PROMINENCE IN YUCATEC MAYA: Emily Kidder

PROMINENCE IN YUCATEC MAYA: Emily Kidder
PROMINENCE IN YUCATEC MAYA:
THE ROLE OF STRESS IN YUCATEC MAYA WORDS
by
Emily Kidder
__________________________
Copyright © Emily Kidder 2013
A Dissertation Submitted to the Faculty of the
DEPARTMENTS OF ANTHROPOLOGY AND LINGUISTICS
In Partial Fulfillment of the Requirements
For the Degree of
DOCTOR OF PHILOSOPHY
In the Graduate College
THE UNIVERSITY OF ARIZONA
2013
2 THE UNIVERSITY OF ARIZONA
GRADUATE COLLEGE
As members of the Dissertation Committee, we certify that we have read the dissertation
prepared by Emily Kidder, titled Prominence in Yucatec Maya: The role of stress in
Yucatec Maya words and recommend that it be accepted as fulfilling the dissertation
requirement for the Degree of Doctor of Philosophy.
_______________________________________________________________________
Date: 4/8/2013
Michael Hammond
_______________________________________________________________________
Date: 4/8/2013
Norma Mendoza-Denton
_______________________________________________________________________
Date: 4/8/2013
Jane Hill
_______________________________________________________________________
Date: 4/8/2013
Stacey Oberly
_______________________________________________________________________
Date: 4/8/2013
Amy Fountain
Final approval and acceptance of this dissertation is contingent upon the candidate’s
submission of the final copies of the dissertation to the Graduate College.
I hereby certify that I have read this dissertation prepared under my direction and
recommend that it be accepted as fulfilling the dissertation requirement.
________________________________________________ Date: 4/8/2013
Dissertation Director: Michael Hammond
3 STATEMENT BY AUTHOR
This dissertation has been submitted in partial fulfillment of the requirements for
an advanced degree at the University of Arizona and is deposited in the University
Library to be made available to borrowers under rules of the Library.
Brief quotations from this dissertation are allowable without special permission,
provided that an accurate acknowledgement of the source is made. Requests for
permission for extended quotation from or reproduction of this manuscript in whole or in
part may be granted by the copyright holder.
SIGNED: Emily Kidder
4 ACKNOWLEDGEMENTS
There were so many moments during which I was almost convinced that the completion
of this dissertation would never come to pass, and I might well have been totally
convinced if not for the many people whose help, advice, encouragement and patience
helped it happen.
First and foremost, thank you to my amazing husband Paul, who has been so incredibly
supportive throughout the last decade of grad school, work and dissertation-writing
insanity. Without your cheerful encouragement, invaluable statistics advice, and
unending patience for weekends and evenings spent writing, there is no doubt in my mind
that I would have given up long ago. I am so lucky to have you. Now we can finally go
see that movie! And go hiking, and to the beach!
Thanks to my family, especially mom and dad, for being such fun and interesting people,
for making me think learning was fun, and that other cultures and languages were
fascinating, for putting me through college and encouraging me to do what I love, and for
being almost happier than I am that this dissertation is a reality. Thanks as well to my
brother and sister-in-law Dan and Erica and my adorable nephews, and to everyone else
in my family for all their support, you guys are the greatest.
A million thanks to my committee members; Mike Hammond, Jane Hill, Norma
Mendoza-Denton, Amy Fountain and Stacey Oberly. Without Mike’s support and
guidance, I would undoubtedly have been left sitting at my computer dumbfounded by
the sheer magnitude of data in front of me, with no idea how to tackle it. Working on this
dissertation remotely while working full time was a huge challenge, and Mike’s prompt
and encouraging communications were instrumental in helping me get this done despite
being far away and working. Jane Hill, thank you for your inspiration and encouragement
over the years. I am lucky and honored to be among the last advisees to benefit from your
knowledge, humor and general awesomeness. Thanks to Norma Mendoza-Denton for
keeping me challenged and giving me such great learning opportunities at the Ling-Anth
lab during my early years in grad school. Thanks to Stacey Oberly for the support and
advice regarding data collection and fieldwork in general, and many thanks as well to
Amy Fountain who is the ultimate encourager, and who always thought my ideas were
cool well before I did.
Thanks also to Alberto Perez Rendon from Asociación MAYAB, and all the folks at
Chan Kahal, both of which do amazing work for the Mayan communities in the Bay
Area, for helping me get to know the community of Mayan speakers and helping to
facilitate data collection. Many thanks as well to the folks at the Yucatec Maya Institute
at the University of North Carolina, Chapel Hill and the Foreign Language Area Studies
Fellowship program for allowing me to begin my study of Yucatec Maya many years
ago. And thank you to Martha Macri from UC Davis for teaching the linguistics course
that got me hooked.
5 Thank you so much to all the amazing friends and family that have encouraged me and
kept me going all these years. Many thanks to Serge for always being there, for the
unwavering support, and for all the reminders that I really would become Dr. Em
someday. Thanks to cousin Ben for being so inspiring, amazing and full of energy.
Thanks to my best friend Christina and her family for being my second family. Thanks to
everybody at work for being my surrogate nerdy family these last 6 years. Especially
thanks to Charlise Tiee and Scott Grieder for their flexibility and support, and to Nina
Chan, Kim Corry, Chuck Smith, Holly Boston, Nyesha Petrushansky, Lauren McFall,
Francesca Sapien and everyone else for being the coolest coworkers ever. Thanks to
Jaime Parchment, for all the linguistic debates which often ended in one of us having
some kind of breakthrough or epiphany, and for the advice and encouragement over the
years. Thanks also to Jeff Punske for letting me crash on his couch during all those trips
back to Tucson, and to Shannon Bischoff for all the cheerful emails of ‘are you done
yet???’, as well as to Dave Medeiros, Scott and Heather Jackson, Dan and Julie Siddiqi,
Michael Anderson, Amy LaCross, Hannah Jones, AJ Vonarx and Damon Vukasovic,
Peter and Erin Richtsmeier, Erin Good-Ament, and everyone else at U of A that shared in
all the good times. I’m sure there are people I have missed and I am sure I’ll feel bad
about that later, but in general if you’re reading this, thank you!
And finally, thanks to my adorable pups Mooch and Oliver, for constantly sitting directly
on my laptop keyboard, staring over the computer screen at me with big sad eyes, and
pawing at my hands as they typed these words. You were right, you are way more
interesting than this piece of machinery, and now hopefully you won’t have to spend so
much time befuddled by how I could stare at it instead of playing with you.
6 DEDICATION
For Paul, my favorite person,
and for Moochie and Ollie, my favorite fluffballs.
7 TABLE OF CONTENTS
LIST OF FIGURES .......................................................................................................... 9
LIST OF TABLES .......................................................................................................... 11
ABSTRACT ..................................................................................................................... 12
CHAPTER 1 INTRODUCTION.................................................................................. 13
1.1 Outline of Dissertation ................................................................................................... 14
1.2 Theoretical Background ................................................................................................ 15
1.3 Perceptual Cues of Stress .............................................................................................. 16
1.4 Definition of Stress ......................................................................................................... 17
1.4.1 Lexical Stress vs. Metrical Stress ............................................................................. 18
1.4.2 Rhythmic vs. Morphological Stress .......................................................................... 19
1.5 Intonation, Stress and Accent ....................................................................................... 19
1.6 Syntagmatic Nature of Stress ........................................................................................ 21
1.7 Grouping Properties of Stress ....................................................................................... 23
1.8 Stress Cross-Linguistically ............................................................................................ 24
1.9 Stress in Yucatec Maya.................................................................................................. 26
CHAPTER 2 OVERVIEW OF YUCATEC MAYA & METHODOLOGY ............ 28
2.1 Bilingualism with Spanish ............................................................................................. 29
2.2 Yucatecan Spanish Dialect ............................................................................................ 29
2.3 Immigration to San Francisco Bay Area ..................................................................... 31
2.4 Methodology of Current Study ..................................................................................... 32
2.5 Yucatec Maya Phonology, Phonotactics and Morphophonemics .............................. 37
2.5.1 Sound Inventory ........................................................................................................ 37
2.5.2 Orthography .............................................................................................................. 45
2.5.3 Phonotactics .............................................................................................................. 45
2.6 Hiatus Avoidance and Pronouns .................................................................................. 47
2.7 Syllable Structure ........................................................................................................... 49
2.7.1 Syllabification ........................................................................................................... 49
CHAPTER 3 PREVIOUS WORK IN YUCATEC MAYA PROSODY................... 54
3.1 Introduction .................................................................................................................... 54
3.2 History of Spanish impact on YM ................................................................................ 55
3.3 Early Descriptions of Yucatec Prosody ........................................................................ 57
3.3.1 Tozzer’s 1921 Account of Yucatec Phonetics .......................................................... 59
3.3.2 Early Observations of the role of pitch in YM .......................................................... 60
3.3.3 Phonemic Pitch ......................................................................................................... 62
3.4 Modern Orthography and Tonal Analysis .................................................................. 64
3.5 Current Account of Phonetics in YM ........................................................................... 67
3.5.1 Frazier (2009) ............................................................................................................ 67
3.5.2 Moras as Tone and Stress Bearing Units .................................................................. 69
3.5.3 Foot Structure in YM ................................................................................................ 72
3.6 Summary of Previous Work in YM .............................................................................. 73
CHAPTER 4 SPANISH LOAN WORD INCORPORATION .................................. 75
8 4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
Introduction .................................................................................................................... 75
Syncretism in Language ................................................................................................ 76
Previous Discussions of Default Stress in YM ............................................................. 78
Loan Word Usage........................................................................................................... 79
The Syncretic Continuum in YM ................................................................................. 81
Pitch in YM Syncretic Shifts ......................................................................................... 83
Length in YM Syncretic Shifts ...................................................................................... 87
Testing the Predictions .................................................................................................. 90
Summary of Spanish Loan Words in YM ................................................................... 94
CHAPTER 5 STRESS IN YUCATEC MAYA ........................................................... 95
5.1 Diagnosing Stress in YM ............................................................................................... 95
5.2 Hypotheses ...................................................................................................................... 96
5.3 Native Speaker Intuitions .............................................................................................. 98
5.3.1 Statistical analysis of Intuition Data ....................................................................... 101
5.3.2 Summary of Intuition Data ..................................................................................... 106
5.4 Phonetic Data from Words in Isolation and Context ............................................... 107
5.4.1 Monosyllabic Words ............................................................................................... 109
5.4.2 Bisyllabic Words ..................................................................................................... 122
5.4.3 Trisyllabic Words .................................................................................................... 141
5.5 Summary of Stress in YM ........................................................................................... 161
CHAPTER 6 CONCLUSION .................................................................................... 162
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
Hypothesis 1: Duration cues Prominence .................................................................. 162
Hypothesis 2: Non-Neutral vowels are prominent .................................................... 165
Hypothesis 3: Weight is prominent ............................................................................ 165
Hypothesis 4: Initial syllables are prominent ............................................................ 166
Hypothesis 5: Phonetic data will accord with speaker intuitions ............................ 166
Comparison to previous research ............................................................................... 167
Areas of future study ................................................................................................... 168
Summary ....................................................................................................................... 168
WORKS CITED............................................................................................................ 171
Figure 1-1
Figure 1-2
Figure 1-3
Figure 1-4
Figure 2-1
Figure 2-2
Figure 2-3
Figure 2-4
Figure 2-5
Figure 2-6
Figure 2-7
Figure 2-8
Figure 2-9
Figure 2-10
Figure 2-11
Figure 2-12
Figure 2-13
Figure 2-14
Figure 2-15
Figure 2-16
Figure 2-17
Figure 2-18
Figure 2-19
Figure 3-1
Figure 3-2
Figure 3-3
Figure 3-4
Figure 3-5
Figure 3-6
Figure 3-7
Figure 3-8
Figure 3-9
Figure 3-10
Figure 3-11
Figure 4-1
Figure 4-2
Figure 4-3
Figure 4-4
Figure 4-5
Figure 4-6
Figure 4-7
Figure 4-8
Figure 4-9
Figure 4-10
9 LIST OF FIGURES
Minimal stress pairs in English ................................................................... 18
Metrical stress patterns in English .............................................................. 18
Example of intonational phrase construction .............................................. 20
Syntagmatic versus Paradigmatic Features (L & R, 2005: 2) .................... 22
Mayan Language Family (Campbell, 1985: 189) ...................................... 28
Map of Yucatán, Mexico ............................................................................. 33
Data Collected ............................................................................................. 34
Example of Elicitation Materials................................................................. 34
Vowel Inventory (Bricker, Po'ot Yah, & Dzul de Po'ot, 1998: xiii) ........... 40
Vowel Articulations (Bricker, Po'ot Yah, & Dzul de Po'ot, 1998: xiii) ...... 41
F0 of High versus Low Tone....................................................................... 42
Short Vowel – bak’ – ‘meat’ ....................................................................... 43
Glottalized Vowel – ba’al – ‘thing’............................................................. 43
Vowel Adjacency ...................................................................................... 44
Orthographic Representation .................................................................... 45
(Orie & Bricker, 2000: 296) ...................................................................... 46
(Orie & Bricker, 2000: 296) ...................................................................... 47
Person markers in YM .............................................................................. 48
Transitive construction .............................................................................. 48
Intransitive Verbs ...................................................................................... 48
Speaker Pauses Between Syllables ........................................................... 50
Spectrogram of /bul-áak'ab/ ...................................................................... 51
Syllabification of Glottalized Vowels ....................................................... 52
Exceptions to Ultimate Stress (Andrade, 1940) .......................................... 61
Comparison of Andrade & Navarette.......................................................... 61
Minimal Tone Pairs (Pike K. L., 1946) ....................................................... 62
Syllable nuclei (Blair & Vermont-Salas, 1965) .......................................... 64
Four Vowel Types of YM ........................................................................... 65
Inventory of Syllabic Nuclei (Straight, 1976: 38) ....................................... 67
Characteristics of the Western Dialect of YM (Frazier, 2009: p 274) ........ 68
Vowel harmony (Krämer, 2001: 11) ........................................................... 70
Stress assignment (Krämer, 2001: 12) ........................................................ 71
Yucatec Trochaic Feet (Krämer, 2001: 13) ............................................. 72
Foot Structure Examples (Krämer, 2001: 12) ........................................... 73
Loans With Spanish and Mayan Morphology............................................. 80
Trisyllabic Loan Words ............................................................................... 81
YM versus Spanish Context ........................................................................ 81
Hypotheses for placement of YM stress ..................................................... 82
Syllable Pitch in ‘trabajo’ ............................................................................ 84
Syllable Pitch in ‘abuelo’ ............................................................................ 84
Syllable Pitch in ‘palabra’ ........................................................................... 85
Syllable pitch in ‘gravilla’ ........................................................................... 86
Visual Representation of Length Measurements ........................................ 88
Syllable Length –‘trabajo’ ........................................................................ 88
Figure 4-11
Figure 4-12
Figure 4-13
Figure 4-14
Figure 4-15
Figure 4-16
Figure 4-17
Figure 4-18
Figure 5-1
Figure 5-2
Figure 5-3
Figure 5-4
Figure 5-5
Figure 5-6
Figure 5-7
Figure 5-8
Figure 5-9
Figure 5-10
Figure 5-11
Figure 5-12
Figure 5-10
Figure 5-11
Figure 5-12
Figure 5-16
Figure 5-17
Figure 5-18
Figure 5-19
Figure 6-1
Figure 6-2
Figure 6-3
10 LIST OF FIGURES CONTINUED
Syllable Length – ‘abuelo’ ........................................................................ 89
Syllable Length – ‘palabra’ ....................................................................... 89
Syllable Length – ‘gravilla’ ...................................................................... 90
Spectrogram of YM wayeʔ ....................................................................... 91
Spectrogram of YM palabra .................................................................... 92
Spectrogram of YM abuelo ....................................................................... 92
Spectrogram of YM trabajo ...................................................................... 93
Spectrogram of YM gravilla ..................................................................... 93
Hypothesis 1: Duration cues Prominence ................................................... 97
Hypothesis 2: Non-Neutral vowels are prominent ...................................... 97
Hypothesis 3: Weight is prominent ............................................................. 97
Hypothesis 4: Initial syllables are prominent .............................................. 97
Hypothesis 5: Phonetic data will accord with speaker intuitions ................ 98
Variable Coding ........................................................................................ 103
Formula for Dependent Variable............................................................... 103
Yucatec Maya Prominence Rules ............................................................. 106
Characteristics of the Western Dialect of YM (Frazier, 2009: p 274) ..... 108
Mean Vowel Length by Speaker ............................................................. 110
Mean Pitch Trajectory ............................................................................. 118
Pitch Means Graphically ......................................................................... 121
Syntagmatic vs. Paradigmatic features ................................................... 125
Pitch Contours by Vowel Type ............................................................... 131
Differences in f0 in word initial vs. word final position ......................... 135
Mean duration by Word Type ................................................................. 143
Mean duration by Word Type ................................................................. 150
Mean Pitch in Trisyllabic Words by Position ......................................... 156
Yucatec Maya Prominence Rules ........................................................... 160
Possible Scenarios ..................................................................................... 164
Frazier (2009)’s Western dialect ............................................................... 167
Final hypotheses ........................................................................................ 169
Table 2-1
Table 2-2
Table 2-3
Table 3-1
Table 4-1
Table 4-2
Table 5-1
Table 5-2
Table 5-3
Table 5-4
Table 5-5
Table 5-6
Table 5-7
Table 5-8
Table 5-9
Table 5-10
Table 5-11
Table 5-12
Table 5-13
Table 5-14
Table 5-15
Table 5-16
Table 5-17
Table 5-18
Table 5-19
Table 5-20
Table 5-21
Table 5-22
Table 5-23
Table 5-24
Table 5-25
Table 5-26
Table 5-27
Table 5-28
Table 5-29
11 LIST OF TABLES
Population of Mayan speakers in the Yucatán Peninsula in 1990 ................ 29
Study Participants ......................................................................................... 32
Consonantal Phoneme Inventory (Straight, 1976: p 22) ............................. 38
Yucatec Maya Underlying Segment Inventory (Straight, 1976: 45) ............ 66
Pitch Measurements for Spanish Loans ........................................................ 83
Loan Word Syllable Length and Pitch Range Measurements ...................... 87
Native Speaker Prominence Intuitions ....................................................... 100
Number of Strong Judgments by Speaker .................................................. 104
Number of Strong Judgments for Position and Vowel Type ...................... 104
Mean Vowel Length in Isolated Monosyllabic Words ............................... 110
Monosyllabic Duration ANOVA results .................................................... 111
Mean Vowel Length in Monosyllabic Words in Phrases ........................... 112
Monosyllabic Duration ANOVA results for words in phrases ................... 112
Mean Pitch at P3 – Isolation ....................................................................... 115
Monosyllabic Pitch ANOVA results – Words in Isolation ........................ 115
Mean Pitch at P3 – Phrasal Context ......................................................... 116
Monosyllabic Pitch ANOVA results – Phrasal Context........................... 117
Mean pitch for words in isolation ............................................................. 117
Pitch Contour - Words in Isolation ........................................................... 120
Pitch Means for Words in Phrases............................................................ 120
Means for Bisyllabic Words ..................................................................... 126
Results for duration difference between Initial and Final Syllables......... 127
Results for differences in duration for bisyllabic words in phrases ......... 129
Pitch Means for H, L, G and S in Initial versus Final position ................ 130
Results for difference in pitch between Initial and Final Syllables .......... 133
Means for Bisyllabic Pitch by Position .................................................... 135
Mean duration by Word Type .................................................................. 142
Statistical Analyses for Trisyllabic Length .............................................. 145
Long-Short-Long Duration ....................................................................... 146
Long-Short-Short Duration ...................................................................... 147
Short-Long-Short Duration ...................................................................... 147
Short-Short-Long Duration ...................................................................... 148
Means for duration in Trisyllabic words by type ..................................... 149
Statistical Analyses for Trisyllabic Length .............................................. 152
Pitch Measurements at the Midpoint ........................................................ 155
12 ABSTRACT
Yucatec Maya (YM) is an indigenous language of Mexico that features both phonemic
tonal distinctions and phonemic vowel length. These features are primarily associated
with the phonetic cues of pitch and duration, which are also considered the primary
correlates of stress in language. Though scholars have noted the existence of stress or
accent since it was first documented centuries ago, no detailed account of stress as either
a separate or related entity to tone or length has been made. This dissertation presents a
unique view into YM prosody by looking at loan word incorporation in conjunction with
native speaker intuitions, and production data. A case study of Spanish loan words into
Yucatec finds that when Spanish words are incorporated into the YM prosodic system,
the initial syllable undergoes lengthening. Statistical analyses performed on data from
native speaker intuitions and production data, however, find no concrete pattern of
obligatory stress on the word level in Yucatec Maya words.
13 CHAPTER 1 INTRODUCTION
Yucatec Maya (henceforth YM), an indigenous language spoken in the Mexican
states of Quintana Roo and Yucatán, is a prosodically complex and fascinating language.
It has been thoroughly studied by Maya scholars for centuries, and has a well
documented contrast between High and Low tones. What has not yet been thoroughly
documented is whether or not YM contains an obligatory metrical stress system, and if
so, what the nature of that system is. While it has been treated in its possible relation to
the tonal system, stress or accent in Yucatec has not been systematically analyzed as a
feature existing separately from tone. This study attempts to remedy this, by presenting a
comprehensive investigation into whether or not a system of metrical stress may be found
in YM words, by looking for at both phonetic cues and native speaker intuitions. Tone
and stress are features that are difficult to tease apart phonetically, because they often rely
on the same phonetic cues; tone relies on pitch, and stress can rely on a variety of
features, most commonly pitch and duration. Because Yucatec includes both a phonemic
tone distinction, utilizing pitch, and also includes alternations between long and short
vowels, relying on segment duration, the question arises of how exactly stress might be
realized. Since both of the most robust cues of stress are already in use, are there phonetic
cues for stress, and if so, do they use means other than pitch and duration? Does a default
stress system, discrete from the tone and length alternations, exist as feature of YM at all?
In order to gain insight into how the prosodic system works in Yucatec Maya,
both production and intuition data was collected from native speakers. This data was
analyzed in order to discern whether or not a pattern of prominence that could be
indicative of stress was present in the intuitions of speakers or in the phonetic cues of
14 their speech. As will be discussed in detail in the sections to follow, the data from native
speaker intuitions as well as production data collected for this study showed no strong
evidence for the existence of a default stress system on YM words that is cued by either
pitch or duration. The following analysis builds on the work of previous scholars to
describe the prosodic characteristics of Yucatec Maya in order to gain a more concrete
view into how the prosodic system works.
1.1
Outline of Dissertation
In Chapter 1, the theoretical background of studies in stress and accent are
discussed, in order to build a framework for the analysis of YM to follow. Chapter 2
gives an in-depth description of the phonetic, phonological and prosodic properties, as
well as the relevant morphological, and syntactic properties of YM, and presents the
details and methodology for the data on native speaker intuitions and production data for
monosyllabic, bisyllabic and trisyllabic words. In Chapter 3 I describe the evolution of
important work on Yucatec Mayan prosody, specifically that which relates to stress, from
the first Spanish scholars who documented it in the 17th century, to the modern linguists
who have studied the phenomenon. Chapter 4 presents data from Spanish loan words into
the language, which gives a unique insight into what forces might be at play which adapt
and incorporate a word into Yucatec from a language without tone or vowel length
distinctions. Chapter 5 presents the data from native speaker intuitions, as well as the
statistical analysis of vowel length and pitch in Yucatec Mayan words, in order to
elucidate how tone and length interact, and lays out statistical analyses of this data which
show no strong patterns that indicate the present of default stress on YM words. Finally,
15 Chapter 6 summarizes these findings and discusses their implications for both studies of
Yucatec and studies of stress in general.
1.2
Theoretical Background
Prosody, a much studied but enduringly difficult to define area of linguistics, is
nonetheless an incredibly important one. Loosely defined by scholars as a range between
“the structure that organizes sound”, and the realization of the ‘suprasegmental’ entities
of pitch, rhythm, loudness, and pause (Cutler, Dahan, & van Donsellar, 1997: 142), the
study of prosody covers a wide range of linguistic phenomena that interact in complex
ways. Given that it has the unique characteristic of lying at the nexus of phonetics,
phonology, pragmatics, semantics, morphology, syntax, and sociolinguistics, coming to a
more concrete and systematic way of viewing how prosody works in language has great
opportunity for shedding light on as yet obscure areas of linguistic study. It is
underlying, or overlaying all of what we say, and the variety of uses it is put to in
different languages show its immense usefulness to the goal of communicating messages,
from the simple to the socially complex. Languages utilize various prosodic tools to
convey meanings at every level of language, from the segmental to the syllabic,
morphological, lexical and up to phrasal levels. One of the most important prosodic tools
in use in language is stress, which is the main topic of investigation in this dissertation.
Stress and accent provide the rhythmic backbone for the prosodic system. The following
sections outline what exactly is meant by stress, what characterizes it, and how it can
surface from one language to another.
1.3
16 Perceptual Cues of Stress
For all the possible variety of meanings that stress can convey cross-linguistically,
these many meanings are predominantly realized by only a few physical, acoustic cues,
namely pitch (fundamental frequency), duration (temporal elements of segments or
syllables), and to a lesser degree intensity (loudness), sometimes in combination with an
increase in phonological precision of a stressed element or decrease in precision of
unstressed elements (such as English schwa). In many cases language specific cues are
known to play a supporting role.
Of all of these possible cues for stress, it has been demonstrated by many
scholars, including Fry (1955, 1958), Morton & Jassem (1965), and Nakatami & Aston
(1978), that intensity does not play as strong of a perceptual role as pitch and duration,
that it tends to co-occur with the other two main cues, and can provide additional help for
perceiving a stressed syllable, but that semantic load does not rest solely with intensity in
any language so far studied. In other words, languages do not have phonemes that differ
only in loudness, or signal stress only by loudness, though they can do both with pitch
and duration. Taking this into consideration, there are then two primary acoustic cues for
all of prosody to work with, i.e. pitch, and duration, in combination with whatever cues
may be utilized on a language-specific basis. Though languages may combine these two
cues with other things, such as intensity, or allophonic features (Lehiste, 1970) to aid in
indicating prosodic distinctions, they appear to lack the same amount of perceptual
weight as pitch and duration in most case studies. For these reasons, the analysis that
follows focuses on these two main cues of pitch and duration as the most likely phonetic
cues for stress in YM.
17 Attempting to pin down and concretely define the notion of stress or accent in
language is especially complex, precisely because it surfaces in different forms from
language to language, and because when it surfaces can be assigned a range of semantic
roles, in some cases serving to distinguish between word meanings, and in others as only
a system of metrical organization. In some languages it is perceived mainly using the cue
of segmental duration, while in others pitch plays the fundamental role, and in some they
work in tandem to cue stress.
The existence of stress in some form or other was long considered to be a
language universal, though languages without identifiable stress systems have been
proposed, such as in Hausa (Leben, 1980), (Leben, 1977) (Leben, 1996). Navajo has also
been a subject of debate amongst scholars in regards to the existence of a stress system
(McDonough, 2003) (Kidder, 2008). What is agreed upon by most scholars is that stress
can be a notoriously nebulous entity, and it is very often not exactly the same beast from
one language to the next.
1.4
Definition of Stress
What is or are the underlying entities that scholars have referred to as stress? The
loosest definition of stress, that might work to capture all the various ways that it appears
in different languages, is to say it is some kind of emphasis, a differentiation in
prominence between elements that is structurally or linguistically meaningful, and that
can be realized in a variety of ways phonetically. As discussed above, these include
features such as pitch, duration, intensity, as well as a contrast in precision between it and
the surrounding syllables, which are realized differently across languages.
1.4.1
18 Lexical Stress vs. Metrical Stress
Some types of emphasis are used on a lexical level, and it is the only
characteristic that differentiates one word from another. This type of stress, typically
called lexical or distinctive stress, is similar to a phonemic distinction, in that minimal
pairs of words are created that would otherwise be functionally the same. For example,
the following are often considered minimal stress pairs in English.
Figure 1-1
a.
b.
Minimal stress pairs in English
perMIT versus PERmit
conVICT versus CONvict
Metrical stress, on the other hand, does not provide meaningful distinctions between
words, but instead provides the units of a language’s rhythmic structure. Examples of
metrical stress in English can be seen in Figure 1-­‐2.
Figure 1-2
a.
b.
c.
Metrical stress patterns in English
HOCKey
MISSiSSIpi
FANCiEST
Minimal pairs cannot be found for stress of this sort, and instead of being specified on the
lexical level, it can often be reliably predicted by context. Hayes describes this distinction
as ‘free’ versus ‘fixed’ stress.
“Fixed stress is predictable in its location, and may be derived by rule, while free
stress is unpredictable, and must be lexically listed.” (Hayes, 1995: 31)
As English shows, languages can have a combination of lexical and metrical stress, or
they can have only one or the other, or could have a metrical stress system that is
metrically restricted, such as Polish and Macedonian (Hammond, 1989). Fixed stress is
often termed the ‘default’ stress system, or the pattern that can often be generalized to the
19 whole language (commonly with exceptions), and which is considered obligatory on the
word level (Hyman, 2006).
1.4.2
Rhythmic vs. Morphological Stress
Another distinction in the typology of stress that Hayes (1995) argues for is that
between Rhythmic and Morphological Stress. Rhythmic stress is argued to be based
solely on phonological factors such as syllable weight. In morphological stress systems,
“stress serves to elucidate the morphological structure of a word (Hayes, 1995: 32)”, i.e.
the morphological constituents of stem, root or affix can be integral to the assignment of
stress. An example of this is English, in which affixes are subordinate to stems in stress
placement, i.e. the word ‘unboundedness’, which has antepenultimate stress due to the
location of the stem, not the default stress pattern. Many stress systems are a mix of these
two types, just as languages can display a mix of lexical and metrical stress features.
1.5
Intonation, Stress and Accent
Patterns of emphasis can be found throughout different domains of speech. The
broadest domain in which emphasis plays a role is in the intonational system on the level
of the utterance. The intonation system, as defined by Gussenhoven (2004), is the use of
"pitch variation contrastively for the expression of discoursal meaning and for marking
phrases (p 22)”. Intonational contours are made up of accented positions and intonational
tunes. For example, the following sentence, from Gussenhoven (2004: 23), is made up of
two intonational phrases (the boundaries of which are marked with subscript letter i),
marked to show the locations of the accents (*) and intonational tones (H, L).
20 Figure 1-3
Example of intonational phrase construction
{In PRESS reports of the incident} {he wasn’t MENtioned}
|
|
| |
|
|
Li
H*L
Hi Li
H*L
Li
As you can see in Figure 1-3, in the first intonational phrase, the only intonational
accent is placed on the word ‘press’, while in the second phrase it lies on the initial
syllable in ‘mentioned’. The other words in this utterance have their own word level
stress associated with them, (i.e. the final syllable in ‘rePORTS’, and the initial in
‘INcident’), however they do not receive an intonational accent in this utterance.
The term accent was often used interchangeably with stress in early studies on the
subject, but many scholars today have distinguished the two as different but related
phenomena. Beckman (1996) discusses this distinction in detail, and describes accent as
“a system of syntagmatic contrasts used to construct prosodic patterns which divide an
utterance into a succession of shorter phrases and to specify relationships among these
patterns which organize them into larger phrasal groupings (Beckman, 1986: 1).”
Stress on the other hand is defined as “a phonologically delimitable type of accent in
which the pitch shape of the accentual pattern cannot be specified in the lexicon but
rather is chosen for a specific utterance from an inventory of shapes provided by the
intonation system (Beckman, 1986: 1).” This distinction Beckman describes between
accent and stress can be summarized as the difference between emphasis in different
domains, i.e. on the utterance level, intonational contours display accentual peaks and
valleys, and these can be defined as accents, while the words that make up those
utterances have their own patterns of emphasis which can be defined as stress. For
21 example, in Figure 1-3, the word ‘incident’ has initial stress, but no accent, while the
word ‘mentioned’ has both initial stress and initial accent. These two phenomena are
understandably very closely related, because intonation level accents are very often1
chosen from the set of lexical level stresses, though not all segments that have a lexical
stress will also have an intonational accent.
In some languages, including Spanish, accents have been argued to be cued in a
slightly different way than stresses. Ortega-Llebaria (2006) studied the difference in cues
between stresses and accents in Spanish, looking at example sentences such as ‘Malena
miraba la luna’ (Malena looked at the moon) versus ‘Mira la luna – dijo Malena’ (Look at
the moon – said Malena). In the first sentence the second syllable in the name Malena has
both a word level stress and a phrase level accent, while it has only the word level stress
in the second. She found that when vowels had stress, they had a flat pitch contour but
increased duration, while those that had accents had both increased duration and a rise in
pitch. The fact that the two often co-occur means that they are often treated as the same
entity, but they may have different cues and behaviors in some languages.
1.6 Syntagmatic Nature of Stress
Crucially lexical stress differs from other phonological features that can create a minimal
pair and hence prove the existence of a phoneme (i.e. backness, rounding, continuant) in
that it has to exist in relation to its surroundings. Looking at stress on a purely lexical
level, a stressed syllable cannot be found on a one syllable word spoken in isolation, there
must be another syllable to compare to the stressed one for prominence to exist, because
in essence, prominence is contrast. And to have contrast, more than one ‘unit’ has to be
1
Lindstrom & Remijsen (2005) argue that the case of Kuot, described below, is one
exception to this tendency that is considered by some to be a universal.
22 present. Without the unstressed syllable, the stressed syllable no longer stands out, and
prominence loses its power to be perceived.
Beckman (1986) and Lindstrom & Remijsen (2005) describe this characteristic of
stress as a syntagmatic feature versus a paradigmatic feature.
Figure 1-4
Syntagmatic versus Paradigmatic Features (L & R, 2005: 2)
As is shown in the schematic of Figure 1-­‐4, ‘a syntagmatic feature distinguishes a
syllable from those preceding or following it, while a paradigmatic feature contrasts a
syllable from other syllables that may appear in the same position.’ (Lindstrom &
Remijsen, 2005: 2). In other words, syntagmatic features look horizontally at what comes
before and after a segment in question. Paradigmatic features differ in that they look at
one segment, and compare it with the other possible segments that might fill that position.
Stress, as discussed above, is syntagmatic because it requires the presence of the
unstressed in order to be meaningful. For example in the word ‘canine’ the initial
syllable is more prominent in relation to the final syllable. Tone, however, is
paradigmatic, because tones are contrasted with a set of other possible tones, which are
present in the phonological inventory, but do not have to be present in the word in
question. For example, a word with a tone, such as the in YM míis – ‘broom’ versus miis
– ‘cat’, is judged not in comparison to the syllable next to it, but in comparison to the
other possible tone(s) that could fall in its place.
23 Looking purely at the lexical level rather than the phrasal level, tone can, and
often does exist in one syllable words, because a tonal distinction is perceived
paradigmatically, i.e. in relation to other possible tones that could occur instead. It does
not necessarily need to be in contrast horizontally, or syntagmatically. One consequence
of stress being syntagmatic is that it can and often does effect the surrounding syllables as
well, in that the addition of a stress to a syllable may not raise its pitch and duration, but
shorten those of the surrounding syllables, hence enhancing the contrast.
1.7
Grouping Properties of Stress
Since stress must have surrounding context to be realized, it is natural that
scholars have analyzed stress in terms of its grouping properties. The notion of metrical
‘feet’, defined by Hayes (1995) as the ‘minimal bracketed units of metrical theory’, is
crucial to the analysis of the rhythmic structure of a language. Feet are generally groups
of 2 syllables or moras2, with stress falling on either the initial (trochee) or the final
(iamb). The language specific patterns and constraints on how syllables or moras are
grouped into feet, and how feet are grouped into larger constituents, determines the
fundamental rhythmic properties of a language.
Further evidence of the importance of grouping to stress is found in the crosslinguistic tendency to disallow ‘degenerate feet’, or feet with only one syllable or mora.
Hayes (1995), Prince (1980) and Kager (1989) all discuss why degenerate feet are
avoided in most languages. The syntagmatic nature of stress seems like it would naturally
2
In moraic languages, light and heavy syllables are treated differently, therefore the
quantitative unit of ‘mora’ is used. Light syllables contain one mora, while heavy
syllables contain two.
24 avoid this type of grouping, because the lone syllable or mora in a degenerate foot has
nothing to be compared to.
1.8
Stress Cross-Linguistically
As was discussed above, stress can be realized in a variety of different ways
cross-linguistically, and can be prosodically flexible. In non-tonal languages, pitch most
often plays a major role as a cue for stress, sometimes also combined with longer
duration, strengthening of the articulation of the syllable or weakening of the syllables
around it, and higher intensity. In tonal languages, the balance shifts from pitch to
duration as the typical key signal for stress (De Lacy, 1999). Because of this flexibility,
stress is unique as a phonological or phonetic entity, because it is not only parasitic, or in
some way recycling the phonetic cues used to signal other meaningful entities, but it can
shift the perceptual load from one building block to another in order to avoid interfering
with other meaningful elements of the system. However, this same flexibility makes it
harder for scholars to define and pinpoint, because it may show up looking differently
from one language to another, and there is often no obvious single cue to look for to
diagnose the stress patterns of a language.
As discussed above, Ortega-Llebaria (2006) describes how metrical stress is cued
mainly by duration in Spanish, while intonational accent which often, but not always,
appears in conjunction with metrical stress, is cued by pitch contours. Using native
speaker productions of nonce words with a particular metrical stress, Ortega-Llebaria
showed that when a syllable occurred in a position that lacked both metrical stress and
intonational accent, it was realized with flat pitch contours and was the shortest segment.
The syllables which had metrical stress but did not have intonational accent, the pitch
25 contour was flat, but the duration increased. And finally syllables with intonational
accents had increased duration and changes in F0 trajectories (Ortega-Llebaria, 2006:
116).
The Kuot language shows another instance of a language with a unique utilization
of phonetic cues for stress and accent. In Lindstrom & Remijsen (2005), the prosodic
systems of the Kuot language are described, and they argue that in Kuot, a non-tonal
language featuring lexically distinctive stress, pitch is solely utilized in the domain of
intonation, and duration is the sole cue for stress. There is no metric or default stress
patterning in Kuot; when stress appears, it is lexically conditioned. Interestingly, they
also find that intonation does not rely on the stressed syllables for intonational accents,
and in fact may not have any intonational accents at all. Intonation in Kuot is argued to
contain boundary tones (marked by subscript i in Figure 1-3), but no accent anchors
(marked by * in Figure 1-3). This shows counterevidence to a language universal posited
by Ladd (1996), that all languages have pitch accents, which act as the anchors for
intonation contours. As Kuot shows, there is a difference between lexical stress, a
feature that must be lexically specified, and which can be the sole means of
distinguishing two different words, and metrical stress, the hierarchical structuring of
syllables based on prominence. Kuot appears to lack the second type of stress altogether,
which though rare, is not unheard of. It also then lacks intonational stress, a third type of
prominence at the utterance level, which is usually chosen from the set of lexically or
metrically stressed syllables.
26 The way languages such as Spanish and Kuot differ in their utilization of phonetic
cues and prosodic distinctions shows how differently these systems can be organized
from one language to another.
1.9
Stress in Yucatec Maya
Although each language is undoubtedly a fascinating case study in the realization
of how the two acoustic cues of duration and pitch are put to work, so to speak, in the job
of providing meaningful linguistic distinctions, this dissertation delves more deeply into
untangling the prosodic puzzle in Yucatec Maya, a language of particular interest in this
area. The specific question being asked is: does “stress”, i.e. a metrical or default system
of lexical emphasis, exist in Yucatec Maya? YM has been thoroughly documented as
having lexically encoded tone, but not lexically encoded stress. The question of whether
or not YM has a system of metrical stress, and if so how it is cued, has only been treated
lightly in previous literature. This question may seem straightforward, since stress is a
phonological entity that has been studied and described in thousands of languages and in
great detail. The nature of stress, however, is much more difficult to pin down as
concrete, because as was discussed above, it surfaces in so many different ways in
different languages. Hayes (1995), terms stress ‘parasitic’ (p. 7), because of its use of two
cues that are used phonemically for other means in language. This parasitic nature makes
it very tricky to concretely outline in a given case, because it often is difficult to tease it
apart from the phonemic elements whose cues are being parasitized.
The fact that Yucatec Maya is a tonal language makes this question even more
interesting, because another important prosodic entity is at play that is not in languages
such as Spanish and Kuot. Shedding more light on the nature of stress in Yucatec Maya
27 can help illuminate a notoriously shadowy area of prosodic inquiry. The following
chapter will present the relevant facts of Yucatec Maya phonology and prosody, and
present the methodology used to collect data for the inquiry into the topic of stress in
YM. Chapter 3 then delves into the previous research undertaken on these topics, and in
chapters 4 and 5 I propose and test multiple hypotheses regarding the placement and
phonetic realization stress in both Spanish loan words and native YM words. Finally
chapter 6 will summarize the findings, and analyze the implications for future research.
28 CHAPTER 2 OVERVIEW OF YUCATEC MAYA & METHODOLOGY
The Yucatec Mayan language belongs to the Yucatecan branch of the Mayan
language family, which is made up of 5 sub-families that are spoken in a wide area of
Central America. These sub-families include Ch'ol-Tzotzil, Huastecan, ChujeanKanjobal, Quichean-Mamean, and Yucatecan, and are spoken in the Mexican states of
Chiapas, Tabasco, Yucatán, Quintana Roo, Campeche, San Luis Potosí and Veracruz,
throughout Guatemala, Belize and into Honduras. These 5 sub-families are thought to
include at least 31 distinct languages or dialects (Campbell, 1985), which were spoken by
approximately 1,695,000 speakers, as of 1996 (Gordon, 2005). Yucatec Maya is one of
only three Mayan languages to develop a tonal distinction, the others being Uspantec and
one dialect of Tzotzil from San Bartolo (Suarez, 1983: p 51), and the only one in the
Yucatecan branch of Maya to have tone. Figure 2-­‐1 shows the branches of the Mayan
language family.
Figure 2-1
Mayan Language Family (Campbell, 1985: 189)
29 The Yucatecan branch of the Mayan family is spoken on the Yucatan Peninsula in
Mexico, in the states of Quintana Roo, Yucatán and Campeche, as well as in Northern
Belize and parts of Guatemala. There are roughly 700,000 speakers of Yucatec Maya in
these areas (Gordon, 2005).
2.1
Bilingualism with Spanish
The number of monolingual Mayan speakers is declining, as Spanish is becoming
more and more common in Mayan communities. Table 2-1 details the number of YM
speakers in Mexico by state, as well as the number of monolingual speakers as of 1990
(Güémez Pineda, 1994).
Table 2-1
Population of Mayan speakers in the Yucatán Peninsula in 1990
State
%
Yucatán
Maya
Speakers
520,440
%
43.8
MonoLingual
40,355
Campeche
71,247
15.4
4,197
5.9
Quintana
Roo
Total
123,146
29.2
10,052
2.4
714,883
3.4
54,604
According to census results over the past decades, the percentage of the Yucatec Mayan
speaking population that is monolingual has fallen from 29.3% in the 1930’s to about 3%
in the 1990’s. Bilingualism is more common in the urban centers of the Yucatán such as
Mérida, in which census figures show a monolingualism rate of 4% in 1940 fall to 0.1%
in 1990 (Güémez Pineda, 1994).
2.2
Yucatecan Spanish Dialect
The Yucatecan dialect of Spanish grew from the unique history and geographical
and cultural isolation of the Yucatán peninsula from the rest of Mexico. When Spanish
30 speaking colonists arrived in the Yucatán they settled in the major cities of Mérida,
Valladolid and Campeche, which were small pockets surrounded by a Mayan speaking
populace. According to colonial records the percentage of Spanish speakers grew from
1% directly after conquest, to 10% in 1700, to 40% by 1794. Even as late as the 1970’s
the percentage of monolingual Spanish speakers was not a majority, at 45% (Mosley,
1980). This was not the case in central Mexico with speakers of Nahuatl, another widely
spoken indigenous language spoken in that area, which was spoken in an area that had a
much higher rate of immigration from Europe and other countries. Another factor that
sets Yucatecan Spanish apart from other dialects in Mexico is that the indigenous
population spoke one relatively consistent language across a widespread area, whereas in
other areas of Mexico, such as Oaxaca, indigenous communities spoke diverse OtoManguean languages such as Zapotec, Mixtec and others. These communities tended to
have smaller populations with a higher degree of variations between them.
For these and other reasons, Yucatecan Spanish has been described as having a
particularly high degree of influence from YM (Michnowicz, 2008). The effects of this
close interaction include the existence of a variety of phonetic and phonological elements
from YM , including distinct intonational patterns, the use of the voiced stops [b, d, g] in
place of the fricative [β, ð, ɣ] of standard Spanish, the aspiration of voiceless stops, the
insertion of glottal stop between word boundaries (as in ‘ocho años’ [otʃoʔaɲos]), and the
variable realization of final nasal consonants, often seen as final [n]-->[m] (Michnowicz,
2008). This final characteristic has been shown to be a salient marker of identity for
speakers of this dialect, and has become a linguistic stereotype in popular culture.
Although this dialect is an object of stereotype, within the community of speakers it is a
31 point of pride and is often viewed positively. According to Michnowicz (2008), all of the
markers listed above are seeing a decline in usage amongst younger speakers, with the
exception of the final nasal variation which is increasing in usage.
2.3
Immigration to San Francisco Bay Area
In recent years, there has been a significant growth in immigration from YM
speaking communities of the Yucatán to the United States, and the main locus of this new
community of Yucatecans is in the San Francisco Bay Area of California. As of 2010,
there were an estimated 10,000 Yucatec Mayans that have resettled in Northern
California, the majority of them having come during the last 6-7 years (Delugan, 2010).
Two main communities have developed, one in the Mission District of San Francisco,
and the other in the Canal District of San Rafael, north of San Francisco, though both
have ties with each other and participate in community events and gatherings together.
The neighborhood centered around the intersection of 16th and Mission in the Mission
District of San Francisco has become known as Little Yucatán3, and Yucatecan
restaurants and cultural events are common in this area, although it remains an incredibly
diverse population with immigrants from cultures all over the world. In both San
Francisco and San Rafael, grassroots organizations have developed which advocate for
the community, organize cultural events both for the enjoyment and maintenance of
Yucatecan culture and language, and to educate those outside of the community.
Asociación MAYAB (Maya Yucateca de la Area de la Bahía) in San Francisco, and Chan
3
Unlike some culturally distinct neighborhoods in San Francisco, signs and other
instances of Yucatec writing are not commonly found in public spaces, except inside of
the cultural centers that promote the use of Yucatec.
32 Kahal in San Rafael provide a variety of important services to members of this
community, and help to maintain cultural ties to the Yucatán.
The San Francisco community of Yucatecans primarily come from the area
surrounding Oxcutzcab, Mexico, while those in San Rafael mainly come from the areas
surrounding nearby Peto. There is also a community of immigrants from Southern
Mexico and Guatemala, who speak the related but quite distinct Mayan languages of
Quiche, Tzeltal, Chol, Tzotzil and Mam, that have settled in the Mission District and
across the San Francisco Bay in Oakland (Delugan, 2010).
2.4
Methodology of Current Study
In order to investigate the existence of metrical stress in a comprehensive way,
data was collected for this study from 8 native speakers of Yucatec Maya living in the
Yucatecan communities of the San Francisco Bay Area. The following table shows the
speaker breakdown by gender, hometown, age, and number of years living in the United
States, and Bay Area community they lived in at the time of the study.
Table 2-2
Study Participants
Gender
1
F
City of
Origin
Akil
Age
2
3
F
M
Chumayel ?4
Oxcutzcab 31
9
5
San Francisco
San Francisco
4
5
6
7
8
M
F
M
M
M
Oxcutzcab
Peto
Peto
Oxcutzcab
Oxcutzcab
30
15
25
18
11
San Francisco
San Rafael
San Rafael
San Rafael
San Francisco
30
53
39
41
43
36
Years in
US
2
Bay Area
Community
San Francisco
A map of the locations in Table 2-2 are seen in Figure 2-2.
4
The age of speaker 2 was not given, but was likely between 45 and 60.
33 Figure 2-2
Map of Yucatán, Mexico
As can be seen from the map above, the speakers all originated in the general area
surrounding Oxcutzcab, which is south of Mérida and just east of Santa Elena. Frazier
(2009) describes the geographical area shown in the map above as the home of the
‘Western dialect’ of YM, which can be distinguished from the Eastern dialect by the
presence of tonal distinctions.
All 8 speakers were fully bilingual in Spanish, and report that they speak Spanish
outside the home and to their children, grandchildren, nieces and nephews, while they
speak YM to their spouses, parents, aunts and uncles, and older friends. They worked in a
variety of fields, and had differing amounts of exposure to English, and though most
knew some English, they were not fluent and did not speak it often in their daily lives.
34 Those that had children reported that the language spoken by and to the younger
generation is Spanish, and for this reason few children of YM speaking parents in these
communities are fluent in YM themselves, though they are often fluent in English.
All participants were contacted through the Asociación MAYAB community
center in San Francisco and the Chan Kahal Mayan Association in the Canal District of
San Rafael, and recording took place at the headquarters of these organizations, which
was deemed to be the most comfortable and convenient location for participants, in one
instance a public place near the organization headquarters was used upon request of the
participant. Figure 2-3 shows the types of data recorded for this study.
Figure 2-3
Data Collected
a. 98 elicited words in isolation, followed by the word used in a meaningful
phrase, and words placed in a frame sentence. (all speakers)5
b. Conversations elicited through given topics (2 speakers)
c. Data on stress intuitions for 55 words (4 speakers)
The data was recorded using a Mini DV video recorder, enhanced with a Sennheiser
shotgun microphone. Participants were asked whether they were comfortable being
videotaped as well as audiotaped, and 5 opted to not be recorded visually. All interviews
were conducted in Spanish.
For the data collected of type (a), YM words and phrases were displayed in both
YM orthography and in Spanish translations, as in Figure 2-4.
Figure 2-4
Example of Elicitation Materials
- carbon
- Juan vende carbon
(chúuk)6
(Juane' ku konik chúuk)
5
The data collected in the elicitation portion of the study, which will be detailed in
Chapter 6, focuses on the speech from only 5 of the 8 participants, as there were sound
quality issues for the other 3 due to unforeseen circumstances during recording.
6
This translates in English to ‘charcoal’, and the following sentence translates to ‘Juan
sells charcoal’.
35 The participants were asked to read an additional frame sentence, in order to determine
whether or not phrasal intonation might have an effect on the word readings. The frame
sentence used was ‘él dice la palabra ___ en todo el tiempo’ – ‘he says the word ___ all
the time’. To put this sentence in some context for participants, they were told to imagine
a child was just learning to speak and that he was repeating the same word over and over.
The words and sentences were taken from a Yucatec Maya-Spanish Dictionary,
(Navarrete, 2009), and were chosen to be representative of different syllable types and
number of syllables, different parts of speech, and different tones and vowel nuclei. 50
tone pairs were recorded, in which the only difference between two words was the tone
on the vowel nuclei.
All speakers were literate in Spanish, and the majority of the speakers were more
comfortable reading the Spanish words and sentences, and translating them into YM. The
YM orthography is not in common use among the communities, and though they could
make it out with effort, they all either preferred to read and translate the Spanish or have
the words and sentences read aloud to them in Spanish for them to translate into YM.
Each speaker was given a practice example and instructions, and were then asked
to say aloud each word, sentence, and frame sentence in YM, or to translate the Spanish
into YM if that was preferred. In order to elicit the most natural speech patterns, they
were also given the instruction that they should translate the words and sentences in a
way that felt natural (i.e. if the words chosen were not commonly used or archaic, which
turned out to be the case for a subset of the chosen words, they were encouraged to use a
different word that was more natural). They were also instructed that it was acceptable to
substitute a Spanish word if it was what they would normally use in a particular context.
36 During recording sessions, speakers often gave their own intuitions on which words were
the most natural, which were uncommon, and which were not used at all, which was
taken into account during the analysis of this data. Of the 8 speakers, 3 of the recordings
for the elicitation section of the study had sound quality issues, and therefore these 3 were
excluded from the phonetic analyses that follow in chapter 5, however their intuitions and
conversation data were utilized.
Conversational data was collected from 2 speakers, who were asked to converse
with each other on topics related to language use and education in YM. Specifically they
were asked to discuss which situations they speak Spanish versus Maya, what kind of
educational aspirations they have for their children, and whether they plan or desire to
return to the Yucatán in the future. The conversational data included many instances of
code-switching between YM and Spanish, and some particularly interesting cases of
Spanish loan word incorporation will be discussed in chapter 4.
Finally, intuitions on syllabification and stress were collected from 4 speakers.
They were given a list of 55 words, which were chosen from the set of 98 that were given
to them in the elicitation section of the study. After being given instruction and practice
examples from Spanish, they were asked to underline each syllable of the given words,
and place an X above the one they felt was the ‘strongest’, or had the most ‘emphasis’,
like those in Spanish they had seen in the instructions. They were also told that it was
acceptable if more than one, or all the syllables in a word were equal.
Various methods of eliciting native speaker intuitions were used, including the
tapping method (Fitzgerald, 1999) (Hayes, 1995), in which the speakers were asked to tap
their finger against a surface for the strongest syllable, as well as the repetition method
37 (Fitzgerald, 1999) (Everett & Everett, 1984), in which I repeated the words back to the
participants with emphases on different syllables and elicited judgments on which
pronunciation sounded the best. Though the speakers were trained first on Spanish
words, in which the tapping and repetition methods were used, when judgments on native
YM words were elicited, the tapping method was often abandoned by the speakers,
preferring instead to repeat the words to themselves in order to make a judgment. In
many cases, the speaker identified two equally strong syllables within a word, and in
other cases, no strong syllables were identified, and also in many cases the speakers
stated that they had no strong opinion about which syllable was stronger. The data from
this portion of the study, which did not result in any significant pattern of stress
placement amongst the speakers, will be analyzed in chapter 5.
2.5
Yucatec Maya Phonology, Phonotactics and Morphophonemics
The following section gives an overview of the basic phonological, morphological
and syntactic properties of YM, to allow the reader more insight into the way words and
phrases are assembled. Knowledge of these features of the language will provide a
framework for understanding the analysis of the stress system.
2.5.1
Sound Inventory
2.5.1.1 Consonants
The YM consonantal system contains the sound inventory seen in Table 2-3.
38 Table 2-3
Consonantal Phoneme Inventory (Straight, 1976: p 22)7
Bilabial
voice
-
+
Stop
p
(b)
Fricative
(f)
ApicoAlveoDental
+
t
(d)
Affricate
Ejective
Nasal
p’
b’
t’
Lamino
Alveolar
+
CentroPalatoVelar
+
d’
m
DorsoPalatoVelar
+
k
s
š
ts
č
ts’
č'
(ɡ)
Laryngeal
-
+
ʔ
h
k’
ɡ’
n
Lateral
l
Flap
r
Glide
w
y
Non-glottalized stops are articulated without aspiration, similar to those in the
English words ‘spill’ or ‘still’, unless they appear word finally, where they are aspirated
(Orie & Bricker, 2000: 297). The consonant [r] is found in a limited context, and almost
never appears at the beginning of a non-onomatopoetic root word of Mayan origin. In
word-final position, nasals assimilate to the following consonant in place of articulation.
7 Words in parentheses are only found in Spanish loan words
39 The laryngeal consonants [h] and [ʔ] exhibit different characteristics than the
other consonants. When these two segments fall between vowels in a morpheme, the
vowels must be identical. They also frequently undergo deletion in rapid speech. The
glottal stop is always transparent to vowel assimilation, and both are utilized as
epenthetic consonants when loan words are incorporated. For these reasons, Orie &
Bricker (2000), argue that they should be considered placeless laryngeals8. These two do
not pattern identically, however, in that /h/ (represented as ‘j’ in YM orthography) is
often featured both word initially (as in jo’ol – ‘head’), word internally (as in pajtal – ‘to
ferment’) and word finally (as in áantaj – ‘to help’), and unlike the glottal stop is not
argued to be epenthetic to avoid initial and final vowels. Therefore /h/ may carry more
phonemic weight than [ʔ], and is described by Orie & Bricker (2000) as chameleon-like
in its properties, sometimes behaving like [ʔ] and sometimes like other consonants (Orie
& Bricker, 2000: 301).
2.5.1.2
Vowels
There are 5 vowels present in YM, which have been argued to surface in 4
different articulations (see discussion below for an in depth look at work done on this
topic). The vowel inventory is seen in Figure 2-5.
8
The segment [h], according to Orie & Bricker (2000) exhibits two distinct behavioral
patterns, which they explain by positing that one of these patterns represents a historical
shift from [x] to [h].
40 Figure 2-5
Vowel Inventory (Bricker, Po'ot Yah, & Dzul de Po'ot, 1998: xiii)
front
high
central
i
back
u
Rounded
low
e
a
o
There is a 2-way distinction between high and low9, and a 3-way distinction between
front, central and back vowels. This vowel inventory is very similar if not identical to that
found in Spanish.
2.5.1.2.1 Vowel Shapes
The vowel shapes of YM will be discussed in detail in the following section, as
they have been a matter of debate throughout the scholarship of YM. Vowels in YM are
argued to fall into 4 phonemically distinct types of articulation in the YM word, which
combine pitch, duration, and laryngealization features, and will be called Short, Low,
High and Glottalized. Short vowels are also called Neutral, because they lack a tonal
element found in the other vowel types. Examples of the vowel articulations are found in
Figure 2-6.
9
Henceforth I will denote tone on vowels upper case letters ‘High/Low’, and vowels with
a feature of high or low in the vowel space as ‘high/low’ with a lower case letter.
41 Figure 2-6
Short
a
i
e
o
u
Low
aa
ii
ee
oo
uu
Vowel Articulations (Bricker, Po'ot Yah, & Dzul de Po'ot, 1998: xiii)
High
áa
íi
ée
óo
úu
Glottalized
áʔa
íʔi
éʔe
óʔo
úʔu
Short vowels are realized with shorter durations than the other three articulation types,
and lack any phonemic tonal distinctions. Short vowels may sometimes be the result of
neutralization of High, Low or Glottalized vowels that undergo reduction of length, pitch
and glottalization features in rapid speech. Glottalized vowels10 have underlyingly longer
durations, but the initial rising vowel can be either interrupted medially by a glottal stop,
or be realized with creaky voice (Frazier, 2009). In rapid speech, these are often
shortened to High Long vowels, or shortened further to High short vowels (Blair &
Vermont-Salas, 1965: p 15).
Vowels that are marked for either Low or High tone are phonetically longer, and
orthographically are written as geminates. Henceforth Low tones will be designated as
geminate vowel with no diacritic, while High tones will be designated with a geminate
vowel with an acute accent (this is consistent with the orthography, discussed in the
section below, adopted by the Mayan community in the Yucatán for its language
materials), as in Blair & Vermont-Salas (1965), Límon Rojas (1997), Navarrete (2009).
The following section shows acoustic data collected for this study of the different vowel
shapes in YM.
10
So called Glottalized vowels have historically been referred to by many names,
including ‘double vowels’, ‘glottalized’, and ‘rearticulated’.
42 2.5.1.3 Realization of Vowel Shapes in YM
The acoustic difference between a High and Low tone, áak’ – ‘turtle’ and aak’ –
‘vine’, can be seen in the spectrogram in Figure 2-711.
Figure 2-7
F0 of High versus Low Tone12
High tones can be realized in YM as steadily rising, as in Figure 2-7, or as
contours which rise and then fall (or vice versa), depending on the intonation contours
present in speech. Low vowels are more uniformly level in F0, though they can also fall
or rise depending on location in the intonational phrase. Short vowels with neutral tone
are realized with shorter durations than the other three articulation types, and lack any
phonemic tonal distinctions, though they can appear in surface forms as the result of
reduction of High or Glottalized vowels. Figure 2-8 and 2-9 show the acoustic features
of Short and Glottalized vowels.
11
The ejective k’ is clearly seen at the end of each utterance by the period of closure
followed by a short burst.
12
This spectrogram was created using Praat software from the data collected for this
study. The lower line represents f0, and the upper line represents amplitude.
43 Figure 2-8
Short Vowel – bak’ – ‘meat’13
Figure 2-9
Glottalized Vowel – ba’al – ‘thing’14
Glottalized vowels have long durations, but the initial rising vowel is interrupted
medially by a glottal stop or more often by creaky voice, as it is in Figure 2-9, which is
documented in Frazier (2009), who also documents that the initial period of modal voice
13
The length distinction between the vowels in (10) through (12) are not readable from
these spectrograms. The length of the vowels are as follows: aak’ –0.255243, áak’ 0.230341, bak’ - 0.092760, ba’al – 0.209306 (all measurements are in seconds). The
neutral vowel is less than half as long as the other three vowel types, which echoes
findings in Frazier (2009) and is consistent with findings in chapter 5.
14
In this instance the glottalized vowel appears with creaky voice or glottalization, and
therefore the f0 stops being readable after a short time.
44 in glottalized vowels often carries a rising pitch. In rapid speech, these can be shortened
to High Long vowels, or shortened further to High short vowels (Blair & Vermont-Salas
1965: p 15), or are realized with creaky voice (Frazier 2011). The nature of the duration
and pitch differences between vowel shapes in YM for monosyllabic, bisyllabic and
trisyllabic words will be treated in depth in chapter 5.
2.5.1.4 Diphthongs
In the description above regarding the four types of vowel nuclei in YM words, no
mention is made of diphthongs. This is because they are argued by many to be absent
from YM words. According to Navarrete’s (2009) corpus, however, a relatively small
number of YM words are documented which orthographically show nuclei with more
than one adjacent vowel of differing vowel quality. These cases are treated differently in
(Bricker, Po'ot Yah, & Dzul de Po'ot, 1998), where they are transcribed with glides
between the vowels, which if accurate preserves the pattern seen in the majority of YM
words, in which nuclei consist of only one vowel type. Examples from both sources can
be seen in Figure 2-10.
Figure 2-10
Vowel Adjacency
(Navarrete, 2009)
a. miats – ‘wise’
b. ajau - ‘king’
c. baili’ – ‘remain’
d. jo’oljeak – ‘yesterday’
e. utia’al – ‘his/hers’
(Bricker, Po'ot Yah, & Dzul de Po'ot, 1998)
miyats
ahaw
bayli’
jo’oljeyak
utya’al
There are no cases in which a diphthong vowel is orthographically written with a tone or
glottalization present, and it is likely that adjacent vowels are realized with medial glides
as in Bricker et. al. (1998)’s description.
45 2.5.2 Orthography
A variety of orthographies have been utilized through the history of Mayan scholarship.
The standardized orthography was adopted in 1984, after a meeting of the Yucatan
Delegation from the National Institute for Adult Education that involved the Southeast
Regional Center of the National Institute of Anthropology and History, the State
Delegation of the National Institute for Adult Education, the Regional Center of Popular
Cultures, the Yucatec Maya Language Academy, and the Department of Indigenous
Education of the Secretary of Public Education. The primary difference between the
symbols used in the consonant inventory above and the orthographic representations are
listed in Figure 2-11.
Figure 2-11
Orthographic Representation
/š/ is represented as x, as in bajux (/bajuš/) – ‘how much’
/č/ is represented as ch, as in chiich (/či:č/)– ‘grandmother’
/ʔ/ is represented as ’, as in ki’ (/ kiʔ/)– ‘delicious’
/h/ is represented as j, as in janal – ‘food’
glottalized vowels are written V’V, as in ju’un (/huʔun/) – ‘paper’
long vowels are represented as VV, as in kaan (/ka:n/) – ‘snake’
2.5.3
Phonotactics
2.5.3.1 Initial-C, Final-C
Among the most stringent phonotactic rules evident in YM is that all words in the
language must begin and end with a consonant. Words that have historically been written
as vowel-initial have been claimed to be glottal-stop initial15 (Bricker, Po'ot Yah, & Dzul
de Po'ot 1998). Words that appear to end in a vowel are posited by some to be in reality
glottal-stop final or with an epenthetic /h/ occurring in final position. Using the
framework of Optimality Theory, Orie & Bricker (2000) posit that the constraint Initial 15
The Bricker et. al. dictionary of YM is organized accordingly, though other Mayan
dictionaries that are based on Mayan orthography do not include the initial glottal stop.
46 C, Final-C is applicable to YM, and that the constraint “applies exceptionlessly in the
language”. Words in YM that are incorporated from Spanish, such as ‘arroz’ – ‘rice’,
acquire epenthetic glottal stop initially so that they comply with the Initial-C, Final-C
constraint. Orie & Bricker (2000) argue that in cases of Spanish loans, glottal stops are
epenthesized in initial postion, causing a word like ‘albahaca’ – ‘basil’ to be incorporated
as /ʔáalbahakah/.
As a result of this constraint, consonant clusters are produced across root and
word boundaries, which are dealt with in a variety of ways (Orie & Bricker, 2000: 2956). Although clusters of different consonants are allowed in YM, as is seen in words such
as /junp’íit/ - ‘a little bit’, if the combination of two words produces adjacent identical
consonants, these undergo either deletion or debuccalization. If the identical consonants
are anterior sibilants, glides, liquids, nasals or laryngeals, one of the segments is deleted,
as in Figure 2-12.
Figure 2-12
a.
b.
c.
d.
e.
(Orie & Bricker, 2000: 296)
Input
k’a’aw wíinik
páapay yúuk
t’a’ah hanal
p’is sakal
maas sa’
Output
k’a’a wíinik
páapa yúuk
t’a’a hanal
p’i sakal
maa sa’
Gloss
skinny man
spotted fawn
well-fed
measure cloth
more gruel
If the identical consonants are stops or ejectives, rather than deletion, the first of
the two segments undergoes debuccalization, or loses its place of articulation. Nonejective stops reduce to [h], and ejectives reduce to [ʔ], as seen in Figure 2-13.
47 Figure 2-13
(Orie & Bricker, 2000: 296)
Ejectives
Input
a.
‘éek’ k’at
b.
sop’ paach16
c.
xot’ tuunich
Output
‘ée’ k’at
so’ paach
xo’ tuunich
Gloss
black clay
pile on one’s back
break stone
Non-Ejectives
Input
d.
páap p’aak
e.
ts’op paach
f.
pit ta’
Output
páah p’aak
ts’oh paach
pih ta’
Gloss
spicy tomato
punch back/rear side
scoop up excrement
Orie & Bricker (2000) argue that this suggests that YM has an OCP17 restriction against
adjacent segments with identical place features, and that it solidifies the argument that the
laryngeals [h] and [ʔ] are placeless laryngeals, since they are acceptable alternatives.
2.6
Hiatus Avoidance and Pronouns
As was discussed above, diphthongs are not found in YM, and in some cases of
morphological concatenation when two vowels are found in adjacent positions due to the
placement of morphemes, a glide is inserted between them. In particular, this is found to
occur systematically when constructing phrases with pronouns. The system of pronouns
in YM is considered to be split-ergative, and verbs are inflected for transitivity. There are
2 sets of person markers in YM, which are shown in Figure 2-14.
16
Example 2b shows that this process occurs not only when two ejectives are adjacent,
but when an initial ejective is followed by a non-ejective of the same place. Orie &
Bricker (2000) argue that this is evidence that the restriction of identical adjacent
consonants refers specifically to place of articulation, and manner is irrelevant.
17
The Obigatory Contour Principle (Leben 1973) states that certain consecutive identical
features are banned in underlying representations.
48 Figure 2-14
1st
2nd
3rd
Person markers in YM
Set A
Singular
Plural
in(w)
k
a(w)
a(w)...e’ex
u(y)
u(y)...o’ob
Set B
Singular
Plural
-en
-o’on
-ech
-e’ex
ø
-o’ob
Transitive verbs always use Set A markers as subjects, and Set B markers as Objects, as
in Figure 2-15.
Figure 2-15
Transitive construction
k-uy-il-ik-en
Hab-3SetA-see-Imp-1SetB
‘He/she sees me’
There is a split in Intransitive verbs, where in the Imperfective aspect, intransitive verbs
are marked with Set A markers as subjects, while in the Perfective and Subjunctive
aspects, intransitive verbs take Set B markers as subjects, as in Figure 2-16.
Figure 2-16
Intransitive Verbs
Imperfective Intransitive
k-in- hóok’-ol
Hab-1SetA-to go out-IMP
‘I go out’
Perfective Intransitive
(h) hóok’-ø-en
Comp-to go out-Perf-1SetB
‘I went out’
The Set A singular and plural markers for 2nd and 3rd person undergo phonological
alternations, in that they have an epenthetic glide (/w/ follows 1st and 2nd, /y/ follows 3rd)
that appears when there is a following stem that is vowel initial.
This hiatus avoidance technique is robust in YM, and can even be found in the
Yucatecan dialect of Spanish, which as described in Section 2.2, has a characteristic in
49 which glottal stops are inserted between word boundaries that would produce hiatus (as
in ‘ocho años’  [otʃoʔaɲos]) (Michnowicz 2008).
2.7
Syllable Structure
Lexical stress patterns, because they are paradigmatic in nature, in that emphasis
inherently relies on multiple units to compare, is understandably intertwined with word
length, quantity and syllable structure. The syllable, or in some languages the mora, is
argued by most scholars to be the unit of emphasis, and the domain of lexical stress.
According to Gussenhoven & Teeuw (2008) stressed syllables are also the tone-bearing
units in YM. In order to present a comprehensive analysis on whether or not metrical
stress exists in YM, therefore, we must first determine how YM words are syllabified.
Section 2.7.1 describes the way native speakers syllabify words in YM.
2.7.1
Syllabification
Scholars have noted that languages tend to syllabify in order to maximize onsets
rather than codas, a tendency referred to as Onset Maximization (Kahn 1976, Selkirk
1982). Data from YM speakers confirms that this tendency is upheld in YM. In this
study, 4 native speakers were given 50 YM words and asked to give their own intuitions
on how many syllables each word contained, and which were stronger or weaker. The
speakers were encouraged to say the words aloud repeatedly during this process, and they
repeated the words with pauses between syllables. The speech during this portion of the
experiments was transcribed, and pauses between syllables were recorded. The data from
this portion of the study shows that speakers consistently paused in places that support
Onset Maximization and avoidance of consonant clusters in onsets or codas, examples of
which are outlined in Figure 2-17.
50 Figure 2-17
Speaker Pauses Between Syllables
Word
a. táanchumuk - ‘half’
b. seenkech – ‘very’
c. yoochel – ‘shadow’
d. chambel – ‘slow’
f. máakalmaak –‘anyone’
g. uuchben – ‘old’
7. báalam – ‘jaguar’
h. awat - ‘to yell’
i. jats’uts – ‘cute’
j. junp’íit – ‘a little bit’
k. máasewaal - ‘indigenous’
l. sitriiyo – ‘species of bird’
m. buláak’ab -‘all night’
Pauses
táan.chu.muk
seen.kech
yoo.chel
cham.bel
máa.kal.maak
uuch.ben
báa.lam
a.wat
ja.ts’uts
jun.p’íit
máa.se.waal
si.trii.yo
bul.áa.k’ab
With the exception of (l) and (m)18, the pattern evident in this data is that consonantal
phonemes are aligned as onsets if possible, but not if they will create an unacceptable
consonant cluster. So for example, (j) is syllabified with a coda in the initial syllable to
avoid the cluster /np/. Example (l) shows the only acceptable consonant cluster, /tr/,
which is exceedingly rare in native Mayan words, as is the sound /r/, which is thought to
be a borrowing from Spanish.
Example (m) is irregular, since if it followed the pattern of the others it would be
syllabified /bu.láa.k’ab/ instead of /bul.áak’.ab/. However this word is unique in this data
set for its morphology, because it is not a stem with inflectional or derivational
morphemes attached, but can more accurately be described as a compound of the
adjective ‘all’ and the noun ‘night’, creating the compound word ‘bul-áak’ab – all-night’.
If this word is actually being treated as a compound or even as two separate words, we
might expect that a glottal stop would be inserted before the high vowel in the word
18
These were the only anomalous cases present in the data, the rest show the pattern
found in examples (a)-(k).
51 ‘áak’ab’, due to the Initial-C constraint. The spectrogram in Figure 2-18 shows that this
is indeed the case.
Figure 2-18
Spectrogram of /bul-áak'ab/
In this example of a speaker’s utterance of ‘buláak’ab’, you can clearly see a
glottal stop before the high vowel /áa/, which as discussed in section above on the InitialC, Final-C requirement in YM, is inserted word initially. This is strong evidence that the
word ‘áak’ab’ is treated as a separate lexical item, and therefore the anomaly in
syllabification is to be expected.
Another trend that can be seen in the native speaker syllabification data is that
most speakers treated the glottalized vowels as one syllable, for three out of four speakers
assigned this task. This data as can be seen in Figure 2-19.
52 Figure 2-19
Syllabification of Glottalized Vowels19
Word
a. mina’an – ‘none’
b. kankala’as – ‘lizard’
c. koja’anil - ‘illness’
d. ba’alche’ – ‘tree’
e. chala’at – ‘side’
f. jela’an – ‘different’
Syllabification
mi.na’an
kan.ka.la’as
ko.ja’a.nil
ba’al.che’
cha.la’at
je.la’an
This gives support to the idea argued in Frazier (2009) and Gussenhoven &
Teeuw (2008), that this vowel type does not consist of two separate vowels separated by
a full glottal stop, but can instead be more accurately analyzed as one long glottalized
vowel. The trend for speakers to treat them as one syllable rather than two is evidence in
favor of this, because if a true glottal stop existed, that would create a break deriving two
syllables.
In summary, the way the speakers appear to syllabify words in YM is consistent
with Onset Maximization, and glottalized vowels tend to be treated as monosyllables.
Both of these facts are important for the following analysis, since knowing exactly what a
syllable consists of in YM is integral to analyzing whether a particular type of syllable
bears stress.
Before diving into the analysis of the data, the following chapter outlines the body
of existing work on YM, and illuminates the gap in the knowledge about YM stress that
the analyses will attempt to fill in. After describing the work of previous scholars,
examples of Spanish loan words from both the data elicitation and conversational
sections of the study will be analyzed in Chapter 4. Chapter 5 will then present statistical
19
The fourth speaker treated most glottalized vowels as two syllables, i.e. /mi.na’.an/
rather than one, while the others consistently treated them as a single syllable.
analyses on the native speaker intuition data, as well as the data from words spoken in
isolation and words in frame sentences.
53 54 CHAPTER 3 PREVIOUS WORK IN YUCATEC MAYA PROSODY
3.1
Introduction
Yucatec Maya has a history of extensive study that not many Native American
languages have had. It has been a subject of interest to historians, anthropologists,
ethnographers and linguists studying the Classical Mayan culture and its hieroglyphic
writing system that has fascinated so many since Europeans first came to Central
America. (Tozzer, 1921) cites over 700 texts that are in YM or deal with YM directly or
indirectly, some of which date back to the 16th century, and a great deal more has been
learned since Tozzer’s time. As important as this bulk of literature is for the
documentation and history of the YM language, it is important to keep in mind when
attempting to use this data for scientific purposes that the data collected by early scholars
has a variety of issues, as Straight states in his 1976 work on language acquisition in YM:
“the vast bulk of this literature is of more antiquarian than scientific value. Most
of it is characterized by poor phonetic transcription (and worse phonological
analysis), Latinizing grammatical sketches, extensive plagiarism…and
frustratingly superficial and incomplete data-gathering.” (Straight, 1976: xviii)
Though a vast number of texts spanning centuries of study are available on YM, few of
these provide scientifically rigorous analyses or data to aid the student of modern
Yucatec speech. In the past few decades, the gaps left in our knowledge of the YM
sound system have begun to be filled in, though significant gaps remain.
In particular, the area of prosody remains an area of relative obscurity in the
centuries of work on YM. Only in the last few decades has YM prosody been treated as a
topic in need of serious exploration, with most texts and grammars treating it fleetingly, if
at all. An area of particular obscurity in early Mayan research is the treatment of tonal
and accent features.
55 In the early days of Mayan scholarship, the tonal properties of the language were
unclear. Many early scholars of YM noticed some kind of distinction in vowel type, but
did not look far enough to determine its nature. In general, attempting to disentangle the
notions of ‘tone’ ‘stress’ and ‘accent’ in the early literature on YM is very challenging.
Many scholars used these terms interchangeably, and without explaining what phonetic
features define them.
In this chapter, I provide a brief history of the Spanish colonization and
missionization practices that had a large impact on the YM language. I then give an
overview of the influential work that has been done on the sound system and prosody of
Yucatec Mayan, from the observations of those who first studied it, to the most current
analyses, and illustrate the areas in which more work is needed to properly illuminate the
question of whether the canonical notions of stress, accent or tone indeed exist in YM.
3.2
History of Spanish impact on YM
The beginning of the study of Yucatec Maya can be traced back to the initial
figures in the Spanish conquest and missionization of the Yucatán peninsula, starting in
the 16th century. Hanks (2010) describes in great detail the process of missionization that
the Spanish undertook in the Mayan world, and pays particular attention to the effect this
process had on the language and linguistic practices of the people. According to Hayes
(2006, 2012), there were roughly 240,000 speakers of Yucatec Maya living in the
Yucatán peninsula when the Spanish arrived in the 1540’s, and this population underwent
dramatic struggle during this time period, and loss of life due to disease, famine and
warfare was common. The population of Yucatec Mayan people did not regain the level
of population found before the Spanish conquest until the 18th century, and during that
56 time period the language was highly affected by the political, social and religious changes
put in place by the Spanish.
Hanks describes the process of conquest and conversion of the Maya as
reducción, which denotes a re-ordering, subjugation and missionization of the society as
a whole. This process of reducción utilized language as its most important tool in reshaping the lives of the Mayan people to think and behave according to the Spanish
model. The process of conversion was undertaken in the native language in the Yucatán,
meaning those holding religious office were required to speak and translate religious
materials from Spanish into YM. The language used for translation and incorporation of
the YM people into the Spanish domain is referred to by Hanks as Maya reducido,
because of the vast changes that took place in the language when it was transformed into
a tool for religious pedagogy. This newly formed version of the YM language, in many
ways crafted by the Spanish missionaries in order to better communicate and persuade
the people that their belief system and ways of living were correct, was propagated
throughout the YM population.
Not only did the YM speakers which the Spanish had successfully incorporated
into their societal and religious structure use this Maya reducido, the YM people that
were living in un-subjugated areas were shown to be highly influenced by it as well. The
flow of people back and forth from the Spanish enclaves to la montaña, or the unsubjugated lands, led to the Spanish-influenced language being used even in the YM
57 writings that were considered idolatrous and forbidden by the Spanish, including works
such as the Books of Chilam Balam20.
So for almost five centuries, the YM language has been influenced by Spanish,
not just because of passive contact between the two cultures but due to active
modification of YM by the early Spanish missionaries. In order to fulfill their goals of
conversion of the native populace, these missionaries had to become the first scholars of
YM, and the first detailed descriptions of the language come from these sources. These
early descriptions will be summarized in the following section.
3.3
Early Descriptions of Yucatec Prosody
A plethora of early grammars of Yucatec exist today, starting with works
recorded by early Spanish settlers in the Yucatán described above. These texts are often
invaluable and fascinating views into YM as it was spoken in the early days after the
arrival of the Spanish, seen through the lens of Spanish conquistadors and missionaries
who recorded these documents. Though because they primarily documented the language
in order to be better able to translate religious texts into the language of those they meant
to convert to Christianity, they are of limited value to students of modern speech,
particularly of prosody. As discussed in Tozzer (1921), this is often because the methods
used to present YM in these works were based on what he calls the “Latin model”
(Tozzer, 1921: 7), meaning they used Spanish as a baseline for comparison, rather than
looking at YM as the unique and quite unrelated language that it is. Because their main
purpose was translation and conversion, they tended to look specifically for sounds and
20
The Books of Chilam Balam include nine texts written in the 17th and 18th centuries in
the YM language using the Latin script. These books were written on topics having to do
with pre-conquest YM traditions, calendar and mythology, and were considered highly
forbidden by the Spanish (Hanks 2006).
58 structures that existed in Spanish. When a direct translation for a particular word or
phrase did not exist, as was frequent due to fundamental differences in the two languages
(examples being a higher degree of polysynthesis in YM, and a vastly different tense and
aspect system), these authors often recorded something that later scholars would realize
was only vaguely related to its gloss, or in some cases a complete misunderstanding.
Though today we can see that these texts are full of errors and misstatements, they remain
the basis upon which further study evolved, and they fundamentally shaped the way YM
was studied by future scholars.
Three of the most important early grammars were those of Juan Coronel (1620),
Gabriel de San Buenaventura (1684) and Pedro Beltran de Santa Rosa María (1746),
which are described in detail in Hanks (2006) and Tozzer (1921), though the texts
themselves are rare and difficult to find today. Though the antiquity of these works
means the language they describe may be quite different than the Yucatec Maya spoken
today, particularly due to the higher degrees of bilingualism with Spanish in modern
times, the descriptions can still prove to be useful to the modern study of YM. Tozzer
says that of these three, only that of Beltran mentions the phonetics of the language at all,
though he does not go into detail on what phonetic topics were discussed. In one
footnote, Tozzer briefly explains what he calls the ‘double vowel’ in Yucatec (what
scholars today would refer to as the High or Low toned vowels which co-occur with a
length distinction), which he says none of the early Spanish grammarians had mentioned
(Tozzer, 1921: 20). Though it is possible that when the early grammars were written, this
tone/length distinction did not exist, it seems more likely that the authors overlooked this
aspect of YM, and in general did not go in to a great amount of detail on the sound
59 system at all. Tozzer’s work itself, which was a major compendium of all that had been
previously done on the YM language, shows a focus on the structure and morphology of
YM rather than the sound system, with only 10 pages dedicated to phonetics (which will
be discussed below), and around 80 focusing on syntactic and morphological
constructions.
3.3.1
Tozzer’s 1921 Account of Yucatec Phonetics
Tozzer’s 1921 Maya grammar was among the first to systematically and
comprehensively lay out and compare the bulk of all the available work on YM that
existed, and was an invaluable tool for most modern scholars of this language. Not only
does he go into great detail on what was known about YM, but he presents data of his
own which he collected from a single speaker while living in Mérida.
The vowel system Tozzer describes shows a common area of confusion amongst
scholars; how to categorize what had been recognized as distinct vowel “types”. Exactly
what causes this distinction is not clear in the early literature, though Tozzer, among
others (Andrade, 1940), attributes it to a difference between long and short vowels, while
also noting a third distinction between what he calls ‘double vowels’. What they appear
to be describing might be one of two features that have been classified by modern YM
scholars: the High-Low tonal distinction, which co-occurs with a lengthened vowel, and
which contrasts to a short Neutral vowel, or the glottalized or rearticulated vowel
discussed in section 2.5.1.2.1. Tozzer’s vowel inventory consists of [a], [i], [e], [o], and
[u], and a long version for all of these except for [o] (which he says he did not observe,
but which is common in modern dictionaries of YM), as well as these ‘double vowels’
which appear to correspond to today’s ‘glottalized’ vowels. A distinction between long
60 and short was noted as well, but Tozzer’s system of YM vowel types appears to be
lacking the observation that the long vowel type is made up of both High and Low tones.
On the topic of ‘accent’, Tozzer devotes one paragraph, in which he says he has
not marked it orthographically because ‘it is in part dependent upon the length of the
vowel’, and describes his impression that ‘contracted syllables usually seem to have
greater stress of voice laid upon them’ (Tozzer, 1921: 27).
3.3.2
Early Observations of the role of pitch in YM
Another valuable account in the area of accent in YM comes from Andrade
(1940), who gives a general overview of accentual patterns he has observed. Andrade, as
Tozzer (1921), does not make any reference to a pitch based phonemic distinction in YM,
and maintains the same model Tozzer details, with a contrast between long, short and the
so-called ‘double vowel’. Andrade, however, does make an observation regarding what
part pitch may play in YM:
“The relationship between pitch and accent in Modern Yucatec has not been
studied sufficiently to permit generalization; it may be said, however, that
frequent examples occur in which greater stress accompanies lower pitch and
lesser stress accompanies higher pitch (e.g., le ˈnuukulo:boʔ, 'the utensils', with
high pitch on –lo:-, lower pitch on -boʔ)” (Andrade, 1940)
This description seems to go against the pattern we would normally expect, which Hayes
(1995) and De Lacy (2002) have noted to be common in the world’s languages, in which
higher tones are attracted to stressed positions. It is difficult to discern, however, exactly
what cue he uses to determine stressed syllables, and does not give more data to allow us
to replicate this finding.
Andrade goes on to summarize the features he has observed regarding ‘accent’ in
YM, which he states ‘appears to be determined lexically’, and which are likely referring
61 to glottalized vowels as well. Other observations about accent made by Andrade (1940)
include that disyllabic words tend to have ‘accents’ on the ultimate syllable (including in
Spanish loan words), but that there are many exceptions in which a disyllabic word has
stress on the initial syllable, which can be seen in the following figure.
Figure 3-1
Exceptions to Ultimate Stress (Andrade, 1940)
Gloss
a. 'grove'
b. 'turkey'
c. 'pig'
d. 'machete'
e. 'husband'
f. 'drink'
Andrade 1940
ˈmakan,
ˈulum
ˈk'ek'en
ˈmaskab
ˈiʧam
ˈuk'ul
Looking at these exceptions from a modern perspective, one might wonder
whether introducing the notion of phonemic pitch distinctions may shed light on some of
these irregularities. If we assume that the disyllabic words listed by Andrade and seen in
Figure 3-1 as breaking from the norm and having stress on the penultimate syllable might
correspond to words that are today classified as having a High-toned long first syllable,
we find a much cleaner set, as can be seen in Figure 3-2.
Figure 3-2
Comparison of Andrade & Navarette
Word
Andrade 1940
'grove'
'turkey'
'pig'
'machete'
'husband'
'drink'
ˈmakan,
ˈulum
ˈk'ek'en
ˈmaskab
ˈiʧam
ˈuk'ul
Navarette 2009
not found
úulum
k’éek’en
máaskab
íicham
uk’ul
All but one of the words Andrade lists as irregularities that are found in the modern
dictionary correspond to cases of high-tones in the initial syllable. This analysis makes
one start to believe that there is a high likelihood that early scholars of YM such as
62 Andrade heard and interpreted high-toned syllables as stress, and have based their
arguments for default stress patterns partially on lexically specified tonal distinctions. If
this is the case, then their accounts of what kind of accent patterns occur in YM becomes
less and less clear, because of the possibility that high tone and accent have been
conflated.
3.3.3
Phonemic Pitch
The first scholar to posit the possibility of phonemic pitch in YM was Kenneth L.
Pike in 1946, who undertook the task of clearing up some of this confusion over vowel
length, pitch and glottalization. Pike (1946) puts forward what he posits to be minimal
tone pairs, which had previously been categorized as long and/or accented, for both short
and long vowels.
Figure 3-3
Minimal Tone Pairs (Pike K. L., 1946)
Short Vowels
a. móčtal -‘to be crippled’
b. k’át tu bel – ‘someone has closed
off the road’
c. múts’ a wič – ‘your eyes are
closed’
d. páhtal – ‘be able’
e. p’íl a wič - ‘your eyes opened’
Long Vowels
f. lúubul – ‘to fall’
g. líikil – ‘to raise’
h. láakal – ‘to detach’
močtal – ‘to double onself up’
k’at tu bel – ‘put it in the road’
muts’ a wič – ‘close your eyes’
pahtal – ‘turn sour’
p’il a wič – ‘open your eyes’
luubul – ‘fallen’21
liikil – ‘raised’
laakal – ‘detached’
In current dictionaries the words in (d) ‘páhtal’ – ‘be able’ and (a) ‘móčtal’ – ‘to be
crippled’ become ‘páahtal’ and ‘móočtal’, while ‘muts’’ and ‘p’il’ surface in modern
21
See discussion above for Beltran’s 1746 observation of this pair.
63 dictionaries as ‘muuts’ – ‘to close the eyes’ and ‘p’iil’ – ‘to open the eyes’, both with
long low vowels.
The main distinction between this account and current work is that there is that he
argues for tone found on short vowels, which according to modern scholars, is not found,
but instead the underlying form is one in which High and Low tone are underlyingly
always associated with long vowels. With the exception of (d) in Figure 3-3 all other
examples are instances of semantically related verbs, and (f) through (g) are specifically
instances of an infinitive verb and its past participle, a tonal phenomenon which is
supported throughout literature of YM. And on the topic of short vowels, he admits that
he did not find any monosyllabic short vowels that had a High tone, which gives evidence
for the later theories that short vowels are underlyingly tone-neutral, but that long High
and Low vowels can sometimes shorten in rapid speech. As for the phonetic cues that
Pike is using to determine these distinctions, he describes that long vowels come with a
low pitch that is ‘quite level…with no noticeable glide up or down’, or with a high pitch
that ‘tends to glide upwards a bit, higher than their (already high) starting point (Pike
1946: 84).
Interestingly, Pike argues that though phonemic pitch certainly exists in YM, he
would ‘hesitate to call Maya a “tone language”, due to what he terms a non-symmetrical
pitch distribution, making it fundamentally different from a canonical tone language like
Chinese or Mixteco (Pike 1946: 87). His final assessment is that “the whole pitch
problem, for Maya proper and its related languages, needs thorough investigation.”
Throughout this analysis of phonemic pitch, Pike makes little mention of the possibility
64 of accent or stress interacting with the tone system, and what role it might play (if any) in
the YM prosodic system.
3.4
Modern Orthography and Tonal Analysis
Based mainly on the work of Blair (1964), a textbook was created to help English
speakers learn YM, put together by Blair and Refugio Vermont-Salas in 1965 and later
edited several times over the years by Vermont-Salas and McQuown. This textbook is
still one of the major references used for students learning YM. It is of course written for
the language learner as opposed to the linguist, and so does not go into detail regarding
the prosodic phonemes and their allophonic variations, although there are some
fundamental properties implied by the design of the text. It starts out explaining that
there are 5 different types of syllable nuclei, which can be seen in Figure 3-4
Figure 3-4
Syllable nuclei (Blair & Vermont-Salas, 1965)
Vowels of YM
V (Neutral)
V’ (Glottalized)
VV (Low Long)
VV (High Long)
V’V (Rearticulated)
a
a’
aa
áa
a’a
e
e’
ee
ée
e’e
i
i’
ii
íi
i’i
o
o’
oo
óo
o’o
u
u’
uu
úu
u’u
In the instructions as to how to pronounce these vowel types, the text describes three
types of ‘Maya accents’; one which is neutral in pitch and short in duration, and is always
denoted by a single vowel, one which is high (or gliding) in pitch and long in duration,
written as a double vowel with a rising accent, and one which is low (or level) in pitch
and long in duration, written as a double vowel with no accent. This orthography was
determined in 1984 in a meeting organized by the Yucatán Delegation from the National
Institute for Adult Education (INEA). At this meeting, representatives of various
65 institutions created an orthography specifically for those who are learning and using the
language, which was appropriate for people of all backgrounds to use.
The vowel distribution used in this text, as can be seen in Figure 3-4, includes a
category for a short vowel followed by a glottal closure, For the vowel type V’ listed
here, there is a question as to whether or not this could be analyzed as an allophonic
variant of the Short vowel when it occurs in word final position. It appears to be a
method to explain pronunciation to a language learner, since in word final position the
glottalization of the vowel is an important rule for pronunciation in YM. Otherwise these
four categories are implied to be all in phonemic contrast, for which minimal pairs occur
for most if not all vowels.
(Blair & Vermont-Salas, 1965) describe an alternation in the pronunciation of
vowels, in which Glottalized syllable nuclei may interchange with High Long nuclei, and
in rapid speech may shorten to a single High toned short vowel. So for example /a’a/
could become [áa] in some situations, and /áa/ could become /á/. Though it is not stated
explicitly, it seems apparent that the authors of this text do not view the short High toned
vowel as phonemic, but rather allophonic, which explains why it was listed by Pike
(1946), therefore leaving the following vowel types as the predominant underlying vowel
types.
Figure 3-5
Four Vowel Types of YM
V (Neutral)
VV (Low Long)
VV (High Long)
V’V (Rearticulated)
a
aa
áa
a’a
e
ee
ée
e’e
i
ii
íi
i’i
o
oo
óo
o’o
u
uu
úu
u’u
3.5
66 Formalizing of the underlying YM Nuclei Types
Straight (1976), in his work on child language acquisition, outlines the ‘four
underlying types of syllabic nucleus’ present in YM: Neutral, Low-tone, High-tone and
Laryngeal (or glottalized). These correspond to those that are implied in the teaching
materials of Blair & Vermont-Salas (1965), shown in Figure 3-5. Table 3-1 shows
Straight’s feature matrix for these prosodic types:
Table 3-1
Yucatec Maya Underlying Segment Inventory (Straight, 1976: 45)
Prosodic Types
Neutral
Low
High
Glottalized
-
+
+
+
contour
-
-
+
+
glottal
constriction
-
-
-
+
tonality
He confirms a claim from Blair (1964) that the ‘double vowel’ noted in previous works is
not realized phonetically as a vowel followed by the complete closure of a glottal stop
with the vowel repeated after, but instead Straight argues is closer to a vowel pronounced
with creaky voice (Straight, 1976: 36). He also argues that the four types of syllabic
nuclei undergo ‘rules of stress and tone modification’ that lead to 9 observable variations
(not unlike McQuown 1968’s list of syllable types, though whether or not these variations
are allophonic, and if so what rules generate them, is not explicated here), listed in Figure
3-6.
67 Figure 3-6
Inventory of Syllabic Nuclei (Straight, 1976: 38)
Long High
Broken High
Long Glottalized
Short High
Short High Glottalized
Long Low
Short Low
Short Neutral Glottalized
Neutral
This is a crucial phonological observation, which most likely was derived from the Blair
& Salas (1965) textbook Spoken (Yucatec) Maya, in that he is saying that the many ways
in which a syllable nucleus might be realized (due to rapid speech, intonation, or other
processes) all originally stem from 4 underlying forms. This variety of realizations for
these 4 types of underlying syllabic nuclei most likely contributed to YM’s prosodic
system being analyzed as non-symmetric by Pike (1946), and considerably clears up the
notion of tone in YM. It does not, however, clear up the notion of whether or not YM has
a separate or intertwined system of stress or accent.
3.5
Current Account of Phonetics in YM
More recent attempts have been made to characterize aspects of prosody in YM,
though none of these have specifically and comprehensively dealt with lexical stress. The
following sections summarize the findings of recent scholars on the topic of YM prosody.
3.5.1
Frazier (2009)
Amongst the most recent and comprehensive is Frazier (2009), which provides an
extensive account of the acoustics of the 4 types of vowel nuclei, which are argued to
differ based on dialect, using results from two production studies designed to elucidate
the phonetic nature of the vowel shapes in YM. Looking only at the production of
monosyllabic words and focusing on the phonetic variation between the four vowel
68 shapes, Frazier’s study detailed the results of production data collected from speakers in
two different areas of the Yucatán peninsula. Results showed that speakers from Mérida
and Santa Elena areas are argued to comprise a Western dialect area in which all 4 types
of vowels are realized in speech, whereas speakers from Sisbicchen in the Eastern
Yucatán have characteristics of an Eastern dialect area in which tonal contrasts are not
found at all. Glottalized nuclei are found in both dialects, and are realized primarily as
creaky voice, rather than a full glottal stop in both dialects. The main phonetic
characteristics of the Western dialect of YM are laid out in Figure 3-7.
Figure 3-7
Characteristics of the Western Dialect of YM (Frazier, 2009: p 274)
a. Long vowels (High tone, Low tone and Glottalized) are about twice as long as
Short (neutral) vowels
b. Low tone vowels are produced with fairly steady low pitch
c. Short tone vowels have steady mid or mid-high pitch
d. Glottalized and High vowels have initial high pitch and a falling contour
e. In non-phrase final position only High vowels reverse to a rising
contour
f. Glottalized vowels do not contain a medial glottal stop as described in much of
the literature, but instead a more accurate description is that they start with High
pitch and end in creaky voice22
The Eastern dialect was argued to be lacking the consistent distinction between High and
Low tones that the Western dialect still preserves. The speakers I recorded for the current
study were also from the Western dialect area, and the analysis of monosyllabic words in
chapter 5 demonstrates that the characteristics in Figure 3-7 are also found in their
speech. The comparison between the findings from this study and Frazier (2009) will be
discussed in more detail in chapters 5 and 6.
A dictionary of YM as spoken in Hocabá, Yucatán was compiled by Bricker,
Po'ot Yah, & Dzul de Po'ot (1998), who also gives a brief description of the morpho 22
Stress in YM is not analyzed in Frazier (2009).
69 syntactic features of the language, as well as the prosodic features. They state that there
is a default stress pattern in YM, in which stress in a word with two or more syllables
falls on the long vowel. If the word has two long vowels, stress falls on the first syllable.
If the word has no long vowels, the last syllable is stressed.
(Gussenhoven & Teeuw, 2008), who analyzed YM prosody in order to discern the
nature of the H-tone, have a different assessment of the default stress pattern. They argue
that YM has ‘three long syllables, Long High, Long Low, and Glottalized, which are
stressed, and a word-initial short syllable, which is stressed. (p 17)’. This specifically
differs from Bricker, Po'ot Yah, & Dzul de Po'ot (1998) in two ways: 1) that the tone or
glottalization features are what attract stress, rather than length, and 2) words with only
short syllables have initial, rather than final stress.
3.5.2
Moras as Tone and Stress Bearing Units
Gussenhoven & Teeuw (2008) also make the claim that stressed syllables are the
Tone-Bearing Units (TBUs) for the Long High and Long Low syllable types, while for
the Glottalized type they argue that the mora is the TBU. Short syllables are only stressed
when in initial position, and non-initial short syllables are said to be incapable of bearing
tone. This provides an important argument for a description of YM stress, i.e. that stress
and tone always co-occur, that means stress and length always co-occur, and that stress
defaults to the initial in words with short syllables. The claim that moras are the TBUs for
Glottalized nuclei also suggests that syllable weight may play an important role in the
assignment of stress in YM, because if moras are integral to tone, they may play a salient
role in stress as well.
70 (Krämer, 2001) gives further evidence for the potential importance of the mora in
YM. Quantity sensitivity, which he says is an uncontroversial notion among Mayan
scholars, although it has not been exhaustively documented, plays a crucial role in his
analysis of vowel harmonic suffixes that surface in YM. Certain suffixes undergo vowel
harmonic alternations, but harmony is blocked when more than one consonant intercedes
between the suffix and the root vowel. Examples of this phenomenon can be seen in the
following figure.
Figure 3-8
Vowel harmony (Krämer, 2001: 11)
Subjunctive Suffixes With Vowel Harmony
a. ah-ak’
wake.up-SUBJ
b. ok-ok
enter-SUBJ
c. lub'-uk
fall-SUBJ
d. wen-ek
sleep-SUBJ
e. kíim-ik
die-SUBJ
Subjunctive Suffixes Where Vowel Harmony is Blocked
f. tùukul-n-ak
think-N-SUBJ
g. hèek'-n-ak
break-N-SUBJ
h. ts'íib'-n-ak
write-N-SUBJ
In examples a-e, the suffixal vowel corresponds to the root vowel, but in f-h, an
intervening consonant appears to block the harmony from occurring, in which case the
default vowel /a/ is used.
In other words, coda consonants are argued to act as a blocking mechanism for
harmony, and therefore must be counted as having quantity. In the examples above which
do undergo harmony, the final consonant of the root is syllabified as the onset of the final
syllable, (i.e. a.hak, o.kok, etc.), and onsets do not usually count as bearing quantity.
Therefore, according to Krämer’s analysis, the syllabification of f-h would create a heavy
71 syllable preceding the vowel harmonic suffix, (i.e. túu.kul.nak, héek.nak, etc.), and the
presence of this extra mora is what blocks the harmonic process.
A consequence of this analysis is that final syllables in Yucatec would always be
counted as heavy, because of the phonotactic restraint in YM that all words must end in a
consonant. This constraint states that if there is no final consonant, a glottal stop or glottal
fricative is added to the lexical item, which creates a situation in which words always end
in a heavy syllable. Additionally, Kramer claims a stress system exists in YM that is
quantity sensitive, so it follows that because of the constraint that requires all final
syllables end in a consonant, final syllables must bear stress in YM. Kramer proposes
this, and also claims that initial syllables attract stress.
Using the data seen in Figure 3-9 below, which are intonational contours taken
from Blair & Vermont-Salas (1965), Krämer argues that stress is assigned to the initial
and final syllables in a phrase, as well as to any heavy syllables intervening between
them.
Figure 3-9
Stress assignment (Krämer, 2001: 12)23
a. 2ku. me. 2yah2→
PR.3SG work
'he works'
b. 2h me. 2yah.-n-a. 2k-en1↓
PAST work-N-SUBJ-1SG
'I have worked'
c. 2way. ka. me.ya. 2h-eʔ2↑
Q PR-2 work-TERM
'Are you working here?'
Superscript numbers, representing levels of intonational accent assignment on the phrase,
are always assigned to the initial and final syllables in both single words (as in example
23
The superscript numbers in this figure represent the following: 3 indicates high pitch
and intensity, 2 indicates medially high, 1 lower than medial, and ∅ neutral intonation.
The arrow at the end of each phrase in the examples below indicates whether the terminal
intonation contour of a phrase is stable, falls or rises.
72 a), or more complex phrases, (in examples b and c). In b, the syllable ‘yah’, which is
heavy, is assigned an accent, while in example c, many intervening light syllables are not
assigned an accent.
3.5.3
Foot Structure in YM
Krämer (2001) claims that foot structure in YM consists of trochaic, bimoraic,
symmetric feet, a diagram of which can be seen in Figure 3-10.
Figure 3-10
Yucatec Trochaic Feet (Krämer, 2001: 13)
(s = strong, w = weak, µ = mora)
a.
(C
F
b.
F
s
w
s
w
µ
µ
µ
µ
V
V/C)
(C
V
C
V)
So bimoraic feet are built, using either two light syllables or one heavy syllable, at both
boundaries of the word or phrase, and intervening heavy syllables also attract stress.
Krämer’s analysis of the foot structure of YM also argues that light syllables that appear
in between the initial and final feet are not footed, and therefore do not receive any stress.
The foot structure of the examples in Figure 3-10, using Krämer’s analysis, can be seen
in Figure 3-11 below.
73 Figure 3-11
Foot Structure Examples (Krämer, 2001: 12)
a. 2ku. me. 2yah2→
PR.3SG work
'he works'
(ku.me)F (yah)F
b. 2h me. 2yah.-n-a. 2k-en1↓
PAST work-N-SUBJ-1SG
'I have worked'
(h.me)F (yah)F na (ken)F
c. 2way. ka. me.ya. 2h-eʔ2↑
Q PR-2 work-TERM
'Are you working here?' (way)F ka.me.ya (heʔ)F
Interestingly, Krämer purposefully omits data in which tones are present, because
the analysis makes no claims on whether or not tone would also attract stress, as it does in
many languages. This topic, along with whether or not Krämer’s proposals are supported
in the YM data collected for this study will be discussed Chapters 4 and 5.
3.6
Summary of Previous Work in YM
Through the early years of Mayan language study, the prosodic elements in YM
were often left unclear and undocumented. From the beginning, some kind of prosodic
prominence was noted, whether it was called accent, stress, or tone. What has become
clear over the years is that a tonal distinction between High and Low exists in YM, and
according to (Frazier, 2009) specifically in the Western dialect of the language. Stress, on
the other hand, has been an enduringly mysterious and hard to define entity in YM. Most
scholars agree that Long vowels, i.e. High, Low and Glottalized, are stressed (Bricker,
Po'ot Yah, & Dzul de Po'ot, 1998) (Gussenhoven & Teeuw, 2008) (Krämer, 2001).
Some have also noted that initial vowels can be stressed (Bricker, Po'ot Yah, & Dzul de
Po'ot, 1998), (Krämer, 2001), and that final syllables bear stress (Krämer, 2001). Finally,
(Krämer, 2001) argues that heavy syllables are also in the mix when it comes to stress
assignment.
74 One more element that could be confounding the notion of stress in all the
accounts described in Chapter 3, is the distinction between lexical/metrical stress and
intonational accent. Krämer (2001) describes stress as ‘intonational accent’, which may
well mean that it is completely separate from a metrical stress found on lexical items.
(Bricker, Po'ot Yah, & Dzul de Po'ot, 1998) describe the notion as simply stress, which
may be referring to the more basic lexical level rather than prominence at the intonational
level. What seems clear is that there is still a lot of untangling that needs to be done when
it comes to stress in Yucatec Maya. According to previous research on YM prosody, it is
unclear whether or not there is actually a metrical stress system present in the language,
or whether it might be similar to Kuot in which there are no lexical accents at all. The
analysis of speech data from speakers of Yucatec Mayan is needed in order to shed light
on the complex issue of prosody in YM, and in the following chapter, I begin this
analysis by looking at some case studies of loan words from Spanish into YM.
75 CHAPTER 4 SPANISH LOAN WORD INCORPORATION
4.1
Introduction
During the data collection process, multiple instances of Spanish loan word
incorporation were recorded. When Spanish words were incorporated into Yucatec
morphological and syntactic processes, they were often, but not always, modified to fit
into the native Yucatec prosodic system. This phenomenon of linguistic 'syncretism', in
which bilingual communities of speakers can choose to make their utterances follow the
rules of either language as an act of claiming membership in one community or another,
was described for the Nahuatl (or Mexicano) speaking communities by Hill & Hill
(1968), Hill (1999). These works describe syncretic practices as falling along a
continuum from 'more Mexicano' to 'more Spanish', and where an utterance lies can give
insight into how much indigenous identity is being claimed by the speaker. The same
process appears to be occurring in YM, where Spanish words are incorporated to varying
degrees into YM phonological and morphological systems. Instances of phonological
syncretism (when Spanish loan words are prosodically modified) are especially
interesting to the study of stress, specifically because the Spanish input words do not
have underlying tone, nor do they have phonemic length. When they are uttered within
the YM 'accent', the prosodic changes they undergo can provide crucial information on
what the default stress patterning of YM might be, information which may be just the
clearer window needed to understand the intricacies of YM stress.
When the rules of a particular system are unclear, looking at syncretic practices
when utterances shift from one set of rules to another can be an invaluable tool in the
discernment of the underlying form. This chapter presents data from natural syncretic
76 practices of Spanish loan word adaptation into YM into whether or not metrical
prominence can be found in YM words.
I will first discuss the notion of ‘syncretism’ and why it could prove useful for the
diagnosis of stress in YM. Following this I will look specifically at examples from my
data in which the speakers utter the same word in both a Spanish and YM ‘accent’ to
determine whether or not evidence exists to support or disconfirm the arguments from
previous scholars about the existence and placement of stress in YM.
4.2
Syncretism in Language
According to Hill (1999), syncretism is defined as “suppression of a relevant
opposition under certain determined conditions” (p. 1), and was originally from the study
of religion, but was used by Kurylowicz (1964) to describe linguistic practices. This act
of suppressing one side of an opposition or the other plays out linguistically when
communities are highly bilingual, and therefore have a choice between two linguistic
repertoires when speaking. In Hill (1999) and Hill & Hill (1986)’s work on syncretic
practices among Mexicano (Nahuatl) speakers, they describe a linguistic situation in
which a dichotomy between “legitimo Mexicano” and Castellano (Spanish) is highly
relevant to members of this bilingual speech community. They were found to use
syncretic practices in all aspects of speech including prosodically. Hill (1999) describes
an instance of phonological syncretism in the following excerpt.
‘For instance, the Spanish word cajon 'coffer', 'chest' could be pronounced
in Mexicano as [kaxón], or it could be shifted toward the Mexicano end of the
syncretic continuum by shifting the stress to [káxon], reflecting the invariant
penultimate stress of indigenous Mexicano words, or, even further, by
pronouncing the [x] as [š], reflecting the phonological pattern of the earliest
borrowings from Spanish. Thus [kášón] might be a self-conscious performance of
indigenous identity, especially on the part of a young or middle-aged male
77 speaker, while [kaxón] might be an equally self-conscious gesture of political
potency and forward-looking urbanity (Hill, 1999: 245).
Although they are culturally and linguistically distinct from Mexicano speakers, an
analogous situation can be argued to exist among Yucatec Mayan speakers. YM speakers
interviewed for this study overtly made the distinction between utterances that were 'puro
Maya', from those that were not, and recognized and commented on the influence of
Spanish on their language. In incorporating loan words into their speech, they can choose
whether to shift them into a more Mayan style prosodic system, or maintain the Spanish
system, in order to identify with one or the other dichotomies of the syncretic situation.
In a similar vein to the Mexicano example given above, when words from Spanish
are incorporated into YM utterances, they can undergo prosodic shifting from a 'Spanish
pronunciation' to one that obeys YM rules of prosody. However, unlike the situation in
Mexicano described by Hill (1999), the rules of YM prosody, particularly the rules of
stress and accent, are not clear-cut and well understood. By looking at words we know to
have undergone a prosodic shift from Spanish to YM accent structure, we can work
backwards to better understand what rules might have been applied to create the changes
we see.
Spanish has its own default system of lexical stress and phrasal accent that is
unaffected by either a tonal system (which is not present in Spanish), or a phonological
length distinctions (Spanish does not distinguish between long and short vowels) as in
YM. Studying the output of syncretic processes that act on Spanish loans into YM, i.e.
looking into exactly what happens to Spanish words when they are made to sound more
Mayan, can give us a different and possibly clearer understanding of the system of stress
in YM than we might otherwise get from looking at native words alone.
4.3
78 Previous Discussions of Default Stress in YM
As described in Chapter 3, the literature on YM that does look at stress is not
consistent as to what exactly the 'default' stress system looks like, if there is one at all.
(Bricker, Po'ot Yah, & Dzul de Po'ot, 1998) claims that a default system exists in YM,
and they describe it as falling on the final syllable if a word has no 'long vowels', if one
long vowel is present the stress falls on that syllable, and if two long vowels are present it
falls on the first long syllable. (Gussenhoven & Teeuw, 2008) who analyzed YM
prosody in order to discern the nature of the High tone, have a different assessment of the
default stress pattern. They argue that YM has “three long syllables, Long High, Long
Low, and Glottalized, which are stressed, and a word-initial short syllable, which is
stressed” (p. 17). This differs from (Bricker, Po'ot Yah, & Dzul de Po'ot, 1998) in that
the tone or glottalization features are claimed to be what attracts stress, rather than length,
and that words with only short syllables have initial, rather than final stress. Gussenhoven
and Teeuw do not state what happens when there is more than one syllable in a word with
tone or glottalization, and this is admittedly a rare situation, because the majority of YM
stems are monosyllabic, and the majority of suffixes contain Short vowels.
Krämer (2001) is among the few scholars to put forth a detailed analysis on foot
structure in YM and how it might be constrained. As discussed in chapter 3, he claims
that YM is quantity sensitive (which he also states would be uncontroversial to scholars
of YM), based on the phenomenon of blocking of particular vowel harmonic alternations.
If YM is quantity sensitive, then final syllables would always be counted as heavy,
because of the phonotactic constraint that exists in YM; that all words must end in a
79 consonant24. This constraint states that if there is no final consonant, a glottal stop or
glottal fricative is added to the lexical item, which creates a situation in which the final
syllable must always be heavy. Krämer claimed a stress system exists in YM that is
quantity sensitive, so it follows that because of the constraint that causes heavy final
syllables, that final syllables must bear stress in YM. In addition to this observation,
Krämer also proposes that initial syllables attract stress. Krämer’s argument can be
thought of as a combination of the observations of Gussenhoven & Teeuw (2008) and
Bricker, Po'ot Yah, & Dzul de Po'ot (1998), in that it claims stress is attracted to both the
initial and final syllable in an utterance given the right environments. The following
sections will look at how loan words are incorporated into YM, and explore the changes
Spanish words undergo when incorporated, and analyze whether these changes are
consistent with these scholars’ observations.
4.4
Loan Word Usage
The use of Spanish loan words is not a rare occurrence in most Mayan
communities, in fact, according to Bricker, Po'ot Yah, & Dzul de Po'ot (1998) “Spanish
loans are so common in Maya that an unlimited number of them occur in everyday
speech.” Syncretic practices are to be expected in a language with such a degree of
mixing and bilingualism. Loan words were found to be introduced sporadically
throughout the speech data collected for this study, often the speakers explicitly called
out the words as being of Spanish origin, though they were not always acknowledged as
24
Gussenhoven & Teeuw (2008) and Blair & Vermont-Salas (1965), & Bricker et. al.
(1998) all describe YM as having a system in which word initial onsets are obligatory, as
are word final codas. If a vowel is found in word final position due to phonological
processes, a glottal stop or glottal fricative is epenthesized, creating an obligatorily heavy
final syllable.
80 such, and some may have undergone incorporation to such a degree that the speakers
were unaware they originated in Spanish. As in (Hill, 1999)’s description of the variable
realizations of the Spanish word ‘cajón’ by Mexicano speakers described above, there
appear to be a range of different levels of incorporation of Spanish words into Yucatec.
In some cases, the words retain Spanish morphology, while in others they occur with
Mayan morphology.
Figure 4-1 shows examples of some loan words within the YM phrases.
Figure 4-1
Loans With Spanish and Mayan Morphology
word
a. pecados
b. odiarme
c. nietos
d. pretender
YM context
taak in kopesatik25 pecados
jach odiarme
in nietoso’obe’
baax ka pretendertik
gloss
‘I must confess my sins’
‘(he/she) really hates me’
‘my grandchildren’
‘what do you intend’
As can be seen in examples a and b in Figure 4-1, some of the Spanish loans retain their
morphology, i.e. pecado is pluralized by adding ‘-s’ instead of the YM ‘-o’ob’, and the
Spanish object clitic morpheme ‘me’ remains in place. While in examples c and d, the
YM plural (/o’ob/) and YM causative morphology (/-tik/) are incorporated onto the loan
word.
The most interesting instances of loan word incorporation in this data include
examples in which individual speakers naturally uttered the same word on both sides of
the syncretic continuum between Spanish and Yucatec. This happened in a subset of
cases, which will be discussed in detail below.
25
The verb in this sentence is very likely a loan word as well, from ‘confesar’ that has
undergone more phonological changes for this speaker and includes the YM morphology
‘-tik’, just as the verb ‘pretender’ does in example d.
4.5
81 The Syncretic Continuum in YM
Four separate instances of one speaker naturally uttering a single word twice,
once on both sides of the syncretic continuum, were present in the data collected for this
study. There were obvious prosodic differences between the two pronunciations. Two of
these instances came from one female speaker during the spontaneous conversation
portion of the data collection, and two came from a male speaker during the single word
elicitation portion. The examples in Figure 4-2 show the 4 words and their standard
Spanish pronunciations, while Figure 4-3 shows the contexts in which they surfaced in
the data:
Figure 4-2
Trisyllabic Loan Words26
Word
a. trabajo
b. abuelo
c. palabra
d. gravilla
Meaning
‘job’, or ‘I work’
‘grandfather’
‘word’
‘gravel’
Spanish Pronunciation
/tra.bá.ho/
/a.bwé.lo/
/pa.lá.βra/
/gra.βí.ya/
Figure 4-3
YM versus Spanish Context
a. YM:
Spanish:
Context
trabajo k’áax ta meyaj waye’
yo puedo porque tiene trabajo
English Gloss
‘I work (in the) forest, you work here’
‘I can because he has a job’
b. YM:
in abuelo desde uuche maya
Spanish:
mi abuelo desde antiguamente
‘my grandfather (spoke) maya since long
ago’
‘my grandfather since long ago’
c. YM:
leti’e’ ku yak le’ palabra chúuk
Spanish:
d. YM:
Spanish:
‘he says the word charcoal all the
time’
el dice la palabra carbón en todo ‘he says the word charcoal all the
el tiempo
time’
yáan gravilla yaan tu jool in
‘there is a lot of gravel in the doorway of
wotoch
my house’
hay mucha gravilla en la puerta ‘there is a lot of gravel in the doorway of
de mi casa
my house’
26
Accent marks denote the stressed syllables, and periods denote syllable boundaries.
82 These four words all happen to be trisyllabic and have stress on the second syllable in the
origin language. If the assertions of previous scholars are accurate, such as those
described in Chapter 3, words of this type would be expected to undergo prosodic shifts
when incorporated into the foot structure of YM. Figure 4-4 outlines what would be
predicted based on the previous work described in Chapter 3.
Figure 4-4
Hypotheses for placement of YM stress
a. The initial syllable will be stressed
b. A glottal stop or fricative will be epenthesized word initially for words
beginning in a vowel, and also word finally creating a final heavy syllable.
c. That final heavy syllable will be stressed
If these hypotheses were proved valid, we would expect that the initial and final syllables
will show some sort of phonetic cue for prominence on the initial and final syllables, and
also show an epenthetic element word finally.
Statistical tests were not performed on this data due to the small sample size of
only 2 subjects and 4 items. Small sample sizes such as this provide difficulty for finding
and interpreting statistical significance because tests such as ANOVA assume that data is
normally distributed, and with only 2 subjects a normal distribution is difficult or
impossible to determine. Also, even if tests were performed, the small sample size would
make the power of the test much lower, and therefore effects would have to be very
dramatic to achieve a significant finding. Because of this, the case study analyzed in this
chapter can be looked at as exploratory in nature, but further study of this process with a
larger data set would be needed to determine whether the patterns described in the
following sections are robust for loan words incorporated into Spanish. The following
section analyzes the data from the instances of incorporated loan words.
83 4.6
Pitch in YM Syncretic Shifts
First we will investigate what role pitch might play in YM prosodic placement by
looking at the pitch of each vowel in the 4 incorporated loan words and their Spanish
counterparts. Measurements were taken at 5 points along the length of the vowel, and the
chart in Table 4-1 shows the pitch range measurements for point 1 to point 5 for each
word pronounced on both sides of the continuum.
Table 4-1
Word
trabajo
abuelo
palabra
gravilla
Pitch Measurements for Spanish Loans27
Language
Spanish
YM
Spanish
YM
Spanish
YM
Spanish
YM
S1
281-231
206-204
159-154
194-201
114-99
120-115
111-106
103-126
Pitch Range in HZ
S2
242-281
202-186
183-188
192-178
105-123
122-117
116-138
116-95
S3
216-145
186-175
186-178
185-178
126-125
111-129
131-118
93-83
You can see the data with all 5 pitch points represented visually in Figure 4-5 through 48.
27
Pitch measurements were taken using Praat software from the vowel of the nucleus of
each syllable. The first data point was taken from the initial pitch of the vowel, and the
second was taken from the pitch peak or valley reached in the vowel nucleus.
84 Figure 4-5
Syllable Pitch in ‘trabajo’
trabajo 300 250 Hz 200 150 Spanish Maya 100 50 0 p1 p2 p3 p4 p5 p1 p2 p3 p4 p5 p1 p2 p3 p4 p5 Syl 1 Figure 4-6
Syl 2 Syl 3 Syllable Pitch in ‘abuelo’
abuelo 250 200 Hz 150 Spanish 100 Maya 50 0 p1 p2 p3 p4 p5 p1 p2 p3 p4 p5 p1 p2 p3 p4 p5 Syl 1 Syl 2 Syl 3 85 Figure 4-7
Syllable Pitch in ‘palabra’
palabra 140 120 HZ 100 80 Spanish 60 Maya 40 20 0 p1 p2 p3 p4 p5 p1 p2 p3 p4 p5 p1 p2 p3 p4 p5 Syl 1 Syl 2 Syl 3 86 Figure 4-8
Syllable pitch in ‘gravilla’
gravilla 160 140 120 Hz 100 80 Spanish 60 Maya 40 20 0 p1 p2 p3 p4 p5 p1 p2 p3 p4 p5 p1 p2 p3 p4 p5 Syl 1 Syl 2 Syl 3 As is evident in these graphs, there appears to be a pattern that when words are
pronounced with Spanish prosody, the pitch rises on the stressed second syllable. This
finding for the words on the Spanish side of the syncretic continuum is in agreement with
work on Spanish prosody that argues that stress and accent are cued by a rise in pitch and
increase in duration on the stressed and accented syllable. In YM, there appears to be no
regular pitch pattern discernable from these three examples, and likely this is due to the
fact that these words fall at different positions in the phrase.
In Figure 4-5, the three syllables appear to be all roughly declining at similar rates, in
Figure 4-6 the pitch rises on the first syllable and falls on the second, in Figure 4-7 the
pitch in YM declines slightly on the first two syllables and rises slightly on the final,
while in Spanish the first syllable sharply falls, the second sharply rises, and the third
remains level. And finally in Figure 4-8 the initial pitch rises in both but more sharply in
87 YM, the second syllable rises sharply in Spanish but falls sharply in YM, and the final
syllable falls in both. There is no consistent rise or fall of pitch on a particular syllable in
any of the words spoken in the YM accent. From this initial investigation, pitch appears
to be an unclear representative of stress in Spanish loan words into YM.
4.7
Length in YM Syncretic Shifts
Since in this preliminary investigation, pitch does not appear to be a clear marker
of stress shift into the YM syncretic continuum, the next logical assumption is that
duration may be a more robust cue. In fact, it might be a more logical predictor than
pitch, since YM has been demonstrated to be phonemically tonal, while length always cooccurs with either High, Low or Glottalized vowels. Table 4-2 shows the measurements
segment duration for ‘trabajo’, ‘abuelo’, ‘palabra’ and ‘gravilla’ in both Spanish and YM.
Table 4-2
Word
trabajo
abuelo
palabra
gravilla
Loan Word Syllable Length and Pitch Range Measurements
Language
Spanish
YM
Spanish
YM
Spanish
YM
Spanish
YM
S1
0.139
0.199
0.171
0.233
0.056
0.100
0.098
0.163
Length in seconds
S2
0.180
0.124
0.295
0.137
0.144
0.060
0.135
0.074
S3
0.194
0.140
0.242
0.162
0.070
0.092
0.050
0.116
This data conforms to a more regular pattern than the pitch data described above. In the
Spanish cases, each conforms to the general rule that S2 is longer than S1, while in YM,
the opposite is the case. The graph in Figure 4-9 represent the length data visually.
88 Figure 4-9
Visual Representation of Length Measurements
0.3"
0.25"
0.2"
0.15"
Syllable"1"
0.1"
Syllable"2"
0.05"
Syllable"3"
Sp
ay
a"
M
an
is
h"
/"
tr
a .b
a .j
o"
/
"
tr
Sp
a
an
.b
a .j
is
h"
o"
/"a
.
bu
M
e.l
ay
o"
a"
/"a
Sp
.
an
bu
is
e.l
h"
o"
/"p
a
M
.l a
ay
.b
ra
a"
/"p
"
Sp
a
.l a
an
.b
is
ra
h"
/"g
"
ra
M
.
vi
ay
.y
a"
a"
/"g
ra
.v
i.y
a"
0"
The graphs in 4-10 through 4-13 show the same data for each individual word.
Figure 4-10
Syllable Length –‘trabajo’
0.25&
0.2&
0.15&
0.194309&
0.180492&
0.139039&
0.199454&
0.13973&
0.123954&
Syllable&1&
Syllable&2&
0.1&
Syllable&3&
0.05&
0&
Spanish&
Yucatec&
89 Figure 4-11
Syllable Length – ‘abuelo’
0.35'
0.3'
0.294948'
0.241789'
0.25'
0.2'
0.233391'
0.171481'
0.161768'
0.137071'
0.15'
Syllable'1'
Syllable'2'
Syllable'3'
0.1'
0.05'
0'
Spanish'
Figure 4-12
0.16%
Yucatec'
Syllable Length – ‘palabra’28
0.144439%
0.14%
0.12%
0.100279%
0.1%
Syllable%1%
0.08%
0.06%
0.091859%
0.07008%
0.056%
0.060474%
Syllable%2%
Syllable%3%
0.04%
0.02%
0%
Spanish%
Yucatec%
28
The speaker who uttered the word ‘palabra’ on both sides of the syncretic continuum
did this in repeated instances in the data, in most cases with the YM morpheme ‘-oʔ’
attached to the end. The instance of ‘palabra’ used for these measurements was the only
one in which this morpheme was not attached, and was chosen so that the analyses would
be unobscured by the addition of another YM morpheme.
90 Figure 4-13
Syllable Length – ‘gravilla’
0.18&
0.163258&
0.16&
0.14&
0.12&
0.1&
0.134935&
0.11595&
0.097998&
Syllable&1&
0.074208&
0.08&
Syllable&3&
0.05&
0.06&
Syllable&2&
0.04&
0.02&
0&
Spanish&
Yucatec&
It is evident from the data on syllable length that when moving from Spanish
pronunciation to YM pronunciation, the length ratio between the first and second
syllables changes distinctly. This is regardless of what position in the phrase it falls in,
unlike the data on pitch described above. If length was a cue for lexical stress, and if
stress is placed in YM according to Krämer’s predicted foot structure described in the
previous chapter and shown in Figure 3-10, we would expect the first and last syllable in
these words to be demonstrably and consistently longer than the second, and this
prediction is borne out in these examples.
4.8
Testing the Predictions
Of the three predictions in Figure 4-4 for how this data would surface after
shifting to the YM side of the syncretic continuum, when incorporated into the YM
prosodic system, both 4-4a (initial syllables will be stressed) and 4-4c (final syllables will
be stressed) could potentially be validated in this data set. However, prediction 4-4b, that
the final syllable would undergo final glottal epenthesis, has not yet been looked at.
91 According to both the attested prosodic rules of YM and the predicted foot structure seen
in Figure 3-10, all of these YM words should be realized with final glottalization in order
for the final syllable to be counted as heavy. The data, however, does not appear to show
any signs of final glottalization in any of the three instances described thus far.
Final glottalization can be seen in the spectrogram as an abrupt stop, or more
commonly in YM as a series of glottal pulses known as creaky voice. An example of the
YM word ‘wayeʔ - here’, which has a final glottal stop, can be seen in Figure 4-14 below.
Figure 4-14
Spectrogram of YM wayeʔ29
If the Spanish loan words in question undergo final glottal epenthesis, we would expect
to see similar evidence of creaky voice or closure at the end of the words, but as you can
see in Figures 4-15 through 4-18, this is not evident.
29
The upper line in this and the following spectrogram denotes intensity and the lower
line denotes fundamental frequency.
92 Figure 4-15
Spectrogram of YM palabra
Figure 4-16
Spectrogram of YM abuelo
93 Figure 4-17
Spectrogram of YM trabajo
Figure 4-18
Spectrogram of YM gravilla
These spectrograms show no signs at all of glottalization in final position akin to that
found in the native YM word ‘wayeʔ’ from Figure 4-14, although ‘abuelo’ in Figure 4-16
does show evidence of an initial glottal stop before the first syllable. The glottal pulses in
initial position in ‘abuelo’ show that for this speaker the YM rule for glottal epenthesis
94 appears to be valid in initial position, but not word finally. It is possible that the final
glottal epenthesis is occurring phrase-finally rather than word finally, and is not apparent
in these spectrograms because none of these words were spoken in a phrase-final context.
However, the duration data suggests that the final syllable is undergoing lengthening, and
indeed the ratio of syllable duration flips from the Spanish Short-Long-Short to the YM
Long-Short-Long.
4.9
Summary of Spanish Loan Words in YM
The data from this case study gives evidence for some kind of prosodic change
occurring on Spanish loan words when they are incorporated in the YM accent, in which
the first and third syllables lengthen, while the second syllable shortens. If this
phenomenon was found to persist in native YM words as well, this might give further
evidence that duration is the salient cue for stress in YM, since the language seems to
have an obvious system in place that functions to adapt Spanish words into the YM
prosodic system, and this system adds length to the segments that were predicted to be
stressed.
However, it is also possible that this initial and final lengthening is a process
specific to loan words, and is not generalizable to the rest of the language. Another
possibility is that there are positional length restrictions in YM that are unrelated to stress
or accent, which the Spanish words are modified to satisfy. The following chapter
investigates the data collected from the production section of the study, and gives
evidence that this phenomenon may in fact not be indicative of the underlying stress
patterns in YM, but instead have an outside explanation.
95 CHAPTER 5 STRESS IN YUCATEC MAYA
5.1
Diagnosing Stress in YM
Diagnosing the stress pattern of a language is a complex and multifaceted process.
Due to the parasitic nature of stress, i.e. that it usually utilizes phonetic cues that are
already in use by the phonological system, one cannot simply read the stress patterns
from a sample of speech in an automated fashion. Hayes (1995) describes various ways
in which stress can be determined, including measuring segments for duration or pitch
fluctuations that are not due to phonemic structures, as well as discovering phonetic
alternations that appear to occur only in metrically prominent positions (or in those
preceding or following metrically prominent positions). For example, in English, the
phenomenon of aspirating voiceless stops only before stressed positions can give clues to
where those stressed positions are. However there is also some debate about whether or
not prominent or stressed positions will reliably and consistently appear in the phonetic
production of speech that does not have phonemic stress. (Hyde, 2007) notes that for
languages that have mora sensitivity, i.e. in which a distinction is made between syllables
with codas or long vowels, and those without, extra moras in a syllable are often thought
to equate to extra duration in a syllable. However Hyde is careful to describe moras, as
well as stress itself, as abstract phonological entities that may or may not be represented
phonetically in a consistent way. This chapter statistically analyzes both native speaker
intuitions as well as phonetic data from the speech of native speakers to discern what
acoustic cues might be present that point to metrically prominent positions.
As described in the previous chapters, existing work done on YM prosody has
noted what positions may be the most prominent, i.e. that if stress exists, High and Low
96 tones and glottalized vowels are always stressed, and that YM may be quantity sensitive,
but these observations leave many questions unanswered. Do native speakers have
intuitions about where stress lies? Do those intuitions correlate to positions with phonetic
cues for stress? If there is a prominence hierarchy in YM words, what acoustic cues
signal those prominent positions? Assuming that tones and glottalized vowels do always
bear stress, what happens when a single word contains two of these vowels, or when a
word contains none? Can some words in YM have no prominent position, because they
contain no tone or glottalized vowel, while others contain two or more adjacent
prominent positions (going against the OCP and other cross linguistic tendencies towards
spacing of prominence)? If both initial position and heavy syllables attract stress in YM,
as some previous scholars have claimed, are they both stressed or does one have priority
over the other? This chapter will investigate whether or not evidence can be found in this
data to answer these questions.
5.2
Hypotheses
There are a variety of hypotheses that can be posited on what rules govern stress
in YM, based both on the observations of previous scholars, as well as on the data from
Spanish loan words incorporated into YM from the previous chapter. First, because the
behavior of fundamental frequency is what cues both tone and glottalization, both of
which have been thoroughly documented in this language, while duration differences can
all be tied to the presence or absence of tone and glottalization, I posit that the cue for
stress is more likely to be duration than pitch.
Figure 5-1
97 Hypothesis 1: Duration cues Prominence
Prominent positions are cued by relatively longer duration (rather than
consistent changes in pitch) on the vowel of the prominent syllable,
however the increase in duration due to prominence will not obscure the
phonemic boundary between short and long vowels.
Second, the most universal observation in the YM prosodic literature is that syllables that
have a tone or glottalized vowel are prominent, while short non-tonal vowels are not.
This would lead to the hypothesis in Figure 5-2.
Figure 5-2
Hypothesis 2: Non-Neutral vowels are prominent
Glottalized, High and Low vowels are prominent.
As described in the previous chapter, Krämer argued that heavy syllables in YM are also
prominent, as are initial syllables regardless of weight, which would give us the
following 2 hypotheses.
Figure 5-3
Hypothesis 3: Weight is prominent
When no Glottalized, High or Low tones are present in a word, heavy
syllables are prominent.
Figure 5-4
Hypothesis 4: Initial syllables are prominent
Words have initial syllable prominence by default.
The current chapter investigates the data collected for this study to determine whether it
can provide evidence in support of or against these hypotheses. Building upon the
evidence presented in the previous chapter that found cases of Spanish loan words in
which the duration of certain syllables underwent changes when incorporated into the
YM prosodic system, the following section focuses on the native YM words, and starts
by looking at the results of the native speaker intuition section of the study.
5.3
98 Native Speaker Intuitions
A crucial aspect in determining the presence or absence of consistent default
stress patterns in a language is in understanding the intuitions of the native speakers
themselves. If the YM language has rules such as those posited by Krämer that long
vowels (tones and glottalized vowels) are always stressed, and that heavy syllables and
initial syllables also attract stress, native speaker intuitions may show evidence of these
tendencies. For languages that have robust metrical stress, native speakers even when not
linguistically trained, can often readily identify the stressed positions in words when
asked to identify the strongest syllable, or to tap out the rhythm of a word.
This leads to the introduction of an additional hypothesis, seen in the following
figure.
Figure 5-5
Hypothesis 5: Phonetic data will accord with speaker intuitions
The results of the phonetic data will match with the native speaker’s
intuitions of which syllables are prominent.
To test Hypothesis 5, the current section will statistically analyze the native
speaker intuition data, which will later be compared to phonetic data. Four of the YM
speakers recorded for this study were asked to complete a task in which they gave their
own intuitions about which syllable in a given word was the most prominent, or
strongest. Because all the speakers were also fluent in Spanish which has a robust stress
system, they were first trained on Spanish words such as ‘médico’, ‘abuélo’ and
‘Yucatán’, which each represented a different stressed syllable. They were then told to
determine which syllables were strong or weak, for mono, bi and trisyllabic words, and
99 were encouraged to use the tapping method of finding the word’s rhythm30. They were
instructed to then underline each syllable in a list of words (to determine how many
syllables they intuited were present), and then mark strong syllables with an X. Because
of the possibility that stress is assigned in a quantity sensitive way, they were also given
the instruction that if no syllable seemed more prominent, that they did not need to mark
any syllable, or if both seemed equally strong, that both could be marked with an X.
Some of the speakers preferred to pronounce the word in question differently than
it was written, often by omitting a final morpheme, and there were also a few
inconsistencies with how the words were syllabified, mainly when the word contained a
glottalized vowel (speaker 4 often marked these as two separate syllables, while the rest
of the speakers consistently marked them as 1). There were also some cases in which a
speaker simply did not have an opinion about which syllable was stronger, and therefore
did not mark it at all, which is represented by a question mark. An S denotes that the
speaker recorded a ‘strong’ syllable, and a W denotes those that lacked prominence.
There was a high degree of variation in the way these were judged, and all the speakers
expressed sentiments during this portion of the study that these intuitions were ‘only their
opinion’, and implied that others may well disagree or view the strong positions
differently.
The data in Figure 5-6 shows the full set of native speaker intuitions that were
gathered for this study. This section will analyze the intuition data to analyze what
patterns are found that support the hypotheses in Figure 5-1 through Figure 5-5.
30
This method did not prove to be intuitive for the participants, and they often declined
to tap in favor of repeating the syllables of the words to themselves.
100 Table 5-1
Native Speaker Prominence Intuitions
Word
luu.bul31
lúu.bul
k’áak.náab
táan.chúu.muk
yáax.k’iin
úuch.ben
xch’úu.pal
oo.chel
tee.p’el
uu.sáan
xtuux.kuuts
ba.jux
ja.ts’uts
jun.p’íit
k’a.jóol
chúu.ka’an
ba’al.che’
je.la’an34
ch’a.la’at
búu.ki.naj
bu.láa.k’ab
ju.jun.wáal
ta.k’an.kuuns
maa.kal.máak
S1
sw
sw
sws
sww
ss
sw
sw
ww
sw
ww
ww
ww
sw
ww
sw
sw
sw
ws
ss
sww
wsw
www
ssw
sww
S2
?32
?
?
sss
ss
sw
sw
?
ww
ws
ww
ww
ww
ws
ww
ww
ww
ws
ww
www
wsw
wws
wws
www
S3
S4
sw
sw
sw
sww
ss
sw
sw
sw
s33
ss
sw
sw
sw
ws
sw
ws
sw
ws
ws
wws
wss
wsw
?
wsw
sw
sw
ss
sww
sw
sw
sw
ww
ww
ww
ww
ww
ww
ws
ws
sw
ww
www
ww
sww
wsw
wws
wws
wws
máa.se.wáal
kan.ka.la’as
sww
sww
wws
www
sww
?
?
?
sww
www
w
www
ji’i.ji.nak
ssw
Word
xíim.bal
yóo.k’ol
xmáa.kech
púu.ts’ul
yée.tel
kaam.bal
kii.mil
seen.kech
chaam.béel
ko.sam
a.wat
chi.chan
k’u.xub
mix.máak
ich.kíil
sée.ba’an
chi’ich.nak
mi.na’an
jóo.ke.saj
áa.k’ab.tal
ja.kúun.taj
si.t’rii.yo
sáa.bu.kaan
ko.ja’a.nil
S1
S2
S3
S4
sw
sw
ws
sw
sw
sw
ww
ww
ww
ww
sw
ww
sw
ww
ww
sw
sw
ws
sww
sw
sww
sww
sww
ssw
sw
ww
ww
ww
ww
ww
sw
ww
ww
ss
ww
ww
ww
ws
ws
ww
sw
ws
ww
?
?
www
www
www
sw
sw
ss
sw
sw
sw
sw
sw
sw
ws
ss
sw
sw
ws
ws
ws
sw
ws
sw
sww
?
?
sww
wsw
báa.lam
si’is
sw
s
se
w
sw
s
sw
sw
sw
sw
sw
ww
sw
ww
ws
ww
ww
ww
ww
ws
ws
ws
www
www
sww
sww
wsw
www
sww
www
w
sw
ww
ba’al
s
w
s
ww
The only instances of monosyllabic words that the speakers were presented with
for this section of the study included the words ‘ba’al - thing’ and ‘si’is - cold’.
Monosyllables are less meaningful when looking for a relative distinction in prominence,
since there is only one unit available, and therefore there is no comparison to make,
31
The syllables are separated by a ‘.’, these boundaries were consistent for all speakers,
who marked the syllabification by underlining each syllable.
32
Question mark represents that a speaker declined to make a judgment.
33
Speaker pronounced this as monosyllabic /teep’/.
34
Speaker 4 often judged glottalized vowels as representing 2 separate syllables, unlike
the other 3 speakers.
101 however these two words were introduced in order to determine whether or not the
speakers treat glottalized vowels as 1 or 2 syllables. If these vowels truly contain a full
glottal stop, it would be probable that they would be parsed as two separate syllables. If,
as the data from Frazier (2009) argues, glottalized vowels are better characterized as
displaying creaky voice rather than a full glottal closure, they may be more often
analyzed as a single syllable only. As the chart above shows, speakers 1, 2 and 3 parsed
these glottalized vowels as consisting of a single syllable rather than 2, while speaker 4
was the only one to parse them as separate and equally prominent syllables during this
portion of the study. The result from the other 3 speakers is consistent with Frazier
(2009), which defines the glottalized vowels as a single vowel shape. Overall, speakers
tend to treat these vowels as single syllabic entities.
In order to determine whether any trends in this native speaker intuition data for
bisyllabic and trisyllabic words are significant, the data was coded and statistical analyses
were performed, which will be described in the following section.
5.3.1
Statistical analysis of Intuition Data
A statistical analysis of this data can help to test the hypotheses laid out in
section 5.2 above. Because this native speaker intuition data is categorical in nature rather
than continuous, i.e. judgments are Strong or Not-Strong, as opposed to a numerical
measurement, the data was transformed into a continuous variable in order to perform
statistical analyses. For each condition, the judgments were aggregated by taking the
number of judgments of Strong for that syllable type, divided by the total number of
judgments. This results in a continuous numerical variable between 0 and 1, for the
likelihood of a particular type of syllable to be deemed Strong.
102 In testing hypotheses 2 (non-neutral vowels are prominent), 3 (heavy syllables are
prominent and 4 (initial syllables are prominent), there are 3 possible variables that are
relevant to analyze: Length (Long vs. Short), Position (Initial vs. Medial vs. Final), and
Weight (Heavy vs. Light). The variables for Length and Position are truly independent
variables, meaning that both Long and Short can be found in all three positions, and the
presence of one does not condition the presence of another.
The variable for Weight, however, is slightly different, in that it is affected by
both Length and Position. Weight is affected by Length because Long vowels are very
likely to contribute to weight, and therefore Long would necessarily equal Heavy.
Similarly, due to the Initial-C, Final-C constraint discussed in the previous chapters,
syllables in final position must have a coda, and therefore Final would also necessarily be
Heavy. In fact, the instances of Heavy syllables that are not either long or in final position
are rare in the data. Because the variable weight is intertwined and often predictable by
the length and position of the syllable, it was not coded as a separate value.
Each syllable of the intuition data was coded for whether it contained a Long or
Short vowel, and whether it fell in word Initial, Medial or Final position35. Because both
bisyllabic and trisyllabic words are found in this data, syllables were coded as medial
only if they fell in the middle of a trisyllabic word. The following chart shows how words
were coded for both the variables of Length and Position.
35
Syllables were coded as Initial, Medial and Final rather than position 1, 2 or 3, because
the latter method would conflate final syllables in bisyllabic words with medial syllables
in trisyllabic words. Since Final position results in obligatorily heavy syllables, it is
important to maintain that distinction.
103 Figure 5-6
Variable Coding
Syllable
word: /luubul/
1.
luu
bul
word: /jun.p’íit/
2.
jun
p’íit/
word: /saabukaan/
3.
saa
bu
kaan
Length
Position
Long
Short
Initial
Final
Short
Long
Initial
Final
Long
Short
Long
Initial
Medial
Final
After each syllable in each word was coded for the independent variables Position and
Length, the value of the dependent variable was calculated for each syllable using the
following formula.
Figure 5-7
Formula for Dependent Variable
For each item/subject within each combination of Position and Length:
Number of Strong Judgments for this Item/Subject
Total number of Judgments for this Item/Subject
This calculation resulted in a number between 0 and 1 for each item or subject and each
combination of Length and Position.
The data from the native speaker judgments from the bisyllabic and trisyllabic
was highly variable in nature, though there was a basic trend towards judging Long
syllables as Strong. The following chart shows the breakdown of strong judgments by
Position and Length alone.
104 Table 5-2
Number of Strong Judgments by Speaker
a. By Position36
S1
26/40 = 65%
3/13 = 23%
6/36 = 16%
Initial
Medial
Final
S2
8/36 = 22%
2/11 = 18%
10/33 = 30%
S3
25/36 = 68%
4/9 = 44%
11/32 = 34%
S4
16/37 = 43%
2/12 = 16%
7/32 = 22%
S2
16/33 = 48%
4/48 = 8%
S3
28/33 = 84%
12/44 = 27%
S4
25/35 = 71%
0/47 = 0%
b. By Vowel Type
S1
24/38 = 63%
11/51 = 22%
Long
Short
Table 5-3
Initial - Long
Medial - Long
Final - Long
Initial – Short
Medial – Short
Final - Short
Number of Strong Judgments for Position and Vowel Type
S1
18/23 = 78%
1/4 = 25%
5/11 = 45%
8/17 = 47%
2/9 = 22%
1/25 = 4%
S2
7/19 = 37%
1/3 = 33%
8/11 = 72%
1/17 = 6%
1/9 = 11%
2/22 = 9%
S3
20/21 = 95%
2/2 = 100%
6/9 = 67%
5/14 = 36%
2/8 = 25%
5/23 = 22%
S4
16/22 = 72%
2/3 = 67%
7/8 = 88%
0/14 = 0%
0/9 = 0%
0/24 = 0%
Table 5-3 shows the differences in judgment behavior between speakers. Speaker
1 is most likely to judge a syllable strong if it is Initial and Long, while speaker two is
much more likely to judge Final Long syllables as strong. Speaker 3 judged 100% of
Medial Long syllables as strong, though there were very few instances which makes this
less concrete, and is the most likely to judge Initial Long syllables as strong. Speaker 4
patterns with speaker 2, in that they are both most likely to judge Final Long syllables as
strong, though is also almost as likely to find the same for Initial and Medial.
36
The denominator is different for each speaker because cases in which that speaker
declined to make a judgment, or the judgment was anomalous (i.e. was a word with more
or less than the default number of syllables), were not included.
105 Looking at the effects of Position and Length on their own, position is
inconsistent with speaker 2 as the only one who judged Final strong more often than
Initial. The more robust tendency seen in this chart is that Short syllables are less likely
to be judged strong for all speakers, with speaker 4 in particular judged no single short
syllable as strong. Though looking simply at percentages does not tell us whether or not
the patterns described above are statistically significant, this distribution of judgments
would tell us that if statistical tests are performed, we would expect that Long to be more
prominent than Short, and perhaps Initial would be more prominent than Final, if any
significant results were found. However the high degree of variation between speakers
makes a significant finding more unlikely to achieve.
To determine whether these patterns are in fact significant, a 2-factor ANOVA
was performed on this data, in order to determine whether or not the factors of Vowel
Type, Position, or the interaction of the two had a significant effect on whether or not the
speaker judged the syllable as Strong. This 2 factor ANOVA was performed both by
subjects and by items, which is the standard practice for psycholinguistic research, and is
meant to ensure that the significant effect is homogeneous both across the speakers and
across the items themselves. Typically, a significant value for both tests is required to
claim that the overall effect is significant.
The results of the 2 factor ANOVA had no significant interaction effect between
Position and Length, by subjects F(2,6) = 1.82, p = .241, by items F(2,83) = .558, p =
.574. By subjects, the main effects of Length and Position were also not found to be
significant, ps > .665, though they were by items, Position: F(2,83) = 15.31, p < .001,
Length: F(1,83) = 78.20, p < .001.
106 The lack of a consistent significant result for this analysis is not altogether
surprising given how variable the data was between the 4 speakers. The charts that show
the breakdown of strong judgments by speaker do show some tendencies by speaker that
are interesting to this discussion, particularly that Long syllables seem much more likely
to be strong than Short syllables. This variability suggests that stress may not be a
consciously salient phenomenon to speakers of YM, particularly as a separate entity to
the phonemic concepts of tone, length and glottalization.
5.3.2
Summary of Intuition Data
In the beginning of this chapter, I introduced 5 hypotheses about stress in YM,
and then presented data from native speaker intuitions to test these hypotheses. After
establishing that YM speakers most likely parse glottalized vowels as a single syllable
nucleus rather than as two separate ones, data was presented from bisyllabic and
trisyllabic words. The following figure shows the hypotheses presented in figure 1, along
with the evidence for or against them found in the native speaker intuition data.
Figure 5-8
Yucatec Maya Prominence Rules
Hypothesis 1: Duration cues Prominence
This will be discussed in the following section.
Hypothesis 2: Non-Neutral vowels are prominent
Though there was a tendency for speakers to judge Long vowels stronger
than Short vowels, the effect was not found to be significant.
Hypothesis 3: Weight is prominent
If Final syllables and/or Long syllables had been found to be significantly
more likely to be judged strong than Initial and/or Short, this would have
been supporting evidence for hypothesis 3. Neither the effects of Length
nor Position were significant, and the pattern in the data leaned towards
the Initial rather than the Final for 3 of the 4 speakers.
107 Hypothesis 4: Initial syllables are prominent
Three out of four speakers were more likely to judge the Initial syllable as
stronger than the Final, however this effect was not significant.
Hypothesis 5: Phonetic data will accord with speaker intuitions
Statistical analyses for native speaker intuition were not significant by
either Length, Position, or the interaction of the two, indicating that
speakers may not perceive a robust and consistent system of prominence.
If phonetic data is found in the following sections has a significant effect
of Position or Length, that would be counter-evidence for hypothesis 5. If
no significant effects are found, this would suggest that stress may not be a
robust phenomenon in the language, and that native speaker intuitions
and phonetic data would be in accord.
In the next section, data from the production facet of the study will be presented in order
to determine whether or not there is evidence within the acoustic signal to support any of
these hypotheses.
5.4
Phonetic Data from Words in Isolation and Context
As a starting point to investigate the full range of prosodic prominence in YM, as
well as to test hypothesis 1, that duration is a cue for prominence, the following section
discusses prosody at the lexical level. This investigation will begin by outlining the pitch
and duration patterns found in monosyllabic, bisyllabic and trisyllabic words of YM
origin, in order to discern if there are regularities not due to the presence of phonemic
pitch or vowel length that may be signals of the surface metrical structure.
The production study of native speakers of YM included the elicitation of
monosyllabic, bisyllabic and trisyllabic words in isolation, as well as in phrases. Though
so far the evidence we have seen from Spanish loan words points to length being a more
salient cue of prominent positions than pitch, both the acoustic measurements of duration
and pitch will be analyzed. For each syllable in the native YM words that were recorded,
108 the vowel nucleus’s duration and 5 points along the vowel’s pitch contours were
recorded, excluding onsets and codas of those syllables.
As discussed in the previous chapter, Frazier (2009) notes a high level of
idiolectal variation amongst speakers, which is also present in the data collected for this
study. However, two different overarching patterns emerged from Frazier’s data
indicating that the Eastern and Western sides of the Yucatán Peninsula demonstrated
enough distinctions to be separate dialects of Yucatec. Vowel length and tonal distinction
played an important role in determining this dialectal split. All of the speakers recorded
for this study are from areas that fall under the Western dialect of Yucatec Maya, which
centers around Mérida and Santa Elena, in the state of Yucatán, Mexico. Frazier’s study
looked primarily at production and perception data from words of the CVC type (i.e.
heavy, monosyllabic) in a variety of phrasal positions, a particular set of patterns
common to the speakers from this dialect was found, and is summarized below.
Figure 5-9
Characteristics of the Western Dialect of YM (Frazier, 2009: p 274)
a. Long vowels (High tone, Low tone and Glottalized) are about twice as long as
Short (neutral) vowels
b. Low tone vowels are produced with fairly steady low pitch
c. Short tone vowels have steady mid or mid-high pitch
d. Glottalized and High vowels have initial high pitch and a falling contour
e. High vowels in non-phrase final position only High vowels reverse to a rising
contour
f. Glottalized vowels do not contain a medial glottal stop as described in much of
the literature, but instead a more accurate description is that they start with High
pitch and end in creaky voice37
In the following section I will outline the findings from my data, collected from 5
speakers originally from areas that fall within Frazier’s Western dialect, on CVC type
words, and compare them to the findings from Frazier when applicable. The data for this
37
Stress in YM is not analyzed in Frazier 2011.
109 study included both words in isolation as well as words in phrases, and both contexts will
be analyzed in the following sections.
5.4.1
Monosyllabic Words
We will begin by looking at monosyllabic words uttered in isolation. Though
there is no comparison to be made between syllable prominence in words with only one
syllable, establishing the baseline behavior of monosyllables is important in testing
whether or not hypothesis 1 is valid, and to ensure that the speech of the participants is in
line with the Western dialect of YM and has tonal distinctions.
5.4.1.1 Length
5.4.1.1.1 Length in Isolated Words
43 monosyllabic words uttered by each of 5 speakers were measured for vowel
length. This measurement was made using Praat Software, and onsets and codas were
excluded. The 43 words contained all 4 vowel shapes, including 4 Short/Neutral, 7
Glottalized, 16 Low and 16 High vowels38. The following chart shows the mean vowel
lengths for each speaker for each of the 4 vowel shapes, and the total mean for each
vowel type across speakers.
38
More words were originally recorded, but there were many variations in pronunciation
from speaker to speaker, and some of the words were unknown to some speakers. The
tokens with obvious irrelevant variation and those that were of unknown origin to some
speakers were not included. This set contains less Neutral vowels because this is a rare
case in monosyllabic YM words, the majority of which are High, Low or Glottalized.
110 Mean Vowel Length in Isolated Monosyllabic Words39
Table 5-4
Speaker
S1
S2
S3
S4
S5
Mean
StDev
Short
0.081
0.094
0.077
0.065
0.073
0.078
0.011
Low
0.203
0.175
0.140
0.138
0.147
0.161
0.028
High
0.204
0.191
0.169
0.166
0.159
0.178
0.019
Glottalized
0.281
0.213
0.198
0.219
0.160
0.214
0.044
This data can be better viewed graphically in the Figure 5-10.
Figure 5-10
Mean Vowel Length by Speaker
Mean Vowel Length by Speaker 0.3 Seconds 0.25 0.2 Neutral Low 0.15 High 0.1 Glottalized 0.05 0 S1 S2 S3 S4 S5 A one-way ANOVA was conducted to determine if there were any significant differences
in duration between Short, Low, High and Glottalized vowel nuclei, with subjects entered
as a random factor. The results provide strong evidence for group differences in length:
F(3, 12) = 33.3, p < .001 by subjects, F(3,45) = 35.52, p < .001 by items. This result,
however, does not provide information on the specific nature of the differences. In order
to shed light on these differences, six individual ANOVAs were run comparing G to H, G
39
Length measurements are in seconds.
111 to L, G to S, H to L, H to S and L to S, by subjects and by items. The following table
shows the results of these comparisons, asterisks indicate significant findings.
Table 5-5
Monosyllabic Duration ANOVA results
Comparison
G vs. H
G vs. L
G vs. S
H vs. L
H vs. S
L vs. S
Results
By Subjects
F(1,4) = 6.084, p = .5
F(1,4) = 13.79, p = .02*
F(1,4) = 50.89, p = .002 *
F(1,4) = 7.565, p = .05 *
F(1,4) = 266.8, p < .0001 *
F(1,4) = 53.33, p < .002 *
G, H >> L >> S
By Items
F(1,23) = 11.07, p < .003 *
F(1, 27) = 19.87, p < .0002 *
F(1,11) = 170.3, p < .0001 *
F(1,34) = 6.137, p < .02 *
F(1,22) = 101.4, p < .0001 *
F(1,26) = 37.91, p < .0001 *
G >> H >> L >> S
For both by-subjects and by-items, the trend is G,H >> L >> S. Neutral/Short vowels
were on average 42% as long as the mean of Glottalized+High+Low vowels, and despite
a somewhat high degree of inter-speaker variation, both by-subjects and by-items
ANOVAs show that the vowel shapes Glottalized, High and Low are significantly longer
than Short, which mirrors the characteristic of the Western YM dialect in Frazier’s
analysis described above. This finding backs up the claim that these YM speakers
distinguish between long and short vowels, a finding that has been documented for others
in their dialect area. The next section will look at the same words in a phrasal context.
5.4.1.1.2 Length in Phrasal data
Table 5-6 shows the means and standard deviations of length by vowel shape for
the same monosyllabic words uttered in a phrasal context.
Table 5-6
Vowel Shape
Short
Low
High
Glottalized
112 Mean Vowel Length in Monosyllabic Words in Phrases40
Mean
0.089
0.132
0.156
0.149
StDev
0.024
0.041
0.046
0.041
A one-way ANOVA was performed to determine if length differed significantly by vowel
shape by subjects and by items, and significant differences were obtained, F(3,11) =
9.357, p = .002 by subjects, and F(3,75) = 8.122, p < .001. In order to determine which
vowel shapes were significantly longer others, six one-way ANOVAs were performed
comparing each condition to each other condition. The results can be seen in Figure 5-7.
Table 5-7
Monosyllabic Duration ANOVA results for words in phrases
Comparison
G vs. H
G vs. L
G vs. S
H vs. L
H vs. S
L vs. S
Results
By Subjects
F(1,4) = .116, p = .751
F(1,4) = 1.844, p = .246
F(1,4) = 23.37, p = .0169*
F(1,4) = 12.48, p = .0242*
F(1,4) = 31.34, p = .0113*
F(1,4) = 36.57, p = .009*
H >> L and G, H, L >> S
By Items
F(1,37) = .473, p = .496
F(1,48) = 3.205, p = .0797
F(1,21) = 26.81, p < .001 *
F(1,57) = 7.634, p = .0077 *
F(1,30) = 18.45, p < .001 *
F(1,38) = 9.485, p = .00384 *
H >> L and G, H, L >> S
For words in a phrasal context, the significant distinction in duration between G/H/L and
S is maintained, and H is also significantly longer than L. Looking at the means from
Table 5-6, the Glottalized vowels are shorter than High when uttered in a phrasal context,
which may indicate that Glottalized vowels are more likely to undergo shortening phrase
medially. This data in conjunction with the data from words in isolation shows that the
duration facts for the speakers recorded for this study are consistent with those from the
Western dialect of YM, by maintaining a significant distinction in length between G/H/L
40
Length measurements are in seconds.
113 and Short vowels. The next section will investigate whether or not a tonal phonemic
distinction is also present for the speakers recorded for this study.
5.4.1.2 Pitch
The second phonetic cue we will be investigating is pitch, or fundamental
frequency (f0). Pitch has been shown to be a cue for stress in a variety of languages, and
in many a combination of pitch and duration are the primary cues for stress. However,
because YM is thought to have a tonal phonemic distinction between High and Low tone,
pitch seems to be an unlikely cue for stress in YM. Not only is pitch already being
utilized to cue tone, but it is also cuing intonational contours on the phrase as a whole.
For these reasons, as well as from the evidence found in Spanish loan word data
discussed in chapter 4, in hypothesis 1 I posited that duration will be a more salient cue
than pitch as a cue for prominent positions.
However, Frazier (2009) notes that speakers can fall into two dialects of YM, and
so far we have established that those recorded for this study come from the Western
dialect area geographically, and maintain a distinction between Long and Short vowels.
However one important aspect of the Western dialect is that a High and Low tonal
distinction is present, while in the Eastern dialect that distinction is unclear or not present.
It is important to establish that these speakers do in fact have a tonal distinction, or the
assumption which led to hypothesis 1 would be invalid. In this section, pitch in
monosyllabic words will be analyzed in order to determine the behavior of pitch for the
speakers of YM recorded for this study, and to establish that these YM speakers do in
fact have a tonal distinction.
114 Because pitch is a continuous element that has a changing trajectory over time, it
is a fundamentally different kind of measurement from duration. Because of this
distinction, I looked at pitch in two separate ways in order to elucidate the possible
distinctions. Pitch measurements were recorded at 5 distinct points along the pitch
contour, including at the beginning, one quarter, midpoint, three quarters and end of the
segment41. This method has been used by Frazier (2009)’s analysis of YM and (Xu,
1997)’s analysis of Mandarin, and may have the effect of slightly simplifying the
contour. The following section will look at how the mean data from these native YM
speakers compares to the analysis in Frazier (2009).
The first method used to analyze the pitch contrasts was to take only the midpoint
of the vowel, which should show the highest degree of separation, and compare that
between each type of vowel. This method cannot give information about any contour that
may be distinctive between vowel types, and so I also employed a second method:
looking at the difference between point 2 and point 4. Using this method allows us to
determine whether there is any significant difference in contour trajectory between the
vowel types. The following section analyzes the data from both words in isolation and
words in phrases using the midpoint of the vowel, and the subsequent section looks at
both sets of data for the contour differences.
5.4.1.2.1 Pitch Midpoint Analysis
5.4.1.2.1.1 Words in isolation
41
Start and end points of the vowels were taken from after or before any apparent
formant transitions from the previous consonant or the following consonant, to avoid
consonantal effects on the vowel.
115 Table 5-8 shows the mean pitch point values for the 4 different vowel shapes at
pitch point 3.
Table 5-8
Vowel Shape
Short
Low
High
Glottalized
Mean Pitch at P3 – Isolation
P3
145.35
128.60
145.62
134.11
Short and High are the highest at the midpoint, while Low is the lowest. In order to
determine whether or not these vowels differ significantly in pitch, a one-way ANOVA
was performed by subjects and by items on the pitch measurement from the mid-point of
the vowel. Results were significant both by-subjects; F(3,12) = 6.08, p = .009, and by
items F(3,79) = 6.75, p < .001. In order to determine the specific effects, as well as to
determine whether or not these speakers can be analyzed as having a tonal distinction
between High and Low, individual one-way ANOVAs were run between each pair of
conditions. The following chart shows the results of these ANOVAs.
Table 5-9
Monosyllabic Pitch ANOVA results – Words in Isolation
Comparison
G vs. H
G vs. L
G vs. S
H vs. L
H vs. S
L vs. S
Results
By Subjects
F(1,4) = .544, p = .514
F(1,4) = 2.70, p = .176
F(1,4) = 1.162, p = .342
F(1,4) = 11.40, p = .028 *
F(1,4) = .214, p = .668
F(1,4) = 7.74, p = .049 *
H >> L, S >> L
By Items
F(1,39) = 5.46, p = .025 *
F(1,46) = 1.43, p = .238
F(1,26) = 1.904, p = .179
F(1,54) = 20.95, p < .001 *
F(1,34) = .001, p = . 998
F(1,42) = 6.39, p = .015 *
H >> L, H >> G, S >> L
The distinction in pitch at the midpoint between H and L is significant by subjects and by
items, indicating that these YM speakers do have a distinction between High and Low
pitch at the midpoint of the vowel, which follows the patterns Frazier describes for their
116 dialect area. The following section outlines the length and pitch data at the midpoint for
YM words recorded in phrases, to determine whether or not there are any significant
differences from words in isolation.
5.4.1.2.1.2 Pitch Midpoint in Phrases
The means for pitch point 3 for words in a phrasal context are seen in the following chart.
Table 5-10
Vowel Shape
Short
Low
High
Glottalized
Mean Pitch at P3 – Phrasal Context
P3
133.13
133.19
141.27
149.10
High is still the highest in this data set, though the difference between High and Low is
smaller for words in phrases than it was for those in isolation. Just as was done above
with the data from words in isolation, a one-way ANOVA was performed to determine
whether the pitch at the midpoint of the vowel differed significantly, so we can see
whether the tonal distinction between High and Low remains consistent or is neutralized
in a phrasal context. The results of the ANOVA for words in a phrasal context was not
significant by-subjects; F(3,11) = 2.191, p = .147, or by-items; F(3,73) = 2.296, p =
.0848. Table 5-11 shows the results of the individual one-way ANOVAs performed on
each pair of vowel shapes.
117 Table 5-11
Monosyllabic Pitch ANOVA results – Phrasal Context
Comparison
G vs. H
G vs. L
G vs. S
H vs. L
H vs. S
L vs. S
Results
By Subjects
F(1,4) = 1.126, p = .348
F(1,4) = 3.373, p = .14
F(1,3) = 17.89, p = .0242 *
F(1,4) = 12.55, p = .024 *
F(1,3) = .106, p = .766
F(1,3) = .585, p = .5
G >> S, H >>L
By Items
F(1,36) = 1.278, p = .266
F(1,46) = 7.806, p = .00757 *
F(1,20) = 1.481, p = .238
F(1,56) = 2.693, p = .106
F(1,30) = .43, p = .517
F(1,37) = .404, p = .529
G >> L
None of these pairwise comparisons show consistent significant results for both bysubjects and by-items. The results from the analysis of monosyllabic words in phrasal
contexts show that the pitch distinctions between vowel shapes may get neutralized in a
phrasal context, since there was no main effect, nor where there any consistent individual
effects of pitch at the midpoint in the phrasal context.
5.4.1.2.2 Contour Analysis
High and Low vowels in YM have been described not only as differing in pitch
height, but also differing in contour over the course of the vowel. Low vowels are
described as level in pitch, while High vowels rise word and phrase initially and fall word
and phrase finally. Table 5-12 shows the average pitch at each pitch point for each
speaker, and Figure 5-11 shows this data represented graphically.
Table 5-12
Vowel Shape
Short
Low
High
Glottalized
Mean pitch for words in isolation
P1
P2
143.37 144.46
133.99 132.83
147.53 149.39
156.52 150.32
P3
145.35
128.84
145.36
134.11
P4
P5
142.25
139.85
124.93
122.03
137.13
128.30
130.19
122.45
118 Figure 5-11
Mean Pitch Trajectory
Mean Pitch: All Speakers 190 170 150 Hz Short 130 Low High 110 Glottalized 90 70 50 P1 P2 P3 P4 P5 The mean pitch for all speakers shows that each type of vowel nucleus has a
distinct pitch trajectory, and that they correspond to Frazier’s description of the Western
Dialect of YM. Glottalized vowels fall in pitch starting around pitch point 3, often
moving into creaky voice toward the end of the vowel. This is consistent with Frazier’s
work on glottalization, which describes this long vowel type as a High-toned vowel
followed by a creaky voiced vowel. Creakiness is known to sharply alter the pitch of a
vowel, and often causes no fundamental frequency to be recordable. As Frazier (2009)
argues, creakiness towards the end of a Glottalized vowel is its most salient cue, which is
consistent with the findings from this study.
119 Short vowels appear to be relatively stable in pitch from beginning to end, which
is also consistent with the Western dialect, and as was determined in the previous section,
are significantly shorter in duration than the other two vowel shapes.
In order to determine whether or not the data collected for this study is consistent
with these descriptions, statistical analyses were done for the difference between pitch
point 4 and 2, both for words in isolation and in phrases. This section discusses the results
of these analyses.
5.4.1.2.2.1 Words in Isolation
As in the previous section, statistical analyses were done on whether or not there
is a significant difference in pitch level between Glottalized, High, Low and Short
vowels, however in order to measure specifically the pitch contour, a value was
calculated to represent the trajectory of the pitch along the course of the vowel. This
value was calculated by taking the value of pitch point 4 and subtracting it from pitch
point 2. If the two values were close to the same, the value would be close to 0, and that
would represent a relatively level contour, if the number was negative that represents a
falling contour, and positive indicates a rising contour.
A one-way ANOVA was performed using the value of pitch point 4 minus pitch
point 2, and the results were significant, F(3,12) = 6.11, p = .009 by subjects, F(3,78) =
25.39, p < .001 by items. In order to determine the nature of the specific interactions,
individual one-way ANOVAs were done comparing each set of values. Table 5-13 shows
the results of these analyses.
Table 5-13
Comparison
G vs. H
G vs. L
G vs. S
H vs. L
H vs. S
L vs. S
Results
120 Pitch Contour - Words in Isolation
By Subjects
F(1,4) = 6.31, p = .066
F(1,4) = 6.78, p = .060
F(1,4) = 7.57, p = .051
F(1,4) = 5.45, p = .080
F(1,4) = 5.42, p = .080
F(1,4) = .04, p = .848
N/A
By Items
F(1,38) = 29.82, p < .001 *
F(1,45) = 45.54, p < .001 *
F(1,26) = 19.90, p < .001 *
F(1,53) = 9.04, p = .004 *
F(1,33) = 9.04, p = .005 *
F(1,41) = .004, p = .950
G >> H >> L,S
The results of the contour analysis for monosyllabic words in isolation was not
significant for both by subjects and by items for any of the pairwise comparisons.
Though the by items analysis was significant for all comparisons other than L vs. S,
overall there was too much variation amongst subjects for a significant result.
5.4.1.2.2.2 Pitch Contour in Phrasal Data
The same data was looked at for words uttered in phrases in order to determine
whether or not the pitch distinction between High and Low was distinguishable by
contour, and if so whether or not it is maintained when uttered phrase-medially.
The mean values for each of the five pitch points are shown in Table 5-14 and Figure 512.
Table 5-14
Vowel Shape
Short
Low
High
Glottalized
Pitch Means for Words in Phrases
P1
131.90
137.73
139.60
156.65
P2
133.49
135.68
140.37
152.28
P3
133.13
133.19
141.27
149.10
P4
129.77
131.31
139.85
142.37
P5
124.76
131.71
136.71
138.32
121 Figure 5-12
Pitch Means Graphically
In Context: Pitch by Vowel Shape 190 170 150 Hz Short 130 Low High 110 Glottalized 90 70 50 P1 P2 P3 P4 P5 A one-way ANOVA was performed to determine if these values differed significantly by
contour, and the results were not significant, F(3,11) = 2.39, p = .124 by subjects, F(3,74)
= 3.30, p = .025 by items.
5.4.1.3 Summary
Monosyllabic words produced for this study showed significant differences
between duration for all 4 types of vowels, and is in line with the analysis of the Western
dialect of YM in which the length of short neutral vowels was demonstrably shorter than
the length of toned and glottalized vowels, with the following pattern present G, H, L >>
S for words in isolation. For words spoken in phrases, these distinctions appear to have
been somewhat neutralized, which is different from the findings of Frazier (2009). This
difference likely indicates that the variability of the speakers for this study was too high
122 for the significance level to be maintained in a phrasal context, where intonational factors
are more likely to play an important role.
Production of pitch on one syllable words by these 5 speakers is also in line with
Frazier (2009)’s Western dialect in which a tonal distinction was evident between High
and Low, with words spoken in context displaying some slight neutralization of the tonal
distinction. Glottalized vowels, in addition to being the longest of the vowel shapes in
isolated words, may be more likely to shorten when in phrasal contexts. They also
display a tendency towards high pitch initially followed by sharply falling pitch, which
backs up the claim that glottalized vowels can be argued to be phonologically [v́ ̰v], or a
high vowel followed by a creaky vowel. Short vowels are pronounced with a steady pitch
that does not fall towards the end of the vowel as the other vowel shapes do.
This data from monosyllabic words, though it does not give us any evidence in
support of the posited hypotheses, establishes a consistency with previous analyses on the
phonetics of YM vowel shapes in monosyllables. It also helps to establish that the speech
patterns of speakers recorded for this study are consistent with the descriptions of the
Western dialect of YM, which importantly has a distinction between High and Low
tones, and a distinction between Short and Long vowels. In the next section, an analysis
of duration and pitch in bisyllabic words will be presented, in a first step at determining
whether or not phonetic cues are present that may signal the presence of stress and test
the hypotheses laid out in the beginning of the chapter.
5.4.2
Bisyllabic Words
An analysis of duration and pitch in bisyllabic words allows a view into the
underlying system of prominence in YM we have not yet seen. So far, we’ve looked at
123 native speaker intuition data, which gave us clues about how the speakers themselves
view prominence, and we’ve looked at monosyllabic words to determine the baseline
behavior. Looking at bisyllabic words gives us the first view into what phonetic cues
may be present that signal that a position in a word is prominent.
Between 33 and 38 bisyllabic words are analyzed from each of the 5 speakers that
participated in this section of the study42. The words were chosen from YM dictionaries
because they represented a broad a range of vowel shape combinations. The vast majority
of YM stems are monosyllabic, and those that are made up of more than one syllable
often contain inflectional and derivational morphology, most of which is suffixed to the
stem. Because of this, the majority of YM bisyllabic words take the form Long-Short,
because the majority of stems have long vowels, and the majority of suffixes have short
vowels. Bisyllabic words that contain two adjacent long vowels are rare, and can often be
analyzed as compound words, and words with two short vowels are likewise rare. All the
different combinations of bisyllabic vowel shapes (Short-Short, Long-Long, Long-Short
and Short-Long) were elicited, though the words that were chosen of the rarer varieties
often proved to either be unknown to the speakers, or pronounced differently than
expected often by omitting or shortening a syllable. The following figure shows a
representative sample of the types of bisyllabic words from this list.
Figure 5-9
Bisyllabic Word Types
a) Short-Short
/kosam/
‘swallow’
/chichan/
‘small’
/k’uxub/
‘achiote’
/bajux/
‘how much’
/jats’uts’/
‘pretty’
42
Some words that were recorded were excluded due to the speaker’s unfamiliarity with
them, or because the speaker uttered them with more or less than 2 syllables.
124 b. Short-Long
/k’ahó:l/
/junp’í:t/
/mixmá:k/
/junja:ts/
/jela’an/
/mina’an/
‘to know’
‘a little bit’
‘nobody’
‘a group of’
‘different’
‘none’
c. Long-Short
/xi:mbal/
/xch’ú:pal/
/ú:chben/
/ka:mbal/
/chi’ichnak/
/ba’alche’/
‘to walk’
‘woman’
‘old’
‘to learn’
‘worried’
‘animal’
d. Long-Long
/sé:ba’an/
/cha:mbéel/
/k’á:kná:b/
/yá:xk’i:n/
/u:sá:n/
/xtu:xku:ts/
‘fast’
‘slowly’
‘sea’
‘drought’
‘rash’
‘pheasant’
5.4.2.1 Coding of the Data
The data for this section of the study was coded in a different manner than the
native speaker intuition data, due to the vast difference in nature between categorical
judgment data and continuous duration and pitch data. For native speaker intuitions, a
value for the dependent variable was aggregated from the number of times a syllable in a
particular position with a particular vowel type was judged as strong. For the data
extracted from the production facet of this study, no such mechanism was needed,
because a numerical measurement was available.
Also, because the phonetic cues of duration and pitch are being analyzed for
whether or not they differ significantly by position and vowel type, the notion of stress as
a syntagmatic phenomenon (discussed in chapter 1) are relevant. Syntagmatic
125 phenomena gain their meaning from being compared to the other segments within the
same word, rather than with other possible segments that could be found in that slot.
Tones were described as an example of a paradigmatic entity, because a tone is judged by
a hearer in relation to other tones from the lexicon that could be there, while stress is
paradigmatic because a syllable is considered stressed when it stands out from those
surrounding it, a visual representation of this, repeated from chapter 1, can be seen in the
following diagram.
Figure 5-13
Syntagmatic vs. Paradigmatic features
Because stress is syntagmatic in nature, it is unlikely that a stressed syllable will always
reach a particular duration or pitch target, but rather that it will be relatively longer or
higher in pitch than the syllables surrounding it in the same word. For example, if a
speaker happened to be speaking at a slower pace at the beginning of the recording
session, and each syllable was looked at separately from the syllables surrounding it, and
stress is cued by duration, the stressed syllables recorded early on might be significantly
longer than those recorded at the end. It would appear as if the words recorded at the
beginning had stressed syllables, while those at the end did not, even if in every word one
syllable was longer in relation to the syllables around it.
For this reason, it is specifically the difference in duration and pitch between
syllables within a word that we should look at in order to determine whether either of
these might cue the prominent position(s) in the word. In order to capture this difference
126 between syllables, the data for this part of the study was coded by ‘word type’, instead of
by position alone. Each bisyllabic word was categorized as a combination of vowel types
in all possible positions as described in Figure 5-9 above: SS (Short-Short), SL (ShortLong), LS (Long-Short) or LL (Long-Long). The following section outlines the analyses
for duration in words in isolation and words in context.
5.4.2.1.1 Length
Vowel length measurements were taken of each vowel nucleus in each syllable of
the 33-38 words, and the initial and final syllables were compared to determine whether
or not one was significantly longer than the other.
5.4.2.1.1.1 Words in isolation
Each word was coded for word type, as described above. Table 5-15 shows the
means and standard deviations for each of the 4 word types.
Table 5-15
Means for Bisyllabic Words
Type
Stat
LL
Mean
N
SD
Mean
N
SD
Mean
N
SD
Mean
N
SD
LS
SL
SS
S1
Duration
0.124
29
0.044
0.139
74
0.033
0.069
34
0.027
0.073
32
0.032
S2
Duration
0.132
29
0.045
0.078
74
0.018
0.157
34
0.040
0.089
32
0.021
127 A two-way ANOVA was performed to determine if there were any significant
effects of Position, Vowel Type, or an interaction of the two. The results found a
significant interaction between Position and Vowel Type, F(3,12) = 76.37, p < .001 by
subjects, F(3,42) = 42.26, p < .001 by items. In order to determine the specific nature of
the interaction effect, a repeated measures ANOVA was conducted for each subset of the
conditions to determine whether or not for each type of vowel there was a significant
difference in duration between the initial and final syllable. The results are found in Table
5-16, asterisks indicate significant results.
Table 5-16
Type
LL
LS
SL
SS
Results for duration difference between Initial and Final Syllables
By-Subjects
F(1,4) = .593,
p = .484
F(1,4) = 28.74,
p = .006 *
F(1,4) = 211.4,
p < .001 *
F(1,4) = 3.05,
p = .156
Result
Initial, Final
Initial >> Final
Final >> Initial
Initial, Final
By-Items
F(1,7) = 0.147,
p = .712
F(1,20) = 155.7,
p < .001 *
F(1,7) = 44.29,
p < .001 *
F(1,8) = 5.936,
p = .0408 *
Result
Initial, Final
Initial >> Final
Final >> Initial
Final >> Initial
If the patterns present in monosyllabic words are maintained in bisyllabic words,
and if hypothesis 2 is correct, and Low, High and Glottalized vowels are always
prominent, the prediction follows that Long-Short words would be significantly longer on
the initial syllable and Short-Long words would be significantly longer on the final
syllable. The findings of the ANOVAs show that, as predicted, for Long-Short words,
the initial syllable was significantly longer than the final syllable, and for Short-Long
words the final syllable was significantly longer than the initial. Both hypotheses 2 and 3
would predict that for Long-Long words, the two syllables will be equally prominent,
since both have tone/glottalization, and weight (assuming that long vowels are heavy).
128 The results for Long-Long words show no significant differences in duration between the
initial and final syllable.
Finally, the Short-Short words, which have no Low, High or Glottalized vowels
present, should give us insight into whether or not prominence that is not due to the
presence of Long vowels may exist in YM. However there are 2 distinct predictions that
could be made given hypotheses 3 and 4. Because all of the Short-Short words end in a
consonant, as per YM phonotactic rules, the final syllables all contain codas and are
therefore heavy. Hypothesis 3, which states that heavy syllables are prominent, would be
supported by evidence that the final syllable of Short-Short words is significantly longer
than the initial. Conversely, hypothesis 4, which states that YM words have initial
prominence, would be supported by evidence that the initial syllable was significantly
longer than the final.
The results for the Short-Short words showed that, while the by-subjects analysis
did not find significance, there was a small but statistically significant difference in which
the final syllable was longer than the initial in the by-items analysis. This non-significant
overall finding does not provide evidence for either hypothesis 3 or 4. The lack of a
significant distinction between may also be evidence in line with hypothesis 2. If no long
vowel is present, no prominence would be present. However the lack of significance may
also be due to the small sample size of Short-Short words, and the relative rarity of words
of this type in general.
5.4.2.1.1.2 Words in Phrases
Data for vowel duration in bisyllabic words in phrases was also analyzed, to
determine whether or not the duration findings from the previous section are maintained
129 when words are uttered phrase-medially. A two-way ANOVA was performed, which
once again found a significant interaction effect between Vowel Type and Position,
F(3,12) = 46.50, p < .001 by subjects, F(3,52) = 58.53, p < .001 by items. To analyze the
nature of the interaction effect, a repeated measures ANOVA was run on each subset of
the 4 vowel shape patterns LL, LS, SL and SS, and the results can be seen in Table 5-17.
Table 5-17
Type
LL
LS
SL
SS
Results for differences in duration for bisyllabic words in phrases
By-Subjects
F(1,4) = 1.048,
p = .364
F(1,4) = 173.6,
p < .001 *
F(1,4) = 35.05,
p = .004 *
F(1,4) = 1.015,
p = .371
Result
Initial, Final
Initial >> Final
Final >> Initial
Initial, Final
By-Items
F(1,7) = .466,
p = .517
F(1,27) = 169.5,
p < .001 *
F(1,10) = 38.92,
p < .001 *
F(1,8) = 1.28,
p = .291
Result
Initial, Final
Initial >> Final
Final >> Initial
Initial, Final
The results are identical to those from words in isolation. Both SL and LS continue to
show a significantly longer duration on the Long syllable. SS and LL words in sentences
are also consistent with those in isolation, in that in both contexts the two syllables are
not significantly different from each other.
The next section will look at the phonetic cue of pitch to understand what role it
may or may not play in the cuing of prominence in YM.
5.4.2.1.2 Pitch
Pitch measurements were also taken for each syllable in bisyllabic words, at 5
points along the f0 continuum. Looking at the pitch means for each vowel type
aggregated by speaker can show whether or not the data for bisyllabic words is also
consistent with Frazier (2009)’s description of YM tone and glottalization. The following
130 charts show the means and standard deviations for High, Low, Glottalized and Short in
initial versus final position.
Table 5-18
Position
Initial H
Final H
Initial G
Final G
Initial L
Final L
Initial S
Final S
Pitch Means for H, L, G and S in Initial versus Final position
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
P1
138.26
32.64
147.13
41.43
158.61
34.20
155.66
51.53
139.23
27.37
130.78
27.08
140.50
38.64
135.49
30.33
P2
140.91
35.44
146.54
40.16
154.74
35.81
143.59
40.22
136.11
26.78
131.71
33.40
139.44
37.20
132.47
28.48
P3
143.78
41.09
140.69
36.52
140.88
32.01
126.33
25.98
134.34
26.80
128.92
32.45
137.99
36.24
129.61
27.83
P4
147.07
46.06
133.63
32.20
160.22
53.00
120.02
24.24
132.25
26.75
128.01
32.76
136.33
35.98
127.13
27.25
These trajectories are represented visually in the graphs in Figure 5-11 below.
P5
146.57
46.84
125.35
27.13
147.70
46.65
112.92
20.09
131.48
26.46
125.61
33.24
134.26
35.51
124.97
26.54
131 Figure 5-14
Pitch Contours by Vowel Type
a. High Tone
High Tone in Initial vs. Final Position 150 145 140 Hz 135 130 Initial 125 Final 120 115 110 1 2 3 4 5 b. Low Tone
Low Tone in Initial vs. Final Position 145 140 Hz 135 130 Initial 125 Final 120 115 1 2 3 4 5 132 c. Glottalized
Hz Glottalized in Initial vs. Final Position 180 160 140 120 100 80 60 40 20 0 Initial Final 1 2 3 4 5 d. Short
Short in Initial vs. Final Position 145 140 Hz 135 Initial 130 Final 125 120 115 1 2 3 4 5 The pitch trajectories for the words with Long vowels are all as expected for the speakers
of the Western dialect, i.e., High tones in initial position rise, and in final position the fall.
Low remain relatively level in both initial and final position, with both positions showing
a slight downward trend. Glottalized vowels show sharp peaks and valleys due to
creakiness in both positions, but Glottalized vowels in initial position remain higher,
133 while in final position they fall sharply. Finally the Short syllables are very similar in
slope in both positions, and are actually slightly higher towards the end of the word.
5.4.2.1.2.1 Midpoint Analysis
Now that consistent patterns have been established with what is expected from the
dialect, we will look at whether or not any statistically significant differences between the
pitch of the Initial and Final syllable is evident.
So far we have looked only at the average behavior of S, L, H and G in Initial
versus Final position, without reference to how Initial and Final positions compare within
the word. In order to test how the two syllables compare in pitch for each type of word, a
two-way ANOVA was performed using Vowel Type and Position as independent
variables, and the pitch value at the midpoint as dependent variables. The results of the
ANOVA were significant, F(3,11) = 6.06, p = .011 by subjects, (3,45) = 9.26, p < .001 by
items. A repeated measures ANOVA was conducted which compared the difference
between the midpoint pitch measurement for each subset of conditions, and the following
diagram shows the results for each vowel type.
Table 5-19
Type
LL
LS
SL
SS
Results for difference in pitch between Initial and Final Syllables
By-Subjects
F(1,3) = 3.149,
p = .174
F(1,4) = 10.36,
p = .03 *
F(1,4) = .919,
p = .392
Result
Initial, Final
F(1,4) = 11.45,
p = .03 *
Final >> Initial
Initial >> Final
Final, Initial
By-Items
F(1,6) = 4.245,
p = .085
F(1,25) = 31.48,
p < .001 *
F(1,8) = .455,
p = .519
Result
Initial, Final
F(1,9) = 7.545,
p = .02 *
Final >> Initial
Initial >> Final
Final, Initial
134 Interestingly, significant findings were achieved only for LS and SS word types, for the
both by-items and by-subjects analyses. The significant finding for the LS type words is
not surprising, since the initial syllables are Long and therefore likely to contain tonal
phonemes that affect the pitch level of the syllable. However, the significant finding that
SS words have a significantly higher pitch at the midpoint of the final vowel is a
surprising find. When tonal phonemes are not present, pitch tends to fall naturally over
time, and towards the end of a word due to the lowering of glottal pressure as air leaves
the lungs, however in the bisyllabic Short-Short words in this study, the pitch of the final
syllable is significantly higher than the initial which is the reverse of the behavior one
would expect, since pitch levels tend to fall naturally over the course of the word or
phrase.
If this finding holds up in the phrasal data as well, this may be an indication that
the final position in YM words, when no Long vowel is present, is the default prominent
position. This would be in favor of hypothesis 3, in which heavy vowels are prominent,
since in all the Short-Short words collected for this study, the Initial syllable is Short
while the Final is Heavy. The following section will look at the data collected for words
in frame sentences to determine whether or not the pitch findings are consistent.
5.4.2.1.2.1.1 Pitch at the midpoint in Phrasal Data
The following section outlines the same data points we looked at earlier in the
words spoken in isolation, but for words in phrasal context. The following table shows
the means and standard deviations for f0 measured in Hertz for each of the 5 pitch points
by position.
135 Table 5-20
Position
Initial H
Means for Bisyllabic Pitch by Position
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Final H
Initial G
Final G
Initial L
Final L
Initial S
Final S
P1
148.11
38.71
151.17
43.61
199.47
97.22
178.66
72.78
142.69
34.49
132.72
20.91
152.58
48.49
143.08
38.46
P2
149.32
41.32
150.08
44.77
194.64
102.22
171.40
64.44
140.41
32.58
129.32
20.70
151.35
47.50
141.31
38.28
P3
149.72
45.61
148.08
44.42
205.32
110.21
168.42
57.02
137.45
31.10
125.63
22.23
156.24
60.77
140.07
38.88
P4
151.08
49.50
147.79
46.83
162.42
50.06
162.02
51.46
136.05
31.41
124.47
23.80
149.54
45.99
138.30
39.25
Figure 5-12a-d show these means graphically.
Figure 5-15
Differences in f0 in word initial vs. word final position
a.
High in Initial vs. Final Position 152 151 Hz 150 Initial 149 Final 148 147 146 1 2 3 4 5 P5
149.70
53.53
147.83
50.38
161.57
47.39
157.39
48.29
134.86
31.79
122.49
26.08
148.04
44.10
136.84
39.02
136 b.
Glottalized in Initial vs. Final Position 250 200 150 Hz Initial 100 Final 50 0 1 2 3 4 5 c.
Low in Initial vs. Final Position 145 140 Hz 135 130 Initial 125 Final 120 115 110 1 2 3 4 5 137 d.
Short in Initial vs. Final Position 160 155 Hz 150 145 Initial 140 Final 135 130 125 1 2 3 4 5 As can be seen in Figure 5-12a, in initial position, High tones rise dramatically whereas
Low tones fall slightly. High tones also behave slightly differently for words in phrases
than they did for words in isolation. For words in isolation, High tones in the first
syllable rise, and those in the final syllable fall. For words in phrases, initial High tones
rise sharply, and Final high tones rise less steeply. This makes sense given Frazier’s
description of High tones as falling ‘phrase-finally’ and rising ‘non-phrase-finally’. For
words in isolation, the Final syllables are phrase final, while in the phrasal context the
words are embedded phrase medially, so would therefore not be expected to fall in pitch
in the same way.
In final position, Low tones fall more sharply and High tones continue to rise, but
just slightly. Glottalized vowels in initial position remain high towards the beginning of
the vowel and then fall, whereas in final position they fall gradually. And finally, the
Short vowels interestingly seem to have a small rise vowel medially in initial syllables
that then levels off.
138 In order to compare the pitch behavior within the two syllables of the word, and
discern whether any of the difference seen in the graphs above represent significant
distinctions, a two way repeated measures ANOVA was run just as it was for words in
isolation. The results showed no significant interaction effect of Position and Vowel
Type, F(3,12) = .93, p = .455 by subjects, F(3,51) = .46, p = .711 by items. The main
effect of Position was also not significant, F(3,9) = .075, p = .975 by subjects, F(1,48) =
3.58, p = .065 by Items. Finally, the main effect of Vowel Type also failed to show a
significant result, F(3,9) = 3.94, p = .048 by subjects, F(3,48) = 1.83, p = .155 by items.
Since none of these results were significant, no statistics were performed on the subsets
of the data.
The finding from bisyllabic words in isolation in which the pitch on SS words was
significantly higher word finally is not replicated here. The fact that this result was not
replicated may weaken the support for hypothesis 3, in that pitch may not be signaling a
prominent position word-finally after all.
5.4.2.1.2.2 Contour Analysis
5.4.2.1.2.2.1 Pitch Contours for Bisyllabic words in Isolation and in Phrases.
An analysis of the pitch contour, calculated from pitch point 4 minus 2, was
performed on this data set as well, for both words in isolation and in phrases. The two
way repeated measures ANOVA did not find a significant interaction between Position
and Vowel Type, F(3,11) = 5.27, p = .017 by subjects, F(3,46) = 2.38, p = .082 by items
on words in isolation. The main effect of Position was not significant, F(3,9) = 2.89, p =
.402 by subjects, F(1,43) = 7.2, p = .010 by items. The main effect of Vowel Type was
also not significant, F(3,9) = 1.74, p = .229 by subjects, F(3,43) = 1.17, p = .333. The
same analysis run on the data from words in phrases also showed no significant
139 interaction effect, F(3,12) = 1.28, p = .327 by subjects, F(3,51) = 1.67, p = .186. These
results show that the data for pitch contours for bisyllabic words is not a salient cue for
prominence.
5.4.2.2
Summary
The previous section looked at length and pitch distinctions in bisyllabic words in
isolation and in context. For bisyllabic words containing one long vowel (High, Low or
Glottalized) and one short vowel, the duration proved to be significantly longer on the
non-short vowel in both contexts. This, along with the evidence found in monosyllabic
words, further bolsters the idea that tone and glottalization in YM go hand in hand with
duration, and the lack of these features corresponds to a shorter duration. Words that
contain two long vowels were not significantly different in duration, suggesting that if
duration is a cue for prominence, both positions in the word can be prominent
simultaneously.
Figure 5-13 summarizes the results from the analyses discussed in this section,
and shows what evidence has accumulated thus far in support of the hypotheses stated at
the beginning of the chapter. For each hypothesis, the relevant results from any analyses
from the sections on native speaker intuitions (NSI), monosyllabic words (Mono), and
bisyllabic words (Bi) are listed.
140 Figure 5-13
Yucatec Maya Prominence Rules
Hypothesis 1: Duration cues Prominence
NSI43: No relevant results.
Mono: The baseline duration was found to be significantly longer
for Long vowels than Short vowels. High vowels are significantly higher
in pitch at the midpoint than Low vowels for words in isolation,
establishing that the tonal distinction is relevant for these speakers.
Bi: Duration was found to differ significantly by position for
Vowel Types LS and SL in both words in isolation and in phrases. Pitch
did not show consistent results that might be a cue for prominence, for
either the analysis of the pitch at the midpoint or for the contour.
Hypothesis 2: Non-Neutral vowels are prominent
NSI: Though there was a tendency for speakers to judge Long vowels
stronger than Short vowels, the effect was not found to be significant.
Mono: N/A
Bi: Words of type LS and SL were significantly longer on the non-neutral
syllable.
Hypothesis 3: Weight is prominent
NSI: If Final syllables and/or Long syllables had been found to be
significantly more likely to be judged strong than Initial and/or Short, this
would have been supporting evidence for hypothesis 3. Neither the effects
of Length or Position were significant, and the pattern in the data leaned
towards the Initial rather than the Final for 3 of the 4 speakers.
Mono: N/A
Bi: No significant effects were found in the bisyllabic data that could be
explained only by weight. If there was an effect found that SS words were
significantly longer or higher in pitch on the Final syllable, this may have
been attributable to weight, however, no such effect was found.
Hypothesis 4: Initial syllables are prominent
NSI: Three out of four speakers were more likely to judge the Initial
syllable as strong than the Final, however this effect was not significant.
Mono: N/A
Bi: No effects were found in which Initial position was stronger in either
duration or pitch without a Long vowel also being present.
43
NSI is an abbreviation for native speaker intuition.
141 Hypothesis 5: Phonetic data will accord with speaker intuitions
NSI: Statistical analyses for native speaker intuition were not significant
by either Length, Position, or the interaction of the two, indicating that
speakers may not perceive a robust and consistent system of prominence.
Mono: N/A
Bi: Because no significant effects were found for NSIs, a finding of no
significant effects for bisyllabic data would be in accord with this.
However there were some significant effects, namely that Long vowels are
significantly longer in duration. This may or may not be evidence for
hypothesis 5.
The following section will look at trisyllabic words to see if any additional evidence can
be added to the summary above that tells us whether or not a system of prominence
amongst syllables can be found in YM, and if so, what cues are utilized.
5.4.3
Trisyllabic Words
Trisyllabic words are often if not always made up of more than one morpheme, since root
words in YM are overwhelmingly CVC in nature. These words were less frequent in the
data, and in some cases were seen as unfamiliar to the subjects. Sometimes the subjects
preferred to pronounce them with less than three syllables, by omitting one of the
morphemes word finally, and these examples were omitted from the data set. All of these
words were also analyzed for length and pitch distinctions in isolation and in phrases.
The 10 trisyllabic words in Figure 5-14 below were elicited from each speaker,
which were one of four patterns: Long-Short-Short, Long-Short-Long, Short-Long-Short
and Short-Short-Long. 44
44 These four word types represented all of the words recorded for this study, however
this does not preclude a different word type from being valid in YM. These four may
represent some of the more common combinations of Long and Short vowels, but
determining the exhaustive list of possible tryisyllabic word pairs is left for future
research.
Figure 5-14
142 Trisyllabic Words
Long-Short Short
a. /táanchumuk/
b. /búukinaj/
c. /áak'abtal/
‘half’
‘to get dressed’
‘end of day/life’
Short-Long-Short
d. /buláak'ab/
e. /sit'ríiyo/
f. /k'oja'anil/
‘all night’
‘cricket’
‘illness’
Long-Short-Long
g. /maakalmáak/
h. /sáabukaan/
i. /máasewáal/
‘anybody’
‘bag’
‘indigenous person’
Short-Short-Long
j. /jujunwáal/
k. /kankala’as/
l. /takankuuns/
‘page by page’
‘lizard’
‘to cook’
These words were analyzed for duration and pitch for both words in isolation and words
in context, which will be discussed in the following sections.
5.4.3.1 Length
5.4.3.1.1 Words in Isolation
Table 5-21 shows the means and standard deviations for length of each syllable.
Table 5-21
Pattern
LSL
LSS
SLS
SSL
Mean duration by Word Type
Stat
Mean
StDev
Mean
StDev
Mean
StDev
Mean
StDev
S1 Length
.10193
.02947
.11535
.02156
.07958
.02883
.08541
.02752
This data is represented graphically in Figure 5-16.
S2 Length
.06541
.01876
.05741
.01620
.17137
.02463
.07631
.03370
S3 Length
.13312
.03024
.08255
.01907
.07838
.01885
.14328
.04280
143 Figure 5-16
Mean duration by Word Type
a.
Long-­‐Short-­‐Long 0.14000 0.12000 Seconds 0.10000 0.08000 0.06000 0.04000 0.02000 0.00000 S1 S2 S3 b.
Long-­‐Short-­‐Short 0.14000 0.12000 Seconds 0.10000 0.08000 0.06000 0.04000 0.02000 0.00000 S1 S2 S3 144 c.
SLS 0.18000 0.16000 Seconds 0.14000 0.12000 0.10000 0.08000 0.06000 0.04000 0.02000 0.00000 S1 S2 S3 S2 S3 d.
SSL 0.16000 0.14000 Seconds 0.12000 0.10000 0.08000 0.06000 0.04000 0.02000 0.00000 S1 In order to determine whether or not there were significant differences in length between
each of the three syllables, a two way repeated measures ANOVA was performed
Position (Initial, Medial, Final) and Word Type (LSS, SLS, SSL, LSL) as the
independent variables. The results show a significant interaction effect between the two
145 independent variables, F(6,24) = 98.55, p < .001 by subjects, F(6,20) = 49.65, p < .001 by
items.
Because there are three syllables to compare in this data, three repeated measures
ANOVAs were performed for each pair of syllables to determine the nature of the
interaction effect, comparing Syllable 1 to 2, Syllable 2 to 3 and Syllable 1 to 3. The
following table shows the results of these ANOVAs.
Table 5-22
Type
LSL
Results
LSS
Results
SLS
Trend
SSL
Results
Statistical Analyses for Trisyllabic Length
S1 vs. S2
By-Subjects
F(1,4) =
32.21,
p = .005 *
Initial >>
Medial
F(1,4) =
99.69,
p = .0006 *
Initial >>
Medial
F(1,4) =
845.1,
p < .0001 *
Medial >>
Initial
F(1,4) =
4.528,
p = .1
Initial,
Medial
By-Items
F(1,5) =
69.46,
p = .0004 *
Initial >>
Medial
F(1,2) =
102,
p = .0096 *
Initial >>
Medial
F(1,2) =
44.01,
p = .022 *
Medial >>
Initial
F(1,2) =
1.391,
p = .359
Initial,
Medial
S2 vs. S3
By-Subjects
F(1,4) =
117.7, p =
.0004 *
Final >>
Medial
F(1,4) =
84.6,
p = .0008 *
Final >>
Medial
F(1,4) =
454.4,
p < .0001 *
Medial >>
Final
F(1,4) =
207.5,
p < .0002 *
Final >>
Medial
By-Items
F(1,5) =
90.91,
p = .0002 *
Final >>
Medial
F(1,2) =
22.33,
p = .042
Final,
Medial
F(1,2) =
53.18,
p = .0183 *
Medial >>
Final
F(1,2) =
113.9,
p = .0087 *
Final >>
Medial
S1 vs. S3
By-Subjects
F(1,4) =
13.8,
p = .02 *
Final >>
Initial
F(1,4) =
37.9,
p = .004 *
Initial >>
Final
F(1,4) =
.076,
p = .797
Initial, Final
F(1,4) =
47.79,
p = .002 *
Final >>
Initial
By-Items
F(1,5) =
24.24,
p = .004 *
Final >>
Initial
F(1,2) =
4749,
p = .0002 *
Initial >>
Final
F(1,2) =
.015,
p = .915
Initial, Final
F(1,2) =
16.79,
p = .0547
Initial, Final
Because trisyllabic words are rare in the data to begin with, and also because speakers
occasionally truncated the trisyllabic words to two syllables or were unfamiliar with
them, the data for these word types is sparse, and therefore significant findings are more
difficult to obtain. Despite this, many of the results for comparisons between syllable
146 length in trisyllabic words are significant. I will discuss the results for each word type
separately in the following sections.
5.4.3.1.1.1 LSL
The following table shows the results for only LSL type words.
Table 5-23
Type
LSL
Results
Long-Short-Long Duration
S1 vs. S2
By-Subjects
F(1,4) =
32.21,
p = .005 *
Initial >>
Medial
By-Items
F(1,5) =
69.46,
p = .0004 *
Initial >>
Medial
S2 vs. S3
By-Subjects
F(1,4) =
117.7, p =
.0004 *
Final >>
Medial
By-Items
F(1,5) =
90.91,
p = .0002 *
Final >>
Medial
S1 vs. S3
By-Subjects
F(1,4) =
13.8,
p = .02 *
Final >>
Initial
By-Items
F(1,5) =
24.24,
p = .004 *
Final >>
Initial
In LSL words, as expected, the Initial and Final syllables are significantly longer than the
Medial in both by-subjects and by-items analyses. Both of these results are expected
because the Initial and Final are Long syllables. We might expect the Initial and Final to
be equally long, but instead the Final syllable is significantly longer than the Initial even
though both are Long vowels. This suggests that the Final position may be more
prominent than the initial, even when both contain Long vowels, and may be further
evidence that word-final position is prominent. Another possibility is that the longer final
syllable could also be due to phrase final lengthening, which would be discussed in the
following section which looks at the same data in a phrase medial context.
5.4.3.1.1.2 LSS
The following table shows the results for the LSS words alone.
147 Table 5-24
Type
LSS
Results
Long-Short-Short Duration
S1 vs. S2
By-Subjects
F(1,4) =
99.69,
p = .0006 *
Initial >>
Medial
By-Items
F(1,2) =
102,
p = .0096 *
Initial >>
Medial
S2 vs. S3
By-Subjects
F(1,4) =
84.6,
p = .0008 *
Final >>
Medial
By-Items
F(1,2) =
22.33,
p = .042*
Final >>
Medial
S1 vs. S3
By-Subjects
F(1,4) =
37.9,
p = .004 *
Initial >>
Final
By-Items
F(1,2) =
4749,
p = .0002 *
Initial >>
Final
For LSS words, as expected, the Initial is significantly longer than the Medial and the
Final, in both by-subjects and by-items analyses. In the comparison of Medial to Final,
which in the case of LSS are two adjacent Short vowels, the Short vowel in Final position
is significantly longer than the one in Medial position. This also suggests that the Final
position may either be prominent or be affected by phrase final lengthening.
5.4.3.1.1.3 SLS
The following chart shows the results for the SLS type words alone.
Table 5-25
Type
SLS
Results
Short-Long-Short Duration
S1 vs. S2
By-Subjects
F(1,4) =
845.1,
p < .0001 *
Medial >>
Initial
By-Items
F(1,2) =
44.01,
p = .022 *
Medial >>
Initial
S2 vs. S3
By-Subjects
F(1,4) =
454.4,
p < .0001 *
Medial >>
Final
By-Items
F(1,2) =
53.18,
p = .0183 *
Medial >>
Final
S1 vs. S3
By-Subjects
F(1,4) =
.076,
p = .797
Initial, Final
By-Items
F(1,2) =
.015,
p = .915
Initial, Final
For SLS words, the Medial syllable is significantly longer than both the Initial and Final
syllables. Interestingly, when comparing the Initial and Final syllables, which are both
Short, there is no significant distinction with a very high p level. This indicates that for
trisyllabic words that contain a Long vowel word-medially, that neither Initial nor Final
position is prominent. If the results that showed possible Final prominence in LSS and
LSL words is due to phrase final lengthening, it is curious that SLS words do not undergo
148 the same lengthening phrase finally. This will be discussed further in the following
section.
5.4.3.1.1.4 SSL
The following table shows the results for SSL words alone.
Table 5-26
Type
SSL
Results
Short-Short-Long Duration
S1 vs. S2
By-Subjects
F(1,4) =
4.528,
p = .1
Initial,
Medial
By-Items
F(1,2) =
1.391,
p = .359
Initial,
Medial
S2 vs. S3
By-Subjects
F(1,4) =
207.5,
p < .0002 *
Final >>
Medial
By-Items
F(1,2) =
113.9,
p = .0087 *
Final >>
Medial
S1 vs. S3
By-Subjects
F(1,4) =
47.79,
p = .002 *
Final >>
Initial
By-Items
F(1,2) =
16.79,
p = .0547
Initial, Final
For SSL words, the Initial and Medial words show no significant difference in either
analysis. The Final syllable is significantly longer than the Medial with a high degree of
significance in both by-subjects and by-items analyses, and is significantly longer than
the Initial in the by-subjects but not by-items analyses. This indicates that the Final
position that contains the long vowel is unsurprisingly more prominent than the Medial,
and also more prominent than the Initial but to a slightly lesser degree.
5.4.3.1.2 Summary
The results for all of these patterns indicate that the position containing the Long
vowel, as expected from the nature of long vowels, is longer than syllables that contain a
Short vowel. The results from the SSL and LSS word types are interesting, in that they
contain two Short vowels in a row, and therefore if they showed a significant distinction
between them this might be evidence of prominence. As described above, for LSS
words, the Final was significantly longer than the Medial, but for SSL, the comparison
between Initial and Medial did not render a significant finding. This could indicate that
149 the Final position is more prominent, or that phrase final lengthening is in effect.
However, the results for SLS words are also informative, because comparing the Initial
and Final Short syllables shows no distinction whatsoever, in fact it is the only trisyllabic
word type to show only one position in the word to be durationally prominent. If phrase
final lengthening was in effect, and was present for all final syllables regardless of
phonemes present, we might expect that in SLS words the Final syllable would be longer
than the Initial, which was not the case.
In order to look more closely at the role phrase-final lengthening may or may not
play, the following section will analyze the duration data in trisyllabic words within
phrases.
5.4.3.2 Trisyllabic Duration in Phrasal Data
In order to rule out phrase final lengthening as a confounding factor in the
analyses above, data from trisyllabic words in context was analyzed as well. The means
and standard deviations for length of each syllable in the trisyllabic words can be found
in the following table.
Table 5-27
Pattern
LSL
LSS
SLS
SSL
Means for duration in Trisyllabic words by type
Stat
Mean
StDev
Mean
StDev
Mean
StDev
Mean
StDev
S1
0.106
0.017
0.125
0.023
0.075
0.019
0.090
0.031
This data is represented visually in Figure 5-17.
S2
0.067
0.015
0.053
0.017
0.148
0.033
0.074
0.031
S3
0.119
0.020
0.099
0.021
0.082
0.019
0.143
0.043
150 Figure 5-17
Mean duration by Word Type
a.
Long-­‐Short-­‐Long in Context 0.14 0.12 Seconds 0.1 0.08 0.06 0.04 0.02 0 S1 S2 S3 b.
Long-­‐Short-­‐Short in Context 0.14 0.12 Seconds 0.1 0.08 0.06 0.04 0.02 0 S1 S2 S3 151 c.
Short-­‐Long-­‐Short in Context 0.16 0.14 Seconds 0.12 0.1 0.08 0.06 0.04 0.02 0 S1 S2 S3 d.
Short-­‐Short-­‐Long in Context 0.16 0.14 Seconds 0.12 0.1 0.08 0.06 0.04 0.02 0 S1 S2 S3 In order to determine whether or not there were significant differences in length
between each of the three syllables in this context, a two way ANOVA was performed
using the independent variables of Word Type and Position. The results showed a
significant interaction effect, F(6,24) = 23.91, p < .001 by subjects, F(6,45) = 22.43,
p<.001 by items. The main effect of Position was not significant, F(1,11) = .120, p = .736
by subjects, F(1,22) = 3.54, p = .073 by items. The main effect of Vowel Type was also
152 not significant, F(3.11) = 1.64, p = .237 by subjects, F(3,22) = .80, p = .508 by items. Just
as in the analysis from words in isolation, three repeated measures ANOVAs were
performed, comparing Syllable 1 to 2, Syllable 2 to 3 and Syllable 1 to 3. For ease of
comparison, the results from the previous analysis of words in isolation can be seen in
Table 5-28a, and the results from words in frame sentences can be seen in Table 5-28b.
Table 5-28
Statistical Analyses for Trisyllabic Length
a. Words in Isolation
Type
LSL
Results
LSS
Results
SLS
Results
SSL
Results
S1 vs. S2
By-Subjects
F(1,4) =
32.21,
p = .005 *
Initial >>
Medial
F(1,4) =
99.69,
p = .0006 *
Initial >>
Medial
F(1,4) =
845.1,
p < .0001 *
Medial >>
Initial
F(1,4) =
4.528,
p = .1
Initial,
Medial
By-Items
F(1,5) =
69.46,
p = .0004 *
Initial >>
Medial
F(1,2) =
102,
p = .0096 *
Initial >>
Medial
F(1,2) =
44.01,
p = .022 *
Medial >>
Initial
F(1,2) =
1.391,
p = .359
Initial,
Medial
S2 vs. S3
By-Subjects
F(1,4) =
117.7, p =
.0004 *
Final >>
Medial
F(1,4) =
84.6,
p = .0008 *
Final >>
Medial
F(1,4) =
454.4,
p < .0001 *
Medial >>
Final
F(1,4) =
207.5,
p < .0002 *
Final >>
Medial
By-Items
F(1,5) =
90.91,
p = .0002 *
Final >>
Medial
F(1,2) =
22.33,
p = .042 *
Initial >>
Final
F(1,2) =
53.18,
p = .0183 *
Medial >>
Final
F(1,2) =
113.9,
p = .0087 *
Final >>
Medial
S1 vs. S3
By-Subjects
F(1,4) =
13.8,
p = .02 *
Final >>
Initial
F(1,4) =
37.9,
p = .004 *
Initial >>
Final
F(1,4) =
.076,
p = .797
Initial, Final
F(1,4) =
47.79,
p = .002 *
Final >>
Initial
By-Items
F(1,5) =
24.24,
p = .004 *
Final >>
Initial
F(1,2) =
4749,
p = .0002 *
Initial >>
Final
F(1,2) =
.015,
p = .915
Initial, Final
F(1,2) =
16.79,
p = .0547
Initial, Final
153 b. Words in Phrases
Type
LSL
Results
LSS
Results
SLS
Results
SSL
Results
S1 vs. S2
By-Subjects
F(1,4) =
145.9,
p < .001 *
Initial >>
Medial
F(1,4) =
98.02,
p < .001 *
Initial >>
Medial
F(1,4) =
103.7,
p < .001 *
Medial >>
Initial
F(1,4) =
.505,
p = .517
Initial,
Medial
By-Items
F(1,8) =
59.02,
p < .001 *
Initial >>
Medial
F(1,6) =
41.15,
p < .001 *
Initial >>
Medial
F(1,5) =
44.42,
p = .001 *
Medial >>
Initial
F(1,4) =
.479,
p = .527
Initial,
Medial
S2 vs. S3
By-Subjects
F(1,4) =
193.6,
p < .001 *
Final >>
Medial
F(1,4) =
39.78,
p = .003 *
Final >>
Medial
F(1,4) =
40.45,
p = .003 *
Medial >>
Final
F(1,4) =
15.31,
p = .017 *
Final >>
Medial
By-Items
F(1,8) =
173.8,
p < .001 *
Final >>
Medial
F(1,6) =
28.24,
p = .002 *
Final >>
Medial
F(1,4) =
10.24,
p = .033 *
Medial >>
Final
F(1,4) =
17.45,
p = .014 *
Final >>
Medial
S1 vs. S3
By-Subjects
F(1,4) =
5.049,
p = .0879
Initial, Final
F(1,4) = 11,
p = .0295 *
Initial >>
Final
F(1,4) =
.636,
p = .047 *
Final >>
Initial
F(1,4) =
9.822,
p = .035 *
Final >>
Initial
By-Items
F(1,8) =
5.135,
p = .0532
Initial, Final
F(1,6) =
5.917,
p = .051
Initial, Final
F(1,4) =
.674,
p = .458
Initial, Final
F(1,4) =
10.43,
p = .032 *
Final >>
Initial
The data for words in phrases is generally in line with the data for words in isolation. The
strongest pattern discernable from this data is that hypothesis 2 is supported, i.e. the
position with the Long syllable is significantly longer those with Short syllables.
LSS words also show an interesting trend, in that for both contexts the Final Short
syllable is significantly longer than the Medial Short. The fact that the result is consistent
when the words are spoken in phrasal context indicates that perhaps the distinction exists
apart from phrase-final lengthening effects. The results for SSL data for both words in
phrases and words in isolation are also interesting, because if hypothesis 4 was valid,
154 initial syllables should have prominence. However there is no difference in duration
between the Initial and Medial syllables, so there is no evidence in support of hypothesis
4 in this data.
In summary, hypothesis 2 is supported in the data so far from trisyllabic words,
because Long vowels are in all cases significantly longer than Short vowels. There was a
tendency seen in the data from words in isolation that final Short syllables were
significantly longer than those in other positions, which could have been evidence in
favor of hypothesis 3, however looking at the data in a phrasal context, these effects are
neutralized, indicating that they may have been the result of phrase-final lengthening. The
next section will discuss whether or not any effects of pitch are present which may signal
prominence.
5.4.3.3 Pitch
Finally, we will analyze trisyllabic words in terms of pitch. The following
sections look at pitch at the midpoint and the pitch contour for both words in isolation
and in phrases.
5.4.3.3.1 Midpoint Analysis
5.4.3.3.1.1 Words in Isolation
The following tables show the mean pitch measurements for Glottalized, High,
Low and Short by position in trisyllabic words in isolation.
Table 5-29
Position
Initial H
Medial H
Final H
Medial G
Final G
Initial L
Final L
Initial S
Medial S
Final S
155 Pitch Measurements at the Midpoint
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
Mean
STDev
P1
138.1
35.4
166.9
38.1
124.6
20.8
145.4
37.6
128.0
24.8
139.2
37.7
131.2
33.3
145.0
33.4
156.7
50.8
127.9
31.5
P2
143.0
36.6
156.5
46.5
126.4
18.9
129.7
14.7
131.5
26.4
141.6
37.9
129.7
29.6
143.2
36.6
152
50.4
124.1
28.6
P3
147.1
42.3
159.2
46.2
124.9
19.2
125.4
13.1
135.0
24.8
145.8
39.2
126.6
26.8
143.5
38.0
146.7
47.5
120.5
27.9
P4
153.6
50.6
158.8
43.7
126.8
22.6
117.8
11.2
139.7
31.4
149.0
43.2
123.0
24.0
144.1
38.6
144.3
45.8
117.2
25.3
P5
154.6
55.0
157.7
45.7
120.5
21.4
113.7
13.1
135.0
31.4
146.6
42.4
116.7
21.1
142.3
37.9
142.3
43.8
114.9
24.2
156 Figure 5-18
Mean Pitch in Trisyllabic Words by Position45
a.
Glottalized by Position 160.0 140.0 120.0 Hz 100.0 80.0 Medial G 60.0 Final G 40.0 20.0 0.0 1 2 3 4 5 b.
High by Position 180.0 160.0 Axis Title 140.0 120.0 Initial H 100.0 80.0 Medial H 60.0 Final H 40.0 20.0 0.0 1 2 3 4 5 45
There were no reliable instances of Initial Glottalized or Medial Low tones recorded.
157 c.
Low by Position 160.0 140.0 120.0 Hz 100.0 80.0 Initial L 60.0 Final L 40.0 20.0 0.0 1 2 3 4 5 d.
Short by Position 180.0 160.0 140.0 Hz 120.0 Initial S 100.0 80.0 Medial S 60.0 Final S 40.0 20.0 0.0 1 2 3 4 5 As can be seen in these graphs, High and Low tones tend to behave in a similar fashion in
trisyllabic and bisyllabic words. The most interesting cases to look at for our purposes
are those involving Short vowels, since they do not have any phonemic pitch attributes,
and as can be seen in Figure 5-18d, there are no drastic changes in the behavior of Short
158 vowels in all three positions, though final short vowels are lower in pitch and falling
towards the end, just as in all other vowel shapes besides Glottalized.
In order to determine whether any of these changes between vowels were
significant, a two way ANOVA was performed on pitch at the midpoint using Word Type
and Position as the independent variables. The results showed no significant effect of the
interaction of Word Type and Position, F(6,24) = .47, p = .825 by subjects, F(6,17) = .48
p = .815 by items. The main effect of Position was not significant, F(4,8) = 2.12, p = .170
by subjects, F(3,4) = .03, p = .992. And the main effect of Condition was also not
significant, F(3,8) = 8.03, p = .009 by subjects, F(3,4) = 2.60, p = .190. This lack of an
effect suggests that pitch does not play a salient role in the cuing of prominence, if
prominence is cued at all. The next section looks at pitch for trisyllabic words in phrases.
5.4.3.3.1.2 Trisyllabic Pitch Midpoint in Phrases
A 2 way ANOVA was run on the data set from pitch at the midpoint in trisyllabic
words in phrases, and just as in the previous section, no significant interaction effect was
found, F(6,24) = .88, p = .527 by subjects, F(6,44) = .797, p = .578 by items. The main
effect of Position was not significant, F(6,6) = .71, p = .659 by subjects, F(2,17) = 2.09, p
= .155. The main effect of Vowel Type was also not significant, F(3,6) = .61, p = .631 by
subjects, F(3,17) = 1.02, p = .408. So far, no evidence that pitch is a cue for prominence
of any particular position has been found. The next section will look at whether any
significant effects were found of the pitch contour by position.
159 5.4.3.3.2 Contour Analysis
5.4.3.3.2.1 Trisyllabic Pitch 4-2 Words in Isolation and in Phrases
A two way ANOVA was performed using pitch point 4 minus pitch point 2 from
the words in isolation as the dependent variable, and Word Type and Position as the
independent variables. The results showed no significant interaction effect, F(6,21) =
1.77, p = .153 by subjects, F(6,18) = .91, p = .509 by items. The main effect of Position
was not significant F(5,7) = .20, p = 952 by subjects, F(2,4) = .64, p = .572. The main
effect of Vowel Type was also not significant, F(3,7) = 1.32, p = .341 by subjects, F(3,4)
= 11.57, p .019 by items.
The same test was performed for words in phrases, and similarly, no significant
interaction effect was found, F(6,24) = 1.82, p = .138 by subjects, F(6,45) = 1.59, p =
.172 by items. The main effect of Position was not significant, F(4,8) = .12, p = .937 by
subjects, F(2,19) = .15, p = .858 by items. The main effect of Vowel Type was also not
significant, F(3,8) = .66, p = .601 by subjects, F(3,19) = .90, p = .462. These results
indicate that pitch contours are also not a salient cue for prominence by position in
trisyllabic words.
5.4.3.4 Summary of Trisyllabic Data
The data from trisyllabic words added some interesting findings to those
described in the previous sections. Figure 5-19 shows the hypotheses with supporting
evidence from native speaker intuitions, monosyllabic, bisyllabic and trisyllabic pitch and
duration data.
160 Figure 5-19
Yucatec Maya Prominence Rules
Hypothesis 1: Duration cues Prominence
NSI: No relevant results.
Mono: The baseline duration was found to be significantly longer
for Long vowels than Short vowels. High vowels are significantly higher
in pitch at the midpoint than Low vowels for words in isolation,
establishing that the tonal distinction is relevant for these speakers.
Bi: Duration was found to differ significantly by position for
word types LS and SL in both words in isolation and in phrases. Pitch
did not show consistent results that might be a cue for prominence, for
either the analysis of the pitch at the midpoint or for the contour.
Tri: Duration was found to differ significantly by position and word type.
Long vowels were significantly longer than Short vowels regardless of
place. There were some significant effects of duration in Final vowels that
was not due to the presence of a Long vowel, however phrase final
lengthening effects for these was not ruled out. Pitch was not found to be a
salient cue for prominence in any of the analyses.
Hypothesis 2: Non-Neutral vowels are prominent
NSI: Though there was a tendency for speakers to judge Long vowels
stronger than Short vowels, the effect was not found to be significant.
Mono: N/A
Bi: Words of type LS and SL were significantly longer on the non-neutral
syllable.
Tri: Words of type LSS, LSL, SSL and SLS were all significantly longer
on the non-neutral vowel.
Hypothesis 3: Weight is prominent
NSI: If Final syllables and/or Long syllables had been found to be
significantly more likely to be judged strong than Initial and/or Short, this
would have been supporting evidence for hypothesis 3. Neither the effects
of Length or Position were significant, and the pattern in the data leaned
towards the Initial rather than the Final for 3 of the 4 speakers.
Mono: N/A
Bi: No significant effects were found in the bisyllabic data that could be
explained only by weight. If there was an effect found that SS words were
significantly longer or higher in pitch on the Final syllable, this may have
been attributable to weight, however, no such effect was found.
Tri: LSL words in isolation were found to have a longer duration on the
final Long vowel, and LSS words were found to have a longer duration on
the final Short vowel. These effects were found to be inconsistent for the
words spoken in phrases, which suggests this effect is not due to weight
being prominent, but instead due to phrase final lengthening.
161 Hypothesis 4: Initial syllables are prominent
NSI: Three out of four speakers were more likely to judge the Initial
syllable as strong than the Final, however this effect was not significant.
Mono: N/A
Bi: No effects were found in which Initial position was stronger in either
duration or pitch without a Long vowel also being present.
Tri: No effects were found in which Initial position was stronger in either
duration or pitch without a Long vowel also being present.
Hypothesis 5: Phonetic data will accord with native speaker intuitions
NSI: Statistical analyses for native speaker intuition were not significant
by either Length, Position, or the interaction of the two, indicating that
speakers may not perceive a robust and consistent system of prominence.
Mono: N/A
Bi: Because no significant effects were found for NSIs, a finding of no
significant effects for bisyllabic data would be in accord with this.
However there were some significant effects, namely that Long vowels are
significantly longer in duration. This may or may not be evidence for
hypothesis 5.
Tri: Same results as for Bisyllabic.
5.5
Summary of Stress in YM
This chapter looked at both native speaker intuition data and production data to
determine whether or not evidence exists for the 5 hypotheses outlined in the beginning
of the chapter. A variety of evidence for and against these hypotheses was obtained by
analyzing both sets of data, and is summarized in Figure 5-19 above. The implications of
this evidence, along with opportunities for further research, will be discussed in the
conclusion that follows.
162 CHAPTER 6 CONCLUSION
In chapters 1 through 5 of this dissertation, I introduced the notion of prosody in
YM and illustrated how the notion of stress was an understudied and unclear area of
research, then gave an overview of the language and the previous work done on the topic,
introduced data from Spanish loan words that showed a change from the Spanish side of
the syncretic continuum to the Mayan, and finally discussed data from native speaker
intuitions and productions of YM words, all in order to shed some light on whether or not
stress exists as a measurable or salient entity in YM.
Six separate hypotheses were posited for how stress would likely be realized if it
is present at all, inspired by the data done by previous scholars, as well as by clues
apparent in the incorporation of words from Spanish into YM. Overall, no strong pattern
for stress placement that is cued by duration or pitch emerged from the data. The
following sections 6.1 through 6.5 summarize the results relevant to each of the posited
hypotheses, and then in section 6.6 I compare these results to those of previous scholars.
Finally in section 6.7 I discuss what the lack of significant stress pattern found in either
intuition or production data implicates for the future study of the language.
6.1
Hypothesis 1: Duration cues Prominence
Chapter 3 looked at evidence from Spanish loan words into YM, and found that
when Spanish words were incorporated into the YM accent, they underwent initial
lengthening and final lengthening. This was the main evidence for positing hypothesis 1,
along with the anecdotal evidence posited by scholars of YM that the initial position
might be stressed. Chapter 4 laid out production data for monosyllabic, bisyllabic and
trisyllabic words, both in isolation and in context, with the goal of determining whether
163 there were any distinctions between syllables that could be the result of prominence
rather than the effect of an underlying tonal or glottal phoneme.
Data from monosyllabic words established that duration and pitch were both cues
that were actively being used by the YM speakers to signal a difference in vowels
containing tone or glottalization, from those that don’t. Duration was found to be
significantly longer for High/Low/Glottalized vowels than for those without any tone or
glottalization. Pitch was found to be a salient cue that distinguished High Low and
Glottalized from each other.
Data from bisyllabic words showed that this trend continues to be salient in longer
lexical items. In other words, tones and glottalization co-occur with duration that is
significantly longer than short vowels, regardless of their position in the word. These
length distinctions are not neutralized when the words are spoken within a phrasal
context. Trisyllabic words show a parallel effect, in which each Long syllable in the word
is significantly longer than any Short syllable, regardless of position within the word.
However, does the fact that Long syllables are consistently longer than Short
syllables mean that duration is a cue for stress? The answer to this question is unclear.
What we know is that High tones, Low tones and Glottalized vowels are always longer
than vowels without those phonemic elements. This length could be encoded as part of
the vowel itself, in which case the findings from the data presented for this study do not
show any evidence that duration is a cue for a prominent position, but that it is instead a
cue for only the presence of a tone or glottal vowel.
On the other hand, another possible scenario is that duration is a cue for
prominence, and if YM has a strict rule that tones and glottalization must be stressed,
164 then we would also see High, Low and Glottalized as always being realized as
significantly longer. Though if the latter explanation was valid, and a system of
prominence exists in YM that is cued by duration, we might also expect to see
prominence in words that lack a High, Low or Glottalized vowel. There was only one
instance in the study of two Long vowels or two Short vowels differing significantly at
all, which was when we looked at duration in trisyllabic words in isolation. In that
context, LSL words were significantly longer word finally than word initially, and in LSS
words the final S was longer than the medial S. However this finding was neutralized
when we looked at those same words in a phrasal context, which means that they may be
explained by phrase-final lengthening effects, rather than the signaling of prominence.
But what about the evidence from Spanish loan words? It is possible that the
initial lengthening of Spanish loan words in YM is a separate process that is unrelated to
prominence. Overall, the findings from this study indicate that either of the two following
situations is valid regarding duration as a cue for stress in YM.
Figure 6-1
Possible Scenarios
1. Lexical stress is not present and tonal and glottalized vowels are obligatorily
long in duration. In Spanish loan words, initial position is lengthened by a
phonological process that affects only loan words.
2. Lexical stress is present, is cued by duration, and is strictly applied only to
High, Low and Glottalized vowels regardless of position. In Spanish loan words
only, Initial position is prominent.
These two scenarios would produce identical results in the production data, and therefore
so far it is not possible to say concretely whether or not stress exists in YM and is cued
by duration. However, the results from the production study on bisyllabic words showed
no significant difference between syllables for Long-Long or Short-Short type words in
either pitch or duration. Similarly, there were often cases in which native speakers
165 judged either both or neither of the syllables in these words as strong. The fact that
words seem to exist in YM that do not have a prominent position goes against the notion
of stress being obligatory on words, therefore scenario two seems unlikely.
6.2
Hypothesis 2: Non-Neutral vowels are prominent
In the data from native speaker intuitions, the speakers most often identified the
syllables containing High, Low or Glottalized vowels as being the strongest, though the
effect was not significant. In the data from bisyllabic and trisyllabic words there was a
distinct trend that showed High, Low and Glottalized vowels are significantly longer than
Short vowels in all positions. This could be evidence in favor of hypothesis 2, if stress
does indeed exist in YM, however, it could also be explained by scenario 1 in Figure 6-1.
If High/Low and Glottalized vowels must always co-occur with longer duration, then the
findings from this study would be expected and explained without the presence of
prominence at all.
6.3
Hypothesis 3: Weight is prominent
Due to the YM phonotactic constraint that causes all words to obligatorily end in
a consonant and therefore be arguably closed or heavy syllables, evidence that shows that
final positions are prominent when initial syllables are open light syllables would be
evidence in support of this hypothesis. The native speaker intuition data is not in support
of this hypothesis, as the final position was not consistently judged stronger. The
production data does not provide concrete support for this hypothesis because most
findings of significantly longer duration on the Final syllable were neutralized when
looking at the words in phrasal context. Trisyllabic words in isolation also showed some
evidence for Final syllable prominence cued by duration, but that evidence was not
166 upheld in the phrasal context. Overall, the findings from this study show no evidence that
weight attracts any kind of prominence, and so far does not even concretely prove that
prominence exists in YM at all.
6.4
Hypothesis 4: Initial syllables are prominent
The data from Spanish loan words into YM appeared to provide evidence that the
Initial position in YM is prominent, because words shifted from the Spanish
pronunciation into the YM pronunciation were elongated on the initial syllable. The data
from the native speaker intuition section of this study showed that though for some
speakers there was a tendency to judge Initial syllables as prominent, each speaker had a
slightly different pattern, and the data was too variable to find significance. The
production section of this study does not show any support for prominence in the initial
position cued by duration, and evidence for initial prominence cued by pitch is
inconsistent, and the initial lengthening found in the case studies of Spanish loan words
may likely be a loan word specific process.
6.5 Hypothesis 5: Phonetic data will accord with speaker intuitions
Finally, the hypothesis that the patterns found in the data from the production study will
accord with the intuitions of the native speakers can be argued to be supported in the
results of this study. As discussed above, there is no strong evidence that a lexical stress
system exists, which applies an obligatory word-level stress to each word, and is cued by
either pitch or duration. The native speaker intuitions also did not show any significant
pattern emerging for the judgments of emphasis on YM words. The phonetic data and
speaker intuitions are in accord in that they both show no consistent word level
prominence.
6.6
167 Comparison to previous research
As was discussed in chapter 5, the results from the production of monosyllabic
words is in accord with Frazier (2009)’s findings for the Western dialect of YM, which
are repeated in the following figure.
Figure 6-2
Frazier (2009)’s Western dialect
a. Long vowels (High tone, Low tone and Glottalized) are about twice as long as
Short (neutral) vowels
b. Low tone vowels are produced with fairly steady low pitch
c. Short tone vowels have steady mid or mid-high pitch
d. Glottalized and High vowels have initial high pitch and a falling contour
e. High vowels in non-phrase final position only High vowels reverse to a rising
contour
f. Glottalized vowels do not contain a medial glottal stop as described in much of
the literature, but instead a more accurate description is that they start with High
pitch and end in creaky voice
All of the characteristics listed in Figure 6-2 were present in the speech recorded for this
study, who all importantly showed a significant distinction between High and Low tones,
and Long and Short vowels. Frazier’s study does not argue for any particular stress
pattern on YM words, and because it looks only at monosyllabic words, is unable to
provide evidence for phonetic prominence on the word level.
As discussed in chapter 3, a variety of scholars have argued for the placement of
stress in YM in various locations. Krämer (2001), Bricker et. al. (1998) and
Gussenhoven & Teeuw (2008) taken together posited that either tones, glottalized
vowels, initial or heavy syllables might be the bearers of stress. The data collected for
this study show only that tones and glottalized vowels are consistently longer in duration
than other vowels, but not that they receive any word level prominence. It can be argued
that if word level prominence was a feature of YM, the results of either the native speaker
intuitions or the production data should have shown some distinction between syllables of
either LL or SS bisyllabic words, or a distinction between syllables of the same length in
168 trisyllabic words. No patterns of this sort emerged from the data. Similarly, there was no
evidence from native YM words that initial position was prominent in either pitch or
duration distinctions or native speaker intuition. Weight was also not shown to be a
significant factor in word-level prominence from either intuitions or phonetic data.
6.7
Areas of future study
The analysis presented here is an initial step into understanding the nature of the
prosodic system in YM, however there are a variety of areas in which future study could
give us a much better and more concrete understanding of prominence in YM. In
particular, a study with more participants that focused on whether prominence might be
cued by something other than pitch or duration would be enlightening. It is possible that
consonantal effects, vowel quality, or other segmental effects of prominent or nonprominent syllables might lead to a pattern of prominence not found in this analysis.
A comprehensive look into the intonational system of YM would also prove most
useful, particularly in looking at how the intonational accent system works. It is possible
that the patterns noted by scholars such as Krämer and Bricker et. al. are in fact at play on
the level of intonational accent, though not on the word level. A deeper understanding of
the intonation system of YM is the next important step in understanding how prosody
works in this language, and would prove invaluable for the study of prosody as a whole.
6.8
Summary
By looking comprehensively at both the production of bisyllabic and trisyllabic
words, as well as the collection and analysis of intuitions on prominence, this study is
unique in the current literature on Yucatec Maya. Data has been presented from native
speaker intuitions, as well as production of YM by native speakers, to try to untangle the
169 complicated system of prosody. Specifically, the aim was to determine whether or not
the concept of metrical stress exists for YM words, and if so, to elucidate the nature of
the metrical system, to determine if past arguments for stress placement could be verified.
The results of this study show that the only reliable pattern found throughout the
data is that for YM words that contain a High, Low or Glottalized vowel, that vowel is
significantly longer in duration compared to those that contain Short or Neutral vowels,
which is not evidence in support of the existence of word level stress. Prominence that
exists apart from Long vowels was not apparent in the phonetic data collected for this
study, and words without any long vowels in them showed no emergent pattern of
prominence.
The results of this study, that the only significant distinctions from one syllable to
another can be predicted by the presence of a tone or glottalized phoneme, could signal
that either of the hypotheses in the following figure are correct.
Figure 6-3
Final hypotheses
Hypothesis 1: Metrical stress exists in YM, but is signaled neither by pitch or
duration, and is instead signaled by an as yet undiscerned phonetic cue.
Hypothesis 2: No metrical stress exists in YM.
It is possible the future research will uncover that metrical stress in YM does in fact exist,
and is cued by a phonetic feature other than pitch or duration, such as vowel quality or
weakening of unstressed syllables. However the data collected for this study leads to the
conclusion that the two default cues for the presence of stress cross-linguistically,
duration and pitch, cannot be concretely connected to an obligatory system of word level
metrical prominence. This would place Yucatec Maya in a rare class of languages, those
that lack a metrical stress system, and further study of this and other such languages is
170 important to fully understand the range of how prosody and prominence are realized in
human speech.
171 WORKS CITED
Andrade, M. J. (1940). A grammar of modern Yucatec. Manuscripts on Middle American
Cultural Anthropology , 7 (41).
Beckman, M. (1986). Stress and Non-Stress Accent. Dordrecht, Holland: Walter de
Gruyter.
Beckman, M. (1996). The Parsing of Prosody. Language and Cognitive Processes , 11,
17-67.
Beckman, M., & Ayers Elam, G. (1993). Guidelines for ToBI Labeling. The Ohio State
University Research Foundation.
Beckman, M., & Pierrehumber, J. (1986). Intonational structure in Japanese and English.
Phonology Yearbook , 3, 225-309.
Berinstein, A. (1979). A Cross-Linguistic Study on the Perception and Prodcution of
Stress. UCLA Working Papers in Phonetics , 47.
Blair, R. W. (1964). Yucatec Maya noun and verb morpho-syntax. Dissertation submitted
to Indiana University.
Blair, R. W., & Vermont-Salas, R. (1965). Spoken (Yucatec) Maya. Chicago: Department
of Anthropology, University of Chicago.
Boersma, P., & Weenink, D. (2006). Praat: doing phonetics by computer. Retrieved from
Praat.org: http://www.praat.org/
Bolinger, D. (1958). A theory of pitch-accent in English. Word , 14, 109-149.
Bricker, V., Po'ot Yah, E., & Dzul de Po'ot, O. (1998). A Dictionary of the Maya
Language as Spoken in Hocabá, Yucatán. Salt Lake City: The University of Utah Press.
Campbell, L. (1985). Mayan Linguistics: Where are we now? Annual Review of
Anthropology , 14, 187-198.
Clements, G. (1985). The Geometry of Phonological Features. Phonology Yearbook 2 ,
187-198.
Cutler, A., Dahan, D., & van Donsellar, W. (1997). Prosody in the Comprehension of
Spoken Language: A Literature Review. Language and Speech , 40 (2), 141-201.
Dainora, A. (2001). Eliminating Downstep in Prosodic Labeling of American English.
Proceedings of the ISCA Workshop on Prosody and Speech Recognition and
Understanding, (pp. 41-46). Red Bank, NJ.
De Lacy, P. (2002). The Interaction of Tone and Stress in Optimality Theory. Phonology
, 1-32.
De Lacy, P. (1999). Tone and Prominence. Retrieved from Rutgers Optimality Archive
#333.
Delugan, R. M. (2010). Indigeneity Across Borders: Hemispheric MIgrations and
Cosmopolitan Encounters. American Ethnologist , 37 (1), 83-97.
Eberhard, D. (1995). Mamaindé Stress: The Need for Strata. The Summer Institute of
Linguistics and the University of Texas, Arlington.
Everett, D., & Everett, K. (1984). On the relevance of syllable onsets to stress placement.
Linguistic Inquiry , 705-711.
Fitzgerald, C. (1999). Loanwords and Stress in Tohono O'odham. Anthropological
Linguistics , 41 (2), 193-208.
172 Fitzpatrick, J. (2000). On intonational typology. In Siemund (Ed.), Methodological Issues
in Language Typology (pp. 88-96).
Frazier, M. (2009). The Production and Perception of Pitch and Glottalization in Yucatec
Maya. Dissertation submitted to the University of North Carolina at Chapel Hill.
Frazier, M. (2011). Tonal Dialects and Consonant-Pitch Interaction in Yucatec Maya. In
Avelino (Ed.), New Persepectives in Mayan Linguistics (Vol. 59, pp. 21-55). Cambridge:
Cambridge University Press.
Fry, D. (1955). Duration and Intensity as Physical Correlates of Linguistic Stress. Journal
of the Acoustical Society of America , 35, 765-769.
Fry, D. (1958). Experiments in the Perception of Stress. Language and Speech , 1, 120152.
Güémez Pineda, M. (1994). La lengua maya en Yucatán: una perspectiva
sociodemográfica. I'inaj, Semilla de Maíz, Revista de divulgación del patrimonio cultural
de Yucatán .
Geluykins, R., & Swerts, M. G. (1994). Prosodic cues to discourse boundaries. Speech
Communication , 15, 69-77.
Gordon, R. (2005). Ethnologue: Languages of the World, 15th Edition. Dallas: SIL
International.
Gussenhoven, C. (1999). Discreteness and gradience in intonational contrasts. Language
and Speech , 42, 281-305.
Gussenhoven, C. (2004). The Phonology of Tone and Intonation. Cambridge: Cambridge
University Press.
Gussenhoven, C., & Teeuw, R. (2008). A moraic and a syllabic H-tone in Yucatec Maya.
In Herrera, & Butragueño (Eds.), Fonología Instrumental: Patrones Fónicos y Variación
(pp. 49-72). D.F., MX: El Colegio de México.
Hammond, M. (1989). Lexical Stresses in Macedonian and Polish. Phonology , 6 (1), 1938.
Hanks, W. (2012). Birth of a Language: The formation and spreas of Colonial Yucatec
Maya. Journal of Anthropological Research , 68 (4), 449-471.
Hanks, W. (2010). Converting words: Maya in the age of the cross (Vol. 6). University of
California Press.
Hayes, B. (1995). Metrical Stress Theory: Principles and Case Studies. Chicago:
University of Chicago Press.
Herman, R., & Tevis McGory, J. (2002). The conceptual similarity of Intonational Tones
and its effects on intertranscriber reliability. Language and Speech , 45 (1), 1-36.
Hill, J. (1999). Sycretism. Journal of Linguistic Anthropology , 9 (4), 244-246.
Hill, J., & Hill, K. C. (1986). Speaking Mexicano: Dynamics of syncretic language in
Central Mexico. Tucson: The University of Arizona Press.
Hill, J., & Hill, K. C. (1968). Stress in the Cupan (Uto-Aztecan) Languages. International
Journal of American Linguistics , 34 (4), 233-241.
Hirst, D., & Di Cristo, A. (1998). Intonation systems: a survey of twenty languages.
Cambridge: Cambridge University Press.
Hyde, B. (2007). Non-finality and weight-sensitivity. Phonology , 24 (2), 287-334.
Hyman, L. (2006). Word-prosodic typology. Phonology , 23 (2), 225-257.
Inkelas, S., & Leben, W. R. (1988). Serbo-Croation Pitch Accent: The Interaction of
Tone, Stress and Intonation. Language , 34 (4), 233-241.
173 Inkelas, S., & Zec, D. (1988). Serbo-Croation Pitch Accent: The Interaction of Tone,
Stress and Intonation. Language , 64 (2), 227-248.
Kager, R. (1989). A metrical theory of stress and destressing in English and Dutch (Vol.
14). Walter de Gruyter.
Kahn, D. (1976). Syllable-based generalizations in English phonology. MIT.
Keating, P. (1996). The Phonology-Phonetics Interface. In U. Kleinhenz (Ed.), Interfaces
in Phonology. Akademie Verlag.
Kidder, E. (2008). Tone, Intonation, Stress & Duration in Navajo. Coyote Papers:
Working Papers in Linguistics, Linguistic Theory at the University of Arizona , 16, 55-66.
Krämer, M. (2001). Yucatec Maya Vowel Alternations - Harmony as Syntagmatic
Identity. Zietschrift für Sprachwissenschaft , 20, 175-217.
Ladd, R. D. (1996). Intonational Phonology. Cambridge Studies in Linguistics , 79.
Leben, W. (1980). A Metrical Analysis of Length. Linguistic Inquiry , 11 (3), 497-509.
Leben, W. (1977). Length and Syllable Structure in Hausa. Studies in African Linguistics
, Supplement 7, 137-143.
Leben, W. (1996). Tonal Feet and the Adaptation of English Borrowings Into Hausa.
Studies in African Linguistics , 25 (2), 139-154.
Lehiste, I. (1970). Suprasegmentals. Cambridge: The M.I.T. Press.
Límon Rojas, M. (1997). Diccionario de la lengua Maya. MX: Instituto Nacional para la
Educación de los Adultos, Secretaria de Educación Pública.
Lindstrom, E., & Remijsen, B. (2005). Aspects of the prosody of Kuot, a language where
intonation ignores stress. Linguistics , 43 (4), 839-870.
McDonough, J. (2003). The Navajo Sound System. Dordrecht: Kluwer Academic
Publshing.
McQuown, N. (1968). La estructura tonal de las sílabas del maya yucateco. Estudios de
Cultura Maya , 7, 293-301.
Michnowicz, J. (2008). Final Nasal Variation in Merida, Yucatan. Spanish in Context , 5
(2), 278-303.
Morton, J., & Jassem, W. (1965). Acoustical correlates of Stress. Language and Speech ,
159-181.
Mosley, E. H. (1980). From Conquest to Independence: Yucatan under Spanish Rule,
1521-1821. In E. Mosley, & E. Terry, Yucatan: A World Apart (pp. 83-121). Alabama:
University of Alabama Press.
Nakatami, L., & Aston, C. H. (1978). Acoustic and Linguistic Factors in Stress
Perception. Murray Hill, NJ: ATT Bell Laboratories.
Navarrete, J. A. (2009). Diccionario Introductorio Español - Maya Maya-Español.
Quintana Roo, MX: Universidad de Quintana Roo.
Niang, M. O. (1997). Constraints on Pulaar Phonology. Lanham, MA: University Press
of America, Inc.
Orie, O., & Bricker, V. (2000). Placeless and Historical Laryngeals in Yucatec Maya.
International Journal of American Linguistics , 66 (3), 283-317.
Ortega-Llebaria, M. (2006). Phonetic cues to stress and accent in Spanish. Selected
proceedings of the 2nd conference on Laboratory Approaches to Spanish Phonetics and
Phonology, (pp. 104-118). 2006.
174 Pierrehumbert, J. B., & Hirschberg, J. (1990). The meaning of intonation contours in the
interpretation of discourse. In Cohen, Morgan, & Pollack (Eds.), Plans and Intentions in
Communication and Discourse (pp. 271-311). Cambridge, MA: MIT Press.
Pierrehumbert, J. (1980). The phonology and phonetics of English intonation. Doctoral
Dissertation, MIT.
Pike, E., & Cowan, J. H. (1967). Huahuapan Mixtec phonology and morphophonemics.
Anthropological Linguistics , 9 (4), 1-15.
Pike, K. L. (1946). Phonemic Pitch in Maya. International Journal of American
Linguistics , 12 (2), 82-88.
Prince, A. S. (1980). A metrical theory for Estonian quantity. Linguistic Inquiry , 511562.
Selkirk, E. O. (1981). English Compounding and the Theory of Word-structure. In M.
Moortgat, H. Van der Hulst, & T. Hoestra, The Scopt of Lexical Rules. Dordrecht: Foris.
Selkirk, E. (1993). Sentence Prosody: Intonation, Stress and Phrasing. In J. Goldsmith
(Ed.), Handbook of Phonological Theory. Oxford: Blackwell.
Straight, S. H. (1976). The Acquisition of Maya Phonology: Variation in Yucatec Child
Language. New York: Garland Publishing, Inc.
Suarez, J. (1983). The Mesoamerican Indian Languages. Cambridge: Cambridge
University Press.
Tozzer, A. M. (1921). A Maya Grammar: With Bibliography and Appraisement of Works
Noted. Cambridge, MA: Peabody Museum of American Archaeology and Ethnology.
Varga, L. (2002). Intonation and Stress: Evidence from Hungarian. New York: Palgrave
McMillan.
Xu, Y. (1997). Contextual tonal variations in Mandarin. Journal of Phonetics , 25 (1), 6183.
Yip, M. (1989). Feature Geometry and Cooccurrence Restrictions. Phonology , 6 (2),
349-374.
Yip, M. (2002). Tone. Cambridge: Cambridge University Press.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement