
this violates anything. Hence I do not adopt a gestural version of *ASSOCIATE. My
approach also differs from those such as Jiang-King (1996), where it is proposed that
tones link directly to a mora via the prosodic anchor hypothesis (1996:77-80). In the
approach I adopt here, it is only coupling relationships that are established, rather than
association to a segment, syllable, or mora.
To make the constraint PRES-COUP-IO and its function clearer, I will present a
tableau with four basic candidates and show how each violates or obeys this constraint
and why. Here, and in all following tableaux, a tone gesture’s activation is shown via a
rightward pointing dashed arrow if it extends beyond 360 degrees of its own clock. That
is, whenever a tone extends past the syllable where it is coupled (due to its failure to
deactivate at 360 degrees of its own clock), I will illustrate it with an arrow. This is
illustrated in Table 3.1:
174
Table 3.1 PRES-COUP-IO
ta bá
H
1
PRESCOUPLE-IO
a. ta bá
H
1
b. tá ba
H
1
*
c. tá bá
H
1
*
d. tá bá
H
2
H
1
In Table 3.1, I have used subscripts to notate correspondence relations among the tones.
First, consider the input given in Table 3.1. The vertical line connecting the H tone with
the /bá/ syllable shows the coupling relationship. This vertical represents that I assume
the H tone gesture is coupled in-phase to the vowel gesture of this syllable.
51
Thus the H
tone is realized on the /ba/ syllable. Candidate a) preserves this coupling relationship and
consequently does not violate PRES-COUP-IO. Candidate b) shows a H tone which is
connected to the /ta/ syllable. This incurs a violation of the constraint because the
coupling relationship has been changed. Candidate c) incurs a violation of PRES-COUP-IO
51
Earlier, I discussed that a tone couples in-phase to a vowel but anti-phase to the onset
consonant. For simplicity I will not show the anti-phase coupling relationships in
tableaux or violations of C-T coupling when a tone moves.
175
because it changes the coupling relationship between input and output. The H tone
gesture is coupled with the initial vowel and then stays active through the second
syllable. The rightward facing dashed arrow indicates that a gesture remains active. The
gesture is active for as long as the arrow shows. It is important to notice in candidate c)
that the H tone gesture is not coupled to /ba/ but merely that it is active during this
syllable as shown via the rightward pointing arrow. As the vertical line shows, the H tone
gesture in candidate c) is coupled in-phase to the vowel. Candidate d) illustrates the fact
that inserting a new tone gesture will not violate PRES-COUP-IO. Observe that the original
H tone, which is marked as H
1
preserves its coupling relationship with the second
syllable in d). A separate H tone which I have marked as H
2
has been inserted on the
syllable /ta/. This does not violate PRES-COUP-IO because the H
2
had no coupling
relationship in the input. Thus PRES-COUP-IO serves to prevent leftward spreading,
because leftward spreading similar to what occurs in candidate b) implies a change in
coupling relationship. I will discuss the implications of my theory for the rightward bias
in tone spreading in this chapter as well as in chapter 4.
A constraint such as PRES-COUP-IO is needed to account for the floating tone in
SJZ because while the floating tone is drawn to the stressed syllable and to the initial
syllable, this is not so for other tone gestures in the input. PRES-COUP-IO will prevent
tone gestures which are already coupled from altering their coupling relationships. Thus
while activation and deactivation of tones in SJZ is more flexible as I will discuss further,
this is not so for the coupling relationships which govern the onset of tone gestures. The
effects of PRES-COUPLE-IO allow various other constraints I propose here to be visibly
176
active only for the floating tone since the floating tone has no coupling relationship
established in the input. As I will argue in this chapter, the location where the floating
tone ultimately gets coupled is determined by the constraints that operate in SJZ.
To explain these ideas more fully I will first present some additional constraints
and then go over examples with relevant tableaux that illustrate how PRES-COUP-IO
interacts with these constraints to give rise to the patterns observed in the SJZ data.
The constraint which is responsible for drawing the floating tone to the vowel of the
stressed syllable is shown in (4).
(4) H-TONE-TO-STRESS V: Assign a violation to any H tone gesture whose
deactivation is triggered other than by the vowel gesture that is
coupled in-phase with a stress oscillator.
52
This constraint will assign violations to any H tone gesture in the output that does not
deactivate with the vowel of the stressed syllable. The constraint presumes that a separate
stress oscillator is responsible for making the stressed syllable stressed, and that the stress
oscillator couples in-phase with the vowel in its respective syllable. Evidence for a
separate stress oscillator has been discussed and analyzed by Tilsen (2011, 2009). The
constraint I use here called H-TONE-STRESS-V makes no reference to coupling, so the
constraint is applicable to any H tone gesture that is present in the output without
52
Recent work in the AP model has found evidence for a syllable or foot-level oscillator
(Tilsen 2009, 2011, Goldstein, personal communication). Future work could incorporate
the gestural approach to constraints that reference a syllable. For example, a constraint
like “H-TONE-STRESS-SYLLABLE”.
177
evaluating the coupling of this tone gesture to any other gesture. Instead, H-TONE-TO-
STRESS-V penalizes any H tone that deactivates anywhere except at the vowel gesture
coupled in-phase to the stress oscillator. This is necessitated by data that show a tone can
couple to a vowel prior to the stressed vowel, but remain active up into the stressed
syllable’s vowel. In a given candidate set, we can determine whether H-TONE-TO-
STRESS-V is satisfied by observing the location of the H tones. If they do not appear
during the stressed vowel, then a violation is incurred. A tone can satisfy the constraint
either by being coupled to the vowel of the stressed syllable and also deactivating with
said vowel, or by being coupled to some vowel gesture to the left of the stressed vowel
but remaining active through the realization of the stressed vowel gesture and then
deactivated with it.
This constraint is similar in its effect to those that have been proposed previously
which show that H tones are often drawn to stressed syllables or prominent positions. The
constraint I propose differs only in its specificity to the gestural endpoint. For example,
de Lacy (2002) proposes a theory of the interaction between tone and prominence. He
notes that prosodic heads prefer higher tone over lower tone, while non-heads have the
opposite preference. He proposes a tonal prominence scale where higher tone is more
prominent than lower tone: H > M > L (2002:2). The generalizations de Lacy makes are
expressed in OT constraints. The tonal prominence scale he proposes is combined with
the structural positions foot head (HD) and foot non-head (NON-HD) to form constraints in
a ranking. For example, the ranking *HD/L >> *HD/M means that low-toned foot heads
will incur a violation of a higher-ranked constraint than a mid-toned foot head. Since
178
there is no constraint against a high-toned foot head, that type is assumed to be the least-
marked (2002:3).
The ranking *NON-HD/H >> *NON-HD/M works to rule out higher-toned foot
non-heads. The fixed ranking of foot non-head constraints gives an implicational
hierarchy so that high-toned foot non-heads are least harmonic, mid-toned non-heads are
relatively more harmonic, and low-toned non-heads are the most harmonic (2002:4). As
de Lacy points out, apart from these fixed rankings, any *NON-HD/TONE constraint can
outrank any *HD/TONE constraint and vice versa (2002:4).
The constraints proposed in de Lacy (2002) are presented with stringent form in
de Lacy (2007:297):
(5) Tone-head and non-head constraints (after de Lacy 2007)
(a) *Hd
α
/L, *Hd
α
/L⋅M , *Hd
α
/L⋅Μ⋅Η
(b) *non-Hd
α
/H, *non-Hd
α
/H⋅Μ, ∗non-Hd
α
/H⋅Μ⋅L
For example, the first constraint in (a), *Hd
α
/L, would read: ‘incur a violation for every
head of constituent α that contains a L vowel’. These constraints are in a subset-superset
relation in terms of their violation marks, so they are in a stringency relation. As de Lacy
(2007) shows, the stringent ranking of constraints can account for cases such as Ayutla
Mixtec, where the foot is attracted to the left edge of a word, but will appear elsewhere if
the ‘perfect toned foot’ can be formed, where the head has a high tone and the non-head
has a low tone (2007:316).
De Lacy’s work is important to the constraint I propose here because he shows
that it is cross-linguistically common for foot heads (stressed syllables) to have a H tone.
179
Yip points out that the preference for H tones to appear on stressed or prominent syllables
is reflected in various languages: a H tone is inserted on stressed syllables in Lithuanian,
a H tone moves to a stressed syllable in Zulu and Digo, and stress tends to avoid L toned
syllables in Golin and Mixtec (Yip, 2002: 97). Jennifer Smith also discusses
augmentation of stressed syllables, including via H tone insertion (2002). In light of these
facts, it is not surprising that the floating H tone of SJZ is attracted to the stressed
syllable.
53
Ranking PRES-COUP-IO above H-TONE-TO-STRESS-V means that only the
floating tone will be drawn to the stressed syllable. This means we do not predict that a
lexical H tone will move its coupling just to satisfy H-TONE-TO-STRESS-V. The ranking
does predict, however, that a lexical H tone which precedes the stressed syllable could
spread to the stressed syllable and stop there. As shown in following examples, this
pattern happens in MacZ when the H tone of the potential aspect marker spreads
rightward to the stressed syllable and stops there. In SJZ, as I showed in chapter 2, the
floating H tone of the first person singular may couple to the initial vowel and stay active
up to the stressed syllable, deactivating there. There are no examples from SJZ that have
a toneless, verb-root initial syllable followed by a H-toned syllable to illustrate that a
lexical tone will not move its coupling to satisfy H-TONE-TO-STRESS-V. Instead, we can
observe the interaction between these two constraints by considering its application to a
hypothetical example, as shown in Table 3.2:
53
I have also observed that when loanwords from Spanish are used in SJZ, the stressed
syllable is assigned a lexical H tone.
180
Table 3.2 PRES-COUP-IO >> H-TONE-TO-STRESS V
CV CV CV
M X H
PRESCOUPLE-IO H-TONE-TO-
STRESS-V
a. CV CV CV
M H X
*!
☞ b. CV CV CV
M X H
*
c. CV CV CV
M H
*! *
The example candidates have lines connecting the tone to the vowel. These lines
indicate in-phase coupling between the tone and the vowel. If the tones are coupled to the
vowel they are also active during the vowel. Table 3.2 shows what happens when PRES-
COUP IO >>H-TONE-TO-STRESS-V. The ranking has the effect that underlying lexical
tones will not move just to satisfy H-TONE-TO-STRESS-V. Candidate a) has moved the
final H tone onto the stressed syllable in order to satisfy the constraint, but violates PRES-
COUP-IO and loses. Candidate b) violates H-TONE-TO-STRESS-V but avoids a violation of
the higher-ranked PRES-COUP-IO. Candidate c) shows that a H tone has moved to couple
to the vowel in the stressed syllable and then stays active through the vowel of the final
syllable. This incurs a violation of PRES-COUP-IO and the candidate loses. Candidate c)
also violates H-TONE-STRESS-V because the tone does not deactivate with the vowel of
181
the stressed syllable. Due to this ranking, only floating H tones will be drawn to the
stressed syllable, while underlying lexical tones will not move. This ranking also shows
that the PRES-COUP-IO prevents leftward spreading by prohibiting the movement of
coupling relationships. In upcoming tableaux that make use of H-TONE-TO-STRESS-V I
generally do not show PRES-COUP-IO since I have established here that it must outrank
H-TONE-TO-STRESS-V, and I will not consider candidates that have lost coupling
relationships of the input.
MAX-T must also outrank H-TONE-TO-STRESS-V because the latter constraint will
not drive tones to delete underlying, lexical tones. MAX-T is defined in (5), and the
relationship with H-TONE-TO-STRESS-V is illustrated in Table 3.3.
(6) MAX-T: Tone gestures in the input have output correspondents.
54
54
This definition of MAX-T is identical to the MAX-T constraint presented in Yip 2002
with the exception that it makes reference to a tone gesture and not an autosegment.
182
Table 3.3 MAX-T >> H- TONE-TO-STRESS-V
ú- ka#$ná=lú$
H L H H
MAX-T H-TONE-TO-
STRESS-V
☞ a. ú- ka#$ná=lú$
H L H H
***
b. ú- ká$na=lú$
H H H
*! **
c. ú- ká$ná=lú$
H H H
*! **
d. ú- ká$ná=lú$
H
*!** *
Table 3.3 illustrates that MAX-T can prevent tones from moving their onset so they can
deactivate with the vowel in the stressed syllable. That is, it prevents deletion of tones in
order to improve satisfaction of H-TONE-TO-STRESS-V. In candidate a) there are three
violations of H-TONE-TO-STRESS-V, and the stressed syllable has an underlying L lexical
tone. Candidate b) incurs a violation of MAX-T by moving a neighboring tone onto the
stressed syllable and deleting the underlying lexical tone of the stressed syllable. A
violation would also be incurred if the tone gesture of the initial syllable remained active
through the stressed syllable and caused deletion of the underlying lexical tone in that
position, as shown via candidate c). Therefore MAX-T dominates H-TONE-TO-STRESS-V.
183
Next, recall that the floating tone of the first-person singular is drawn to the initial
syllable of the prosodic word. The floating tone shows up in this location only if there is
no underlying tone already occupying that position. To account for appearance of the
floating tone on a toneless aspect marker, a constraint is needed which will call tone
gestures to couple to the initial vowel of the prosodic word. I call this constraint
COUPLE-INITIAL-VOWEL and define it in (7):
(7) COUPLE-INITIAL-VOWEL: Assign a violation mark to any tone gesture
which is not coupled to the initial vowel gesture of the prosodic
word.
This constraint will be violated by any candidate with tones that are not coupled to the
vowel gesture in the initial syllable. Recall that when a tone gesture is coupled to a vowel
gesture in a given syllable it is also active during that vowel (and possibly in vowels that
follow it but not vowels that come before). The function of COUPLE-INITIAL-V is to draw
the floating tone to the initial syllable’s vowel. In all candidates I show, a straight line
indicates that the tone and vowel are coupled and therefore that the tone is active during
the vowel to which it is coupled. In a case where a tone stays active longer than the vowel
to which it is coupled, a rightward pointing arrow is shown in the candidate. PRES-COUP-
IO must outrank COUPLE-INITIAL-V in order to ensure that only the floating tone is drawn
to the initial vowel. Observe the interaction between these two constraints in Table 3.4.
Here, the input is a word with a tone gesture already coupled to a vowel gesture in the
initial syllable, and all other tones in the word are coupled with gestures in other
184
syllables. Since PRES-COUP-IO is ranked above COUPLE-INITIAL-V, no coupling
relationships are changed between input and output.
Table 3.4 PRES-COUP-IO >> COUPLE-INITIAL-VOWEL
í - la%tsu#$=lu%$
H M L M
PRESCOUP-IO COUPLE-INITIAL-V
☞ a. í - la%tsu#$=lu%$
H M L M
***
b. i !-la tsu#$=lu%$
H M L M
*! **
Candidate a) is faithful to the input and incurs three violations of COUPLE-INITIAL-V,
since there are three tones which are not active during the initial vowel /í/. Candidate b)
has altered the coupling of the M tone on the stressed syllable. In the input this M tone is
coupled to the vowel in the syllable /la/, while in the output this tone is coupled to the
initial vowel /í/. This coupling change incurs fewer violations of COUPLE-INITIAL-V but it
causes the candidate to violate the higher-ranked PRES-COUP-IO. Therefore candidate b)
loses.
To further illustrate, I include a tableau showing that the ranking PRES-COUP-IO
>> COUPLE-INITIAL-V can draw a floating tone to the vowel of the initial syllable if the
initial syllable is toneless. To accomplish this I will use a hypothetical input. The
hypothetical input is needed because there are no cases in SJZ where the floating tone is
185
drawn only to the initial syllable and not also to the stressed syllable. Both COUPLE-
INITIAL-V and H-TONE-TO-STRESS-V are needed to account for the SJZ data. Table 3.5
illustrates the interaction between PRES-COUP-IO and COUPLE-INITIAL-V:
Table 3.5 COUPLE-INITIAL-V and toneless initial syllable
CV- CV CV
L M
PRES-COUP-IO COUPLE-INITIAL-
VOWEL
☞ a. CV- CV CV
H
1
L M
**
b. CV CV CV
H
1
L M
*! *
Candidate a) has two violations of COUPLE-INITIAL-V for the L and M tones that are non-
initial. The floating tone is attracted to the initial syllable since it is toneless. Candidate a)
illustrates that placing the floating H tone of the first-person singular in initial position
does not violate PRES-COUP-IO. Candidate b) shows that moving an underlying lexical
tone in order to have fewer violations of COUPLE-INITIAL-V is non-optimal, since this
violates higher-ranked PRES-COUP-IO.
While COUPLE-INITIAL-V and H-TONE-TO-STRESS-V are needed to account for
the behavior of the floating tone in SJZ, PRES-COUP-IO must outrank both of them. This
section has argued that tone gestures will not change their coupling relationships between
input and output despite pressure from constraints calling for tones to be coupled to the
H
1
186
word initial vowel or to be active during the vowel of the stressed syllable. In the
approach I have taken here, PRES-COUP-IO will hold tones with established coupling
relationships in place while only affecting the floating tone.
Next, I will present a more complicated example where the vowel in the stressed
syllable of the verb root has a specified tone gesture, but where the vowel is short. As
described in chapter 2, in these cases the floating H tone of the first-person singular does
not form a contour by combining with the underlying lexical tone. Instead, the floating H
deletes the underlying lexical tone and replaces it. To account for this pattern, a few more
constraints are necessary, which I will define below.
The first constraint that is needed in this case is MAX-UNCOUPLED-T. This
constraint is necessary to rule out simple deletion of the floating tone in the output of
first-person singular forms. I also propose a constraint called *UNCOUPLED-TONE. These
constraints are defined formally in (8) and (9) below:
(8) MAX-UNCOUPLED-TONE – Uncoupled tone gestures in the input have
output correspondents.
(9) *UNCOUPLED-T: Assign a violation mark for each tone gesture in the
output that is not coupled with another gesture in the
output.
I do not show *UNCOUPLED-TONE in tableaux in this chapter because I assume this
constraint is part of GEN. This is based upon the theory of Articulatory Phonology
discussed in detail in chapter 1. Recall that coupling is the glue that holds gestures
187
together. I propose that a gesture cannot exist in the output if it is not coupled and
therefore integrated into the coupling graph that governs gestural coordination. Therefore,
no candidate outputs are generated in which a tone remains “floating” in the output. This
is similar to the *FLOAT constraint
55
used by Yip (2002). The constraint MAX-
UNCOUPLED-TONE is similar to the MAXFLOAT
56
constraint proposed by Wolf (2007)
57
and to the MAX-SUBSEG constraints proposed by Zoll (1996). For the SJZ data, the
constraint MAX-UNCOUPLED-TONE is necessary to prevent deletion of the floating tone.
Three more constraints are needed for the example showing failure to form a
contour tone on a syllable with a short vowel. These are *MULTI-TONE-SHORT-VOWEL,
*Hd/L, and *MULTI-TONE. First I define the constraints in (10) – (12), and then I show
how they are ranked, in some cases with hypothetical inputs. After showing how these
constraints are ranked with respect to each other and to PRES-COUP-IO, I will present a
more complex example from SJZ where all of these constraints are needed to correctly
predict the output form.
(10) *MULTI-TONE-SHORT-VOWEL- Assign a violation mark for each short
vowel gesture whose activation overlaps the activation of more
than one tone gesture.
(11) *Hd/L : ‘No L on head syllables’ (de Lacy 2002:2)
55
*FLOAT: A tone must be associated with a tone bearing unit (Yip, 2002: 83).
56
MAXFLOAT: All autosegments that are floating in the input have output correspondents
(Wolf 2007: 2).
57
Unlike previous analyses that posit a floating tone which may remain floating in the
output and cause downstep, I assume that if a tone is present at all in the output, it must
be coupled. This topic merits analysis in the future.
188
(12) *MULTI TONE: Assign a violation mark for each vowel gesture whose
activation overlaps the activation of more than one tone gesture.
*MULTI-TONE-SHORT-VOWEL will prevent two tone gestures from appearing in syllables
which are comprised of short vowels.
58
The constraint is violated if two tone gestures are
coupled with a vowel gesture or whenever the activation of more than one tone gesture
overlaps the activation of the vowel gesture. The percent overlap must be 25% of the
activation duration of a gesture. This constraint makes the claim that EVAL can reference
overlap of the gestural score. This constraint is needed because, as I present in chapter 2,
in the SJZ data, a contour fails to form if the vowel is short. Only long vowels permit the
formation of contour tones in SJZ . *MULTI-TONE is the gestural version of NOCONTOUR
(Yip, 2002).
59
*HD/L will break a tie between two candidates and will penalize candidates with a
L tone on the vowel of the stressed syllable. While de Lacy writes this constraint as “No
L on head syllables”, we can understand the constraint to mean that L tone gestures
coupled to the vowel which is coupled in-phase to a stress oscillator will incur violations.
To make the relationships between these constraints clear, I will first establish
their ranking using pairwise ranking comparisons. First, I show that several of these
constraints must dominate MAX-T.
58
Vowel gestures deactivate soon after their targets are reached, so even for vowels of
the same quality in adjacent syllables, the constraint only references the vowel of one
syllable at a time.
59
*NOCONTOUR: A tone bearing unit may be associated with at most one tone (Yip,
2002: 83).
189
The constraint *MULTI-TONE-SHORT-VOWEL must dominate MAX-T because in
cases where the stressed syllable of the verb root has a short vowel, a contour tone is not
formed and the lexical tone is deleted in favor of the floating H tone. This relationship is
illustrated in Table 3.6:
Table 3.6 *MULTI-TONE-SHORT-VOWEL >> MAX-T
í- la%tsu#$-a#$
H M L L
*MULTI-TONE-
SHORT-VOWEL
MAX-T
☞ a. í- látsu#$-a#$
H H
1
L L
*
b. í- la! tsu#$-a#$
H H
1
M L L
*!
In Table 3.6 I focus solely on the relationship between MAX-T and *MULTI-TONE-
SHORT-V. Candidate b) is not optimal because it violates *MULTI-TONE-SHORT-V. Here,
no extended activation is illustrated in the candidate (via an arrow) because both tones are
coupled to the vowel gesture /a/ in candidate b). This means that both tone gestures
overlap the vowel gesture and thus that the constraint *MULTI-TONE-SHORT-V is
violated. Candidate a) is the winner because it has deleted the underlying lexical tone to
avoid forming a contour on a short vowel.
H
1
190
MAX-T must dominate *Hd/L. Lexical L tones will not be deleted under ordinary
circumstances just to satisfy *Hd/L. It is only in cases where the floating tone is added
that there are changes to the tone of the vowel in the stressed syllable. In Table 3.7, I
show that MAX-T dominates *HD/L.
Table 3.7 MAX-T >> *HD/L
ta% - ne#llé
M L H
MAX-T *HD/L
☞ a. ta% -ne# llé
M L H
*
b. ta% -ne llé
M H
*!
Table 3.7 illustrates that *HD/L is a low-ranking constraint and will not cause lexical
tones to be deleted just to satisfy the prohibition on L tones coupled to the vowel of a
stressed syllable. This constraint will come into play in a following tableau where it is
needed to break a tie between two candidates that both violate MAX-T. Because MAX-T
outranks it, we do not see the effects of *HD/L in most cases. Rather, *HD/L is important
only when it conspires to place the floating H tone on the stressed syllable.
Finally, I illustrate in Table 3.8 that MAX-T dominates *MULTI-TONE. This
ranking applies because in cases where the stressed syllable of the verb root has a long
vowel, a falling contour tone emerges instead of deletion of the underlying tone. That is,
191
it is better to violate *MULTI-TONE than to delete an underlying lexical tone. This is
illustrated in Table 3.8:
Table 3.8 MAX-T >> * MULTI-TONE
CV- CV"
H M
MAX-T *MULTI-TONE
☞ a. CV- CV "
H H
1
M
*
b. CV- CV "
H H
1
*!
Table 3.8 shows an input with a stressed long vowel. As will become evident in
following tableaux from SJZ, when the stressed syllable of the verb root has a long
vowel, the floating H tone forms a contour tone when it combines with the underlying
lexical tone. Table 3.8 illustrates the ranking that predicts this: a violation of MAX-T, as
shown in candidate b), is dispreferred to a violation of *MULTI-TONE.
Now that I have illustrated these basic rankings I move to more complex
examples from SJZ. As will become apparent, almost every example from SJZ involves
the interaction of several constraints to give the correct output form.
Putting all of these constraints together, I show that the form/ ú-ka#$ná/ becomes
[ú-ká$ná] after the floating H of the first-person singular is coupled. In this example the
H
1
192
floating H of the first-person singular replaces the underlying tone of the stressed
syllable. The input and possible outputs are illustrated in Table 3.9.
Table 3.9 Floating H replaces underlying L tone on verb root in SJZ
ú- ka#$ ná=$
H
2
L H
3
MAX-
UNCPLED
-TONE
*MULTI-
TONE-
SHORT-
VOWEL
MAX-
T
H-
TONE-
TO-
STRESS-
V
*HD/L
a. ú- ka#$ ná=$
H
2
L H
3
*!
*
**
*
☞ b. ú- ká$ ná=$
H
2
H
1
H
3
*
**
c. ú- ka#$ ná=$
H
2
L H
1
*
**
*!
d. ú- ka#$ ná=$
H
1
L H
3
* ** *!
e. ú- ká #$ ná=$
H
2
H
1
L H
3
*!
**
*
Table 3.9 shows an input with a floating H tone, and with a specified L tone on the vowel
in the syllable that receives stress. The aspect marker in this case is H-toned /ú/, so the
initial syllable already has a tone gesture in the input. The floating H tone of the first-
H
1
193
person singular cannot appear on the aspect marker due to its underlying tone and instead
is drawn only to the vowel of the stressed syllable of the verb root. In the winning
candidate b), the underlying L tone of the input is replaced with the floating H tone. This
violates MAX-T but avoids a violation of *MULTI-TONE-SHORT-VOWEL and MAX-
UNCOUPLED-TONE. Candidate a) shows that MAX-UNCOUPLED-T will rule out deletion of
the floating tone before it is coupled. Again, I argue that since the floating tone always
appears in first-person singular transitive forms, MAX-UNCOUPLED-T is high-ranked in
SJZ. There are no cases where the constraint is dominated since the floating tone always
appears in output forms.
Next, consider candidate c), which ties with winning candidate b) except for the
constraint *HD/L. In candidate c) the floating H tone of the first-person singular replaces
the underlying lexical tone of the final syllable, which is deleted. Therefore both winning
candidate b) and loser candidate c) violate MAX-T. These two candidates are
distinguished by the constraint penalizing a head syllable with a L tone. In candidate c),
the floating tone appears on the final syllable instead of on the initial or stressed syllable,
and the candidate loses. Candidate d) shows that replacing the lexical tone of the initial
vowel with the floating H tone will incur violations of MAX-T, H-TONE-STRESS-V and
*HD/L, eliminating the candidate. Finally, candidate e) shows that *MULTI-TONE-SHORT-
VOWEL will prevent the floating H tone of the first-person singular from appearing on the
stressed syllable when this syllable is comprised of a short vowel. The resulting pattern is
to delete the underlying lexical tone of the stressed syllable and replace it with the H of
194
the first-person singular. This is the result that emerges from the constraint ranking
shown in Table 3.9.
The same effects are obtained when the stressed syllable of the verb root has an
underlying H tone. The constraint ranking I present predicts that the floating H of the
first-person singular will replace an underlying H tone of the verb root. As I discussed in
the data chapter, in these cases there is no change evident in the output when the floating
H tone is realized, but I propose that the output is reflecting a structure which contains
the first-person singular H tone. A tableau illustrating this is given in Table 3.10:
Table 3.10 Floating H replaces underlying H on stressed syllable of verb root
be%-júnna%=$
M H
2
M
MAX-
UNCPLED-
TONE
*MULTI-
TONE
SHORT-
V
MAX-
T
H-TONE-
TO-
STRESS-V
a. be%-júnna%=$
M H
2
M
*! *
☞ b. be%-júnna%=$
M H
1
M
*
c. be%-júnna%=$
MH
1
H
2
M
*!
Candidate a) is ruled out because the floating H tone is deleted. This violates MAX-
UNCOUPLED-TONE. Candidate c) is ruled out by the constraint *MULTI-TONE-SHORT-V,
H
1
195
which prohibits more than one active tone gesture in a syllable with a short vowel.
Candidate b) emerges as the winner, where the floating H of the first-person singular has
replaced the underlying H lexical tone on the vowel of the stressed syllable.
A ranking may be used to explain a case where the floating H of the first-person
singular fails to form a falling contour on an open syllable with a short vowel. In the
example shown in Table 3.10 we did not observe that more than one tone gesture
appeared on the stressed syllable (the H tone did not become extra-long on that syllable).
In the example in Table 3.10 I assume that the geminate (fortis) consonant was following
the vowel in question. To show that no contour forms with a short vowel and that this is
not related to a following geminate or fortis consonant, I include the example in Table
3.11. The constraint *MULTI-TONE-SHORT-VOWEL will prevent two tone gestures from
being active on a short vowel regardless of whether these have the same or different
articulatory targets. The constraint from de Lacy called *HD/M is needed for this
example and is defined below:
(13) *HD/M: No M on head syllables (de Lacy 2007: 2)
*HD/M will break a tie between two candidates and will penalize candidates with a M
tone coupled to the vowel of the stressed syllable. While de Lacy writes this constraint as
“No M on head syllables”, we can understand it to mean that a M tone gesture coupled to
the vowel which is coupled in-phase to a stress oscillator will incur a violation. MAX-T
dominates *HD/M just as it does *HD/L. Under normal circumstances, it is fine for a
196
stressed syllable to have a M or L lexical tone. It is only in cases involving the floating H
tone of the first-person singular that we see the effects of *HD/M.
Table 3.11 shows a verb whose root is /kka%bí/. The first syllable is stressed,
contains a short vowel and has a specified M tone. The second syllable has an underlying
H tone. The clitic /á$/ is H due to a separate spreading process in the language. To
simplify this example I have illustrated the clitic with its own tone gesture. I will examine
tone spreading and relevant examples in following sections. The candidates in Table 3.11
focus on the placement of the floating H of the first-person singular, which is shown as
H
1
in the input.
Table 3.11 No contour on open syllable with short vowel
é - kka% bí=á$
H
2
M H
3
H
4
*MULTI-
TONE-
SHORT-
VOWEL
MAX-
T
H-TONE-
TO-STRESS-
V
*MULTI-
TONE
*HD/M
a. é - kka% bí=á$
H
2
M H
1
H
4
* *** *!
☞ b. é - kká bí=á$
H
2
H
1
H
3
H
4
* ***
c. é - kka! bí=á$
H
2
H
1
M H
3
H
4
*!
*** *
H
1
197
In candidate a) the specified lexical H tone of the second syllable of the verb root has
been deleted and replaced with the floating H tone. This incurs a violation of MAX-T.
Candidate a) also violates H-TONE-TO-STRESS-V three times for the other tones not
active during the stressed syllable. The constraint *HD/M breaks the tie between
candidates a) and b), and candidate b) emerges as the winner. In this case the underlying
M tone of the stressed syllable is deleted and the floating H tone of the first-person
singular takes its place. Candidate c) shows the effects of the constraint *MULTI-TONE-
SHORT-VOWEL. In candidate c) the floating H tone of the first-person singular has
coupled to the stressed syllable and has avoided deletion of the underlying M tone. This
forms a falling contour. Candidate c) loses because it contains two active tone gestures on
a short vowel.
As I presented in chapter 2, when the vowel of the stressed syllable in the verb
root is long then contour tones do form in this location when the floating H of the first-
person singular is added. This suggests that it is better in SJZ to allow contours on the
stressed syllable and avoid deleting underlying tones, so long as the syllable in question
has a long vowel. Zhang (2004) showed via crosslinguistic survey that contour
restrictions are a function of rime type, not directionality. Likewise, Zoll (2003)
recognizes the same pattern whereby long vowels are more likely to have contour tones
as compared to short vowels, and proposes a markedness constraint that only licenses
contours on long vowels (2003: 234). The data I present on SJZ agree with the
observations of Zoll (2003) and Zhang (2004), although I analyze them in a different
way. To obtain the SJZ pattern, the same ranking that was shown in Table 3.11 is
198
applicable, although we do not need to make reference to as many constraints. Here the
ranking *MULTI-TONE-SHORT-VOWEL >> MAX-T>> H-TONE-TO-STRESS-V, *MULTI-
TONE is sufficient. I already illustrated the ranking MAX-T >> *MULTI-TONE in an earlier
example. This ranking establishes that it is better to avoid deletion of a tone and to allow
a contour to form, than the reverse. Since the syllable in question does not contain a short
vowel, the constraint banning more than one tone gesture in this position is not relevant.
Table 3.12 illustrates the ranking:
Table 3.12 Contour forms when stressed syllable of verb root has long vowel
ú ta%" bi# = a#$
H
2
M L L
*MULTI-
TONE-
SHORT-V
MAX-T H-TONE-
TO-
STRESS-
V
*MULTI-TONE
☞ a. ú ta!" bi# =a#$
H
2
H
1
M L L
* *
b. ú ta%" bi! = a#$
H
2
M H
1
L L
*!
**
*
c. ú tá" bi# =a#$
H
2
H
1
L L
*! *
The input to Table 3.12 shows a high-toned potential marker /ú/ followed by the verb root
/ta#bi#/. The clitic /a#$/ is L due to a separate spreading process that I discuss in a later
section of this chapter. Here I show the clitic with its own tone gesture to simplify the
H
1
199
example. Winning candidate a) incurs one violation of H-TONE-TO-STRESS-V because
there is one tone gesture which is not active during the vowel of the stressed syllable.
One violation of *MULTI-TONE is incurred because the floating H of the first-person
singular has coupled to the stressed syllable and instead of deleting the underlying lexical
tone, it forms a contour in this location. Candidate b) shows a candidate where the
floating tone has coupled to a location other than the stressed syllable. Since the location
where the tone couples is a syllable with a short vowel, candidate b) violates *MULTI-
TONE-SHORT-VOWEL. Candidate b) also incurs an additional violation of H-TONE-TO-
STRESS-V. Finally, candidate c) shows a case where the floating H of the first-person
singular has deleted the underlying lexical tone and replaced it. This candidate loses due
to a violation of MAX-T. These candidates and their evaluation in Table 3.12 illustrate
that only long vowels permit the formation of contour tones in SJZ. One important
question that this brings out is why a falling contour results from the addition of a
floating H tone, and not a rising contour tone. Thus far the constraints I present penalize
contours or two tone gestures with the same target without specific reference to whether
these contours are falling or rising. I will argue that the falling contours result from the
coupling graph and that this coupling graph emerges due to markedness constraints on
tone gesture coupling.
3.3 Falling Contour Tones as Result of the Floating H in SJZ
I propose that the falling contour results from the way that the floating tone is
coupled to the underlying vowel gesture of the syllable where its onset is anchored. This
200
proposal relies upon markedness constraints that penalize coupling relationships between
tone gestures and C, V, and other tone gestures. The ranking between these three
constraints in SJZ causes the floating tone to couple to a vowel gesture directly rather
than become entangled with already established coupling relationships between the V, C,
and lexical tone gestures. To explain this proposal I will first review the coupling graphs
that Gao (2008) found for Mandarin tones. Then, I will explain the coupling graph that is
formed when the floating H attaches in SJZ and the coupling graphs I assume for the
underlying lexical tones in SJZ. Finally, I propose the constraints that penalize coupling
relationships between tones and other gestures and show how their ranking obtains a
different coupling pattern for the floating tone as compared to lexical tones.
For Mandarin Chinese, Gao (2008) found two basic coupling graphs were needed
to account for the four lexical tones of Mandarin Chinese. These four tones are level L,
level H, rising LH and falling HL. For level H, level L, and rising LH tones, Gao (2008:
95) proposes the coupling graph shown in Figure 3.2:
Figure 3.2 Coupling between C, V, and T in Mandarin Tones 1, 2, and 3
C-----------T
V
Here, the solid lines indicate in-phase coupling while the dashed line indicates anti-phase
coupling. For the rising tone, then, “T” in the figure above would be filled with both a L
and H tone. These tones would be in-phase with each other and with the vowel, but anti-
201
phase to the consonant gesture. Gao argues that in Mandarin rising tones, the L tone
gesture is intrinsically shorter, which gives rise to the LH rise. This scenario is illustrated
in Figure 3.3:
Figure 3.3 Mandarin rising tone
C---------L
H
V
Figure 3.3 shows two tone gestures, L and H, both in-phase with the vowel and with each
other. The onsets of both H and L tone gestures are synchronous. The shorter duration of
the L tone gesture is what causes the pitch to rise according to Gao’s proposal.
For the falling contour tone in Mandarin, Gao proposes a coupling graph like the
one shown in Figure 3.4. This graph is different from the one shown for the rising
contour tone because it proposes that the H and L tone gestures are anti-phase coupled to
one another.
Figure 3.4 Mandarin Tone 4 (Gao 2008: 96)
C-----------H----------L
V
202
Figure 3.4 shows that the tone gestures are coupled anti-phase to one another and to the
consonant but in-phase to the vowel. The anti-phase coupling is what makes the tone
gestures sequential –the H tone is realized first, then the L tone gesture.
I assume that lexically specified falling and rising tones in SJZ have similar
coupling graphs to the Mandarin examples I have presented here, with the exception that
I assume both contours in SJZ result from tone gestures that are anti-phase coupled to one
another. The falling and rising lexical tones for SJZ are illustrated in Figure 3.5 and
Figure 3.6 with the respective coupling graphs I assume.
60
Articulatory study of the onset
and alignment patterns between C, V, and T gestures in Zapotec are needed to verify
these coupling graphs.
Figure 3.5 Proposed coupling graph for SJZ lexical rising tone
C-----------M----------H
V
Figure 3.6 Proposed coupling graph for SJZ lexical falling tone
C-----------H----------M
V
60
Recall that Mandarin does not have coupling graphs like these for its rising tone based
on the articulatory data collected by Gao (2008).
203
As shown, I assume that the HM and MH patterns both emerge when the tone gestures
are anti-phase coupled to one another but still in-phase coupled to the vowel.
61
As
illustrated, for the rising tone, only the M tone gesture is anti-phase coupled to the
consonant gesture, while for the falling tone, the H tone gesture is anti-phase coupled to
the consonant. I propose that these lexical contour tones differ from the falling contour
that is formed when the floating H tone appears in a syllable. The structure I propose for
the falling tone resulting from the appearance of the floating H tone is illustrated in
Figure 3.7:
Figure 3.7 SJZ falling contour as result of floating H
C------------M
V
H
In Figure 3.7 the floating H tone has coupled directly to the vowel and has no relationship
with the consonant gesture or the underlying M tone gesture. The structure of the
coupling graph will cause the onset of the floating H tone to begin prior to the onset of
the lexical M tone, which will cause a HM pattern in the output. That is, the V gesture
and H tone gesture will begin simultaneously, with the M tone gesture following. I
61
Articulatory study of the onset and alignment patterns between C, V, and T gestures in
Zapotec should be part of future work in order to verify the proposals made here.
204
propose this structure because it allows the floating H to avoid changing underlying
coupling relationships between the lexical tone and C and V gestures.
I propose that this structure arises due to the interaction of the constraints *C-T,
*T-T, and *V-T with PRES-COUP-IO. These three constraints are defined below in (14)-
(16):
(14) *C-T: Assign a violation mark to each coupling relationship between a
consonant and a tone gesture in the output.
(15) *V-T: Assign a violation mark to each coupling relationship between a
vowel and tone gesture in the output.
(16) *T-T: Assign a violation mark to each coupling relationship between two
tone gestures in the output.
In SJZ, I propose that these three constraints are dominated by PRES-COUP-IO. This
ranking relationship will prevent underlying lexical tone coupling graphs from moving or
changing relationships between tone gestures and other gestures, as illustrated in Table
3.13:
205
Table 3.13 PRES-COUP-IO >> *C-T, *T-T >> *V-T
C-----------M
V
PRES-COUP-
IO
*C-T *T-T *V-T
☞ a. C-----------M
V
* *
b. C
V
M
*! *
The ranking between *C-T, *V-T and *T-T is explained in following tableaux. In Table
3.13, a tone gesture is shown coupled in-phase to the vowel gesture of its same syllable
and anti-phase coupled to the consonant gesture. Candidate a) maintains this same
structure in the output and incurs a violation of *C-T and *V-T. Candidate b) has
eliminated the relationship between the M tone gesture and the C gesture to avoid
violation of *C-T, but in the process has violated PRES-COUP-IO. Because PRES-COUP-IO
is ranked higher than the markedness constraints, the lexically coupled tone is not
affected. The same ranking applies to hold lexical contour tones in place with their
respective C and V gestures, as illustrated in Table 3.14:
206
Table 3.14 Lexical contour tone is unchanged due to PRES-COUP-IO
C-----------H--------M
V
PRES-COUP-
IO
*C-T *T-T *V-T
☞ a. C-----------H--------M
V
* * **
b. C-------H---------M
V
*! * * *
c. C-----------H
V
M
*! * **
This tableau illustrates that a lexical rising tone will emerge with the same coupling graph
in the output according to the ranking I propose. Candidate a) is faithful to the input, and
despite a violation of *C-T, and two each of *T-T and *V-T, emerges as the winner.
Candidate b) has deleted a coupling relationship between the input and output and
consequently loses due to a violation of PRES-COUP-IO. Candidate c) couples the second
tone gesture directly to the vowel, but eliminates its anti-phase coupling relationship to
the first tone gesture. This violates PRES-COUP-IO and the candidate loses.
207
MAX-T also dominates the markedness constraints *C-T, *T-T and *V-T.
Deletion of a tone gesture in order to satisfy these three constraints is not an optimal
solution, so MAX-T >> *C-T, *T-T >> *V-T, as illustrated in Table 3.15:
Table 3.15 MAX-T >> *C-T, *T-T >> *V-T
C-----------M
V
MAX-T *C-T *T-T *V-T
☞ a. C-----------M
V
* *
b. C
V
*!
In Table 3.15 candidate b) has deleted the tone gesture, and therefore incurs no violations
of the other three constraints. Candidate a) is still the winner, however, with no violations
of MAX-T.
Now consider the ranking amongst the constraints *C-T, *T-T, and *V-T. I
propose that both *C-T and *T-T dominate *V-T in SJZ and for this reason the floating
tone is drawn to couple directly with the vowel gesture. I also illustrate that MAX-
UNCOUPLED-T dominates these three constraints. This ranking is illustrated in Table 3.16
208
Table 3.16 *C-T, *T-T >> *V-T
62
C-----------M
H
V
MAX-
UNCOUPLED-T
PRES-
COUP-IO
*C-T *T-T *V-T
☞ a. C-----------M
V
H
* **
b. C-------H------M
V
*! * ** **
c. C-----------M-------H
V
* *! *
d. C-----------M-------H
V
* *! **
e. C-----------M
V
*! * *
Table 3.16 illustrates a syllable with a lexical M tone coupled in-phase to the vowel and
anti-phase to the consonant. The floating H tone of the first-person singular is also
62
The ranking *C-T >> *V-T predicts that when a syllable is toneless in the input and a
floating tone attaches to it, it will couple only to the vowel. This is discussed in Figure
3.9.
209
present in the input. Candidate a) shows the winning candidate, where the floating tone
couples directly to the vowel gesture. In this candidate there is a violation of *C-T and
two for *V-T. Candidates b) through d) show other options for the floating tone. In each
case, these other possibilities are ruled out by *C-T ,*T-T or PRES-COUP-IO. For
example, in candidate b), the floating tone has been inserted into the coupling graph so
that it eliminates the anti-phase coupling relationship between the lexical M tone and the
consonant gesture. This incurs a violation of PRES-COUP-IO. Candidate b) also has
violations for *C-T, *T-T, and *V-T. Candidates c) and d) show other options for
attaching the floating H that do not violate PRES-COUP-IO. In candidate c) the floating
tone has coupled anti-phase to the underlying lexical tone but has no relationship with the
vowel gesture. This candidate has a violation of *C-T (for the lexical tone), two
violations for *T-T (one for each tone gesture), and one violation of *V-T. Candidate d)
incurs the same violations as candidate c), with the exception that it incurs an additional
violation of *V-T. Therefore, in SJZ the best option is to couple the floating tone directly
to the vowel gesture. This avoids a violation of PRES-COUP-IO and incurs a violation of
lower-ranked *V-T. As shown by candidate e), deletion of the floating tone gesture is not
optimal, so MAX-UNCOUPLED-T also outranks the tone coupling markedness constraints.
While all candidates which have tone gestures will violate the constraints *C-T,
*T-T and *V-T, the ranking *C-T, *T-T >> *V-T can account for the different coupling
graph that emerges when the floating H tone of the first-person singular appears in a
syllable. In most circumstances, violations of these constraints will not be relevant to
select a winner. This is because PRES-COUP-IO dominates the constraints and will work
210
to hold all relationships of lexical tones to their respective C and V gestures. It is only
when the floating tone is introduced that *C-T, *T-T and *V-T become relevant. In
upcoming tableaux I do not show these constraints but assume that the floating tone will
always couple in-phase to the vowel gesture of a syllable, as I have shown in this section.
Future work should examine how these three markedness constraints may be ranked in
other languages and how this could predict different outputs where floating tones are part
of the representation.
63
An additional prediction of this approach is that syllables with no underlying tone
gesture which acquire a floating H tone gesture will have a different structure from an
underlying, lexically coupled tone gesture. These two graphs are given in Figure 3.8 and
Figure 3.9:
Figure 3.8 Lexical H tone gesture in SJZ
C-----------H
V
Figure 3.9 Floating H tone gesture coupled in syllable with no lexical tone
C
V
H
63
One case where the ranking may differ is in Coatec Zapotec, where Beam de Azcona
(2004) reports that addition of a floating tone results in a rising MH contour.
211
Figure 3.9 is predicted to emerge due to the ranking *C-T, *T-T >> *V-T. This makes a
prediction that merits experimental verification in the future. If lexical H tones have a
different coupling graph, as shown in Figure 3.8, then the onsets of the gestures should
occur in the order consonant, then vowel, then H tone, due to the C-center effect. For a
syllable where there is no underlying tone but the floating tone has coupled, as shown in
Figure 3.9, then the consonant, vowel, and H tone gesture should begin simultaneously.
As a side point, coming back to an issue that was raised in chapter 1, more
fieldwork could demonstrate that H tones on long vowels are systematically absent. The
work I presented in chapter 1 suggested that long vowels do occur, and that there are long
H vowels in the language. If it were found that such vowels are not consistently present
in the idiolects of other speakers, then it is possible that what looks like a falling tone on
long vowels is actually just a H tone that deactivates itself. That would of course demand
revisions to the analysis I have presented here, and would instead suggest that floating H
tones couple the same way that lexical tones do. As discussed in section 1.2.8, more
fieldwork with more speakers is the best way to resolve this question for future work.
I expand upon questions for future research pertaining to the constraints *C-T,
*T-T, and *V-T in chapter 4, as well as possible interactions between these constraints
and others I present in this chapter. This section has illustrated how the ranking of *C-T
and *T-T with respect to *V-T can account for the formation of a falling contour tone
when the floating H appears in a syllable with an underlying lexical tone.
212
3.4 Floating tone appears on toneless aspect marker in SJZ
The data presented in chapter 2 showed that the floating H tone of the first-person
singular will appear on toneless aspect markers if they are available. This means that
when the aspect marker has a specified tone, the floating H is realized only on the
stressed syllable, but when the aspect marker is toneless, that the floating tone is realized
there as well as on the stressed syllable. This section will show how a constraint ranking
can account for this behavior.
I propose that MAX-T dominates COUPLE-INITIAL-V, because in cases where the
aspect marker has a tone underlyingly, the floating tone does not appear on the initial
syllable. This is illustrated in Table 3.17:
Table 3.17 MAX-T >> COUPLE-INITIAL-V
CV- CV"
H
2
M
MAX-T COUPLE-INITIAL-
VOWEL
☞ a. CV- CV "
H
2
H
1
M
**
b. CV- CV "
H
1
M
*! *
Candidate b) in Table 3.17 shows that if the floating tone deletes the underlying tone of
the initial syllable, the candidate is eliminated. Candidate a) illustrates that it is better to
H
1
213
violate COUPLE-INITIAL-V and avoid deleting the underlying tone gesture of the initial
syllable.
Now that I have established that MAX-T is ranked above COUPLE-INITIAL-V,
consider the tableau in Table 3.18. This tableau illustrates how the constraint calling the
floating tone to an initial syllable interacts with the other constraints I have proposed thus
far. In this tableau, I illustrate via a box labeled “V” the vowel gesture’s activation for
some candidates where a tone spreads rightward from the aspect marker onto the root.
This helps make clear whether a given vowel gesture has two tone gestures active in it or
not.
214
Table 3.18 Floating H appears on toneless aspect marker
ru di ba%" gá$
M H
MAX-T H-TONE-TO-
STRESS-V
*MULTI-
TONE
COUPLE-
INITIAL-V
☞ a. rú dí ba!" gá$
H
1
M H
* * **
b. ru di bá" gá$
H
1
H
*! * **
c. ru di ba!" gá$
H
1
M H
* * ***!
d. rú dí ba%" gá$
H
1
M H
**! **
For this particular verb, the aspect marker /ru/ is toneless, as is the causative marker /di/.
The underlying tones of the root are M and H. There is also the floating H of the first-
person singular. In this case the floating tone is drawn to the initial syllable of the
prosodic word, due to the effects of the constraint COUPLE-INITIAL-V. Since there is no
tone gesture present to block satisfaction of COUPLE-INITIAL-V, the floating tone couples
to the initial vowel of the prosodic word and is active during the initial vowel. This tone
remains active up through the stressed syllable due to the effects of the constraint H-
TONE-TO-STRESS-V. This type of tone duration and activation is illustrated by the
H
1
V
V
215
rightward dashed arrow in candidate a). The box showing the gestural activation of the
vowel gesture in the stressed syllable shows that the H tone, the M tone, and this V
gesture overlap – which violates * MULTI-TONE. The behavior of tones that stay active
for longer than one syllable is discussed in following sections explicitly. For the example
shown in Table 3.18 I do not show constraints that penalize tones that maintain their
activation. For this example my primary focus is on the behavior of the floating tone and
its attraction to both the initial and stressed syllable.
Candidate a) emerges as the winner, with only one violation of H-TONE-TO-
STRESS-V. Since both the floating H tone of the first-person singular, H
1
, and the
underlying M tone of the stressed syllable are active during the stressed syllable’s vowel,
the only tone that can violate H-TONE-TO-STRESS-V is the final H tone. Candidate a) also
violates *MULTI-TONE once for having formed a contour, and COUPLE-INITIAL-V twice
for the M and H tones which are not coupled to the initial syllable /ru/. Candidate a) is the
optimal candidate according to this ranking because the floating tone is active both
during the initial syllable and during the stressed syllable.
Candidate b) shows that the floating tone of the first-person singular has replaced
the underlying lexical tone. This incurs an unmotivated violation of MAX-T and the
candidate loses. Candidate c) has the floating tone of the first-person singular only on the
stressed syllable, where a contour has formed. Placing the floating H on the stressed
syllable satisfies H-TONE-TO-STRESS-V and avoids a violation of MAX-T (since the
underlying lexical tone is preserved), but it incurs an additional violation of COUPLE-
INITIAL-V. Candidate d) illustrates the need for the H-TONE-TO-STRESS-V to require that
216
the H tone gesture and the vowel of the stressed syllable deactivate together. Candidate d)
couples the floating tone to the initial V of the prosodic word, and the tone remains active
through the following toneless vowel. However, in candidate d) the tone gesture does not
deactivate with the vowel of the stressed syllable. This violates the constraint and
candidate d) loses. The box for the V gesture helps to illustrate that the H tone gesture
and the V gesture do not overlap or deactivate together in candidate (d).
The same ranking I discussed for Table 3.18 also explains why a form with a mid-
toned aspect marker and a long vowel on the verb root shows the floating H tone only on
the verb root and not also on the initial syllable. To illustrate this I show that the input
/u%-ta%"ní=á$/ plus the floating H tone becomes [u%-ta!"ní=á$] in the output. This is
illustrated in Table 3.19:
217
Table 3.19 Mid-toned aspect marker and long vowel on verb root
u%- ta%"ní=á$
M M H
2
H
3
*MULTI-
TONE-
SHORT -
V
MAX-
T
H-
TONE-
TO-
STRESS-
V
*MULTI-
TONE
COUPLE-
INITIAL-
V
☞ a. u%- ta!" ní=á$
M H
1
M H
2
H
3
** * ****
b. u!- ta!" ní=á$
H
1
M H
1
M H
2
H
3
*! *** ** ****
c. ú ta%" ní=á$
H
1
M H
2
H
3
*! *** ***
Table 3.19 shows how my analysis prevents a tone gesture from coupling in two
locations in the output. In candidate b), the floating tone of the first-person singular
couples in initial position and also appears in the stressed syllable.
64
By placing the
floating H tone in the same syllable as the aspect marker and its underlying tone,
candidate b) incurs a violation of *MULTI-TONE-SHORT-V and an additional violation of
H-TONE-TO-STRESS-V. Candidate b) is then eliminated. Candidate c) shows another
possibility, where the floating H tone has coupled to the vowel in the initial syllable. This
causes deletion of the underlying lexical tone of the initial syllable, and candidate c) is
64
This candidate also violates a constraint on duplicating a tone. The continuation of H
1
without duplicating it could not produce falling contours on two consecutive syllables.
H
1
218
ruled out. The optimal candidate is shown in a), where the floating H tone appears on the
stressed syllable of the verb root to form a falling contour.
These examples and the ranking illustrated in Table 3.19 account for the pattern
where both the stressed syllable and the initial syllable are important locations for the
floating tone of the first-person singular. As I have shown, the floating H will appear on
the aspect marker (in initial position) only if this syllable is toneless. Otherwise, the
floating H will appear just on the stressed syllable of the verb root.
3.4.1 Main Constraint Rankings Summary
The constraint rankings I have established thus far are summarized in (17) and the
effects they produce on the tone patterns are briefly described.
(17) Some main constraint interactions in SJZ
a. PRES-COUP-IO >> H-TONE-TO-STRESS-V, COUPLE-INITIAL-V
Only the floating tone is drawn to the initial syllable or to the stressed syllable.
Other tone gestures do not alter their input coupling relationships.
b. PRES-COUP-IO >> MAX-T >> *C-T, *T-T >> *V-T
The floating tone is drawn only to couple in-phase with a vowel gesture. Other
tone coupling relationships are not affected. Tones are not deleted to satisfy
markedness constraints on tone coupling relationships.
219
c. *MULTI-TONE-SHORT-V >> MAX-T
The floating H tone can replace an underlying lexical tone tone so as to avoid
forming a contour on a short vowel.
d. MAX-T >> H-TONE-TO-STRESS-V >>*MULTI-TONE, COUPLE-INITIAL-V, *HD/M,
*HD/L, COUPLE-H-PROM-V
Formation of a contour tone is preferable to deletion of the lexical tone provided
the vowel on which the tones appear is long. The floating tone never appears in
initial position when the intial vowel gesture has an underlying tone gesture.
65
A summary of the main interactions is also given in Figure 3.10, where I show constraint
lattices for floating tones in SJZ.
65
The ranking I present here does not rule out contour tones on the initial syllable,
although there is not evidence to suggest that aspect markers ever have long vowels in
SJZ.
220
Figure 3.10 Constraint lattice for SJZ floating tone
PRES-COUP-IO *MULTI-TONE-SHORT-V
MAX-T
H-TONE-STRESS-V *MULTI-TONE
COUPLE-INITIAL-V *HD/L
*C-T *HD/M
*T-T
*V-T
3.4.2 Tone Deactivation: Preview
It is important to address what motivates the lengthened activation duration of
tones such as when the tone of the aspect marker spreads rightward onto a following
toneless syllable. To do this I consider constraints that govern the deactivation of tone
gestures. Recall that deactivation refers to the point in time when a gesture turns off.
When a gesture turns off, it means the articulators that control it are no longer actively
engaged in producing their constriction actions. As I have explained thus far, coupling
has to do with the onset of a gesture and not its deactivation, which is a crucial part of my
analysis of the directional asymmetry in tone spreading. Rightward spreading of a
lexically coupled tone does not incur violations of coupling but leftward spreading will. I
devote a separate section to these constraints in this chapter (see section 3.5) but
introduce them here. I propose two additional constraints shown in (18) – (19). The first
constraint makes reference to specific deactivators and the second is a general constraint
221
penalizing a tone gesture that spans more than one syllable. These are defined in (18) and
(19):
(18) *SELF DEACTIVATE BY CLOCK: Assign a violation mark to any tone that
deactivates other than at a deactivator, where a deactivator is
either a boundary (end of utterance) or a following anti-phase
coupled tone.
(19) *SPAN-SYLL – assign a violation mark for each tone gesture that
deactivates later than at 360 degrees of its own clock.
Neither of these constraints is specific to the floating tone because tone spreading and
coupling of the floating tone are distinct in SJZ, although sometimes related. The first of
these constraints, *SELF-DEACTIVATE-BY-CLOCK, will penalize a tone gesture if it “turns
off” or becomes deactivated before reaching a following tone gesture or the end of an
utterance (if the tone deactivates on its own at 360 degrees of its clock). Under this
approach I assume that tones in a sequence are anti-phase coupled to one another and that
this is imposed by a constraint in GEN. I assume that the anti-phase coupling relationships
between sequential tones causes each to be deactivated by the following tone. Thus
*SELF-DEACT penalizes cases where this does not occur. This constraint necessitates a
disjunction (a deactivator can be either a following tone or the end of the utterance) and
future work must examine whether the cohort of ‘deactivators’ can vary by language and
whether it might be expanded. For one possible way of handling certain disjunctive
phenomena using Harmonic Grammar, see Jesney and Tessier (2011), and Potts et al.
222
(2010). Under my approach, *SELF-DEACT motivates rightward tone spreading onto
syllables in the output that have no underlying lexical tone. If *SELF-DEACT were active
in the grammar and no other constraints came into play, it would predict unbounded tone
spreading up until the end of an utterance or a following anti-phase coupled tone.
The constraint *SPAN-SYLL will penalize any type of tone spreading, since it
assigns a violation for any tone gesture that is active for longer than one cycle (360
degrees of its clock). For simplicity, this constraint is named *SPAN-SYLL, since tones
which remain active longer than for one cycle of their clock will persist into following
syllables. (Recall from the discussion of H-TONE-STRESS-V that work by Tilsen (2009,
2011) finds evidence for a syllable level oscillator). The constraint *SPAN-SYLL is similar
to the Crisp Edge constraints discussed by Walker (2011) as well as earlier work by Ito
and Mester (1999), Walker (2001), and Kawahara (2008). Crisp Edge constraints
penalize a feature such as [+round] from being shared across a syllable boundary. As
Walker points out, the constraint “prohibits any chain that contains feature specification
occurrences that are dominated by different syllables” (2011, 86). Crisp Edge constraints
could also refer to tone gestures, but a difference between my approach and Crisp Edge
constraints is that the constraints I propose make reference to either the onset or the
deactivation point of a gesture, rather than to features. The constraint *SPAN-SYLL is very
similar to Yip’s constraint NOLONGT, which assigns violations to any tone that is
associated with more than one tone bearing unit (2002: 83). In my approach there are no
constraints that penalize toneless syllables that acquire a tone or toneless syllables that
remain toneless in the output. I assume that toneless syllables lack a tone gesture and that
223
the insertion of a tone gesture either via a new coupling relationship or via lengthened
activation does not incur any violations.
To show how the constraints I propose for tone deactivation assign violations, I
will first present a basic tableau with three general candidates. This will help make the
evaluation of candidates clearer. After the basic tableau, I then present a tableau for
/be%-di%-ba!"gá$/ with the additional constraints motivating tone spreading. I will show that
*SELF-DEACT is needed to account for the correct output form.
First, consider the tableau shown in Table 3.20 where two schematic outputs are
evaluated with regard to the constraints *SELF-DEACT, and *SPAN-SYLL. Here my
purpose is not to show a ranking between these constraints for SJZ, but to show how they
would assign violations to various candidates.
224
Table 3.20 Tone deactivation constraints and evaluation
CV CV CV
M H
*SELF
DEACTIVATE
BY CLOCK
*SPANSYLL
a. CV CV CV
M H
*
b. CV CV CV
M H
*
c. CV CV CV
M H
* *
The constraint *SELF-DEACT is violated by candidates a) and c). In candidate a) the M
tone fails to spread onto the following toneless syllables and deactivates via its own clock
upon reaching its mid-tone target. In candidate c), the M tone has deactivated neither via
a following tone which is anti-phase coupled to it, nor via an utterance boundary, and a
violation of *SELF-DEACT is incurred. In candidate b), however, only a violation of
*SPAN-SYLL is incurred, since the tone gesture of the M tone is deactivated by the onset
of the tone which is anti-phase coupled to it (in this case the H tone). Although not shown
in candidates or tableaux, I propose, as argued earlier, that all tones in the output are anti-
phase coupled to one another. This is not illustrated in tableaux for reasons of space. The
constraint *SELF-DEACT will work to ensure that tones spread in the conventional sense
onto following syllables.
225
An interesting side effect of the representations I assume and the constraints I
have presented here for tone deactivation is that they fit with other constraints that have
been proposed for tones. For example, Yip mentions the constraint NOGAP “multiply
linked tones cannot skip tone bearing units” (2002: 84). NOGAP calls for the same type of
structure that a gestural approach presupposes.
66
For instance, gapping or tone spreading
that skips intervening syllables is not possible from a gestural point of view, since once a
tone is deactivated, it will not be reactivated. Thus a sequence of toneful, toneless, toneful
implies that there are two tone gestures present, and not one that has skipped an
intervening syllable. I will discuss the relationship between my gestural approach to tone
deactivation and previous approaches such as autosegmental phonology in more detail in
chapter 4.
Returning to the SJZ data, I will now present a tableau showing a case where the
floating tone is attracted to the stressed syllable of the verb root. In a separate tableau, I
will then show how the constraints I propose for tone spreading are needed to account for
the winning candidate. First consider what happens when the aspect marker (the initial
syllable) has an underlying tone gesture. In these cases, I do not predict that the floating
tone will appear on the initial syllable and delete the underlying lexical tone. This is
because MAX-T dominates COUPLE-INITIAL-V. Table 3.21 shows an example of this
interaction. A completive aspect marker with a M tone is followed by a toneless syllable
and then by the verb root. When this form is realized in the first-person singular, a
floating H tone is also part of the representation, and the floating H tone is drawn only to
66
Yip’s approach differs in that NOGAP is a violable constraint, whereas the prohibition
on gapping in gestures comes from a restriction on GEN in my approach.
226
the stressed syllable, not to the initial syllable nor to the intervening toneless syllable /di/.
An additional constraint called COUPLE-H-TO-PROM-V is needed in this particular case. I
define the constraint below:
(20) COUPLE-H-TO-PROM-V: Assign a violation to any H tone gesture not
coupled in-phase with a prominent vowel gesture in the output.
The constraint will penalize coupling of the H tone gesture unless it is with the initial
vowel or to the vowel in the stressed syllable. Again, this is a constraint similar in its
effects to the constraints proposed by de Lacy that I discussed earlier. In requiring a H
tone in particular to couple to a prominent vowel, it finds support in the work of Smith
(2002) and de Lacy (2007). This is a markedness constraint in that it is imposing a
structural well-formedness requirement on the output rather than demanding that the
output form match the input form. This constraint is needed because, as shown in Table
3.21, the constraints penalizing a M or L tone on the stressed syllable cannot account for
the spreading behavior of the floating tone in the example illustrated by the tableau. Even
the winning candidate violates *HD/M in this particular example. In this case, the ranking
MAX-T >> H-TONE-TO-STRESS-V >> COUPLE-INITIAL-V, *MULTI-TONE, COUPLE-H-TO-
PROM-V can account for the pattern. This is illustrated in Table 3.21. First I will show
these constraints in a tableau, and then I will present a tableau which incorporates the
*SELF-DEACT, and *SPAN-SYLL constraints.
227
Table 3.21 Floating H tone goes to stressed syllable and aspect marker spreads
rightward
be% - di - ba%" gá$
M X M H
MAX-T H-TONE-
TO-
STRESS-V
COUPLE-
INITIAL-V
*MULTI-
TONE
COUPLE-
H-TO-
PROM-V
☞ a. be% - di%- ba!" gá$
M H
1
M H
* *** * *
b. bé -dí -ba!" gá$
H
1
M H
*! * ** * *
c. be% - dí -ba%" gá$
M H
1
M H
**!
***
**
d. be% -dí - ba!" gá$
M H
1_
M H
* *** * **!
Candidate a) emerges as the winner, with one violation of H-TONE-TO-STRESS-V. In this
candidate a box for the V gesture of the stressed syllable shows that the M tone of the
aspect marker does not overlap with the vowel of the stressed syllable or its tones. Three
violations are assigned for the constraint COUPLE-INITIAL-V and one for *MULTI-TONE.
In this candidate the floating H of the first-person singular has coupled to the vowel in the
stressed syllable of the verb root to form a contour. Also notice that the M tone of the
H
1
V
V
V
228
completive aspect marker stays active through the toneless syllable /di/ in candidate a).
This is shown via the arrow.
Candidate b) demonstrates why MAX-T must outrank the constraint calling a
floating tone to the initial syllable. If the floating tone does delete the lexical tone of the
aspect marker in order to satisfy COUPLE-INITIAL-V then the candidate loses, as shown in
b). The box showing the V gesture of the stressed syllable makes it clear that the tone has
remained active and overlaps with the V and T gesture of the stressed syllable. Candidate
c) shows that if the floating tone goes only to the toneless causative marker /di/, then an
extra violation of COUPLE-H-TO-PROM-V is incurred and the candidate loses. Candidate
c)’s solution might seem like an intuitively appealing one, since there is no tone gesture
underlyingly which would cause a violation of MAX-T. The presence of the constraint
COUPLE-H-TO-PROM-V rules this out, however. Candidate d) is slightly more
complicated. Here, the floating tone of the first-person singular has coupled to the vowel
in the toneless syllable /di/, but stayed active through the stressed syllable. The
lengthened activation through the duration of the vowel of the stressed syllable is shown
via the overlapping box and rightward pointing arrow. Candidate d) avoids a violation of
H-TONE-TO-STRESS-V since the constraint H-TONE-TO-STRESS-V only requires a gesture
to deactivate with the vowel of the stressed syllable, not couple to it. Candidate d) ties
with the winning candidate a) on all constraints except for COUPLE-H-TO-PROM-V. This
constraint is violated by candidate d) and thus the distinction is made between winner and
loser.
229
Now, consider the same example taking into consideration the constraints that
govern deactivation of tone gestures. I show the constraint *SELF-DEACT, which will ban
the initial M tone of the aspect marker from turning off before it reaches the specified
tone of the verb root. This will ensure that the causative marker /di/ which separates these
two specified tones is also mid-toned. First, observe the tableau shown in Table 3.22.
230
Table 3.22 Tone of aspect marker spreads rightward in SJZ
be% - di - ba% gá$
M X M H
MAX
-T
H-
TONE-
TO-
STRES
S-V
COUPLE
INITIAL-
V
COUPLE-
H-TO-
PROM-V
*SELF
DEAC
T
*SPAN
-SYLL
☞ a. be% - di% -ba!" gá$
M H
1
M H
* *** * *
b. bé - dí - ba!" gá$
H
1
M H
*! * ** * *
c. be% - dí - ba%" gá$
M H
1
M H
**! ***
**!
d. be% - dí - ba!" gá$
M H
1_
M H
**! *** **!
e. be% - di - ba!" gá$
M H
1
M H
* *** * *!
In Table 3.22 the only new candidate is shown in e). The M tone gesture of /be%/ has
deactivated without the presence of either a following anti-phase coupled tone or an
utterance boundary. This incurs a violation of *SELF-DEACT. Candidate e) has exactly the
same violations as winning candidate a) for the constraints H-TONE-TO-STRESS-V,
COUPLE-H-TO-PROM-V and COUPLE-INITIAL-V. Candidate e) demonstrates that *SELF-
H
1
V
V
V
231
DEACT >> *SPAN-SYLL because the winning candidate a) violates *SPAN-SYLL but has
fewer violations of *SELF-DEACT. More examples of tone spreading and the effects of
these constraints are found in upcoming examples. *SELF-DEACT and *SPAN-SYLL do not
interact with the other constraints shown in the tableau in the sense that they only come
into play for candidates like the one shown in e). Although it may appear that *SELF-
DEACT is ranked below H-TONE-STRESS-V, in fact the two do not interact. Under most
circumstances, the effects of *SELF-DEACT and *SPAN-SYLL are not noticeable since they
do not cause tone deletion, etc. *SELF-DEACT just works to keep tones on until
appropriate deactivators are encountered, either by default (end of utterance) or via
another constraint (such as H-TONE-STRESS-V).
Next, consider the situation when a high-toned potential aspect marker spreads
rightward onto a toneless causative marker. In this case, just as we saw in the previous
example, I propose that the floating H tone of the first-person singular will be attracted
only to the vowel of the stressed syllable of the verb root. This case is illustrated in Table
3.23:
232
Table 3.23 High-toned aspect marker spreads onto toneless syllable
gú - di - ba%" gá$
H X M H
MAX-
T
H-TONE-
TO-
STRESS
COUPLE-
INITIAL-V
*MULTI-
TONE
COUPLE-
H-TO-
PROM-V
*SELF
DEACT
☞ a. gú- dí -ba!" gá$
H H
1
M H
** *** * *
b. gú -dí – ba%" gá$
H
1
M H
*! ** ** *
c. gú - dí -ba!" gá$
H
H
1
M H
** *** * **!
d. gú -di -ba!" gá$
H H
1
M H
** *** * * *!
In this tableau, boxes for the V gesture of the stressed syllable have been added to show
where the H spreading tone gesture overlaps for candidates a) through c). Candidate b) is
eliminated due to a violation of MAX-T. The candidate has deleted the initial H tone of
the potential aspect marker and instead has coupled the floating H of the first-person
singular in this location. The floating H tone then stays active up through the toneless
causative marker /di/. MAX-T prevents this situation from occurring. The competition
between candidates a) and c) is decided by COUPLE-H-TO-PROM-V. The preferred
solution is for the winning candidate to realize the floating tone of the first-person
H
1
V
V
V
233
singular on the stressed syllable only, forming a contour when it combines with the
underlying lexical tone. In the winning candidate I assume that the causative marker /di/
is high due to rightward spreading from the high-toned potential aspect marker. Since /di/
has no underlying tone of its own, this rightward spreading does not incur any violations
of MAX-T. As shown via candidate d), the causative marker /di/ will become H in the
winning candidate due to the effects of *SELF-DEACT. This constraint prevents the
specified H tone of the potential aspect marker from deactivating via its own clock and
failing to spread onto /di/.
Thus far I have examined several tableaux and constraints which account for the
placement of the floating tone of the first-person singular. As I have shown, there is also
evidence of rightward tone spreading, which I have accounted for via the constraints
*SELF-DEACT, and *SPAN-SYLL. Although I have commented briefly on this rightward
spreading in the previous examples, I will now move to a section that examines
constraints on tone deactivation more closely. These constraints are needed to account for
the observed pattern that tones spread rightward onto toneless syllables in the SJZ data I
have collected. Some of these constraints are applicable to the tableaux I have already
presented, but as I have shown, they do not affect the outcomes which have already been
discussed. In the following section on tone spreading, my examples will include verbs
and pronouns other than the first-person singular. I include other persons because tone
spreading and coupling of the floating tone of SJZ are separate phenomena. This does not
imply that the processes of rightward spreading do not occur in the first-person singular,
as should be clear from the sections I have already presented in this chapter. Rather, it is
234
easier to isolate the interaction of constraints pertaining to spreading in forms that do not
also involve coupling of a floating tone.
3.5 Tone Spreading in SJZ
In languages with rightward spreading onto toneless syllables, I propose that
*SELF-DEACT >> *SPAN-SYLL. In languages where there is no rightward spreading onto
toneless syllables, I propose that the opposite ranking applies *SPAN-SYLL >> *SELF-
DEACT. To illustrate how the ranking *SELF-DEACT >> *SPAN-SYLL accounts for a
language like SJZ, I discuss several tableaux in this section.
Recall from chapter 2 that there are several processes of tone spreading in SJZ
which are unrelated to the floating H tone of the first-person singular. One we have
already seen, and this is that tones spread rightward onto toneless syllables. I will present
additional examples of this process to make the phenomena and the relevant constraint
rankings clearer in this section. A second process of spreading occurs even when
following syllables have specified tones. Spreading happens when a verb root ends in a H
tone. The H tone will spread rightward onto attached clitics and make the clitics high-
toned as well – regardless of their underlying, specified tone. First, consider an example
where the mid-toned aspect marker spreads rightward onto a toneless syllable. This is
illustrated in Table 3.24:
235
Table 3.24 Tone of aspect marker spreads rightward due to *SELF-DEACT
be%- di- ba#" gá$= lu%$
M X L H M
* SELF
DEACT
*MULTI-
TONE
*SPAN-
SYLL
☞ a. be% -di% -ba#" gá$ =lu%$
M L H M
*
b. be% -di -ba#" gá$ lu%$
M L H M
*!
c. be% - di% - ba!" gá$ lu%$
M L H M
*! *
For the input in Table 3.24, a mid-toned aspect marker stays active until it encounters the
specified L tone of the verb root. This is illustrated in candidate a). Candidate a) has only
one violation of *SPAN-SYLL, since the initial M tone stays active through the syllables
/be%/ and /di/. No violations of *SELF-DEACT are incurred because the tone of the aspect
marker deactivates upon reaching the onset of the following, anti-phase coupled tone.
Candidate b) incurs a violation of *SELF-DEACT. It loses to candidate a) because
the M tone of the aspect marker deactivates via its own clock. Candidate c) shows a case
where the tone of the aspect marker stays active so long that it forms a contour tone with
the underlying L tone of the verb root. This incurs a violation of *MULTI-TONE and the
candidate loses. Here I assume that the L tone of the verb root (the following tone) is still
V
V
236
what causes deactivation for the tone of the aspect marker, but that this deactivation
occurs later than in the other cases such as the one shown in the winning candidate a).
The constraint *MULTI-TONE is needed to rule out such cases where more than one tone
gesture is active in the same syllable. This tableau shows how *SELF -DEACT can drive
rightward spreading, and shows that this constraint must outrank *SPAN-SYLL.
Another case where the ranking of *SELF-DEACT and *SPAN-SYLL is needed is
when a clitic marker is toneless and the verb-final tone spreads onto this clitic. In this
case there is no following anti-phase coupled tone to cause deactivation, so instead the
end of utterance deactivates the tone gesture. Observe the tableau shown in Table 3.25:
Table 3.25 Verb-final H tone spreads onto toneless clitic
wá$ ní = a)
H H X
DEP-T *SELF
DEACT
*SPANSYLL
a. wá$ ní =a)
H H M
*!
☞ b. wá$ ní =á
H H
*
c. wá$ ní = a)
H H X
*!
This tableau adds the constraint DEP-T, which will prevent the insertion of a new tone
gesture. I define this constraint formally in (21):
237
(21) DEP-T: A tone gesture in the output has an input correspondent.
The DEP-T constraint I propose is identical to the DEP-T constraint used by Yip (2002)
with the exception that I refer to a tone gesture rather than to a tone that is an
autosegment. Table 3.25 shows that in the input the third-person singular non-formal has
no underlying tone gesture. If, as candidate c) shows, the tone gesture of the final syllable
were to turn off via its own clock, this would incur a violation of *SELF-DEACT.
Candidate a) presents another solution. In this case, a M tone gesture is inserted onto the
clitic. Candidate a) violates DEP-T and consequently loses. The winning candidate b)
shows that the tone gesture of the penultimate syllable remains active in the output until it
reaches the end of the utterance.
To illustrate that this spreading behavior is not particular to a H tone, I will show
the same ranking for a different input. This time, I will illustrate that the same rightward
spreading occurs in a verb that ends in a L tone. Consider the verb ‘to paint’. In the
potential aspect, this is /ka!"bi#=a) #/ for the third-person singular non-formal. Here, I
propose that the third-person singular clitic is underlyingly toneless as discussed in
chapter 2, but that it has become L due to the final tone gesture of the verb root. In Table
3.26 below, I show how the ranking DEP-T ,*SELF-DEACT >> *SPAN-SYLL can account
for this pattern.
238
Table 3.26 Verb-final L tone spreads onto toneless clitic
67
ka!" bi# = a)
HM L X
DEP-T *SELF
DEACT
*SPANSYLL
a. ka!" bi# = a) %
HM L M
*!
☞ b. ka!" bi# = a) #
HM L
*
c. ka!" bi# = a)
HM L X
*!
In Table 3.26, the verb-final L tone spreads rightward onto the toneless clitic. I propose
that this process is driven by failure to deactivate, as I show in Table 3.26. The constraint
*SELF-DEACT will cause rightward spreading from a toneful syllable onto a toneless
syllable. This is shown in winning candidate b). Candidate a) shows that inserting a M
tone onto the toneless syllable is ruled out by the constraint DEP-T, while candidate c)
illustrates that a violation of *SELF-DEACT is fatal.
Ranking *SELF-DEACT above *SPAN-SYLL predicts that when the first syllable of
the verb root is toneless, the aspect marker which precedes it should spread rightward.
My analysis predicts that if the second syllable of a verb is toneless, it too should undergo
spreading. While I have not found examples where the second syllable of the root is
67
In this case I assume that both tone gestures in the initial syllable are in-phase coupled
to the vowel but anti-phase coupled to one another and to the consonant, as discussed
earlier in this chapter.
239
toneless and undergoes spreading, there are examples where the first syllable of the root
is toneless. I will begin with an example where a mid-toned aspect marker spreads
rightward onto the first syllable of a verb root. Then I will examine cases of rightward H
spreading.
The tableau in Table 3.27 shows a case where the ranking DEP-T, *SELF-DEACT
>>*SPAN-SYLL obtains.
Table 3.27 M tone spreads onto a toneless syllable
be% de t(u%=lu%$
M X M M
DEP-T *SELF-
DEACT
*SPAN-
SYLL
a. be%- de t(u%=lu%$
M X M M
*!
☞ b. be%- de% t(u%=lu%$
M M M
*
c. be%- de% t(u%=lu%$
M M M M
*!
Table 3.27 illustrates the same patterns I presented earlier, albeit in a different location. I
have already shown a verb-final H tone that stays active through a toneless syllable, as
well as the floating H tone of the first-person singular that may stay active for longer than
one syllable. Table 3.27 shows that in a verb with a second-person singular clitic, the
mid-toned aspect marker /be%/ stays active through the first syllable of the verb root.
240
Candidate a) is ruled out due to a violation of *SELF-DEACT. Candidate c) is ruled out by
a violation of DEP-T, and the winner is candidate b), where the tone gesture of /be%/
remains active. These examples have illustrated the ranking *SELF-DEACT, DEP-T >>
SPAN-SYLL.
Now that I have shown examples where only these three constraints are needed, I
move on to more complex examples. As presented in the data chapter, there is also a
separate process of rightward tone spread when the verb root ends in a H tone. This is
separate from the spreading onto toneless syllables because it occurs even when clitics
have an underlying tone gesture. I propose that this is due to the effects of a separate
constraint which I define below in (22):
(22) DEACTIVATE-H-AT-UTTERANCE-BOUNDARY: Assign a violation mark for
each H tone gesture which deactivates anywhere other than at an
utterance boundary.
This constraint penalizes any H tone gesture that deactivates anywhere except at an
utterance boundary and, as such, will drive the H tone of a verb to stay active across a
morpheme boundary and onto the attached clitics. This constraint will also require that
the H tone of the aspect marker stay active. Since we have seen that the H tone of the
aspect marker does not spread onto the verb root unless the verb root begins with a
toneless syllable, a ranking is needed to account for the difference between the aspect
marker’s H tone and the final H tone of the verb root. I will examine the ranking that
makes this possible in following paragraphs. First, consider a case where a verb-final H
241
tone spreads onto a clitic which already has a tone. I assume these effects are driven by
DEACT-H-AT-UB. Table 3.28 shows that a mid-toned clitic in the input becomes H in the
output. This is driven by the effects of DEACT-H-AT-UB.
Table 3.28 Verb-final H tone spreads onto clitic
wá$ ní=lu%$
H H M
DEACT-
AT-UB
*MULTI-
TONE-
SHORT V
MAX-T *MULTI-
TONE
*SPAN-
SYLL
a. wá$ ní =lu%$
H H M
**!
☞ b. wá$ ní=lú$
H H
* * *
c. wá$ ní =lu!$
H H M
**(!) *(!) * * *
Table 3.28 shows a second-person singular informal, and therefore that there is no
floating H tone in the input. Candidate a) has two violations of the constraint DEACT-H-
AT-UB since there are two H tones that deactivate before the utterance boundary.
Candidate b) has only one violation of DEACT-H-AT-UB and incurs a violation of MAX-
T, since the tone gesture of the clitic has been deleted. Candidate b) also incurs a
violation of *SPAN-SYLL. For this case I assume that the rightward spread or failure to
deactivate is not driven by *SELF-DEACT but rather by the effects of DEACT-H-AT-UB.
V
242
This is because we have seen examples previously which show that when a tone follows
another tone, the first deactivates under normal circumstances and no spreading results.
The appearance of verb-final H tones on all clitics is a separate process from the
deactivation effects we observe in the presence of toneless syllables. Finally, candidate c)
is ruled out due to the contour tone which is formed on the clitic. This candidate incurs
two violations of DEACT-H-AT-UB and a violation of MAX-T, and it is penalized for
forming a contour on a short vowel and for having a tone active for longer than one
syllable.
While Table 3.28 showed the effects of adding the constraint DEACT-H-AT-UB, it
is also helpful to observe the constraint interaction when we incorporate the constraints
that govern deactivation. This is shown in Table 3.29 below:
243
Table 3.29 Verb-final tone spread and deactivation
wá$ ní=lu%$
H H M
DEACT
H- AT-
UB
*MULTI-T-
SHORT-V
MAX-
T
*SELF-
DEACT
*MULTI-
TONE
*SPAN-
SYLL
a. wá$ ní =lu%$
H H M
**!
☞ b. wá$ ní=lú$
H H
* * *
c. wá$ ní =lu!$
H H M
**! * *
Here, the winner is still candidate b). This candidate violates MAX-T and *SPAN-SYLL.
Here I argue that the final H tone of the verb stem stays active up until the speaker
reaches the end of the utterance, in which case the tone is deactivated. Candidate a) loses
due to two violations of DEACT-H-AT-UB. Candidate c) also incurs two violations of this
constraint and one for *MULTI-TONE-SHORT-V. Here, my analysis differs from the one
presented in Zoll (2003). Zoll’s approach, called Optimal Tone Mapping (OTM),
analyzes tone spread as the result of high-ranking markedness constraints that penalize
sequences of identical tones in combination with morphological directionality
imperatives (2003: 250). For instance, where the tone in question is part of an affix, if
alignment is more important than qualitative constraints, morphologically induced
V
244
directionality will prevail (2003: 255-256). That is, when qualitative constraints are
ranked low, prefixing tones will show evidence of left-to-right association, while
suffixing tones would show a right-to-left pattern. My constraints instead predict only
left-to-right tone spreading, as shown in Table 3.29.
Now I will present a case where the H tone of the potential aspect marker does not
continue into the verb root. This will violate DEACT-H-AT-UB, but I propose that this is
motivated by the constraint MAX-ROOT-T, which bans deletion of tone gestures in the
root. This constraint is ranked higher than DEACT-H-AT-UB. The ranking relationship
between these constraints accounts for the patterns of SJZ. First I will show a verb root
with a specified H tone on its first syllable, and then I show an example with a specified
L tone.
First, consider the definition of MAX-ROOT-T, shown in (23) below:
(23) MAX-ROOT-T: Tone gestures in the root in the input have output
correspondents.
This constraint is similar to MAX-ROOT constraints posed by others (McCarthy and
Prince (1993, 1995), Urbanczyk 1995, Futgai 1997, among others) with the exception
that it refers specifically to tone gestures. As I show in the following tableaux, this
constraint will work to prevent deletion of tone gestures of the verb root, which excludes
aspect markers and clitics. To show how this constraint interacts with DEACT-H-AT-UB, I
include Table 3.30:
245
Table 3.30 Tone gesture of aspect marker does not spread onto root
68
gú - di# a# =lu%$
H LL M
MAX-
ROOT
-T
DEACT-
H AT -
UB
*MULT
TONE-
SHORT-
V
MAX-
T
*SELF
DEACT
*SPAN
SYLL
a. gú - d i! a# =lu%$
H L L M
* *! *
☞ b. gú - di# a#=lu%$
H L L M
*
c. gú -dí a# =lu%$
H L M
*! * * *
Here I have shown DEACT-H-AT-UB in addition to the other constraints that govern
deactivation of tone gestures. Candidate b) emerges as the winner, and the high-toned
aspect marker does not stay active through the first syllable of the verb root. This solution
is better than that shown in candidate c) where MAX-ROOT-T is violated. Candidate a)
has kept the H tone gesture of the potential aspect marker active long enough to form a
contour. The V box shows that both tone gestures are active during the V gesture in this
candidate. This incurs a violation of DEACT-H-AT-UB, and also violates *MULTI-TONE-
68
In a few cases in the appendix where a potential aspect marker consists of a consonant
plus a H tone, this H tone is realized on the vowel-initial verb root. This causes deletion
of the underlying tone at times. In these special cases, I assume a constraint requiring
realization of the aspect marker morpheme dominates MAX-ROOT-T.
V
246
SHORT-V and candidate a) loses. Table 3.30 illustrates the ranking MAX-ROOT-T >>
DEACT-H-AT-UB, *MULTI-TONE-SHORT-V >> MAX-T, *SELF-DEACT >> *SPAN-SYLL.
3.5.1 Glottal Stops Block Rightward Tone Spreading
This section will present one more source of deactivation for tones: glottal stops.
As I presented in the data chapter, in SJZ glottal stops block rightward spreading from a
verb-final H tone onto the clitics. I account for this by proposing a constraint that
penalizes tone spreading through glottal stops. This constraint is called *ACTIVE-
THROUGH-GLOTTAL-STOP and is defined below in (24):
(24) *ACTIVE THROUGH GLOTTAL STOP – Assign a violation mark to each tone
gesture whose activation overlaps a glottal constriction gesture.
Earlier in this chapter I gave an example showing that verb-final H tones spread onto the
clitics even when these have an underlying tone. I have also shown examples illustrating
that tones spread rightward onto toneless syllables. The example in Table 3.31 shows a
case where rightward spreading is thus far predicted. The monosyllabic verb root has a H
tone in this particular case, and the third-person singular non-formal clitic is toneless. As
I showed in chapter 2, no spreading occurs in this case, nor in any other case where a
glottal stop intervenes. The tableau here has all of the constraints that were used earlier,
with the exception that I have added *ACTIVE-T-GS. This constraint needs to be ranked
higher than DEACT-H-AT-UB because a H tone does not persist through to an utterance
boundary when a glottal stop intervenes. In cases where the tone spreads through a
247
glottal stop, I will illustrate the C gesture (for the glottal constriction gesture) with a box,
just as I did previously to show V gestures with more than one active tone gesture when
this resulted from spreading. For example, if the rightward pointing arrow and the C box
overlap, this indicates that the gestures have overlapped by more than 25% of their
activation duration. For reasons of space, if the tone and glottal constriction gesture do
not overlap, I will not include a C box.
Table 3.31 H tone spread blocked by glottal stop
gó$ = a)
H X
*ACTIVE
T-GS
DEACT-
H-AT-UB
DEP-T *SELF-
DEACT
*SPAN
SYLL
a. gó$ = á
H
*! *
☞ b. gó$ = a)
H
* *
c. gó$= a)
H M
* *!
In candidate a) the tone gesture of the verb root stays active across the morpheme
boundary and onto the toneless following syllable. However, this candidate violates
*ACTIVE-T-GS because the tone gesture remains active through a glottal stop. Candidate
b) is the optimal candidate, where no spreading occurs. Candidate c) shows another
C
248
option, where a M tone is inserted onto the toneless syllable. This violates DEP-T, and
candidate c) loses. This example illustrates how the constraint *ACTIVE-T-GS works to
prevent tone activation through a glottal stop.
Now I will present an additional example of failure to spread a tone through a
glottal stop. This time however, I will give an example of a verb that ends with the
sequence /v.$/ where the attached clitic has an underlying tone. Recall from previous
examples shown here and in the data chapter that normally, when a toneful clitic follows
a verb that ends in a high-toned vowel, this H tone spreads onto the clitic and deletes its
underlying tone. As I show in Table 3.32 this is not the case when the verb root ends in a
glottal stop:
Table 3.32 Clitic does not undergo spreading from verb root ending with glottal
stop
be%-di-ba##"gá$=lu%$
M X L H M
*ACTIVE
T-GS
DEACT-
H
AT-UB
MAX-
T
*SELF
DEACT
*SPAN
SYLL
a. be%-di%-ba##"gá$=lú$
M L H
*! * **
☞ b. be%-di%-ba##"gá$=lu%$
M L H M
* *
c. be%-di-ba##"gá$=lu%$
M L H M
* *!
C
249
In this example there are two locations where we might expect rightward spreading due
to failure to deactivate a tone gesture. The first is from the specified M tone gesture onto
the toneless causative marker /di/. The second is spreading from the high-toned final
vowel of the verb root onto the clitic. Candidate a) has kept the tone gesture of the initial
syllable active through the toneless syllable, as predicted from the effects of *SELF-
DEACT. However, candidate a) has also kept the final H tone gesture active through a
glottal stop. This obeys the constraint DEACT-H-AT-UB but violates *ACTIVE-T-GS, and
the candidate loses. Candidate b), which is the winning candidate in this tableau, shows
that the final H tone of the verb root does not spread onto the clitic. This avoids a
violation of *ACTIVE-T-GS at the cost of DEACT-H-AT-UB. Candidate c) has lost due to
violation of *SELF-DEACT, since the tone gesture of the aspect marker has deactivated as
soon as its target is reached.
This concludes the section on SJZ and its floating tone and spreading tones. I
have shown several processes thus far, and I have described how the new constraints I
present here can account for these patterns. The constraint lattice showing the constraints
and their relationships is found in Figure 3.11:
250
Figure 3.11 Constraint lattices for SJZ tone deactivation
MAX-ROOT-T
*ACTIVE-T-GS
H-TONE-STRESS-V
DEACT-H-AT-UB
*MULTI-TONE-SHORT-V *SELF-DEACT
MAX-T *SPAN-SYLL
I have shown that these constraints account for the position of the floating tone in the
output on the initial syllable and or the stressed syllable of the verb root, and I have
shown that tones spread rightward onto toneless syllables. I have described and
accounted for the fact that verb-final H tones spread rightward onto the clitics, but that
this process is blocked when a glottal stop intervenes. Now that I have finished
accounting for the patterns in SJZ, I will discuss how these constraints can account for
very similar patterns in Macuiltianguis Zapotec, a closely related dialect.
251
3.6 Macuiltianguis Zapotec Tones
Macuiltianguis Zapotec (MacZ) is spoken in the town of Macuiltianguis in the
Ixtlán district of Oaxaca, Mexico. This language has been described in a manuscript by
Broadwell (2000) as well as in a dissertation by Foreman (2006). As of the year 2000,
there were only around 1,135 remaining speakers of MacZ (Foreman, 2006). The overall
phonology of the language is quite similar to SJZ, and the two varieties may be
considered dialects rather than separate languages (Foreman, 2006). In this section, I
examine processes of rightward spreading, and a floating H tone that is associated with
the first-person singular that are similar to those found in SJZ. I refer the reader to
Broadwell’s work as well as to the dissertation of John Foreman (2006) for a more
detailed overview of MacZ. I will first present some examples of tone spreading and
show how the constraints that I proposed for SJZ can successfully account for them.
Then, I will discuss the patterns and analysis of the floating H tone in MacZ. I include a
discussion of MacZ to show that the constraints introduced for the analysis of tone
patterns in SJZ are also applicable to similar but distinct tone patterns in MacZ, with the
differences resulting from different constraint rankings and different lexical
representations.
3.6.1 Tone spreading in MacZ
Recall that in SJZ, the focus of this chapter and chapter 2, a verb-final H tone
spreads rightward onto a clitic, even if this clitic has an underlying tone. I motivated this
behavior using the constraint DEACT-H-AT-UB. I also showed that in SJZ, the H tone of
252
the potential aspect marker does not spread rightward onto the verb root. I explained this
as the result of ranking MAX-ROOT-T over DEACT-H-AT-UB. That is, in SJZ, even
though H tones would normally spread rightward across a morpheme boundary, they do
not do so when this would result in deletion of a tone in the verb. Interestingly, MacZ
has a similar process of rightward spreading of H tones across morpheme boundaries, but
in this language the ranking of the constraints is slightly different and consequently a
different pattern emerges.
First, I will present two examples of verbs which are given in Broadwell (2000).
Then, I will show how my constraints can account for these patterns. The two examples
in (25a) and (25b) are from the verb “to roll up”. I will show the underlying
representations first, and then consider the output forms.
(25a) underlying form /be#-tu#"bí=na#/ Broadwell (2000, 6)
(25b) output form be#-tu#"bí=ná
COM-roll.up=3s
‘He rolled (it) up’
In example (25b), Broadwell notes that the completive aspect marker /be#/ has an
underlying L tone, as does the first syllable of the verb root /tu#/. The second syllable of
the verb root has an underlying H tone, and the third-person singular clitic has an
underlying L tone. The example in (25b) shows this rightward H spread from the verb-
final H tone onto the clitic – just as was shown for SJZ. However, the two languages
253
differ when it comes to rightward spread of the potential H tone, as shown in example
(26), below.
(26a) underlying form /gú-tu#"bí=na#/ Broadwell (2000, 6)
(26b) output form gú-tu!"bí=ná
POT-roll.up=3s
‘He will roll (it) up’
This example shows that the H tone of the potential aspect marker has spread rightward
onto a following L tone to form a falling contour. This is because the first syllable of the
verb root contains a vowel that is both open and long. As Broadwell shows, when the first
syllable of the verb root is a short vowel in a closed syllable, then no contour tone is
formed in this position. This is shown in examples (27a) through (27c):
(27a) underlying form
69
/tu#ppí$=na#/
(27b) output form be#-tu#ppí$=na# Broadwell (2000, 6)
COM-pull.out=3s
‘He pulled (it) out’
(27c) output form gú-túppí$=na# Broadwell (2000, 6)
POT-pull.out=3s
‘He will pull (it) out’
69
This is the underlying form of the verb root plus clitic.
254
These examples illustrate that the verb root has an underlying L tone on its first syllable.
Example (27c) shows that H tone of the potential aspect marker will spread rightward,
and that if it is not possible to form a contour on the following syllable when this occurs,
then rightward spreading results in deletion of the tone gesture on the first syllable of the
verb root. I propose that in MacZ, the constraint MAX-ROOT-T is not ranked as highly as
it is in SJZ. Therefore, in MacZ, the constraint DEACT-H-AT-UB is more highly ranked
than MAX-ROOT-T. The examples presented in (27a) through (27c) also show that a final
glottal stop blocks rightward spread from the verb-final H tone onto the clitic. This is
apparent because in the underlying form the third-person clitic has a L tone, and this L
tone is retained in both output representations. Recall that I showed the same blocking
effect of a glottal stop for the SJZ data.
Given these data, I will now present tableaux using the same constraints that were
applied to the SJZ data I examined earlier in this chapter. I will show how the same
constraints can also account for the patterns found in the neighboring dialect of MacZ.
The first example I illustrate in a tableau format is /gú-tu#"bí=na#/. This form has a high-
toned potential aspect marker, a verb root with the tones L-H, and a low-toned clitic.
There are two processes which modify this input. The H tone of the potential aspect
marker spreads rightward onto the first syllable of the verb root to form a contour tone
and the verb-final H tone spreads rightward onto the clitic /na#/, which becomes high. The
constraint H-TONE-STRESS-V will drive rightward spreading of the aspect marker’s tone
255
onto the verb root, while DEACT-H-AT-UB will drive spreading from the root-final H tone
onto the clitic. This is illustrated in Table 3.33:
Table 3.33 Potential H tone spreads rightward onto root in MacZ
gú-tu#"bí=na#
H L H L
DEACT
H AT UB
*MUL
TI
TONE-
SHORT
-V
H-TONE-
STRESS-
V
MAX-
ROOT
-T
MAX-
T
*MULT
I-
TONE
*SPA
N
SYLL
a. gú-tu#"bí=na#
H L H L
**! **
☞ b. gú-tu!" bí=ná
H L H
* * * * **
c. gú-tu!" bí=na#
H L H L
**! * * *
d. gú-tu#"bí=ná
H L H
* **! * *
e. gú-tú:bí=ná
H H
* * *(!) *(!) **
V
V
256
Candidates where the stressed vowel has more than one active tone gesture due to
spreading are shown with an extra V box as in the SJZ examples.
70
In Table 3.33,
candidates a) and c) incur extra violations of DEACT-H-AT-UB and are eliminated.
Candidate a) fails to spread the H tone of the aspect marker as well as the final H tone of
the verb root and loses. Candidate c) fails to spread the verb-final H tone, and loses.
Candidate d) and winning candidate b) are tied until H-TONE-TO-STRESS-V comes into
play. Candidate d) incurs an extra violation of this constraint because the H tone of the
aspect marker deactivates before it reaches the vowel of the stressed syllable. Candidate
(e) ties with the winner except that it violates MAX-ROOT-T and MAX-T. This shows that
at least one of these constraints dominates *MULTI-TONE. Candidate b) emerges as
winner and shows that the rightward spreading of the potential H tone in MacZ is driven
by the effects of H-TONE-TO-STRESS-V. Since the stressed syllable of the verb root has a
long vowel, there are no violations of *MULTI-TONE-SHORT-V.
MacZ differs from SJZ, as can be seen in the next tableau, where a tone of the
root is deleted. This is shown in Table 3.34 where the input /gú-tu#ppí$=na#/ is
considered. For this verb, the potential H tone spreads rightward and causes the
underlying tone of the first syllable of the root to delete. This deletion occurs because the
syllable is short and no contour can be formed. The verb-final H tone fails to spread
rightward in this example because a glottal stop will block the persistence of a tone
gesture. The interactions amongst the constraints are shown in Table 3.34:
70
Note that the box itself does not signal overlapping tones, although the two cases where
it appears in the tableau have overlapping tones.
257
Table 3.34 Potential H tone spreads rightward to delete tone of root in MacZ
gú-tu#ppí$=na#
H L H L
*ACTIVE
-
T-GS
DEACT
H AT
UB
*MULT
-T-
SHORT-
V
H-
TONE-
STRESS
-V
MAX
ROOT
T
MAX
-T
*SPAN
SYLL
a. gú-tu! ppí$=na#
H L H L
** *! * *
b. gú-tú ppí$=ná
H H
*! * * * ** **
☞ c. gú-tú ppí$=na#
H H L
** * * * *
d. gú-tu#ppí$=na#
H L H L
** **!
e. gú-tu#ppí$=ná
H L H
*! * ** * *
Here, V gestures with overlapping tones are shown with a V box, while tones that spread
through glottal stops are indicated in part with a C box. From this tableau the ranking for
MacZ becomes clearer. Candidates b) and e) are immediately ruled out due to violations
of *ACTIVE-T-GS. Candidates a), c), and d) all incur two violations of DEACT-H-AT-UB.
Candidate a) loses due to its formation of a contour on a short vowel (by spreading from
V
C
C
258
the aspect marker to the verb root). Candidates c) and d) are tied until H-TONE-TO-
STRESS-V comes into play. By failing to spread the H tone of the aspect marker rightward
to deactivate with the vowel of the stressed syllable, candidate d) incurs a fatal violation
and is eliminated. The winning candidate shown in c) spreads the aspect marker’s tone
rightward to satisfy H-TONE-STRESS-V while at the same time failing to spread the final
H tone of the verb root onto the clitic marker due to the effects of *ACTIVE-T-GS. Recall
that the different ranking for SJZ disallows the H tone of the aspect marker from deleting
a root tone in the way it occurs for MacZ. In SJZ, MAX-ROOT-T is ranked above H-
TONE-STRESS-V, and this prevents the aspect marker tone from spreading into the root to
deactivate with the stressed vowel. In MacZ on the other hand, H-TONE-STRESS-V is
ranked above MAX-ROOT-T. A constraint ranking for MacZ is shown in Figure 3.12.
259
Figure 3.12 Constraint lattice for MacZ
*ACTIVE-T-GS
MAX-UNCOUPLED-T DEACT-H-AT-UB
*MULTI-TONE-SHORT-V
H-TONE-STRESS-V
MAX-T
MAX-ROOT-T
COUPLE-INITIAL-V
*MULTI-TONE
*SPAN-SYLL
MacZ has low-toned aspect markers, which makes it an excellent testing ground
for the theory of tone deactivation I have presented. In this chapter I have shown that M
and H tones in SJZ spread rightward onto toneless syllables, but I have not shown any
examples where a L tone spreads rightward this way. This is because I did not uncover
any examples of verbs with a low-toned aspect marker in SJZ. The analysis I have
presented for SJZ using *SELF-DEACT predicts that tones will spread rightward onto
toneless syllables, and it predicts that this will happen regardless of whether the
articulatory target is H, M, or L for a tone. MacZ complements the SJZ data because it
260
does have low-toned aspect markers, and because it shows the same pattern as SJZ: tones
spread rightward onto toneless syllables.
In what follows I present some selected examples from Broadwell (2000) that
illustrate the pattern of rightward spread. Then I will relate this phenomenon to my
constraints governing tone deactivation. First, note that MacZ has stems where the initial
syllable of the verb root is toneless, and others where it has a specified tone. In the cases
where the first syllable is toneless, spreading is observed from the aspect marker onto the
root. In the cases where the first syllable of a verb root has a tone, spreading is only
observed when the verb root is preceded by a high-toned potential aspect marker. This is
evidence that both *SELF-DEACT and H-TONE-STRESS-V are at work. Consider the
examples in (28), below:
(28a) underlying form /det(u%=na#/
71
(28b) habitual output form ru%-de%t(u%=na# Broadwell (2000: 9)
HAB-fold=3s
‘He is folding (it)’
(28c) completive output form be#-de#t(u%=na# Broadwell (2000: 9)
COM-fold=3s
‘He folded (it)’
71
Underlying forms for aspect markers are /ru%/, /be#/, and /gu./.
261
(28d) potential output form gú-de.t(u%=na# Broadwell (2000: 9)
POT-fold=3s
‘He will fold (it)’
These examples are very exciting in terms of their implications for a theory of tone
deactivation. They provide evidence that tones spread rightward onto toneless syllables.
MacZ fills out the picture by giving an example where this happens when the articulatory
target of the tone is L, as I show in example (28c). For the habitual aspect shown in
(28b), the first syllable of the verb root has a M tone. I posit that this is due to rightward
spread from the habitual aspect marker, which is also mid-toned. Example (28c) shows
the completive aspect. Here, the first syllable of the root has a L tone, just like the aspect
marker /be#/. Again, this suggests rightward spreading from the aspect marker onto the
verb root. Finally, the potential aspect is shown in (28d), where the first syllable of the
verb root is also H. I have used examples from the third-person singular to avoid
confusion from the floating H of the first-person singular, which is also present in MacZ.
Because there is no floating H tone which could appear here, I argue that example (28d)
shows a H tone on the first syllable of the root due only to spreading from the potential
aspect marker.
Now I will examine how the constraints I have presented thus far can account for
this rightward spreading from an aspect marker onto a verb root. The tableau shown in
Table 3.35 has a low-toned aspect marker preceding the verb root. The stressed syllable is
underlined.
262
Table 3.35 Rightward L tone spread onto toneless syllable in Mac Z
be#-det(u%=na#
L X M L
DEP-T *SELF-
DEACT
*SPAN-
SYLL
a. be#-det(u%=na#
L X M L
*!
☞ b. be#-de#t(u%=na#
L M L
*
c. be#-det(u%=na#
L M M L
*!
This tableau shows the same pattern that I found for the SJZ data: DEP-T, *SELF-DEACT
>> *SPAN-SYLL. This means that tones will stay active until they reach a following tone.
In Table 3.35, candidate a) loses for a violation of *SELF-DEACT. Candidate c) also is
eliminated, but for inserting a M tone. This violates DEP-T and the candidate loses. The
winning option is shown in candidate b), where the L tone of the aspect marker stays
active through the stressed syllable and deactivates once it encounters a following tone.
Just like SJZ, the case where a H tone of the potential aspect marker spreads
rightward onto a toneless syllable in the root is ambiguous. This is because such
spreading could be driven by *SELF-DEACT as I have shown in Table 3.35 or it could be
due to the effects of H-TONE-STRESS-V. The tableau shown in Table 3.36 has both
constraints.
263
Table 3.36 Rightward H spread onto toneless syllable in MacZ
gú-det(u%=na#
H X M L
H-TONE-
STRESS-V
DEP-T *SELF-
DEACT
*SPANSYLL
a. gú-det(u%=na#
H M L
*! *
☞ b. gú dét(u%=na#
H M L
*
c. gú-de%t(u%=na#
H M M L
*! *
The constraint H-TONE-STRESS-V will penalize any candidate that does not deactivate
with the vowel of the stressed syllable (the vowel that is coupled to a stress oscillator).
Candidate a) violates this constraint in addition to *SELF- DEACT because the tone gesture
of the potential aspect marker has deactivated as soon as its target is reached. Candidate
c) also violates H-TONE-STRESS-V because even though a M tone has been inserted onto
the first syllable of the verb root (violating DEP-T), the H tone of the potential aspect
marker still does not deactivate with the vowel of the stressed syllable. Thus, the winning
candidate is shown in b), where the tone gesture of the potential aspect marker stays
active until it reaches a following tone.
Thus far I have shown that SJZ and MacZ are quite similar in terms of tone
spreading onto toneless syllables. The ranking DEACT-H-AT-UB >> Max-T, *SELF-
DEACT >> *SPAN-SYLL can account for both dialects in cases where a root-final H tone
264
spreads onto an attached clitic. I have also shown that the ranking of *H-TONE-STRESS-V
and MAX-ROOT-T is different in the two dialects. A constraint lattice for MacZ is shown
in Figure 3.13:
Figure 3.13 Constraint lattice for MacZ tone spreading
H-TONE-STRESS-V
DEP-T
*SELF-DEACT
*SPAN-SYLL
3.6.2 Floating Tone in MacZ
Just like SJZ, MacZ uses a floating H tone to mark first-person singular transitive
verbs. Although there is no published work on this phenomena, the manuscript by
Broadwell (2000) cites examples of the first-person singular that are quite similar to the
data I have presented in chapter 2 for SJZ. In this section I will show how the constraints
I used for SJZ can also account for the MacZ data.
First I will present data from Broadwell (2000: 9). Consider the forms from the
verb “to iron” which are almost identical to the forms I presented for SJZ. I give data
from the first and third-person in the completive, habitual, and potential aspects using a
265
table. These forms show both the presence of the floating H as well as rightward tone
spreading from the aspect marker onto the verb root.
Table 3.37 Potential and habitual forms for 1s and 3s in MacZ
/&atta%$/
‘to iron’
Potential Habitual Completive
First-person gú-&átta%$=ya#$
POT-iron=1s
‘I will iron’
ru%-&átta%$=ya#$
HAB-iron=1s
‘I am ironing’
be#-&átta%$=ya#$
COM-iron=1s
‘I ironed’
Third-person gú-&átta%$=na#
POT-iron=3s
‘He will iron’
ru%-&a%tta%$=na#
HAB-iron=3s
‘He is ironing’
be#-&a#tta%$=na#
COM-iron=3s
‘He ironed’
The data in Table 3.37 shows a verb root with an initial toneless syllable and a second
syllable that has a specified M tone. In the first-person singular, the first syllable of the
verb root has a floating H tone on the stressed syllable of the verb root in each case – for
the potential, habitual, and the completive. Unlike SJZ, where I showed that the habitual
aspect marker was underlyingly toneless, in MacZ this aspect marker has a specified M
tone, so the floating H of the first-person singular cannot appear in this location. Also
observe in Table 3.37 that for the third-person singular forms, in each case the first
syllable of the verb root takes whatever tone is to its immediate left. When the aspect
marker has a L tone, the first syllable of the verb root is L, and when the aspect marker
has a H tone the first syllable of the verb root is also H.
Now that I have presented these data, consider how the constraints presented in
this chapter and their ranking for SJZ can account for the floating H in MacZ. The
266
tableau shown in Table 3.38 only shows constraints that govern the appearance of the
floating H tone. I will consider deactivation of tone gestures in a separate case. Table
3.38 shows the input /ru%-&atta%$=ya#$/ with a floating H tone. The initial syllable of the
verb root is toneless.
Table 3.38 Floating H appears on stressed syllable in MacZ
ru%-&atta%$=ya#$
M X M L
MAX-
UNCPLED-
TONE
H-
TONE-
STRESS-
V
MAX-
T
COUPLE-
INITIAL-
V
a. ru%- &atta%$=ya#$
M X M L
*! * **
☞ b. ru%- &átta%$=ya#$
M H
1
M L
***
c. ru%- &atta%$=yá$
M X M H
1
*! * **
d. rú- &atta%$=ya#$
H
1
X M L
*! * **
e. rú- &átta%$=ya#$
H
1
M L
*! **
This particular example is handled easily by the constraints shown in Table 3.38, since
the initial syllable of the verb root (the stressed syllable) is toneless in the input. The
H
1
267
floating tone may not be deleted, as shown via candidate a), since this violates MAX-
UNCOUPLED-TONE. Candidate b) shows the best solution, where the floating tone has
coupled to the vowel of the stressed syllable. This does not violate MAX-T, since no
underlying tone is deleted. Candidates c) and d) are eliminated for violating H-TONE-
STRESS-V and both also violate MAX-T. Candidate e) manages to escape violation of H-
TONE-STRESS-V by keeping the tone gesture active until it reaches the vowel of the
stressed syllable, but the candidate loses for having violated MAX-T via deletion of the
initial tone on the aspect marker. Table 3.38 shows one important difference between
MacZ and SJZ. In SJZ, the habitual aspect marker is toneless, and as I showed, whenever
there is a toneless aspect marker the floating tone shows up in that location in addition to
the stressed syllable. In MacZ, at least for the data presented by Broadwell, there are no
toneless aspect markers. This means that if the floating tone were to couple in initial
position in the prosodic word, a violation of MAX-T would occur. As I have shown in
Table 3.38 this would result in the candidate losing. The constraints and their ranking are
different in the two dialects, and the inputs are slightly different. In MacZ it will never be
the case that a floating tone can appear on the initial syllable because MAX-T >>
COUPLE-INITIAL-VOWEL and because there are no toneless aspect markers. Also, as I
have shown, in MacZ, H-TONE-STRESS-V >> MAX-ROOT-T whereas in SJZ, the ranking
is reversed. Any one of the examples given in Table 3.37 could be substituted into the
tableau in Table 3.38 with the same result: the floating tone will be drawn to the stressed
syllable of the verb root and thus will avoid deleting an underlying tone in another
location.
268
Next, it is useful to consider what happens in MacZ when the vowel in the
stressed syllable of the verb root already has an underlying tone coupled to it. Just as I
showed for SJZ, in MacZ the floating H tone of the first-person singular will appear with
the underlying lexical tone to form a falling contour tone. To test the constraints I have
developed for SJZ on data from MacZ, I will use data from a verb with an underlying L-
M form in the verb root. First, observe the pairs shown in (29) and (30) below:
(29a) underlying form 1s /be#-da#"ga%=ya#$/
(29b) completive form 1s be#-da!"ga%=ya#$ Broadwell (2000:11)
COM-run.into=1s
‘I ran into (him)’
(30a) underlying form 3 s /be#-da#"ga%=na#/
(30b) completive form 3 s be#-da#"ga%=na# Broadwell (2000:11)
COM-run.into=3s
‘He ran into (him)’
Example (29b) shows that for the first-person singular, the first syllable of the verb root
has a falling contour tone. I argue that this results from the floating H of the first-person
singular. As shown, I posit that the floating H tone is not coupled in the underlying form
but that it is coupled to the stressed syllable and appears to form a falling contour tone in
the output form, as shown in (29b). Now, consider the form of the first-person singular in
H
1
269
tableau format with the constraints I have proposed in this chapter, and again, their
ranking as in SJZ. This is shown in Table 3.39:
Table 3.39 Floating tone forms a falling contour on stressed syllable in MacZ
be#-da#"ga%=ya#$
L L M L
MAX-
UNCPLED-
TONE
*MULTI-
TONE-
SHORT-
V
H-TONE-
STRESS-
V
MAX-
T
*MULTI-
TONE
a. be#-da#"ga%=ya#$
L L M L
*! *
☞ b. be#-da!" ga%=ya#$
L H
1
L M L
*
c. be#-da#"ga!= ya#$
L L H
1
M L
*!
*
*
d. be#-dá"ga%=ya#$/
L H
1
M L
*!
This tableau looks very similar to others I presented earlier for the SJZ data. Here, the
vowel of the stressed syllable of the verb root has an underlying L tone coupled to it. This
syllable has a falling tone after the floating tone of the first-person singular appears in the
output. Candidate a) loses for having deleted the floating tone, which violates MAX-
UNCOUPLED-T. Candidate c) loses for having formed a contour on a short vowel and thus
for violating *MULTI-TONE-SHORT-VOWEL. Candidate d) loses for having deleted the
H
1
270
underlying L tone of the stressed syllable in favor of the floating H tone. The winning
candidate is b), where the floating H tone of the first-person singular appears on the
stressed syllable without deleting the underlying tone in that position. Since the vowel is
long, a contour forms. This example shows that the floating H tone of MacZ is quite
similar in its behavior to the floating H tone discussed in chapter 2 of this dissertation. A
constraint lattice for the MacZ floating tone is shown in Figure 3.14:
Figure 3.14 Constraint lattice for MacZ floating tone
MAX-UNCOUPLED-T *MULTI-TONE-SHORT-V
H-TONE-STRESS-V
MAX-T
COUPLE-INITIAL-V
*MULTI-TONE
In summary, it appears based on the data given in Broadwell (2000) that MacZ is
quite similar to SJZ with regard to its tonal behavior. More work to document the tonal
properties of MacZ would certainly be useful to determine how closely the tone
paradigms of the two dialects overlap. In this section I have discussed two separate
processes in MacZ. One is the rightward spreading of tone gestures onto toneless
syllables and the spreading of H tones rightward. I motivated such spreading via two
271
constraints: one which penalizes deactivation via a gesture’s own clock, and one which
penalizes deactivation of a H tone gesture anywhere except at an utterance boundary.
Then, I examined the floating H tone of the first-person singular, which I suggested is
uncoupled in the input but obtains its coupling location via the effects of the same
constraints as in SJZ, ranked in a slightly different order. The main differences I propose
here between MacZ and SJZ are different inputs in MacZ, and the ranking between H-
TONE-STRESS-V and MAX-ROOT-T. In SJZ, MAX-ROOT-T is ranked above H-TONE-
STRESS-V, while in MacZ, the ranking is reversed.
3.7 Conclusion
This chapter has presented two core analyses. One covers tone gesture
deactivation and its effects in the grammar. As I have shown via the presentation of
constraints and their ranking, tone gesture deactivation occurs in SJZ at the onset of a
following anti-phase coupled tone or at an utterance boundary. A constraint may also
require deactivation with the vowel gesture of a stressed syllable. When no deactivators
are present it results in rightward spreading from toneful to toneless syllables. I argue that
this type of tone deactivation may cause spreading in the output (and violations of
*SPAN-SYLL) and yet at the same time not involve insertion of a new tone gesture. At the
gestural level, tones are still controlled via coupling. My contribution adds a component
in the grammar which can control when tones deactivate. By making use of the
representation of tones as gestures, my approach has the flexibility to explain the
rightward bias in tone spreading, and it makes predictions for the Zapotec data I have
272
presented in chapter two of this dissertation. The gestural approach to tone spreading
opens an explanation for a rightward asymmetry that is not otherwise available.
I have also argued that a set of constraints operates to determine where floating
tones are realized. In SJZ, I argue that this placement occurs always on the stressed vowel
of the root, and sometimes on toneless aspect markers and on causative markers that
intervene between the toneless aspect marker and the stressed syllable of the verb root.
The constraints I have proposed to account for these patterns do not involve any type of
new activation paradigms for gestures. I chiefly adapt pre-existing grammatical
constraints such as those that call tones to prominent syllables or to initial positions and
make them compatible with a gestural framework. That is, my constraints operate over
the grammar to require gestures to couple with vowels in certain positions or to
deactivate at certain points. I also propose constraints that prohibit overlapping activation
of gestures. The one exception is the constraint PRES-COUP-IO, which serves to hold
gestures to their same coupling locations between input and output.
To summarize some of the theoretical proposals made in this chapter, it is
important to note that there are six main constraint types which have been proposed.
They belong to the two main constraint families: markedness constraints and faithfulness
constraints. Most all of these constraints are markedness, with the exception of four.
The faithfulness constraints are PRES-COUP-IO, MAX-T, DEP-T, and MAX-ROOT-
T. Recall that PRES-COUP-IO requires input coupling relationships to remain unchanged
in the output. MAX-T and MAX-ROOT-T are more familiar faithfulness constraints, with
the exception that they refer specifically to gestures and require that a tone gesture
273
present in the input also be present in the output. DEP-T prohibits insertion of a tone
gesture.
The rest of the constraints proposed in this chapter are markedness constraints.
First, there are constraints which require coupling in a certain position, such as COUPLE-
INITIAL-V and COUPLE-H-PROM-V. These constraints penalize coupling relationships that
do not occur in certain positions such as with the vowel of the initial vowel or with a
prominent vowel gesture.
Second, there are constraints which penalize specific coupling relationships such
as *C-T, *T-T, *V-T, *HD/L, *HD/M. These are also markedness constraints because
they operate over the output coupling graph to penalize specific coupling structures in the
graph.
Third, there are constraints which require deactivation in specific locations, such
as H-TONE-STRESS-V and DEACT-H-AT-UB. These constraints are markedness
constraints in that they operate over the output candidates and penalize deactivation that
does not occur in specific locations.
Fourth, there are two constraints which penalize deactivation in various locations,
such as *SELF-DEACT and *SPAN-SYLL. Both are markedness constraints. *SELF-DEACT
penalizes deactivation that occurs at 360 degrees of a gesture’s own clock if there is no
other deactivator present such as a following anti-phase coupled tone or an utterance
boundary, while *SPAN-SYLL penalizes deactivation that occurs at more than 360 degrees
of a tone gesture’s own clock.
274
Finally, there are constraints that penalize overlapping activation of gestures.
These constraints are markedness constraints that operate on the first level output of the
coupling graph – the gestural score. These constraints are *MULTI-TONE, *MULTI-TONE-
SHORT-V, and *ACTIVE-T-GS. All three constraints penalize a vowel gesture that
overlaps with another gesture such as multiple tone gestures or a glottal constriction
gesture. I proposed via these constraints that EVAL can reference overlap of the gestural
score without evaluating the entire gestural score itself (which includes real time).
By grouping the constraints proposed in this dissertation into ‘families’, we can
see that there are six main types that belong to the two constraint types of OT.
As I have shown in this chapter, most verbs when conjugated for the first-person
singular show both the floating H tone of the first-person singular in addition to the
effects of *SELF-DEACT or DEACT-H-AT-UB. Thus we need both the new approach to
tone deactivation that allows for rightward spreading in addition to constraints that
govern the placement of the floating H tone. My approach incorporates both these aspects
and then, as I have shown, can also be applied to the neighboring dialect of
Macuiltianguis Zapotec (MacZ). As I will suggest in the following chapter which
compares my approach to traditional autosegmental analyses, these types of constraints
can also explain data from other languages that are not related to either Sierra Juárez
Zapotec or Macuiltianguis Zapotec
275
3.8 Summary of Constraint Definitions
(31) *ACTIVE THROUGH GLOTTAL STOP – Assign a violation mark to each tone
gesture whose activation overlaps a glottal constriction gesture.
(32) *C-T: Assign a violation mark to each coupling relationship between a
consonant and a tone gesture in the output.
(33) COUPLE-H-PROM-V: Assign a violation to any H tone gesture not
coupled in-phase with a prominent vowel gesture in the output.
(34) COUPLE-INITIAL-VOWEL: Assign a violation mark to any tone gesture
which is not coupled to the initial vowel of the prosodic word.
(35) DEACTIVATE-H-AT-UTTERANCE-BOUNDARY: Assign a violation mark for
each H tone gesture which deactivates anywhere other than at an
utterance boundary.
(36) DEP-T: A tone gesture in the output has an input correspondent.
(37) *Hd/L : ‘No L on head syllables’ (de Lacy 2007:283). = No L tone gesture coupled
on a head syllable.
(38) *Hd/M : ‘No M on head syllables’ (de Lacy 2007:283). = No M tone gesture
coupled on a head syllable.
(39) H-TONE-TO-STRESS V: Assign a violation to any H tone gesture whose
deactivation is triggered other than by the vowel gesture that is
coupled in-phase with a stress oscillator.
(40) MAX-T: Tone gestures in the input have output correspondents.
276
(41) MAX-ROOT-T: Tone gestures in the root in the input have output
correspondents
(42) MAX-UNCOUPLED-TONE – Uncoupled tone gestures in the input have
output correspondents.
(43) *MULTI TONE: Assign a violation mark for each vowel gesture whose activation
overlaps the activation of more than one tone gesture.
(44) *MULTI-TONE-SHORT-VOWEL- Assign a violation mark for each short vowel
gesture whose activation overlaps the activation of more than one tone
gesture.
(45) PRESERVE-COUPLING –IO: Let X and Y be gestures in the input, and X’ and Y’
be correspondents of X and Y in the output, respectively. If X is coupled
with Y in the input, then X’ is coupled to Y’ in the output.
(46) *SELF DEACTIVATE BY CLOCK: Assign a violation mark to any tone that
deactivates other than at a deactivator, where a deactivator is either a
boundary (end of utterance) or a following anti-phase coupled tone.
(47) *SPAN-SYLL – assign a violation mark for each tone gesture that deactivates later
than at 360 degrees of its own clock.
(48) *T-T: Assign a violation mark to each coupling relationship between two
tone gestures in the output.
(49) *UNCOUPLED-T: Assign a violation mark for each tone gesture in the
output that is not coupled with another gesture in the
output.
(50) *V-T: Assign a violation mark to each coupling relationship between a
vowel and tone gesture in the output.
277
Chapter 4 Case Studies and Future Directions
4.1 Introduction
This chapter has several goals. First, I will explore typological predictions made
by the deactivation constraints presented in chapter 3. Then, I compare other approaches
to tonal spreading and movement with the approach developed in this dissertation. In
chapter 3 I argue that tones are articulatory gestures, and that their deactivation can be
controlled via constraints in the phonological grammar. In this chapter I will argue that
my approach enhances the representations for tone beyond that of autosegmental
approaches by allowing tone deactivation to be separately controlled from the coupling
location of the onset of a tone. I will argue that the constraints I propose in chapter 3
have an advantage over a rule-based approach to tone spreading in that they help explain
the rightward bias in the typology of tone spreading. Although Hyman (2007) has noted
that there seems to be a rightward bias in spreading of tones and although this
observation goes back as far as Goldsmith (1976), thus far there is no particular
understanding for why this is the case, aside from the observation that tone targets tend to
be reached late in their respective syllables. Kaplan (2008) has proposed a convincing
argument for movement of tones one syllable to the right called Peak Delay Theory. As
outlined in chapter 3, I argue that the tendency for tones to spread rightward or to stay
active longer than one syllable is due to their specific deactivation typology. Thus in
some languages, tones may never spread rightward because they are deactivated upon
278
reaching their targets. In other languages, they may stay active until reaching a following
anti-phase coupled tone, or the end of the utterance.
In this chapter I will first discuss the typology predicted by the two constraints
governing tone deactivation: *SELF-DEACT and *SPAN-SYLL. I discuss the types of
grammars that these constraints predict according to their respective ranking. I also
discuss the typology that is predicted when four constraints are considered: PRES-COUP-
IO, COUPLE-INITIAL-VOWEL, *SELF-DEACT, and *SPAN-SYLL. I then examine a typology
predicted by the relationship of DEP-T, *SELF-DEACT and *SPAN-SYLL. Next, I illustrate
how these constraints can be applied to various languages by presenting case studies. I
then compare and contrast my approach with autosegmental phonology. I discuss
alternate approaches such as Optimal Domains Theory (Cassimjee and Kisseberth 1998),
and Peak Delay Theory (Kaplan 2008). Finally, I discuss topics for further research.
4.2 Typology of Deactivation Constraints
This section explores typological predictions made by the deactivation constraints
*SELF-DEACT and *SPAN-SYLL in combination with other constraints such as DEP-T,
COUPLE-INITIAL-VOWEL, and PRES-COUP-IO. In OT, the typology of patterns predicted to
be possible in languages results from factorial ranking of the constraint set. In the classic
conception of OT, constraints are the same across languages and there are no language-
particular restrictions on inputs (Richness of the Base; Prince and Smolensky, 1993). The
constraint ranking is therefore the language-particular part of the grammar. Not every
ranking produces a distinct pattern, because multiple rankings may converge on the same
279
outcome. To test the typological predictions of the constraints I present, factorial
typologies for constraint sets were generated using OTSoft, Version 2.1 (Hayes et al.
2003). First, I repeat the definitions of the two deactivation constraints in (1) and (2)
below:
(1) *SELF DEACTIVATE BY CLOCK: Assign a violation mark to any tone that
deactivates other than at a deactivator, where a deactivator is either a
boundary (end of utterance) or a following anti-phase coupled tone.
(2) *SPAN-SYLL – assign a violation mark for each tone gesture that remains
active for longer than 360 degrees of its own clock.
Because there are only two constraints that govern deactivation of tone gestures, there are
only two possible grammars that are predicted just by these two constraints. When
*SELF-DEACT is ranked above *SPAN-SYLL, spreading will occur from a toned syllable to
a toneless syllable. This is the pattern that was presented in chapter 2 for SJZ. When
*SPAN-SYLL is ranked above *SELF-DEACT, then the opposite pattern occurs and no tone
spreading is observed. The approach I adopt using just these two constraints does not
predict spreading only one syllable to the right. As I discuss in a following section, the
type of spreading that occurs only one syllable to the right can be effectively handled by
Peak Delay Theory (PDT) proposed by Kaplan (2008). By themselves, the two
constraints I present in this dissertation predict then that tones should either spread all the
280
way until the onset of a following anti-phase coupled tone, or that they should not spread
at all.
I also consider the factorial typology that is predicted by incorporating COUPLE-
INITIAL-VOWEL and PRES-COUP-IO. These constraints are defined in (3) and (4):
(3) COUPLE-INITIAL-VOWEL: Assign a violation mark to any tone gesture
which is not coupled to the initial vowel of the prosodic word.
(4) PRESERVE-COUPLING –IO: Let X and Y be gestures in the input, and X’
and Y’ be correspondents of X and Y in the output, respectively.
If X is coupled with Y in the input, then X’ is coupled to Y’ in the
output.
As presented in chapter 3, COUPLE-INITIAL-VOWEL will draw tones to couple to the intial
vowel of the prosodic word, while PRES-COUP-IO will hold underlying coupling
relationships in place. In SJZ, PRES-COUP-IO prevented all tone gestures except the
floating tone from moving into initial position. However, in a language where PRES-
COUP-IO is less highly ranked, a different result could emerge.
Inputting these four constraints to OTSoft determined that the number of logically
possible grammars is 24, while there were only 4 different output patterns. I chose these
four constraints in particular because their interaction has the capacity to drive both
movement of tones (potentially looking like leftward spread) and to cause rightward
spreading. I will discuss each grammar below, along with its sample candidates. The
presentation of each grammar is preceded by a table showing the input and output
candidates that were used with the OTSoft program. The three candidates are shown
281
across the top row of the table, with their potential outcomes listed below. In these
grammars I did not consider a floating tone as part of the input candidates. Candidates
were chosen to demonstrate possible ways of arranging the tones that would illustrate any
ranking effects between constraints. The winning candidates for Grammar 1 are marked
with a pointing hand in Table 4.1.
Table 4.1 Input/Output candidates for OTSoft
Input Candidates CV- CV-CV-CV
M X X H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H X
Possible Output 1 ☞ CV- CV-CV-CV
M X X H
CV- CV-CV-CV
M X X X
☞ CV- CV-CV
X H X
Possible Output 2 CV- CV-CV-CV
M H
☞ CV- CV-CV-CV
X X M X
CV- CV-CV
X H
Possible Output 3 CV- CV-CV-CV
M X H
CV- CV-CV-CV
M
CV- CV-CV
H X X
Possible Output 4 CV- CV-CV-CV
M H
CV- CV-CV-CV
X X M
CV- CV-CV
H
The ranking of constraints shown in Sample Grammar 1 gives the output candidates
shown in Figure 4.1 as winners. To save space, I only show the tones and do not include
the CV representations that were included in Table 4.1.
282
Figure 4.1 Sample Grammar 1
*SPAN-SYLL, PRES-COUP-IO >> *SELF-DEACT, COUPLE-INITIAL-V
a. M---X---X---H M-----X----X-----H
b. X----X----M---X X----X----M-----X
c. X---H---X X----H-----X
The grammar illustrated in Figure 4.1 is one where PRES-COUP-IO holds all coupling
relationships in place. *SPAN-SYLL prevents tone gestures from staying active once their
targets are reached. Three sample input and output pairs are given in a), b) and c) and,
since both PRES-COUP-IO and *SPAN-SYLL are in the top tier of the hierarchy, no changes
occur between input and output candidates. The grammar shown in Figure 4.1 predicts a
language such as Lithuanian, which I discuss in my case studies – tones do not move to
initial position nor do they spread.
Grammar 2, illustrated in Figure 4.2, predicts tone movement to the initial
syllable but no tone spreading. This results from the ranking of *SPAN-SYLL above
*SELF-DEACT. For this ranking, the same candidate set was put into OTSoft, albeit with
different winners. This is shown in Table 4.2.
283
Table 4.2 Input/Output candidates for OTSoft Grammar 2
Input Candidates CV- CV-CV-CV
M X X H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H X
Possible Output 1 ☞ CV- CV-CV-CV
M X X H
☞ CV- CV-CV-CV
M X X X
CV- CV-CV
X H X
Possible Output 2 CV- CV-CV-CV
M H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H
Possible Output 3 CV- CV-CV-CV
M X H
CV- CV-CV-CV
M
☞ CV- CV-CV
H X X
Possible Output 4 CV- CV-CV-CV
M H
CV- CV-CV-CV
X X M
CV- CV-CV
H
Figure 4.2 illustrates the output of the grammar with simplified candidate input and
outputs.
Figure 4.2 Sample Grammar 2
*SPAN-SYLL, COUPLE-INITIAL-VOWEL >> *SELF-DEACT, PRES-COUP-IO
a. M----X----X----H M----X----X----H
b. X------X----M---X M----X----X----X
c. X----H ---X H---X----X
I do not have an example like this as part of my case studies. However, a language like
this would have movement of tones to prominent position (initial), but, due to the ranking
284
of *SPAN-SYLL, no spreading. As shown by the second and third input/output pairs, a
tone would move to initial position under this ranking and then deactivate via its own
clock.
72
Grammar 3 is the ranking that predicts a pattern like that in SJZ or MacZ. As
illustrated in Figure 4.3, this ranking predicts rightward spreading due to the effects of
*SELF-DEACT, but no movement of tones to initial position, due to ranking PRES-COUP-
IO above COUPLE-INITIAL-VOWEL. The input candidates and winners are illustrated in the
table below:
Table 4.3 Input/Output candidates for OTSoft Grammar 3
Input Candidates CV- CV-CV-CV
M X X H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H X
Possible Output 1 CV- CV-CV-CV
M X X H
CV- CV-CV-CV
M X X X
CV- CV-CV
X H X
Possible Output 2 ☞ CV- CV-CV-CV
M H
CV- CV-CV-CV
X X M X
☞ CV- CV-CV
X H
Possible Output 3 CV- CV-CV-CV
M X H
CV- CV-CV-CV
M
CV- CV-CV
H X X
Possible Output 4 CV- CV-CV-CV
M H
☞ CV- CV-CV-CV
X X M
CV- CV-CV
H
72
Additional constraints such as *MULTI-TONE or *CONTOUR could rule out multiple
tone gestures on the same vowel, which would prevent all tone gestures from ending up
on the initial syllable’s vowel.
285
The constraint ranking and simplified input/output pairs are shown in Figure 4.3.
Figure 4.3 Sample Grammar 3
*SELF-DEACT, PRES-COUP-IO >> *SPAN-SYLL, COUPLE-INITIAL-VOWEL
a. M-----X----X----H M---------------H
b. X-----X----M---X X----X----M-----]
73
c. X----H----------] X-----H--------]
As discussed in the case studies I present, Grammar 3 can account for a language with
unbounded tone spreading. Tone movement is not predicted due to the effects of PRES-
COUP-IO. In Figure 4.3, tones stay active until they reach a following anti-phase coupled
tone or the utterance boundary.
Finally, Grammar 4, illustrated in Figure 4.4 predicts a language with what looks
like leftward spreading of a tone. This occurs when a tone moves to initial position and
then remains active due to the effects of *SELF-DEACT. Table 4.4 illustrates the input and
output candidates to OTSoft.
73
Brackets indicate the end of a tone via the utterance end in the examples for this
sample grammar.
286
Table 4.4 Input/Output candidates for OTSoft Grammar 4
Input Candidates CV- CV-CV-CV
M X X H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H X
Possible Output 1 CV- CV-CV-CV
M X X H
CV- CV-CV-CV
M X X X
CV- CV-CV
X H X
Possible Output 2 ☞ CV- CV-CV-CV
M H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H
Possible Output 3 CV- CV-CV-CV
M X H
☞ CV- CV-CV-CV
M
CV- CV-CV
H X X
Possible Output 4 CV- CV-CV-CV
M H
CV- CV-CV-CV
X X M
☞ CV- CV-CV
H
Figure 4.4 Sample Grammar 4
*SELF-DEACT, COUPLE-INITIAL-V >> *SPAN-SYLL, PRES-COUP-IO
a. M-----X----X----H M---------------H
b. X-----X----M---X M----------------]
c. X------H-----X H---------]
Figure 4.4 illustrates that if COUPLE-INITIAL-V and *SELF-DEACT are both high-ranked,
then a tone will move to initial position but remain active until it encounters a following
anti-phase coupled tone or an utterance boundary. As illustrated in candidates (b) and (c),
this will cause what looks like leftward spreading, because a tone will move its coupling
287
(and onset) to initial position, but then remain active through following toneless syllables.
As I discuss in the case studies, this is the ranking that obtains for Margi.
I also consider a factorial typology for just three constraints: DEP-T, which
penalizes the insertion of a tone gesture, and the two deactivation constraints *SELF-
DEACT and *SPAN-SYLL.
74
I considered these three constraints separately from the
constraints in the previous paragraphs to simplify the output from OTSoft. A language’s
relative ranking of DEP-T and the deactivation constraints can determine whether
insertion of tone gestures is preferred to spreading them or leaving syllables toneless in
the output. OTSoft generated 3 possible output patterns with these constraints. Thus,
although there are six logically possible grammars, the rankings only result in 3 different
output patterns.
74
I did not consider a ranking with MAX-T, although the possibility of cases where tones
or syllables are deleted in order to satisfy *SELF-DEACT or *SPAN-SYLL is not ruled out
by my approach. Bickmore and Broadwell (1998) discuss insertion of a tone gesture on a
toneless syllable, so I include a possible ranking with DEP-T to account for that
possibility. If *SELF-DEACT were ranked above MAX-T, I would predict deletion of the
tone that would otherwise self-deactivate.
288
Table 4.5 Input/Output candidates for OTSoft Grammar 1 with Dep-T
Input Candidates CV- CV-CV-CV
M X X H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H X
Possible Output 1 ☞ CV-CV-CV-CV
M X X H
CV- CV-CV-CV
X X M
☞ CV- CV-CV
X H X
Possible Output 2 CV- CV-CV-CV
M H
☞ CV- CV-CV-CV
X X M X
CV- CV-CV
X H
Possible Output 3 CV- CV-CV-CV
M X H
CV- CV-CV- CV
M M M M
CV- CV-CV
M H M
Possible Output 4 CV- CV-CV-CV
M M M H
Grammar 1 with DEP-T has the ranking shown in Figure 4.5:
Figure 4.5 Grammar 1 with DEP-T
*SPAN-SYLL, DEP-T >> *SELF-DEACT
a. M-----X----X----H M---X----X----H
b. X-----X----M---X X----X----M----X
c. X------H-----X X-------H------X
In a language like the one shown here, no tones are inserted and no spreading occurs.
Syllables with no tone gesture coupled to them remain toneless in the output, and tone
289
gestures do not spread due to the effects of *SPAN-SYLL. A language with a pattern like
this is discussed in the Lithuanian case study in a following section of this chapter.
For Grammar 2 with DEP-T, a table with the candidate inputs to OTSoft is shown
below, with winning candidates marked via a pointing hand.
Table 4.6 Input/Output candidates for OTSoft Grammar 2 with Dep-T
Input Candidates CV- CV-CV-CV
M X X H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H X
Possible Output 1 CV- CV-CV-CV
M X X H
CV- CV-CV-CV
X X M
CV- CV-CV
X H X
Possible Output 2 CV- CV-CV-CV
M H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H
Possible Output 3 CV- CV-CV-CV
M X H
☞ CV- CV-CV- CV
M M M M
☞ CV- CV-CV
M H M
Possible Output 4 ☞CV-CV-CV-CV
M M M H
Grammar 2 with DEP-T is illustrated in Figure 4.6:
290
Figure 4.6 Grammar 2 with DEP-T
*SPAN-SYLL, *SELF-DEACT >> DEP-T
M-----X----X----H M---M----M-----H
X-----X----M---X M----M----M----M
X------H-----X M------H-------M
In a language like the one shown in Figure 4.6, no tone gestures will spread due to the
effects of *SPAN-SYLL. However, since *SELF-DEACT also comes into play, the grammar
inserts a M tone to prevent toneless syllables in the output.
75
This violates DEP-T but
obeys the higher-ranked constraints. A grammar like this one is discussed for Weinan
Chinese in a following section.
Finally, consider the third grammar generated by OTSoft for these constraints,
shown in Figure 4.7. The table showing input and output candidates is given first.
75
I have arbitrarily chosen a M tone as the one inserted. Note that DEP-T makes no
reference to the target of a tone – a H or L tone could also be inserted.
291
Table 4.7 Input/Output candidates for OTSoft Grammar 3 with Dep-T
Input Candidates CV- CV-CV-CV
M X X H
CV- CV-CV-CV
X X M X
CV- CV-CV
X H X
Possible Output 1 CV- CV-CV-CV
M X X H
☞CV- CV-CV-CV
X X M
CV- CV-CV
X H X
Possible Output 2 ☞CV- CV-CV-CV
M H
CV- CV-CV-CV
X X M X
☞CV- CV-CV
X H
Possible Output 3 CV- CV-CV-CV
M X H
CV- CV-CV- CV
M M M M
CV- CV-CV
M H M
Possible Output 4 CV- CV-CV-CV
M M M H
Figure 4.7 Grammar 3 with DEP-T
*SELF-DEACT , DEP-T >> *SPAN-SYLL
M-----X----X----H M----------------H
X-----X----M---X X----X-----M------]
X------H-----X X-------H---------]
A grammar such as the one shown in Figure 4.7 will ban insertion of tone gestures, but
also leave tone gestures active until either a following anti-phase coupled tone or an
292
utterance boundary. A grammar like this was discussed for SJZ, and is also presented in
upcoming case studies for Margi.
To summarize the section on typology, I want to emphasize that the two
deactivation constraints predict either unbounded rightward spreading or no spreading at
all. As I have explained, *SELF-DEACT >> *SPAN-SYLL predicts unbounded spread, while
*SPAN-SYLL >> *SELF-DEACT predicts that tone gestures will deactivate upon reaching
their targets and fail to spread onto toneless syllables. Spreading only one or two
syllables to the right but deactivating prior to the onset of a following tone is ruled out
under my approach. The outputs I presented from OTSoft suggest that when other
constraints intervene – such as a constraint that calls a tone to initial position, then tones
can move their onset if PRES-COUP-IO is ranked low enough in the grammar. Because
PRES-COUP-IO references the coupling or onset of a tone, it does not interact with the
deactivation constraints to cause leftward or rightward spreading. Rather, the interaction
of PRES-COUP-IO with constraints calling tones to specific positions can drive this
apparent leftward spreading. As I argue in this dissertation, if GEN requires tones to be
anti-phase coupled to one another in the output, then deactivation or the end of a tone will
always occur via deactivation at target, or via deactivation by following tone or utterance
boundary.
76
In summary, OTSoft demonstrated that *SPAN-SYLL and *SELF-DEACT can
create three different grammars when combined with DEP-T. It was also shown that
combining the deactivation constraints with PRES-COUP-IO and COUPLE-INITIAL-V
76
A topic for further work is whether the set of deactivators is larger.
293
resulted in 4 different grammars. These different patterns are expressed in table format
below:
Table 4.8 Grammars discovered by OTSoft
Grammar Ranking of Constraints Output Pattern
1 *SPAN-SYLL, PRES-COUP-IO
>> *SELF-DEACT, COUPLE-
INITIAL-VOWEL
No tone spreading or movement of tones
occurs. Tones deactivate once their
targets are reached.
2 *SPAN-SYLL, COUPLE-INITIAL-
V >> *SELF-DEACT, PRES-
COUP-IO
Tones may move to initial position to
satisfy COUPLE-INITIAL-V, but they
deactivate upon reaching their targets. No
spreading is observed.
3 *SELF-DEACT, PRES-COUP-IO
>> *SPAN-SYLL, COUPLE-
INITIAL-V
Tones do not move to initial position.
Tones stay active until reaching an anti-
phase coupled tone or an utterance
boundary.
4 *SELF-DEACT, COUPLE-
INITIAL-V >> *SPAN-SYLL,
PRES-COUP-IO
Tones both move to initial position where
possible, and stay active until a following
anti-phase coupled tone or utterance
boundary is reached.
5 *SPAN-SYLL, DEP-T >> *SELF-
DEACT
Tones deactivate upon reaching their
targets and no tone gestures are inserted.
6 *SPAN-SYLL, *SELF-DEACT
>>DEP-T
Tones are inserted wherever there is a
toneless syllable in the input. No tone
spreading is observed.
7 *SELF-DEACT, DEP-T >>
*SPAN-SYLL
Tones deactivate only at a following anti-
phase coupled tone or at an utterance
boundary. No tones are inserted on
toneless syllables.
The approach I adopt in this dissertation predicts an asymmetry for left and right
spreading. My approach, as shown via the rankings in OTSoft, predicts that rightward
spreading will occur in an unbounded fashion, stopping at the end of the utterance or at
the onset of an anti-phase coupled tone. Leftward spreading, on the other hand, is
294
predicted to occur only if a tone is drawn to initial position (due to the effects of Couple-
Initial-V). The same could be true for the stressed vowel, where a tone could move to that
position and stay active, although I did not include that constraint in the OTSoft
grammars.
To illustrate how the deactivation constraints *SPAN-SYLL and *SELF-DEACT
interact in additional languages I present four brief case studies. These suggest how my
approach can be applied to cases from languages families outside of Zapotec, and give
examples showing how the constraint typology plays out.
4.3 Case Studies
This section will present four case studies. I examine languages that exhibit
processes of tone spreading that are similar to the patterns presented in chapter 2 for SJZ,
as well as cases where the pattern is reversed and where there is no tone spreading onto
toneless syllables. For Margi, I use data from Pulleyblank (1986). Margi exhibits
rightward spreading from toned syllables onto toneless syllables just like SJZ. I will
argue that in Margi, *SELF-DEACT is ranked above *SPAN-SYLL, and that this accounts
for the rightward spreading behaviors described by Pulleyblank. Margi also exhibits what
looks like leftward spreading from a toned suffix onto a toneless verb root. I will account
for this pattern with a constraint that drives a tone to couple with the initial syllable of the
prosodic word, (COUPLE-INITIAL-VOWEL) and have this outrank the constraint PRES-
COUP-IO. As I discussed in chapter 3, I assume that leftward spreading results in a new
coupling relationship, and that this is a separate process from the one that leads to
rightward spreading in the output.
295
I then discuss data from Chilungu, using data from Bickmore (1996). Chilungu is
a language with unbounded tone spreading. A H tone will spread rightward through any
number of syllables until it reaches the penultimate syllable of the prosodic word. I
assume this occurs due to the effects of *SELF-DEACT, and that only upon reaching a L
boundary tone does the H tone deactivate.
I discuss the ‘neutral tone’ of Chinese dialects and explain how my approach can
account for the variable way in which this tone is realized in Chinese. In Mandarin,
toneless syllables receive their phonetic realization via spreading from the preceding
syllable. This suggests that in the Mandarin dialect at least, *SELF-DEACT is ranked above
*SPAN-SYLL. In the Weinan dialect, however, ‘toneless’ syllables get a default L tone
instead of a tone that depends on the surrounding tone environment. I show that the two
dialects differ in their ranking of constraints that govern deactivation, which is then what
causes their different resolutions of the neutral tone problem.
Finally, I discuss Lithuanian, which is a pitch accent language with no spreading.
I assume that for Lithuanian, stressed syllables are assigned a target H tone which then
deactivates via its own clock and fails to spread.
4.3.1 Margi
This section will present and analyze data from Margi (Pulleyblank 1986). Here I
will first present examples, and then I discuss tableaux with relevant constraints and
candidates. I will discuss the rightward tone spreading first, and then I discuss the effects
of a toned suffix when it combines with a toneless verb stem.
296
Margi is reported to have H and L tones underlyingly. First, notice in the
examples in (5) through (8) that when a verb has a tone gesture underlyingly, and when
the suffix also has a tone gesture, no tone interactions occur.
(5) H-verb stem + H-suffix (Pulleyblank 1986: 72-73)
tá + bá ➝ tábá
‘to cook’ ‘to cook all’
(6) H-verb stem and L-suffix
ná + da# ➝ náda#
‘give’ ‘me’ ‘give me’
(7) L-verb stem and H-suffix
mbu# + 1g2.rí ➝ mbu#1g2.rí
to sew onto ‘to sew onto’
(8) L-verb stem and L-suffix
ptsa# + ya# ➝ ptsa#ya#
roast us ‘roast us’
Examples (5) – (8) show that no tone spreading occurs from the verb stem onto the
suffix. The situation is different, however, when a verb stem does not have a tone gesture
as part of its underlying representation, or when a suffix lacks a tone gesture. First,
consider the cases where a suffix has no underlying tone of its own:
297
(9) H-verb stem and toneless suffix
a) tsá + ri ➝ tsárí
beat ‘to knock at’
b) tá + nya ➝ tányá
cook ‘to use all in cooking’
(10) L-verb and toneless suffix
a) n2# + ri ➝ n2#ri#
say ‘to tell a person’
b) fa#f2# +na ➝ fa#f2#na#
wipe to wipe off
The examples in (9) and (10) show that when the verb root has a tone gesture it spreads
rightward onto the suffix, and that this happens whether the tone in question is H or L.
This is consistent with a model of gestural deactivation that does not deactivate gestures
until they encounter either a following anti-phase coupled tone or an utterance boundary.
I propose that in Margi, *SELF-DEACT is ranked above *SPAN-SYLL.
To illustrate this ranking, I will now present relevant tableaux using the
constraints *SELF-DEACT, *SPAN-SYLL, MAX-T, and DEP-T. Using these constraints, I
show that when there is a suffix with its own tone gesture, spreading does not occur.
Table 4.9 illustrates that when a high-toned syllable combines with a low-toned suffix no
spreading or deletion occurs in Margi:
298
Table 4.9 H verb with L suffix in Margi
ná + da#
H L
MAX-T DEP-T
GEST
*SELF-
DEACT
*SPAN-
SYLL
a. ná dá
H
*! *
☞ b. ná da#
H L
This tableau shows no violations for the winning candidate. If, as shown in candidate a),
the tone of the first syllable does stay active to the point that it causes deletion of the tone
of the second syllable, this will violate MAX-T and the candidate loses.
77
This example
shows that *SELF-DEACT does not come into play in this language when there are no
toneless syllables. The winner is candidate b), where both tones are present, and where
the first tone does not deactivate prior to the onset of the second. I assume the second
tone is deactivated at the end of the utterance. This ranking is very similar to the ranking I
presented in chapter 3 for the SJZ data.
In Table 4.10 I show that when a toneless syllable follows a toned syllable, then
we observe spreading from the root to the suffix.
77
The crucial ranking of MAX-T with the other constraints is illustrated in a following
tableau.
299
Table 4.10 Toneless suffix attaches to toned root in Margi
n2# + ri
L
DEP-T
GEST
*SELF-
DEACT
*SPAN-
SYLL
a. n2# + ri
L
*!
☞ b. n2# + ri#
L
*
c. n2# + ri%
L M
*!
Table 4.10 shows three possible candidates. In a), there is no tone spreading from the
toned verb root onto the toneless following syllable. The tone on the first syllable is thus
deactivated upon reaching its own target. This violates *SELF-DEACT and the candidate
loses. Candidate c) shows a possible solution, where a M tone is inserted on the toneless
syllable. This violates DEP-T and candidate c) loses. The winner is candidate b), where
the tone gesture of the initial syllable stays active until the end of the utterance.
In Margi, a ranking of *SELF-DEACT, DEP-T >> *SPAN-SYLL can account for the
pattern where toneless suffixes acquire their tone via the preceding verb root. This looks
quite similar to the rankings I presented in chapter 3 for SJZ. However, Margi has an
additional complication in that a L or H tone suffix will appear to spread leftward onto a
verb root with no tone. In cases where there is already a tone gesture present in initial
position on the verb root, this coupling will be blocked. However when no tone is present
300
in the verb stem, the coupling will occur and the gesture will remain active until it
encounters either a phrase boundary or another tone gesture. Examples from Pulleyblank
(1986: 72-73) illustrate the pattern:
(11a) Toneless verb stem and H-tone suffix
m2l+ ía m2.lía
make ‘to make someone ready’
1al + bá 1álbá
bite ‘to bite a hole’
(11b) Toneless verb stem and L-tone suffix
h2r + da# h2#rda#
bring me ‘bring me’
sk2 + da# sk2#da#
wait me ‘wait for me’
In (11a) and (11b), I assume that a tone gesture has moved to initial position and coupled
there, and that it will remain active until it encounters another gesture to deactivate it or
the end of the utterance is reached. The constraint COUPLE-INITIAL-VOWEL can drive this
coupling. This constraint must outrank PRES-COUP-IO because the suffix tone will
change its coupling and then stay active.
78
The interaction of these constraints is
illustrated in Table 4.11:
78
If there were roots that are toneless on the second syllable only, I would predict
rightward spreading.
301
Table 4.11 Verb root takes tone of suffix in Margi
m2l+ ía
H
1
DEP-T COUPLE-
INITIAL-
VOWEL
*SELF-
DEACT
PRES-
COUP-IO
*SPAN-
SYLL
a. m2l+ ía
H
1
*!
☞ b. m2.l+ ía
H
1
* *
c. m2.l+ ia
H
1
*!
d. m2.l+ ía
H
2
H
1
*(!) * (!)
As predicted by the factorial typology I presented earlier, this table shows that PRES-
COUP-IO can be violated in some languages. Candidate a) is ruled out because the suffix
tone is not coupled to the vowel of the initial syllable. Candidate b) is the winner, and has
violated PRES-COUP-IO for changing the H tone so that it is coupled to the vowel in the
initial syllable of the verb root. Importantly however, candidate b) also keeps this same
tone active during the suffix syllable due of the effects of *SELF-DEACT. As we can see
via candidate c), a candidate that deactivates this H tone after having coupled it to the
initial syllable will still lose out, since I argue that gestures do not deactivate via their
own clocks in Margi.
302
To show how COUPLE-INITIAL-V interacts with other constraints when there is a
tone gesture already in place on the verb root, I include Table 4.12. Here, I show that
MAX-T must outrank COUPLE-INITIAL-V.
Table 4.12 Verb root does not take suffix tone in Margi
ná + da#
H L
MAX-T COUPLE-
INITIAL-
V
*SELF-
DEACT
PRES-
COUP-IO
*SPAN-
SYLL
a. na#+ da#
L
*! * *
☞ b. ná + da#
H L
*
Table 4.12 shows that ranking MAX-T above the constraint COUPLE-INITIAL-V will
prevent the suffix tone from coupling to the root in cases where the root already has a
tone gesture.
79
As shown in this tableau, the root has a H tone, and the suffix has a L
tone. When the suffix tone couples to the verb root and then stays active, as shown in
candidate a), this incurs violations of MAX-T, PRES-COUP-IO, and *SPAN-SYLL. Thus this
option is not ideal. The winning candidate in b) has a violation of COUPLE-INITIAL-V, but
avoids violating higher-ranked MAX-T. In this candidate, the suffix and root each keep
their original tones.
79
Since there are no contours in Margi, a constraint such as *MULTI-TONE could rule out
formation of a contour tone on the initial syllable.
303
These brief examples have shown that Margi, like SJZ, shows spreading onto
toneless suffixes. I attribute these effects to the constraint *SELF-DEACT, as I have shown
here. Margi differs from SJZ in that a toneless verb root can obtain its tone via a suffix. I
assume this is due to the effects of COUPLE-INITIAL-V, which is ranked above PRES-
COUP-IO in Margi. A ranking lattice is shown in Figure 4.8
Figure 4.8 Ranking Lattice for Margi
MAX-T
COUPLE-INITIAL-V
*SELF –DEACT
DEP-T
PRES-COUP-IO
*SPAN SYLL
4.3.2 Chilungu
The Bantu language Chilungu also has a rightward spreading process that can be
explained using the gestural deactivation model I present in this dissertation. The data I
examine here are from Bickmore (1996). Chilungu has what Bickmore calls “unbounded
tone spreading”, meaning that tones spread all the way up until the L boundary tone of
the final syllable. Bickmore accounts for the pattern using an ALIGN-R constraint. In
cases where a high-toned prefix attaches to a verb root which is underlyingly toneless,
rightward spreading occurs. This is illustrated in (12):
304
(12) Infinitival prefix /kú/ + toneless verb roots (Bickmore 1996: 11)
kú-vúl-a# ‘to be enough’
kú-víímb-a# ‘to thatch’
kú-fúlúmy-a# ‘to boil over’
kú-sáákúl-a# ‘to comb’
kú-sóóbólól-a# ‘to sort out’
The H tone spreads to all syllables except for the last. In accordance with Bickmore
(1996), I assume that a L boundary tone is present on the final syllable, which then
deactivates the H tone gesture of the infinitival prefix. Here, we can say that the H tone of
the infinitival prefix remains active because there is no tone gesture to deactivate it until
the last syllable of the prosodic word.
As I show in the Chilungu data set, a high-toned suffix will spread all the way
through a toneless verb root until it reaches the L boundary tone. The constraints DEP-T,
*SELF-DEACT, and *SPAN-SYLL are ranked such that both tone insertion and self-
deactivation are prohibited in the language. Therefore, since *SPAN-SYLL is ranked
lowest, a tone will remain active until it reaches either the utterance boundary or a
following anti-phase coupled tone. Table 4.13 illustrates how the constraints I propose
are arranged in tableau format for Chilungu:
305
Table 4.13 H tone spread in Chilungu
kú-saakul-a
H
1
L
DEP-T *SELF-
DEACT
*SPANSYLL
a. kú-saakul-a#
H
1
L
*!
☞ b. kú-sáákál- a#
H
1
L
*
c. kú-áákul- a#
H
1
L
*! *
d. kú-sáákúl-a#
H
1
H
2
H
3
L
*!*
In Table 4.13 the input has a high-toned prefix and a toneless verb root. I have marked
the L boundary tone in the input. Notice that candidate a), which has no tone spreading, is
ruled out by *SELF-DEACT. I assume that in this candidate, the H tone of the prefix
deactivates as soon as its target is reached. Candidate c) shows spreading, but in an
unexpected way. Here, the H tone of the prefix has spread onto the following syllable but
no further. This also violates *SELF-DEACT. Candidate d) loses for inserting two
additional H tone gestures. Note that a candidate would also lose if it inserted M tone
gestures, since this would still incur violations of DEP-T. Thus, candidate b) is the
winner, where the H tone gesture of the prefix stays active until it reaches the L boundary
tone gesture.
306
The example candidates shown in Table 4.13 illustrate how the constraints I
develop in this dissertation can account for a language like Chilingu, which despite being
unrelated to SJZ, shows similar patterns of rightward tone spread onto toneless syllables.
Table 4.13 suggests that an appropriate ranking for this language is DEP-T, *SELF-DEACT
>> *SPAN-SYLL. For this language, PRES-COUP-IO does not come into play since no tone
movement is reported. I propose then that PRES-COUP-IO is ranked higher than
constraints that would call for the onset of a tone to move. Although I do not show
examples or analysis for Shona, another Bantu language, it has very similar rightward
spreading patterns which could be accounted for using the same constraints and ranking.
These patterns of rightward spread are discussed in Myers (1990)
80
and
Kenstowicz (1994: 331-338).
4.3.3 Chinese Neutral Tone
In this section, I will present data on Chinese dialects and the ‘neutral tone’ of
Chinese. Yip (2002) points out that there is a certain subset of syllables in Chinese which
never acquire a phonological tone. That is, the syllables in question obtain their output
tone via spreading from the surrounding syllables. She notes that these are mainly affixes,
but also non-initial syllables of some bisyllabic words. For example, the suffix /de/ which
is used to mark possessives has no tone of its own in any context. Perhaps not
surprisingly, Yip notes that in Mandarin, these toneless syllables obtain their tone via the
preceding syllable. This is summarized in Figure 4.9 below:
80
Future work could present analyses of unbounded tone spreading in several languages,
Shona among them.
307
Figure 4.9 Toneless syllables in Mandarin (Yip, 2002: 181)
Preceding Tone Toneless Syllable
55 H high level starts high, then falls
35 LH high rise
81
starts high, then falls, but not as low as after 55
21 L low starts fairly low, then rises
53 HL high fall
82
starts fairly low, and falls even lower
These patterns are all consistent with an analysis where a tone gesture stays active
through a ‘toneless’ syllable. Of course, phonetic or articulatory studies that examine the
onset and offset of tone gestures in these contexts are needed. The first line of the
diagram above shows that a toneless syllable, when following a 55 H level tone is also H,
although it begins to fall during the toneless syllable. This is consistent with a tone
gesture that is coupled to the vowel gesture of the first syllable but that stays active up
through the second (toneless) syllable. Perhaps in Mandarin, following tones deactivate
tone gestures, meaning that *SELF-DEACT is ranked above *SPAN-SYLL. Thus while
previous studies have argued that no phonological tone is assigned to toneless syllables
but that their tones are determined by preceding syllables, my analysis suggests that this
fits into a scenario where the constraints are ranked much like they are in SJZ. In cases
where there is no tone gesture coupled to a particular syllable (the V gesture of that
syllable), the tone gesture of the preceding syllable will stay active and cause what looks
like phonetic spreading in the output.
81
Recall that for Mandarin, Gao (2008) found that LH syllables had two tone gestures
coupled in-phase, but the L tone gesture had a shorter activation duration.
82
For the HL tone, Gao found that the two tone gestures were anti-phase coupled to each
other.
308
First, consider the examples in (13). These show the toneless syllable /de/ as
discussed by Yip. (Examples are from David Li and Xiao He, personal communication)
(13a) wo de mao (13b) wo de gou
LH H LH LH
I poss cat I poss dog
‘my cat’ ‘my dog’
Here, the possessive marker /de/ is between a LH syllable and a H syllable in example
(13a). In (13b) it falls between a LH word and another LH word. In both cases, Yip
points out that the tone of /de/ has a similar realization, which is H. This is consistent
with a model where a tone gesture stays active until a following anti-phase coupled tone.
In both (13a) and (13b), I assume that the preceding H tone stays active until it reaches
the onset of the following tone – in (13a) this is a H tone gesture, while in (13b) it is a L
tone gesture.
To observe how the constraints I propose here would work on Chinese ‘neutral
tone’ syllables, consider Table 4.14. This illustrates that *SELF-DEACT is ranked above
*SPAN-SYLL since a tone will stay active until it reaches another tone gesture.
309
Table 4.14 Mandarin neutral tone syllables
wo#. de go#u.
LH LH
DEP-T
GEST
*SELF-
DEACT
*SPAN-
SYLL
a. wo#. de go#u.
LH LH
*!
☞ b. wo#. dé go#u.
LH LH
*
c. wo#. de% go#u.
LH M LH
*!
Table 4.14 does not show a ranking between DEP-T and *SELF-DEACT, but it does show
that both of these constraints must outrank *SPAN-SYLL in this language. Candidate a)
has deactivated the H tone at the point its target is reached, and thus loses for a violation
of *SELF-DEACT. Candidate c), on the other hand, has inserted a M tone. This violates
DEP-T and the candidate loses. The best solution according to this ranking is candidate
b), where the H tone of the LH rise stays active onto the following toneless syllable and is
deactivated at the onset of the following tone gesture (a LH rise). I propose that this tone
spreading (or failure to deactivate) is a possible explanation of what is occurring in
Mandarin and therefore that the ranking illustrated in Table 4.14 can account for it.
While Mandarin neutral tones get their phonetic or output realization via the tone
of a preceding syllable, other dialects of Chinese differ. For instance, the Weinan dialect
310
of Chinese presented in Li (2007) has a LL realization of neutral tone. Thus the neutral
tone, when it follows LL, LH, HL, and MM, is always realized as LL in this dialect (Li,
2007: 22). This sort of pattern is obtained by ranking *SPAN-SYLL and *SELF-DEACT
above DEP-T, because in this case a tone gesture is failing to stay active through a
toneless syllable. Unfortunately, Li does not give lexical items to illustrate this pattern, so
I will illustrate this ranking in Table 4.15 with sample syllables and tones to show the
ranking that accounts for the patterns he describes in the Weinan dialect.
Table 4.15 L tone realization of toneless syllable in Weinan Chinese
CV + CV
H
*SELF-
DEACT
*SPAN-
SYLL
DEP-T
a. CV + CV
H
* !
☞ b. CV + CV
H L
*
c. CV + CV
H
* !
I assume the input is a high-toned syllable followed by a toneless syllable. As shown,
faithful candidate a) is eliminated under this ranking for a violation of *SELF-DEACT. The
H tone in candidate a) has turned off once its target is reached. In candidate c), a H tone
stays active through the toneless syllable and causes it too to become H. Candidate c) is
311
ruled out by a violation of the constraint *SPAN-SYLL. The winning candidate in this case
is b), where a L tone has been inserted. I assume that for this dialect of Chinese, a gesture
with a L target is inserted on the toneless syllable, and that this violates DEP-T. Perhaps
in this particular dialect, L tone is the unmarked tone. Under this scenario, the initial H
tone is deactivated by this newly inserted L tone. More data would be useful to help show
exactly how the toneless syllable behaves, and why Li analyzes it as toneless.
A question that arises is how we know that this tone is truly toneless and not just
specified L, since it is realized with a L tone in all contexts. Li argues, however, that this
tone really has no target but that the default realization is L as opposed to some other
tone. Perhaps even more intriguing, Li notes that in the Nayong dialect (Guizhou
province), the neutral tone is realized as H, again regardless of the preceding tone (2007:
25). Thus, while more data on this topic would be valuable, it seems that in Chinese
dialects, toneless syllables can be handled by the grammar in one of two ways: either the
toneless syllable acquires its output tone via spreading from a preceding syllable (due to
the effects of *SELF-DEACT), or a tone gesture is inserted (due to the effects of *SPAN-
SYLL). These examples from Chinese have illustrated how the constraints I propose in
this dissertation could account for both types of languages.
4.3.4 Lithuanian Pitch Accent
Accentual languages are another area where the approach I suggest for tone
deactivation can be successfully applied. For some accentual languages the H tone
gestures that are coupled to lexically specified syllables do not spread rightward. This
312
suggests that they are deactivated upon reaching their targets, which suggests a different
ranking of the constraints I present in this dissertation. In cases where a H tone gesture is
active for one syllable and then deactivates, we can assume that *SPAN-SYLL is ranked
above *SELF-DEACT. This is the opposite ranking from languages like SJZ.
Yip notes that accentual languages typically have a lexical contrast between tone
and no tone, with each morpheme having a maximum of one tone or tonal complex
whose location must be lexically specified (2002: 258). She discusses Lithuanian nominal
accent as a tonal system, noting that accented morphemes have a H tone linked to the first
mora, while unaccented morphemes do not. In some morphemes, the tone appears on the
second mora instead of the first. Some examples are shown in (14) below:
(14a) lízdas ‘nest’
H
(14b) kúunas ‘body’
H
(14c) laápas ‘leaf’
H
These examples illustrate that there is only one tone gesture present for each word. Since
no tone spreading occurs onto following syllables, it must be the case that the tone
gestures are deactivating upon reaching their articulatory targets. After Yip (2002) I
assume that the other syllables in each word that are realized with a non-H pitch simply
do not have an underlying tone. Yip even points out that a tonal analysis of these data is
313
convincing, and that “strictly speaking, languages termed accentual are simply a subclass
of tone languages” (2002: 259). She notes that several other languages have similar
patterns, including Tonga, New Chongming Chinese, Serbo-Croation, Swedish, and
Japanese. Under my approach tone deactivation is determined in a language by how the
constraints that govern it are ranked. In accentual languages with no spreading I assume
that the tone of accented syllables is deactivated upon reaching its target due to the
effects of *SPAN-SYLL. In these languages, *SELF-DEACT is violated by winning
candidates. To illustrate in tableau format, I analyze a Lithuanian example in Table 4.16:
Table 4.16 Lithuanian pitch accent
lízdas
H
1
DEP-T *SPAN-
SYLL
*SELF-DEACT
a. lízdás
H
1
H
2
*!
☞ b. lízdas
H
1
*
c. lízdas
H
1
*!
d. lízda%s
H
1
M
*!
314
The input in Table 4.16 shows the lexical item with just one H tone coupled to it. The
other syllable is toneless in the input. In Table 4.16, candidates a) and d) both lose for
having inserted a tone gesture and for having violated DEP-T. Candidate c) is ruled out
by having kept the tone gesture of the first syllable active until the word edge, which
violates *SPAN-SYLL. The winner is shown in b), where the first syllable has its H tone,
and the H tone gesture is then deactivated as soon as its target is reached.
Now that I have shown a few brief case studies illustrating how the constraints
proposed in this dissertation can be applied to other languages, I will discuss
Autosegmental Phonology and consider Optimal Domains Theory (Cassimjee and
Kisseberth 1998). First I present an overview of Autosegmental Phonology and then I
compare and contrast it to my approach.
4.4 Autosegmental Phonology
In this section I discuss Autosegmental Phonology and related approaches to tone.
Then, I comment on how my approach is similar and how it differs. The fundamental
difference between my approach and earlier models is that I propose that the deactivation
or endpoint of tones can be controlled and that this deactivation differs across grammars.
A fundamental insight of autosegmental tonology is that tones are semi-
autonomous from the tone-bearing units on which they are realized. This treatment of
tones as distinct goes back to Leben (1973) and Goldsmith (1976). As I will discuss in
following paragraphs, tones are represented in this model as autosegments that may be
connected to segments or prosodic elements via association lines. The entity to which a
315
tone attaches is known as a tone-bearing-unit (TBU). Rules can then apply to the tones
which insert, delete, flop or spread them. This model views tones as entities in their own
right which reside on a separate tier from the segmental material of an utterance. An
autosegmental treatment of tones in Optimality Theory employs the same
representations, but instead of using a set of rules to account for tone patterns, a set of
constraints operates over the tones to produce the tone patterns of language.
There are two general observations about properties of tones which have
supported an analysis of tones as semi-autonomous from their tone bearing units which I
will present before moving to the basics of Autosegmental Phonology. The first argument
in favor of tones being separate from the segmental material is that there is a non-
isomorphism: more than one tone can link to the same TBU, which results in a contour
tone, and the opposite can also occur, where a tone is associated with more than one
TBU.
Another property of tone that has been discussed as the result of the semi-
autonomy of tones from their TBUs are known as stability effects (Hyman 2007: 494).
That is, when a TBU is deleted, the tone is not necessarily deleted but instead may either
reassociate (appear in a different position) or “float” i.e. remain in the representation but
unassociated and thus affect other tones. In the autosegmental approach and in the
discussion that Hyman brings to floating tones, it is assumed that tones can de-link from
their autosegmental tier and then remain floating in the output to cause downstep or
upstep of other tones.
316
Given these three ways that autosegmental phonology offers an appealing
approach to tones, I will now present the basics of the model laid out by Goldsmith
(1976).
Autosegmental Phonology is based on the work of Goldsmith (1976), who built
upon the work of Leben (1973). In this model, tones occupy a separate tier from the
segmental and prosodic material. Because they reside on this separate tier, tones can only
be realized on the in the output if they are associated with a segment or prosodic unit
such as a syllable or a mora. The associations between syllables and tones are represented
using association lines, which look similar to the representations I used in my tableaux in
chapter 3. For instance, a classic autosegmental representation is found in Figure 4.10:
Figure 4.10 Autosegmental representation
σ σ σ
T T T
The σ symbol stands for syllable, while T can represent any tone. The associations
between the tones and their respective syllables are represented by the vertical lines.
83
In
Figure 4.10 there are three syllables, three tones, and three associations with each tone
linked to its own syllable. Work in Autosegmental Phonology has assumed that in the
underlying representation, tones are not necessarily associated to specific syllables since
83
Although these look similar to the coupling lines I use, autosegmental lines do not
imply in-phase coupling relationships. They represent association between tone and
segment.
317
they can be a property of the morpheme and not any particular syllable.
84
In the cases
where tones are not associated underlyingly, then the grammar must supply these
associations during the course of the derivation.
Autosegmental Phonology analyzes tonal mobility as a change in association. A
change in association is illustrated in Figure 4.11(a) and (b), where the dashed line shows
a new association between tone and segment:
Figure 4.11 Change in association
(a) σ σ σ σ (b) σ σ σ σ
T T
In Figure 4.11, the line with two horizontal marks through it indicates an association has
been eliminated. Therefore in (b), the tone is no longer associated with the first syllable
and is instead associated with the final syllable. In the autosegmental model, there is no
limit on how far a tone can travel, as long as no other tones block it.
Autosegmental Phonology allows for the re-association of tones in cases where
segmental material is deleted. Since tones in the autosegmental model reside on a
separate tier, there is no reason to assume that deletion of segmental material would result
in deletion of a tone. Figure 4.12 illustrates a case where a vowel is deleted and the tone
84
In some languages, associations between tones and syllables must be underlying
because they are lexically distinctive.
318
re-associates as a result. This example shows that a tone may associate to a vowel within
the syllable.
Figure 4.12 Vowel deletion and tone reassociation
CV-VCV CVCV CVCV
T
1
T
2
T
3
T
1
T
2
T
3
T
1
T
2
T
3
This example shows vowel hiatus where the second vowel deletes to avoid an onsetless
syllable. T
2
then re-associates to the first vowel in the sequence which is also associated
with T
1
. The fact that tones are on a separate tier allows for this type of representation
and process to occur. As pointed out by Yip (2002: 75), this type of behavior is rare with
phonological features, with the exception of [nasal]. Usually, when segments delete, any
features associated with them are deleted also.
Another trait of tones that autosegmental representations can capture is the one-
to-many, where one tone spreads onto many syllables. This is handled by the
autosegmental model as multiple association. This is illustrated in Figure 4.13:
Figure 4.13 Multiple association
(a) σ σ σ (b) σ σ σ
T T
319
Here, a tone which was associated with the first syllable has spread onto all of the other
syllables to the right. The autosegmental model handles this by adding association lines.
As I discussed in chapter 3, I propose that tone spreading occurs when a tone gesture
stays active for longer than one syllable. Thus, my model takes a similar approach in that
I assume it is one entity spreading rightward. New coupling relationships are not added
between the tone gesture and each successive syllable where it is active. Instead, it is
coupled to the vowel in the syllable where it is initiated, and then it stays active through
the following syllables and their respective gestures.
The opposite situation, where we have “many-to-one”, is also handled by the
Autosegmental model via association. In this case, more than one tone can be associated
to the same segment. This is illustrated in Figure 4.14 (a) and (b):
Figure 4.14 Many-to-one
(a) σ σ σ (b) σ σ σ
T T T T T T T T
In (a), there is an “extra” tone that is not connected to the segmental material. In (b),
however, it has acquired an association to the final segment, and this syllable now has
two tones associated to it.
85
Not all languages allow contours like this, and very few
allow more than two tones per syllable, although these are attested (see Yip, 2002).
85
Note that my approach can also capture the many-to-one tone structure.
320
Finally, autosegmental representations can account for toneless syllables by
representing them as lacking an association line to any tone. Tones can also exist without
an association to any syllable in the autosegmental model in the output. Such tones are
known in autosegmental terms as floating tones.
It is important to consider the well-formedness conditions that Goldsmith (1976)
proposed for his autosegmental theory of tones in addition to the representations that he
assumes. Here I will briefly present these well-formedness conditions.
Goldsmith (1976) proposed that association of tones to segments could occur as
long as some basic conditions were met. First, he proposed that every tone-bearing unit
must have a tone, and that every tone must be associated to a tone bearing unit. This
effectively prevents floating tones as well as toneless syllables. He also proposed that
association must proceed one-to-one, from left to right. This means that syllables cannot
be skipped, as tones were associated to their respective syllables. Finally, he proposed
that association lines must not cross. A visual representation of cases where these
conditions are both met and violated are shown in Figure 4.15 a) through e):
Figure 4.15 Well-formedness and violations
(a) σ σ σ (b) σ σ σ (c) σ σ σ
T T T T T T T T T
321
(d) σ σ σ (e) σ σ σ
T T T T T T
Out of this data set, only example (a) satisfies all of the well-formedness conditions
proposed by Goldsmith. (b) is an unwanted output because not all TBUs (Tone-Bearing-
Units) have a tone, not all tones have a TBU, and because association has not proceeded
one-to-one from left to right. Example (c) has violations because not all of the tones are
associated to a TBU, and because association has not proceeded one-to-one from left to
right. Example (d) is an unwanted output too, since not every TBU has a tone and
because association has not proceeded one-to-one from left to right. Finally, example (e)
is unwanted because the association lines cross and because association has not
proceeded one-to-one from left to right.
The importance of the association occurring one-to-one from left to right is
evident when we consider cases where there are more syllables than tones. Presented with
this situation, many languages show spreading of the final tone onto the last syllable
which gives a tonal plateau. This is illustrated in Figure 4.16 example (a), below.
Examples (b) and (c) show other possibilities:
322
Figure 4.16 More syllables than tones
(a) σ σ σ (b) σ σ σ (c) σ σ σ
T T T T T T
In (a), the second tone in the sequence has spread rightward onto the final syllable,
creating a tonal plateau. This is illustrated by the additional association line shown in (a).
This satisfies the well-formedness conditions, since all tones are associated to a TBU, all
TBUs have a tone, and association has proceeded from left to right. As I discussed in
chapter 3, I assume that the process illustrated in (a) is precisely what happens at the
gestural level. I assume that in an example like (a), the second tone gesture stays active
through the final syllable since there is no underlying tone gesture in that position. In my
model a case like (b) is possible, depending on how the constraints that govern
deactivation are ranked. I show that examples like (b) are predicted according to the
rankings of the constraints *SELF-DEACT and *SPAN-SYLL. Example (c), also shows a
representation not permitted under the autosegmental well-formedness conditions,
although autosegmental representations could produce a representation like (c) as long as
the well-formedness conditions of Goldsmith were not strictly enforced, for example
through constraint interaction in OT. In this case, association did not proceed one-to-one
from left to right.
Now that I have presented the basic overview of autosegmental phonology and its
basic assumptions, I will discuss how such an approach differs from the one I have
323
presented in chapter 3 for the rightward spreading in the SJZ data. I then compare and
contrast other aspects of my approach with the autosegmental model. First, consider an
example I presented in chapter 2 that shows rightward spreading of a tone onto a toneless
syllable. This is shown in (15a) and (15b)
(15a) tá-ne#llí + a) (15b) tá-ne#llí=a) .
POT-visit=3s.NFML
‘He will visit’
Example (15b) shows that the aspect marker /a)/, which has no underlying tone of
its own, has become H after a verb root that ends in a H tone. The autosegmental rule that
could account for this is shown below in (16). This rule is called Rightward H Spread:
86
(16) Rightward H Spread
[- constricted glottis]
σ + σ]
H
Example (16) shows that a H tone spreads rightward, as indicated by the dashed line. The
[-constricted glottis] means that this rightward H spread will only occur when no glottal
stop intervenes, since as I showed in chapter 2, glottal stops seem to block rightward tone
86
This is similar to a rule proposed by Bickmore and Broadwell (1998), who use
autosegmental rules to analyze the tonal patterns in this language.
324
spreading in SJZ. This rule will apply to H tones that precede a toneless syllable and
cause the toneless syllable to also become H in the output.
Autosegmental rules written this way can therefore describe the tonal spreading,
but as I argue in this chapter, they do not say why such spreading should happen in the
first place. Likewise, OT constraints such as ALIGN-TONE, which calls a specified edge of
a tone span to align with the head or edge of a morphological or prosodic unit (Yip, 2002;
Akinlabi, 1996 ; Wolf, 2005) can model the patterns without saying why a particular tone
should spread. Neither of these approaches explicitly predicts a rightward bias for tone
spreading across languages.
The well-formedness conditions of Goldsmith which state that tone association
should occur from left to right have remained largely unchallenged, since it seems clear
based on many languages that tones do in fact spread rightward far more frequently than
they spread leftward. The tone association direction that Goldsmith proposes does offer a
possible explanation for a rightward bias in tone spreading, although I argue that it is not
as concrete as the approach I present in this dissertation. Goldsmith proposed that tones
must associate from left to right in a word without saying why this must be the case. Thus
the autosegmental approach (or the OT approach that assumes autosegmental
representations) to the tone spreading example I gave in (16) would be to use either a
rightward spreading rule or a tonal alignment constraint. In chapter 3, I argued instead
that tone deactivation is what controls rightward spreading onto either toneless or toned
syllables. In cases where an underlying tone is deleted in favor of a preceding tone that
maintains its activation, a violation of faithfulness is incurred. Here I will not repeat the
325
analysis given in chapter 3, but instead remind the reader that my approach to tone
spreading is different from an autosegmental approach that views spreading as the
addition of association lines. My analysis is that tone spreading results from failure to
deactivate a tone.
Under the gestural representation that I present in this dissertation, the mobility of
tones and their capacity for association and dissociation are compatible with a gestural
representation. The gestural representation assumes that all segments are comprised of
gestures as action units. In the case of a contour tone, two gestures are coupled to the
same vowel gesture (either competitively but in-phase to the vowel, or anti-phase to one
another). This structure was demonstrated by Gao (2008) for Mandarin Chinese. As
discussed in chapter 3, in cases where a tone is realized on more than one vowel, I
propose that this results from a specific deactivation typology for tone gestures, where
tones are deactivated either by the onset of a following anti-phase coupled tone gesture or
by the end of an utterance, and not via their own clock. Tones can behave this way
because a tone gesture can be produced and maintained even while other gestures are
being articulated. I assume that, if driven by higher ranked constraints, a tone can lose its
coupling relationship to one vowel gesture (syllable) and couple in a new location
instead.
I assume that the gestures to which tones are coupled and the tone gestures
themselves are separate entities, so it is possible to delete one and not the other. I
postulate that tones can be independent in this sense because deleting a tone gesture does
not affect the V or C gesture of its syllable. Deletion of a tone gesture would just cause a
326
toneful syllable to become toneless. I argue that the constraint ranking of a particular
language would determine whether or not a tone was deleted (in violation of MAX-T) or
maintained but coupled in a new location (possibly in violation of PRES-COUP-IO).
Regarding the well-formedness restrictions proposed by Goldsmith, I propose that
they too can also be the result of articulatory tone gestures which are coupled in specific
ways to C and V gestures, as I have argued in this dissertation. For instance, if tones are
coupled to the V or C gesture where they are initiated and then stay active until
something deactivates them, it would appear as rightward spreading. I do not argue that
coupling must proceed from left to right, but spreading or activation will necessarily
operate from left to right. This is a given if a tone is active because it will stay on through
all of the intervening syllables until something turns it off. Likewise, as I argue in this
chapter and in chapter 3, leftward spreading is only possible if PRES-COUP-IO is violated
by higher-ranked constraints and the gesture remains active, given the representations
that I assume.
The effect of crossing of association lines is also ruled out in my approach
because I propose that a gesture is active in the same syllable where it is coupled or in a
following syllable due to lengthened activation. A gesture cannot be active before its
onset. According to the approach I adopt in this dissertation, tones can move, but this
requires that a new coupling relationship be established, which will violate the constraint
I have called PRES-COUP-IO.
An interesting difference between my approach and the autosegmental approach
proposed by Goldsmith is that I do allow for the possibility that a given vowel gesture
327
(and thus syllable) may not have a tone gesture coupled to it. This was shown in chapters
2 and 3 where aspect markers are allowed to emerge with no tone. Although they are
realized with a mid-like pitch since there is neither a H nor a L target, I do not assume
that a M tone has to be inserted as the autosegmental model would require. Thus my
approach allows for both toneless syllables, and for rightward spreading from toned to
toneless syllables as I discussed in chapter 3.
An additional difference between the autosegmental model and the approach I
adopt is that I do not allow tones to be uncoupled in the output. I propose that tones must
have coupling relationships in the output in order to exist at all in a spoken utterance.
Although I do not analyze downstep as the result of floating tonal phenomena here, I
hypothesize that when downstep occurs it is due to a gesture that is coupled to some other
gesture in the output.
In summary, gestural representation of tone can be considered to include aspects
of Autosegmental Phonology and elaborate elements of it. Both autosegmental and
gestural approaches assume a relationship between a tone and other segments, whether
this is another gesture or a syllable, or a mora. Whether this relationship is modeled via
association lines or via coupling does not matter. In both cases the onset of the tone is
predicted to occur in the places where the association (or coupling) directs it. The main
difference in the gestural approach is that the end point of the tone can be controlled.
There is no mention of the endpoint of a tone in autosegmental representations, and
spreading is modeled by inserting new association lines. The gestural approach that I
adopt proposes that both the onset and the endpoint of a tone can be controlled. The onset
328
is controlled via coupling, while the offset may be controlled via a constraint that
penalizes deactivation that occurs prior to a following anti-phase coupled tone or an
utterance boundary.
Now that I have briefly discussed autosegmental approaches to tone, I move to a
discussion of an approach called Optimal Domains Theory (Cassimjee and Kisseberth
1998; Cole and Kisseberth 1995)
4.5 Optimal Domains Theory
In this section I will briefly present one other proposal that relates to tone. I
discuss the Optimal Domains Theory (ODT) of Cassimjee and Kisseberth (1998) and
Cole and Kisseberth (1994) as another approach to tone spreading. I argue that ODT fits
less well with the gestural approach I adopt, primarily in cases where tone spreading
occurs over long distances.
4.5.1 Optimal Domains Theory
In this section I give a brief overview of Optimal Domains Theory (ODT)
(Cassimjee and Kisseberth, 1998; Cole & Kisseberth 1994) as an alternate approach to
tone spreading phenomena. In this approach, phonological strings are parceled into
domains. Each domain corresponds to one feature or tone. A segment can belong to many
domains, each of which may or may not be coextensive with the next domain. Domains
specify a string of segments that can express a certain feature, and constraints determine
329
the number and extent of domains as well as whether or not a feature will be expressed in
its domain.
There are two kinds of Alignment constraints that regulate domain size under the
ODT approach. The first is Basic Alignment (BA) which is responsible for building a
domain around a feature or tone’s underlying ‘host’. The host is the segment with which
a feature (or tone) is underlyingly affiliated. The constraints shown in (17) are BA
constraints. Each calls for one edge of a sponsor (the segment that is specified with the
feature F is called the sponsor of F) to be aligned with the same edge of some domain for
the relevant feature (tone):
(17) BA-Left: Align (Sponsor , L; F-domain, L)
BA-Right: Align (Sponsor, R; F-domain, R)
These constraints prohibit spreading or shifting of features and tones. Wide-Scope
Alignment (WSA) is responsible for extending a domain’s reach. WSA constraints are
standard Alignment constraints (McCarthy and Prince, 1993) that require one or the other
edge of a domain to coincide with some edge of a morphological or prosodic category.
When a WSA constraint for one edge of a domain outranks the BA constraint for that
same edge, spreading of the domain occurs. This is illustrated in Table 4.17. Parentheses
show domain edges and the ‘sponsor’ is underlined.
330
Table 4.17 WSA and BA constraints in ODT
/xxxxx/
BA-LEFT ALIGN-
RIGHT
BA-
RIGHT
ALIGN-
LEFT
a. xx(x)xx *!* **
b. (xxx)xx *! **
☞ c. xx(xxx) * **
Here, the x’s represent syllables which are potential hosts of the tone or feature.
Parentheses delimit domains, and the tone or feature itself is indicated via underlining.
BA-Left calls the left edge of the sponsor to be aligned with the left edge of the domain.
This blocks leftward spreading. However, ALIGN-R outranks BA-RIGHT, so spreading the
domain all the way to the right edge of the word is better than limiting the domain to just
its sponsor. This ranking and its outcome is illustrated in Table 4.17. The winning
candidate c) has a feature (or tone) that spreads to the right edge of the word.
Whether spreading or shifting actually occurs will be the result of other
constraints. For example, EXPRESS stipulates that every feature (or tone) in a domain for
the feature (or tone) F expresses F. So in Table 4.17, a candidate like (xx)(x́x́x́) with H
tones on the last three syllables would satisfy EXPRESS and have spreading. Shift would
happen when the constraint *(F,NONHEAD) applies. In this constraint, F could also
represent a tone as *(T, NONHEAD). This constraint bans expression of a domain’s feature
on all elements except for the domain’s head. So in this case, *(H, NONHEAD) would
favor shifting rather than spreading. A candidate like (xx)(x́x́x́) would be suboptimal
because there are two non-heads which have a high tone.
331
The Wide Scope Alignment constraints proposed by Cassimjee and Kisseberth
will not give spreading or shifting by just one unit unless the sponsor is only one unit
away from the domain edge which is specified by the WSA constraints. They posit a
constraint to account for this, given in (18)
(18) *MonoHD: A high-tone domain must not be monomoraic/monosyllabic.
*MONOHD can trigger spreading of a high tone’s domain if the basic alignment
constraints outrank WSA constraints, because extending the domain to two syllables will
escape violation of *MONOHD. By changing the rankings of EXPRESS and *(H,
NONHEAD) the approach can account for whether shifting by one syllable occurs or
spreading by one syllable.
A potential difficulty with ODT is its failure to give evidence for the domains that
are posited by the theory. As Kaplan rightly notes, in languages like Chichewa and
Kikuyu this is less problematic since tones are expressed throughout each domain. But in
languages where tone shift occurs over longer distances, ODT would predict very large
domains.
Yet another potential problem for ODT is the status of floating tones in
particular.
87
This is especially relevant given the floating tone that I examine in chapters
2 and 3. ODT rejects autosegments overtly, and the BA constraints that define where
each domain starts are dependent on those tones being underlyingly associated with their
sponsors. This is similar to the assumption made in the autosegmental model. As Kaplan
87
Cole and Kisseberth (1995) do discuss a floating [+nasal] feature in Terena which can
be analyzed in ODT. However the question of floating tones appears more challenging.
332
(2008: 255) notes, tones can appear on morphemes other than the ones that contributed
them, and the simplest analysis is one that does not require these tones to be underlyingly
linked to their contributing morphemes. In light of the analysis and data I present for SJZ,
this point seems even more important. For the floating tone of SJZ, an analysis that
posited the floating tone to be underlying associated with its morpheme is problematic
because the H tone of the 1 sg morpheme may, as I showed in chapter 2, appear on the
stressed vowel of the verb root, or on the stressed vowel of the verb root and the aspect
marker. The H tone of the 1 sg does not appear on the segmental material of the 1 sg
clitic.
Finally, it is less clear how to analyze contour tones in ODT. As argued by Gao
(2008), contours can be formed by tone gestures that are coupled anti-phase to each other
and to a consonant gesture. In autosegmental terms, contours can form when two tones
are linked to the same TBU. In ODT, however, contours must be formally distinct from
level tones and thus have their own special domains, or syllables with contour tones must
belong to overlapping domains for different tone levels. The problem with a theory where
tone domains overlap is how to determine what would result in a rising vs. a falling tone,
and how to know why the overlapping would not result in a tonal blend of some sort.
Also, if tone contours have their own, special domains, this seems unmotivated because
of the evidence that contours result from the combination of two distinct level tones. For
these reasons, I suggest that a gestural view of tone gestures is better able to capture
spreading patterns than is ODT. While future work should more fully explore the
implications of the model I propose in this dissertation, a gestural approach can more
333
simply account for contour tones and floating tones, and does not require a domain-based
system.
4.6 Future Work
Due to its combination of fieldwork and two theoretical approaches, this
dissertation opens up several exciting avenues for future work which should incorporate
additional data, and insights drawn from both Optimality Theory and Articulatory
Phonology. In this section, I outline areas where my work could be expanded upon, and
discuss Peak Delay Theory (Kaplan 2008) as an augmentation of the proposals I make in
this dissertation.
This dissertation can be extended into future work in several ways: additional
fieldwork on SJZ and related Zapotec dialects is one area where my research can be
applied and expanded, and applications for tone spreading and deactivation constraints is
another area. The incorporation of gestural representations into OT allows for a rich
representation of tone and tonal contrast, and merits future examination in lexical tone
languages.
Future work on SJZ could include an expanded data set with more verbs than I
present in the appendix to this dissertation. I would like to include more verbs and at the
same time incorporate data from additional speakers in order to fully understand how the
tone sandhi processes of the language function. One issue that was raised by the data I
gathered thus far is whether transitivity is the conditioning factor in the appearance of the
floating H tone of the first-person singular. My database supports this analysis, but would
334
be more complete if I had more verbs to provide examples – especially from verb classes
other than 1 (1A). An additional way that this data set could be expanded would be to
compare SJZ data to the related dialect of MacZ, which I also discussed in chapter 3. As I
suggested in chapter 3, it is likely that there are some differences which we could explain
as a slightly different ranking of constraints in the MacZ grammar. More data would be
useful for such a project. Incorporating historical work on Zapotec might also be useful to
determine how a transitive marking for first-person singular developed in these
languages.
88
I would also like to examine more case studies where the tone deactivation
constraints I present in this dissertation can be applied. It would be valuable to develop a
typology of spreading where I present a large number of languages that exhibit tone
spreading. The constraints I have proposed, *SELF-DEACT and *SPAN-SYLL predict only
two languages – those with no spreading and those with unbounded spreading, so it
would be worthwhile to investigate whether all cases that behave differently could be
attributed to Peak Delay, as argued by Kaplan (2008). It would also be interesting to
determine whether language families would show particular patterns (such as Bantu vs.
Otomanguean).
One important difference that my approach makes which merits future
investigation is that I predict a more constrained spreading pattern than the spreading
predicted via OT approaches to tone. As discussed in chapter 3 when I present *SELF-
DEACT and *SPAN-SYLL, my constraints can interact with constraints that call a tone
88
At the moment I know of no other language families where transitivity is marked via
tone.
335
gesture to couple to a prominent position. For example, COUPLE-INTIAL-V can call a tone
to the initial vowel and will result in movement of a tone if PRES-COUP-IO is ranked
lower. A key difference between my approach and others is that mine predicts leftward
tone spreading will occur only when it reaches and is bounded by a prominent position.
In an OT approach, leftward tone spreading up to but not including a prominent position
could be obtained using a gradient left alignment constraint to drive unbounded tone
spreading. This gradient alignment constraint could be dominated by a tone faithfulness
constraint for the prominent position. My approach does not predict this type of leftward
spreading. As shown in this chapter, my approach only predicts movement to a prominent
syllable and then possibly rightward spreading from that position based on the ranking of
*SPAN-SYLL and *SELF-DEACT.
My approach also predicts a different outcome for a sequence such as
[σ σ σ
H
σ σ], where only the antepenultimate syllable has a tone specification. In this
case, a leftward spreading rule or left alignment constraint could drive leftward tone
spreading from a toneful syllable followed by a toneless syllable to give [σ
H
σ
H
σ
H
σ σ].
My approach does not predict the effect of leftward spreading alone in such a case.
COUPLE-INITIAL-VOWEL and *SELF-DEACT could obtain the effect of leftward spreading
from the initial syllable, but it could not drive spreading to stop at the antepenult. Instead,
my approach would yield a pattern where tone spreading would appear to be bidirectional
from the medial syllable – leftward to the initial syllable and unbounded to the right until
reaching a toneful syllable or an utterance boundary i.e. [σ
H
σ
H
σ
H
σ
H
σ
H
]. The failure of
tones to spread at all if they can not go all the way to an utterance boundary or an anti-
336
phase coupled tone is reminiscent of the “sour grapes spreading” discussed in Walker
2010 and originally proposed by Padgett 1995. As presented by Walker 2010, sour grapes
spreading is a type of harmony that must be fully achieved or not at all. In this sense it is
similar to the tone spreading I predict, which will either be all the way to a deactivator or
not occur at all. The more constrained predictions I make for tone spreading leftward to a
prominent vowel or otherwise only rightward merit future analysis to see how they play
out with larger data sets from diverse language families.
I claim that GEN requires tones in the output be anti-phase coupled to one another.
Another area of research should pursue whether this results from the fact that coda
consonant gestures are anti-phase coupled with vowel gestures. Since tones are associated
with vowels, the coupling relationship between tones might fall out from this relationship
between coda consonant gestures and vowel gestures of a syllable. Or, this property of
being anti-phase coupled to one another in the output may be something unique about
tones. It would also be helpful to consider cases of directionality in vowel harmony,
which tend to be leftward rather than rightward (Hansson 2001, Hyman 2002), and
compare the vowel gestures involved in this type of spreading with the tone gestures I
examine in this dissertation.
In this regard, work on Bantu would be useful because Bantu languages
demonstrate many varieties of tone spreading and tone movement. It would also be
valuable to investigate the approach I have developed here in relation to more dialects of
Chinese, preferably with larger data sets than the ones I discuss here. It would be useful
to examine phonetic data from Chinese ‘neutral tone’ syllables to determine whether tone
337
spreading occurs at the phonetic level, and if so, how this might support the theory of
tone deactivation that I present in this dissertation. I would expect to find patterns similar
to those discussed by Yip, but nonetheless experimental verification would lend support
to my ideas. The different coupling graphs I propose for the floating vs. lexical tones in
SJZ also merit instrumental study, because by proposing different coupling structures I
allow for the possibility that there are phonetic (timing) diffences in how these types of
graphs are realized in the output.
The implications of my constraints which penalize coupling relationships between
tones and C, V, and other T gestures need to be more fully explored. I showed in chapter
3 that the ranking *C-T, *T-T- >> *V-T can account for the formation of a falling
contour tone when the floating H of the first-person singular attaches. This raises the
question as to whether other varietes of Zapotec which form a rising tone upon the
addition of a floating tone (as in Coatec Zapotec) would have a different ranking, and
how the factorial typology of these constraints might play out in other languages. Thus in
addition to the factorial typology of *SELF-DEACT and *SPAN-SYLL, future work could
examine predictions and other languages due to different rankings of PRES-COUP-IO,
*C-T, *V-T and *T-T. It is possible that there is some sort of universal preference for
tones to couple with vowel gestures, or that this is a language-specific feature that could
vary.
Another area of investigation is why tone gestures might deactivate differently
from other gestures, and how this affects phonological grammars. It is possible that tones
are fundamentally different from other gestures because they can be produced separately
338
without affecting other articulatory gestures. The physical structures used for tone
production are more difficult to observe, but we can assume that the articulators which
control pitch production are independent in some sense from other parts of the vocal
tract, such as the lips, tongue body, etc. The relationship between laryngeal contrasts and
tonal contrasts is less clear, although a sequencing effect between the laryngeal gestures
and tones has been proposed by Silverman (1997) and is discussed for a dialect of
Zapotec in Tejada (2009).
Another important point to consider is that C and V gestures may differ from each
other in terms of their intrinsic activation periods. It is possible that the intrinsic
activation period of T gestures is more similar to V gestures rather than C gestures, and
that this accounts for the tendency of peak delay. Section 4.6.1 explains Peak Delay
Theory. This question of intrinsic activation periods could mesh very well with PDT and
is another avenue that merits future work.
4.6.1 Peak Delay Theory
Peak Delay is considered by both Kaplan (2008) and Myers (1999). By delay the
authors refer to a tone whose full realization or target is delayed – often until a following
syllable. Myers gives experimental evidence for peak delay, while Kaplan proposes an
OT constraint called PEAK DELAY which takes into consideration the experimental work
completed by Myers.
Myers (1999: 224) gives one explanation for peak delay: “the vocal fold
adjustments that determine F0 modulation are more sluggish than the supralaryngeal
339
gestures that define the syllable”. This implies that executing a high tone’s rise is
inherently more difficult (taking longer) than gestures associated with other phonological
units. In support of this idea, Myers (1999) examines data from Chichewa. Chichewa has
a contrast between H and ∅ (Kanerva 1990, Myers 1999). An illustration of rightward
tone spread in Chichewa is given in (17) to illustrate that H tones spread rightward by
only one syllable that is not one of the last three syllables in a phrase.
(17) a. mitsíkaana ‘girl’
mitsíkána uuyu ‘this girl’
b. chigawée
n
ga ‘terrorist’
chigawé
n
gá iichi ‘this terrorist’
c. zidzábeera ‘they will steal x for/with you’
zidzábéraana ‘they will steal x for each other’
d. mte
n
go wá dee
n
gu ‘price of basket’
mte
n
go wá dé
n
gu iili ‘price of this basket’
mte
n
go wá
n
thíwatiiwa ‘price of this ostrich’
e. tinabá dee
n
gu ‘We stole the basket’
tinabá dé
n
gu iili ‘We stole this basket’
tinabá
n
thíwatiiwa ‘We stole the ostrich
Examples illustrating that no spreading occurs during the last three syllables of a phrase
are found in (18). There are no examples of phrase-final tones because these tones always
retract to the penultimate syllable:
(18) dzíina ‘name’
mtée
n
go ‘tree’
340
Kaplan notes that the important phrasal category here is Kanerva’s (1990) Focal Phrase.
The right edge of this phrase is characterized by the lack of tone spread, by lengthening
of phrase-penultimate vowels, and by the retraction of phrase-final tones to the
penultimate syllable. Kaplan points out that both Myers (1999) and Kanerva (1990)
transcribe retracted tones over the second half of lengthened penultimate vowels, as
shown in (19a) and (19b).
(19) a. mle
n
dó uuyu ‘this visitor’
mleé
n
do ‘visitor’
b. pezá nyaama ‘find the meat!’
peéza ‘find!’
The experimental data collected by Myers suggest that what looks like high tone
spread in Chichewa is due to peak delay. For high-toned syllables from which H could
spread, the tone’s peak delay is normally realized only in the following syllable.
However, the timing of the peak is still correlated with properties of the first syllable. We
would expect that if Chichewa had genuine phonological spread, then peak duration
would be longer (perhaps twice as long) in spreading contexts as compared to non-
spreading contexts. Myers did not find any evidence that this is the case, however. He
identified two properties that might be expected of spreading tones – articulations that are
timed with respect to their new hosts and greater durations compared to non-spreading
tones. Neither property was found for Chichewa, however. Therefore, Myers concludes
that H tones do not actually spread in Chichewa. He argues that the H tone is formally
341
associated only with its original host syllable but that phonetic implementation of the H
target leads to the impression that rightward spreading has occurred.
Myers also gives a good explanation for why there is no spreading from the final
three syllables in a phrase. He argues that spreading from phrase-penultimate syllables is
not reported via transcriptions because these syllables are lengthened. Due to this
lengthening, even when a penultimate syllable’s H tone target has peak delay, the peak is
still contained within its host syllable. Therefore, peak delay will not result in the peak
appearing in a following syllable and transcribers place it in its same syllable.
Kaplan makes the important argument that if Myers’ analysis is correct, there is
no tone spread in Chichewa to be analyzed. The only phonological tone phenomena is
that tones retract from final syllables, and this can be handled as a type of Non-Finality
effect. This case is then one which supports Kaplan’s claim that there is no non-iterative
tone spreading in Chichewa. The Chichewa data, along with Myers’ analysis of them
explain rightward spread by only one syllable as a separate case from rightward spread
due to the effects of *SELF-DEACT.
To account for peak delay and consequent tone shift, Kaplan proposes a constraint
called PEAK DELAY (2008: 228). He defines the constraint as in (20):
(20) PEAK DELAY: The F0 rise or fall for a tone must be allotted an adequate
duration.
This constraint will ensure that an output allots enough time for a tone’s rise or fall in
pitch to be successfully executed. An effect of this constraint is that the onset of the pitch
excursion (the point where the rise or fall in F0 begins) and the F0 peak must be
342
sufficiently separated. If the onset is anchored to the tone’s host syllable, this can produce
peak delay because the F0 peak must be held back until the constraint PEAK DELAY is
satisfied.
Kaplan cites Myers’ linear regression models that formalize the function of
“adequate duration” in this case. Myers (1999: 222) states that peak delay “varies
systematically as a function of syllable duration” and gives a model like the one shown in
(21):
(21) Peak delay = (((-.88P) +l.43) *S) – 3.89
where P=syllable position (0 for medial, 1 for penult) and
S=syllable duration
Kaplan assumes that the constraint PEAK DELAY references a function like the one shown
in (21), and that violations are assigned to candidates whose peak delay is not within
some window around this function’s output. Kaplan then makes the assumption that a
violation is incurred for PEAK DELAY when the peak and onset are contained within the
same light syllable or the same half of a heavy syllable. Importantly, he points out that in
languages where there is no tone spread or shift, PEAK DELAY may be not be ranked high
enough to trigger displacement, or that the separation between onsets and peaks required
by that language’s version of PEAK DELAY may not be long enough to move those
landmarks into neighboring syllables (2008: 230).
Thus Peak Delay and Kaplan’s formalization of the constraint can account for
languages like Chichewa where a tone gesture is perhaps delayed relative to the C or V
gestures of its own syllable, but where no lengthened activation occurs. Peak Delay can
343
account for languages with spreading (or ‘delay’) of just one syllable, while the effects of
*SELF-DEACT can account for languages with unbounded tone spreading. *SPAN-SYLL
accounts for languages with no spreading. Languages with peak delay would not violate
*SPAN-SYLL. Future work should consider the intrinsic activation duration of T gestures
to determine whether tone gestures may stay active longer than C gestures and in that
sense be similar to V gestures, which are also active for a period after their targets have
been reached.
In addition to the evidence and analysis presented by Kaplan and Myers that I
discussed here, work by scholars such as Mu0cke (to appear), shows that the highest pitch
associated with a H pitch accent in Catalan is often not achieved on the stressed syllable
in which it is coupled, but rather on the next syllable. Mu0cke and Grice (2006) found that
for speakers of Vienna German, the target (H) of the prenuclear pitch accent coincides
with the target of the post-accented vowel, and that the target H of the nuclear pitch
accent coincides with the target of the consonant immediately following the accented
syllable. This also shows that the target H is often reached after the syllable in which the
tone gesture itself is coupled.
Future articulatory studies of Zapotec and of languages with pitch accent will be
extremely valuable in discovering the differences between intrinsive deactivation times of
gestures.
344
4.7 Summary
This dissertation presented new data from Sierra Juárez Zapotec and proposed a
new approach to tones combining the representations and coordination structures of
Articulatory Phonology and Optimality Theory. The analysis presented here proposed
that both the onset location (coupling point) as well as the end point or place of
deactivation of a tone gesture can be controlled via the grammar. Several areas of
research are opened up by this approach with much potential for future research and
analysis.
Chapter 1 began with an introduction to three main topics: the language SJZ and
the theories of Articulatory Phonology and Optimality Theory. I presented previous work
on Zapotec that relates to the questions I examine in this dissertation, and suggested that
my work contributes both to the data set available on Zapotec as well as to the theoretical
literature on tone. Chapter 1 laid the groundwork for the theoretical proposals made in
chapter 3 by explaining the current understanding of tone gestures and their deactivation
properties.
Chapter 2 presented several tone processes of interest. First, I examined the
special H tone which is affiliated with the first-person singular and which appears either
on the stressed syllable of the verb root or on the toneless aspect marker in addition to the
stressed syllable of the verb root. Next, I examined processes of rightward spreading in
SJZ that suggest that toneless syllables obtain their output tone via perseveratory
spreading from a previous syllable. I also showed that verb-final H tones spread
rightward onto the clitics regardless of whether these have an underlying tone. This
345
chapter adds much valuable data to the pool of data on Zapotec, and especially to
varieties from the Sierra Juárez region. This dissertation is the first analysis of SJZ tones
to use instrumental observation to complement the transcriptions, and it is also the first to
focus exclusively on an analysis of the tone of the first-person singular via fieldwork with
a native speaker.
Chapter 3 analyzed the tone patterns presented in chapter 2 and presented several
constraints which make reference to articulatory gestures rather than to segments. I
suggest in chapter 3 that floating tones are just tone gestures with no coupling
relationships established in the input, and that these floating tones may obtain their
coupling relationships due to the effects of constraints in the phonological grammar.
Even more important in chapter 3, I present constraints that govern the deactivation of
tone gestures. The most important of these was *SELF-DEACT, which prevents a tone
from deactivating prior to the onset of a following anti-phase coupled tone or an utterance
boundary. As I show in chapter 3 and in chapter 4, these constraints capture the patterns
in SJZ, and their re-ranking allows us to account for languages where toneless syllables
are treated differently. Chapter 3 also makes the important contribution of proposing six
different types of constraints that reference tone gestures. These constraints govern
coupling location, deactivation location, and overlap of gestural activation.
As I suggest in chapter 4, the flexibility afforded by the representation of tones as
gestures rather than just as autosegments provides an excellent way to account for the
rightward bias observed in tone spreading and noted by Hyman (2001, 2007), among
others. Future work on this topic could examine whether these grammatical properties in
346
many languages arise from the nature of tone deactivation. Other areas for future research
include more fieldwork with additional native speakers of SJZ, articulatory analysis of
contour tones in SJZ, and elaborated typologies of the constraint families proposed in this
dissertation. By bringing together the insights obtained via articulatory studies and the
theoretical implications of Articulatory Phonology, this dissertation brings a new
dimension to the analysis of tone in Optimality Theory.
347
References
Akinlabi, A. (1996). Featural affixation. Journal of Linguistics, 32, 239-289.
Avelino, H. (2004). Topics in Yalálag Zapotec, with particular reference to its phonetic
structures. Ph.D. dissertation, University of California, Los Angeles.
Bartholomew, D. (1983). Gramática Zapoteca. In N. Nellis, & J. G. Nellis (Eds.),
Diccionario zapoteco: Zapoteco de Juárez (pp. 337-484). Mexico City: Instituto
Lingüístico de Verano.
Beam de Azcona, Rosemary G. (2004). A Coatlán-Loxicha Zapotec grammar. Ph.D.
dissertation, University of California, Berkeley.
Beckman, J. (1998). Positional faithfulness. Ph.D. dissertation, University of
Massachusetts, Amherst.
Bickmore, L. (1996). Bantu tone spreading and displacement as alignment and minimal
misalignment. ROA # 161-1196,
Bickmore, L., & G.A. Broadwell. (1998). High tone docking in Sierra Juárez Zapotec.
International Journal of American Linguistics, 64(1), 37-67.
Black, C. A. (2000). Quiegolani Zapotec Syntax: A principles and parameters account.
SIL International and University of Texas at Arlington.
Borowsky, T. (2000). Word-faithfulness and the direction of assimilation. The Linguistic
Review, (17), 1-28.
Broadwell, G. A. (2000). Macuiltianguis Zapotec Tone Paradigms. Unpublished
manuscript. SUNY Buffalo.
Browman, C. P., & Goldstein, L. (2000). Competing constraints on intergestural
coordination and self-organization of phonological structures. Bulletin De La
Communication Parlee, 5, 25-34.
Browman, C. P., & Goldstein, L. (1986). Towards an articulatory phonology. Phonology
Yearbook, 3, 219-252.
Browman, C. P., & Goldstein, L. (1988). Some notes on syllable structure in articulatory
phonolgy. Phonetica, 45, 140-155.
Browman, C. P., & Goldstein, L. (1989). Articulatory gestures as phonological units.
Phonology, 6, 201-251.
348
Browman, C. P., & Goldstein, L. (1990). Gestural specification using dynamically-
defined articulatory gestures. Journal of Phonetics, (18), 299-320.
Browman, C. P., & Goldstein, L. (1990). Tiers in articulatory phonology with some
implications for casual speech. In J. Kingston, & M. E. Beckman (Eds.), Papers in
Llaboratory Phonology I : Between the Grammar and Physics of Speech.
Cambridge, UK: Cambridge University Press.
Browman, C. P., & Goldstein, L. (1992). Articulatory phonology: An overview.
Phonetica, 49, 155-180.
Browman, C. P., & Goldstein, L. (1995). Dynamics and articulatory phonology. In T.
Van Gelder, & R. F. Port (Eds.), Mind as motion. Cambridge, MA: MIT Press.
Butler, I. M. (1980). Gramatica Zapoteca: Zapoteco de Yatzachi el Bajo (4th ed.).
Mexico City: Instituto Lingüístico de Verano.
Butler, I. M. (1997). Diccionario Zpoteco de Yatzachi: Yatzachi el Bajo, Yatzachi el Alto,
Oaxaca. Tucson, Arizona: Instituto Lingüístico de Verano.
Byrd, D. (1995). C-centers revisited. Phonetica, (52), 263-282.
Byrd, D. (1996). Influences on articulatory timing in consonant sequences. Journal of
Phonetics, (24), 209-244.
Casali, R. (1997). Vowel elision in hiatus contexts: Which vowel goes? Language, (73),
493-533.
Cassimjee, F. and C.W. Kisseberth (1998). Optimal Domains Theory and Bantu
tononology: A case study from Isixhosa and Shingazidja. In L. Hyman and C.
Kisseberth, (Eds). Theoretical aspects of Bantu tone. Stanford: CSLI.
Clements, G. N. (1981). Akan vowel harmony. Harvard Studies in Phonology, 2, 108-
177.
Clements, G.N. (1990). The role of the sonority cycle in core syllabification. In J.
Kingston, & M. E. Beckman (Eds.), Papers in Laboratory Phonology I : Between
the Grammar and Physics of Speech. Cambridge, UK: Cambridge University Press.
Cole, J. & C. Kisseberth (1994). An Optimal Domains Theory of Harmony. Studies in the
Linguistic Sciences 24 (2): 101-114.
Cole, J. & C. Kisseberth (1995). Restricting multi-level constraint evaluation: Opaque
rule interaction in Yawelmani vowel harmony. In K. Suzuki and D. Elsinga (Eds.)
Proceedings of the South Western Optimality Theory Workshop. Tucson: University
of Arizona Coyote Papers. 18-38.
349
Córdova, Fray. J. d. (1886 [1578a]). Arte del idioma zapoteco. Morelia, Mexico:
Imprenta del Gobierno.
Córdova, F. J. d. (1987 [1578b]). Vocabulario en lengua çapoteca . Mexico: Ediciones
Toledo (INAH [Instituto Nacional de Antropología e Historia]).
Cysouw, M. (2003). The paradigmatic structure of person marking. Oxford: Oxford
University Press.
de Lacy, P. (2002). The Interaction of Tone and Stress in Optimality Theory. Phonology
19: 1-32.
de Lacy, P. (2007). The Interaction of tone, sonority, and prosodic structure. In Paul de
Lacy (Ed.) The Cambridge Handbook of Phonology. Cambridge: Cambridge
University Press, Ch. 12 (pp. 281-308).
Downing, Laura J. (2009). Optimality Theory and African Language Phonology. In
Selected Proceedings of the 38
th
Annual Conference on African Linguistics. Ed.
Masangu Matondo, Fiona McLaughlin, and Eric Potsdam, 1-16. Somerville, MA:
Cascadilla Proceedings Project.
Foreman, J. (2006). The morphosyntax of subjects in Macuiltianguis Zapotec. Ph.D
dissertation. University of California, Los Angeles.
Futgai, Y. (1997). Root-reduplicant Faithfulness. Proceedings of WCCFL 16.
Galant, M. R. (1998). Comparative constructions in Spanish and San Lucas Quiaviní
Zapotec. Ph.D. dissertation, University of California, Los Angeles.
Gao, M. (2008). Mandarin tones: An articulatory phonology account. Ph.D. dissertation,
Yale University.
Gibbs, W. P. (1977). Discourse elements in Sierra Juarez Zapotec. Unpublished M.A.,
University of Texas at Arlington,
Goldsmith, J. (1976). Autosegmental phonology. Indiana University Linguistics Club.
Goldstein, L. (2011). Back to the past tense in English. In Gutiérrez-Bravo, R.,
Mikkelsen, L., & Potsdam, E (Eds.), Representing language: Essays in honor of
Judith Aissen. (pp. 69-88.) Santa Cruz, CA: Linguistics Research Center.
Goldstein, L., Byrd, D., & Saltzman, E. (2006). The role of vocal tract gestural action
units in understanding the evolution of phonology. In M. A. Arbib (Ed.), Action to
language via the mirror neuron system (pp. 215-249). Cambridge: Cambridge
University Press.
350
Gordon, M. (2001). A typology of contour tone restrictions. Studies in Language 25:405-
44.
Gordon, R. G. (2005). In Gordon R. G. (Ed.), Ethnologue: Languages of the world (15th
ed.). Dallas, Texas: SIL International.
Greenberg, J. H. (1957). Essays in linguistics. Chicago: University of Chicago Press.
Greenberg, J.H. (1988). The first person dual as an ambiguous category. Studies in
Language 12. 1-18.
Grimes, B. F. (Ed.). (1996). Ethnologue. Dallas, TX: Summer Institute of Linguistics.
Hansson, G. O. (2001). Theoretical and typological issues in consonant harmony. Ph.D
dissertation, University of California, Berkeley,
Hayes, B., Tesar, B., & Zuraw, K. (2003). OTSoft 2.1
Hyman, L. M. (2002). Is there a right-to-left bias in vowel harmony? 9
th
International
Phonology Meeting.Vienna. 1-36.
Hyman, L. M. (2008). Directional asymmetries in the morphology and phonology of
words, with special reference to Bantu. Linguistics, 46(2), 309-350.
Hyman, L. (2001). On the limits of phonetic determinism in phonology: *NC revisited. In
E. Hume, & K. Johnson (Eds.), The role of speech perception phenomena in
phonology (pp. 141-185). San Diego: Academic Press.
Hyman, L. (2007). Tone: Is it different? In J. Goldsmith, J. Riggle and A. Yu (Eds.), The
Handbook of Phonological Theory 2
nd
Ed. (pp. 483-528) Blackwell.
Ito, J., & Mester, A. (1999). Realignment. In R. Kager, H. van der Hulst & W. Zonneveld
(Eds.), The prosody-morphology interface (pp. 217). Cambridge: Cambridge
University Press.
Jakobson, R., G. Fant & M. Halle. (1963). Preliminaries to Speech Analysis. MIT Press.
Javkin, H. (1979). Phonetic universals and phonological change. Report of the Phonology
Laboratory, (4) UC Berkeley.
Jesney, K., & Tessier, A. (2011). Biases in harmonic grammar: The road to restrictive
learning. Natural Language and Linguistic Theory 29: 251-290.
Jiang-King. P. (1996). An Optimality Account of Tone-Vowel Interaction in Northern
Min. Ph.D dissertation: Unversity of British Columbia.
351
Kanerva, J. M. (1990). Focus and phrasing in chichewa phonology. New York: Garland.
Kaplan, A. (2008). Noniterativity is an emergent property of grammar. Ph.D dissertation.,
University of California, Santa Cruz.
Kawahara, S. (2008). On the proper treatment of non-crisp edges. Japanese/ Korean
Linguistics, 13: 55-67.
Keating, P. (1988). Underspecification in phonetics. Phonology 5: 275-92.
Kenstowicz, M. (1994). Phonology in generative grammar. Malden, MA: Blackwell
Publishing.
Kingston, J. (2003). Mechanisms of tone reversal. In S. Kaji (Ed.), Cross-linguistic
studies of tonal phenomena (pp. 57-120). Tokyo: ILCAA.
Kiparsky, P. (1981). Remarks on the metrical structure of the syllable. In W. V. Dressler,
O. E. Pfeiffer & J. E. Rennison (Eds.), Phonologica 1980 Innusbruck.
Kubozono, H. (2010). Accentuation of alphabetic acronyms in varieties of Japanese.
Lingua, 120, 2323-2335.
Leben, W. (1973). Suprasegmental phonology. Indiana University Linguistics Club.
Lee, F. (1999). Antisymmetry and the syntax of San Lucas Quiaviní Zapotec. Ph.D.
dissertation, University of California, Los Angeles.
Li, D., & He, Xiao. (2011). Personal Communication.
Li, M. (2007). Neutral tones in disyllabic sequences across Chinese dialects: An OT
account. ROA 854,
Long C., R., & Cruz M., S. (1999). Diccionario zapoteco de San Bartolomé Zoogocho
Oaxaca (1.th ed.). Coyoacán, D.F.: Instituto Lingüístico de Verano.
Long, C., Cruz, R., & Cruz, S. M. (1999). Diccionario Zapoteco de San Bartolomé
Zoogocho Oaxaca. Coyoacan, D.F.: Instituto Lingüístico de Verano.
López, L., Newberg, R., & Newburg, F. (1990). La conjugación del verbo zapoteco.
Mexico: Instituto Lingüístico de Verano.
Marks, D. (1976). Zapotec verb morphology: Categories and tonomechanics with special
attention to Sierra Juárez Zapotec. (M.A., University of Texas at Arlington).
McCarthy, J. (1979). Formal problems in Semitic phonology and morphology. Ph.D,
dissertation .Massachusetts Institute of Technology.
352
McCarthy, J. (1981). A prosodic theory of nonconcatenative morphology. Linguistic
Inquiry, (10), 443-465.
McCarthy, J. (2002). A thematic guide to Optimality Theory. Cambridge: Cambridge
University Press.
McCarthy, J., & Prince, A. (1993). Generalized Alignment. In G. Booij and J. van Marle,
(Eds.) Yearbook of Morphology 1993(pp. 79-153). Dordrecht: Kluwer.
McCarthy, J., & Prince, A. (1995). Faithfulness and reduplicative identity. In J.
Beckman, L. W. Dickey & S. Urbanczyk (Eds.), Papers in optimality theory.
university of massachusetts occasional papers (pp. 249-384). Amherst, MA:
Graduate Linguistic Student Association.
Mu0cke, D., Nam, H., Hermes, A., & Goldstein, L. (to appear). Coupling of tone and
constriction gestures in pitch accents. In Hoole, Bombien, Pouplier, Mooshammer &
Kuehnert. (Eds.), To appear.
Mu0cke, D. & Grice, M. (2006). Articulatory anchors for tonal targets in rising pitch
accents. Poster presentation at Laboratory Phonology 10, Paris.
Munro, P., & Lopez, F. (1999). Di'csyonaary XːTee'n dii'zh sah Sann Lu'uc. Los
Angeles: UCLA Chicano Studies Research Center.
Myers, S. (1990). Tone and the structure of words in Shona. New York: Garland Press.
Myers, S. (1999). Tone association F0 timing in Chichewa. Studies in African
Linguistics, 28(2), 215-239.
Myers, S. (2003). F0 timing in kinyarwanda. Phonetica, 60(2), 71-97.
Nam, H., Goldstein, L., & Saltzman, E. (2005). A coupled oscillator model of
intergestural timing within syllables. Presentation at the 150th Meeting of the
Acoustical Society of America.
Nellis, N., & Nellis, J. G. (1983). Diccionario zapoteco de Juárez. Mexico: Instituto
Lingüístico de Verano.
Noyer, R. (1992.) Features, positions, and affixes in an autonomous morphological
structure. Ph.D. Dissertation MIT.
Ohala, J. J. (1990). The phonetics and phonology of aspects of assimilation. In J.
Kingston, & M. Beckman (Eds.), Papers in laboratory phonology I : Between the
grammar and physics of speech. Cambridge: Cambridge University Press.
353
Padgett, J. (1995). Partial class behavior and nasal place assimilation. In K. Suzuki and
D. Elsinga (Eds.) Proceedings of the South Western Optimality Theory Workshop.
Tucson: University of Arizona Coyote Papers. 145-183
Pater, J. (2009). Morpheme-specific phonology: Constraint indexation and inconsistency
resolution. Phonological argumentation: Essays on evidence and motivation (pp.
123-154). London: Equinox.
Pickett, V. B. (1959). Vocabulario zapoteco del istmo: Castellano-zapoteco, zapoteco-
castellano (3rd ed.). Mexico: Instituto Lingüístico de Verano.
Pickett, V. B., Black, C. A., & Marcial, V. C. (1998). Gramática popular del zapoteco
del istmo. Tucson, Arizona: Instituto Lingüístico de Verano.
Pickett, V., Black, C., & Marcial Cerqueda, V. (1998). Gramatica popular del zapoteco
del istmo (1. en español ed.). Juchitán, México; Tucson, AZ: Centro de Investigación
y Desarrollo Binnizá; Instituto Lingüístico de Verano.
Potts, C., Pater, J., Jesney, K., Bhatt, R., & Becker, M. (2010). Harmonic grammar with
linear programming: From linear systems to linguistic typology. Phonology, 27(1),
77-117.
Prince, A., & Smolensky, P. (1993). Optimality theory: Constraint interaction in
generative grammar. Technical Reports of the Rutgers University Center for
Cognitive Science TR-2.,
Prince, A., & Smolensky, P. (2004). Optimality theory: Constraint interaction in
generative grammar. Oxford: Blackwell.
Pulleyblank, D. (1986). Tone in lexical phonology. Dordrecht: D. Reidel Publishing
Company.
Saltzman, E. (1986). Task dynamic co-ordination of the speech articulators: A
preliminary model. In H. Heuer, & C. Fromm (Eds.), Generation and modulation of
action patterns (pp. 129-144). Berlin: Springer-Verlag.
Saltzman, E., & Kelso, J. A. S. (1987). Skilled actions: A task dynamic approach.
Psychology Review, 94, 84-106.
Saltzman, E. L. (1986). Task dynamic coordination of the speech articulators: A
preliminary model-generation and modulation of action patterns. In H. Heuer, & C.
Fromm (Eds.), Experimental brain research (pp. 129-144). New York: Springer-
Verlag.
354
Saltzman, E. L., & Munhall, K. G. (1989). A dynamical approach to gestural patterning
in speech production. Ecological Psychology, 1, 333-382.
Siewierska, A. (2004). Person. Cambridge: Cambridge University Press.
Silverman, D. (1997). Laryngeal complexity in Otomangeaun vowels. Phonology 14:
235-262.
Smith, J. (2002). Phonological Augmentation in Prominent Positions. Ph.D. dissertation.
University of Massachussetts, Amherst.
Sonnenschein, A. H. (2004). A descriptive grammar of Zoogocho Zapotec on a
typological basis. Ph.D. dissertation, University of Southern California.
Steriade, D. (2001). Directional asymmetries in place assimilation: A perceptual account.
In E. Hume, & K. Johnson (Eds.), The role of speech perception in phonology. (pp.
219-250). New York: Academic Press.
Stubblefied, M., & Stubblefied, C. (1991). Diccionario zapoteco de Mitla. Mexico:
Instituto Lingüístico de Verano.
Stubblefield, M., & Hollenbach, E. E. d. (1991). Gramática zapoteca: Zapoteco de mitla,
oaxaca. Diccionario zapoteco de mitla (31st ed., pp. 191-248). Mexico, D.F.:
Instituto Lingüístico de Verano.
Tejada, L. (2009). Phonation Categories in Cajonos Zapotec. 157
th
Meeting of the
Acoustical Society of America. Portland, Oregon.
Tilsen, S. (2009). Multitimescale dynamical interactions between speech rhythm and
gesture. Cognitive Science 33. 839-879.
Tilsen, S. (2011). Effects of syllable stress on articulatory planning observed in a stop-
signal experiment. Journal of Phonetics.
Urbanczyk, S. (1996). Patterns of Reduplication in Lushootseed. Ph.D. Dissertation.
University of Massachusetts, Amherst.
Walker, R. (2001). Round licensing, harmony, and bisyllabic triggers in Altaic. Natural
Language and Linguistic Theory, (19), 827-878.
Walker, R (2010). Nonmyopic harmony and the nature of derivations. Linguistic Inquiry
41: 169-179.
Walker, R. (2011). Vowel patterns in language. Cambridge University Press.
355
Wechsler, S. 2010. What ‘You’ and ‘I mean to each other: Person Indexicals, Self-
Ascription, and Theory of Mind. Language: 86:2. 332-365.
Wolf, M. (2007) For an autosegmental theory of mutation. MS, University of
Massachusetts, Amherst.
Yip, M. (2002). Tone. Cambridge: Cambridge University Press.
Zhang, J. (2004). The role of contrast-specific and language-specific phonetics in contour
tone distribution. In Hayes, Bruce, Kirchner, Robert, Kirchner, Martin, and Donca
Steriade (Ed.), Phonetically based phonology (pp. 157-190) Cambridge University
Press.
Zoll, C. C. (1996). Parsing below the segment in a constraint based framework. Ph.D.
dissertation, University of California, Berkeley.
Zoll, C.C. (2003). Optimal Tone Mapping. Linguistic Inquiry 34:2 :225-268.
356
Appendix: SJZ Verb Reference
Zapotec root English Spanish Class Page
/&a%tta%/ ‘iron’ ‘planchar’ 1 (I-A) 358
/ka#$ná/ ‘leave’ ‘dejar’ 1 (I-A) 359
/+i#a#/ ‘close’ ‘cerrar’ 1 (I-A) 360
/ni + bíá$/ ‘know’ ‘conocer’ 1 (I-A) 361
/di + tsa#$/ ‘fill’ ‘llenar’ 1 (I-A) 362
/di#a# + lo!"/ ‘cross oneself’ ‘persignarse’ 1 (I-A) 363
/&a%tta%/ ‘scrunch up’ ‘abollar’ 1 (I-A) 364
/di + ba#gá$/ ‘cover up’ ‘cubrir’ 1 (I-A) 365
/det(u%/ ‘fold, twist’ ‘doblar’ 1 (I-A) 366
/di#a#/ ‘write’ ‘escribir’ 1 (I-A) 367
/da%" + na%"ga%$/ ‘listen’ ‘escuchar’ 1 (I-A) 368
/di +bi%"tsi%/ ‘dry’ ‘secar’ 1 (I-A) 369
__________________________________________________________________________________________________________________
/e%$/ ‘hit, strike’ ‘pegar’ 2 (I-B) 370
/a$ní/ ‘haul, carry’ ‘llevar, cargar’ 2 (I-B) 371
__________________________________________________________________________________________________________________
/lá$ + ni/ ‘see something’ ‘ver algo’ 3 (II-A) 372
__________________________________________________________________________________________________________________
/bi&i%/ ‘fall down’ ‘caerse’ 4 (II-B) 373
/ja%/ ‘drink’ ‘beber’ 4 (II-B) 374
/láni%/ ‘hug’ ‘abrazar’ 4 (II-B) 375
/la%tsu#$/ ‘tear up, pull out’ ‘arrancar’ 4 (II-B) 376
__________________________________________________________________________________________________________________
/júnna%/ ‘return something’ ‘devolver’ 5 (II-C) 377
/kka%bí/ ‘answer’ ‘contestar’ 5 (II-C) 378
/ba%ni%/ ‘awaken’ ‘despertarse’ 5 (II-C) 379
357
Zapotec root English Spanish Class Page
/ a%"ní/ ‘dig’ ‘cavar’ 6 (III) 380
/kába#/ ‘carve (wood)’ ‘labrar’ 6 (III) 381
/a##"bi#/ ‘paint’ ‘pintar’ 6 (III) 382
/edá/ ‘wait for’ ‘esperar’ 6 (III) 383
/o#a#/ ‘take away’ ‘quitar’ 6 (III) 384
__________________________________________________________________________________________________________________
/ada%/ ‘sow crops’ ‘sembrar’ 7 (IV-A) 385
/kku#$ + 'o!"/ ‘get dressed’ ‘vestirse’ 7 (IV-A) 386
/++a%/ ‘get up’ ‘levantarse’ 7 (IV-A) 387
__________________________________________________________________________________________________________________
/a#$a%/ ‘enter’ ‘entrar’ 8 (IV-B) 388
/o/ ‘eat’ ‘comer’ 8 (IV-B) 389
/a%"'i#/ ‘call’ ‘llamar’ 8 (IV-B) 390
/á++i%/ ‘sleep’ ‘dormir’ 8 (IV-B) 391
__________________________________________________________________________________________________________________
/ía/ ‘go’ ‘ir’ 9 (V-A) 392
/a##"ji% / ‘burn oneself’ ‘quemarse’ 9 (V-A) 393
/appi%/ ‘climb up’ ‘subirse’ 9 (V-A) 394
__________________________________________________________________________________________________________________
/á$/ ‘strike, beat’ ‘azotar’ 10 (V-B) 395
/o$o%/ ‘buy’ ‘comprar’ 10 (V-B) 396
__________________________________________________________________________________________________________________
/a% + la%tsi#$/ ‘love’ ‘querer’ 11 (VI) 397
/a/ ‘be’ ‘ser’ 11 (VI) 398
__________________________________________________________________________________________________________________
/ne#llé/ ‘visit’ ‘visitar’ 12 (VII) 399
/&ákka#$/ ‘find unexpectedly’ ‘hallar’ 12 (VII) 400
358
Class: 1 (1A)
English: ‘to iron’
Spanish: ‘planchar’
Posited verb root: /&a%tta%/
Aspect markers: /ú/, /be%/, /ru/
Nellis and Nellis page: 109
POTENTIAL COMPLETIVE HABITUAL
1 S ú-&átta%=$ be%-&átta%=$ rú-&átta%=$
2 S INFORMAL ú-&a%tta%=lu%$ be%-&a%tta%=lu%$ ru-&a%tta%=lu%$
2 S FORMAL ú-&a%tta%=é be%-&a%tta%=é ru-&a%tta%=é
3 S FAMILIAR ú-&a%tta%=bí be%-&a%tta%=bí ru-&a%tta%=bí
3 S FORMAL ú-&a%tta%=é be%-&a%tta%=é ru-&a%tta%=é
3 S NON-FORMAL ú-&a%tta) % be%-&a%tta) % ru-&a%tta) %
1 P INCLUSIVE ú-&a%tta%=rí$u% be%-&a%tta%=rí$u% ru-&a%tta%=rí$u%
1 P EXCLUSIVE ú-&a%tta%=tu%$ be%-&a%tta%=tu%$ ru-&a%tta%=tu%$
2 P INFORMAL ú-&a%tta%=lé be%-&a%tta%=lé ru-&a%tta%=lé
2 P FORMAL ú-&a%tta%=lé be%-&a%tta%=lé ru-&a%tta%=lé
3
P FAMILIAR ú-&a%tta%=ga%bí be%-&a%tta%=ga%bí ru-&a%tta%=ga%bí
3 P FORMAL ú-&a%tta%=gé be%-&a%tta%=gé ru-&a%tta%=gé
3 P NON-FORMAL ú-&a%tta%=ga) # be%-&a%tta%=ga) # ru-&a%tta%=ga) #
359
Class: 1 (1A)
English: ‘to leave behind’
Spanish: ‘dejar’
Posited verb root: /ka#$ná/
Aspect markers: /ú/, /be%/, /ru/
Nellis and Nellis page: 80
POTENTIAL COMPLETIVE HABITUAL
1 S ú-ká$ná=$ be%-ká$ná=$ rú-ká$ná=$
2 S INFORMAL ú-ka#$ná=lú$ be%-ka#$ná=lú$ ru-ka#$ná=lú$
2 S FORMAL ú-ka#$ná=e be%-ka#$ná=e ru-ka#$ná=e
3 S FAMILIAR ú-ka#$ná=bí be%-ka#$ná=bí ru-ka#$ná=bí
3 S FORMAL ú-ka#$ná=e be%-ka#$ná=e ru-ka#$ná=e
3 S NON-FORMAL ú-ka#$na) . be%-ka#$na) . ru-ka#$na) .
1 P INCLUSIVE ú-ka#$ná=rí$u% be%-ka#$ná=rí$u% ru-ka#$ná=rí$u%
1 P EXCLUSIVE ú-ka#$ná=tú$ be%-ka#$ná=tú$ ru-ka#$ná=tú$
2 P INFORMAL ú-ka#$ná=lé be%-ka#$ná=lé ru-ka#$ná=lé
2 P FORMAL ú-ka#$ná=lé be%-ka#$ná=lé ru-ka#$ná=lé
3
P FAMILIAR ú-ka#$ná=gábí be%-ka#$ná=gábí ru-ka#$ná=gábí
3 P FORMAL ú-ka#$ná=gé be%-ka#$ná=gé ru-ka#$ná=gé
3 P NON-FORMAL ú-ka#$ná=ga) . be%-ka#$ná=ga) . ru-ka#$ná=ga) .
360
Class: 1 (1A)
English: ‘to close
Spanish: ‘cerrar’
Posited verb root: / +i#a#/
Aspect markers: /ú/, /be%/, /ru/
Nellis and Nellis page: 105
POTENTIAL COMPLETIVE HABITUAL
1 S ú-+ía#=$ be%-+ía#=$ rú-+ía#=$
2 S INFORMAL ú-+i#a#=lu%$ be%-+i#a#=lu%$ ru-+i#a#=lu%$
2 S FORMAL ú-+i#=é be%-+i#=é ru-+i#=é
3 S FAMILIAR ú-+i#a#=bí be%-+i#a#=bí ru-+i#a#=bí
3 S FORMAL ú-+i#=é be%-+i#=é ru-+i#=é
3 S NON-FORMAL ú-+i#a) # be%-+i#a
# ) ru-+i#a) #
1 P INCLUSIVE ú-+i#a#=rí$u% be%-+i#a#=rí$u% ru-+i#a#=rí$u%
1 P EXCLUSIVE ú-+i#a#=tu%$ be%-+i#a#=tu%$ ru-+i#a#=tu%$
2 P INFORMAL ú-+i#a#=lé be%-+i#a#=lé ru-+i#a#=lé
2 P FORMAL ú-+i#a#=lé be%-+i#a#=lé ru-+i#a#=lé
3 P FAMILIAR ú-+i#a#=ga%bí be%-+i#a#=ga%bí ru-+i#a#=ga%bí
3
P FORMAL ú-+i#a#=gé be%-+i#a#=gé ru-+i#a#=gé
3 P NON-FORMAL ú-+i#a#=ga) # be%-+i#a#=ga) # ru-+i#a#=ga) #
361
Class: 1 (1A)
English: ‘to know/ be familiar with a subject'
Spanish: ‘conocer’
Posited verb root: / ni + bíá$ /
Aspect markers: /gú/, /be%/, /ná/
Nellis and Nellis page: 96
POTENTIAL COMPLETIVE HABITUAL
1 S gú-níbí$=a#$ bé-níbí$=a#$ nábíá$=te#$
2 S INFORMAL gú-níbíá$=lu%$ be%-ni%bíá$=lu%$ nábíá$ni=lu%$
2 S FORMAL gú-níbí$=e. be%-ni%bí$=é nábíá$ni=é
3 S FAMILIAR gú-ní-bíá$=bí be%-ni%bíá$=bí nábíá$ni=bí
3 S FORMAL gú-níbí$=é be%-ni%bí$=é nábíá$ni=é
3 S NON-FORMAL gú-níbí$=a)) be%-ni%bí$=a) nábíá$ni=a)
1 P INCLUSIVE gú-níbíá$=rí$u% be%-ni%bíá$=rí$u% nábíá$ni=rí$u%
1 P EXCLUSIVE gú-níbíá$=tu%$ be%-ni%bíá$=tu%$ nábíá$ni=tu%$
2 P INFORMAL gú-níbíá$=lé be%-ni%bíá$=lé nábíá$ni=lé
2 P FORMAL gú-níbíá$=lé be%-ni%bíá$=lé nábíá$ni=lé
3 P FAMILIAR gú-níbíá$=ga%bí be%-ni%bíá$=ga%bí nábíá$ni=ga%bí
3 P FORMAL gú-níbíá$=gé be%-ni%bíá$=gé nábíá$ni=gé
3 P NON-FORMAL gú-níbíá$=ga) # be%-ni%bíá$=ga) # nábíá$ni=ga) #
Notes: This verb is irregular in that the /ni/ comes before /bíá$/ in the potential and
completive, but after it in the habitual.
362
Class: 1 (1A)
English: ‘to fill’
Spanish: ‘llenar’
Posited verb root: / di + tsa#$ /
Aspect markers: /gú/, /be%/, /ru/
Nellis and Nellis page: 87
POTENTIAL COMPLETIVE HABITUAL
1 S gú-dí-tsá$=a#$ be%-di%-tsá$=a#$ rú-dí-tsá$=a#$
2 S INFORMAL gú-dí-tsa#$=lu%$ be%-di%%-tsa#$=lu%$ ru-di-tsa#$=lu%$
2 S FORMAL gú-dí-tse#$=é be%-di%-tse#$=é ru-di-tse#$=é
3 S FAMILIAR gú-dí-tsa#$=bí be%-di%-tsa#$=bí ru-di-tsa#$=bí
3 S FORMAL gú-dí-tse#$=é be%-di%-tse#$=é ru-di-tse#$=é
3 S NON-FORMAL gú-dí-tsa#$=a) be%-di%-tsa#$=a) ru-di-tsa#$=a)
1 P INCLUSIVE gú-dí-tsa#$=rí$u% be%-di%-tsa#$=rí$u% ru-di-tsa#$=rí$u%
1 P EXCLUSIVE gú-dí-tsa#$=tu%$ be%-di%-tsa#$=tu%$ ru-di-tsa#$=tu%$
2 P INFORMAL gú-dí-tsa#$=lé be%-di%-tsa#$=lé ru-di-tsa#$=lé
2 P FORMAL gú-dí-tsa#$=lé be%-di%-tsa#$=lé ru-di-tsa#$=lé
3 P FAMILIAR gú-dí-tsa#$=ga%bí be%-di%-tsa#$=ga%bí ru-di-tsa#$=ga%bí
3 P FORMAL gú-dí-tsa#$=gé be%-di%-tsa#$=gé ru-di-tsa#$=gé
3 P NON-FORMAL gú-dí-tsa#$=ga) # be%-di%-tsa#$=ga) # ru-di-tsa#$=ga) #
Notes: This verb includes the causative /di/.
363
Class: 1 (1A)
English: ‘to cross onself’
Spanish: ‘persignarse’
Posited verb root: / di#a# + lo!" /
Aspect markers: /gú/, /be%/, /ru/
Nellis and Nellis page: 86
POTENTIAL COMPLETIVE HABITUAL
1 S gú-día#-loá=$ be%-día#-loá=$ rú-día#-loá=$
2 S INFORMAL gú-di#a#-lo!"=lu%$ be%-di#a#-lo!"=lu%$ ru-di#a#-lo!"=lu%$
2 S FORMAL gú-di#a#-lo!"=é be%-di#a#-lo!"=é ru-di#a#-lo!"=é
3 S FAMILIAR gú-di#a#-lo!"=bí be%-di#a#-lo!"=bí ru-di#a#-lo!"=bí
3 S FORMAL gú-di#a#-lo!"=é be%-di#a#-lo!"=é ru-di#a#-lo!"=é
3 S NON-FORMAL gú-di#a#-lú=i# be%-di#a#-lú=i# ru-di#a#-lú=i#
1 P INCLUSIVE gú-di#a#-lo!"=rí$u% be%-di#a#-lo!"=rí$u% ru-di#a#-lo!"=rí$u%
1 P EXCLUSIVE gú-di#a#-lo!"=tu%$ be%-di#a#-lo!"=tu%$ ru-di#a#-lo!"=tu%$
2 P INFORMAL gú-di#a#-lo!"=lé be%-di#a#-lo!"=lé ru-di#a#-lo!"=lé
2 P FORMAL gú-di#a#-lo!"=lé be%-di#a#-lo!"=lé ru-di#a#-lo!"=lé
3 P FAMILIAR gú-di#a#-lo!"=ga%bí be%-di#a#-lo!"=ga%bí ru-di#a#-lo!"=ga%bí
3 P FORMAL gú-di#a#-lo!"=gé be%-di#a#-lo!"=gé ru-di#a#-lo!"=gé
3 P NON-FORMAL gú-di#a#-lo!"=gi) # be%-di#a#-lo!"=gi) # ru-di#a#-lo!"=gi) #
Notes: This verb includes an obligatorily possessed noun. These require genitive
subjects (Foreman 2006), as evident in the third-person non-formal foms.
364
Class: 1 (1A)
English: ‘to scrunch up’
Spanish: ‘abollar’
Posited verb root: / &a%tta% /
Aspect markers: /ú/, /be%/, /ru/
Nellis and Nellis page: not given
POTENTIAL COMPLETIVE HABITUAL
1 S ú-&átta%=$ be%-&átta%=$ rú-&átta%=$
2 S INFORMAL ú-&a%tta%=lu%$ be%-&a%tta%=lu%$ ru-&a%tta%=lu%$
2 S FORMAL ú-&a%tta%=é be%-&a%tta%=é ru-&a%tta%=é
3 S FAMILIAR ú-&a%tta%=bí be%-&a%tta%=bí ru-&a%tta%=bí
3 S FORMAL ú-&a%tta%=é be%-&a%tta%=é ru-&a%tta%=e
3 S NON-FORMAL ú-&a%tta) % be%-&a%tta) % ru-&a%tta) %
1 P INCLUSIVE ú-&a%tta%=rí$u% be%-&a%tta%=rí$u% ru-&a%tta%=rí$u%
1 P EXCLUSIVE ú-&a%tta%=tu%$ be%-&a%tta%=tu%$ ru-&a%tta%=tu%$
2 P INFORMAL ú-&a%tta%=lé be%-&a%tta%=lé ru-&a%tta%=lé
2 P FORMAL ú-&a%tta%=lé be%-&a%tta%=lé ru-&a%tta%=lé
3 P FAMILIAR ú-&a%tta%=ga%bí be%-&a%tta%=ga%bí ru-&a%tta%=ga%bí
3 P FORMAL ú-&a%tta%=gé be%-&a%tta%=gé ru-&a%tta%=gé
3 P NON-FORMAL ú-&a%tta%=ga) # be%-&a%tta%=ga) # ru-&a%tta%=ga) #
365
Class: 1 (1A)
English: ‘to cover’
Spanish: ‘cubrir'
Posited verb root: /di + ba#gá$/
Aspect markers: /gú/, /be%/, /ru/
Nellis and Nellis page: 118
POTENTIAL COMPLETIVE HABITUAL
1 S gú-dí-ba!"gá$ be%-di%-ba!"gá$ rú-dí-ba!"gá$
2 S INFORMAL gú-dí-ba##"gá$=lu%$ be%-di%-ba##"gá$=lu%$ ru-di-ba#"gá$=lu%$
2 S FORMAL gú-dí-ba##"gé$=é be%-di%-ba##"gé$=é ru-di-ba#"gé$=é
3 S FAMILIAR gú-dí-ba#"gá$=bí be%-di%-ba#"gá$=bí ru-di-ba#"gá$=bí
3 S FORMAL gú-dí-ba#"gé$=é be%-di%-ba#"gé$=é ru-di-ba#"gé$=é
3 S NON-FORMAL gú-dí-ba#"gá$=a) be%-di%-ba#"gá$=a) ru-di-ba##"gá$=a)
1 P INCLUSIVE gú-dí-ba#"gá$=rí$u% be%-di%-ba#"gá$=rí$u% ru-di-ba##"gá$=rí$u%
1 P EXCLUSIVE gú-dí-ba#"gá$=tu%$ be%-di%-ba##"gá$=tu%$ ru-di-ba#"gá$=tu%$
2 P INFORMAL gú-dí-ba#"gá$=lé be%-di%-ba#"gá$=lé ru-di-ba#"gá$=lé
2 P FORMAL gú-dí-ba#"gá$=lé be%-di%-ba#"gá$=lé ru-di-ba##"gá$=lé
3 P FAMILIAR gú-dí-ba#"gá$=ga%bí be%-di%-ba#"gá$=ga%bí ru-di-ba#"gá$=ga%bí
3 P FORMAL gú-dí-ba#"gá$=gé be%-di%-ba#"gá$=gé ru-di-ba#"gá$=gé
3 P NON-FORMAL gú-dí-ba##"gá$=ga) # be%-di%-ba##"gá$=ga) # ru-di-ba#"gá$=ga) #
Notes: This verb includes the causative marker /di/.
366
Class: 1 (1A)
English: ‘to fold, twist’
Spanish: ‘doblar / torcer'
Posited verb root: / det(u%/
Aspect markers: /gú/, /be%/, /ru/
Nellis and Nellis page: 85
POTENTIAL COMPLETIVE HABITUAL
1 S gú-dét(
w
=a#$ be%-dét(
w
=a#$ rú-dét(
w
=a#$
2 S INFORMAL gú-dét(u%=lu%$ be%-de%t(u%=lu%$ ru-det(u%=lu%$
2 S FORMAL gú-dét(u%=é be%-de%t(
w
=é ru-det(
w
=é
3 S FAMILIAR gú-dét(u%=bí be%-de%t(u%=bí ru-det(u%=bí
3 S FORMAL gú-dét(
w
=é be%-de%t(
w
=é ru-det(
w
=é
3 S NON-FORMAL gú-dét(
w
=a) be%-de%t(
w
=a) ru-det(
w
=a)
1 P INCLUSIVE gú-dét(u%=rí$u be%-de%t(u%=rí$u% ru-det(u%=rí$u%
1 P EXCLUSIVE gú-dét(u%=tu%$ be%-de%t(u%=tu%$ ru-det(u%=tu%$
2 P INFORMAL gú-dét(u%=lé be%-de%t(u%=lé ru-det(u%=lé
2 P FORMAL gú-dét(u%=lé be%-de%t(u%=lé ru-det(u%=lé
3 P FAMILIAR gú-dét(u%=ga%bí be%-de%t(u%=ga%bí ru-det(u%=ga%bí
3 P FORMAL gú-dét(u%=gé be%-de%t(u%=gé ru-det(u%=gé
3 P NON-FORMAL gú-dét(u%=ga) # be%-de%t(u%=ga) # ru-det(u%=ga) #
367
Class: 1 (1A)
English: ‘to write’
Spanish: ‘escribir’
Posited verb root: /di#a#/
Aspect markers: /gú/, /be%/, /ru/
Nellis and Nellis page: 86
POTENTIAL COMPLETIVE HABITUAL
1 S gú-día#=$ be%-día#=$ rú-día#=$
2 S INFORMAL gú-di#a#=lu%$ be%-di#a#=lu%$ ru-di#a#=lu%$
2 S FORMAL gú-di#=é be%-di#=é ru-di#=é
3 S FAMILIAR gú-di#a#=bí be%-di#a#=bí ru-di#a#=bí
3 S FORMAL gú-di#=é be%-di#=é ru-di#=é
3 S NON-FORMAL gú-di#a) # be%-di#a) # ru-di#a) #
1 P INCLUSIVE gú-di#a#=rí$u% be%-di#a#=rí$u% ru-di#a#=rí$u%
1 P EXCLUSIVE gú-di#a#=tu%$ be%-di#a#=tu%$ ru-di#a#=tu%$
2 P INFORMAL gú-di#a#=lé be%-di#a#=lé ru-di#a#=lé
2 P FORMAL gú-di#a#=lé be%-di#a#=lé ru-di#a#=lé
3 P FAMILIAR gú-di#a#=ga%bí be%-di#a#=ga%bí ru-di#a#=ga%bí
3
P FORMAL gú-di#a#=gé be%-di#a#=gé ru-di#a#=gé
3 P NON-FORMAL gú-di#a#=ga) # be%-di#a#=ga) # # ru-di#a#=ga) #
368
Class: 1 (1A)
English: ‘to listen’
Spanish: ‘escuchar’
Posited verb root: / da%" + na%"ga%$/
Aspect markers: /gú/, /be%/, /ru/
Nellis and Nellis page: 82
POTENTIAL COMPLETIVE HABITUAL
1 S gú-da!"-na%"gá$ be%-da!"-na%"gá$ rú-da!"-na%"gá$
2 S INFORMAL gú-da%"-na%"ga%$=lu%$ be%-da%"-na%"ga%$=lu%$ ru-da%"-na%"ga%$=lu%$
2 S FORMAL gú-da%"-na%"ga%$=é be%-da%"-na%"ga%$=é ru-da%"-na%"ga%$=é
3 S FAMILIAR gú-da%"-na%"ga%$=bí be%-da%"-na%"ga%$=bí ru-da%"-na%"ga%$=bí
3 S FORMAL gú-da%"-na%"ga%$=é be%-da%"-na%"ga%$=é ru-da%"-na%"ga%$=é
3 S NON-FORMAL gú-da%"-na%"g=i) # be%-da%"-na%"g=i) #
ru-da%"-na%"g=i) #
1 P INCLUSIVE gú-da%"-na%"ga%$=rí$u% be%-da%"-na%"ga%$=rí$u% ru-da%"-na%"ga%$=rí$u%
1 P EXCLUSIVE gú-da%"-na%"ga%$=tu%$ be%-da%"-na%"ga%$=tu%$ ru-da%"-na%"ga%$=tu%$
2 P INFORMAL gú-da%"-na%"ga%$=lé be%-da%"-na%"ga%$=lé ru-da%"-na%"ga%$=lé
2 P FORMAL gú-da%"-na%"ga%$=lé be%-da%"-na%"ga%$=lé ru-da%"-na%"ga%$=lé
3
rd
P FAMILIAR gú-da%"-na%"ga%$=ga%bí be%-da%"-na%"ga%$=ga%bí ru-da%"-na%"ga%$=ga%bí
3
rd
P FORMAL gú-da%"-na%"ga%$=gé be%-da%"-na%"ga%$=gé ru-da%"-na%"ga%$=gé
3 P NON-FORMAL gú-da%"-na%"ga%$=gi) # be%-da%"-na%"ga%$=gi) # ru-da%"-na%"ga%$=gi) #
Notes: This verb includes an obligatorily possessed noun. These require genitive
subjects, as evident in the third-person non-formal foms.
369
Class: 1 (1A)
English: ‘to dry (something)’
Spanish: ‘secar (algo)’
Posited verb root: / di +bi%"tsi% /
Aspect markers: /gú/, /be%/, /ru/
Nellis and Nellis page: 56
POTENTIAL COMPLETIVE HABITUAL
1 S gú-dí-bi!"tsi%=a#$ be%-di%-bi!"tsi%=a#$ rú-dí-bi!"tsi%=a#$
2 S INFORMAL gú-dí-bi%"tsi%=lu%$ be%-di%-bi%"tsi%=lu%$ ru-di-bi%"tsi%=lu%$
2 S FORMAL gú-dí-bi%"tsi%=é be%-di%-bi%"tsi%=é ru-di-bi%"tsi%=é
3 S FAMILIAR gú-dí-bi%"tsi%=bí be%-di%-bi%"tsi%=bí ru-di-bi%"tsi%=bí
3 S FORMAL gú-dí-bi%"tsi%=é be%-di%-bi%"tsi%=é ru-di-bi%"tsi%=é
3 S NON-FORMAL gú-dí-bi%"tsi%=a
% ) be%-di%-bi%"tsi%=a
% ) ru-di-bi%"tsi%=a
% )
1 P INCLUSIVE gú-dí-bi%"tsi%=rí$u% be%-di%-bi%"tsi%=rí$u% ru-di-bi%"tsi%=rí$u%
1 P EXCLUSIVE gú-dí-bi%"tsi%=tu%$ be%-di%-bi%"tsi%=tu%$ ru-di-bi%"tsi%=tu%$
2 P INFORMAL gú-dí-bi%"tsi%=lé be%-di%-bi%"tsi%=lé ru-di-bi%"tsi%%=lé
2 P FORMAL gú-dí-bi%"tsi%=lé be%-di%-bi%"tsi%=lé ru-di-bi%"tsi%=lé
3
rd
P FAMILIAR gú-dí-bi%"tsi%=ga%bí be%-di%-bi%"tsi%=ga%bí ru-di-bi%"tsi%=ga%bí
3
rd
P FORMAL gú-dí-bi%"tsi%=gé be%-di%-bi%"tsi%=gé ru-di-bi%"tsi%=gé
3 P NON-FORMAL gú-dí-bi%"tsi%=ga) #
be%-di%-bi%"tsi%= ga) #
ru-di-bi%"tsi%= ga) #
Notes: This verb includes the causative /di/
370
Class: 2 (IB)
English: ‘to hit or strike’
Spanish: ‘pegar’
Posited verb root: /e%$/
Aspect markers: /w + H tone/, /be/, /ru/
Nellis and Nellis page: 113
POTENTIAL COMPLETIVE HABITUAL
1 S wé$=a#$ bé$=a#$ ré$=a#$
2 S INFORMAL wé-e%$=lu%$ be-e%$=lu%$ ru-e%$=lu%$
2 S FORMAL wé-e%$=é be-e%$=é ru-e%$=é
3 S FAMILIAR wé-e%$=bí be-e%$=bí ru-e%$=bí
3 S FORMAL wé-e%$=é be-e%$=é ru-e%$=é
3 S NON-FORMAL wé-e%$=a) be-e%$=a) ru-e%$=a)
1 P INCLUSIVE wé-e%$=rí$u be-e%$=rí$u% ru-e%$=rí$u%
1 P EXCLUSIVE wé-e%$=tu%$ be-e%$=tu%$ ru-e%$=tu%$
2 P INFORMAL wé-e%$=lé be-e%$=lé ru-e%$=lé
2 P FORMAL wé-e%$=lé be-e%$=lé ru-e%$=lé
3
rd
P FAMILIAR wé-e%$=ga%bí be-e%$=ga%bí ru-e%$=ga%bí
3
rd
P FORMAL wé-e%%$=gé be-e%$=gé ru-e%$=gé
3 P NON-FORMAL wé-e%%$=ga) #
be-e%$=ga) # ru-e%$=ga) #
Notes A falling contour appears for the potential on the other persons (HM) and the
vowel is longer in the completive, which suggests that it be- + plus another /e/
vowel.
371
Class: 2 (IB)
English: ‘to carry/haul’
Spanish: ‘llevar/cargar’
Posited verb root: /a$ní/
Aspect markers: /w + H tone/, / be/, /ru/
Nellis and Nellis page: 112
POTENTIAL COMPLETIVE HABITUAL
1 S wá$ní=á$ bí-á$ní=á$ rú-á$ní=á$
2 S INFORMAL wá$ní=lú$ bi-a$ní=lú$ ru-a$ní=lú$
2 S FORMAL wá$ní=e bi-a$ní=e ru-a$ní=e
3 S FAMILIAR wá$ní=bí bi-a$ní=bí ru-a$ní=bí
3 S FORMAL wá$ní=e bi-a$ní=e ru-a$ní=e
3 S NON-FORMAL wá$ní=a) . bi-a$ní=a) . ru-a$ní=a) .
1 P INCLUSIVE wá$ní=rí$u% bi-a$ní=rí$u% ru-a$ní=rí$u%
1 P EXCLUSIVE wá$ní=tú$ bi-a$ní=tú$ ru-a$ní=tú$
2 P INFORMAL wá$ní=lé bi-a$ní=lé ru-a$ní=lé
2 P FORMAL wá$ní=lé bi-a$ní=lé ru-a$ní=lé
3
rd
P FAMILIAR wá$ní=gábí bi-a$ní=gábí ru-a$ní=gábí
3
rd
P FORMAL wá$ní=gé bi-a$ní=gé ru-a$ní=gé
3 P NON-FORMAL wá$ní=ga) . bi-a$ní=ga) . ru-a$ní=ga) .
Notes: There is a segmental change in the completive aspect marker: /be/ becomes /bi/
before the vowel /a/.
372
Class: 3 (IIA)
English: ‘to see (something)’
Spanish: ‘ver (algo)’
Posited verb root: /lá$ + ni/
Aspect markers: /í/, / bi/, /ri%/
Nellis and Nellis page: 123
POTENTIAL COMPLETIVE HABITUAL
1 S í-lá$=te#$ bí-lá$=te#$ ri%-lá$=te#$
2 S INFORMAL í-lá$-ni=lu%$ bi-lá$-ni=lu%$ ri%-lá$-ni=lu%$
2 S FORMAL í-lá$-ni=é bi-lá$-ni=é ri%-lá$-ni=é
3 S FAMILIAR í-lá$-ni=bí bi-lá$-ni=bí ri%-lá$-ni=bí
3 S FORMAL í-lá$-ni=é bi-lá$-ni=é ri%-lá$-ni=é
3 S NON-FORMAL í-lá$-ni=a) bi-lá$-ni=a) ri%-lá$-ni=a)
1 P INCLUSIVE í-lá$-ni=rí$u% bi-lá$-ni=rí$u% ri%-lá$-ni=rí$u%
1 P EXCLUSIVE í-lá$-ni=tu%$ bi-lá$-ni=tu%$ ri%-lá$-ni=tu%$
2 P INFORMAL í-lá$-ni=lé bi-lá$-ni=lé ri%-lá$-ni=lé
2 P FORMAL í-lá$-ni=lé bi-lá$-ni=lé ri%-lá$-ni=lé
3
rd
P FAMILIAR í-lá$-ni=ga%bí bi-lá$-ni=ga%bí ri%-lá$-ni=ga%bí
3
rd
P FORMAL í-lá$-ni=gé bi-lá$-ni=gé ri%-lá$-ni=gé
3 P NON-FORMAL í-lá$-ni=ga) #
bi-lá$-ni=ga) #
ri%-lá$-ni=ga) #
Notes: This verb takes the special /te$/ for subject oriented stems, as well as /ni/. See
Bartholomew (1983: 349) for a discussion of these verbs.
373
Class: 4 (IIB)
English: ‘to fall down’
Spanish: ‘caerse’
Posited verb root: / bi&i%/
Aspect markers: /í/, / u/, /ri/
Nellis and Nellis page: 119
POTENTIAL COMPLETIVE HABITUAL
1 S í-bí&i%=a%$ u-bi&i%=a%$ ri-bi&i%=a%$
2 S INFORMAL í-bí&i%=lu%$ u-bi&i%=lu%$ ri-bi&i%=lu%$
2 S FORMAL í-bí&i%=é u-bi&i%=é ri-bi&i%=é
3 S FAMILIAR í-bí&i%=bí u-bi&i%=bí ri-bi&i%=bí
3 S FORMAL í-bí&i%=é u-bi&i%=é ri-bi&i%=é
3 S NON-FORMAL í-bí&i%=a
% ) u-bi&i%=a
% ) ri-bi&i%=a) %
1 P INCLUSIVE í-bí&i%=rí$u% u-bi&i%=rí$u% ri-bi&i%=rí$u%
1 P EXCLUSIVE í-bí&i%=tu%$ u-bi&i%=tu%$ ri-bi&i%=tu%$
2 P INFORMAL í-bí&i%=lé u-bi&i%=lé ri-bi&i%=lé
2 P FORMAL í-bí&i%%=lé u-bi&i%=lé ri-bi&i%=lé
3
rd
P FAMILIAR í-bí&i%%=ga%bí u-bi&i%=ga%bí ri-bi&i%=ga%bí
3
rd
P FORMAL í-bí&i%=gé u-bi&i%=gé ri-bi&i%=gé
3 P NON-FORMAL í-bí&i%=ga) #
u-bi&i%=ga) #
ri-bi&i%=ga) #
Notes: I assume this verb lacks a floating tone because it is intransitive. The verb also
shows rightward spread onto toneless syllable.
374
Class: 4 (IIB)
English: ‘to drink’
Spanish: ‘beber’
Posited verb root: /ja%/
Aspect markers: /gí/, /gu/, /ri/
Nellis and Nellis page: 138
POTENTIAL COMPLETIVE HABITUAL
1 S gí$-já=$ gú$-já=$ rí$-já=$
2 S INFORMAL gí$-ja%=lu%$ gu$-ja%=lu%$ ri$-ja%=lu%$
2 S FORMAL gí$-jé gu$-jé ri$-jé
3 S FAMILIAR gí$-ja%=bí gu$-ja%=bí ri$-ja%=bí
3 S FORMAL gí$-jé gu$-jé ri$-jé
3 S NON-FORMAL gí$-ja) gu$-ja) ri$-ja)
1 P INCLUSIVE gí$-ja%=rí$u% gu$-ja%=rí$u% ri$-ja%=rí$u%
1 P EXCLUSIVE gí$-ja%=tu%$ gu$-ja%=tu%$ ri$-ja%=tu%$
2 P INFORMAL gí$-ja%=lé gu$-ja%=lé ri$-ja%=lé
2 P FORMAL gí$-ja%=lé gu$-ja%=lé ri$-ja%=lé
3
P FAMILIAR gí$-ja%=ga%bí gu$-ja%=ga%bí ri$-ja%=ga%bí
3 P FORMAL gí$-ja%=gé gu$-ja%=gé ri$-ja%=gé
3 P NON-FORMAL gí$-ja%= ga) #
gu$-ja%= ga) #
ri$-ja%= ga) #
Notes: It appears that there is an epenthetic glottal stop added after the aspect markers.
375
Class: 4 (IIB)
English: ‘to hug, wrap around’
Spanish: ‘abrazar’
Posited verb root: / láni%/
Aspect markers: /í/, /gu/, /ri/
Nellis and Nellis page: 123
POTENTIAL COMPLETIVE HABITUAL
1 S í-láni%=a%$ gú-láni%=a%$ rí-láni%=a%$
2 S INFORMAL í-láni%=lu%$ gu-láni%=lu%$ ri-láni%=lu%$
2 S FORMAL í-láni%=é gu-láni%=é ri-láni%=é
3 S FAMILIAR í-láni%=bí gu-láni%=bí ri-láni%=bí
3 S FORMAL í-láni%=é gu-láni%=é ri-láni%=é
3 S NON-FORMAL í-láni%=a) % gu-láni%=a) % ri-láni%=a) %
1 P INCLUSIVE í-láni%=rí$u% gu-láni%=rí$u% ri-láni%=rí$u%
1 P EXCLUSIVE í-láni%=tu%$ gu-láni%=tu%$ ri-láni%=tu%$
2 P INFORMAL í-láni%=lé gu-láni%=lé ri-láni%=lé
2 P FORMAL í-láni%=lé gu-láni%=lé ri-láni%=lé
3 P FAMILIAR í-láni%=ga%bí gu-láni%=ga%bí ri-láni%=ga%bí
3 P FORMAL í-láni%=gé gu-láni%=gé ri-láni%=gé
3 P NON-FORMAL í-láni%=ga) #
gu-láni%=ga) #
ri-láni%=ga) #
376
Class: 4 (IIB)
English: ‘to pull out ; tear up’
Spanish: ‘arrancar’
Posited verb root: /la%tsu#$/
Aspect markers: /í/, /gu/, /ri/
Nellis and Nellis page: 123
POTENTIAL COMPLETIVE HABITUAL
1 S í-látsu#$=a#$ gú-látsu#$=a#$ rí-látsu#$=a#$
2 S INFORMAL í-la%tsu#$=lu%$ gu-la%tsu#$=lu%$ ri-la%tsu#$=lu%$
2 S FORMAL í-la%tsu#$=é gu-la%tsu#$=é ri-la%tsu#$=é
3 S FAMILIAR í-la%tsu#$=bí gu-la%tsu#$=bí ri-la%tsu#$=bí
3 S FORMAL í-la%tsu#$=é gu-la%tsu#$=é ri-la%tsu#$=é
3 S NON-FORMAL í-la%tsu#$=a) gu-la%tsu#$=a) ri-la%tsu#$=a)
1 P INCLUSIVE í-la%tsu#$=rí$u% gu-la%tsu#$=rí$u% ri-la%tsu#$=rí$u%
1 P EXCLUSIVE í-la%tsu#$=tu%$ gu-la%tsu#$=tu%$ ri-la%tsu#$=tu%$
2 P INFORMAL í-la%tsu#$=lé gu-la%tsu#$=lé ri-la%tsu#$=lé
2 P FORMAL í-la%tsu#$=lé gu-la%tsu#$=lé ri-la%%tsu#$=lé
3 P FAMILIAR í-la%tsu#$=ga%bí gu-la%tsu#$=ga%bí ri-la%tsu#$=gabí
3 P FORMAL í-la%tsu#$=gé gu-la%tsu#$=gé ri-la%tsu#$=gé
3 P NON-FORMAL í-la%tsu#$=ga) #
gu-la%tsu#$=ga) #
ri-la%tsu#$=ga) #
377
Class: 5 (IIC)
English: ‘to return (something)’
Spanish: ‘devolver (algo)’
Posited verb root: /júnna%/
Aspect markers: /é/, /be%/, /re/
Nellis and Nellis page: 68
POTENTIAL COMPLETIVE HABITUAL
1 S é-júnna%=$ be%-júnna%=$ ré-júnna%=$
2 S INFORMAL é-júnna%=lu%$ be%-júnna%=lu%$ re-júnna%=lu%$
2 S FORMAL é-júnné be%-júnné re-júnné
3 S FAMILIAR é-júnna%=bí be%-júnna%=bí re-júnna%=bí
3 S FORMAL é-júnné be%-júnné re-júnné
3 S NON-FORMAL é-júnna) be%-júnna) re-júnna)
1 P INCLUSIVE é-júnna%=rí$u% be%-júnna%=rí$u% re-júnna%=rí$u%
1 P EXCLUSIVE é-júnna%=tu%$ be%-júnna%=tu%$ re-júnna%=tu%$
2 P INFORMAL é-júnna%=lé be%-júnna%=lé re-júnna%=lé
2 P FORMAL é-júnna%=lé be%-júnna%=lé re-júnna%%=lé
3 P FAMILIAR é-júnna%=ga%bí be%-júnna%=ga%bí re-júnna%=ga%bí
3
P FORMAL é-júnna%=gé be%-júnna%=gé re-júnna%=gé
3 P NON-FORMAL é-júnna%=ga) #
be%-júnna%=ga) #
re-júnna%= ga) #
378
Class: 5 (IIC)
English: ‘to answer’
Spanish: ‘contestar’
Posited verb root: / kka%bí/
Aspect markers: /é/, /be%/, /re/
Nellis and Nellis page: 56
POTENTIAL COMPLETIVE HABITUAL
1 S é-kkábí=á$ be%-kkábí=á$ ré-kkábí=á$
2 S INFORMAL é-kka%bí=lú$ be%-kka%bí=lú$ re-kka%bí=lú$
2 S FORMAL é-kka%bí=e be%-kka%bí=e re-kka%bí=e
3 S FAMILIAR é-kka%bí=/ be%-kka%bí=/ re-kka%bí=/
3 S FORMAL é-kka%bí=e be%-kka%bí=e re-kka%bí=e
3 S NON-FORMAL é-kka%bí=a) . be%-kka%bí=a) . re-kka%bí=a) .
1 P INCLUSIVE é-kka%bí=rí$u% be%-kka%bí=rí$u% re-kka%bí=rí$u%
1 P EXCLUSIVE é-kka%bí=tú$ be%-kka%bí=tú$ re-kka%bí=tú$
2 P INFORMAL é-kka%bí=lé be%-kka%bí=lé re-kka%bí=lé
2 P FORMAL é-kka%bí=lé be%-kka%bí=lé re-kka%bí=lé
3
P FAMILIAR é-kka%bí=gábí be%-kka%bí=gábí re-kka%bí=gábí
3 P FORMAL é-kka%bí=gé be%-kka%bí=gé re-kka%bí=gé
3 P NON-FORMAL é-kka%bí=ga) .
be%-kka%bí=ga) . re-kka%bí=ga) .
Notes: The third-person singular familiar /bí/ is dropped here. An OCP effect?
379
Class: 5 (IIC)
English: ‘to wake up’
Spanish: ‘despertarse’
Posited verb root: / ba%ni%/
Aspect markers: /é/, /be%/, /re/
Nellis and Nellis page: 55
POTENTIAL COMPLETIVE HABITUAL
1 S é-ba%ni%=a%$ be%-ba%ni%=a%$ re-ba%ni%=a%$
2 S INFORMAL é-ba%ni%=lu%$ be%-ba%ni%=lu%$ re-ba%ni%=lu%$
2 S FORMAL é-ba%ni%=é be%-ba%ni%=é re-ba%ni%=é
3 S FAMILIAR é-ba%ni%=bí be%-ba%ni%=bí re-ba%%ni%=bí
3 S FORMAL é-ba%ni%=é be%-ba%ni%=é re-ba%ni%=é
3 S NON-FORMAL é-ba%ni%=a)) % be%-ba%ni%=a) % re-ba%%ni%=a) %
1 P INCLUSIVE é-ba%ni%=rí$u% be%-ba%ni%=rí$u% re-ba%ni%=rí$u%
1 P EXCLUSIVE é-ba%ni%=tu%$ be%-ba%ni%=tu%$ re-ba%ni%=tu%$
2 P INFORMAL é-ba%ni%=lé be%-ba%ni%=lé re-ba%ni%=lé
2 P FORMAL é-ba%ni%=lé be%-ba%ni%=lé re-ba%ni%=lé
3
P FAMILIAR é-ba%ni%=ga%bí be%-ba%ni%=ga%bí re-ba%ni%=ga%bí
3 P FORMAL é-ba%ni%=gé be%-ba%ni%=gé re-ba%ni%=gé
3 P NON-FORMAL é-ba%ni%=ga) #
be%-ba%ni%=ga) #
re-ba%ni%=ga) #
Notes: No floating H because intransitive.
380
Class: 6 (III)
English: ‘to dig’
Spanish: ‘cavar’
Posited verb root: / a%:ní/
Aspect markers: /k + H tone/, /u%/, /ri/
Nellis and Nellis page: 28
POTENTIAL COMPLETIVE HABITUAL
1 S ka!"ní=á$ u%-ta!"ní=á$ rí-ga!"ní=á$
2 S INFORMAL ka!"ní=lú$ u%-ta%"ní=lú$ ri-ga%"ní=lú$
2 S FORMAL ka!"ní=e u%-ta%"ní=e ri-ga%"ní=e
3 S FAMILIAR ka!"ní=bí u%-ta%"ní=bí ri-ga%"ní=bí
3 S FORMAL ka!"ní=e u%-ta%"ní=e ri-ga%"ní=e
3 S NON-FORMAL ka!"ní=a) . u%-ta%"ní=a) . ri-ga%%"ní=a) .
1 P INCLUSIVE ka!"ní=rí$u% u%-ta%"ní=rí$u% ri-ga%"ní=rí$u%
1 P EXCLUSIVE ka!"ní=tú$ u%-ta%"ní=tú$ ri-ga%"ní=tu%$
2 P INFORMAL ka!"ní=lé u%-ta%"ní=lé ri-ga%"ní=lé
2 P FORMAL ka!"ní=lé u%-ta%"ní=lé ri-ga%"ní=lé
3
P FAMILIAR ka!"ní=gábí u%-ta%"ní=gábí ri-ga%"ní=gábí
3 P FORMAL ka!"ní=gé u%-ta%"ní=gé ri-ga%"ní=gé
3 P NON-FORMAL ka!"ní=ga) . u%-ta%"ní=ga) .
ri-ga%"ní=ga) .
Notes: A falling tone present when the 1 s is added. The H of the potential creates a
falling contour for all pronouns. Also, there is a segmental change. There are
different stems for the different aspects.
381
Class: 6 (III)
English: ‘to carve (wood)’
Spanish: ‘raspar (madera) ; labrar’
Posited verb root: / kába#/
Aspect markers: ///, /be%/, /ri/
Nellis and Nellis page: 28
POTENTIAL COMPLETIVE HABITUAL
1 S kába#=$ be%-kába#=$ rí-kába#=$
2 S INFORMAL kába##=lu%$ be%-kába#=lu%$ ri-kába##=lu%$
2 S FORMAL kábé be%-kábé ri-ga%"ní=e
3 S FAMILIAR kába#=bí be%-kába#=bí ri-kába##=bí
3 S FORMAL kába#=é be%-kába#=e ri-ga%"ní=e
3 S NON-FORMAL kába) # be%-kába) # ri-kába) #
1 P INCLUSIVE kába#=rí$u% be%-kába#=rí$u% ri-kába##=rí$u%
1 P EXCLUSIVE kába#=tu%$ be%-kába#=tu%$ ri-kába##=tu%$
2 P INFORMAL kába#=lé be%-kába#=lé ri-kába##=lé
2 P FORMAL kába#=lé be%-kába#=lé ri-kába##=lé
3
P FAMILIAR kába#=ga%bí be%-kába#=ga%bí ri-kába##=ga%bí
3 P FORMAL kába#=gé be%-kába#=gé ri-kába##=gé
3 P NON-FORMAL kába#=ga) # be%-kába#=ga) #
ri-kába##=ga) #
382
Class: 6 (III)
English: ‘to paint’
Spanish: ‘pintar’
Posited verb root: /a##"bi#/
Aspect markers: /k + H tone/, /u%t/, /r/
Nellis and Nellis page: 30
POTENTIAL COMPLETIVE HABITUAL
1 S ka!"bi#=a#$ u%-ta!"bi#=a#$ ra!"bi#=a#$
2 S INFORMAL ka!"bi#=lu%$ u%-ta##"bi#=lu%$ ra##"bi#=lu%$
2 S FORMAL ka!"bi#=é u%-ta##"bi#=é ra#"bi#=é
3 S FAMILIAR ka!"bi#=bí u%-ta##"bi#=bí ra##"bi#=bí
3 S FORMAL ka!"bi#=é u%-ta#"bi#=é ra#"bi#=é
3 S NON-FORMAL ka!"bi#=a) # u%-ta##"bi#=a) # ra#"bi#=a) #
1 P INCLUSIVE ka!"bi#=rí$u% u%-ta#"bi#=rí$u% ra#"bi#=rí$u%
1 P EXCLUSIVE ka!"bi#=tu%$ u%-ta#"bi#=tu%$ ra#"bi#=tu%$
2 P INFORMAL ka!"bi#=lé u%-ta#"bi#=lé ra##"bi#=lé
2 P FORMAL ka!"bi#=lé u%-ta##"bi#=lé ra##"bi#=lé
3
P FAMILIAR ka!"bi#=ga%bí u%-ta##"bi#=ga%bí ra#"bi#=ga%bí
3 P FORMAL ka!"bi#=gé u%-ta#"bi#=gé ra##"bi#=gé
3 P NON-FORMAL ka!"bi#=ga) # u%-ta##"bi#=ga) # ra##"bi#=ga) #
383
Class: 6 (III)
English: ‘to wait for’
Spanish: ‘esperar’
Posited verb root: /edá/
Aspect markers: /k
w
+ H tone/, /u%/, /ri/
Nellis and Nellis page: 35
POTENTIAL COMPLETIVE HABITUAL
1 S k
w
édá=$ u%-lédá=$ rí-bédá=$
2 S INFORMAL k
w
édá=lú$ u%-le%dá=lú$ ri-bedá=lú$
2 S FORMAL k
w
édé=e u%-le%dé=e ri-bedé=e
3 S FAMILIAR k
w
édá=bí u%-le%dá=bí ri-bedá=bí
3 S FORMAL k
w
édé=e u%-le%dé=e ri-bedé=e
3 S NON-FORMAL k
w
éda) . u%-le%da) . ri-beda) .
1 P INCLUSIVE k
w
édá=rí$u% u%-le%dá=rí$u% ri-bedá=rí$u%
1 P EXCLUSIVE k
w
édá=tú$ u%-le%dá=tú$ ri-bedá=tú$
2 P INFORMAL k
w
édá=lé u%-le%dá=lé ri-bedá=lé
2 P FORMAL k
w
édá=lé u%-le%dá=lé ri-bedá=lé
3 P FAMILIAR k
w
édá=gábí u%-le%dá=gábí ri-bedá=ga.bí
3 P FORMAL k
w
édá=gé u%-le%dá=gé ri-beda.=gé
3 P NON-FORMAL k
w
édá=ga) . u%-le%dá=ga) . ri-beda.=ga) .
Notes: Irregular verb – note the segmental changes: an /l/ is added in the completive,
and a /b/ in the habitual.
384
Class: 6 (III)
English: ‘to take away; take out ; remove’
Spanish: ‘quitar’
Posited verb root: /o#a#/
Aspect markers: /k+ H tone/, /gu%/, /ri/
Nellis and Nellis page: 58
POTENTIAL COMPLETIVE HABITUAL
1 S kóa#=$ gu%-tóa#=$ rí-góa#=$
2 S INFORMAL kóa#=lu%$ gu%-to#a#=lu%$ ri-go#a#=lu%$
2 S FORMAL kóe#=é gu%-to#e#=é ri-go#e#=é
3 S FAMILIAR kóa#=bí gu%-to#a#=bí ri-go#a#=bí
3 S FORMAL kóe#=é gu%-to#e#=é ri-go#e#=é
3 S NON-FORMAL kóa) # gu%-to#a) # ri-go#a) #
1 P INCLUSIVE kóa#=rí$u% gu%-to#a#=rí$u% ri-go#a#=rí$u%
1 P EXCLUSIVE kóa#=tu%$ gu%-to#a#=tu%$ ri-go#a#=tu%$
2 P INFORMAL kóa#=lé gu%-to#a#=lé ri-go#a#=lé
2 P FORMAL kóa#=lé gu%-to#a#=lé ri-go#a#=lé
3 P FAMILIAR kóa#=ga%bí gu%-to#a#=ga%bí ri-go#a#=ga%bí
3
P FORMAL kóa#=gé gu%-to#a#=gé ri-go#a#=gé
3 P NON-FORMAL kóa#=ga) # gu%-to#a#=ga) # ri-go#a#=ga) #
385
Class: 7 (IV-A)
English: ‘to sow (crops)’
Spanish: ‘sembrar’
Posited verb root: /ada%/
Aspect markers: /gá/, /gu/, /rá/
Nellis and Nellis page: 70
POTENTIAL COMPLETIVE HABITUAL
1 S gá-ádá=$ gú-údá=$ rá-ádá=$
2 S INFORMAL gá-áda%=lu%$ gu-uda%=lu%$ rá-áda%=lu%$
2 S FORMAL gá-áde%=é gu-ude%=é rá-áde%=é
3 S FAMILIAR gá-áda%=bí gu-uda%=bí rá-áda%=bí
3 S FORMAL gá-áde%=é gu-ude%=é rá-áde%=é
3 S NON-FORMAL gá-áda) % gu-uda) % rá-áda) %
1 P INCLUSIVE gá-áda%=rí$u% gu-uda%=rí$u% rá-áda%=rí$u%
1 P EXCLUSIVE gá-áda%=tu%$ gu-uda%=tu%$ rá-áda%=tu%$
2 P INFORMAL gá-áda%=lé gu-uda%=lé rá-áda%=lé
2 P FORMAL gá-áda%=lé gu-uda%=lé rá-áda%=lé
3 P FAMILIAR gá-áda%=ga%bí gu-uda%=ga%bí rá-áda%=ga%bí
3 P FORMAL gá-áda%=gé gu-uda%=gé rá-áda%=gé
3 P NON-FORMAL gá-áda%=ga) # gu-uda%=ga) # rá-áda%=ga) #
Notes: the /u/ aspect marker also causes the verb root’s first vowel to change.
386
Class: 7 (IV-A)
English: ‘to get dressed’
Spanish: ‘vestirse’
Posited verb root: / kku#$ + 'o!" /
Aspect markers: /gá/, /gu/, /rá/
Nellis and Nellis page: 70
POTENTIAL COMPLETIVE HABITUAL
1 S gá-kkú$-'ó=á$ gú-kkú$-'ó=á$ rá-kkú$-'ó=á$
2 S INFORMAL gá-kku#$-'o!"=lu%$ gu-kku#$-'o!"=lu%$ rá-kku#$-'o!"=lu%$
2 S FORMAL gá-kku#$-'o!"=é gu-kku#$-'o!"=é rá-kku#$-'o!"=é
3 S FAMILIAR gá-kku#$-'o!"=bí gu-kku#$-'o!"=bí rá-kku#$-'o!"=bí
3 S FORMAL gá-kku#$-'o!"=é gu-kku#$-'o!"=é rá-kku#$-'o!"=é
3 S NON-FORMAL gá-kku#$-'ú=i) #
gu-kku#$-'ú=i) #
rá-kku#$-'ú=i) #
1 P INCLUSIVE gá-kku#$-'o!"=rí$u% gu-kku#$-'o!"=rí$u% rá-kku#$-'o!"=rí$u%
1 P EXCLUSIVE gá-kku#$-'o!"=tu%$ gu-kku#$-'o!"=tu%$ rá-kku#$-'o!"=tu%$
2 P INFORMAL gá-kku#$-'o!"=lé gu-kku#$-'o!"=lé rá-kku#$-'o!"=lé
2 P FORMAL gá-kku#$-'o!"=lé gu-kku#$-'o!"=lé rá-kku#$-'o!"=lé
3
P FAMILIAR gá-kku#$-'o!"=ga%bí gu-kku#$-'o!"=ga%bí rá-kku#$-'o!"=ga%bí
3
P FORMAL gá-kku#$-'o!"=gé gu-kku#$-'o!"=gé rá-kku#$-'o!"=gé
3 P NON-FORMAL gá-kku#$-'o!"=gi) # gu-kku#$-'o!"=gi) # rá-kku#$-'o!"=gi) #
Notes: This verb includes an obligatorily possessed, inalienable noun.
387
Class: 7 (IV-A)
English: ‘to get up’
Spanish: ‘levantarse’
Posited verb root: / ++a%/
Aspect markers: / t(á/, /gu/, /rá/
Nellis and Nellis page: 46
POTENTIAL COMPLETIVE HABITUAL
1 S t(á-++á=$ gú-++á=$ rá-++á=$
2 S INFORMAL t(á-++a%=lu%$ gu-++a%=lu%$ rá-++a%=lu%$
2 S FORMAL t(á-++é gu-++é rá-++é
3 S FAMILIAR t(á-++a%=bí gu-++a%=bí rá-++a%=bí
3 S FORMAL t(á-++é gu-++é rá-++é
3 S NON-FORMAL t(á-++a) gu-++a) rá-++a)
1 P INCLUSIVE t(á-++a%=rí$u% gu-++a%=rí$u% rá-++a%=rí$u%
1 P EXCLUSIVE t(á-++a%=tu%$ gu-++a%=tu%$ rá-++a%=tu%$
2 P INFORMAL t(á-++a%=lé gu-++a%=lé rá-++a%=lé
2 P FORMAL t(á-++a%=lé gu-++a%=lé rá-++a%=lé
3 P FAMILIAR t(á-++a%=ga%bí gu-++a%=ga%bí rá-++a%=ga%bí
3
P FORMAL t(á-++a%=gé gu-++a%=gé rá-++a%=gé
3 P NON-FORMAL t(á-++a%=ga) # gu-++a%=ga) # rá-++a%=ga) #
388
Class: 8 (IVB)
English: ‘to enter’
Spanish: ‘entrar’
Posited verb root: /a#$a%/
Aspect markers: / g + H tone /, /gu%/, /r/
Nellis and Nellis page: 70
POTENTIAL COMPLETIVE HABITUAL
1 S gá$=a#$ gu%-tá$=a#$ r-á$=a#$
2 S INFORMAL gá$a%=lu%$ gu%-ta#$a%=lu%$ r-a#$a%=lu%$
2 S FORMAL gá$=é gu%-ta#$=é r-a#$=é
3 S FAMILIAR gá$a%=bí gu%-ta#$a%=bí r-a#$a%=bí
3 S FORMAL gá$=é gu%-ta#$=é r-a#$=é
3 S NON-FORMAL gá$=a)) gu%-ta#$=a) r-a#$=a)
1 P INCLUSIVE gá$a%=rí$u% gu%-ta#$a%=rí$u% r-a#$a%=rí$u%
1 P EXCLUSIVE gá$a%=tu%$ gu%-ta#$a%=tu%$ r-a#$a%=tu%$
2 P INFORMAL gá$a%=lé gu%-ta#$a%=lé r-a#$a%=lé
2 P FORMAL gá$a%=lé gu%t-a#$a%=lé r-a#$a%=lé
3 P FAMILIAR gá$a%=ga%bí gu%-ta#$a%=ga%bí r-a#$a%=ga%bí
3 P FORMAL gá$a%=gé gu%-ta#$a%=gé r-a#$a%=gé
3 P NON-FORMAL gá$a%=ga) # gu%-ta#$a%=ga) # r-a#$a%%=ga) #
Notes: In the potential, the H tone of the potential replaces an underlying L lexical tone.
Also: segmental change – addition of /t/ in the completive. Perhaps the stem is
different in the completive.
389
Class: 8 (IVB)
English: ‘to eat’
Spanish: ‘comer’
Posited verb root: /o/
Aspect markers: / g + H tone /, /gu%/, /r/
Nellis and Nellis page: 77
POTENTIAL COMPLETIVE HABITUAL
1 S g-ó=á$ u%-tó=á$ r-ó=á$
2 S INFORMAL g-ó=lú$ u%-to%=lu%$ r-o=lu%$
2 S FORMAL g-ó=é u%-to%=é r-o=é
3 S FAMILIAR g-ó=bí u%-to%=bí r-o=bí
3 S FORMAL g-ó=é u%-to%=é r-o=é
3 S NON-FORMAL g-ó=a) .
u%-to%=a) % r-o=a)
1 P INCLUSIVE g-ó=rí$u% u%-to%=rí$u% r-o=rí$u%
1 P EXCLUSIVE g-ó=tú$ u%-to%=tu%$ r-o=tu%$
2 P INFORMAL g-ó=lé u%-to%=lé r-o=lé
2 P FORMAL g-ó=lé u%-to%=lé r-o=lé
3 P FAMILIAR g-ó=gábí u%-to%=ga%bí r-o=ga%bí
3 P FORMAL g-ó=gé u%-to%=gé r-o=gé
3 P NON-FORMAL g-ó=ga) . u%-to%=ga) # r-o=ga) #
Notes: Addition of /t/ in the completive, or possibly the stem is different in the
completive.
390
Class: 8 (IVB)
English: ‘to call’
Spanish: ‘llamar’
Posited verb root: /a%"'i# /
Aspect markers: / g + H tone /, /u%/, /r/
Nellis and Nellis page: 77
POTENTIAL COMPLETIVE HABITUAL
1 S g-a!"'i#=a#$ u%-ta!"'i#=a#$ r-a!"'i#=a#$
2 S INFORMAL g-a!"'i#=lu%$ u%-ta%"'i#=lu%$ r-a%'i#=lu%$
2 S FORMAL g-a!"'i#=é u%-ta%"'i#=é r-a%'i#=é
3 S FAMILIAR g-a!"'i#=bí u%-ta%"'i#=bí r-a%'i#=bí
3 S FORMAL g-a!"'i#=é u%-ta%"'i#=é r-a%'i#=é
3 S NON-FORMAL g-a!"'i#=a
# )
u%-ta%'i#=a) # r-a%'i#=a) #
1 P INCLUSIVE g-a!"'i#=rí$u% u%-ta%"'i#=rí$u% r-a%'i#=rí$u%
1 P EXCLUSIVE g-a!"'i#=tu%$ u%-ta%"'i#=tu%$ r-a%'i#=tu%$
2 P INFORMAL g-a!"'i#=lé u%-ta%"'i#=lé r-a%'i#=lé
2 P FORMAL g-a!"'i#=lé u%-ta%"'i#=lé r-a%'i#=lé
3 P FAMILIAR g-a!"'i#=ga%bí u%-ta%"'i#=ga%bí r-a%'i#=ga%bí
3 P FORMAL g-a!"'i#=gé u%-ta%"'i#=gé r-a%'i#=gé
3 P NON-FORMAL g-a!"'i#=ga) # u%-ta%"'i#=ga) # r-a%'i#=ga) #
Notes: segmental change in that /t/ is added in the completive.
391
Class: 8 (IVB)
English: ‘to sleep’
Spanish: ‘dormir’
Posited verb root: /á++i%/
Aspect markers: / g + H tone /, /u%t/, /r/
Nellis and Nellis page: 70
POTENTIAL COMPLETIVE HABITUAL
1 S g-á++i%=a%$ u%-tá++i%=a%$ r-á++i%=a%$
2 S INFORMAL g-á++i%=lu%$ u%-tá++i%=lu%$ r-á++i%=lu%$
2 S FORMAL g-á++i%=é u%-tá++i%=é r-á++i%=é
3 S FAMILIAR gá++i%=bí u%-tá++i%=bí rá++i%=bí
3 S FORMAL g-á++i%=é u%-tá++i%=é r-á++i%=é
3 S NON-FORMAL g-á++i%=a
% ))
u%-tá++i%=a) % r-á++i%=a) %
1 P INCLUSIVE g-á++i%=rí$u% u%-tá++i%=rí$u% r-á++i%=rí$u%
1 P EXCLUSIVE g-á++i%=tu%$ u%-tá++i%=tu%$ r-á++i%=tu%$
2 P INFORMAL g-á++i%=lé u%-tá++i%=lé r-á++i%=lé
2 P FORMAL g-á++i%=lé u%-tá++i%=lé r-á++i%=lé
3 P FAMILIAR g-á++i%=ga%bí u%-tá++i%=ga%bí r-á++i%=ga%bí
3 P FORMAL g-á++i%=gé u%-tá++i%=gé r-á++i%=gé
3 P NON-FORMAL g-á++i%=ga) # u%-tá++i%=ga) # r-á++i%=ga) #
Notes: Again, /t/ epenthetic in the completive or possibly a different stem in the
completive.
392
Class: 9 (VA)
English: ‘to go’
Spanish: ‘ir’
Posited verb root: /ía/
Aspect markers: / ts + H tone /, /g
w
/, /d/
Nellis and Nellis page: 214
POTENTIAL COMPLETIVE HABITUAL
1 S tsá=$ g
w
á=$ día=$
2 S INFORMAL tsíá=lú$ g
w
ía=lu%$ díá=lú$
2 S FORMAL tsí=é g
w
í=é dí=é
3 S FAMILIAR tsíá=bí g
w
ía=bí díá=bí
3 S FORMAL tsí=é g
w
í=é dí=é
3 S NON-FORMAL tsía) .
g
w
ía) . día) .
1 P INCLUSIVE tsíá=rí$u% g
w
ía=rí$u% díá=rí$u%
1 P EXCLUSIVE tsíá=tu%$ g
w
ía=tu%$ díá=tú$
2 P INFORMAL tsíá=lé g
w
ía=lé díá=lé
2 P FORMAL tsíá=lé g
w
ía=lé díá=lé
3 P FAMILIAR tsíá=ga%bí g
w
ía=ga%bí díá=gábí
3 P FORMAL tsíá=gé g
w
ía=gé díá=gé
3 P NON-FORMAL tsíá=ga) # g
w
ía=ga) # díá=ga) #
Notes: The tone of the second syllable seems to differ in the completive. In the potential
and habitual this vowel has a H tone –in the completive the second syllable emerges
as M.
393
Class: 9 (VA)
English: ‘to burn oneself’
Spanish: ‘quemarse’
Posited verb root: / a##"ji% /
Aspect markers: / ts + H tone /, /g
w
/, /r/
Nellis and Nellis page: 211
POTENTIAL COMPLETIVE HABITUAL
1 S ts-á"j=a#$ g
w
-a!"j=a#$ r-a!"ja#$
2 S INFORMAL ts-á"ji%=lu%$ g
w
-a##"ji%=lu%$ r-a#"ji%=lu%$
2 S FORMAL ts-á"jé g
w
-a#"jé r-a##"jé
3 S FAMILIAR ts-á"ji%=bí g
w
-a##"ji%=bí r-a##"ji%=bí
3 S FORMAL ts-á"jé g
w
-a##"jé r-a##"jé
3 S NON-FORMAL ts-á"ja) . g
w
-a#"ja
# ) r-a#"ja) #
1 P INCLUSIVE ts-á"ji%=rí$u% g
w
-a##"ji%=rí$u% r-a#"ji%=rí$u%
1 P EXCLUSIVE ts-á"ji%=tu%$ g
w
-a##"ji%=tu%$ r-a#"ji%=tu%$
2 P INFORMAL ts-á"ji%=lé g
w
-a##"ji%=lé r-a##"ji%=lé
2 P FORMAL ts-á"ji%=lé g
w
-a##"ji%=lé r-a#"ji%=lé
3 P FAMILIAR ts-á"ji%=ga%bí g
w
-a#"ji%=ga%bí r-a#"ji%=ga%bí
3 P FORMAL ts-á"ji%=gé g
w
-a#"ji%=gé r-a#"ji%=gé
3 P NON-FORMAL ts-á"ji%=ga) # g
w
-a##"ji%=ga) # r-a#"ji%=ga) #
Notes: The H tone associated with the potential aspect replaces the underlying tone of
the verb root.
394
Class: 9 (VA)
English: ‘to climb up’
Spanish: ‘subirse’
Posited verb root: / appi% /
Aspect markers: / ts + H tone /, /w/, /r/
Nellis and Nellis page: 211
POTENTIAL COMPLETIVE HABITUAL
1 S ts-áppi%=a%$ w-áppi%=a%$ r-áppi%=a%$
2 S INFORMAL ts-áppi%=lu%$ w-appi%=lu%$ r-appi%=lu%$
2 S FORMAL ts-áppi%=é w-appi%=é r-appi%=é
3 S FAMILIAR ts-áppi%=bí w-appi%=bí r-appi%=bí
3 S FORMAL ts-áppi%=é w-appi%=é r-appi%=é
3 S NON-FORMAL ts-áppi%=a
% ) w-appi%=a
% ) r-appi%=a) %
1 P INCLUSIVE ts-áppi%=rí$u% w-appi%=rí$u% r-appi%=rí$u%
1 P EXCLUSIVE ts-áppi%=tu%$ w-appi%=tu%$ r-appi%=tu%$
2 P INFORMAL ts-áppi%=lé w-appi%=lé r-appi%=lé
2 P FORMAL ts-áppi%=lé w-appi%=lé r-appi%=lé
3 P FAMILIAR ts-áppi%=ga%bí w-appi%=ga%bí r-appi%%=ga%bí
3 P FORMAL ts-áppi%=gé w-appi%=gé r-appi%=gé
3 P NON-FORMAL ts-áppi%=ga) # w-appi%=ga) # r-appi%=ga) #
395
Class: 10 (VB)
English: ‘to strike or beat’
Spanish: ‘azotar’
Posited verb root: / á$ /
Aspect markers: / g+ H tone /, /w/, /r/
Nellis and Nellis page: 69
POTENTIAL COMPLETIVE HABITUAL
1 S gá$=a#$ wí$=a#$ rá$=a#$
2 S INFORMAL gá$=lu%$ wí$=lu%$ rá$=lu%$
2 S FORMAL gá$=é wí$=é ré$=é
3 S FAMILIAR gá$=bí wí$=bí rá$=bí
3 S FORMAL gá$=é wí$=é ré$=é
3 S NON-FORMAL gá$=a) wí$=a) rá$=a)
1 P INCLUSIVE gá$=rí$u% wí$=rí$u% rá$=rí$u%
1 P EXCLUSIVE gá$=tu%$ wí$=tu%$ rá$=tu%$
2 P INFORMAL gá$=lé wí$=lé rá$=lé
2 P FORMAL gá$=lé wí$=lé rá$=lé
3 P FAMILIAR gá$=ga%bí wí$=ga%bí rá$=ga%bí
3 P FORMAL gá$=gé wí$=gé rá$=gé
3 P NON-FORMAL gá$=ga) # wí$=ga
# ) rá$=ga) #
Notes: A segmental change in the completive aspect, and for the 2 S FORMAL in the
habitual aspect.
396
Class: 10 (VB)
English: ‘to buy’
Spanish: ‘comprar’
Posited verb root: /o$o%/
Aspect markers: / g+ H tone /, /g
w
/, /r/
Nellis and Nellis page: 78
POTENTIAL COMPLETIVE HABITUAL
1 S g-ó$=a#$ g
w
-í$=a#$ r-ó$=a#$
2 S INFORMAL g-ó$o%=lu%$ g
w
-i$i%=lu%$ r-o$o%=lu%$
2 S FORMAL g-ó$=é g
w
-i$=é r-o$=é
3 S FAMILIAR g-ó$o%=bí g
w
-i$i%=bí r-o$o%=bí
3 S FORMAL g-ó$=é g
w
-i$=é r-o$=é
3 S NON-FORMAL g-ó$=a) g
w
-i$=a) r-o$=a)
1 P INCLUSIVE g-ó$o%=rí$u% g
w
-i$i%=rí$u% r-o$o%=rí$u%
1 P EXCLUSIVE g-ó$o%=tu%$ g
w
-i$i%=tu%$ r-o$o%=tu%$
2 P INFORMAL g-ó$o%=lé g
w
-i$i%=lé r-o$o%=lé
2 P FORMAL g-ó$o%=lé g
w
-i$i%=lé r-o$o%=lé
3 P FAMILIAR g-ó$o%=ga%bí g
w
-i$i%=ga%bí r-o$o%=ga%bí
3 P FORMAL g-ó$o%=gé g
w
-i$i%=gé r-o$o%=gé
3 P NON-FORMAL g-ó$o%=ga) # g
w
-i$i%=ga) # r-o$o%=ga) #
Notes: Again, there are segmental changes in the completive aspect.
397
Class: 11 (VI)
English: ‘to love’
Spanish: ‘querer’
Posited verb root: /a% + la%tsi#$/
Aspect markers: / k + H tone /, /k
w
/, /k/
Nellis and Nellis page: 31
POTENTIAL COMPLETIVE HABITUAL
1 S ká-látsa#$ k
w
a%-látsa#$ ka%-látsa#$
2 S INFORMAL ká-la%tsi#$=lu%$ k
w
a%-la%tsi#$=lu%$ ka%-la%tsi#$=lu%$
2 S FORMAL ká-la%tsi#$=é k
w
a%-la%tsi#$=é ka%-la%tsi#$=é
3 S FAMILIAR ká-la%tsi#$=bí k
w
a%-la%tsi#$=bí ka%-la%tsi#$=bí
3 S FORMAL ká-la%tsi#$=é k
w
a%-la%tsi#$=é ka%-la%tsi#$=é
3 S NON-FORMAL ká-la%tsi#$=i) # k
w
a%-la%tsi#$=i) # ka%-la%tsi#$=i) #
1 P INCLUSIVE ká-la%tsi#$=rí$u% k
w
a%-la%tsi#$=rí$u% ka%-la%tsi#$=rí$u%
1 P EXCLUSIVE ká-la%tsi#$=tu%$ k
w
a%-la%tsi#$=tu%$ ka%-la%tsi#$=tu%$
2 P INFORMAL ká-la%tsi#$=lé k
w
a%-la%tsi#$=lé ka%-la%tsi#$=lé
2 P FORMAL ká-la%tsi#$=lé k
w
a%-la%tsi#$=lé ka%-la%tsi#$=lé
3 P FAMILIAR ká-la%tsi#$=ga%bí k
w
a%-la%tsi#$=ga%bí ka%-la%tsi#$=ga%bí
3 P FORMAL ká-la%tsi#$=gé k
w
a%-la%tsi#$=gé ka%-la%tsi#$=gé
3 P NON-FORMAL ká-la%tsi#$=gi) # k
w
a%-la%tsi#$=gi) # ka%-la%tsi#$=gi) #
Notes: the noun /la%tsi#$/ is obligatorily possessed, hence the different markings for the
third person singular and plural non-formal. This verb takes genitive subjects.
398
Class: 11 (VI)
English: ‘to be’
Spanish: ‘ser’
Posited verb root: /a/
Aspect markers: / kká /, /ek
w
/, /n/
Nellis and Nellis page: 29
POTENTIAL COMPLETIVE HABITUAL
1 S kká=$ ek
w
a=$ ná=$
2 S INFORMAL kká=lú$ ek
w
a=lu%$ ná=lú$
2 S FORMAL ekké=e k
w
é née
3 S FAMILIAR kká=bí ek
w
a=bí ná=bí
3 S FORMAL ekké=e k
w
é née
3 S NON-FORMAL kka) . ek
w
a) náa)
1 P INCLUSIVE ekká=rí$u# k
w
a=rí$u% ná=rí$u#
1 P EXCLUSIVE kká=tú$ k
w
a=tu%$ ná=tú$
2 P INFORMAL kká=lé k
w
a=lé ná=lé
2 P FORMAL kká=lé k
w
a=lé ná=lé
3 P FAMILIAR kká=ga%bí k
w
a=ga%bí ná=ga%bí
3 P FORMAL kká=gé k
w
a=gé ná=gé
3 P NON-FORMAL kká=ga) . k
w
a=ga) # ná=ga) .
Notes: Irregularity in aspect markers especially in the potential and completive
399
Class: 12 (VII)
English: ‘to visit’
Spanish: 'visitar a’
Posited verb root: / ne#llé/
Aspect markers: / tá /, /ta%/, /rita/
Nellis and Nellis page: 123
POTENTIAL COMPLETIVE HABITUAL
1 S tá-néllí=á$ ta%-néllí=á$ rítá-néllí=á$
2 S INFORMAL tá-ne#llé=lú$ ta%-ne#llé=lú$ rita-ne#llé=lú$
2 S FORMAL tá-ne#llí=e ta%-ne#llí=e rita-ne#llí=e
3 S FAMILIAR tá-ne#llé=bí ta%-ne#llé=bí rita-ne#llé=bí
3 S FORMAL tá-ne#llí=e ta%-ne#llí=e rita-ne#llí=e
3 S NON-FORMAL tá-ne#llí=a) . ta%-ne#llí=a) . rita-ne#llí=a) .
1 P INCLUSIVE tá-ne#llé=rí$u% ta%-ne#llé=rí$u% rita-ne#llé=rí$u%
1 P EXCLUSIVE tá-ne#llé=tú$ ta%-ne#llé=tú$ rita-ne#llé=tú$
2 P INFORMAL tá-ne#llé=lé ta%-ne#llé=lé rita-ne#llé=lé
2 P FORMAL tá-ne#llé=lé ta%-ne#llé=lé rita-ne#llé=lé
3 P FAMILIAR tá-ne#llé=ga%bí ta%-ne#llé=ga%bí rita-ne#llé=ga%bí
3 P FORMAL tá-ne#llé=gé ta%-ne#llé=gé rita-ne#llé=gé
3 P NON-FORMAL tá-ne#llé=ga) . ta%-ne#llé=ga) . rita-ne#llé=ga) .
400
Class: 12 (VII)
English: ‘to find unexpectedly’
Spanish: ‘hallar inesperadamente’
Posited verb root: / &ákka#$/
Aspect markers: / tá /, /ta%/, /rita/
Nellis and Nellis page: 193
POTENTIAL COMPLETIVE HABITUAL
1 S tá-&ákka#$=a#$ ta%-&ákka#$=a#$ rítá-&ákka#$=a#$
2 S INFORMAL tá-&ákka#$=lu%$ ta%-&ákka#$=lu%$ rita-&ákka#$=lu%$
2 S FORMAL tá-&ákka#$=é ta%-&ákka#$=é rita-&ákka#$=é
3 S FAMILIAR tá-&ákka#$=bí ta%-&ákka#$=a) rita-&ákka#$=a)
3 S FORMAL tá-&ákka#$=é ta%-&ákka#$=é rita-&ákka#$=é
3 S NON-FORMAL tá-&ákka#$=a) ta%-&ákka#$=a) rita-&ákka#$=a)
1 P INCLUSIVE tá-&ákka#$=rí$u% ta%-&ákka#$=rí$u% rita-&ákka#$=rí$u%
1 P EXCLUSIVE tá-&ákka#$=tu%$ ta%-&ákka#$=tu%$ rita-&ákka#$=tu%$
2 P INFORMAL tá-&ákka#$=lé ta%-&ákka#$=lé rita-&ákka#$=lé
2 P FORMAL tá-&ákka#$=lé ta%-&ákka#$=lé rita-&ákka#$=lé
3 P FAMILIAR tá-&ákka#$=ga%bí ta%-&ákka#$=ga%bí rita-&ákka#$=ga%bí
3 P FORMAL tá-&ákka#$=gé ta%-&ákka#$=gé rita-&ákka#$=gé
3 P NON-FORMAL tá-&ákka#$=ga) #
ta%-&ákka#$=ga) #
rita-&ákka#$=ga) #
Abstract (if available)
Abstract
This dissertation examines floating tones and tone gesture deactivation in Sierra Juárez Zapotec (SJZ), and provides an Optimality Theoretic account of tonal spreading and placement using insights from Articulatory Phonology. While the data portion of the dissertation is drawn from SJZ, the approach has broader implications for theories of tonal representation and analysis. ❧ Constraints that prohibit tone deactivation except in specific locations such as at an utterance boundary, at a following anti-phase coupled tone, or with the vowel of a stressed syllable are proposed. Constraints that reference the coupling point (onset) of a tone are proposed which can draw a tone gesture to couple to a prominent position such as the initial syllable. Constraints are also proposed which prohibit the overlap of activation of tone gestures with other vowel or consonant gestures. In this way, the dissertation proposes that both the onset and endpoint of a tone may be controlled via constraints in the grammar. ❧ Fieldwork by the author is presented which illustrates that in SJZ the first-person singular form of transitive verbs has a high tone that does not appear on the first-person singular suffix. Instead, this high tone appears on the stressed syllable of the verb root, and, in cases where the aspect marker is toneless, on the aspect marker as well as the stressed syllable of the verb root. Because the tone appears in locations separate from the first-person singular suffix, it is called a floating tone. The constraints proposed in the dissertation account for the placement of this floating tone while making testable predictions for other languages. ❧ The proposed constraints suggest that tone spreading should normally be rightward due to the unique deactivation properties that tone gestures possess (as distinct from consonant or vowel gestures). Leftward spreading is predicted to occur only when a tone is drawn to couple to a prominent position such as with the vowel of the initial syllable. In cases where a tone changes its coupling (onset) and then spreads rightward, the output pattern will appear to be bidirectional. In cases where the onset of a tone does not move, spreading is predicted to be rightward until the tone encounters a deactivator. The predictions made by the constraints are tested via OT analyses of other languages. Several areas where these constraints could be applied in future research are discussed.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Harmony in gestural phonology
PDF
The Spanish feminine el at the syntax-phonology interface
PDF
Sources of non-conformity in phonology: variation and exceptionality in Modern Hebrew spirantization
PDF
Minimal contrast and the phonology-phonetics interaction
PDF
Sound sequence adaptation in loanword phonology
PDF
Beatboxing phonology
PDF
Dynamics of consonant reduction
PDF
Articulatory dynamics and stability in multi-gesture complexes
PDF
Individual differences in phonetic variability and phonological representation
PDF
Prosodic recursion and syntactic cyclicity inside the word
PDF
Syntax-prosody interactions in the clausal domain: head movement and coalescence
PDF
Effects of speech context on characteristics of manual gesture
PDF
Speech production in post-glossectomy speakers: articulatory preservation and compensation
PDF
Prosody in contact: Spanish in Los Angeles
PDF
The phonological dimension of grammatical markedness
PDF
Soft biases in phonology: learnability meets grammar
PDF
Generalized surface correspondence in reduplicative opacity
PDF
Signs of skilled adaptation in the co-speech ticking of adults with Tourette's
PDF
Binding and scope dependencies with 'floating quantifiers' in Japanese
PDF
The phonology and phonetics of Turkish intonation
Asset Metadata
Creator
Tejada, Laura
(author)
Core Title
Tone gestures and constraint interaction in Sierra Juarez Zapotec
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Linguistics
Publication Date
04/24/2012
Defense Date
04/24/2012
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
Mexico,OAI-PMH Harvest,phonetics,phonology,zapotec
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Walker, Rachel (
committee chair
), Frazier, Melissa (
committee member
), Goldstein, Louis (
committee member
), Munro, Pamela (
committee member
), Saltarelli, Mario (
committee member
)
Creator Email
lauratejada3@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-10202
Unique identifier
UC11289268
Identifier
usctheses-c3-10202 (legacy record id)
Legacy Identifier
etd-TejadaLaur-628.pdf
Dmrecord
10202
Document Type
Dissertation
Rights
Tejada, Laura
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
phonetics
phonology
zapotec